VORTEX GENERATION AND BUBBLE DYNAMICS IN

A MICROFLUIDIC CHAMBER WITH ACTUATIONS

SHANG XIAOPENG

SCHOOL OF MECHANICAL & AEROSPACE ENGINEERING

2017
VORTEX GENERATION AND BUBBLE DYNAMICS IN

A MICROFLUIDIC CHAMBER WITH ACTUATIONS

SHANG XIAOPENG

School of Mechanical & Aerospace Engineering

A thesis submitted to Nanyang Technological University in partial fulfillment of

the requirement of the degree of
Doctor of Philosophy

2017
 Abstract

Abstract

This dissertation presents a comprehensive study on vortex generation, vortex control and
bubble dynamics in a microfluidic chamber with PZT (lead-zirconate-titanate) actuations, which
can be applied to mixing enhancement and manipulations of microparticles and nanoparticles in
microfluidics. The microfluidic chamber is featured with an acoustic resonator shape, and the
PZT disk is formed as part of the chamber walls. The working fluid (DI water) is pumped into
the chamber with a controllable flowrate. Both vortices and bubbles are generated inside the
chamber, depending on the actuation conditions.
In the study on vortex generation, it is found that, at proper actuation conditions, large-scale
vortices are formed inside the chamber, and the vortex generation and intensity can be regulated
by the operation conditions, including actuation frequency, voltage and the flowrate through the
microfluidic chamber. There is an actuation frequency window between 1.5 kHz and 3 kHz for
the vortex generation and at the actuation frequency out of this window, the vortex is very weak
or even unable to occur. The vortex intensity is found to increase with actuation voltage, and
decreases with the flowrates.
The direction of the generated vortex can, in principle, be clockwise or counter-clockwise
due to the symmetry of the chamber and the PZT actuator. However, it is found that only one
type of vortex can occur in an individual microfluidic chamber. This is probably due to the
imperfection in the chamber fabrication. The control of the vortex direction has been carried out.
The microfluidic chamber is modified by dividing the piezoelectric transducer into two parts
which can be actuated separately, so that the actuation becomes asymmetric. In the modified
microfluidic chamber, the vortex direction can be shifted on-demand from clockwise to
counterclockwise by a switch to alter the working transducer.
The mechanism of the vortex generation in the present study is investigated. It is found that
the vibration of the PZT disk induces a large-amplitude pulsatile flow at the outlet channel of the
chamber, where the geometry and configuration are similar to a sudden expansion from a narrow
channel to a wide chamber. It is proposed that the vortex generation in the present study is due to
the instability of the actuation-induced pulsatile flow through the sudden expansion part of the
outlet channel at high Reynolds number. The flow velocity, corresponding to the actuation, at the

I
 Abstract

outlet channel is measured to confirm that the flow Reynolds number at the working conditions
is high enough for the vortex generation.
Bubbles can be generated inside the chamber when the actuation voltage is larger than a
threshold and the actuation frequency is within the range between 0.5 kHz and 5 kHz. The
bubbles can be generated in forms of a single bubble, or as a cloud of bubbles, depending on the
applied voltage. Such type of bubble generation is novel compared to the conventional bubble
generations in other microfluidic devices reported in open literatures.
The dynamics of multi-bubbles in the chamber with actuations at 1 kHz is investigated both
experimentally and numerically. It is found that, under actuations, the bubbles are generated near
the center of the chamber, grow up in size and move upstream against the hydrodynamic flow
from inlets. Along with the bubbles’ motions, coalescences and breakups happen frequently and
cause vigorous motions of surrounding liquids, which has a major contribution to the mixing
enhancement. The pressure variation and distribution in the chamber are calculated by numerical
simulation. The simulation results show that inside the chamber there is a low-pressure zone,
which is consistent with the area of bubble generation observed in the experiment. The amplitude
of the actuation exceeds the Blake threshold of nuclei (> 10 μm), which leads to the explosive
growth of the nuclei in the working fluids, and hence the bubble are generated and observed in
the experiments.
The dynamics of a single bubble is investigated both experimentally and analytically. It is
found that the bubble, after being generated, translates upstream against the flow at an average
velocity of about 0.23 m/s, and undergoes significant expansion and compression with a ratio of
the maximum to minimum volume up to 22. In order to understand the bubble translation under
actuations, a simplified analytical model which describes the dynamics of a bubble nearby two
intersected boundaries is constructed by using the Lagrangian scheme and the method of images.
The analytical results of the model are found to be agreeable with the experimental observations
in terms of the bubble expansion, compression and translation. It is concluded that the bubble
translation is caused by the attractive secondary Bjerknes force of image bubbles behind the two
inclined walls, which is the so-called wall effect.

II
 Acknowledgements

Acknowledgments

First and above all, I would like to express my special thanks of gratitude to my supervisor,
Associate Professor Huang Xiaoyang, for his patient help, inspiring guidance in my research and
sincere solicitude in my personal life during the past years.
I would like to thank Associate Professor Yang Chun, Charles, my co-supervisor, for his
enthusiastic enlightenment, continuous motivation and endless support in my study and research
during past years.
I would also like to express my appreciation to our technicians Mr. Yuan Kee Hock, Mr.
Cheo Hock Leong for their help in experiments. I also thank Dr. Cui Xinguang, Yan Zhibin, Zhu
Guiping and Huang Yi and other friends, who share their knowledge and experience with me in
these days.
Finally, I would like to thank my wife Li Chong and my daughter Linda, whose love and
support have made this work completed.

III
 Abbreviations

Abbreviations

AC alternating current
DI water de-ionized water
ICEO induced-charge electroosmosis
LDV Laser Doppler Vibrometer
PCR polymerase chain reaction
PIV Particle image velocimetry
PMMA polymethylmethacrylate
PZT lead-zirconate-titanate
RD rectified diffusion
Re Reynolds number
TOMV twin opposing microvortice
UDF user-defined-function

IV
 Contents

Contents

Abstract ............................................................................................................................................ I

Acknowledgments......................................................................................................................... III

Abbreviations ................................................................................................................................ IV

Contents ......................................................................................................................................... V

List of Figures ............................................................................................................................... IX

List of Tables ............................................................................................................................ XVII

Chapter 1 Introduction .................................................................................................................... 1

1.1 Background and motivation .................................................................................................. 1

1.2 Objectives .............................................................................................................................. 3

1.3 Outline of the Dissertation .................................................................................................... 3

Chapter 2 Literature Review ........................................................................................................... 5

2.1 Introduction ........................................................................................................................... 5

2.2 Microfluidic vortex generator ............................................................................................... 6

2.2.1 Vortex in microfluidics ................................................................................................... 6

2.2.2 Passive vortex generators ............................................................................................... 7

2.2.3 Active vortex generators ............................................................................................... 10

2.3 Cavitation inception and thresholds .................................................................................... 17

2.3.1 Cavitation nuclei as bubble sources.............................................................................. 17

2.3.2 Cavitation threshold ...................................................................................................... 18

2.3 Radial dynamics of a bubble in liquid ................................................................................. 20

2.3.1 Rayleigh-Plesset equation and extensions .................................................................... 21

2.3.2 Linear oscillation of a bubble in oscillatory pressure fields ......................................... 23

2.3.3 Nonlinear oscillation of a bubble in oscillatory pressure fields ................................... 23

V
 Contents

2.4 Translational dynamics of a bubble in oscillatory pressure fields ...................................... 26

2.4.1 Bjerknes force and drag force ....................................................................................... 27

2.4.2 Decoupled model of an oscillating bubble with weak actuations ................................ 28

2.4.3 Coupled model of an oscillating bubble with strong actuations ................................... 29

2.5 Bubble boundary interaction ............................................................................................... 31

2.6 Bubble dynamics in microfluidics....................................................................................... 33

Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization ................................ 37

3.1 Experimental methods ......................................................................................................... 37

3.1.1 Design and fabrication of the microfluidic chamber for vortex generation ................. 37

3.1.2 Experimental setup and characterization ...................................................................... 39

3.2 Results and discussion......................................................................................................... 41

3.2.1 Flow pattern development in the microfluidic chamber ............................................... 41

3.2.2 Effects of actuation frequency on the generated vortices ............................................. 44

3.2.3 Effects of the applied voltage on the generated vortices .............................................. 47

3.2.4 Effects of flowrates on the generated vortices.............................................................. 49

3.2.5 The vortex generation in a circular chamber ................................................................ 51

3.2.6 The counter-clockwise vortices .................................................................................... 52

3.3 Summary ............................................................................................................................. 54

Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism ................... 55

4.1 Introduction ......................................................................................................................... 55

4.2 Experimental methods ......................................................................................................... 56

4.2.1 Modification of the microfluidic chamber and experimental setup ............................. 56

4.2.2 Measurement of the flow velocity in the outlet channel .............................................. 58

4.3 Results and discussion......................................................................................................... 59

4.3.1 Visualization of vortex generation and control by fluorescent dyes ............................ 59

VI
 Contents

4.3.2 Effects of actuation conditions on the generated vortices ............................................ 61

4.3.3. Mechanism of vortex generation and control .............................................................. 64

4.3.4. The flow velocity in the outlet channel ....................................................................... 68

4.3.5. Vibration measurement of the PZT disk...................................................................... 71

4.4 Summary ............................................................................................................................. 73

Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber .............................................. 75

5.1 Introduction ......................................................................................................................... 75

5.2 Experimental methods ......................................................................................................... 76

5.3 Results and discussion......................................................................................................... 77

5.3.1 Visualization of bubble generation by fluorescent dyes ............................................... 77

5.3.2 Motions of bubbles in the nozzle.................................................................................. 79

5.3.3 Bubble coalescence and breakup .................................................................................. 81

5.3.4 Bubble-wall interactions ............................................................................................... 85

5.4 Numerical simulation .......................................................................................................... 87

5.4.1 Governing equations and boundary conditions ............................................................ 88

5.4.2 Bubble generation by low pressure .............................................................................. 90

5.4.3 Volume pulsation by pressure fluctuation .................................................................... 92

5.5 Summary ............................................................................................................................. 93

Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber ............................................ 95

6.1 Introduction ......................................................................................................................... 95

6.2 Experimental results and discussion ................................................................................... 96

6.2.1. Bubble trajectory of translation ................................................................................... 97

6.2.2. Radial oscillation and its interaction with translation ................................................. 98

6.3 Analytical modelling ......................................................................................................... 100

6.3.1. Analytical modelling for bubble oscillation and translation ..................................... 101

VII
 Contents

6.3.2. Modelling results of bubble oscillation and translation............................................. 109

6.3.3. Further discussion of bubble translation – wall effect ............................................... 111

6.4 Summary ........................................................................................................................... 115

Chapter 7 Conclusions and Recommendations........................................................................... 117

7.1 Conclusions ....................................................................................................................... 117

7.2 Recommendations for Future Work .................................................................................. 119

Publication List ........................................................................................................................... 120

References ................................................................................................................................... 121

VIII
 List of Figures

List of Figures

Fig. 2.1 Illustration of vortex generators with ridge arrays and generated vortices in
microchannels. (a) sloped ridges; (b) herringbone ridges. [17] .............................................. 7
Fig. 2.2 (a) Dimensions of the vortex generator integrating “Y” junctions with zigzag
microchannels. (b) Velocity vectors obtained at Re=267 for zigzag geometries. [18] ........... 8
Fig. 2.3 (a) Schematic illustration of inertial vortex generator. (b) Illustration of the separation
principle. The symmetric vortices are indicated by the red dashed lines. (c) Stacked images
at various downstream positions illustrating the separation phenomenon at Re=110. [19] .... 9
Fig. 2.4 Flow patterns in the 3D flow-focusing geometry for varying Re. Comparison between
experimental (grey-scale images) and numerical pathlines at the centre plane. (a) Re=2.8; (b)
Re=6.5; (c) Re= 94.2; (d) Re= 113. [20] ................................................................................ 10
Fig. 2.5 Flow patterns and concentration fields (a) with induced-charge electroosmosis effect and
(b) without induced-charge electroosmosis effect. [27] ........................................................ 11
Fig. 2.6 Schematic illustration of the TOMV (twin opposing microvortices) flow with two
microvortices (circular arrows) generated by the laser radiation and two coplanar electrodes
on the top surface of the microchannel. [32] ......................................................................... 12
Fig. 2.7 Stacked images showing various microvortical flows in terms of the location of the laser
spots on two electrodes. The laser spot is focused at the middle (a), left side (b), right side (c)
of exposed electrode A, and (d) middle of exposed electrode B. [32] .................................. 13
Fig. 2.8 Schematic of experimental setup for an acoustic vortex generator. The transducer is
positioned adjacent to the channel; a bubble is trapped inside the horse-shoe structure. [34]
............................................................................................................................................... 14
Fig. 2.9 Characterization of the flow pattern around a single bubble. (a) An air bubble trapped in
the horse-shoe structure and stationary particles in the fluid. (b) Vortices around the air
bubble when the bubble membrane oscillates at its resonance frequency. [34] .................... 14
Fig. 2.10 A pair of vortices generated by the bubble oscillation at different driving frequencies
(a-d). Outline of the oscillatory bubble superposed over one cycle at different frequencies (e)
9.6 kHz, (f) 20.6 kHz, (g) 48.6 kHz, and (h) 100.3 kHz. [35] ............................................... 15
Fig. 2.11 The natural convection-driven flows in a microfluidic chamber. (A to D): trajectories of
fluorescent particles in time sequences. [40] ......................................................................... 15

IX
 List of Figures

Fig. 2.12 Fluorescence images of microparticles during agitation. The flow can be directed in a
clockwise (left-hand panels) or counter-clockwise (right-hand panels) fashion. [41] .......... 16
Fig. 2.13 Radius as a function of time for one cycle of the motion of a bubble of 2 µm. The
bubble dynamics is described by the Rayleigh–Plesset equation. [61] ................................. 24
Fig. 2.14 Upper graph: Radius-time curve of an oscillating bubble calculated by Miksis model.
Bubble radius at rest R0=1 µm, driving frequency f=20 kHz and the pressure amplitude Pa
=142 kPa (dash-dotted line), 143.5 kPa (dashed line), and 144 kPa (solid line). [63] .......... 25
Fig. 2.15 Left: Photographs of a bubble driven at 21.4 kHz and 132 kPa. Right: Comparison of
the numerically calculated curve (solid line) and experimental data (open circles). [67] ..... 26
Fig. 2.16 Free oscillation of a laser bubble in water: the framing rate is 75,000; maximum radius
is 2 mm. [68].......................................................................................................................... 26
Fig. 2.17 Two cases of stable radial and translational response of a bubble (R0 =2.0 mm) after 29
periods of forcing when it undergoes: (a) mild, super-resonant actuation (b) subresonant
excitation. [72] ....................................................................................................................... 30
Fig. 2.18 Jerky motion of two bubbles with strong actuations: vertical stripes from successive
frames are displayed next to each other, time proceeding from left to right. The acoustic
frequency is 20 kHz, and the inter-frame time is 0.4 periods. [77] ....................................... 31
Fig. 2.19 Photograph of a laser bubble collapsing in the vicinity of a glass wall located at the
bottom (time between frames: 5 µm). [78] ............................................................................ 32
Fig. 2.20 Selected frames showing the translations of several bubbles at (a) 0 ms, (b) 20 ms, (c)
40 ms, (d) 60 ms, (e) 100 ms, (f) 200 ms, (g) 300 ms, (h) 400 ms, and (i) 500 ms. The
pressure amplitude was 11.5 kPa. [79] .................................................................................. 33
Fig. 2.21 (a) The microfluidic flow-focusing device (top view). (b) Illustration of a bubble in the
outlet channel. [84] ................................................................................................................ 34
Fig. 2.22 (a) Schematic of apparatus for bubble generation via a flow-focusing section with
external actuations; (b) A series of selected frames of bubble formation captured by
highspeed photography at 7692 frames/s. [85] ...................................................................... 34
Fig. 2.23 (A) Schematic design of the sorter of microbubbles using the primary Bjerknes force.
(B) Photograph of the flow focusing section. (C) Microbubbles are formed in a narrow
orifice. (D) A travelling wave changes the trajectory of a series of bubbles. [87] ................ 35

X
 List of Figures

Fig. 3.1 Schematic illustration of the microfluidic chamber with resonator-shaped nozzle. (a)
Side view of the configuration; (b) Top view of the configuration with channel design and
geometric dimensions. ........................................................................................................... 38
Fig. 3.2 Schematic illustration of the microfluidic chamber without resonator-shaped nozzle. (a)
Side view of the configuration; (b) Top view of the configuration with channel design and
geometric dimensions. ........................................................................................................... 38
Fig. 3.3 The concept of fabrication and the assembled device. .................................................... 39
Fig. 3.4 Schematic illustration of the experimental setup for vortex visualization. ..................... 40
Fig. 3.5 The microfluidic chamber filled with working fluids and the cross line C’-O-C for
vortex characterization. ......................................................................................................... 40
Fig. 3.6 Recorded images in time sequence to show the vortex generated in the microfluidic
chamber with actuations. The actuation voltage and frequency were 100 V and 2.0 kHz, and
the flowrate was 2 mL/h at each inlet. ................................................................................... 42
Fig. 3.7 The dependence of the vortex intensity with time after being formed in the chamber. The
actuation voltage and frequency were 100 V and 2.0 kHz, and the flowrate was 2 mL/h at
each inlet. ............................................................................................................................... 43
Fig. 3.8 Recorded images in time sequence to show the development of the flow pattern under
the actuation of 1.5 kHz. The actuation voltage was 100 V and the flowrate was 2 mL/h at
each inlet. ............................................................................................................................... 44
Fig. 3.9 Recorded images in time sequence to show the development of the flow pattern under
the actuation of 2.5 kHz. The actuation voltage was 100 V and the flowrate was 2 mL/h at
each inlet. ............................................................................................................................... 45
Fig. 3.10 Development of the vortex intensity versus time after the actuation is turned on under
various actuation frequencies. The applied voltage and flowrate were fixed at 100 V and 2
mL/h, respectively. ................................................................................................................ 46
Fig. 3.11 Effect of actuation frequency on vortex formation and intensity after the flow patterns
are fully developed. The inset images were recorded at 50 seconds. The PZT disk was
actuated at the voltage of 100 V and the flowrate was 2 mL/h at each inlet. ........................ 46
Fig. 3.12 The images of the flow patterns under various applied voltages from 40 V to 120 V for
the DI water. The images were recorded at 50 seconds after actuation is switched on. The
driving frequency was 2.0 kHz and the flowrate was 2 mL/h at each inlet. .......................... 47

XI
 List of Figures

Fig. 3.13 Development of the vortex intensity versus time after the actuation is turned on under
various applied voltages. The actuation frequency and flowrate at each inlet were fixed at
2.0 kHz and 2 mL/h, respectively. ......................................................................................... 48
Fig. 3.14 Effect of the applied voltage on vortex formation and intensity after the flow patterns
are fully developed. The inset images were recorded at 50 seconds. The PZT disk was
actuated at the frequency of 2.0 kHz and the flow rate was 2 mL/h at each inlet. ................ 49
Fig. 3.15 Development of the vortex intensity versus time after the actuation is turned on under
various flowrates. The actuation frequency and voltage were fixed at 2.0 kHz and 100 V,
respectively. ........................................................................................................................... 50
Fig. 3.16 Effect of flowrates on vortex formation and intensity in DI water. The inset images
were recorded at 50 seconds after actuation is switched on. The PZT disk was actuated with
frequency 2.0 kHz and the applied voltage was 100 V. ........................................................ 50
Fig. 3.17 Visualization of the generated vortex in a circular chamber at the voltage of 140 V. The
actuation frequency was 2.0 kHz and the flowrate was 2 mL/h at each inlet. ...................... 51
Fig. 3.18 Development of the vortex intensity versus time after the actuation is turned on in a
circular chamber. The actuation frequency and voltage were fixed at 2.0 kHz and 140 V, and
the flowrate at each inlet is 2 mL/h. ...................................................................................... 52
Fig. 3.19 Visualization of the generated vortex in the counter-clockwise direction. The actuation
frequency was 2.0 kHz, the voltage 100 V and the flowrate 2 mL/h at each inlet. ............... 53
Fig. 3.20 The intensity of a counter-clockwise vortex versus time after the actuation is turned on.
The actuation frequency and voltage were fixed at 2.0 kHz and 100 V, and the flowrate 2
mL/h at each inlet. ................................................................................................................. 53
Fig. 4.1 Illustration of the modified microfluidic chamber and experimental setup. (a) Side view
of the configuration. (b) Top view and geometric dimensions of the configuration. (c)
Schematic illustration of the assembled device. (d) Schematic illustration of the
experimental setups. The velocity in the rectangular region – zone 1 is measured. ............. 57
Fig. 4.2 (a) Illustration of the PZT disk. The PZT disk consists of two separate transducers and a
brass sheet. (b) The vibration profile of the PZT disk when Transducer A is working. (c) The
vibration profile of the PZT disk when Transducer B is working. ........................................ 58
Fig. 4.3 Schematic illustration of the measurement of the velocity field in the outlet channel. The
velocity in the rectangular region – zone 1 is measured........................................................ 59

XII
 List of Figures

Fig. 4.4 Recorded images of the flow pattern development with time. Clockwise vortex (CW
vortex) is generated when Transducer A is actuated (a-e) and counter-clockwise vortex
(CCW vortex) is generated when Transducer B is actuated (g-k). The actuation voltage and
frequency are 100 V and 1.5 kHz respectively, and the flowrate is 10 mL/h in total. The
traces of particles of CW vortex (f) and CCW vortex (l) are in accordance with the flow
patterns of fluorescent dyes, respectively. ............................................................................. 61
Fig. 4.5 Effect of actuation frequency on the generated vortex when Transducer A (solid line) or
Transducer B (dashed line) of the PZT disk is actuated. The actuation voltage is 100 V and
the flowrate is 10 mL/h in total. All the flow patterns are recorded at 50 s. ......................... 62
Fig. 4.6 Effect of the actuation voltage on the generated vortex when Transducer A (solid line) or
Transducer B (dashed line) of the PZT disk is actuated. The actuation frequency is 1.5 kHz
and the flowrate is 10 mL/h in total. The inset images are recorded when flows are fully
developed at 50 s. .................................................................................................................. 63
Fig. 4.7 Effect of flowrate on the generated vortex when Transducer A (solid line) or Transducer
B (dashed line) of the PZT disk is actuated. The actuation frequency is 1.5 kHz and the
voltage is 100 V, respectively. The inset images are recorded when flows are fully
developed at 50 s. .................................................................................................................. 63
Fig. 4.8 Illustration of the flow through a sudden expansion or rectangular chamber. (a) The flow
transition from a pair of symmetric vortices to asymmetric at high Reynolds number. [96] (b)
Transition of the flow in a rectangular chamber from symmetric to asymmetric when the
Reynolds number is above the critical value. [98] (c) The asymmetric flow in a sudden
expansion with a large expansion ratio at high Reynolds number. [100] (d) The asymmetric
flow in a rectangular chamber with a large expansion ratio at high Reynolds number. [101]
ER is the expansion ratio. ...................................................................................................... 65
Fig. 4.9 The fluids flow into and out of the chamber through the outlet channel with the vibration
of the PZT disk. ..................................................................................................................... 67
Fig. 4.10 The fluids flow through a suddenly expanded part with a very large expansion ratio
(ER) in the outlet channel when the PZT disk is vibrating. .................................................. 67
Fig. 4.11 The velocity fields in the outlet channel at six moments of one cycle when the actuation
frequency is 1.5 kHz. ............................................................................................................. 68

XIII
 List of Figures

Fig. 4.12 The variation of the cross-section mean velocity in zone 1 of the outlet channel with
time in one cycle when Transducer A (solid line) or Transducer B (dashed line) is working.
............................................................................................................................................... 69
Fig. 4.13 The effect of the actuation frequency on the amplitude of the pulsatile velocity in the
outlet channel when Transducer A (solid line) or Transducer B (dashed line) is actuated. .. 70
Fig. 4.14 Schematic illustration of the experimental setup for measuring the vibration velocity
and displacement of the PZT disk. ........................................................................................ 71
Fig. 4.15 (a) Vibration velocity amplitude of the center O with the actuation frequency. (b)
Vibration displacement amplitude of the center O with the actuation frequency. Solid line –
Transducer A is working; Dashed line – Transducer B is working....................................... 72
Fig. 4.16 Side view of the microfluidic chamber for vortex generation. The fluids are actuated
into and out of the microfluidic chamber due to the volume change corresponding to the
PZT vibration. ........................................................................................................................ 73
Fig. 5.1 Schematic illustration of the microfluidic chamber. (a) Side view of the configuration. (b)
Top view of the chamber with geometric dimensions. The rectangular region zone 2 is
selected to record the bubble’s motion. ................................................................................. 76
Fig. 5.2 Schematic illustration of the experimental setup. ............................................................ 77
Fig. 5.3 Visualization of the bubbles generated in the chamber with the actuation at frequency
1.0 kHz and voltage 150 V. (a) Recorded image before actuation: the flow is laminar
throughout the microfluidic chamber. (b) Recorded image when the actuation is switched on:
the bubbles are generated. (c) Recorded image showing that the bubbles are traveling
upstream against the main stream. (d) Recorded image when the bubbles arrive at some
position close to the straight channel. The flowrate is 10 mL/h in total and the flow direction
is from right to left. ................................................................................................................ 79
Fig. 5.4 Photographic series of the dynamics of bubbles in one cycle after being generated. The
time interval between two successive images is 1/6 T, or 0.167 ms. The actuation voltage
and frequency are 150 V and 1 kHz, respectively. ................................................................ 80
Fig. 5.5 Images of moving bubbles captured at the same phase of different cycles. The time
between two successive images shown in the figure is one period, that is, 0.001 second. The
actuation voltage and frequency are 150 V and 1 kHz, respectively. ................................... 81

