Hydrodynamics and Mass Transfer of Oscillating Gas-Liquid

Flow in Ultrasonic Microreactors

Zhengya Dong, Chaoqun Yao, and Yuchao Zhang
Dalian National Laboratory for Clean Energy, Dalian Institute of Chemical Physics, Chinese Academy of
Sciences, Dalian 116023, China
University of Chinese Academy of Sciences, Beijing 100049, China
Guangwen Chen and Quan Yuan
Dalian National Laboratory for Clean Energy, Dalian Institute of Chemical Physics, Chinese Academy of
Sciences, Dalian 116023, China
Jie Xu
Applied Acoustics Institute, Shaanxi Normal University, Xian, Shaanxi 710062, China

DOI 10.1002/aic.15091
Published online November 17, 2015 in Wiley Online Library (wileyonlinelibrary.com)

Ultrasonic microreactors were used to intensify gas-liquid mass-transfer process and study the intensification mecha-
nism. Fierce surface wave oscillation with different modes was excited on the bubble. It was found that for slug bubbles
confined in smaller microchannel, surface wave oscillations require more ultrasound energy to excite due to the confine-
ment effect. Cavitation microstreaming with two toroidal vortices was observed near the oscillating bubble by a streak
photography experiment. Surface wave oscillation at the gas-liquid interface increases the specific surface area, while
cavitation microstreaming accelerates the interface renewal and thus improves the individual mass-transfer coefficient.
With these two reasons, the overall mass-transfer coefficient was enhanced by 3–20 times under ultrasonication. As for
gas-liquid flow hydrodynamics, ultrasound oscillation disturbs the bubble formation process and changes the initial bub-
C 2015 American Institute of Chemical Engineers AIChE J, 62: 1294–1307, 2016
ble length and pressure drop. V
Keywords: ultrasound, microfluidic, Taylor flow, sonochemistry, acoustic cavitation

Introduction are often designed in gas-liquid microreactor to induce local

vortices or secondary convective flow,16–18 and further
As one of the most promising applications of microreaction
improve the mass-transfer performance.11,13,19,20 Besides these
technology, gas-liquid microreactor has great advantages for
passive methods, active methods using external field17,18 can
various gas-liquid systems, such as gas absorption,1,2 catalytic
also be used to enhance mass-transfer rate in microreactor. We
hydrogenation,3,4 liquid phase oxidation,5,6 and direct fluorina-
proposed previously that ultrasound can be introduced into
tion.7,8 Concerning the design and operation of gas-liquid
microreactor to further intensify the gas-liquid mass-transfer
microreactor, many researches have been devoted to the
process.21
hydrodynamics and mass transfer of the gas-liquid flow in In fact, ultrasound has been widely applied in conventional
microreactors.9–12 It has been confirmed that the overall equipment to intensify multiphase mass-transfer/chemical
volumetric mass-transfer coefficient (kLa) in gas-liquid micro- reaction processes.21–25 Several literatures have studied the
reactor is much greater than that in conventional equip- effect of ultrasound on gas-liquid mass-transfer process.26–30
ment.1,9,10,13 The high kLa is mainly caused by the large Early in 1971, Cadwell and Fogler26 found that the rate of
specific surface area (a) in microreactor, while the individual CO2 absorption by glycerol was significantly enhanced in an
mass-transfer coefficient (kL) itself is still comparable to some ultrasonic batch reactor, due to the ultrasonically induced con-
conventional equipment (such as static mixers and tubular vective streaming in the liquid. Kumar et al.28 compared the
reactors).1,9,13 This is because that laminar flow prevails in overall volumetric gas-liquid mass-transfer coefficient in an
microfluidics, where the mass-transfer (and mixing) process is ultrasonic horn, an ultrasonic bath, and a mechanically agi-
relatively slow compared with the turbulent flow in conven- tated contactor (gas was sparged into the reactors with a
tional equipment. 14–16 Therefore, curved and bent structures sparger). It was found that mass-transfer enhancement in the
ultrasonic horn (about 50–110%) is higher than that in ultra-
Additional Supporting Information may be found in the online version of this
article. sonic bath, but both are much lower than that in mechanically
Correspondence concerning this article should be addressed to G. Chen at agitated contactor. They allotted the mass-transfer enhance-
[email protected].
ment to the reduction of gas bubble size by ultrasound-created
C 2015 American Institute of Chemical Engineers
V turbulence, which is less effective than the convective motion