XIV
 List of Figures

Fig. 5.6 Coalescence of two oscillating bubbles. One small bubble is attracted by a large one and
then the two bubbles merge together to form a single one. The time between two successive
images is 1/12 T, or 0.083 ms. The actuation voltage and frequency are 150 V and 1 kHz,
respectively. ........................................................................................................................... 82
Fig. 5.7 Breakup of one oscillating bubble at the phase of the violent collapse. The time between
two successive images is 1/12 T, or 0.083 ms. The actuation voltage and frequency are 150
V and 1 kHz, respectively. ..................................................................................................... 84
Fig. 5.8 A series of coalescence and breakup of the oscillating bubbles in the process of
translation. The time between two successive images is 1/12 T, or 0.083 ms. The actuation
voltage and frequency are 150 V and 1 kHz, respectively. ................................................... 85
Fig. 5.9 The dynamics of an oscillating bubble nearby the upper boundary (1-13) and the
trajectory of the bubble by stacking three images of the bubble corresponding to the same
phase in a cycle. The time between two successive images is 1/12 T, or 0.083 ms. The
actuation voltage and frequency are 120 V and 1 kHz, respectively. ................................... 86
Fig. 5.10 The dynamics of an oscillating bubble nearby the bottom boundary (1-13) and the
trajectory of the bubble by stacking three images of the bubble corresponding to the same
phase in a cycle. The time between two successive images is 1/12 T, or 0.083 ms. The
actuation voltage and frequency are 120 V and 1 kHz, respectively. ................................... 87
Fig. 5.11 (a) Geometry and dimension of the numerical simulation domain. (b) Meshing scheme
in x-y plane (not to scale). (c) Meshing scheme in the height direction (not to scale). ......... 89
Fig. 5.12 The pressure distributions inside the chamber by numerical simulation. (a) Pressure
distribution without actuation. (b) Pressure distribution under an actuation of frequency f
=1.0 kHz at 0.75 T (T = 1/f = 0.001 s). The pressure at Point A is 469 Pa. .......................... 91
Fig. 5.13 Pressure fluctuation in one cycle at point P (0.013, 0, -0.00015) in zone 2. T = 1/f =
0.001 s. ................................................................................................................................... 93
Fig. 6.1 Schematic illustration of the experimental setup. The motion of a single bubble moving
through zone 2 is recorded by a high-speed camera and then analyzed in MATLAB. ......... 96
Fig. 6.2 Movement of a single bubble towards upstream along a straight line after being
generated in the chamber. The trajectory is obtained by stacking four images of the bubble
corresponding to the same phase in a cycle and the time between two successive images is
two periods, that is, 0.002 seconds. ....................................................................................... 97

XV
 List of Figures

Fig. 6.3 (a) Images of a moving single bubble captured at the same phase in different cycles. (b)
The displacement of the bubble versus time normalized by the actuation period T = 0.001 s.
The average translation velocity of the bubble is estimated to be 0.23 m/s, based on the slope
of the displacement curve. ..................................................................................................... 98
Fig. 6.4 Images of a bubble under the actuation of two successive cycles, showing remarkable
volume oscillation. The time between two successive images is 1/12 T with T = 0.001 s
being the actuation period...................................................................................................... 99
Fig. 6.5 (a) Variation of the bubble radius in two successive cycles. (b) Variation of the bubble
displacement in two successive cycles. The bubble moves forward at the phase of collapse.
T = 0.001 s, is the actuation period. .................................................................................... 100
Fig. 6.6 Illustration of the analytical model. (a) In the experiment, the bubble is in zone 2 of the
microfluidic chamber. (b) In the model, the bubble is located in a position on the bisector
line of two inclined straight walls........................................................................................ 102
Fig. 6.7 The model and image system for a bubble located nearby two inclined walls. ............ 103
Fig. 6.8 The whole process of the motions of the bubble nearby two tapered walls. The whole
bubble movement corresponds to three stages: uniform movement, accelerated movement
and final impact towards the walls. (a) Radius of the bubble with time. (b) Displacement of
the bubble with time. The bubble travels at an average velocity of 0.22 m/s in the stage of
uniform movement. ............................................................................................................. 110
Fig. 6.9 (a) Details of the radius in two successive periods for the bubble nearby two inclined
walls in an oscillatory pressure field. (b) Details of the displacement in two successive
periods for the bubble nearby two inclined walls in an oscillatory pressure field. ............. 111
Fig. 6.10 (a) Radius of a bubble in infinite water under the oscillatory pressure actuations. (b)
Displacement of a bubble in infinite water under the oscillatory pressure actuations. The
bubble remains stationary in the stationary fluid (dashed line) or travels with the main
stream at the stream velocity (solid line). ............................................................................ 113
Fig. 6.11 Illustration of the real bubble and ith image bubble. ................................................... 114

XVI
 List of Tables

List of Tables

Table 5.1 Boundary conditions used in the numerical simulation. ............................................... 89

Table 6.1 The boundary conditions at the bubble and wall surfaces. ......................................... 103
Table 6.2 The parameters used in the analytical model. ............................................................. 109

XVII
 Chapter 1 Introduction

Chapter 1 Introduction

1.1 Background and motivation

Microfluidics, which is involved in the science and technology of manipulating small

amounts of fluids in networks of micro channels, has been broadly applied in different fields,
such as chemistry, medicine and biology, owing to its numerous advantages of simple
manipulation, high sensitivity, low cost, little sample consumption and so on [1]. Vortical flows
are widely applied for flow control in many microfluidic devices, such as cell centrifuges, mixers,
sorters, concentration gradient generators and reactors etc. [2-4]. However, in microfluidics the
flow is usually featured with creeping flow due to the predominance of viscosity over inertia.
Thus, the Reynolds number (Re) for flows is very small, leading to parallel and highly-ordered
streamlines. Besides, the difficulty of vortex generation is greatly increased by the small
geometric dimensions and extremely low flowrates. In fact, it is always challenging to generate
vortices in microfluidic systems for applications because swirling of streamlines is suppressed by
the constraints of small inertia and simple geometries.
To produce vortices, several well-designed configurations have been used in the microfluidic
systems, including herringbone structures, zigzag channels, serpentine channels and flow
focusing systems [2-4]. However, in these methods, the designs of the configurations are very
complex and high flowrates are required. An alternative approach involves introducing external
actuations, such as magnetic force, electrokinetic force, acoustic actuation, thermal effect, even
external air flux etc. However, special equipment and elaborate manipulation are indispensable,
which causes inconvenience in real applications. Another downside is that these methods are
primarily for a specific fluid or particle with given properties. If the requirements for the flowrate
or materials are changed, a new design might be needed. Therefore, despite of many kinds of
vortex generators, the novel and simple methods of vortex generation and control in microfluidic
systems remain to be an attractive research topic. This is one of the objectives of the present
study.
Bubble dynamics in bulk flow has been intensively studied because of its applications
associated with many technologies, such as ultrasonic cleaning, sonochemistry, erosion of bubble

1
 Chapter 1 Introduction

collapse and ultrasonography [5]. Theoretical modelling of the bubble’s radial oscillation
probably was first conducted by Rayleigh, and Plesset extended Rayleigh’s work by including
liquid viscosity and surface tension, leading to the well-known Rayleigh-Plesset equation [6].
Experimental studies revealed that bubbles under external actuations not only undergo volume
oscillations but also translate in space. The strong coupling between radial oscillation and
translation makes it difficult to calculate the radius and displacement of a bubble under
actuations. Crum and Eller obtained the translation velocity of an oscillating bubble by
introducing the averaged acoustic force and drag, which was justified to be valid only in a weak
pressure field [7]. Doinikov derived coupled equations for radial and translational dynamics of a
bubble in infinite liquid and obtained the time-dependent radius and displacement of a bubble
under high-intensity actuations [8]. In addition to oscillation and translation, an oscillating
bubble in a pressure field can also experience a third type of motion – shape deformation when
its volume oscillation becomes unstable. Liu et al investigated the shape stability of an
encapsulated bubble in an ultrasound field and the nonlinear process of higher-order shape mode
was obtained by solving the continuity equation and Navier-Stokes equation directly [9].
Recently, bubble dynamics has been widely applied in lots of microfluidic processes, such as
micropumping, micromixing, sorting, flow control and manipulation of micron-sized objects etc.
However, the challenges of energy concentration, viscous surface forces and low number of
cavitation nuclei make it difficult to generate bubbles directly in microfluidic channels via the
conventional acoustic method [10]. Alternatively, gas bubbles can be formed in microfluidic
channels by introducing external gas into a flow-focusing system or through laser irradiation.
However, the bubble generation by external gas and laser radiation are often inconvenient in real
applications: formation of bubbles in flow-focusing systems requires introduction of external gas
and elaborated manipulation; laser devices are indispensable for the method of laser radiation.
New methods for bubble generation in microfluidic systems are attractive research topics in this
field. Besides, most studies on bubbles dynamics under actuations in microfluidics are focused
on high frequencies, ranging from 20 kHz to several MHz, while there are few investigations on
bubble dynamics at low frequencies. At the low-frequency actuations, the wavelength of
actuation pressure will be long compared to the bubble size and microfluidic channel, which are
normally in 10-2 m – 10-4 m. As a result, the actuated pressure is almost uniformly distributed in
the microfluidic channels and the pressure gradient is very small around the bubbles, leading to a

2
 Chapter 1 Introduction

negligible primary Bjerknes force compared to high frequencies. Therefore, other mechanisms,
such as the wall effects, tend to be the dominant force to propel the bubble forward under the
low-frequency actuation in a microfluidic chamber.

1.2 Objectives

This study aims to investigate vortex generation, vortex control, and bubble dynamics in
microfluidics with the following objectives.
1) To develop a novel method of vortex generation and control in a resonator-shaped
microfluidic chamber with actuations.
2) To study bubble dynamics in a microfluidic chamber with actuations, including the
bubble generation, radial oscillation and spatial translation.
The scope of this research includes:
1) Design and fabricate a microfluidic chamber for vortex generation and regulate the
direction and intensity of the generated vortices on-demand by varying the actuation conditions,
including the working transducers, frequency, voltage and flowrate.
2) Investigate the mechanism of the vortex generation and control in the microfluidic
chamber with actuations.
3) Conduct experiments to study the dynamics of multi-bubbles and a single bubble in a
microfluidic chamber with actuations, including the bubble generation, radial oscillation and
spatial translation etc.
4) Conduct both numerical simulations and analytical modeling to interpret the experimental
findings of bubble dynamics in a microfluidic chamber under the low frequency actuations.

1.3 Outline of the Dissertation

The following are the outlines of the present study.

Chapter 1 gives an introduction to the research, including background, motivation and
objectives of the present study.

3
 Chapter 1 Introduction

In Chapter 2, literature review is conducted on microfluidic vortex generators, cavitation and

bubble dynamics in bulk liquid and microfluidics.
Chapter 3 presents a novel microfluidic chamber with piezoelectric actuations for vortex
generation. Fabrication of the microfluidic chamber and experimental visualization are described
in detail. The effects of the actuation conditions including the frequency, voltage and flowrate
are investigated.
Chapter 4 studies modification of the microfluidic chamber by separating the transducer into
two parts. The direction and intensity of the generated vortex in the chamber are regulated by
varying the actuation conditions.
Chapter 5 investigates the dynamics of multiple bubbles in the microfluidic chamber with
low-frequency actuations. The phenomena of bubble coalescence and breakup are observed.
Numerical simulation is carried out to investigate the mechanisms of the bubble dynamics based
on the pressure variation and distribution inside the chamber.
In Chapter 6, the dynamics of a single bubble is studied experimentally and analytically.
Radial and translational motions of the bubble in the chamber are quantitatively analyzed by
measuring the recorded images. An analytical model is constructed by using the Lagrangian
mechanics and the method of images and the analytical results are compared with the
experimental observations.
Finally, Chapter 7 summarizes the main conclusions and contributions of the present research
and gives some recommendations for future research.

4
 Chapter 2 Literature Review

Chapter 2 Literature Review

2.1 Introduction

Vortex structure is a prevalent feature in most macro hydraulic flows. Many great figures
such as Helmholtz, Kelvin, Prandtl and von Karman, all made substantial contributions to vortex
flows in hydrodynamics [11-13]. Up to date, Karman Vortex Street and Taylor vortex are
favorable tools to investigate the instability of fluid flows and transition from laminar to
turbulence. In microfluidics, due to the low flowrates and small dimensions, viscous force is
dominant relative to inertia, which makes it a great challenge to generate vortices in micro-scaled
channels. To produce vortex in microfluidics, special design or external disturbance must be
introduced to disrupt the highly-ordered streamlines [2-4]. The methods of generating vortices in
microfluidics can be categorized into passive and active according to whether external forces are
used. Passive methods rely on well-designed geometries and special flow conditions, while
active methods require external forces. According to the external forces, the main classes of
active vortex generators are electrokinetic generators, magnetic generators, and acoustic
generators and so on. Active methods have fewer limits on flow conditions and higher efficiency
in contrast to passive methods.
Cavitation, which is defined as rupture of liquid, is widely applied in many areas such as
engineering and medicine. The radial response of a cavitation bubble can be described by the
well-known Rayleigh-Plesset equation and its extensions, which consider various factors, for
instance, damping effects, heat and mass transfer across the interface between liquid and gas, and
the liquid compressibility etc. Under weak external actuation, the bubble undergoes linear
oscillation; while under strong actuations, the bubble oscillates nonlinearly with violent
contraction and even collapse. The bubbles under external actuations not only undergo volume
oscillations but also move in space – translate, and the translational dynamics of an oscillating
bubble is highly coupled with radial pulsation, which is a challenge for researchers to study the
bubble’s motions. The challenges of energy concentration, viscous surface forces and low
number of cavitation nuclei make it difficult to generate bubbles directly in microfluidic
channels via the conventional method. Pulsed laser radiation can provide enough energy to

5
 Chapter 2 Literature Review

evaporate the liquid to generate bubbles, but the generated bubbles are usually transient, and
more frequently external gas needs to be introduced into a flow-focusing system to create
bubbles.
This chapter presents a literature review on microfluidic vortex generators, cavitation
phenomenon and bubble dynamics in bulk liquid, and bubble dynamics in microfluidics,
including the bubble’s radial pulsation and translation. The research gap of current investigation
is also summarized.

2.2 Microfluidic vortex generator

2.2.1 Vortex in microfluidics

Vortices form prominent features in most natural flows, such as atmospheric vortices in
tornadoes, whirlwinds of dust or leaves, smoke rings and ocean currents. Fluid vortices are also
central factors of consideration in many industrial flows: centrifugal separators used in chemical
engineering processes, tip vortices generated by wings, wakes of ships and so on.
Although the notion of a vortex is so widely used in fluid dynamics, few researchers can give
an exact and inclusive definition of vortex. Several authors have proposed rigorous definitions of
vortex. Brachet et al. [14] defines a vortex as a region of negative velocity gradient determinant.
Babiano et al. [15] defines a vortex as any domain of a fluid with vorticity greater than a critical
value. However, no single definition has gained broad acceptance. Herein, we do not explore
deeply the exact mathematical definition of a vortex, but accept the simple and visible
description: any type of swirling flow with closed or spiraling streamlines [16].
Vortical flows are widely applied for flow control in many microfluidic devices, such as cell
centrifuges, mixers, sorters, concentration gradient generators and reactors etc. However, due to
small channel dimensions and low flowrates, the flows in microfluidics tend to be laminar and
dominated by viscosity of the fluids. Thus, it is always challenging to generate controllable
vortices in microfluidic systems for applications. To produce vortices, several well-designed
configurations have been used in the microfluidic systems including herringbone structures,
zigzag channels, serpentine channels and flow focusing systems, and this is so-called passive
method. An alternative approach, active method, can effectively generate vortical flows with

6
 Chapter 2 Literature Review

lower flowrates. It involves the introduction of external actuations, such as electrokinetic force,
magnetic force, acoustic actuation, thermal effect, and even external air flux etc.
Next, we will discuss the two typical microfluidic vortex generators in detail.

2.2.2 Passive vortex generators

The passive vortex generators rely on the advection flow caused by complex configurations
and geometries of the microchannels, such as herringbone structures, zigzag turns, serpentine
channels and flow focusing systems and so on. In this section, the current study of passive vortex
generators is briefly reviewed.
Stroock et al. [17] presented the earliest passive vortex generator to mix fluids under certain
flow conditions. As illustrated in Fig. 2.1, two types of vortex generators are designed by
introducing different ridge arrays at the bottom of the channel. One type is featured with sloped
ridges as shown in Fig. 2.1(a) and the other with staggered herringbone ridges as shown in Fig.
2.1(b). Both microstructures can generate vortices under high enough Reynolds number (Re). For
example, the staggered herringbone structure (Fig. 2.1(b)) resulted in the vortex flow at Re of
about 100.

Fig. 2.1 Illustration of vortex generators with ridge arrays and generated vortices in microchannels. (a)
sloped ridges; (b) herringbone ridges. [17]

7
 Chapter 2 Literature Review

Mengeaud et al. [18] designed a zigzag microchannel integrated with “Y” junctions (Fig.
2.2(a)), which can generate vortices at the corners. They found that below a critical Re of about
80, the flow profile remains parabolic indicating existence of a laminar flow; for higher Re, the
occurrence of vortices at the corners is possible. It is observed that the vortices in Fig. 2.2(b)
occupied the entire arms of the channel after each angle at Re=267.

Fig. 2.2 (a) Dimensions of the vortex generator integrating “Y” junctions with zigzag microchannels. (b)
Velocity vectors obtained at Re=267 for zigzag geometries. [18]

8
 Chapter 2 Literature Review

Fig. 2.3 (a) Schematic illustration of inertial vortex generator. (b) Illustration of the separation principle. The
symmetric vortices are indicated by the red dashed lines. (c) Stacked images at various downstream positions
illustrating the separation phenomenon at Re=110. [19]

Wang et al. [19] proposed an inertial vortex generator with simple geometries for cell
extraction. The configuration of the device, the principle of separation and experimental images
are shown in Fig. 2.3. The design takes advantage of symmetric vortices generated in a high-
aspect-ratio microchannel to separate cells.
Oliveira et al. [20] reported the formation of free vortices through a microfluidic flow-
focusing system. The device is shaped like a cross-slot and comprises three entrances and one
exit. Figure 2.4 shows the flow patterns for varying Re at the channel’s center plane using streak
photography. Vortices are obtained both in numerical calculations and experimental observations
when Re is equal to 94.2 (Fig. 2.4(c)) or 113 (Fig. 2.4(d)).

9
 Chapter 2 Literature Review

Fig. 2.4 Flow patterns in the 3D flow-focusing geometry for varying Re. Comparison between experimental
(grey-scale images) and numerical pathlines at the centre plane. (a) Re=2.8; (b) Re=6.5; (c) Re= 94.2; (d) Re=
113. [20]

Liu et al. [21] reported a three-dimensional serpentine channel which consists of periodic
units with a “C-shaped” structure to produce vortices. In this type of channel, strong vortices can
only be generated at relatively high Reynolds numbers (Re = 70) when the flow patterns become
complicated by combined effects of rotational and backward flows. Vortex generation in a
passive way is also widely investigated by other researchers such as Singh et al. [22], Kang et al.
[23], Simonnet et al. [24] and Camesasca et al. [25] and so on.
Although passive method is easily-manipulated and low cost, it needs a complex geometry
design and very high Reynolds number. If changes are induced to either the velocity or materials,
a new optimized design might be needed.

2.2.3 Active vortex generators

Vortex can be generated by an external energy input to introduce the perturbation of fluid
streamlines in a microfluidic device, which is the so-called active method. Active vortex
generators are categorized with respect to the type of external forcing. In this section, we will
discuss active vortex generators in microfluidics based on the type of external actuations.

10
 Chapter 2 Literature Review

2.2.3.1 Electrokinetic method

Electrokinetic forcing serves as an effective means not only for liquid pumping but also for
flow control. For example, electroosmosis is widely used due to its significant advantages such
as plug-like velocity profile, ease to control and no mechanical moving parts. Squires and Bazant
[26] reported a type of electrokinetic flow called induced-charge electroosmosis (ICEO), which
can generate flow circulations near a highly polarizable conducting object in an external electric
field. Wu and Li [27] obtained an irregular flow field with micro vortices by means of ICEO
flow in experiments and simulations. Figure 2.5 shows the comparison of the flow fields for the
case with and without ICEO effect. Clearly, with the presence of ICEO effect, a pair of vortices
is generated before and after the throat of the channel, respectively.

Fig. 2.5 Flow patterns and concentration fields (a) with induced-charge electroosmosis effect and (b) without
induced-charge electroosmosis effect. [27]

Developments in fabrication of microelectrodes have intrigued new methods of generating

vortices in microfluidics. Wu and Liu [28] reported a vortex generator with the electrodes
embedded in the herringbone structure of a T-shaped channel’s bottom wall. Sasaki et al. [29]
demonstrated a vortex generator due to a meandering electrode pair on the bottom of the “Y”
channel, around which electroosmotic flow is created. More complex patterns of microelectrodes
on the channel walls were also proposed by Huang et al. [30].

11
 Chapter 2 Literature Review

Vortices can also be generated in an opto-electrofluidic platform, which combines a focused

laser beam with alternating current (AC) electric fields. In the presence of electric fields, the use
of intensive laser irradiation can produce strong vortices in a microchannel formed by two plates,
due to a combination of electrothermally-induced flows and electrohydrodynamically-induced
flows. Kumar et al. [31] reported the first opto-electric vortex generator in microfluidics when a
near-infrared laser beam is focused in a microchannel with an AC electric field. Park and
Wereley [32] demonstrated an opto-electric vortex generator with two coplanar electrodes, which
can generate twin opposing microvortices (TOMVs). The configuration is different from the
design of Kumar, as shown in Fig. 2.6: two electrodes A and B are located on the same surface
of the microchannel. When the laser spot is focused on any region of the exposed electrode, a
vortical flow is generated rapidly (Fig. 2.7). The motion is a complex, three-dimensional vortex
flow but remains steady in time. The pattern and direction of twin vortices are controllable by
positioning of the laser radiation on the electrodes. This technique is a promising method to
produce desirable fluid flows and manipulate chemical and biological fluids in microfluidics.

Fig. 2.6 Schematic illustration of the TOMV (twin opposing microvortices) flow with two microvortices
(circular arrows) generated by the laser radiation and two coplanar electrodes on the top surface of the
microchannel. [32]

12
 Chapter 2 Literature Review

Fig. 2.7 Stacked images showing various microvortical flows in terms of the location of the laser spots on two
electrodes. The laser spot is focused at the middle (a), left side (b), right side (c) of exposed electrode A, and
(d) middle of exposed electrode B. [32]

2.2.3.2 Acoustic Method

Vortices can be achieved by means of acoustic actuation, which is introduced into the
channel by integrated piezoelectric ceramic transducers (PZT). Air bubbles can be introduced in
a microfluidic channel and the surface of the air bubbles in liquid exposed to a sound field can
function as a vibrating membrane, which causes a bulk fluid movement at the air–liquid interface.
This effect, known as cavitation microstreaming, has been applied in microfluidic vortex
generation by a single bubble or an array of bubbles.
The first research on acoustic streaming induced by air bubbles is conducted by Liu et al.
[33]. A number of air pockets were used for trapping air bubbles and a PZT disk is attached onto
the wall of the channel. When an integrated PZT actuator is switched on, acoustic streaming was
induced by the vibration of bubbles, resulting in steady vortices around the bubbles. Ahmed et al.
[34] designed a single-bubble-based vortex generator by trapping an air bubble within a “horse-
shoe” structure located between two laminar flows (Fig. 2.8). Acoustic waves induce the
vibration of the trapped air bubble at its resonance frequency, resulting in acoustic streaming.
Figure 2.9 shows the visualization of flow patterns near an acoustically excited microbubble
trapped in a horse-shoe structure. The particle was nearly motionless prior to the acoustic
agitation; as the air bubble was excited, a pair of counter-rotating vortices extended from the

13
 Chapter 2 Literature Review

frontside to backside of the horse-shoe structure, which is made visible through the trajectories
of particles.

Fig. 2.8 Schematic of experimental setup for an acoustic vortex generator. The transducer is positioned
adjacent to the channel; a bubble is trapped inside the horse-shoe structure. [34]

Fig. 2.9 Characterization of the flow pattern around a single bubble. (a) An air bubble trapped in the horse-
shoe structure and stationary particles in the fluid. (b) Vortices around the air bubble when the bubble
membrane oscillates at its resonance frequency. [34]

Hilgenfeldt et al. [35] investigated the frequency dependence of the vortices of an oscillating
bubble in microfluidics, driven by the ultrasounds in the frequency range between 1 kHz and 100
kHz. The oscillation modes and flow patterns are shown in Fig. 2.10. They further developed a
complete asymptotic theory to calculate the streaklines of the fluid flow from cylindrical bubbles
and the theoretical results are found to be agreeable with experimental observations [36].

14
 Chapter 2 Literature Review

Fig. 2.10 A pair of vortices generated by the bubble oscillation at different driving frequencies (a-d). Outline
of the oscillatory bubble superposed over one cycle at different frequencies (e) 9.6 kHz, (f) 20.6 kHz, (g) 48.6
kHz, and (h) 100.3 kHz. [35]

2.2.3.3 Other types of active methods

Vortex can be achieved by stirring of magnetic particles suspended in the fluids [37]. When
the scattered paramagnetic particles are exposed in a magnetic field, they tend to align in a chain,
which is effective in fluid stirring for vortex generation. Calhoun et al. [38] and Franke et al. [39]
verified the feasibility of this method of vortex generation using a rotating magnet and magnetic
chains by simulations and experiments, respectively.
Thermal disturbance can also act as the driving force to produce vortices in the microfluidic
channels. Kim et al. [40] created multi-vortices by using the natural convection in conjunction
with alternating heating and the resultant vortical flow is successfully used for micromixing and
polymerase chain reaction (PCR) tests (Fig. 2.11).

Fig. 2.11 The natural convection-driven flows in a microfluidic chamber. (A to D): trajectories of fluorescent
particles in time sequences. [40]

15
 Chapter 2 Literature Review

Fig. 2.12 Fluorescence images of microparticles during agitation. The flow can be directed in a clockwise (left-
hand panels) or counter-clockwise (right-hand panels) fashion. [41]

Geissler et al. [41] use air flux to induce spiral motions in a small-scale fluid domain by
means of a convenient and efficient manner. The air flux is regulated and projected onto a liquid-
containing reservoir, which leads to spiral motions of the fluid in the reservoir. A suspension of
microbeads follows the spiral motion of the liquid and thus visualizes the flow pattern in the
reservoir. As shown in Fig. 2.12, the generated vortex can be either clockwise (left-hand side) or
counter-clockwise (right-hand side), depending on the orientation of the incoming air in the
supply channel.
Active vortex generators have fewer limits on flow conditions and higher efficiency than
passive vortex generators. However, some active methods are primarily for specific fluids or
particles with given properties, and the special equipment and elaborate manipulation are also
indispensable. For example, the electrokinetic or magnetic actuation has requirements on the
conductivity or magnetism of the working fluids. The thermal method may damage the reagents
and samples which are sensitive to the temperature variation; the general acoustic methods need
to introduce and trap external gas in advance and there is also a problem of temperature rise at
high frequency.

16
 Chapter 2 Literature Review

2.3 Cavitation inception and thresholds

Cavitation is a very common phenomenon that occurs in a fluid flow system. It refers to the
process of nucleation from liquid to gas phase when pressure falls below some critical value,
mostly vapor pressure, at almost constant temperature. From a basic physical viewpoint,
cavitation has no difference from boiling. Rupture of a liquid leading to gas phase is the main
feature in both cavitation and boiling. Cavitation can be further subdivided into hydrodynamic
cavitation – rupture in a low or even negative pressure zone of a liquid flow and acoustic
cavitation – rupture in sound fields. Apfel [42] initially denominated acoustic cavitation to be
formation of a vapor cavity or bubble in response to an acoustic pressure field, but in most
circumstances, the bubble which is activated by a sound field is observed to not simply form
from bulk liquid but grow from some pre-existing nuclei in liquid. Therefore, the cavitation
phenomena might be thought of as the process of creation of new bubbles or
expansion/contraction/distortion of pre-existing ones in a liquid, which is coupled to the pressure
fluctuation.

2.3.1 Cavitation nuclei as bubble sources

The existence of nuclei in liquid such as water can be expected. Those nuclei are of size
between a few microns and some hundreds of microns and operate as starting points for
cavitation. In addition to cavitation, the consideration of nuclei inside a homogeneous liquid is
common in thermodynamics, for example, boiling, condensation and solidification. It is thought
that there are three types of nuclei [43]: some nuclei are freely suspended in liquid; some are
trapped in crevices of the walls; many nuclei are also attached to solid motes in liquid. New
populations of nuclei can also be generated by larger bubbles that undergo breakup or splitting.
An excess pressure would be imposed to the bubble surface due to the existence of surface
tension, which is called surface tension pressure. Under this pressure, bubbles tend to dissolve
because the concentration gradient occurs between the bubble wall and infinity. However, the
experiments show that in real liquids free nuclei exist after long time [44-45]. Besides, when an
external actuation such as a sound field is acting, the oscillation of bubbles can counteract the
dissolution from surface tension and bubbles can grow remarkably over long periods of time.

17
 Chapter 2 Literature Review

Tensile strength is the largest tension which a liquid can suffer without cavitation, which is
equal to the hydrodynamic cavitation threshold of liquids. Considering the cohesion between two
liquid molecules, the theoretical tensile strength of pure liquids is remarkably large. Caupin
predicts [46] that for pure water at 300 K, the tensile strength is about 128 MPa, beyond which
the nucleation of the vapor phase can take place. However, in real liquid, lots of nuclei exist,
which are generally of the magnitude of micron-size. Typically, nuclei or microbubbles, which
are freely suspended or attached to solid boundaries or motes, act as starting points of cavitation
and limit the tensile strength of liquids. Therefore, the real liquid can suffer much smaller tension
and even will be nucleated at positive pressure.
The tensile strength or cavitation threshold is often not the property of the pure liquid, but
highly depends on the degree of cleanness of the liquid, such as the number of impurities, motes
and pre-existing nuclei. The tap water without any further processing begins to be nucleated at
even positive pressure - saturated vapor pressure, and no tension occurs in liquid. Herbert et al.
[47] reported that, after ultrapure water is treated through a series of rigorous cleaning
procedures, the tension strength is found to be 26 MPa at 0 centigrade and 17 MPa at 80
centigrade. Because the tensile strength of laboratory water is greatly less than theoretical
prediction, it can be expected that the cavitation can occur easier than “ideal” liquids.

2.3.2 Cavitation threshold

2.3.2.1 Blake threshold

When the ambient pressure is varied to P∞, one can expect that the bubble radius is evolved
to R from R0. In new equilibrium, the pressure inside a bubble consists of gas and vapor, Pi, is
Pi  Pg  Pv  P  P (2.1)

where Pg, Pv and Pσ are the gas pressure, vapor pressure and surface tension pressure,
respectively. Assuming that the variation is slow enough for isothermal process of the gas, the
evolution of R can be obtained
  R 
3
2S
 P,0   Pv  0   Pv,0  P  P (2.2)
 R0  R 

where S is the surface tension and the subscript 0 denotes the initial state.

18
 Chapter 2 Literature Review

If the external pressure is lowered to a critical value, there is no possible equilibrium radius R.
Beyond this pressure, the bubble will expand to a multiple of its initial size and collapse
violently afterwards. The critical pressure, which depends on R0, is called “Blake threshold” [48].
It can be calculated by seeking the minimal R from the above equation [49]
1/ 2
 4 X3 
Pcrit  P0  P0  

(2.3)
 27 1  X 

where P0 is the ambient pressure and X=2S/P0R0. In an oscillatory pressure field, this expression
also determines the minimum actuation pressure, beyond which nuclei will cavitate, grow
explosively and collapse violently. Therefore, this is the threshold for inertial cavitation, beyond
which stable cavitation cannot occur.

2.3.2.2 Rectified diffusion threshold

Consider a gas bubble undergoing radial oscillation in a liquid with dissolved gas. During the
expansion phase of a cycle, the gas tends to enter the bubble while the bubble loses gas during
the contraction phase. Hence, the concentration gradient of the dissolved gas will oscillate in
liquid, resulting in a pulsating diffusion flux. On average, the net flux within one cycle is non-
zero. Although during one cycle, the net flux is very small, diffusion may cause bubbles to grow
over long periods of time. This phenomenon of gas accumulation in the bubble over many cycles
is referred to “rectified diffusion” or RD [45]. Young [50] proposed the detailed explanations:
Firstly, gas diffusion from liquid to the bubble occurs when the area of exchange is larger during
the expansion phase of a cycle; secondly, the diffusion boundary layer is thinner during the
expansion phase and thus incur a higher inward flux.
An oscillating bubble grows because of rectified diffusion while surface tension promotes the
dissolution. A competition takes place between two effects and there exists a threshold for a
nucleus to grow or dissolve. For weak oscillation, nuclei will dissolve or remain inactive with the
action of surface tension; while under strong actuation, nuclei can aggregate the gas in liquid and
become active bubbles over long time. The expression for the RD threshold is proposed by Hsieh
and Plesset [51]

Pa 3  C  2S 
 1  1   (2.4)
P0 2  C0  R0 P0 

19
 Chapter 2 Literature Review

where Pa is the amplitude of pressure fluctuation, C∞ is the gas concentration in the liquid and C0
is the saturation gas concentration in the liquid. It applies only to bubbles well below resonance
size, and above the threshold, nuclei can grow to become active.

2.3.2.3 Transient threshold

Transient cavitation can be qualitatively distinguished from stable cavitation by whether the
bubble is fragmented or dissolved during one period. However, to define the threshold exactly,
the transition from stable to transient cavitation can be stated quantitatively in terms of the
maximum radial velocity of the bubble surface. When this velocity reaches the sound speed in
liquid, that is, Mach number approaches 1, we specify that transient cavitation occurs. In
incompressible liquid, a gas bubble which reaches a collapse velocity equal to sound speed will
grow to a maximum radius, Rmax, about 2 times its original size. According to this criterion,
Neppiras [52] introduced the expression of the transient threshold
1/ 2
4 2S 3 
Pa  P0   3  (2.5)
3  3R0 Pa  2S / R0  

where Pa is the amplitude of pressure variation, S is the surface tension, P0 is the ambient
pressure and R0 is the initial radius of the bubble.
The RD threshold is usually regarded as the beginning point for a stable cavitation because
below it, the microbubble will be inactive and dissolved afterwards. If a nucleus grows over long
time above RD threshold, it may undergo inertial cavitation when reaching Blake threshold. If
the pressure amplitude become larger or the bubble grows larger, the bubble can suffer transient
cavitation, undergoing fragmentation after violent collapse.

2.3 Radial dynamics of a bubble in liquid

Conventionally, the oscillation of a bubble can be qualitatively classified according to the

magnitude of the bubble’s vibration in response to the imposed actuations. In general, three
regimes are identified [53].

20
 Chapter 2 Literature Review

a) Linear oscillation exists under very small pressure amplitudes. The radial oscillation of a
bubble is non-linear in nature. However, actuated under low amplitude pressure, the bubble’s
response is weak oscillation around equilibrium radius and can be regarded as linear.
b) With increasing amplitude of oscillating pressure, the response of a bubble will become
featured with weak nonlinearity and slight inertia effect. However, the bubble may still oscillate
stably, which can be referred as “stable cavitation”.
c) Under a strong enough pressure field, the bubble size may vary by several orders of
magnitude during one cycle. The bubble will experience explosive growth followed by violent
fragmentation. Such a response is termed to be “transient cavitation”.
Current experiments show that stable bubbles can also exist under such kind of cavitation –
“inertial cavitation”, one of which is the sonoluminescence bubble [54]. Inertial cavitation,
which was termed as transient cavitation in the past, is separated independently since the
sonoluminescence bubble experiences violent collapse but without fragmentation or dissolution.
The cavitation phenomena with low frequencies (in the audible domain and below) approach
hydrodynamic cavitation in a way. Nevertheless, the classification of cavitation at low
frequencies complies with the same law with the cavitation at high frequencies.

2.3.1 Rayleigh-Plesset equation and extensions

The analysis of cavitation and bubble dynamics was firstly made by Rayleigh, who solved
the collapse of a spherical cavity in infinite liquid with the assumption of isothermal change. In
1949, Plesset improved this equation by including the effects of surface tension and viscosity,
and obtained the famous Rayleigh-Plesset equation [6]

  3  2 Pi  P 4 L R 2S
RR R    (2.6)
2  R R

where, over-dots represent the time derivative, ρ is the liquid density, P∞ is the ambient pressure
in liquid, Pi is the pressure inside the bubble, S and νL are the surface tension constant and kinetic
viscosity of the liquid, respectively.
For a mixture bubble composed of permanent gas and vapor of surrounding liquid, the
interior pressure is calculated by Eq. (2.1). Assuming that the gas in the bubble is polytropic so
that when the bubble has radius R, the partial pressure of the gas will be:

21
 Chapter 2 Literature Review

Pg  Pg ,0 R0 R3  P0  2S R0  Pv R0 R 3 (2.7)

where, Pg,0=P0+2S/R0-Pv, the gas pressure at equilibrium and κ is polytropic index, clearly, κ=1
for isothermal process and κ=γ (the ratio of the gas specific heat) for adiabatic behavior.
In a pressure field, the pressure P∞ is transient, for example, sinusoidal driving, and it can be
written as
P  P (t )  P0  Pa sin(t ) (2.8)
where P0 is the static ambient pressure and Pa is the amplitude of driving pressure and ω is the
angular frequency. The driving frequency f= ω/2π=1/T, where T is the driving period.
Therefore, substituting Eqs. (2.7) and (2.8) into Eq. (2.6), an extension equation, which
describes the radial dynamics of an oscillating mixture bubble in liquid, follows
3  2 Pv  P0 4 L R 2S Pa sin(t ) 1 2S R
 
RR R      ( P0   Pv )( 0 )3 (2.9)
2  R R   R0 R

where, all the symbols are defined above.

This extension of Rayleigh-Plesset equation is termed as RPNNP (Rayleigh Plesset Noltingk
Neppiras Porisky) equation and was first derived by Poritsky [55]. When we discuss bubble
dynamics in pressure fields in the next part, this equation is referred to Rayleigh-Plesset equation
or Rayleigh-Plesset model without ambiguity.
The compressibility of the liquid can be neglected when the characteristic length of the
bubble (the radius R) is much smaller than the wavelength of the applied pressure field, λ=c/f,
where c is the sound speed in the liquid. However, for high frequencies, compressibility of the
liquid needs to be considered and the Rayleigh-Plesset equation for a permanent gas bubble
becomes the so-called Gilmore model [56] which is well suited to the situation that
compressibility of liquid plays an important role under high pressure. Different from the Gilmore
model which assumed a pressure dependent sound speed, Keller and Miksis assumed that the
velocity potential around a bubble in acoustic fields satisfies the spherical wave equation with
constant sound speed [57]. Both Gilmore model and Miksis model are extensions of Rayleigh-
Plesset equation with similar form. The connection between the different models of bubble
dynamics is investigated by Prosperetti and Lezzi [58].

22
 Chapter 2 Literature Review

2.3.2 Linear oscillation of a bubble in oscillatory pressure fields

The bubble’s oscillation is nonlinear in essence. However, when the actuation pressure field
is very weak with small amplitude, the bubble can respond with small oscillations around its
equilibrium radius. In this sense, the bubble/liquid system can be regarded as a linear oscillator,
similar to a mass-spring system with a periodic force.
The Rayleigh-Plesset equation (2.9) is used to analyze the linear response of a bubble in a
pressure field with a small amplitude Pa and frequency ω. When thermal and compressible
effects are neglected, we can get a linear relation between the complex amplitude of the bubble
radius ϕ and the driving pressure Pa [59]:
Pa


R02 n2   2  4 jL / R02  (2.10)

where R0 is the equilibrium radius at the pressure P0 and R0|ϕ| is the amplitude of the bubble’s
oscillation and the phase of ϕ is the phase difference between P∞ and R, υL is the kinetic viscosity.
From Eq. (2.10), we are ready to draw the conclusions that [59]
For ω<ωn, the bubble radius is out of phase with the driving forces, which means that the
bubble will expand during the depression of the pressure.
For ω>ωn, the bubble will be in phase with the driving forces, which sounds impossible
because it means that the bubble will contract as the exerted forces increase. This case occurs
due to the predominance of the liquid inertia at high frequency.

2.3.3 Nonlinear oscillation of a bubble in oscillatory pressure fields

The preceding section assumes that the perturbation of the bubble’s radius is small enough to
keep the linear response. However, this approximation is questionable in real cases because the
bubble can exhibit a lot of nonlinear phenomena as reviewed by [60]. As the amplitude increases,
the bubble may experience the process from stable cavitation to transient cavitation.
The governing equation of an oscillating bubble at infinity is a nonlinear ordinary differential
equation, including Rayleigh-Plesset model, Gilmore model or Miksis model. It’s difficult to
solve the equation directly; however, with the progress of computer science, it’s possible to
obtain the solution numerically.

23
 Chapter 2 Literature Review

An example of bubble dynamics calculated by Rayleigh-Plesset equation is presented in Fig.

2.13 by Vignoli et al [61]. A bubble with initial radius of 2 µm is driven by a sound field of 1.42
atm pressure amplitude and 26.5 kHz. The surface tension and viscosity are 72.8 mN/m and
1.002 mPa·s, respectively, while the compressibility is not considered. By solving Rayleigh-
Plesset equation, Hilgenfeldt et al. [62] investigated the role of liquid viscosity on bubble
oscillation. They found that the maximum radius in a cycle decreases with increasing viscosity;
viscosity, as damping, prevents the bubble from violent collapse and afterbounces and above
some threshold, the bubble cannot obtain collapse strong enough for ignition of
sonoluminescence.

Fig. 2.13 Radius as a function of time for one cycle of the motion of a bubble of 2 µm. The bubble dynamics is
described by the Rayleigh–Plesset equation. [61]

When a small bubble is encountering external actuations with increasing pressure, inertial
cavitation will emerge instead of stable cavitation. Lauterborn and Mettin [63] investigated this
transition process of regimes of bubble oscillation by calculating the radius of a 1 µm bubble
driven at 20 kHz and increasing pressure by Keller-Miksis model. As shown in Fig. 2.14, the
transition from stable to inertial cavitation with increasing pressure is clearly displayed: the
maximum radius during one period increases suddenly with a subtle ascent of driving pressure.
The upper curve (solid line) shows the typical feature of inertial cavitation which includes
explosive growth of the bubble and a steep collapse with a series of afterbounces. During this
process, the bubble motion is mainly governed by liquid inertia, which is the reason for the term
“inertial oscillation”. Inertial cavitation is in the region of sonoluminescence experiment and has
the characteristics of a large excursion followed by an extremely steep collapse with a series of

24
 Chapter 2 Literature Review

afterbounces. The radial dynamics of bubble clusters are also studied by researchers by different
models, such as volume-averaged equations [64] and ensemble-averaged equations [65].

Fig. 2.14 Upper graph: Radius-time curve of an oscillating bubble calculated by Miksis model. Bubble radius
at rest R0=1 µm, driving frequency f=20 kHz and the pressure amplitude Pa =142 kPa (dash-dotted line),
143.5 kPa (dashed line), and 144 kPa (solid line). [63]

In addition to numerical simulation, experiments have been extensively conducted about

bubble dynamics to verify the theoretical models and other predictions.
Gompf and Pecha [66] used light scattering method to record the dynamics of a single air
bubble in an oscillatory pressure field in water and made comparison of the radius-time curves
between measurement and calculation for different driving pressures. The calculation model is
based on RP equation. The agreement between experimental result and theoretical model is
remarkable except for the afterbounces of the bubble. Lauterborn et al. [67] made more agreeable
comparison between experimental result and numerical calculation. They recorded one complete
cycle of an oscillating bubble of 8.1 µm radius driven by the external actuation of 21.4 kHz for
frequency and 132 kPa for amplitude. The dependence of the bubble radius on time from
experiments is derived from the images. It is observed in Fig. 2.15 that the bubble shows typical
features of inertial cavitation, expanding slowly and shrinking with a fast collapse and several
afterbounces. Within the error of measurement, the experiments coincide quite well with
numerical calculation.

25
 Chapter 2 Literature Review

Fig. 2.15 Left: Photographs of a bubble driven at 21.4 kHz and 132 kPa. Right: Comparison of the
numerically calculated curve (solid line) and experimental data (open circles). [67]

The bubble can also be generated by a focused laser radiation into a liquid with pulse
duration [68]. This kind of bubble only lasts for a few of its own oscillation cycles, which allows
for the study of transient bubbles in a controlled and flexible way. Figure 2.16 shows a
photographic sequence of the free oscillation of a single laser bubble. It is observed that the
spherical bubble reaches a maximum radius of about 2 mm, then the nonlinear oscillation is
strongly damped after rebound from the first collapse and finally, the bubble is seen to be almost
disappeared after several afterbounces.

Fig. 2.16 Free oscillation of a laser bubble in water: the framing rate is 75,000; maximum radius is 2 mm. [68]

2.4 Translational dynamics of a bubble in oscillatory pressure fields

The bulk of early work on bubble dynamics was aimed at studying the radial motion of a
bubble or bubble population in an actuated pressure field. However, the bubbles under external

26
 Chapter 2 Literature Review

actuations not only undergo volume oscillation but also move in space – translational motion.
The translational dynamics of an oscillating bubble is coupled with the radial motion and cannot
be worked out prior to solving the radial oscillating equation.
There are two ways of obtaining theoretical solution of the translational motion of an
oscillating bubble. One method, first proposed by Crum and Eller [7], is averaging and
decoupling the equations of radial and translational motions to calculate the long term behavior
of the bubble. This approach is based on the averaged Bjerknes force – a kind of radiation force
from the external actuations, buoyancy, gravity and viscous force etc. and justified in very weak
actuation where both radial and translational movements are relatively small. Another way of
studying translation of the bubbles is to solve the coupled equations of radial and translational
motions of bubbles directly. The governing equations are coupled nonlinearly between the
volume pulsation and translation, and therefore numerical calculation is demanded. This method
is applicable to strong external actuations and become popular with the development of
numerical analysis. Prior to discussing the translation of an oscillating bubble, several concepts
are introduced such as Bjerknes force.

2.4.1 Bjerknes force and drag force

There are two physical phenomena which can be categorized as Bjerknes forces, primary
Bjerknes force, which is exerted by the external pressure fields and secondary Bjerknes force,
which is the mutual interaction between multiple oscillating bubbles.
Assuming that a bubble in an ideal liquid is spherical and small, and the spatial variation of
the pressure gradient in the bubble is negligible. When the centroid of the bubble is located at
some point at time t, the Bjerknes force subject to volume pulsation can be calculated in the
following [69]
F    pdV  V (t )p(r , t ) (2.11)
V

where V denotes the volume of the bubble and p is the liquid pressure at the bubble wall. When
the pressure gradient is caused by the external actuation pressures (e.g. travelling wave or
standing wave), this force is the so-called primary Bjerknes force and when the pressure gradient
is caused by other bubbles, this force is the so-called secondary Bjerknes force.

27
 Chapter 2 Literature Review

The average net force on the bubble during one cycle is

4
FB    V (t )p(r , t )    R3 (t )p(r , t )  (2.12)
3
where < > denotes the time average and R(t) is the radius of the bubble.
When considering a real fluid, we must add the viscous drag. The motion of the bubble in
liquid can be discussed according to the Reynolds number. The Reynolds number is determined
as the ratio of inertial forces to viscous forces and quantifies the relative importance of these two
types of forces. In bubble dynamics, if the radial velocity of the bubble wall is introduced as a
second characteristic velocity, there are two types of Reynolds number, ReU  R | U |  ,

ReR  R | R |  [70].

For ReU>>1 or ReR>>1, that is the bubble moves fast or is strongly forced, the viscous drag is
similar to the case of a spherical droplet at high Re [70]
FD  12RU (2.13)
When ReU<<1 and ReR<<1, that is the bubble is slowly translating with slow radial dynamics,
the quasisteady Stokes drag force for an “ideal” bubble (zero shear stress at the interface) follows
FDQS  4RU (2.14)

2.4.2 Decoupled model of an oscillating bubble with weak actuations

A small bubble in an oscillatory pressure field (e.g. standing wave field) oscillates in volume
and translates unsteadily and the two motions are coupled. However, it is of great convenience to
decouple volume oscillation and translation as an approximation. The translational dynamics of a
bubble in a decoupling way was first proposed by Crum and Eller [7]. The two equations are
decoupled by averaging all the forces acting on the bubble during one cycle of vibration and the
decoupling requires a weak external actuation. Consider an oscillating spherical bubble, the
equation of momentum follows
FpB  FA  FD  0 (2.15)

where FpB is the generalized Bjerknes force, FA is the force resulting from added mass and FD is
the drag, taking the form without history force. Substituting Eqs. (2.12) and (2.13) into Eq. (2.15),
it is transformed to [71],

28
 Chapter 2 Literature Review

2 d 3
4R3p 3  ( R U )  12RU  0 (2.16)
3 dt
The pressure gradient and radial oscillation are assumed in the following forms
P( x, t )  Pa ( x) sin(t ) (2.17a)

R(t )  R0 [1   sin(t )] (2.17b)

The equation of motion is rewritten in the time-averaged form
2RT3  4Pa
UT   R 3 sin(t )  12RT U T (2.18)
3 3
where, < > denotes the time average of the quantity during one cycle.
Assuming all the transients expire, the average velocity follows that [71]
U   3k2 2k1 (2.19)
where k1  18 R02 , k2  2Pa  .
By means of this method, Crum and Eller [7] calculated the decoupled solution of averaged
velocity of the bubble and found excellent agreement with experimental measurement.

2.4.3 Coupled model of an oscillating bubble with strong actuations

With the increasing capacity of calculation of computers, it is possible to solve the coupled
equations of radial and translational motions of an oscillating bubble directly. Reddy and Szeri
[72] numerically solved the coupled radial and translational dynamics of a single bubble in a
travelling wave field. The radial and translational equations they used are described as

  3 R 2  1  P  P  R d P  P 
R R (2.20a)
 
2   L  cL dt L  

2 d 3
 
 R U a  X 
3 dt

4 3
3
R P  12R U a  X   (2.20b)

where, X is the displacement of the bubble, Ua is the absolute velocity of liquid. The liquid

pressure at the bubble wall PL is calculated by PL  Pg  R g  4 L R  2S and the liquid
dP
3c g dt R R

pressure at infinity P∞, is calculated by P  P0 1  Pa sin2ft  .

Two cases are calculated for a bubble of 2 um in radius: one in a mild and super-resonant
wave field and the other one in subresonant wave field (Fig. 2.17). The solution for coupled

29
 Chapter 2 Literature Review

collapse and translation dynamics is agreeable with past experiments and previous theories for
decoupled translation dynamics.

Fig. 2.17 Two cases of stable radial and translational response of a bubble (R0 =2.0 mm) after 29 periods of
forcing when it undergoes: (a) mild, super-resonant actuation (b) subresonant excitation. [72]

Doinikov [8, 73] derived coupled equations of radial and translational motions of an
oscillating bubble in an intense standing wave field using Lagrangian formalism. It is found that
the radial equation needs a correction term considering the effect of translation. The new coupled
equations read as [8]
  3  2 PL x
RR R   (2.21a)
2  4

3R x 3Fex
x   (2.21b)
R 2R 3

where Fex is the sum of primary Bjerknes force and viscous drag and PL is the liquid pressure at
2S 4 L R
the bubble wall and, PL  Pg    P .
R R

By solving the new equations, Doinikov proved that a bubble driven below resonance can
reciprocate about the pressure node instead of moving towards pressure antinodes if pressure
amplitude is high enough.

30
 Chapter 2 Literature Review

There is very little experimental investigation about translation of an oscillating bubble with
external actuations since Crum’s experiments [7]. Mettin [74] displayed the “jerky” motion of
two bubbles during its collapse in Fig. 2.18. The centroid positions of two bubbles that move
upwards are marked by white lines. It can be clearly observed that the bubble position frequently
jumps upwards during collapse, and stops during expansion. The relatively net velocity of the
bubbles is 1 to 1.5 m/s.

Fig. 2.18 Jerky motion of two bubbles with strong actuations: vertical stripes from successive frames are
displayed next to each other, time proceeding from left to right. The acoustic frequency is 20 kHz, and the
inter-frame time is 0.4 periods. [74]

2.5 Bubble boundary interaction

Due to the practical importance of bubble-wall interaction in ultrasonic cleaning,

sonochemistry and erosion of bubble collapse as well as in medical applications, some progress
has been achieved about the dynamics of a bubble nearby plane boundaries and confined
configurations [75].
Far from the boundary wall, a bubble in an oscillatory pressure field can be regarded as
spherical and the potential problem can be solved by the method of images that replaces the rigid
boundary by an image bubble equal to the real one and located symmetrically to it with respect
to the boundary.
Cole simulated the global motion of a bubble in the neighborhood of a plane solid wall or a
free interface by introducing a series of images [76]. Using Cole’s strategy, Sato et al. [77]
studied the behavior of a gas bubble near a rigid boundary in an oscillatory pressure field and
they found that the period and amplitude of the bubble oscillation are related not only to the

31
 Chapter 2 Literature Review

actuation pressure but also to the bubble location from a rigid wall. Near a solid surface, the
asymmetry of the flow that the boundary itself introduces gives rise to highly non-spherical
bubble shapes and even leads to the formation of a high-velocity jet directed toward the wall at
its collapsing phase, while the overall characteristics of the sinklike flow of the image attracts the
bubble toward the wall. Kroninger et al. [78] investigated the bubble collapsing aspherically near
a rigid wall and experimentally observed the development of a jet and a vortex ring as shown in
Fig. 2.19. Besides, their simulation results with the boundary integral method are in good
accordance with the experimental observation.