1294 April 2016 Vol. 62, No. 4 AIChE Journal

in mechanically agitated contactor. Herran et al.30 also studied alloy plate. As the four corners of microreactor plate on the pre-
gas-liquid oxygen transfer process in sonicated bubble col- vious USMR undergo flexural vibration and lead to dramatic
umns with different reactor diameter and liquid height. They variation of the vibration displacement,21 we cut them off in the
found that the effects of ultrasound were only significant at present USMR (as shown in Figure 2). The vibration displace-
very high power densities (above 0.4 W/mL), in which only a ment distribution simulated by ANSYS software in Figures 1c,
10–30% increment of the mass-transfer coefficient was d shows a much even acoustic field in the microreactor.
obtained. In these literatures, the effect of ultrasonication on
gas-liquid mass-transfer enhancement has been confirmed; the Experimental setup
macroscopic mechanism associated with ultrasound-induced The USMR was excited by a RF power generator (AG 1016,
turbulence and bubble size reduction has also been proposed. T&C Power Conversion) including an impedance matcher that
But the detailed mechanism—the microscopic scenario of matches the impedance difference between the generator (50 X)
ultrasound-bubble interaction has not been clarified. This is and the USMR (10–35 X). Pure CO2 and deionized water flow
due to the difficulties of observing individual bubble behaviors were put into the microchannels on the microreactor plate,
in the complex two phase flow field and also the uneven which were covered by a transparent polycarbonate film with
acoustic field.31–33 The lack of understanding the intensifica- epoxy glue. The hydrodynamics and mass-transfer behaviors of
tion mechanism would lead to low and even uncontrollable (or the gas-liquid flow were monitored by a high-speed camera
unpredictable) ultrasonic enhancement results,24,34 especially (Phantom M310, Vision Research). The camera was either con-
when compared with mechanically agitated reactor. nected to a macro lens (Macro AF90, Tamron) to record large
Ultrasonic gas-liquid microreactor provides an opportunity to view photo or to a stereo microscopy (SZX16, Olympus) to
solve these problems.21 Compared with conventional reactors, capture local details. To study the influence of channel size on
strong and uniform ultrasonic field can be realized in microreac- the gas-liquid flow hydrodynamics and mass transfer, three
tor due to its much smaller volume.35,36 Stable and regular gas- USMRs (labeled as MR1, MR0.5, and MR0.25) with different
liquid flow can be easily formed in microchannel,9,10,37 where channel size were fabricated, as shown in Figure 2 and Table 1.
the movement of the bubbles in the microchannel is highly con- Typical gas and liquid flow rates during the experiments are
trolled and can be directly observed online. With these advan- also displayed in Table 1, which are set to keep the superficial
tages, ultrasonic microreactor (USMR) provides an ideal gas/liquid velocity in the three USMRs the same (jL 5 0.05 m/s,
environment to study the ultrasound-bubble interaction phe- jG 5 0.1 m/s). To alleviate the thermal effect of ultrasound, a
nomena and control the intensification process.33,36,38,39 In this small air fan was used to cool the system. Moreover, the ultra-
work, USMRs were built to intensify the gas-liquid mass trans- sonic generator was turned on for only a few minutes at each
fer and to study the microscopic mechanism. First, the operating condition, during which the temperature rise was kept
ultrasound-induced bubble oscillation behaviors and the associ- below 38C. Such low-temperature rise has little effect on our
ated cavitation microstreaming were characterized by a high- experimental results. Other details of the experimental setup
speed camera. Then, the influence of ultrasonication on gas- can be found in our previous paper.21
liquid flow hydrodynamics, such as bubble formation processes
Characterization of the experimental system
and pressure drop were investigated. Finally, the effect of ultra-
sound power, microchannel size, and flow rate on the mass- After connection of the whole experiment system, some
transfer coefficient enhancement were studied by an online important properties of the USMRs were characterized. The
measurement method. The mass-transfer enhancement mecha- first goes the admittance curve, which was measured by an
nism was further quantitatively analyzed. All the experiments impedance analyzer (PV70A, Beijing Band Era), as shown in
are based on gas-liquid Taylor (or slug) flow pattern, which has Figure 3a. For each microreactor, only one resonance peak
been considered as the most promising flow regime for applica- was observed at the measured frequency range, suggesting
tions and has been widely studied.9–11,40 that the pure longitudinal resonator described above was
obtained. As the resonance frequency of the USMR is mainly
determined by its overall size (the microchannel size differ-
Experimental Section
ence has little effect), the resonance peaks of the three USMRs
Building the USMR are quite close, which are all around the designed frequency
The design and fabrication process of the USMR has been (20 kHz). The quality factor of the USMRs calculated from
specifically addressed in our previous paper.21 The USMR is the admittance curves is around 500–1000. The high quality
built by directly coupling a microreactor plate with a factor indicates high electroacoustic efficiency.41,43
Langevin-type ultrasonic transducer, which is composed of Another important property is the acoustic power density
piezoelectric ceramic pieces clamped between the front mass (Pth) in the miroreactor. It is generally measured by a calori-
and the back mass, as shown in Figure 1a. The whole USMR metric method, which records the temperature change of the
is designed as a longitudinal half wavelength resonator, guided fluid in the ultrasound-irradiating reactors.23,44,45 The acoustic
by a simplified electro-mechanical equivalent analysis power density is determined from the initial temperature slope

method.21,41 The antinode plane with highest sound intensity DT 
Pth 5qcp (1)
is located at the microreactor, while the node plane with high- Dt 
t50
est stress located at the piezoelectric pieces, as shown in Fig-
ure 1a. This novel design not only realizes highest ultrasound where q is the density of the liquid and cp is its specific heat.
power density in the microreactor but also maximizes the To record the temperature change in our USMRs, a 0.5-mm
energy efficiency and lifespan of the ultrasonic transducer.41,42 diameter T-type thermocouple was inserted into the outlet of
The fabricated USMR with a maximum power of 100 W and a the microchannel, which is first filled with deionized water. As
resonance frequency around 20 kHz is shown in Figure 1b. shown in Figure 4a, the measured acoustic power density
The microreactor is made of a 3-mm thick LY12 aluminum increases linearly with the input power and the channel size

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1295
Figure 1. (a) Design diagram of the USMR. The dashed lines represent the half wavelength standing wave resona-
tor. (b) Photo of the fabricated USMR. The scale bar in the figure is 10 mm. (c, d) The simulated vibration
displacement distribution of the whole USMR (c) and the upper surface (d) at a resonance frequency of
20.15 kHz.
[Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

has little effect. It indicates that power densities in the three leading to less regular geometric deformations and complex
USMRs can be considered the same for a particular input patterns. With further increasing of the acoustic pressure, the
power. The power density is from 0.07 to 0.3 W/mL at power bubble oscillations become highly distorted and disordered,
10–50 W, which is in the same range as that in conventional which is also called chaotic surface mode.52 The large surface
ultrasonic reactors (usually between 0.05 and 0.6 W/mL).27–30 undulations even lead to the formation of a liquid jet penetrat-
ing deep into the bubble, during which a droplet is sometimes
Bubble Oscillation and Cavitation Microstreaming pinched-off from the jet tip.50,53,54 When the acoustic pressure
Characteristics of the bubble oscillation rise to about 100 kPa, the vigorous oscillations would lead to
collapse of the bubble, generating several high energy phe-
Bubble oscillation behaviors under ultrasonic field have nomena, such as shock wave, light emission, and radical for-
been systematically studied in the domain of acoustic cavita- mation.22,34,46,47 These pulsations resulting in the breakup of
tion.31,34,46,47 A bubble in liquid can be viewed as an oscilla- the bubble are often called transient cavitation, while the for-
tory system, which can be driven by acoustic pressure mer oscillations persisting over many cycles are classified as
fluctuations. This bubble oscillator could undergo a number of stable cavitation.34 It is worth mentioning that all surface
different oscillations.48–51 At very low acoustic pressure, the wave modes in stable cavitation are capillary wave, for which
bubble oscillates radically with symmetrical pulsation of the the restoring force is provided by the interfacial tension. There
bubble volume, which is often called breathing (or volume) is a dispersion relation between the capillary wave’s wave-
oscillation. When the acoustic pressure exceeds a threshold,
length and frequency53,55,56
slight nonradial modes set in, which develop into shape oscil-
 1
lation with many surface wave modes. The mode with the low- 2pr 3
est threshold is the Faraday wave, which appears as regular k5 2 (2)
f q
geometric distortion and oscillates at a frequency equal to one
half of the driving frequency. At higher acoustic pressure, The slug bubble does not collapse during the fierce oscilla-
multiple surface wave modes are excited and superimposed, tion in our experiment, implying that it undergoes stable

1296 DOI 10.1002/aic Published on behalf of the AIChE April 2016 Vol. 62, No. 4 AIChE Journal
 Table 1. Width (W), Depth (D) of the Main Channel, Typical
Gas (QG), and Liquid (QL) Flow Rate in the USMRs
USMRs W (mm) D (mm) QL (mL/min) QG (mL/min)
MR1 1 1 3 6
MR0.5 0.5 0.5 0.75 1.5
MR0.25 0.5 0.25 0.38 0.75

tion. A smaller space between the obstacles (channel wall in

this work) and the bubble surface leads to a larger viscous
resistance, and thus, more ultrasound energy is needed to
excite the surface wave oscillation.
In addition to the overall behaviors, some detailed charac-
teristics of the surface wave oscillations could also be quanti-
fied. The regular surface wave oscillation at relatively low
ultrasound power was recorded by the camera at a frame rate
of 80,000 fps, as shown in Figure 6. The oscillation shows a
clear period of 100 ls. For example, the wave peak at the cen-
Figure 2. Structure of the microchannel on the micro- ter of the bubble tip first retracts from 0 to 25 ls, then turns
reactor plate (MR0.5). into a valley from 25 to 75 ls, and finally returns back to the
The size of the main channel is displayed in Table 1. same peak from 75 to 100 ls. It indicates that the surface
For the three USMRs, the size of the outlet channel wave oscillates periodically at half the driving frequency
(1 3 1 mm2) and the width of inlet channel (0.5 mm)
are the same. The depth of the inlet channel is the same
with the main channel. Small stainless steel capillaries
were inserted into the inlet and outlet channels (sealed
by epoxy glue), and then connected to the input/output
flow tubes. [Color figure can be viewed in the online
issue, which is available at wileyonlinelibrary.com.]