Fig. 2.19 Photograph of a laser bubble collapsing in the vicinity of a glass wall located at the bottom (time
between frames: 5 µm). [78]

Xi et al. [79] investigated the dynamics of collective bubbles near a solid surface, especially
the bubble translation, in a weak acoustic standing wave field as shown in Fig. 2.20. It was found
that the bubble translation near a surface in a multi-bubble environment was mainly controlled
by the primary Bjerknes force imposed by the acoustic field, and secondary Bjerknes force
introduced by a surface and neighboring bubble. The primary Bjerknes force dominated the
bubble translation when the bubble was far away from the surface and was outweighed by the
secondary Bjerknes force from the boundary when the bubble was approaching the surface. Xi et
al. also found that the presence of a wall can change a bubble’s oscillation mode and in particular,
the wall shifts the instability pressure thresholds to smaller driving frequencies for fixed bubble
equilibrium radii, or to smaller equilibrium radii for fixed excitation frequency [80].

32
 Chapter 2 Literature Review

Fig. 2.20 Selected frames showing the translations of several bubbles at (a) 0 ms, (b) 20 ms, (c) 40 ms, (d) 60
ms, (e) 100 ms, (f) 200 ms, (g) 300 ms, (h) 400 ms, and (i) 500 ms. The pressure amplitude was 11.5 kPa. [79]

Investigations have also been conducted with the dynamics of bubbles in the proximity of an
elastic boundary or composite material, a free surface, a convex or concave wall, in gaps or tubes
and so on [81-83].

2.6 Bubble dynamics in microfluidics

While bubble dynamics has been observed and studied extensively in bulk fluid flows, the
bubble dynamics, especially acoustic cavitation, has not been widely investigated in
microfluidics so far. The challenges of energy concentration, viscous surface forces and low
number of cavitation nuclei make it difficult to generate bubbles directly in microfluidic
channels via the conventional method [10]. In current studies of bubble dynamics, the generation
of gas bubbles can be formed by the external introduction of gas phase through a flow-focusing
system
By importing external gas as the secondary phase, Garstecki et al. [84] firstly reported a
method for generating monodisperse bubbles in a microfluidic flow-focusing device as
illustrated in Fig. 2.21. The bubbles self-assemble into highly-ordered flowing lattices depending
on flow parameters. Shirota et al. [85] produced submillimeter bubbles from an orifice by means
of a complex system with a loudspeaker (Fig. 2.22) similar to flow-focusing channels. The
33
 Chapter 2 Literature Review

detachment of a bubble is controlled by pulsed actuation pressure and the intensity of actuation
pressure is controllable to generate much smaller bubbles compared to ordinary flow-focusing
systems.

Fig. 2.21 (a) The microfluidic flow-focusing device (top view). (b) Illustration of a bubble in the outlet
channel. [84]

Fig. 2.22 (a) Schematic of apparatus for bubble generation via a flow-focusing section with external
actuations; (b) A series of selected frames of bubble formation captured by highspeed photography at 7692
frames/s. [85]

At high frequency actuations, the wavelength of acoustic fields is short and comparable to
the microfluidic channel and bubble size. As such, acoustic field and pressure gradients will be
established around the bubbles, which will induce significant primary Bjerknes force on the
bubbles. Therefore, the current studies of bubble dynamics in microfluidics are concentrated on
high frequencies, ranging from 20 kHz to several MHz, to obtain large primary Bjerknes force as
the driving force of bubble motion.

34
 Chapter 2 Literature Review

Rabaud et al [86] demonstrated the use of primary Bjerknes force or the acoustic radiation
force to induce the motion of bubbles confined between walls of microchannels. They developed
two applications in microfluidics based on the interplay between the primary Bjerknes force and
drag force: asymmetric bubble breakup to produce very well controlled bidisperse populations
and intelligent switching at a bifurcation. Segers and Versluis [87] presented a simple lab-on-a-
chip method to sort the population of coated microbubbles using the primary Bjerknes force in a
travelling ultrasound wave field as shown in Fig. 2.23. They demonstrated successful sorting of a
commercial ultrasound contrast agent (UCA) and improved the sensitivity of the contrast sound.

Fig. 2.23 (A) Schematic design of the sorter of microbubbles using the primary Bjerknes force. (B)
Photograph of the flow focusing section. (C) Microbubbles are formed in a narrow orifice. (D) A travelling
wave changes the trajectory of a series of bubbles. [87]

Another method of bubble generation is radiation of a pulsed laser into the liquid. Zwaan et
al. [88] reported generation of bubbles in confined microfluidic system by focusing a pulsed
laser into the light-absorbing liquid. The pancake-shaped bubbles expand and collapse radially,
agreeable with a two-dimensional Rayleigh-Plesset model. They also investigated the cavitation
bubble dynamics in microfluidic gaps with variable height and found that the bubble lifetime
increases with decreasing gap height whereas the maximum bubble radius remains constant [89].
Patrascioiu et al. [90] studied the correlation between the laser-induced bubble and liquid
ejection by analyzing the time-resolved behavior of the bubble and liquid jetting and found that
the pulsating behavior of the bubble leads to successive jetting events with different jet
morphologies arising from the particular geometries that the bubble acquires during its evolution.
Their results are in good agreement with the numerical simulation of cavitation bubbles near a
free-surface of a liquid.

35
 Chapter 2 Literature Review

The shortcomings of bubble generation by flow-focusing system or by laser radiation are

prominent: formation of bubbles in flow-focusing systems requires introduction of external gas
and elaborate manipulation; laser-induced bubbles are transient and laser devices are
indispensable for the method of laser radiation. Therefore, the existing methods of bubble
generation by external gas or laser radiation are inconvenient in real applications and new
methods for bubble generation in microfluidic systems are attractive research topics in this field.
In addition, at low-frequency actuations, the acoustic wavelength will be long comparing to
the bubble size and microfluidic channel. For example, the pressure wavelength is 1.5 meter for
the case of 1 kHz actuation in water (taking the sound speed in water as 1500m/s). This will be
much larger than the dimension of the bubble size and microfluidic channels, which are normally
in 10-2 m – 10-4 m. As a result, the actuated pressure is almost uniformly distributed in the
microfluidic channels and the pressure gradient is very small around the bubbles, leading to a
weak primary Bjerknes force exerting on the bubbles. Therefore, other mechanisms, such as the
wall effect, tend to be the dominant factor to propel the bubble forward under the low-frequency
actuation in a microfluidic chamber. In a microfluidic system, the bubbles are enclosed by
multiple walls, and in this case the wall effect will be much more complicated than the case of a
single wall. Therefore, the dynamics of a bubble in the neighborhood of multiple walls needs to
be further investigated.

36
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Chapter 3 Vortex Generation in a Microfluidic Chamber:

Characterization

In this chapter, a novel resonator-shaped microfluidic chamber is designed for vortex

generation. A PZT (Lead Zirconate Titanate) disk is assembled to the microfluidic chamber from
one side to actuate the fluids. The details of the experiment method, including the configuration
and fabrication, experimental setup, and measurements of the performance are clearly presented.
The vortices generated in the chamber are visualized by tracing fluorescent dye and the vortex
intensity is quantitatively analyzed by processing the recorded images in MATLAB. The
influence of the actuation conditions, such as the actuation frequency, voltage and flowrate, on
the performance of vortex generation in the microfluidic chamber is investigated. The
performance of the resonator-shaped chamber is compared with another type of circular chamber
to study the effect of the resonator-shaped nozzle part on the vortex generation.

3.1 Experimental methods

3.1.1 Design and fabrication of the microfluidic chamber for vortex generation

The microfluidic chamber for vortex generation is made by two PMMA

(polymethylmethacrylate) plates sandwiched by a dry adhesive layer with 300 µm in thickness
(Arclad 8102 transfer adhesive, Adhesives Research, Inc.). The configurations and dimensions of
the microfluidic chamber are shown in Fig. 3.1. It mainly consists of two parts, a 16 mm
diameter semi-circular chamber and a nozzle with an acoustic resonator profile. The overall
geometric profile of the chamber can be expressed by

 82  x 2 , 8  x  4;
y (3.1)
0.5e0.1011(30 x) , 4  x  30.


Two inlets are connected at a 60-degree angle to the chamber by a straight channel of 10 mm
in length and 1 mm in width, and another straight channel is connected to the other end of the
chamber to form the outlet. A PZT (Lead Zirconate Titanate) disk is attached to the bottom of the

37
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

chamber to provide actuations. The PZT disk is formed by a piezoelectric ceramic layer of 15
mm diameter glued onto a brass sheet of 22 mm diameter. The brass sheet was painted to avoid
direct contact with the working fluids in the chamber. The geometry of the entire configuration is
precisely manufactured by a laser cutting machine (Universal M-300 Laser Platform, Universal
Laser Systems Inc., Arizona, USA). For comparison, another type of microfluidic chamber
without the resonator-shaped nozzle is designed and the configurations and dimensions are
illustrated in Fig. 3.2.

Fig. 3.1 Schematic illustration of the microfluidic chamber with resonator-shaped nozzle. (a) Side view of the
configuration; (b) Top view of the configuration with channel design and geometric dimensions.

Fig. 3.2 Schematic illustration of the microfluidic chamber without resonator-shaped nozzle. (a) Side view of
the configuration; (b) Top view of the configuration with channel design and geometric dimensions.

The microfluidic chamber is fabricated by lamination technology. Two PMMA plates are
assembled together by a dry adhesive layer in the middle. The height of the channel is controlled

38
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

by the thickness of the adhesive layer. The PZT disk is fitted into the access hole of the top
PMMA plate. The inlets and outlet are assembled to the plate by epoxy glue (Araldite, Huntsman
Advanced Materials, USA). The concept of fabrication and the assembled device are shown in
Fig. 3.3.

Fig. 3.3 The concept of fabrication and the assembled device.

3.1.2 Experimental setup and characterization

The schematic of the experimental setup is illustrated in Fig. 3.4. In the experiment, a
degassed DI water solution was supplied to the microfluidic chamber from inlet A, and the same
solution with fluorescent dye was supplied to inlet B. The flowrate was adjusted by two syringe
pumps (KD Scientific Inc., USA). By using the fluorescent dye as flow tracers, the vortices can
be observed inside the microfluidic chamber after the PZT is switched on under varied voltages
and frequencies. The piezoelectric disk was driven by sinusoidal signals generated by a signal
generator (33120A, Hewlett Packard) and then amplified 50 times by a voltage amplifier (790,
PCB Piezotronics). A high speed CCD camera (Phantom V711, Vision Research, USA) is
mounted on top of the chamber to record the flow field.

39
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.4 Schematic illustration of the experimental setup for vortex visualization.

Fig. 3.5 The microfluidic chamber filled with working fluids and the cross line C’-O-C for vortex
characterization.

When a vortex forms, the working fluids are stirred and swirled, leading to the mixing of
the fluorescent dye in the chamber. Therefore, the vortex intensity is proportional to the
uniformity of the fluorescent dye concentration in the chamber, which is an indicator
associated with the vortical flow and vortex strength. To quantitatively characterize the
intensity of the vortex, a MATLAB code was used to process the captured images assuming that
the intensity of grayscale is related to the fluorescent dye concentration in the recorded images.

40
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

A cross line C’-O-C (Fig. 3.5) through the center of the chamber was chosen to be analyzed. The
vortex intensity is calculated using the following expression:
2
N
 I i  I ave 

1
Vortex Intensity(%)  1    (3.2)
N  1 i 1  I ave 

where, N represents the total number of points examined along the cross-line C-O-C’, I̅ i is the
normalized intensity at each point, I̅ ave is the mean of normalized intensity at all points, which is
 N 

1 
calculated by I ave  Ii  and the square root denotes the standard deviation of the
N  
 i 1 

fluorescent dye distribution. Ideally, this expression will give 0% before vortex is generated and
100% if the fluids are perfectly mixed resulting from the vortex.
The procedures of the experiment are stated in detail as follows:
(1) Fix the device onto a metal holder and then connect the function generator, amplifier and
the PZT disk of the microfluidic chamber. Mount a high-speed camera above the PMMA plate
enclosing the chamber and install the LED (light emitting diode) as the illumination system.
(2) Fill one syringe with the degassed DI water and the other one with the DI water of
fluorescent dye. Connect the syringes with the inlets of the chamber through plastic tubes.
(3) Install two syringes to a pump and switch on the pump to supply the working fluids into
the microfluidic chamber through two inlets.
(4) When the flow in the chamber is stable, turn on the PZT disk and meanwhile trigger the
camera to record the flow patterns inside the chamber until the flow is fully developed.
(5) Adjust the actuation conditions including the frequency, voltage and flowrate to
investigate the influences on the performance of vortex generation. Make sure that the channel is
free from bubbles in the experiment.

3.2 Results and discussion

3.2.1 Flow pattern development in the microfluidic chamber

By tracing the patterns of the fluorescent dyes in the flow, the vortices generated in the
chamber can be visualized and characterized. Figure 3.6 presents the evolvement of the

41
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

generated vortex with time sequences after the PZT disk is actuated. The experimental conditions
are as follows: a flowrate of 2 mL/h at each inlet, an actuation voltage of 100 V and a frequency
of 2.0 kHz. Without actuations (Fig. 3.6(a)), the interface between the upper fluid (without
fluorescent dye from inlet A) and lower fluid (with fluorescent dye from inlet B) is very clear,
indicating that the flow is laminar, without vortices. However, as shown in Fig. 3.6(b), 5 seconds
after the actuation is turned on, the fluorescent-dyed fluid has been swirled up to the upper half
of the chamber and a vortex can be clearly observed. The fluorescent fluid is continuously
swirled (see Fig. 3.6(c-e)) and at 25 seconds, the fluids with fluorescent dye are found to occupy
the whole chamber. Finally, at 45 seconds, the interface of the fluorescent dye in the fluids
disappears with the action of the vortex and the flow pattern is fully developed.

Fig. 3.6 Recorded images in time sequence to show the vortex generated in the microfluidic chamber with
actuations. The actuation voltage and frequency were 100 V and 2.0 kHz, and the flowrate was 2 mL/h at
each inlet.

42
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.7 The dependence of the vortex intensity with time after being formed in the chamber. The actuation
voltage and frequency were 100 V and 2.0 kHz, and the flowrate was 2 mL/h at each inlet.

In terms of Eq. (3.2), the vortex generation is further characterized by the vortex intensity
with time sequence after generation as shown in Fig. 3.7. It shows that the vortex intensity
inceases with time under the action of the vortex and reaches up to 90% at 45 seconds. We
take 30% as an criterion to determine whether the vortex is formed or not inside the chamber
although such criterion is a bit arbitary [91]. Since the vortex is observed by tracing the
fluorescent dye and the time scale is characterized by the fluorescent diffusion in such low
Reynolds number flow, it therefore takes a long time to observe the vortex formation. The
actual vortex developement corresponding to the actuation should be much faster and
characterized by the actuation frequency. Scalling analysis suggests that the characteristic
time for a vortex development is of order  vor  O1 f  . Using f = 2.0 kHz in frequency, we can

obtain the time scale of the vortex development is of the order  vor  O103  second. On the other
hand, the characteristic time of fluorescent dye can be estimated from  dye  OL2c D . Taking the

characteristic dimension Lc= 300 μm in channel height and diffusion coefficient D=10-10 m2/s in
dye diffusivity [92], we can find the fluorescent dye relaxation time of the order  dye  O10

second, which is in good agreement with our experimental observation as shown in Fig. 3.6.

43
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

3.2.2 Effects of actuation frequency on the generated vortices

The vortices generated in the microfluidic chamber are further investigated under various
frequencies. It is found that the pattern of the vortical flow varies at different frequencies.
Figure 3.8 and 3.9 show the flow pattern development with time under the actuation of 1.5 kHz
and 2.5 kHz for actuation frequency, respectively. The applied voltage was fixed at 100 V and
the flowrates were kept 2 mL/h at each inlet. The working fluids are DI water. It is observed that
at 1.5 kHz, the vortex is irregular and the flow is rotating around an axis located at the upper-
right half of the chamber, compared to the vortex at 2.0 kHz. As shown in Fig. 3.9, at 2.5 kHz, in
addition to a large vortex in the middle of the chamber, a smaller one is also observed to occur
on the upper-left side of the chamber, which is different from a single vortex at 2.0 kHz.

Fig. 3.8 Recorded images in time sequence to show the development of the flow pattern under the actuation of
1.5 kHz. The actuation voltage was 100 V and the flowrate was 2 mL/h at each inlet.

44
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.9 Recorded images in time sequence to show the development of the flow pattern under the actuation of
2.5 kHz. The actuation voltage was 100 V and the flowrate was 2 mL/h at each inlet.

The vortex intensity along the line C’-O-C under varous actuation frequencies is
calculated by Eq. (3.2) at different times. The vortex intensity versus time for different
actuation frequencies is shown in Fig. 3.10. The results show that, the time it takes to achieve
the vortex intensity above 30% differs for different actuation frequencies: at 1.5 kHz, 2.0
kHz or 2.5 kHz, the vortex generation is quite easy, indicated by the short times (<10 s) for
the vortx intensity to reach above 30% and at either 3.5 kHz or 1.5 kHz, the vortices are
found to become too weak to achieve the vortex intensity above 30%. The vortex intensity at
50 seconds when the flows at various frequencies are fully developed is depicted in Fig. 3.11.
It is clear that there exists an effective frequency window, within which the strong vortex can
be generated while out of this frequency window, the vortex is very weak or even unable to
occur in the chamber. In the range from 1.5 kHz to 3.0 kHz, strong vortex can be generated
easily with the vortex intensity as high as 70%. At either higher frequencies, e.g. 3.5 kHz or
lower frequency, e.g. 1.0 kHz, the vortex is found to become weaker, even below 10%. Such
actuation frequency window is found to be dependent on the actuation volatge and flowrate.

45
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.10 Development of the vortex intensity versus time after the actuation is turned on under various
actuation frequencies. The applied voltage and flowrate were fixed at 100 V and 2 mL/h, respectively.

Fig. 3.11 Effect of actuation frequency on vortex formation and intensity after the flow patterns are fully
developed. The inset images were recorded at 50 seconds. The PZT disk was actuated at the voltage of 100 V
and the flowrate was 2 mL/h at each inlet.

46
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

3.2.3 Effects of the applied voltage on the generated vortices

The influence of the applied voltages was investigated by varying the actuation voltage from
40 V to 120 V and the actuation frequency was fixed at 2.0 kHz and the flowrates at 2 mL/h at
each inlet. The working fluids are DI water. The flow patterns in Fig. 3.12 were recorded at 50
seconds when the flows are fully developed for all voltages. It is seen that, at 40 V, there were
almost no fluorescent dyes in the upper half of the chamber. From 60 V to 80 V (Fig. 3.12(b-c)),
more and more fluorescent dyes emerge on the upper half of the chamber with increasing
voltages. The higher voltage is applied, the more fluorescent dye is swirled to the upper-half
part of the chamber, indicating a stronger vortex is formed by the actuation. However, at
very high voltages, gas bubbles are observed in the chamber in addition to the strong vortex
flow. As shown in Fig. 3.12(d) for the actuation at 120 V, bubbles are found at some position
close to the upstream straight channel (indicated by a dashed rectangle). The bubble
generation was reported previously, in which flow mixing enhancement was attributed to the
bubble oscillations without mentioning the vortices [93]. The present study clearly reveals
that the mixing enhancement can occur without bubbles at relatively low voltages, only
because of the vortices generated in the chamber, and at very high voltages, the combined
effects of the vortex and bubbles contribute to the mixing enhancement.

Fig. 3.12 The images of the flow patterns under various applied voltages from 40 V to 120 V for the DI water.
The images were recorded at 50 seconds after actuation is switched on. The driving frequency was 2.0 kHz
and the flowrate was 2 mL/h at each inlet.

47
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

The intensity of the generated vortex under various applied voltages is calculated in terms of
Eq. (3.2). Figure 3.13 shows the vortex intensity versus time for different applied voltages,
with the frequency and flowrate being fixed at 2.0 kHz and 2 mL/h at each inlet. It is seen
that, at the low actuation voltage 40 V, the vortex cannot be properly formed and the vortex
intensity remains below 30%. As the voltage is increased to 60 V, the vortex intensity is
above 30% after 12 seconds. The higher voltage is applied, the shorter time is taken to bring
the vortex intensity above 30%, indicating a stronger vortex formed by the actuation. We
demonstrate the vortex intensity under various applied voltages at 50 s when the flows are fully
developed for all cases in Fig. 3.14. Clearly the vortex intensity is increased almost proportional
to the applied voltages. The vortex intensity is very poor (approx. 10%) at 40 V and increases to
90% at 100 V before the bubbles occurs and when the voltage of 120 V is applied, the vortex
intensity reaches 96.4%. The results show that the mixing enhancement can be mainly attributed
to the strong vortices generated inside the chamber rather than the bubbles, because the bubbles
are generated only at higher actuation voltages.

Fig. 3.13 Development of the vortex intensity versus time after the actuation is turned on under various
applied voltages. The actuation frequency and flowrate at each inlet were fixed at 2.0 kHz and 2 mL/h,
respectively.

48
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.14 Effect of the applied voltage on vortex formation and intensity after the flow patterns are fully
developed. The inset images were recorded at 50 seconds. The PZT disk was actuated at the frequency of 2.0
kHz and the flow rate was 2 mL/h at each inlet.

3.2.4 Effects of flowrates on the generated vortices

To study the effects of flowrates on the performance of the generated vortices, the driving
frequency was fixed at 2.0 kHz and the voltage was 100 V while the flowrate at each inlet
was increased from 2 mL/h to 20 mL/h. The working fluids were DI water. The vortex
intensity versus time for different flowrates was plotted in Fig. 3.15. It can be clearly seen
that at low flowrates the strong vortices are relatively easy to be generated, while at high
flowrates the votices are relatively difficult, even unable to be generated if the flowrate at
each inlet is equal to or higher than 10 mL/h. The time that it takes for the vortex intensity to
reach 30% is increased with the flowrates of the fluids. The vortex intensity under various
flowrates is demonstrated in Fig. 3.16 at 50 s when the flows are fully developed for all cases. It
shows that the vortex intensity decreases with the increasing flowrate. At the flowrate of 2 mL/h,
the vortex intensity is up to 90% and when the flowrate at each inlet is increased to 20 mL/h, the
vortex intensity is dropped to approximately 10%. These vortex generation results indicate
strong interactions of the hydrodynamic driving of the incoming flows and PZT actuations
inside the chamber. The hydrodynamic force from the syringe pump is to maintain the
49
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

flowrate and laminar flows in the channel, while the actuation is to impose an external force
on the flow for the vortex generation. It is believed that relatively higher hydrodynamic force,
corresponding to high flowrates, can counterbalance the external force produced by the
actuations so that the vortex becomes weak, is suppressed, or even completely vanishes.

Fig. 3.15 Development of the vortex intensity versus time after the actuation is turned on under various
flowrates. The actuation frequency and voltage were fixed at 2.0 kHz and 100 V, respectively.

Fig. 3.16 Effect of flowrates on vortex formation and intensity in DI water. The inset images were recorded at
50 seconds after actuation is switched on. The PZT disk was actuated with frequency 2.0 kHz and the applied
voltage was 100 V.

50
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

3.2.5 The vortex generation in a circular chamber

Investigation of the vortex generation in a circular microfluidic chamber is also conducted

for comparison to study the role of the nozzle part in vortex generation. Figure 3.17 shows the
flow pattern development in a circular chamber at different times during actuation. By tracing the
patterns of the fluorescent dyes in the flow, the vortices generated in the chamber can be
visualized and characterized. The experimental conditions as follows: a flowrate of 2 mL/h at
each inlet, an actuation voltage of 140 V and a frequency of 2.0 kHz. Similar process of the flow
pattern development is observed in the chamber except that the vortex is smaller than that
generated in the resonator-shaped chamber. The vortex intensity in a circular chamber was
plotted versus time in Fig. 3.18. It can be seen that about 10 seconds after the actuation, the
vortex intensity reaches 30% and finally, at 50 seconds, the vortex intensity is only about 50%
even when the flow is fully developed. It has been demonstrated that the profile of the nozzle
can be used as an acoustic resonator to achieve high amplitude pressure fluctuations and
therefore, the resonator-shaped chamber can achieve stronger vortices than the circular chamber.

Fig. 3.17 Visualization of the generated vortex in a circular chamber at the voltage of 140 V. The actuation
frequency was 2.0 kHz and the flowrate was 2 mL/h at each inlet.

51
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.18 Development of the vortex intensity versus time after the actuation is turned on in a circular
chamber. The actuation frequency and voltage were fixed at 2.0 kHz and 140 V, and the flowrate at each inlet
is 2 mL/h.

3.2.6 The counter-clockwise vortices

In the previous experiments, the generated vortices are rotating in the clockwise direction. In
fact, we have also found that the vortex in the counter-clockwise direction can be generated in
the chamber. The experimental conditions include: a flowrate of 2 mL/h at each inlet, an
actuation voltage of 100 V and a frequency of 2.0 kHz. As shown in Fig. 3.19, the direction of
the generated vortex is counter-clockwise and the process of the flow pattern development is
almost symmetric about the centerline compared to the previous ones. The intensity of the
counter-clockwise vortex was plotted versus time in Fig. 3.19. It is found that about 7
seconds after the actuation is turned on, the vortex intensity reaches 30% and finally, at 50
seconds when the flow is fully developed, the vortex intensity is up to 90%, which is as
strong as the clockwise vortex generated under the same conditions.

52
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

Fig. 3.19 Visualization of the generated vortex in the counter-clockwise direction. The actuation frequency
was 2.0 kHz, the voltage 100 V and the flowrate 2 mL/h at each inlet.

Fig. 3.20 The intensity of a counter-clockwise vortex versus time after the actuation is turned on. The
actuation frequency and voltage were fixed at 2.0 kHz and 100 V, and the flowrate 2 mL/h at each inlet.

We also found that only one regime of vortex can be generated in one specific assembled
device and it is impossible that one device can have two regimes of vortices no matter how the
actuation conditions are adjusted. Besides, the occurrence chance of the two types of vortices is
approximately equal: when we fabricate ten devices, clockwise vortex appears in about five
devices and counter-clockwise vortex in the other five.