cavitation. The overall bubble oscillation behaviors agree well

with the above description. As shown in Figures 4 and 5, with
the increasing of ultrasound power, the bubble oscillation
changed from volume pulsation to regular surface wave oscil-
lation, and finally to chaotic surface distortion. Due to the light
reflection on the bubble surface, the liquid jet formed by the
vigorous surface undulations was difficult to be observed.
However, the droplet pinched-off from the liquid jet was cap-
tured flying in the slug bubble, as shown in the right of
Figure 4. The moving velocity of the droplet was measured to
be 1 6 0.5 m/s, which is on the same order of magnitude as
that measured in Prabowo’s experiment50 (about 2.5 m/s). As
the bubbles in our experiment have a slug shape and are con-
fined in the microchannel, which is quite different from the
sphere or semisphere free bubbles discussed in most of the lit-
erature,46,48–51 some additional phenomena were also found.
As shown in Figures 4 and 5, with the increasing of ultrasound
power, the range of the bubble surface oscillation expands
gradually, which starts from the free bubble tip, then develops
into the bubble surface at the channel corner, and finally
invades the bubble waist confined by the channel wall. It indi-
cates that for confined bubble surface, higher acoustic inten-
sity is required to excite the surface wave than the free bubble
surface. This confinement effect can be further verified by
comparing the oscillation behaviors at different USMRs, as
shown in Figure 5. It can be seen that the surface wave on the
bubble starts at different threshold power in the three USMRs,
which are around 6, 21, 28 W in MR1, MR0.5, and MR0.25,
respectively. The smaller the channel size, the higher the Figure 3. (a) Measured admittance curves (Y) of the
threshold power. It means that for bubbles confined in smaller three USMRs. (b) The measured acoustic
channel, more ultrasound energy is needed to excite the sur- power density (Pth) in the fluid of the three
face wave oscillations. This confinement effect could be USMRs.
explained by the fact that obstacles near the bubble surface [Color figure can be viewed in the online issue, which is
would add more viscous resistance34,46 to the bubble oscilla- available at wileyonlinelibrary.com.]

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1297
 amplitudes in the three USMRs increase distinctly with the
ultrasound power. The increase of the oscillation amplitude
for the three USMRs begins at different ultrasound powers,
which are the very threshold powers observed in Figure 5. The
maximum oscillation amplitude reaches 60–70 lm at ultra-
sound power 40–50 W.

Cavitation microstreaming around the slug bubble

Acoustic streaming, fluid flow generated by the attenuation
of acoustic wave, is a well-known phenomenon within the
acoustics community.57–59 Among the different types of
acoustic streaming, cavitation microstreaming—the steady
flow formed by the viscous dissipation of acoustic energy near
an oscillating bubble, attracts the most attention due to its
large streaming velocity.60–62 The magnitude of the velocity
ranges from mm/s to m/s, depending on the bubble’s shape,
Figure 4. Bubble oscillation behavior at different ultra- environment, and acoustic intensity.59,62 The flow pattern for
sound power in MR1. cavitation microstreaming also differs under different experi-
mental circumstances. The flow pattern associated with hemi-
The inset picture on the right shows a droplet moving in
the bubble. All scale bars in the figure are 500 lm. spherical bubble resting on a solid boundary is usually two
[Color figure can be viewed in the online issue, which is toroidal vortices near the bubble surface.39,52,59 This kind of
available at wileyonlinelibrary.com.] cavitation microstreaming has been widely applied in micro-
fluidics to manipulate particles or cells,39,60 pump fluid,63,64
(around 20 kHz, with a period of 50 ls), which is a key feature and enhance mixing.59,61,65,66 It can be expected that this cavi-
of the Faraday wave. The wavelength of the surface wave tation microstreaming is also present near the oscillating bub-
could also be estimated by measuring the distance between bles in our experiment, although the bubbles are moving and
two peaks in the pictures in Figure 6. The measured result, have a quite different shape and environment. Streak photog-
163 6 10 lm, agrees well with the value calculated by Eq. 2 raphy experiment60–62 was conducted to verify this presump-
(165 lm). These results indicate that the regular surface wave tion. Polystyrene microparticles of diameter 9.0–9.9 lm
on the slug bubble is Faraday capillary wave. With the (Aladdin) were suspended into the water as tracers. Streak
increase of ultrasound power, different surface wave modes photographs that approximately reflect the streamlines of the
are excited and coupled with each other, leading to the losing flow field were obtained by superposing about 120 successive
of the clear periodical and regular shape oscillation behavior. images (captured at a frame rate of 3000 fps), as shown in
Under this circumstance, only the oscillation amplitude can be Figure 8. The superposition of images was conducted by Star-
approximately measured, as shown in Figure 7. The oscillation trails,67 a free software which is developed to get star trail

Figure 5. Bubble oscillation behavior at different ultrasound power in MR1 (the first row), MR0.5 (the second row),
and MR0.25 (the third row).
The number labeled at the bottom of each picture is the ultrasound power (W). All scale bars in the figure are 250 lm. [Color fig-
ure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