53
 Chapter 3 Vortex Generation in a Microfluidic Chamber: Characterization

3.3 Summary

In this chapter, a novel method to generate vortices in a specially designed microfluidic

chamber with PZT actuations is presented. Experimental studies have been conducted on vortex
generation for liquid flows in this chamber. The vortices under the actuation have been
visualized and characterized in terms of the actuation voltages, frequencies and flowrates.
The results show that by applying actuations, large-scale vortices can be generated inside the
chamber, in which only laminar flows would exist otherwise. A frequency window is identified,
in which the vortex generation is possible. The vortex intensity is less than 20% at 40 V, and
increases to 90% with the actuation voltage equal to 100 V. Bubbles may be induced in the
chamber when the actuation voltage exceeds a threshold. It is also found that the resonator-
shaped nozzle which the microfluidic chamber is featured with can contribute to much stronger
vortices than a common circular chamber. The vortices generated in the chamber can be
clockwise or counter-clockwise while only one regime of vortex can be generated in one specific
device and it is impossible that one device can have two regimes of vortices. This method of
vortex generation is advantageous in many aspects: it has no requirements on the flowrates and
properties of working fluids such as conductivity or magnetism; it can generate large-scale
vortices without the need to introduce and trap external gas in advance; it is non-intrusive and
damage-free without obvious temperature rise. All these advantages are favorable for this
method of vortex generation to be applied in flow control of microfluidic manipulations
including mixing enhancement and distribution of microparticles and nanoparticles.

54
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Chapter 4 Vortex Generation in a Microfluidic Chamber:

Control and Mechanism

4.1 Introduction

In last chapter, it has been demonstrated that the vortices can be generated in a specially
designed microfluidic chamber with PZT actuations and the flow pattern and vortex intensity can
be modulated by varying the actuation conditions, including the actuation frequency, applied
voltage and flowrate. The generated vortices can be rotated in either clockwise or counter-
clockwise direction. However, in one specific device, only one regime of vortex can occur no
matter how the actuation conditions are regulated.
In this chapter, the microfluidic chamber is modified to have two regimes of vortices –
clockwise or counter-clockwise in one device. The vortex direction and intensity are regulated
conveniently by varying the operation conditions. The mechanism of the vortex generation and
control is explored by comparing with the instability of the flow through a sudden expansion or a
rectangular chamber. Specifically, the flow velocity in the outlet channel at various actuation
frequencies is measured in order to support our explanation about the mechanism of the vortex
generation. The vibration displacement and velocity of the PZT disk are measured to investigate
the relation between the vibration and the frequency window for vortex generation and flow
velocity.

55
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

4.2 Experimental methods

4.2.1 Modification of the microfluidic chamber and experimental setup

As shown in Fig. 4.1(a-b), the configuration and dimension of the microfluidic chamber used
in the experiment are the same as previously except that the ceramic layer of the PZT disk is
divided into two transducers which can be actuated separately – Transducer A and Transducer B.
The working transducer of the PZT disk can be altered conveniently via a switch without
interrupting the flow or exchanging the device. Three inlets, instead of two inlets, are smoothly
connected to the chamber through a straight channel at an angle of 36 degrees. The modified
microfluidic chamber for vortex generation is illustrated in Fig. 4.1(c).
The microfluidic chamber is fabricated using the same method in last chapter. The schematic
of the experimental setup is illustrated in Fig. 4.1(d). To visualize the flow patterns of the
generated vortices in clockwise or counter-clockwise direction, the DI water with fluorescent dye
is supplied to Inlet 2 in the middle and the degassed DI water without fluorescent dye is supplied
to the microfluidic chamber at Inlets 1 and 3. The vortex intensity along the straight line C-O-C’
is calculated in terms of Eq. (3.2) for quantitative analysis. In the experiment, Transducer A was
triggered at first and meanwhile the flow patterns were recorded; when the flow was fully
developed, the actuation was turned off; after the flow was resumed to its original state,
Transducer B was triggered.

56
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Fig. 4.1 Illustration of the modified microfluidic chamber and experimental setup. (a) Side view of the
configuration. (b) Top view and geometric dimensions of the configuration. (c) Schematic illustration of the
assembled device. (d) Schematic illustration of the experimental setups. The velocity in the rectangular region
– zone 1 is measured.

A piezoelectric actuator converts an electrical signal in terms of voltages into a mechanical

displacement (called a stroke). As shown in Fig. 4.2(a), the two separate transducers are affixed
to a circular brass sheet to form the PZT disk and the PZT disk will be vibrating as a whole when
either transducer is actuated. When the semi-circular transducer – Transducer A or B is actuated,
the vibration of the PZT disk is considered as the small deflection of a thin circular plate with
off-center loads. As illustrated in Fig. 4.2(b-c), the “vibration center” – the point in the PZT disk
with maximum amplitude is not located at its geometric center, so that the flow induced by the
actuations is asymmetric and the asymmetry is controllable. In the experiments, the flows were
recorded by a camera from the top of the PMMA plate opposite to the PZT disk and thus,

57
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Transducers A and B in Fig. 4.2(a) are shown upside down compared to the illustration in Fig.
4.1(c).

Fig. 4.2 (a) Illustration of the PZT disk. The PZT disk consists of two separate transducers and a brass sheet.
(b) The vibration profile of the PZT disk when Transducer A is working. (c) The vibration profile of the PZT
disk when Transducer B is working.

4.2.2 Measurement of the flow velocity in the outlet channel

In the traditional PIV (Particle image velocimetry) technique, the velocity field of a flow can
be quantitatively measured by tracing the motions of the particles in the fluids. The tracer
particles throughout a region are recorded using the multiple exposure method and the velocity
field is obtained by calculating the displacement of the particles during the known time interval.
The illumination is most commonly provided by a laser sheet. However, due to the limits of laser
sampling rate and image transfer capacity, the PIV measurement is not suited to be applied to the
flows actuated at high frequencies. For example, the common 15 Hz dual-head pulsed Nd:YAG
laser can only capture 15 recordings at most within one second, which is far from our demand.
Therefore, we take advantage of the high-speed camera to record the patterns of the flow with
particles at first and then transfer the images to the PIV software platform to calculate the
particle displacement of two consecutive images. In the experiments, at the frequency of 1.5

58
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

kHz, the sampling rate of recording the images is 21,000 fps, namely 14 images captured in one
cycle.
The measurement of the transient velocity in the rectangular region – zone 1 is illustrated in
Fig. 4.3. The flow patterns of the fluids which are seeded by micro-sized particles are recorded
by the CCD camera and then analyzed through a PIV software platform (DynamicStudio, Dantec
Dynamics, Denmark). In order to obtain high quality signal with around 2-5 particles per
interrogation spot, the traceable polystyrene microspheres (Fluoro-Max, Diameter 4.8μm,
Thermo Scientific) was diluted with degassed DI water to act as the working fluid. The velocity
field of the flow was obtained by calculating the particle displacement of two consecutive images
by means of DynamicStudio.

Fig. 4.3 Schematic illustration of the measurement of the velocity field in the outlet channel. The velocity in
the rectangular region – zone 1 is measured.

4.3 Results and discussion

4.3.1 Visualization of vortex generation and control by fluorescent dyes

Using the fluorescent dye as flow tracers, the flow patterns inside the microfluidic chamber
were recorded by the camera after the actuation was switched on. The flowrates for the three
inlets, 1, 2 and 3 were respectively set as 4.5 mL/h, 1 mL/h and 4.5 mL/h such that the total
flowrate through the chamber was 10 mL/h. The PZT disk was actuated at 100 V and 1.5 kHz.
The flow pattern development with time for Transducers A and B is presented in Fig. 4.4(a-e)
and Fig. 4.4(g-k), respectively. It is seen that, without actuations (Fig. 4.4(a) and 4.4(g)), the
fluids flow separately through the chamber and the interface between the fluids is very clear,
which means that the flow is laminar. Within a few seconds of turning on the actuation, the

59
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

fluorescent-dyed fluids which come from Inlet 2 are distorted and swirled to the whole chamber
as shown in Fig. 4.4(b-d) and Fig. 4.4(h-j). In both cases, a large vortex is formed, occupying the
whole chamber. However, the regime of the vortex is different for the two cases in terms of the
vortex direction: when Transducer A is actuated, the vortex rotates in the clockwise direction,
which is denoted as CW vortex; when Transducer B is actuated, the formed vortex is in the
counter-clockwise direction, which is denoted as CCW vortex. In Fig. 4.4(e) and 4.4(k), the
fluorescent fluid is continuously swirled and finally, the interface between the fluids disappears
and the flow is fully developed. In addition, we also used the suspension of microparticles in the
flow as the tracer to display the flow characteristics of the generated vortices by stacking a large
amount of images of the fluid flow at various moments. As shown in Fig. 4.4(f) and 4.4(l), the
traces of particles clearly present the generated vortices in the chamber, in accordance with the
flow patterns of fluorescent dyes. It is noticeable that the generated vortices are three
dimensional in essence because the flow is induced by the up-and-down vibration of the PZT
disk in the height direction.
In the experiments, the two regimes of vortices are similar to those generated in the chamber
actuated by a complete circular transducer; while the vortex direction can be shifted conveniently
by selecting the working transducer without altering the device or interrupting the flow, which is
different from the vortex reported in Chapter 3. These features are novel and useful, which can
promote the maneuverability of microfluidic operations for mixing enhancement of multiple
reagents and distribution of microparticles and nanoparticles.

60
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Fig. 4.4 Recorded images of the flow pattern development with time. Clockwise vortex (CW vortex) is
generated when Transducer A is actuated (a-e) and counter-clockwise vortex (CCW vortex) is generated
when Transducer B is actuated (g-k). The actuation voltage and frequency are 100 V and 1.5 kHz
respectively, and the flowrate is 10 mL/h in total. The traces of particles of CW vortex (f) and CCW vortex (l)
are in accordance with the flow patterns of fluorescent dyes, respectively.

4.3.2 Effects of actuation conditions on the generated vortices

The effects of the actuation conditions such as the actuation frequency, voltage and flowrate
on the generated vortices are investigated to regulate the vortical flow. The dependence of the
vortex on actuation frequency is presented in Fig. 4.5. The frequency is varied from 1 kHz to 4.5
kHz, and the applied voltage and flowrate are 100 V and 10 mL/h respectively. The flow patterns
in Fig. 4.5 were recorded when the flows were fully developed at 50 s, and the vortex intensity at
this time was calculated. From Fig. 4.5, it is clear that when either Transducer A (solid line) or
Transducer B (dashed line) is actuated, there exists an effective frequency window for vortex
generation. Within the frequency window between 1.5 kHz and 3 kHz, the generated vortex is
quite strong, and the vortex intensity indicator is higher than 80%. The intensity of both CW
vortex and CCW vortex reaches a maximal at 2.5 kHz. When the actuation frequency is lower
than 1 kHz or higher than 4 kHz, the vortex becomes very weak or even unable to occur in the
chamber, and thus the vortex intensity is below 15%. It is also found that the vortex direction

61
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

keeps unchanged when varying the applied frequency: CW vortex is generated only when
Transducer A is actuated, and CCW vortex occurs only when Transducer B is actuated.

Fig. 4.5 Effect of actuation frequency on the generated vortex when Transducer A (solid line) or Transducer
B (dashed line) of the PZT disk is actuated. The actuation voltage is 100 V and the flowrate is 10 mL/h in
total. All the flow patterns are recorded at 50 s.

A further investigation was conducted by varying the actuation voltage from 40 V to 100 V
while keeping the frequency and flowrate as 1.5 kHz and 10 mL/h, respectively. The results
when Transducer A (solid line) or Transducer B (dashed line) is actuated are presented in Fig.
4.6. The inset images were recorded when the flow patterns were fully developed at 50 s and the
vortex intensity at this time was calculated. At 40 V, the fluids with fluorescent dye are slightly
distorted in the chamber for both cases, indicating that the vortex is very weak. When the applied
voltage is varied from 60 V to 100 V, a large vortex occurs inside the chamber and the vortex
intensity increases greatly with the applied voltage while the vortex direction keeps unchanged.
At 100 V, the intensity of CW vortex is as high as 90% when Transducer A is actuated and the
intensity of CCW vortex reaches up to 67% when Transducer B is actuated. It is also found that
with increasing actuation voltages, the intensity indicator of CW vortex is larger than that of
CCW vortex on the whole.

62
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Fig. 4.6 Effect of the actuation voltage on the generated vortex when Transducer A (solid line) or Transducer
B (dashed line) of the PZT disk is actuated. The actuation frequency is 1.5 kHz and the flowrate is 10 mL/h in
total. The inset images are recorded when flows are fully developed at 50 s.

Fig. 4.7 Effect of flowrate on the generated vortex when Transducer A (solid line) or Transducer B (dashed
line) of the PZT disk is actuated. The actuation frequency is 1.5 kHz and the voltage is 100 V, respectively.
The inset images are recorded when flows are fully developed at 50 s.

63
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

To study the effects of the flowrate on the performance of the vortex generation, the total
flowrate was set as 6 mL/h, 10 mL/h, 16 mL/h, 22 mL/h and 40 mL/h while the ratio for the
three inlets, 1, 2 and 3 was maintained constant at 4.5:1:4.5 for all cases. The driving
frequency was fixed at 1.5 kHz and the voltage is 100 V. The working fluids are DI water. It
shows that for both cases, the vortex intensity decreases with the increasing flowrate (Fig. 4.7).
At the flowrate of 6 mL/h, the vortex intensity reaches up to 90% and at low flowrates the
strong vortices are relatively easy to be generated. The votices are relatively weak at high
flowrates, even unable to be generated, for example, the vortex intensity is dropped to below
10%, when the total flowrate is increased to 40 mL/h. It is also found that the different flowrates
can only change the vortex intensity and the vortex direction keeps unchanged: CW vortex is
generated only when Transducer A is actuated, and CCW vortex occurs only when Transducer B
is actuated. Therefore, except for the vortex direction, the vortex intensity can also be tuned
conveniently by varying the applied frequency, voltage or the flowrate.
In Fig. 4.5 – 4.7, although the direction of the CW and CCW vortex is opposite, the
vortex intensity of Transducer A and Transducer B should be equal in ideal circumstances.
However, due to the manufacturing challenges in such a small scale, the real device
fabricated in experiments is not perfectly symmetric about the mid-plane where the x-axis is
located and this kind of configuration asymmetry results in the difference of the intensity of
CW and CCW vortex. The configuration asymmetry may originate from: (i) the transducer
may not be perfectly separated into two parts; (ii) the PZT disk may not be concentric with
the chamber; (iii) soldering tin of the wires on the PZT disk results in the asymmetry of the
mass distribution of the PZT disk.

4.3.3. Mechanism of vortex generation and control

It has been demonstrated, in the previous sections, that the vortices can be generated in the
microfluidic chamber incorporated with actuators and the vortex directions can be controlled by
varying the actuation conditions. Regarding the mechanism, we propose that the vortex
generation in the present study is induced by the instability of the actuation-induced pulsatile
flow through the sudden expansion part at the outlet of the chamber, while the vortex control is
realized through the asymmetric flows in the chamber induced by the upper or lower transducers.

64
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Acoustic streaming is a second order nonlinear effect where the presence of an oscillatory flow
generates steady mean vortical flows [94-95]. However, the acoustic streaming relies on
momentum transfer from the acoustic waves – standing waves or travelling waves – to the fluid
[95]. Herein, the typical actuation frequency for vortex generation is 1.5 kHz and the actuation
pressure wavelength is about 1 meter (taking the sound speed in water to be 1500 m/s). While the
largest dimension of the microfluidic chamber in our study is about 0.07 m, which is much
smaller than the pressure wavelength. As a result, the actuation pressure is almost uniformly
distributed in the microfluidic chamber and no acoustic wave exists in the chamber at all. In
other words, the fluid is incompressible inside the chamber in the present actuation frequencies,
which indicates that acoustic streaming is not the mechanism of the vortex generation in the
present microfluidic chamber.

Fig. 4.8 Illustration of the flow through a sudden expansion or rectangular chamber. (a) The flow transition
from a pair of symmetric vortices to asymmetric at high Reynolds number. [96] (b) Transition of the flow in a
rectangular chamber from symmetric to asymmetric when the Reynolds number is above the critical value.
[98] (c) The asymmetric flow in a sudden expansion with a large expansion ratio at high Reynolds number.
[100] (d) The asymmetric flow in a rectangular chamber with a large expansion ratio at high Reynolds
number. [101] ER is the expansion ratio.

65
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

It has been reported that the symmetric flow through a sudden expansion yields to a pair of
stable and asymmetric vortices – a large vortex on one side and a small one on the other side, due
to the instability of the flow at high Reynolds number [96-101]. As shown in Fig. 4.8(a), Fearn et
al. [96] studied the flow in a symmetric expansion experimentally and numerically and showed
that in a channel of expansion ratio (ER) equal to 3:1, the asymmetric vortices arose instead of a
pair of symmetric vortices due to the symmetry-breaking bifurcation when Re was above 40.45.
Shapira and Degani [97] demonstrated that the asymmetric vortices can not only be formed in a
“sudden” expansion but also in a “smooth” expansion, and the critical Reynolds number for the
flow instability increases with the degree of “smoothness” of the expansion. For example, for a
3:1 expansion ratio, the critical Reynolds number becomes 147 for an expansion of 10° semi-
angle. Mizushima et al. [98] investigated the transition of the flow through a rectangular chamber
with ER equal to 3:1 and aspect ratio (AR) equal to 7:3 (Fig. 4.8(b)). They found that the steady
and symmetric flow became two vortices of unequal size when Re was larger than 47.7. Further
investigations show that in the case of a large expansion ratio, the asymmetry of the two uneven-
sized vortices in the flows through a sudden expansion or a rectangular chamber is greatly
enhanced due to the reduction of the flow stability, as illustrated in Fig. 4.8(c-d) [99-101].
Revuelta demonstrated that the asymmetric vortices in the flow through an expansion of large
expansion ratios tended to appear for lower values of Reynolds number due to the momentum
exchange between the incoming jet and the recirculating fluids [100]. Dufresne et al. studied the
flow fields in a rectangular shallow basin at high Reynolds number and found that in the
chamber with ER =16:1, the asymmetric flow had two vortices with distinct sizes: compared to
the case of ER =3:1, a larger vortex was formed inside the chamber and a smaller vortex was
formed on the upper corner [101]. In addition, researchers demonstrated that even for a pulsatile
flow through an expansion channel or a rectangular chamber, the transition to asymmetric
vortices from a pair of symmetric vortices can also occur as a result of the flow instability and
the periodicity of the flow is lost when the Reynolds number exceeds the critical value [102-
103].

66
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Fig. 4.9 The fluids flow into and out of the chamber through the outlet channel with the vibration of the PZT
disk.

In our study, the PZT disk vibrates in the height direction and leads to the variation of the
chamber volume. Since the flowrate at the inlets is maintained constant, the outlet channel
becomes the only passage of the actuation-induced flow. Therefore, due to the conservation law
of mass, the fluids are actuated into and out of the chamber through the outlet channel
corresponding to the sucking and squeezing phase of the PZT disk, as shown in Fig. 4.9. The
ratio of the diameter of the circular chamber (16mm) versus the width of the outlet channel
(1mm) reaches up to 16 which is a sudden expansion of a large expansion ratio. Therefore, the
vortex generation in our study is probably induced by the instability of the actuation-induced
pulsatile flow through the sudden expansion part of the outlet channel at high Reynolds number
and the different vortex directions are realized through the asymmetric flows caused by
asymmetric actuations of the upper or lower transducer as illustrated in Fig. 4.10.

Fig. 4.10 The fluids flow through a suddenly expanded part with a very large expansion ratio (ER) in the
outlet channel when the PZT disk is vibrating.

67
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

4.3.4. The flow velocity in the outlet channel

In order to support our aforementioned assumption of the vortex generation, the transient
velocity of the flow through the outlet channel of the chamber was measured to obtain the
Reynolds number. A rectangular region – zone 1 (indicated in Fig. 4.3) was selected to measure
the time-dependent flow velocity in the outlet channel. The flow patterns of the fluids which
were seeded by micron-sized particles were recorded by the high-speed camera and the images
are analyzed through a PIV software platform to calculate the flow velocities. In the experiments,
the applied voltage and flowrate were 100 V and 10 mL/h, respectively. Figure 4.11 exemplifies
the velocity fields in the outlet channel at six moments of one cycle for the experiment at 1.5
kHz. The sampling rate of the camera was 21,000 fps, which means 14 records every cycle. It is
clear that the flow is similar to a plug flow with very thin boundary layer, and the direction and
magnitude of the flow velocity are periodic with time.

Fig. 4.11 The velocity fields in the outlet channel at six moments of one cycle when the actuation frequency is
1.5 kHz.

68
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

The flow velocity in the channel varies with the positions and thus the mean velocity across
the outlet channel is calculated. The cross-section mean velocity u in the outlet channel is
plotted with time in Fig. 4.12. It can be seen that when the transducer A (solid line) or transducer
B (dashed line) is actuated, the mean flow velocity fluctuates sinusoidally, indicating that the
flow through the outlet is pulsatile corresponding to the vibration of the PZT disk. The peak
velocity u p of the pulsatile flow reaches up to 0.45 m/s for both cases. This flow velocity is

much greater than the mainstream velocity (0.01 m/s, denoted by red dashed line), which means
that the actuation-induced flow by the PZT disk is much greater than the stream flow from inlets.
By taking u p = 0.45 m/s as the characteristic velocity, outlet channel width (= 1 mm) as the

characteristic dimension, and water as the working fluid, the Reynolds number is calculated to be
450, which is large enough for the flow instability in a sudden expansion of ER equal to 16 to
generate vortices. The high Reynolds number justifies the proposed mechanism of vortex
generation from one perspective.

Fig. 4.12 The variation of the cross-section mean velocity in zone 1 of the outlet channel with time in one cycle
when Transducer A (solid line) or Transducer B (dashed line) is working.

The peak-to-peak velocity or the velocity amplitude is defined as a half of the difference
between the maximum value and the minimum value during one cycle. To study the effects of
the actuation frequency on the flow velocity, we plot the flow velocity amplitude versus the
actuation frequency in Fig. 4.13. It can be found that there also exists a frequency window for the

69
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

flow velocity amplitude: within the frequency range between 1.5 kHz and 3 kHz, the amplitude
of the pulsatile velocity is very large, about 0.4 m/s, while out of this frequency range, the
velocity amplitude dramatically drops to 0.1 m/s. This frequency window of the velocity
amplitude is in accordance with the frequency window for the vortex generation shown in Fig.
4.5, which supports our aforementioned assumption of the vortex generation from another
perspective.

Fig. 4.13 The effect of the actuation frequency on the amplitude of the pulsatile velocity in the outlet channel
when Transducer A (solid line) or Transducer B (dashed line) is actuated.

In summary, we can state the mechanism of the vortex generation and control as follows:
when the PZT disk is actuated, the fluids periodically flow into/out of the chamber through a
suddenly expanded part with very large expansion ratio; at the proper actuation conditions, the
velocity of the pulsatile flow is very large that the flow loses its stability at high Reynolds
number; as such, a large vortex is formed in the chamber throughout the entire cycle of the
actuation and the regime of the generated vortex is dependent on the asymmetric flow induced
by the different working transducer of the PZT disk [104].

70
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

4.3.5. Vibration measurement of the PZT disk

The frequency window (1.5 kHz – 3.5 kHz) for the vortex generation or flow velocity in the
channel is associated with the natural frequency of the diaphragm. As shown in Fig. 4.14, to
explore the relation between the vibration of the PZT disk and the frequency window of vortex
generation and flow velocity, we measured the vibration velocity amplitude (maximum vibration
velocity) of the center of the PZT disk (point O) as a function of the forcing frequency by means
of LDV (Laser Doppler Vibrometer, PSV300, Polytec Ltd., UK) technique.

Fig. 4.14 Schematic illustration of the experimental setup for measuring the vibration velocity and
displacement of the PZT disk.

The amplitude of the vibration velocity is plotted with the actuation frequency in Fig.
4.15(a). We can see that there is a frequency window for the vibration velocity amplitude
between 1.5 kHz and 3 kHz and it is agreeable with the frequency range for the vortex generation
or the flow velocity amplitude in the previous sections. The vibration displacement amplitude of
the center O is shown in Fig. 4.15(b). We can see that the frequency dependence in Fig. 4.15(b)
does not match the frequency window for the vortex generation or the flow velocity in the outlet
channel. Thus, the measurement results of the vibration of the PZT disk show that the frequency
window for vortex generation or flow velocity in the outlet channel matches the frequency range
for the vibration velocity of the PZT disk instead of the vibration displacement of the PZT disk.

71
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Fig. 4.15 (a) Vibration velocity amplitude of the center O with the actuation frequency. (b) Vibration
displacement amplitude of the center O with the actuation frequency. Solid line – Transducer A is working;
Dashed line – Transducer B is working.

This agreement can also be explained in terms of the actuation-induced flow by the PZT disk.
We illustrate the side view of the microfluidic chamber in Fig. 4.16. The vibration of the PZT
disk leads to the variation of the chamber volume and the fluids are actuated into and out of the
chamber through the outlet channel since the flowrate at inlets are constant at 10 mL/h controlled
by pumps. It means that due to conservation of mass, the flowrate at the outlet is approximately
equal to the change rate of the chamber volume because the mainstream flow from inlets is
negligible (0.01 m/s versus 0.4 m/s). In another word, the flowrate or the mean velocity in the
outlet channel is proportional to the vibration velocity of the PZT disk rather than the vibration
displacement of the PZT disk. Therefore, the frequency window for vortex generation or flow
velocity in the outlet is agreeable with the frequency range for the vibration velocity of the PZT
disk, rather than the vibration displacement of the PZT disk.

72
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

Fig. 4.16 Side view of the microfluidic chamber for vortex generation. The fluids are actuated into and out of
the microfluidic chamber due to the volume change corresponding to the PZT vibration.

4.4 Summary

This chapter presents a modified microfluidic chamber for vortex generation and control
based on the aforementioned microfluidic chamber with PZT actuations. The transducer of the
PZT disk is divided into two parts which can be actuated separately. The generated vortex is
visualized by tracing fluorescent dyes in the fluid and recorded by a CCD camera. The results
show that two regimes of vortices – CW (clockwise) vortex and CCW (counter-clockwise)
vortex can be formed inside a microfluidic chamber with actuations when the PZT disk is
actuated at proper frequencies and voltages. The direction of the vortex can be shifted on-
demand from clockwise to counterclockwise by simply switching between two different
transducers and the vortex intensity can be tuned by the actuation frequency, voltage and
flowrate.
To explore the mechanism of the vortex generation, PIV measurements have been conducted
to obtain the flow velocity in the outlet channel. The results show that the flow in the outlet
channel is pulsatile with the vibration of the PZT disk and the transient velocity is as high as 0.4
m/s at proper actuation conditions, leading to a large Reynolds number. A frequency window of
the flow velocity amplitude is identified between 1.5 kHz and 3 kHz, which is in accordance
with the frequency range for the vortex generation. The vortex generation in the chamber with
actuations is due to the flow instability when the fluids are driven through the sudden expansion
part of the outlet channel at high Reynolds number. Based on the vibration measurement of the
PZT disk, it is found that the frequency window for vortex generation or flow velocity in the
73
 Chapter 4 Vortex Generation in a Microfluidic Chamber: Control and Mechanism

outlet is agreeable with the frequency range for the vibration velocity of the PZT disk, rather
than the vibration displacement of the PZT disk. The controllable vortex direction via a switch is
a very novel and useful feature, which may promote the maneuverability of microfluidic
operations for mixing enhancement of multiple reagents and distribution of microparticles and
nanoparticles.

74
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Chapter 5 Dynamics of multi-bubbles in a Microfluidic

Chamber

5.1 Introduction

Bubble dynamics has been widely applied in lots of microfluidic processes, such as
micropumping, micromixing, sorting, flow control and manipulation of micron-sized objects etc.
However, the challenges of energy concentration, viscous surface force and low number of
cavitation nuclei make it difficult to generate bubbles directly in microfluidic channels via the
conventional acoustic method. The current methods of bubble formation in microfluidics include
laser radiation and introduction of external gas. There are few reports about direct generation of
bubbles in microfluidic systems. Besides, most studies on bubbles dynamics under actuations in
microfluidics are focused on high frequencies, ranging from 20 kHz to several MHz. At low
frequencies, the cavitation phenomena approach in a way hydrodynamic cavitation and the
pressure wavelength will be long compared to the bubble size and microfluidic channel. As a
result, the driving force to propel the bubble forward at low frequencies greatly differs from
those at high frequencies. Therefore, the dynamics of bubbles in the microfluidic system,
including the radial oscillation, translation and interactions with boundaries, needs to be
thoroughly investigated.
In the previous chapters, we have demonstrated a new design of microfluidic chamber for
vortex generation which takes advantage of PZT actuations. In the experiments, we observed that
bubbles can be induced at the frequency of kilo-Hz range when actuation voltage is high enough.
This phenomenon inspires us that it is an excellent method of bubble generation in microfluidics.
Therefore, we investigate the dynamics of bubbles which are directly generated inside the
microfluidic chamber, including the bubble generation, oscillation and translation. The bubbles
can be generated in the microfluidic chamber within the frequency window between 0.5 kHz and
5 kHz. To reduce the influence of the vortex, we only investigate the flow actuated at the
frequency of 1.0 kHz. Experiments are conducted to observe bubble generation, radial oscillation
and translation by tracing fluorescent dyes in the flow and photographing the motions of the

75
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

bubbles in the nozzle part. Numerical simulation has been carried out using the commercial CFD
software Ansys Fluent to elaborate the mechanism of bubble generation in the chamber.

5.2 Experimental methods

The microfluidic chamber used in the present study has the same configuration and
dimension as described in Chapter 3. As shown in Fig. 5.1, the chamber contains two parts, a 16
mm diameter circular chamber and a nozzle with an acoustic resonator profile to achieve high
amplitude pressure fluctuations. The chamber is made by two PMMA plates sandwiched by a
dry adhesive layer with 300 µm in thickness (Arclad 8102 transfer adhesive, Adhesives Research,
Inc.). Two inlets are connected at a 60-degree angle to the chamber by a straight channel of 10
mm in length and 1 mm in width, and another straight channel is connected to the other end of
the chamber to form the outlet. A piezoelectric (PZT) disk is attached to the bottom of the
circular chamber to provide actuation. The PZT disk consists of a PZT ceramic layer of 15 mm
diameter and a brass sheet of 22 mm diameter. The height of the channel is controlled by the
thickness of the adhesive layer. The geometry of the entire configuration is precisely
manufactured by a laser cutting machine (Universal M-300 Laser Platform, Universal Laser
Systems Inc., Arizona, USA). Finally, the inlets and outlet are assembled to the plate by epoxy
glue (Araldite, Huntsman Advanced Materials, USA).

Fig. 5.1 Schematic illustration of the microfluidic chamber. (a) Side view of the configuration. (b) Top view of
the chamber with geometric dimensions. The rectangular region zone 2 is selected to record the bubble’s
motion.

76
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

The schematic of the experimental setup is illustrated in Fig. 5.2. The working fluid, DI
water, is supplied to the microfluidic chamber through two inlets. The flowrate is controlled by
two syringe pumps (KD Scientific Inc., USA). The piezoelectric disk is driven by an external
signal generator (33120A, Hewlett Packard) and an amplifier (790, PCB Piezotronics). A high
speed CCD camera (Phantom V711, Vision Research, USA) is mounted on top of the chamber to
record the flow field and bubble motion. In the experiments, we focused on the motion of the
generated bubbles moving through zone 2 (indicated in Fig. 5.2). The actuation frequency is
fixed at 1 kHz to weaken the effect of the vortex and the flowrate at each inlet is 5 mL/h.

Fig. 5.2 Schematic illustration of the experimental setup.

5.3 Results and discussion

5.3.1 Visualization of bubble generation by fluorescent dyes

In order to visualize the bubble generation, development and moving tracks, the degassed DI
water solution is supplied to the microfluidic chamber from inlet A (indicated in Fig. 5.1(b)),
and the same DI water solution with fluorescent dye is supplied to inlet B. The total flowrate
through the chamber is 10 mL/h. The PZT is actuated at 150 V and 1 kHz.
The images of the flow at different stages are presneted in Fig. 5.3. It is seen that, without
the actuation (Fig. 5.3(a)), the upper fluid (without fluorescent dye) and lower fluid (with
fluorescent dye) flow separately through the chamber, and no bubbles are observed. As shown in

77
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.3(b), after the actuation is switched on, a group of small bubbles are found to be generated
in the area near the center of the chamber. As shown in Fig. 5.3(c), after being generated, the
bubbles travel upstream against the hydrodynamic flow from the inlets, which is of great
interests. The bubbles arrive finally at some position close to the upstream straight channel and
keep relatively equilibrium (Fig. 5.3(d)). The violent oscillation, coalescence and breakup of the
bubbles cause strong churning of the fluids, so as to enhance the mixing of the fluids inside the
whole chamber. It is observed in our experiment that the bubbles can be generated to form as a
single bubble, or as a group of bubbles in a cloud moving upstream, depending on the applied
voltage and the degree of the fluid degassing. A single bubble tends to occur at low voltage (e.g.
120 V) in heavily degassed water, especially at the immediate beginning when the actuation is
switched on.

78
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.3 Visualization of the bubbles generated in the chamber with the actuation at frequency 1.0 kHz and
voltage 150 V. (a) Recorded image before actuation: the flow is laminar throughout the microfluidic chamber.
(b) Recorded image when the actuation is switched on: the bubbles are generated. (c) Recorded image
showing that the bubbles are traveling upstream against the main stream. (d) Recorded image when the
bubbles arrive at some position close to the straight channel. The flowrate is 10 mL/h in total and the flow
direction is from right to left.

5.3.2 Motions of bubbles in the nozzle

The motions of a group of bubbles moving through zone 2 are recorded by the camera at
12,000 fps, namely 12 images captured in per cycle based on the actuation frequency of 1 kHz.
The time interval between two consecutive images is one sixth of one period in Figs. 5.4 and 5.5.

79
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

It shows that under strong actuations, the bubbles undergo a cycle of explosive growth and
violent collapse, and during each period of oscillation the volumes of bubbles are varied by
several orders of magnitude: in Fig. 5.4(a), the diameter of the largest bubble is 0.45 mm while
in Fig. 5.4(c-e), the bubbles are fragmented or even dissolved by the compression of the
actuation. It can also be found that, the bubbles undergo extensive coalescence and breakup
under the action of actuations. Such a response with the fragmentation or dissolution of bubbles
during the collapse phase is termed as “transient cavitation”, which is distinguished from the
“stable cavitation”. After the diameter of the generated bubbles exceeds the height of the
channel, the bubbles will lose their spherical shape and possess flat surfaces lubricated by thin
liquid films near the top and bottom channel walls.

Fig. 5.4 Photographic series of the dynamics of bubbles in one cycle after being generated. The time interval
between two successive images is 1/6 T, or 0.167 ms. The actuation voltage and frequency are 150 V and 1
kHz, respectively.

80
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

The images of the bubbles at the same phase of different cycles are displayed in Fig. 5.5. The
time interval between two consecutive images is one period, that is, 0.001 second. At high
frequencies, especially in standing waves, bubbles can self-organize into a variety of structures
such as streamers, clusters or bubble layers, and evolve on a timescale much larger than the
acoustic period. However, in the present study, the bubble clouds are distributed randomly in
space without certain structures due to the low frequency. For the case of 1 kHz actuation in
water, the actuated pressure is almost uniformly distributed in the microfluidic channels, and no
acoustic standing waves or travelling waves exist in the microfluidic chamber and thus, the
bubbles cannot be organized in a certain pattern under the actuation.

Fig. 5.5 Images of moving bubbles captured at the same phase of different cycles. The time between two
successive images shown in the figure is one period, that is, 0.001 second. The actuation voltage and frequency
are 150 V and 1 kHz, respectively.

5.3.3 Bubble coalescence and breakup

It is known that at low frequencies, the primary Bjerknes force which results from the
radiation of the oscillatory pressure field is very weak while the secondary Bjerknes force which
is the mutual interaction between oscillating bubbles is dominant at a short distance. The
secondary Bjerknes force can cause the attraction or repulsion of bubbles depending on the phase
81
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

difference of two bubbles. Therefore, for the bubbles subjected to an oscillatory pressure field,
the coalescence is further complicated due to the secondary Bjerknes force.

Fig. 5.6 Coalescence of two oscillating bubbles. One small bubble is attracted by a large one and then the two
bubbles merge together to form a single one. The time between two successive images is 1/12 T, or 0.083 ms.
The actuation voltage and frequency are 150 V and 1 kHz, respectively.

The phenomena of bubble coalescence were photographed involving two bubbles and the
process was exemplified in Fig. 5.6. It is seen that, during the compression phase a small bubble
approaches a large one quickly under the action of the secondary Bjerknes force, and
subsequently at expansion phase the two bubbles grow in size simultaneously until collide with
each other. When the surfaces of the two bubbles touch with each other, a saddle-shaped
“bridge” is built at first and the large bubble immediately “swallow” the small one with the

82
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

volume growth. After the surface area of the resultant cusp decreases, the process of bubble
coalescence is finished. Finally, the consequent bubble contracts and expands with the periodic
actuations. The whole process of bubble coalescence happens during an extremely short time,
which is about one third of a period (0.33 ms).
The bubble model such as the Rayleigh-Plesset equation which is commonly used is based on
the assumption that the oscillating bubble maintains spherical. However, when a bubble is
collapsing strongly, the interface between the dense liquid and less dense gas is violently
accelerated in the inward direction and thus, the bubble’s shape is distorted due to the Rayleigh-
Taylor instability. Under strong actuations, the shape instability of an oscillating bubble can even
become so strong that breakup of the bubble takes place.
An example of the bubble breakup is demonstrated in Fig. 5.7. Contrary to the process of
coalescence, as the bubble is collapsing rapidly, a cusp may be expelled from the base and then
the “neck” connecting the two parts of the bubble is thinning quickly. When the interfacial
tension is no longer able to maintain the bubble intact, a small bubble is shedded from the main
one. The whole process of bubble breakup is even faster than the coalescence, and takes only one
fourth of one period or 0.25 ms. Two mechanisms for the bubble breakup in fluids are reported,
the inertial mechanism leading to sudden breakup by shear stress overcoming the surface tension
and the resonance mechanism when the exciting frequency matches the natural frequency. Here,
the natural frequency of the bubbles is higher than 10 kHz and the actuation frequency is far
below the bubble’s natural frequency and thus, the bubble breakup is caused by the large shear
stress caused by the large velocity gradients.

83
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.7 Breakup of one oscillating bubble at the phase of the violent collapse. The time between two
successive images is 1/12 T, or 0.083 ms. The actuation voltage and frequency are 150 V and 1 kHz,
respectively.

In fact, when the bubbles are translating forward, coalescences and breakups of bubbles
happen frequently. Figure 5.8 shows an example of two bubbles undergoing a series of breakup
and coalescence in very short time (three cycles). At first, one bubble breaks up into two smaller
ones when it is collapsing and immediately the two resultant bubbles merge together into a large
bubble, and in the collapse phase of the next cycle, the large bubble ruptures into several satellite
ones. As such, after alternate breakup and coalescence for several times, the bubble clouds merge
into a single bubble. The whole process of successive breakup and coalescence of the bubbles
causes vigorous motions of the surrounding liquid, which has a major contribution to the mixing
enhancement of the fluids.

84
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.8 A series of coalescence and breakup of the oscillating bubbles in the process of translation. The time
between two successive images is 1/12 T, or 0.083 ms. The actuation voltage and frequency are 150 V and 1
kHz, respectively.

5.3.4 Bubble-wall interactions

Oscillating bubbles near boundaries may not maintain spherical due to the loss of symmetry.
It is reported that a nearby solid wall causes the collapsing bubbles to develop a high-speed
liquid jet towards the solid surface, together with the formation of splash, vortex ring and even a
sequence of shock waves [105].
Figures 5.9 and 5.10 give the example of an oscillating bubble in the proximity of the upper
boundary and lower boundary, respectively. To avoid the breakup of the bubble in the process of

85
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

oscillation, we apply the actuation of lower voltage, 120 V. We can observe that due to the
boundary effect, the bubble in each case cannot maintain perfectly spherical but become oblate
or ellipsoidal. With the rapid contraction of the bubble, a liquid jet impacts towards the boundary
in the 3rd and 4th frame of each case, which causes the bubble surface to sink from one side to
the boundary. The bubble is further squeezed until it reaches the minimum volume in the 7th
frame. Subsequently the bubble expands and goes into the next cycle. It is also noticeable that
when the bubble is contracting quickly, the direction of the liquid jet is not perpendicular to the
boundary wall, but partially directs against the main stream, which leads to the bubble’s
translation upstream in the nozzle.

Fig. 5.9 The dynamics of an oscillating bubble nearby the upper boundary (1-13) and the trajectory of the
bubble by stacking three images of the bubble corresponding to the same phase in a cycle. The time between
two successive images is 1/12 T, or 0.083 ms. The actuation voltage and frequency are 120 V and 1 kHz,
respectively.

86
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.10 The dynamics of an oscillating bubble nearby the bottom boundary (1-13) and the trajectory of the
bubble by stacking three images of the bubble corresponding to the same phase in a cycle. The time between
two successive images is 1/12 T, or 0.083 ms. The actuation voltage and frequency are 120 V and 1 kHz,
respectively.

Strasberg reported that when a bubble is close to a rigid boundary, its natural frequency shifts
toward the lower frequency region [106]. Therefore, another effect of the boundary walls,
including the top and bottom PMMA plates, is to make the nearby pulsating bubbles have lower
resonance frequencies compared to free bubbles in infinite liquid. In addition, the parallel top
and bottom walls can introduce additional drag force onto the bubble translation [107]. However,
the confinement of the two parallel and symmetric walls does not fundamentally change the
characteristics of the bubble dynamics in the x-y plane, such as the translation direction [108].

5.4 Numerical simulation

In order to explore the mechanism of bubble generation and oscillation in the microfluidic
chamber under low frequency actuation, a three dimensional numerical simulation has been
conducted by using the commercial software Ansys Fluent, focusing on pressure distribution and
variation in the chamber.

87
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

5.4.1 Governing equations and boundary conditions

The bubbles generated inside the chamber are associated with the cavitation of the DI water
under the low-frequency actuations. Thus, the pressure variation and distribution are key issues
to understand the mechanisms of bubble generation in the microfluidic chamber. Due to small
dimensions of the microfluidic chamber, it is difficult to measure the pressure experimentally.
Here, numerical simulations using Ansys Fluent are performed to study the pressure variation
and distribution in the chamber. The configuration and dimension of the chamber for the
numerical simulation are identical to the experimental device. The computation domain and
meshing scheme are shown in Fig. 5.11. There are 25 layers of meshes along the channel height
direction and the meshes inside the chamber and at the Y junction are specifically refined.
Under the assumptions of incompressible, Newtonian fluid, the governing equations of
continuity and momentum are expressed as,

 V  0 , (5.1)

V   p 
 V  V     2V . (5.2)
t 

⃑ is the velocity vector, p the pressure, ρ the density and υ the kinetic viscosity.
where 𝑉
In accordance with the experiments, the top wall is stationary and the bottom wall with PZT
disk is set as a moving boundary of the PZT disk. The vibration of the PZT disk is described by
the displacement ζ(x, y, t)
   2 2 2
 x  y2 
 
 A0 sin(2ft ) 1   
 , within disk
 ( x, y , t )     (5.3)
  Rchamb   .
  

 0, outside disk

where the amplitude of vibration A0 is 0.5 μm, the frequency imposed to the PZT disk f is 1.0
kHz, and Rchamb is radius of the chamber, which is 8 mm in our experiment.
The boundary condition for the inlet and outlet is set as velocity inlet and pressure outlet,
respectively. All the boundary conditions are listed in Table 5.1. Based on a meshing
independence study, it is found that the meshing scheme with 877300 cells in total is sufficient
for our simulations.

88
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.11 (a) Geometry and dimension of the numerical simulation domain. (b) Meshing scheme in x-y plane
(not to scale). (c) Meshing scheme in the height direction (not to scale).

Table 5.1 Boundary conditions used in the numerical simulation.

Component Boundary condition Description
Inlet Velocity inlet 0.00926 m/s
Outlet Pressure outlet Reference pressure: 101325 Pa
Fluid DI water Density ρ=998.2 kg/m3
Dynamic viscosity µ= 0.001003 Pa·s
Wall (Bottom) Moving
Wall (Others) Stationary

The governing equations Eqs. (5.1) and (5.2) are solved by the finite volume method (FVM)
in ANSYS FLUENT and the motion of the moving boundary described by Eq. (5.3) is fulfilled
by the user-defined-function (UDF). The following solution methods and numerical schemes
were used in the calculations: PRESTO! for pressure discretion and second order upwind for

89
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

momentum discretion; PISO scheme for pressure-velocity coupling; and the first order implicit
time-dependent solution for unsteady formulations.
5.4.2 Bubble generation by low pressure

The simulation results are plotted in Fig. 5.12 which shows the pressure distributions with
and without the actuation. It can be seen that without actuations (Fig. 5.12(a)), the DI water
flows through the fluidic chamber from the right to left with a constant flowrate (10 mL/h). The
pressure exhibits monotonic decrease from the inlet to outlet, and finally is equal to ambient
pressure (101325 Pa) at the outlet, and the pressure drop is slight in the chamber. When the
actuation is switched on, the pressure in the chamber fluctuates periodically following the
actuation. The minimum pressure, which is our interest, in the chamber occurs at 0.75 T (T = 1/f
= 0.001s) of each period. A typical pressure distribution at 0.75 T is presented in Fig. 5.12(b). It
is found that, instead of decreasing monotonously, a low-pressure zone takes place in the region
of the chamber where we observed the bubble generation in the experiments. The minimum
pressure is as low as 469 Pa, which is very critical to the bubble generation.

90
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.12 The pressure distributions inside the chamber by numerical simulation. (a) Pressure distribution
without actuation. (b) Pressure distribution under an actuation of frequency f =1.0 kHz at 0.75 T (T = 1/f =
0.001 s). The pressure at Point A is 469 Pa.

It is known that nuclei or microbubbles can exist inevitably in real liquids, even in pre-
degassed water, and they act as the starting points for cavitation [109]. In a pressure field with
the amplitude of Pa, the pressure on the bubble is P0  Pa sint  . A microbubble will undergo very
weak oscillation and tiny volume pulsation unless the amplitude of the applied pressure
fluctuation, Pa, is above a threshold, which is so-called Blake threshold [59]
1
 4  S3 2
PaB  P0  Pv  P0   (5.4)
 27 1   
 S 

where, αS = 2σ/(P0R0) is the Laplace tension of the bubble in ambient conditions, P0=101325 Pa
and Pv =3169 Pa are the ambient pressure and vapor pressure respectively, σ = 0.072 N/m is the
surface tension and R0 is the initial radius of the bubble. Once the amplitude of the pressure
fluctuation is higher than the Blake threshold, the microbubbles under the actuations may expand
to a multiple of the initial size and collapse strongly afterwards, undergoing a violent volume
91
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

variation — the bubbles are observed to be “generated” in the liquid. The microbubbles in liquid
have the characteristic sizes ranging from several microns to tens of microns [109]. By taking a
typical microbubble of 10 µm in radius (R0=10 µm), the Blake threshold PaB can be calculated to
be 100111 Pa, corresponding to the threshold pressure on the bubble PL  P0  PaB , equal to 1214
Pa. The minimum pressure (469 Pa) from the simulation in the low-pressure zone is significantly
below this threshold pressure, and the nuclei of the radius larger than 10 µm can be observed to
be generated in the microfluidic chamber in our experiments.
It is noted that the gas dissolved in a liquid tends to accumulate in oscillating microbubbles
over long periods, and this process of mass exchange from liquid to the bubbles is so-called
“rectified diffusion” or RD effect [51]. When the actuation is applied, even the nuclei less than
10 µm can grow up in size continuously over many periods due to the rectified diffusion and
once the microbubbles are large enough, they will become active, undergoing explosive growth
and violent collapse.

5.4.3 Volume pulsation by pressure fluctuation

The pressure fluctuation over one period at the point P of zone 2 is plotted in Fig. 5.13. It
shows that the pressure fluctuates periodically with its amplitude about 1 bar following a
sinusoidal curve. Under this large amplitude pressure fluctuation, the bubble is expanded
corresponding to the minimum pressure, while the bubble is squeezed to a collapse phase at the
maximum pressure. These can be seen by the inset bubble photos in Fig. 5.13. It should be
pointed out that the simulation results are obtained under the conditions of single phase and
incompressible fluid. Once the bubbles are generated in the fluid, these conditions are not
applicable and the pressure fluctuation amplitude can be different. However, as the bubble
photos amply show, the actuated pressure fluctuations in the fluid with bubbles are strong
enough to induce a large volume oscillation for the bubbles. This part will be further
quantitatively studied in next chapter, together with the bubble translation.

92
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

Fig. 5.13 Pressure fluctuation in one cycle at point P (0.013, 0, -0.00015) in zone 2. T = 1/f = 0.001 s.

5.5 Summary

In this chapter the dynamics of multiple bubbles in a microfluidic chamber under low-
frequency actuation has been investigated both experimentally and numerically. The bubble
generation is visualized by tracing fluorescent dyes, and the motions, coalescence and breakup of
bubbles are recorded by a high-speed camera. The mechanisms of the bubble dynamics,
including generation and oscillation, are studied by numerical simulation.
Our experiments show that under the actuation of 1 kHz, the bubbles are observed, starting
near the center of the chamber, growing up in size and even moving upstream against the main
flow. Such type of bubble generation is novel and different from those conventional bubble
generations in acoustic fields, which are normally at high frequencies (> 20 kHz). The generated
bubbles become irregular in shape due to the confinement of channel walls, other nearby bubbles
and the curved nozzle walls. When bubbles are close enough under the action of the secondary
Bjerknes force, the phenomena of bubble coalescence may take place. When a bubble is
collapsing strongly, the shape instability of an oscillating bubble can even become so strong as to
lead to breakup of the bubble. When the bubbles are translating forward, coalescences and
breakups of bubbles happen frequently and cause vigorous motions of the surrounding liquid,
which has a major effect on mixing enhancement. The simulation results show that there is a
low-pressure zone in the chamber in each actuation cycle which corresponds to the area of

93
 Chapter 5 Dynamics of multi-bubbles in a Microfluidic Chamber

bubble generation in the experiment. The minimum pressure in this zone is related to a high-
pressure fluctuation above the Blake threshold and the violent expansion and collapse of the pre-
existing nuclei, which leads to the bubble generation in the experiment. The pressure fluctuation
follows the actuation sinusoidally and maintains constant at 1 bar, resulting in considerably
periodic volume pulsation of the bubble.

94
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Chapter 6 Dynamics of a single bubble in a Microfluidic

Chamber

6.1 Introduction

In the last chapter, we have investigated the generation, motions, coalescence and breakup of
multiple bubbles in a microfluidic chamber with low-frequency actuations. The mechanism of
bubble generation and oscillation is investigated by numerical simulation using Ansys Fluent.
The bubble generation is attributed to the cavitation of nuclei in the liquid and the bubble
oscillation to the pressure fluctuation in the chamber. In the experiments, we observed that
bubbles translate upstream against the main flow after being generated, and this phenomenon has
yet been clarified. At high frequencies, the primary Bjerknes force is the driving force of the
bubble translation. However, at the low-frequency actuations, the wavelength of actuation
pressure will be long compared to the bubble size and microfluidic channel, and hence the
pressure gradient is very small around the bubbles, leading to a weak primary Bjerknes force.
Therefore, other mechanisms, such as the wall effect, tend to be the dominant factor to propel the
bubble forward under the low-frequency actuations in a microfluidic chamber.
The bubble dynamics in the vicinity of a plane wall was widely investigated by experiments
and numerical simulations. However, in a microfluidic system, the bubbles are usually enclosed
by multiple walls, and the wall effect will be much more complicated than the case of a single
wall. Therefore, the dynamics of a single bubble in the neighborhood of multiple walls, such as
two intersected walls, will be further investigated.
In this chapter, the dynamics of a single bubble, especially the translation and its relation
with the oscillation, is investigated experimentally and analytically. An analytical model, taking
into account of the multiple wall effect, is proposed to derive the governing equations of bubble
oscillation and translation under the actuations. The analytical results are compared with the
experimental observations. The mechanism of bubble translation is explored by studying the
interaction force between the real bubble and the image bubbles induced by the boundary wall.

95
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

6.2 Experimental results and discussion

It is observed that the bubbles can be generated to form as a single bubble, or as a group of
bubbles in a cloud, depending on the applied voltage and the degree of the fluid degassing. For
the simplicity of quantitative analysis, the motion of a single bubble is investigated in this
chapter. The single bubble tends to occur under low voltage in heavily degassed water, especially
at the immediate beginning when the actuation is switched on. The experimental setup is the
same as that used in Chapter 5 as illustrated in Fig. 6.1. The motion of a single bubble moving
through zone 2 (indicated in Fig. 6.1) is recorded by a high-speed camera. In the experiments, the
actuation frequency is set as 1 kHz and the total flowrate is 10 mL/h, respectively. The transient
radius and displacement of the bubble are measured by processing recorded images in the
software MATLAB. In the experiments, the bubble volume undergoes a great change within one
actuation cycle and after the actuation is switched off, the bubble’s radius is measured as its
equilibrium radius.