1298 DOI 10.1002/aic Published on behalf of the AIChE April 2016 Vol. 62, No. 4 AIChE Journal
Figure 6. (a) Surface wave oscillation at different time on the bubble tip. The number labeled at the bottom of
each picture is the time (ls). (b) A sine wave with a period of 100 ls, which is used to illustrate the peri-
odically oscillating behaviors of the surface wave in Figure 6a.
The pictures were taken at a frame rate of 80,000 fps in MR0.5, at power of 25 W. The scale bar in the figure is 200 lm. [Color
figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

images. It is worth noting that the streamlines in the streak significantly intensify the mass-transfer process at the gas-
photographs are all relative to the bubble’s motion. liquid interface.
The streamlines of the flow field near the bubble is shown The cavitation microstreaming was further quantitatively
in Figure 8. When the ultrasound is not present, two recircula- analyzed by measuring the particle velocity near the bubble.
tion loops were observed in the liquid slug, as shown in Figure The absolute velocity in the flow direction is shown in Figure
8a (see also video S1 in the Supporting Information). The 9. Without the influence of ultrasound, the velocity is almost
recirculation flow pattern agrees well with that reported in lit- the same along the center line of the liquid slug. At power of
erature.9–11,19 When ultrasound was turned on, the fierce bub- 30 W, the velocity decreases distinctly when getting close to
ble oscillation generates two additional streaming vortices in the bubble surface. This regular reduction of velocity is a good
the flow field, as shown in Figure 8b (see also video S2 in the reflection of the streaming pattern shown in Figure 8b, in
Supporting Information). The streaming vortices are similar to which the fluid near the bubble tip is pulled toward the bubble
the flow pattern around the resting hemispherical bubble as by the two vortices. When ultrasound power rises to 40 W, the
described above. With the increasing of ultrasound power, the velocity points scatter around the points at power of 0 W, due
two vortices get faster and larger. At power of 40 W, the cavi- to the frequently changing and direction-reversing of the
tation microstreaming becomes complex and volatile, leading streaming pattern. The discrepancy of the velocity increases
to relatively discontinuous particle streaks in Figure 8c (see when approaching to the bubble tip, indicating faster stream-
also video S3 in the Supporting Information). As the flow pat- ing near the bubble surface. The highest velocity reaches
tern is always changing and the flow direction reverses fre- 0.03–0.04 m/s at power of 30–40W. The range of the acoustic
quently, the blue lines in Figure 8c just illustrates a typical streaming extends to about 0.4 mm from the bubble surface. It
situation. Such vigorous and dynamic streaming would inter- is worth pointing out that the actual cavitation microstreaming
act with the recirculation motion within the liquid slug, and is a complex three-dimensional flow. Further research is

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1299
 caused by the fierce surface oscillations on the thin emerging
bubble. The changing of breakup point would significantly
influence the bubble formation process. Compared with
breakup point near the inlet, breakup downstream the main
channel results in a shorter expansion step, as the expansion
volume needed for the bubble tip to block the main channel is
smaller. This leads to smaller bubble length at lower ultra-
sound power, as you can see in Figure 11a. In MR0.5, as the
breakup point is a little downstream from the inlet, the effect
of rupture step accelerating and breakup point moving com-
pete with each other, leading to relatively small variation
extent of the bubble length with ultrasound power. As the for-
mation of a bubble is always accompanied with the formation
of a liquid slug, the effect of ultrasonication on liquid slug
length is similar with that on the bubble length, as shown in
Figure 11b. In all the three USMRs, the deviation of the

Figure 7. Measured oscillation amplitudes at different

ultrasound power in the three USMRs.
The oscillation amplitude at each condition was deter-
mined by measuring the height of the peaks. To ensure
accuracy of the measurement, the chosen peaks should
be right at the circumference of the bubble tip. Each
point was measured for at least four times. [Color figure
can be viewed in the online issue, which is available at
wileyonlinelibrary.com.]

needed, especially using micro-PIV (Particle Image

Velocimetry) and confocal microscopy to characterize the
acoustic streaming flow field.