Fig. 6.1 Schematic illustration of the experimental setup. The motion of a single bubble moving through zone
2 is recorded by a high-speed camera and then analyzed in MATLAB.

96
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

6.2.1. Bubble trajectory of translation

An individual bubble moving through the region zone 2 is recorded by the camera set at
12,000 fps, namely 12 images captured in one cycle based on the actuation frequency of 1 kHz.
The images of the oscillating bubble of the radius = 0.075 mm at the same phase of each cycle
are stacked into a particular figure (via the software ImageJ), and the resultant figure is shown in
Fig. 6.2. The time interval between two selected images is two periods, that is, 0.002 second. It
shows that a single bubble moves forward along a straight line, which makes it an excellent case
for quantitative analysis. The bubble translates against the direction of the main stream, which is
agreeable with the phenomenon for multiple bubbles.

Fig. 6.2 Movement of a single bubble towards upstream along a straight line after being generated in the
chamber. The trajectory is obtained by stacking four images of the bubble corresponding to the same phase
in a cycle and the time between two successive images is two periods, that is, 0.002 seconds.

97
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Fig. 6.3 (a) Images of a moving single bubble captured at the same phase in different cycles. (b) The
displacement of the bubble versus time normalized by the actuation period T = 0.001 s. The average
translation velocity of the bubble is estimated to be 0.23 m/s, based on the slope of the displacement curve.

The images of the moving bubble at the same phase of different cycles are displayed in Fig.
6.3(a) and the displacement of the bubble center versus time is plotted in Fig. 6.3(b). The time
interval between two successive images is two periods, that is, 0.002 seconds. It can be seen that
the bubble displacement increases with time and the average bubble velocity is estimated to be
0.23 m/s based on the slope of the displacement curve. The present velocity of bubble translation
is much greater than the mean flow velocity (about 0.0093 m/s) inside the chamber. Crum and
Eller [7] reported that a bubble, which was actuated by a standing wave field in infinite liquid,
moved spatially due to the primary Bjerknes force. At the frequencies ranging from 23.6 kHz to
28.3 kHz, the translation velocity reached its maximum, 0.023m/s, which is much smaller than
that in our experiment. In our study, the translation of the bubble is not attributed to the primary
Bjerknes force but the effect of the two inclined boundary walls, which will be discussed in
detail in Section 6.3.3.

6.2.2. Radial oscillation and its interaction with translation

The images of a bubble under actuation at different phases in two successive cycles are
presented in Fig. 6.4. The time between two successive images is 1/12 T, with T = 0.001 s being
the actuation period. It is observed that the bubble oscillates periodically under the actuation,

98
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

indicating that the bubble undergoes remarkable volume variation, and a complete bubble
pulsation cycle is found to consist of slow growth and fast collapse. The bubble maintains
spherical during most time of the oscillation, except that during a short time when the bubble
contracts to its minimum volume, the spherical shape changes into an ellipsoid.

Fig. 6.4 Images of a bubble under the actuation of two successive cycles, showing remarkable volume
oscillation. The time between two successive images is 1/12 T with T = 0.001 s being the actuation period.

Figure 6.5 shows the radius (Fig. 6.5(a)) and displacement (Fig. 6.5(b)) of the bubble in two
successive cycles. From Fig. 6.5(a), it can be seen that the bubble radius does not follow the
sinusoidal actuation with time, and there is a long expansion and steep collapse in each cycle,
which indicates that the radial response of the bubble is a typical nonlinear process under strong
actuation. The bubble radius varies from Rmin = 0.05 mm to Rmax =0.14 mm during one cycle,
corresponding to a ratio of Rmax/Rmin = 2.8 and the ratio of the maximum volume to the minimum
volume about 22. Accompanied by periodical volume variation, the bubble translates in a “jerk”
way as shown in Fig. 6.5(b). The bubble “jumps” forward during a short time interval of the
collapse phase, and then moves backward slightly when growing up during expansion phase. In

99
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

the collapse phase (as illustrated between the two dot-dashed lines), the bubble moves faster, at
the velocity of about 0.55 m/s, which is greater than the average bubble translation velocity
0.23m/s shown in Fig. 6.3(b). This observation is consistent with the numerical results by Reddy
and Szeri [72] for bubbles driven at high frequencies that the bubble translation velocity was
maximized when the bubble collapsed violently.

Fig. 6.5 (a) Variation of the bubble radius in two successive cycles. (b) Variation of the bubble displacement
in two successive cycles. The bubble moves forward at the phase of collapse. T = 0.001 s, is the actuation
period.

6.3 Analytical modelling

A simplified analytical model for a bubble nearby two intersected walls is proposed to
analyze the bubble oscillation and translation under the actuations, especially to explore the
mechanism of the bubble moving upstream against the main flow. To construct the model, we
assume that the flow is potential, the fluid around the bubble is incompressible, and the bubble is
small compared to the bubble-wall distance. The nozzle part of the chamber is further simplified
into two intersected straight walls which extend to infinity.

100
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

6.3.1. Analytical modelling for bubble oscillation and translation

6.3.1.1 The simplified model and governing equations

Based on our experimental observations, the bubbles are confined within a narrow space (300
μm) between the top and bottom PMMA plates, and the bubble translates in the x-y plane. It is
known that the bubble confined by two parallel boundaries will undergo additional drag force
[107]. However, our model does not include the effect of the top and bottom surfaces for two
reasons: the additional drag force does not fundamentally change the characteristics of the
bubble dynamics in the x-y plane, such as the direction of the bubble translation; the additional
drag force from confinements is estimated to be less than 10% on average [108]. Moreover, the
nozzle profile of the chamber at zone 2 is ideally treated as two inclined straight walls and
intersected at an angle of 2α =30° on the right as illustrated in Fig. 6.6. Since the maximum main
flow velocity is about 0.0093 m/s in the microfluidic chamber, which is much smaller than the
observed bubble translation velocity, 0.23 m/s, the fluid is assumed to be stationary with zero
main flow velocity. Thus, the original problem can be converted to a bubble moving in a still
fluid between two intersected walls as shown in Fig. 6.6. The Lagrangian method used by
Doinikov [8] is adopted to derive the governing equations for the bubble’s radius R and
displacement x in the present study. The method of images is introduced to satisfy the boundary
conditions on the wall and the bubble surface and here, we take the first-order images for
simplicity.

101
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Fig. 6.6 Illustration of the analytical model. (a) In the experiment, the bubble is in zone 2 of the microfluidic
chamber. (b) In the model, the bubble is located in a position on the bisector line of two inclined straight
walls.


The velocity potential, Φ, which is defined as V   for the fluid velocity V in the flow
field, satisfies the following governing equations and boundary conditions
 2  0 in the flow field
  0 in the infinity of the flow field

  R  x cos  on the bubble surface
n

0 on the sphere surface
n

where n is the unit normal towards the flow field from the bubble or the boundary; R and x with
dots are the corresponding time derivatives for the bubble’s radius and displacement,
respectively.
The velocity potential of the bubble nearby a boundary wall can be simulated by introducing
a series of image bubbles to replace the boundary wall, and this is so-called the method of
images. The model for a bubble between two intersected walls is illustrated in Fig. 6.7(a),
together with the coordinate system. The image system which satisfies the boundary condition at
the rigid walls is illustrated in Fig. 6.7(b). There are 2n-1 image bubbles in total, where n=
180/2α (n is an integer). R and x denote the time-dependent radius and displacement respectively,
the variables with dots corresponding to the time derivatives, and di is the distance between the
real bubble and the ith image bubble. Local coordinates, (r, θ) and (ri, θi) are located at the center
of the real bubble and ith image bubbles, respectively.

102
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

The velocity potential in response to the actuation in the model is defined as

  Rr  x x (6.1)
where R and x with dots are the corresponding time derivatives for the bubble’s radius and
displacement, respectively, Φr is the unit potential for bubble radial motion and Φx is the unit
potential for bubble translational motion.

Fig. 6.7 The model and image system for a bubble located nearby two inclined walls.

The boundary conditions which satisfy the no-penetration or zero normal velocity at the
surface of the bubble and walls are listed in Table 6.1.

Table 6.1 The boundary conditions at the bubble and wall surfaces.
Bubble Surface Wall Surface
Pulsation  r  r
 1 0
n n
Translation  x  x
  cos  0
n n

6.3.1.2 The kinetic energy

The kinetic energy T of the system is determined by the surrounding liquid

103
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

 2

2
T  dV (6.2)


which can be expressed in terms of Gaussian Theorem by,


T
2    dV

(6.3)

The Green’s first identity reads as,

       dV      n dS

2 
(6.4)
 


Here, the negative sign results from the inward normal of n . Replacing ψ with Φ, we can
immediately obtain,

      dV     n dS

2 
(6.5)
 

For a potential flow satisfying  2  0 , substituting Eq. (6.5) into Eq. (6.3), we can obtain
the volume integrand of the kinetic energy,
 
T 
2   n dS

(6.6)

Then, the total kinetic energy is given by,

 R n dS   2  R  


T  r
 x x r
 x x  R  x cos  dS (6.7)
2  b

where the surface integral is taken over the real bubble surface.
Thus, the kinetic energy of the fluid, T, can be calculated based on the velocity potential
as [76]
 2
T   dS  2Rx   dS  x   cos dS ]
r x 2 x
[R (6.8)
2 b b b

where  is the fluid density,  is a coordinate angle defined in Fig. 6.7, and in the calculation, the
Green’s second identity is used,
 x  r
b
r
n
dS  b
x
n
dS (6.9)

To calculate the kinetic energy T, in our model, the two inclined walls are replaced by 2n  1
(n = 180/2α) image bubbles to meet the boundary condition of zero normal velocity at the walls.
This is illustrated in Fig. 6.7(b). Since 2α =30° and n = 6 there are 11 image bubbles in our
model.

104
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

The oscillation of the bubble is regarded as a point source with periodic strength and thus, the
velocity potential from the bubble radial pulsation of unit strength is obtained by the
superposition of contributions from all bubbles, including the real one and its images, i.e.
2n 1 R 2 2n1  R 2 
 r  r   ir      
i 1  ri 
i 1 r (6.10)

where r is the velocity potential corresponding to the real bubble, and ir is the velocity
potential responding to the ith image bubble.
The translation of the bubble is analogous to a moving sphere in the liquid and the velocity
potential from the bubble translation of unit strength is obtained by the superposition of
contributions from all bubbles, including the real one and images, i.e.
2n 1 2n 1  R3 cos i 
R3 cos 
 x  x   ix   2r 2
    2ri2 

(6.11)
i 1 i 1 

where  x is the velocity potential for the real bubble, and ix is the velocity potential for the ith
image. ξ and ξi, are the coordinate angles denoted in Fig. (6.7b).
The first integral in Eq. (6.8) can be calculated by substituting the potential Eq. (6.10) into
the integral and evaluating it over the real bubble surface. The integration, denoted as I1, gives
4R 4
b
 r dS  4R 3 
D1
 I1 (6.12)

n1

d
1 2 1
where   .
D1 i 1 i dn

By substituting the velocity potential Eq. (6.11) into the second and third integrals in Eq.
(6.8), the two integrals, denoted as I2 and I3, can then be obtained by evaluating integrals over the
real bubble surface,
2R 5
b
 x dS  
D2
 I2 (6.13)

2R 6

2
 x cos dS  R 3   I3 (6.14)
b 3 D3

n1
2 sin i 1 n1 sin 2 i  1 1

1 1
where, in Eqs. (6.13) and (6.14),   and    .
D2 i 1 d i2 d n2 D3 i 1 di3 d n3

The kinetic energy T can therefore be given as

105
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

 2
T [ R I1  2R xI 2  x 2 I 3 ] (6.15)
2
where I1, I2 and I3 denote the three integrals in Eq. (6.8). In the calculations, the velocity
potentials contributed by the image bubble are solved by transforming the coordinates of the
image bubble to the local coordinates of the real bubble and keeping the first two terms of the
Legendre expansion.

6.3.1.3 The dissipation energy

The potential energy U of the system is caused by the variation of the gas volume,
U   P1Vb (6.16)
where Vb is the time-dependent volume of the bubble and P1 is the pressure at the surface of the
real bubble, P1  Pg  2 R  P , Pg the pressure of gas inside the bubble, σ the surface tension and

P the pressure at infinity.

In the situation that the bubble is translating drastically or the bubble is pulsating violently,
the viscous term is insignificant except in the boundary layer.
The liquid velocity can be represented by
  
V  Vr er  V e (6.17)
where the radial component Vr and tangential component V are

R 2 R R3 x cos 
Vr   (6.18)
r2 r3

R 3 x sin 
V  (6.19)
2r 3

The dissipation potential of the system is calculated by

F 
f d (6.20)

where f is the density of the dissipation potential and Ω is the volume occupied by the liquid.
For the incompressible liquid, the dissipation potential density f can be calculated by [110]
f   (Vij ) 2 (6.21)

where Vij is the rate of strain tensor, written as

1 Vi V j
Vij  (  ) (6.22)
2 x j xi

106
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Therefore, substituting Eq. (6.22) into Eq. (6.21), one obtains

6 R 4 R 2 18R5 R x cos  9R 6 x 2 9 R 6 x 2 cos 2 
f   (Vrr2  V
2
 V
2
 2Vr2 )     (6.23)
r6 r7 2r 8 r8

Then, the dissipation potential F of the system can be integrated over the whole volume
F  8RR 2  6Rx 2 (6.24)
Hence, the drag force of the bubble can be obtained as
F
Fer    16RR (6.25)
R

for radial motion and

F
Fex    12Rx (6.26)
x
for translational motion, which is exactly the Levich viscous drag exerted upon the bubble [111].

6.3.1.4 The Lagrangian equation

The Lagrangian L can be obtained by the kinetic energy T and potential energy U,
 2
L  T U  [ R I1  2R xI 2  x 2 I 3 ]  P1Vb (6.27)
2

Substituting Eqs. (6.24) and (6.27) into the Lagrangian equation

d L L F
( )  (6.28)
dt qi q i qi

and carrying out the differentiations with respect to the generalized coordinates R and x, one can
obtain the governing equations for the oscillation and translation of a bubble between two
inclined walls,
R 2  3 2 R  2 2R 2  R3 2 R 3 1 3R 3 2 P2
(R  )R  (  )R  Rx  x  ( '   ) x  (6.29)
D1 2 D1 D2 2 D2 D3 4 2D3 

R R4 6 R 3  3R 2  2 R 3  Fex
(  ) x  (1  ) Rx  R  R (6.30)
3 D3 D3 D2 D2 2R 2 

where D1, D2, D3 and D3’ are coefficients related to the bubble-wall distance and they are given
as following
n1 n1 n1 n1
2 sin i sin 2 i  1 2 sin 2 i
   
1 2 1 1 1 1 1 1 1
  ,   ,   and   .
D1 i 1
di d n D2 i 1 d i2 d n2 D3 i 1 d i3 d n3 D3' i 1 d i3 d n3

107
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

where, di is the distance between the real bubble and the ith image bubble. The coefficients D1,
n 1
D2, D3 and D3’ in Eqs. (6.29) and (6.30) contain terms with i 1
which denote the effect of the

inclined walls. In a limiting case by setting 2α =180° and n = 1, the two inclined walls will be
n 1
reduced to a single flat wall and all the terms with 
i 1
in coefficients D1, D2, D3 and D3’ will

become zero. Then our derived equations Eqs. (6.29) and (6.30) will be exactly the same as the
analytical results obtained by Cole [76].
In the calculations, the scattered pressure at the surface of the bubble P2 is defined by
P2  P1  4eff R R  Pg  2 R  4eff R R  P , and eff the effect viscosity. Fex is the drag force for

the bubble translation calculated by the dissipation potential of the system, Fex  12Rx . The

gas bubble is polytropic so that the pressure of a bubble of radius R is Pg  Pg 0 R0 R 3 , where the

gas pressure at equilibrium Pg0 is calculated by Pg 0  P0  2 R0 , the bubble radius at equilibrium

R0=0.075 mm and κ is polytropic index, clearly, κ=1 for isothermal process and κ=γ (the ratio of
the gas specific heat) for adiabatic behavior. The applied pressure at infinity is
P  P0  Pa sin(2ft ) , where P0 is the ambient pressure, Pa the amplitude of applied pressure
fluctuation, and f the actuation frequency. Pa is set as 0.93P0 which is obtained by means of
numerical simulation in last chapter. The effective viscosity eff consists of the thermal damping
t and fluid viscous damping  [112]. All the parameters in the model are listed in Table 6.2.

108
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Table 6.2 The parameters used in the analytical model.

Parameter Symbol Value
Liquid density ρ (kg/m3) 998.2
Surface tension σ (N/m) 0.0728
Sound speed c (m/s) 1484.7
Ambient pressure P0 (Pa) 101325
Vapor pressure Pv (Pa) 3169
Pressure amplitude Pa (Pa) 0.93P0
Viscous damping µ (kg/(m∙s)) 0.001
Thermal damping µt (kg/(m∙s)) 0.0057
Driving frequency f (kHz) 1.0
Liquid velocity at Vf (m/s) 0.00926
inlet
Polytropic exponent γ 1.4

6.3.2. Modelling results of bubble oscillation and translation

A typical case of a spherical bubble of R0=0.075 mm under the periodic pressure fluctuation
with Pa=0.93P0 is calculated, and the initial displacement of the bubble, x0, is set at -8.25 mm for
the bubble located nearby two inclined walls, which is analogous to the experimental
observations. The coupled equations Eqs. (6.29) and (6.30) are solved by using the ODE function
in MATLAB and the results for the bubble radius and displacement are plotted in Fig. 6.8 and
more details for the bubble radius and displacement over two periods are shown in Fig. 6.9.

109
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Fig. 6.8 The whole process of the motions of the bubble nearby two tapered walls. The whole bubble
movement corresponds to three stages: uniform movement, accelerated movement and final impact towards
the walls. (a) Radius of the bubble with time. (b) Displacement of the bubble with time. The bubble travels at
an average velocity of 0.22 m/s in the stage of uniform movement.

The whole process of the bubble’s motion is presented in Fig. 6.8. We can see that the bubble
travels upstream from the initial position (x0 = - 8.25 mm) to the intersection point of the tapered
walls, corresponding to three stages: uniform movement, accelerated movement and final impact
towards the walls. Initially, the bubble moves upstream at an almost uniform average velocity,
which is so-called uniform movement. In this stage, the average translation velocity is about 0.22
m/s, which agrees well with the experimental velocity of the bubble in zone 2 (0.23 m/s). In
closer proximity to the walls, the bubble is accelerated to move forward. In the stage of final
impact towards the walls, the bubble is experiencing non-spherical deformation. In this stage, the
analytical model is not applicable.
The details of the bubble’s oscillation and translation over two periods are demonstrated in
Fig. 6.9. As shown in Fig. 6.9(a), the response of the bubble does not follow the sinusoidal
actuation of the pressure fluctuation but appears strong and stable nonlinearity: during one cycle,
the bubble grows up rapidly and then experiences a steep collapse followed by a series of after-

110
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

bounces. The bubble radius varies from 0.05 mm to 0.16 mm in one cycle with a ratio of
Rmax/Rmin about 3.2, which is comparable to the ratio of 2.8 in our experiments. As shown in Fig.
6.9(b), the bubble translation mainly occurs in the collapse phase (indicated by two dot-dashed
lines) with an average velocity about 0.6 m/s. While in the growth phase, the bubble does not
move forward but slightly backward. This feature is also agreeable with our experimental
observations in Fig. 6.5. The bubble’s oscillation and translation predicted by the model
demonstrate that, although several simplifications and assumptions are made in the present
analytical model, the results, including bubble oscillation, translation direction and velocity, are
semi-quantitatively comparable with the experimental observations.

Fig. 6.9 (a) Details of the radius in two successive periods for the bubble nearby two inclined walls in an
oscillatory pressure field. (b) Details of the displacement in two successive periods for the bubble nearby two
inclined walls in an oscillatory pressure field.

6.3.3. Further discussion of bubble translation – wall effect

The results from both the experiments (Figs. 6.3 and 6.5) and analytical model (Figs. 6.8 and
6.9) show that the bubbles under the low-frequency actuation translate upstream towards the
proximity of the walls. To explore the driving mechanism for the bubble translation, we

111
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

investigate the motion of the same bubble in infinite liquid under otherwise the same conditions
and further the bubble-wall interaction is studied in terms of the mutual force between the real
bubble and the image bubbles.

6.3.3.1 Comparison with bubble translation in infinite liquid

It is known that at high-frequency actuations, the wavelength of acoustic pressure is short

and comparable to the microfluidic channel scale and the bubble size, and a strong acoustic field
will be established inside the fluidic chamber. The bubbles in the strong acoustic field will
experience acoustic forces, such as the primary Bjerknes force which is associated with the
pressure gradients. The primary Bjerknes force is normally the main driving force corresponding
to the bubble translations at high frequency actuations. However, the situation is different at the
low-frequency actuations. In the present study, the actuation frequency is 1 kHz and the
actuation pressure wavelength is about 1.5m (the sound speed in water is 1500m/s), which is
much greater than the dimension of the microfluidic chamber (0.02 m-0.07 m). As a result, the
actuated pressure is almost uniformly distributed in the microfluidic chamber and the pressure
gradient is very small in the fluid, leading to a negligible primary Bjerknes force on the bubbles.
In order to further illustrate the wall effect, the model of a single bubble in an infinite liquid
(without any walls) constructed by Doinikov [8] is solved here. All the actuation conditions are
the same as that used in our model except that the bubble is free in the infinite liquid. The radius
and displacement of the free bubble are plotted in Fig. 6.10(a) and 6.10(b), respectively. The
results in Fig. 6.10(a) show that the bubble oscillation is less violent than the results of the
bubble with two inclined solid walls. The bubble experiences a volume oscillation with its radius
varying from 0.05mm to 0.15mm with smaller volume ratio (about 3.0) in one cycle. But the
bubble translation is very different. As shown in Fig. 6.10(b), the free bubble under the actuation
will remain stationary if there is no mainstream flow, or will be drifted following the mainstream
flow from the right to the left when a main flow is imposed. From the comparison of the bubble
translations without walls and nearby two intersected walls, it can be concluded that only the
pressure fluctuation cannot drive the bubble to move forward and the bubble translation within
two inclined walls towards upstream is attributed to the wall effects. In other words, the bubble
translation is due to the action of the nearby walls, not the force from the external actuations. In

112
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

our experiments, the walls of the nozzle at zone 2 play the similar role as the inclined walls in the
model.

Fig. 6.10 (a) Radius of a bubble in infinite water under the oscillatory pressure actuations. (b) Displacement
of a bubble in infinite water under the oscillatory pressure actuations. The bubble remains stationary in the
stationary fluid (dashed line) or travels with the main stream at the stream velocity (solid line).

6.3.3.2 The secondary Bjerknes force

It has been known that the bubble translation within two inclined walls towards upstream is
attributed to the wall effects. However, it is unclarified that why the force between the bubble
and the nearby walls are attractive or how the wall effects on the real bubble work. From the
analytical modelling of the bubble’s oscillation and translation, we can see that the intersected
walls around the bubble are replaced by 11 image bubbles, which impose their effects on the
bubble’s motion. It is known that the oscillation of one bubble can lead to radiation force on the
nearby bubble due to the induced pressure disturbance, which is so-called secondary Bjerknes
force. Herein, we can study the secondary Bjerknes force between the real bubble and image
bubble to explore the nature of the bubble-wall interaction.

113
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

Fig. 6.11 Illustration of the real bubble and ith image bubble.

For simplicity, let us assume that the bubble of radius R together with its all image bubbles is
fixed at some position nearby the intersected boundary walls. As shown in Fig. 6.11, the distance
between the real bubble and the ith bubble is constant. Omitting the high-ordered convective
term of the Navier-Stokes equation, the governing equation of the liquid around the bubble is

V
  p  0 (6.31)
t
where V is the liquid velocity caused by the oscillation of the image bubble, ρ is the liquid
density and p is the pressure emitted by the oscillating image bubble.
The oscillating velocity of the image bubble can be calculated
R 2 R
Vi r  (6.32)
ri2

Substituting Eq. (6.32) into Eq. (6.31), we can obtain the pressure gradient resulting from the
image bubble’s oscillation,
p
r
V r
t
 d 2
  i   2
ri dt
R R   (6.33)

At the center of the real bubble, the pressure gradient caused by the image bubble’s
oscillation can be obtained immediately,
p
r ri  d i
 2  
 d 2
d i dt
R R (6.34)

The secondary Bjerknes force for a bubble of volume Vb in liquid of a pressure gradient p
reads as,

114
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber


FsB  Vbp (6.35)
Substituting Eq. (6.34) into Eq. (6.35), one obtains the secondary Bjerknes force of the image
bubble on the real bubble,

FsB  2b
D dt
 
V d 2  
R R eir 

4D

V V e
2 b b ir
(6.36)

where, eir is the unit of the pressure gradient in the radial direction.
When the bubble oscillates periodically with time, the net force on the real bubble is the time

average of FsB over one period,
1 
 
T
     
FsB  V V e 
2 b b ir
 VV  V 2 eir   V 2 eir (6.37)
4D 4D 2  T 0  4D 2

where, < > denotes the time average value. In the coordinate system, the negative net value of

FsB means the force directed to the image bubble or the real bubble is attracted by the image

bubble. Therefore, the mutual force FsB is always attractive since V 2 is always positive. In

the analytical model, the real bubble is attracted by the image bubble behind the nearby boundary
walls due to the secondary Bjerknes force, which also explains why the bubble moves upstream
against the hydrodynamic flow in the experiments.

6.4 Summary

In this chapter, the dynamics of a single bubble in a microfluidic chamber under low-
frequency actuations has been investigated both experimentally and analytically. A single bubble
is generated in the microfluidic chamber with low-voltage actuation (120 V). After being
generated, the bubble under actuation is found to translate against the hydrodynamic flow at an
average velocity of about 0.23 m/s, and it undergoes significant expansion and compression with
a ratio of the maximum to minimum volume up to 22. The bubble moves forward during a short
time interval of the collapse phase with a velocity of 0.55 m/s.
A simplified analytical model which describes the dynamics of a bubble between two
intersected boundaries is constructed for the first time to explore the mechanism of the bubble
translation. This model extends the case of a bubble near a single flat wall to a bubble located
between two intersected walls. Based on the assumptions of incompressible and potential flows,

115
 Chapter 6 Dynamics of a single bubble in a Microfluidic Chamber

the nearby boundary walls are replaced by 11 image bubbles and the governing equations of the
bubble oscillation and translation are derived by the Lagrangian method. The bubble motions
predicted by the model are agreeable with the experimental observations in terms of bubble
oscillation and translation. The comparison between a bubble in infinite liquid and nearby two
intersected walls shows that the bubble translation is due to the attractive secondary Bjerknes
force induced by multiple image bubbles behind the two inclined walls, which is the so-called
wall effect.