Hydrodynamics of the Gas-Liquid Flow

Bubble formation at the inlet
The hydrodynamic behaviors of gas-liquid flow in microchan-
nel depend highly on the bubble formation process at the
inlet.2,68–70 A typical bubble formation process consists of an
expansion step and a rupture step. First, the emerging bubble
expands axially and radically until its tip blocks the main chan-
nel. Then, under the action of shear stress, gas dynamic pressure
and liquid pressure difference, a neck was formed on the gas-
liquid interface, which eventually leads to breakup of the emerg-
ing bubble.68,69 According to the dominant forces involved in the
breakup process, the formation mechanism is generally classified
into squeezing regime at very low Capillary number (Ca < 0.01)
and shearing regime at high Capillary number.69,70 As the Capil-
lary number in our experiment is 0.0021, bubble formation obeys
the squeezing mechanism. The breakup of the slug bubble is con-
trolled by the liquid-phase pressure drop across the emerging
bubble resulting from its blockage of the liquid flow path, which
can be seen in some pictures in Figure 10.
The influence of ultrasound on the bubble formation process
differs in the three USMRs. In MR1, the breakup of the Figure 8. Streak photographs showing the streamlines
emerging bubble occurs near the inlet, as shown in the first of the flow field around the slug bubble at
column of Figure 10. When ultrasound is turned on, acoustic ultrasound power of 0 W (a), 30 W (b), and 40
pressure oscillation in the liquid and surface oscillation on the W (c) in MR0.5.
interface accelerate the rupture step, which reduces the bubble
The red lines are drawn to illustrate the recirculation
formation time and leads to smaller bubble length. As shown flow in the liquid slug. The blue lines show the cavita-
in Figure 11a, the higher is the ultrasound power, the smaller tion microstreaming. Surfactant SDS (Sodium dodecyl
is the bubble length. In MR0.25, the breakup point of the sulfate) with a density of 0.15% w/w is added to the
emerging bubble is downstream in the main channel, due to water to prevent sticking of the particles. Both the gas
and liquid flow rate in this experiment is 0.125 mL/min.
the stronger effect of channel wall’s surface tension at smaller The scale bar in the figure is 200 lm. [Color figure can
channel size. With the present of sonication, the position of be viewed in the online issue, which is available at
the breakup point moves back to the inlet, which may be wileyonlinelibrary.com.]

1300 DOI 10.1002/aic Published on behalf of the AIChE April 2016 Vol. 62, No. 4 AIChE Journal
Figure 9. The absolute velocity (UZ) of the tracer parti-
cle in the liquid slug at ultrasound power of
0, 30, and 40 W.
UZ is the particle’s velocity component in the flow direc-
tion. The horizontal axis is the distance from the meas-
ured point to the bubble tip. The particle velocity points
were measured from about 20 successive images, using
the high-speed camera control software PCC 2.14. The
chosen particles are all near the center line of the chan-
nel. Experiment condition is the same with that in Fig-
ure 8. [Color figure can be viewed in the online issue,
which is available at wileyonlinelibrary.com.]

bubble and liquid slug length generally increases with the ultra-
sound power (as shown in Figure 11), due to the fluctuation of
bubble formation process caused by the acoustic oscillation.

Pressure drop
Pressure drop is an important parameter for rational design
and application of microreactor.2,71,72 The total frictional pres-
sure drop of Taylor flow can be divided into three parts Figure 11. Initial bubble (a) and liquid slug (b) length at
according to its uniform structure.2,72–74 That is the inlet of the three USMRs under different
DPtot 5nS DPS 1nB ðDPbody 1DPcaps Þ (3) ultrasound power.
Each point was measured at least for nine times.
where DPS is the pressure drop in the liquid slug, DPbody the [Color figure can be viewed in the online issue, which
pressure drop across the bubble body, and DPcaps the excess is available at wileyonlinelibrary.com.]

Figure 10. Pictures showing the break-up of the gas bubble at the inlet of the three USMRs.
Pictures at each row were taken at different ultrasound power, as indicated on the left. Pictures at each column were taken in
different microreactors. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1301
 ultrasound power, with a relatively small extent of 15% at
power 40–50W.

Mass Transfer of the Gas-Liquid Flow

Overall mass-transfer coefficients
The overall volumetric mass-transfer coefficient of the gas-
liquid flow in microchannel is measured by an online method
based on the dissolution rate of gas bubble along the channel
length.77,78 Compared with conventional offline methods
which usually consist of measuring the gas and/or liquid com-
position at the oultet, such online method is more convenient
and accurate, as it avoids the serious end effects.1,12 The
method shows that the relationship between the bubble length
and its moving distance at the channel is exponential as
Figure 12. The measured pressure drops of the gas-
following
liquid flow in the three USMRs at different
ultrasound power.
The pressure drop was measured by inserting a pres-
sure sensor into the gas feeding line. [Color figure can
be viewed in the online issue, which is available at
wileyonlinelibrary.com.]

pressure drop over the bubble’s end caps. Due to the much
lower viscosity of gas than liquid, DPbody is negligible com-
pared with the other two parts. According to the theoretical
analysis by Bretherton,75 Wong,76 and Yue,2,72 the total pres-
sure drop can then be written as
 
L C LS qL US2 Lc r
DPtot 5 1DPcirculation 1 Ca2=3
LB 1LS ReS 2DH LB 1LS DH
(4)
The first term is DPS, which consists of the laminar pressure
drop under fully developed condition and the extra pressure
drop caused by the inner recirculation in the liquid slug
(DPcirculation). The second term is DPcaps, which is highly
dependent on the bubble and liquid slug length.
The effect of ultrasonication on pressure drop can be
summed up into two aspects. On the one hand, ultrasound
oscillation changes the length of bubble and liquid slug, which
would alter the pressure drop. If the length of bubble and liq-
uid slug is reduced by sonication, the pressure drop (mainly
DPcaps) would rise according to Eq. 4, as in the case of MR1
and MR0.5. On the contrary, the increase of bubble and liquid
slug length would lead to a decrease of the pressure drop,
which is the case for MR0.25. Conversely, the bubble surface
oscillations and acoustic streaming intensify the recirculation
in the liquid slug, leading to higher DPcirculation. In MR1, both
these two aspects lead to the obvious increase of the pressure
drop with ultrasound power. According to the experimental
measured results shown in Figure 12, the maximum pressure
drop increment with sonication reaches 1.2 kPa in MR1, about
34% of the value without sonication. In MR0.5, as the Figure 13. (a) The length of the bubble at different
decrease of bubble and liquid slug length under sonication is locations in the channel under different
less significant than that in MR1, the maximum increase of ultrasound power (MR1). (b) The measured
pressure drop is smaller, about 13% of the value without soni- overall volumetric mass-transfer coefficient
cation. In MR0.25, the effect of the above two aspects on pres- in the USMRs at different ultrasound power.
sure drop is opposite. As the bubble and liquid slug length
The dashed lines are fitting curves according to Eq. 5.
increases distinctly under sonication (by 80–120% in maxi- The error bars were based on three times of measure-
mum), the influence of the first aspect would exceed the sec- ment. [Color figure can be viewed in the online issue,
ond. This can explain the pressure drop decreasing with which is available at wileyonlinelibrary.com.]

1302 DOI 10.1002/aic Published on behalf of the AIChE April 2016 Vol. 62, No. 4 AIChE Journal
 increments in MR0.5 and MR0.25 are smaller than that in
MR1, the absolute increments are all around 3–5.4 s21. Such
mass-transfer enhancement is much higher than that in con-
ventional ultrasonic reactors, which is usually between 10 and
110%.27–30 This can be attributed to the better control of gas-
liquid flow field and acoustic field in our USMR. A very inter-
esting phenomenon is that the increase of kLa begins at differ-
ent points of ultrasound power, which are around 5 W, 20 W,
and 25 W in MR1, MR0.5, and MR0.25, respectively, as
shown in Figure 13b. These threshold power values are the
very power that surface wave oscillation starts to be excited in
the USMRs, as discussed in the bubble oscillation section. It
indicates that the enhancement of kLa in USMR is highly
related with the bubble surface oscillation.
The overall mass-transfer coefficient at different flow rates
were also measured, as shown in Figure 14. When the ultra-
sound is not present, kLa increases with the total flow rate
(QTP), which has been reported in other literatures.9,11,12
When the ultrasound power is turned to 20 W, the mass-
transfer coefficient is enhanced obviously. The kLa line at
power of 20 W is almost parallel with the line at 0 W. It indi-
cates that the mass-transfer process can be effectively
enhanced by ultrasound at a relatively wide range of flow rate.
At a given QTP, the effect of gas-liquid flow rate ratio (QG/QL)
on kLa was also investigated, as shown in Figure 14b. When
the ultrasound is not turned on, kLa decreases slightly with
QG/QL. This trend can be explained by the fact that kLa
increases more remarkably with liquid flow rate than gas flow
rate.2,78 When QG/QL increases at a given QTP, the liquid flow
rate decreases, leading to the reduction of kLa. The changing
trend of kLa with QG/QL at power of 20 W is similar to the
line at 0 W, except the last two points, which increase signifi-
cantly at higher QG/QL. This abnormal change is probably due
to the very short liquid slug at high QG/QL. As shown in
Figure 14. (a) The measured overall volumetric mass- Figure 14b, the liquid slug length drops to 0.8–0.9 mm at the
transfer coefficient at different total flow last two power points. As the range of the cavitation micro-
rate (QTP) in MR1 (QG/QL 5 2). (b) The meas- streaming is about 0.4 mm from the bubble tip (as discussed in
ured overall volumetric mass-transfer coef- Figures 8 and 9), streaming vortices at the two sides would
ficient at different gas-liquid flow rate ratio interact with each other at such short liquid slug, which would
(QG/QL) in MR1 (QTP 5 9 mL/min). further enhance the mass-transfer rate.
The dashed line is the liquid slug length at power of 20
W. [Color figure can be viewed in the online issue,
which is available at wileyonlinelibrary.com.]
The mechanism of mass-transfer enhancement
k a
2 jL x
In the above discussion, we have found that the mass-
LB 5m1 1m2 e L (5) transfer enhancement under ultrasonication is highly related to
the bubble surface oscillation and cavitation microstreaming.
where jL is the superficial velocity of liquid, and parameter m1
The detailed mechanism is further analyzed here.
and m2 only depend on the experimental conditions. By curve
The overall volumetric mass-transfer coefficient is the prod-
fitting of the measured bubble length at different locations
uct of specific surface area (a) and individual mass-transfer
with Eq. 3 (as shown in Figure 13a), the overall mass-transfer
coefficient (kL). By assuming the Taylor bubble as a long cyl-
coefficient (kLa) can be determined from the exponential
inder with two hemisphere caps, the specific surface area can
decay rate (kLa/jL).77,78 It can be seen from Figure 13a that the
bubble length decay rate increases with the ultrasound power, be written as
indicating faster mass-transfer rate at higher power. pdB ðLB 2dB Þ1pdB 2
The overall mass-transfer coefficient in the three USMRs a5 (6)
VB
was shown in Figure 13b. When the ultrasound is not intro-
duced, microreactor with smaller channel has higher kLa, where the numerator at right is the surface area of the bubble
mainly due to its higher specific surface area. When the ultra- (SB).20,79 When surface wave oscillation was excited on the
sound is present, kLa in all the USMRs was significantly bubble under sonication, the surface area of the wavy bubble
enhanced. The maximum kLa increment (relative to the kLa becomes larger, as shown in Figure 15. If we simplify the gas-
without sonication) is about 2000%, 300%, and 500% in MR1, liquid interface line at a cross-section as an oscillating sinusoi-
MR0.5, and MR0.25, respectively. Although the percentage dal wave