116
 Chapter 7 Conclusions and Recommendations

Chapter 7 Conclusions and Recommendations

7.1 Conclusions

The vortex generation and control, and bubble dynamics have been investigated in a
resonator-shaped microfluidic chamber with PZT actuations. The following conclusions are
drawn from this study based on the achieved results.
(1) A novel method of generating vortices has been demonstrated in a microfluidic chamber
incorporated with a PZT actuator. Once being switched on at proper conditions, large-scale and
asymmetric vortices are formed inside the chamber, while only laminar flows would exist
otherwise. The generation and intensity of the vortex can be controlled by actuation voltage,
frequency and flowrate. There is a frequency window between 1.5 kHz and 3 kHz, within which
the vortex can be generated with strong intensity and out of this range, the vortex is very weak,
or even unable to occur. The vortex intensity increases with the actuation voltage, from less than
20% at 40 V to 80% at 100 V. At 120 V, bubbles can be generated in addition to the vortices.
The vortex intensity decreases with flowrate. The microfluidic chamber with resonator-shaped
nozzle has much stronger vortices than a common circular chamber. The vortices generated in
the chamber can be either clockwise or counter-clockwise while only one regime of vortex is
generated in one specific device.
(2) By modifying the microfluidic chamber by dividing the PZT disk into two parts which
can be actuated separately, the direction of the generated vortex can be shifted on-demand from
clockwise to counterclockwise by simply switching between two different transducers. Such
device for vortex generation can generate two regimes of vortices – CW (clockwise) vortex and
CCW (counter-clockwise) vortex when the PZT disk is actuated at proper conditions. The vortex
intensity can be tuned by the actuation frequency, voltage and flowrate, which is similar to the
previous microfluidic chamber. PIV measurements show that the flow in the outlet channel is
pulsatile with the vibration of the PZT disk and the transient velocity is as large as 0.4 m/s within
the frequency window between 1.5 kHz and 3 kHz, leading to large Reynolds number. The
vortex generation in the chamber with actuations is due to the flow instability when the fluids are
driven through the sudden expansion part of the outlet channel at high Reynolds number. Based

117
 Chapter 7 Conclusions and Recommendations

on the vibration measurement of the PZT disk, it is found that the frequency window for vortex
generation and flow velocity in the outlet is agreeable with the frequency range for the vibration
velocity of the PZT disk, rather than the vibration displacement of the PZT disk.
(3) Bubbles are induced when actuation voltage exceeds a threshold (120 V), and such type
of bubble generation is novel and different from those conventional bubble generations in
acoustic fields, which are normally at high frequencies (> 20 kHz). The dynamics of multi-
bubbles in the chamber with actuations at 1 kHz has been investigated both experimentally and
numerically. Under the actuations, the bubbles are observed to start near the center of the
chamber, grow up in size and move upstream against the hydrodynamic flow from inlets. When
the bubbles are translating forward, coalescences and breakups happen frequently, leading to
vigorous motions of the surrounding liquid and a significant effect on mass transfer. The
simulation results show that in each actuation cycle there is a low-pressure zone in the chamber
which matches the area of bubble generation observed in the experiment. The minimum pressure
in this zone is related to a high-pressure fluctuation above the Blake threshold, which leads to the
bubble generation.
(4) The bubbles can be generated to form as a single bubble, or as a group of bubbles in a
cloud, depending on the applied voltage and the degree of the fluid degassing. For the simplicity
of quantitative analysis, the motion of a single bubble is investigated both experimentally and
analytically. Under the actuations, the bubble undergoes significant expansion and compression
with a ratio of the maximum to minimum volume up to 22. After being generated, the bubble
under actuations is found to translate along a straight line against the hydrodynamic flow at an
average velocity of about 0.23 m/s. The bubble movement mainly occurs during a short time
interval of collapse phase. A simplified analytical model which describes the dynamics of a
bubble between two intersected boundaries is constructed to simulate the bubble oscillation and
translation. The governing equations of the bubble radius and displacement are derived by using
the Lagrangian method. The motions of the bubble predicted by the model are agreeable with the
experimental observations in terms of oscillation and translation. The comparison between a
bubble in infinitely large liquid domain and nearby two intersected walls shows that the bubble
translation is due to the attractive secondary Bjerknes force of multiple image bubbles behind the
two inclined walls, which is the so-called wall effect.

118
 Chapter 7 Conclusions and Recommendations

7.2 Recommendations for Future Work

In this study, the vortex generation and control, and bubble dynamics have been investigated
in detail by means of experimental visualization and numerical simulation. However, there are
several unsolved problems which are worthwhile to be studied further in future. The following
are two recommendations.

(1) The flow field in the chamber needs to be characterized by means of PIV measurement to
justify the proposed mechanism of vortex generation. It would be useful to conduct numerical
simulation to study the vortex formation in the microfluidic chamber with actuations and make
direct comparison with the PIV results.
(2) The dynamics of the bubble in the chamber with low-frequency actuations has been
studied. The applications of the bubble in microfluidics can be further explored, such as
manipulations of micro/nano particles by means of an oscillating bubble, including the particle
trapping, sorting, migration or even patterning.

119
 Publication List

Publication List

XP Shang, XY Huang, C Yang, “The interactions between a gas bubble and a movable sphere in
an oscillatory pressure field,” (in preparation)

XP Shang, XY Huang, C Yang, “Vortex generation and control in a microfluidic chamber with
actuations,” Physics of Fluids, 28, 122001 (2016).

XP Shang, XY Huang, C Yang, “Bubble dynamics in a microfluidic chamber under low-

frequency actuation,” Microfluidics Nanofluidics, 20, 14 (2016).

XP Shang, XY Huang, C Yang, “Mixing enhancement by the vortex in a microfluidic mixer with
actuation,” Experimental Thermal and Fluid Science, 67, 57 (2015).

XP Shang, XG Cui, XY Huang, C Yang, “Vortex generation in a microfluidic chamber with

actuations,” Experiments in Fluids, 55, 1758 (2014).

XP Shang, XY Huang, C Yang, “Bubble translation at low-frequency actuation in a resonator-

shaped microfluidic chamber,” The 7th International Conference on Fluid Mechanics,
Qingdao/China (2015).

XP Shang, XY Huang, C Yang, “Mixing enhancement by the vortex in a microfluidic mixer with
actuation,” The 5th International Conference on Heat Transfer and Fluid Flow in Microscale,
Marseille/France (2014).

120
 References

References

[1] G. M. Whitesides, “The origins and the future of microfluidics,” Nature 442, 368 (2006).
[2] A. J. Mach, J. H. Kim, A. Arshi, S. C. Hur and D. D. Carlo, “Automated cellular sample preparation using
a Centrifuge-on-a-Chip,” Lab Chip 11, 2827 (2011).
[3] N. Nguyen and Z. Wu, “Micromixers – a review,” J. Micromech. Microeng. 15, R1 (2005).
[4] X. P. Shang, X. Huang and C. Yang, “Mixing enhancement by the vortex in a microfluidic mixer with actuation,”
Exp. Therm. Fluid Sci. 67, 57 (2015).
[5] W. Lauterborn and T. Kurz, “Physics of bubble oscillations,” Rep. Prog. Phys. 73, 106501 (2010).
[6] M. S. Plesset and P. Calif, “The dynamics of cavitation bubbles,” J. Appl. Mech. 16, 277 (1949).
[7] L. Crum and A. Eller, “Motion of bubbles in a stationary sound field,” J. Acoust. Soc. Am. 48, 181 (1970).
[8] A. A. Doinikov, “Translational motion of a spherical bubble in an acoustic standing wave of high intensity,”
Phys. Fluids 14, 1420 (2002).
[9] Y. Liu, K. Sugiyama, S. Takagi and Y. Matsumoto, “Surface instability of an encapsulated bubble induced by an
ultrasonic pressure wave,” J. Fluid Mech. 691, 315 (2012).
[10] Tandiono, S. W. Ohl, D. S. W. Ow, E. Klaseboer, V. V. Wong, R. Dumke and C. D. Ohl, “Sonochemistry and
sonoluminescence in microfluidics,” Proc. Natl. Acad. Sci. 108, 5996 (2011).
[11] M. J. Lighthill, An informal introduction to theoretical fluid mechanics (Oxford University Press, London, 1986)
p.68.
[12] T. V. Karman, Aerodynamics: selected topics in the light of their historical development (Cornel University
Press, New York, 1954) p.67.
[13] G. I. Taylor and A. E. Green, “Mechanism of the production of small eddies from large ones,” Proc. R. Soc.
Lond. A 158, 499 (1937).
[14] M. E. Brachet, M. Meneguzzi, H. Politano and P. L. Sulem, “The dynamics of freely decaying two-dimensional
turbulence,” J. Fluid Mech. 194, 333 (1988).
[15] A. Babiano, C. Basdevant, B. Legras and R. Sadourny, “Vorticity and passive-scalar dynamics in two-
dimensional turbulence,” J. Fluid Mech. 183, 379 (1987).
[16] G. Haller, “An objective definition of a vortex,” J. Fluid Mech. 525, 1 (2005).
[17] A. D. Strook, S. K. W. Dertinger, A. Ajdari, I. Mezic, H. A. Stone and G. M. Whitesides, “Chaotic mixer for
microchannels,” Science 295, 647 (2002).
[18] V. Mengeaud, J Josserand and H. H. Girault, “Mixing Processes in a Zigzag Microchannel: Finite Element
Simulations and Optical Study,” Anal. Chem. 74, 4279 (2002).
[19] X. Wang, J. Zhou and I. Papautsky, “Vortex-aided inertial microfluidic device for continuous particle
separation with high size-selectivity, efficiency, and purity,” Biomicrofluidics 7, 044119 (2013).
[20] M. S. N. Oliveira, F. T. Pinho and M. A. Alves, “Divergent streamlines and free vortices in Newtonian fluid
flows in microfluidic flow focusing devices,” J. Fluid Mech. 711, 171 (2012).

121
 References

[21] R. H. Liu, A. S. Stremler, K. V. Sharp, M. G. Olsen, J. G. Santiago, R. J. Adrian, H. Aref and D. J. Beebe,
“Passive Mixing in a Three-Dimensional Serpentine Microchannel,” J. Microelectromech. Syst. 9, 190 (2000).
[22] M. K. Singh, T. G. Kang, H. E. H. Meijer and P. D. Anderson, “The mapping method as a toolbox to analyze,
design, and optimize micromixers,” Microfluid. Nanofluid. 5, 313 (2008).
[23] T. G. Kang, M. K. Singh, T. H. Kwon and P. D. Anderson, “Chaotic mixing using periodic and aperiodic
sequences of mixing protocols in a micromixer,” Microfluid. Nanofluid. 4, 589 (2008).
[24] C. Simonnet and A. Groisman, “Chaotic mixing in a steady flow in a microchannel,” Phys. Rev. Lett. 94,
134501 (2005).
[25] M. Camesasca, M. Kaufman and I. M. Zloczower, “Staggered passive micromixers with fractal surface
patterning,” J. Micromech. Microeng. 16, 2298 (2006).
[26] T. M. Squires and M. Z. Bazant, “Induced-charge electro-osmosis,” J. Fluid Mech. 509, 217 (2004).
[27] Z. Wu and D. Li, “Micromixing using induced-charge electrokinetic flow,” Electrochimica Acta 53, 5827
(2008).
[28] H. Y. Wu and C. H. Liu, “A novel electrokinetic micromixer,” Sensors Actuat. A 118, 107 (2005).
[29] N. Sasaki, T. Kitamori and H. B. Kim, “AC electroosmotic micromixer for chemical processing in a
microchannel,” Lab chip 6, 550 (2006).
[30] S. H. Huang, S. K. Wang, H. S. Khoo and F. G. Tseng, “Ac electroosmotic generated in-plane microvortices for
stationary or continuous fluid mixing,” Sensors Actuat. B 125, 326 (2007).
[31] A. Kumar, S. Williams and S. T. Wereley, “Experiments on opto-electrically generated microfluidic vortices,”
Microfluid. Nanofluid. 6, 637 (2009).
[32] C. Park and S. T. Wereley, “Rapid generation and manipulation of microfluidic vortex flows induced by AC
electrokinetics with optical illumination,” Lab Chip 13, 1289 (2013).
[33] R. H. Liu, J. Wang, M. Z. Pindera, M. Athavale and P. Grodzinski, “Bubble-induced acoustic micromixing,”
Lab Chip 2, 151 (2002).
[34] D. Ahmed, X. Mao, J. Shi, B. K. Juluri and T. J. Huang, “A millisecond micromixer via single-bubble-based
acoustic streaming,” Lab Chip 9, 2737 (2009).
[35] C. Wang, B. Rallabandi and S. Hilgenfeldt, “Frequency dependence and frequency control of microbubble
streaming flows,” Phys. Fluids 25, 022002 (2013).
[36] B. Rallabandi, C. Wang and S. Hilgenfeldt, “Two-dimensional streaming flows driven by sessile
semicylindrical microbubbles,” J. Fluid Mech. 739, 57 (2014).
[37] M. Grumann, A. Geipel, L. Riegger, R. Zengerle and J. Ducree, “Batch-mode mixing on centrifugal
microfluidic platforms,” Lab Chip 5, 560 (2005).
[38] R. Calhoun, A. Yadav, P. Phelan, A. Vuppu, A. Garcia and M. Hayes, “Paramagnetic particles and mixing in
micro-scale flows,” Lab Chip 6, 247 (2006).
[39] T. Franke, L. Schmid, D. A. Weitz and A. Wixforth, “Magneto-mechanical mixing and manipulation of
picoliter volumes in vesicles,” Lab Chip 9, 2831 (2009).

122
 References

[40] S. J. Kim, F. Wang, M. A. Burns and K. Kurabayashi, “Temperature-Programmed Natural Convection for
Micromixing and Biochemical Reaction in a Single Microfluidic Chamber,” Anal. Chem. 81, 4510 (2009).
[41] M. Geissler, B. Voisin and T. Veres, “Air stream-mediated vortex agitation of microlitre entities on a fluidic
chip,” Lab Chip 11, 1717 (2011).
[42] R. E. Apfel, “Acoustic cavitation inception,” Ultrasonics 22, 167 (1984).
[43] D. H. Trevena, “Cavitation and the generation of tension in liquids,” J. Phys. D: Appl. Phys. 17, 2139 (1984).
[44] D. E. Young, “Skins of varying permeability: a stabilization mechanism for gas cavitation nuclei,” J. Acoust.
Soc. Am. 65, 1429 (1979).
[45] Trevena, Cavitation and tension in liquids (Hilger, Bristol, 1987) p.15.
[46] F. Caupin, “Liquid-vapor interface, cavitation, and the phase diagram of water,” Phys. Rev. E 71, 051605
(2005).
[47] E. Herbert, S. Balibar and F. Caupin, “Cavitation pressure in water,” Phys. Rev. E 74, 041603 (2006).
[48] I. Akhatov and N. Gumerov, “The role of surface tension in stable single bubble sonoluminescence,” Phys. Rev.
Lett. 78, 227 (1997).
[49] R. E. Apfel, “Acoustic cavitation prediction,” J. Acoust. Soc. Am. 69, 1624 (1981).
[50] F. R. Young, Cavitation (Imperial Colledge Press, London, 1989) p. 44.
[51] D. Y. Hsieh and M. S. Plesset, “Theory of rectified diffusion of mass into gas bubbles,” J. Acoust. Soc. Am. 33,
206 (1961).
[52] E. A. Neppiras, “Acoustic cavitation thresholds and cyclic processes,” Ultrasonics 18, 201 (1980).
[53] C. E. Brennen, Cavitation and bubble dynamics (Oxford University Press, London, 1995) p.89.
[54] D. F. Gaitan, L. A. Crum, C. C. Church and R. A. Roy, “Sonoluminescence and bubble dynamics for a single,
stable, cavitation bubble,” J. Acoust. Soc. Am. 91, 3166 (1992).
[55] H. Poritsky, in Proceedings of the First U.S. National Congress on Applied Mechanics, New York, U.S.A.,
1952, edited by E. Sternberg, p. 813-821.
[56] C. C. Church, “A theoretical study of cavitation generated by an extracorporeal shock wave lithotripter,” J.
Acoust. Soc. Am. 86, 215 (1989).
[57] J. B. Keller and M. Miksis, “Bubble oscillation of large amplitude,” J. Acoust. Soc. Am. 68, 628 (1980).
[58] A. Prosperetti and A. Lezzi, “Bubble dynamics in a compressible liquid: part 1. First-order theory,” J. Fluid
Mech. 168, 457 (1986); A. Lezzi and A. Prosperetti, “Bubble dynamics in a compressible liquid: part 2. Second-
order theory,” J. Fluid Mech. 185, 289 (1987).
[59] O. Louisnard and J. G. Garcia, Acoustic cavitation in Ultrasound Technologies for food and bioprocessing
(Springer Science, 2011) p.21.
[60] Z. C. Feng, “Nonlinear bubble dynamics,” Ann. Rev. Fluid Mech. 29, 201 (1997).
[61] L. L. Vignoli, A. L. F. Barros, R. C. A. Thome, A. L. Nogueira, R. C. Paschoal and H. Rodrigues, “Modeling
the dynamics of single-bubble sonoluminescence,” Eur. J. Phys. 34, 679 (2013).
[62] S. Hilgenfeldt, M. P. Brenner, S. Grossmann and D. Lohse, “Analysis of Rayleigh-Plesset dynamics for
sonoluminescing bubbles,” J. Fluid Mech. 365, 171 (1998).

123
 References

[63] W. Lauterborn and R. Mettin, in Proceedings of the NATO Advanced Study Institute, Washington, U.S.A., 18-
29 August 1997. edited by L. A. Crum, T. J. Mason, J. L. Reisse and K. S. Suslick, pp. 63-72.
[64] K. W. Commander and A. Prosperetti, “Linear pressure waves in bubbly liquids: comparison between theory
and experiments,” J. Acoust. Soc. Am. 85, 732 (1989).
[65] D. Z. Zhang and A. Prosperetti, “Ensemble phase-averaged equations for bubbly flows,” Phys. Fluids 6, 2956
(1994).
[66] B. Gompf and R. Pecha, “Mie scattering from a sonoluminescing bubble with high spatial and temporal
resolution,” Phys. Rev. E 61, 5253 (2000).
[67] W. Lauterborn, T. Kurz, R. Mettin and C. D. Ohl, “Experimental and theoretical bubble dynamics,” Adv. Chem.
Phys. 110, 316 (1999).
[68] D. Kroninger, Ph.D. thesis, University of Gottingen, 2008.
[69] T. G. Leighton, A. J. Walton and M. J. W. Pickworth, “Primary Bjerknes force,” Eur. J. Phys. 11, 47 (1990).
[70] J. Magnaudet and D. Legendre, “The viscous drag force on a spherical bubble with a time-dependent radius,”
Phys. Fluids 10, 550 (1998).
[71] D. Krefting, J. O. Toilliez, A. J. Szeri, R. Mettin and W. Lauterborn, “Translation of bubbles subject to weak
acoustic forcing and error in decoupling from volume oscillation,” J. Acoust. Soc. Am. 120, 670 (2006).
[72] A. J. Reddy and A. J. Szeri, “Coupled dynamics of translation and collapse of acoustically driven
microbubbles,” J. Acoust. Soc. Am. 112, 1346 (2002).
[73] A. A. Doinikov, “Translational motion of two interacting bubbles in a strong acoustic field,” Phys. Rev. E 64,
026301 (2001).
[74] R. Mettin, Oscillations, waves, and interactions (University of Gottingen Press, Gottingen, 2007) p.187.
[75] J. R. Blake and D. C. Gibson, “Cavitation bubbles near boundaries,” Ann. Rev. Fluid Mech. 19, 99 (1987).
[76] R. H. Cole, Underwater explosions (Princeton University Press, Clinton, 1948) p.327.
[77] K. Sato, Y. Tomita and A. Shima, “Numerical analysis of a gas bubble near a rigid boundary in an oscillatory
pressure field,” J. Acoust. Soc. Am. 95, 2416 (1994).
[78] D. Kroninger, K. Kohler, T. Kurz and W. Lauterborn, “Particle tracking velocimetry of the flow field around a
collapsing cavitation bubble,” Exp. Fluids 48, 395 (2010).
[79] X. Xi, F. Cegla, R. Mettin, F. Holsteyns and A. Lippert, “Collective bubble dynamics near a surface in a weak
acoustic standing wave field,” J. Acoust. Soc. Am.132, 37 (2012).
[80] X. Xi, F. Cegla, M. Lowe, A. Thiemann, T. Nowak, R. Mettin, F. Holsteyns and A. Lippert, “Study on the
bubble transport mechanism in an acoustic standing wave field,” Ultrasonics 51, 1014 (2011).
[81] E. Klaseboer and B. C. Khoo, “An oscillating bubble near an elastic material,” J. Appl. Phys. 96, 5808 (2004).
[82] P. B. Robinson, J. R. Blake, T. Kodama, T. Shima and Y. Tomita, “Interaction of cavitation bubbles with a free
surface,” J. Appl. Phys. 89, 8225 (2001).
[83] Y. Tomita, P. B. Robinson, R. P. Tong and J. R. Blake, “Growth and collapse of cavitation bubbles near a
curved rigid boundary,” J. Fluid Mech. 466, 259 (2002).

124
 References

[84] P. Garstecki, I. Gitlin, W. Diluzio, G. M. Whitesides, E. Kumacheva and H. A. Stone, “Formation of

monodisperse bubbles in a microfluidic flow-focusing device,” Appl. Phys. Lett. 85, 2649 (2004).
[85] M. Shirota, T. Sanada, A. Sato and M. Watanabe, “Formation of a submillimeter bubble from an orifice using
pulsed acoustic pressure waves in gas phase,” Phys. Fluids 20, 043301 (2008).
[86] D. Rabaud, P. Thibault, J. P. Raven, O. Hugon, E. Lacot, and P. Marmottant, “Manipulation of confined
bubbles in a thin microchannel: Drag and acoustic Bjerknes forces,” Phys. Fluids 23, 042003 (2011).
[87] T. Segers and M. Versluis, “Acoustic bubble sorting for ultrasound contrast agent enrichment,” Lab Chip 14,
1705 (2014).
[88] E. Zwaan, S. L. Gac, K. Tsuji and C. D. Ohl, “Controlled Cavitation in Microfluidic Systems,” Phys. Rev. Lett.
98, 254501 (2007).
[89] P. A. Q. Su, K. Y. Lim and C. D. Ohl, “Cavitation bubble dynamics in microfluidic gaps of variable height,”
Phys. Rev. E 80, 047301 (2009).
[90] A. Patrascioiu, J. M. Fernandez-Pradas, A. Palla-Papavlu, J. L. Morenza and P. Serra, “Laser-generated liquid
microjets: correlation between bubble dynamics and liquid ejection,” Microfluid. Nanofluid. 16, 55 (2014).
[91] X. P. Shang, X. G. Cui, X. Y. Huang and C. Yang, “Vortex generation in a microfluidic chamber with
actuations,” Exp. Fluids 55, 1758 (2014).
[92] J. Veilleux and S. Coulombe, “A total internal reflection fluorescent microscopy study of mass diffusion
enhancement in water-based alumina nanofluids,” J. Appl. Phys. 108, 104316 (2010).
[93] S. Wang, X. Huang and C. Yang, “Mixing enhancement for high viscous fluids in a microfluidic chamber,” Lab
Chip 11, 2081 (2011).
[94] J. Lighthill, “Acoustic streaming,” J. Sound Vib. 61, 391 (1978).
[95] N. Riley, “Steady streaming,” Ann. Rev. Fluid Mech. 33, 43 (2001).
[96] F. Durst, A. Melling and J. H. Whitelaw, “Low Reynolds number flow over a plane symmetric sudden
expansion,” J. Fluid Mech. 64, 111 (1974).
[97] M. Shapira, D. Degani and D. Weihs, ‘‘Stability and existence of multiple solutions for viscous flow in
suddenly enlarged channels,’’ Comput. Fluids 18, 239 (1990).
[98] J. Mizushima and Y. Shiotani, “Transitions and instabilities of flow in a symmetric channel with a suddenly
expanded and contracted part,” J. Fluid Mech. 434, 355 (2001).
[99] A. Quaini, R. Glowinski and S. Canic, “Symmetry breaking and Hopf bifurcation for incompressible viscous
flow in a contraction-expansion channel,” Int. J. Comput. Fluid Dyn. 30, 7 (2016).
[100] A. Revuelta, “On the two-dimensional flow in a sudden expansion with large expansion ratios,” Phys. Fluids
17, 028102 (2005).
[101] M. Dufresne, J. Vazquez, A. Terfous, A. Ghenaim and J. B. Poulet, “Experimental investigation and CFD
modelling of flow, sedimentation, and solids separation in a combined sewer detention tank,” Comput. Fluids 38,
1042 (2009).
[102] I. Sobey, “Observation of waves during oscillatory channel flow,” J. Fluid Mech. 151, 395 (1985).

125
 References

[103] E. Roberts and M. Mackley, “The development of asymmetry and period doubling for oscillatory flow in
baffled channels,” J. Fluid Mech. 328, 19 (1996).
[104] X. P. Shang, X. Y. Huang and C. Yang, “Vortex generation and control in a microfluidic chamber with
actuations,” Phys. Fluids 28, 122001 (2016).
[105] W. Lauterborn and T. Kurz, “Physics of bubble oscillations,” Rep. Prog. Phys. 73, 106501 (2010).
[106] M. Strasberg, “The pulsation frequency of nonspherical gas bubbles in liquids,” J. Acoust. Soc. Am. 25, 536
(1953).
[107] B. F. Espinoza, R. Zenit and D. Legendre, “The effect of confinement on the motion of a single clean bubble,”
J. Fluid Mech. 616, 419 (2008).
[108] X. P. Shang, X. Y. Huang and C. Yang, “Bubble dynamics in a microfluidic chamber under low-frequency
actuation,” Microfluid. Nanofluid. 20, 14 (2016).
[109] T. G. Leighton, The acoustic bubble (Academic Press, San Diego, CA, 1994) p.72.
[110] L. D. Landau and E. M. Lifshitz, Fluid Mechanics (Pergamon, Oxford, 1986).
[111] B. V. Levich, Physicochemical hydrodynamics (Prentice-Hall, Englewood Cliffs, NJ, 1962).
[112] A. Prosperetti, “Thermal effects and damping mechanisms in the forced radial oscillations of gas bubbles in
liquids,” J. Acoust. Soc. Am. 61, 17 (1977).

126