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1303
 rffiffiffiffiffiffi
DL
kL 52 (11)
pt
where t is the two phase exposure (or contact) time of the sur-
face element, which depends on the interface renewal
rate.10,13,81 According to Calderbank’s analysis,82,83 the expo-
sure time (t) in a bubbling system can be derived in terms of
the average bubble size (dB) and the average bubble slip veloc-
ity (US),13 so
Figure 15. Schematic illustrating the mechanism of
rffiffiffiffiffiffiffiffiffiffiffiffi
DL US
mass-transfer enhancement by ultrasound. kL 52 (12)
pdB
[Color figure can be viewed in the online issue, which
is available at wileyonlinelibrary.com.] The bubble slip velocity, the relative velocity between the
    gas and liquid phase, can be estimated from the experimental
2pt 2px holdup and superficial gas and liquid velocities for a gas-
y5Acos sin (7)
T k liquid reactor with cocurrent flow13,84,85
where k is the wavelength and A is the oscillation amplitude of jG jL
US 5 2 (13)
the surface wave. The ratio of this sine line’s length to the rG rL
original straight line can be written as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
ð ð
L0 1 2p 2p 2pA 2 2 2
b5 5 2 11 cos ðtÞsin ðxÞdxdt (8)
L 4p 0 0 k

in which the inner integration is sine line’s length at a time

and the outer integration is the time average over a period.
The surface area has a square relation with the interface line
length, so the surface area of the wavy bubble can be written
as
S0B 5b2 SB (9)
As the volume variation caused by the surface oscillation is
quite small, especially when compared with the total bubble
volume, the specific surface area under sonication can be
expressed as
a0 5b2 a (10)
This equation indicates that the specific surface area of gas-
liquid flow under sonication is increased by b2 times. By sub-
stituting the measured wavelength and oscillation amplitudes
(which have been discussed in the bubble oscillation section)
into Eq. 8, b2 can be calculated, which varies from 1.3 to 2.6,
depending on the oscillation amplitude of the surface wave.
The specific surface area without sonication can be calcu-
lated by inserting the bubble and liquid slug size into Eq. 6.
Then, the specific surface area under different ultrasound
power can be obtained by Eq. 10. By dividing the measured
kLa with this calculated specific surface area, the individual
mass-transfer coefficient (kL) was obtained, which is shown in
Figure 16b. It can be seen that the specific surface area is
improved by 30–160% in the USMRs, depending on the ultra-
sound power. The maximum kL enhancements range from 90
to 650%, which differ in the three USMRs. Nevertheless, the
absolute kL in the three USMRs at high ultrasound power (30–
50 W) are all around 0.32–0.43 mm/s. The changing trend of
the specific surface area and kL with ultrasound power in the
three USMRs is similar to that of kLa in Figure 13b and oscil- Figure 16. The estimated specific surface area (a) and
lation amplitude in Figure 7, in which the enhancement begins mass-transfer coefficient (b) in the three
at different threshold power due to the confinement effect. USMRs at different ultrasound power.
The relation of individual mass-transfer coefficient between
The dashed line in figure (b) is the mass-transfer coef-
the acoustic streaming can also be described quantitatively. ficient of MR0.5 calculated by Eq. 14. [Color figure
According to Higbie’s penetration model,80 the mass-transfer can be viewed in the online issue, which is available at
coefficient can be expressed as wileyonlinelibrary.com.]

1304 DOI 10.1002/aic Published on behalf of the AIChE April 2016 Vol. 62, No. 4 AIChE Journal
where rG and rL is the experimental hold up of the gas and microstreaming lead to higher pressure drops, which is also
liquid phase, which can be calculated from the bubble and liq- influenced by the changing of bubble and liquid slug length.
uid slug volume in our experiment.79 As acoustic streaming The mass-transfer coefficient at different ultrasound power,
also leads to a relative moving of the liquid element to the channel size, and flow rate were studied by an online measure-
bubble surface, it can be deduced that bubble slip velocity ment method. The overall volumetric mass-transfer coeffi-
under sonication should be the sum of the original US and cients in the USMRs were enhanced by 3–20 times (relative to
streaming velocity (UA) (see also Figure 15). Therefore, the the kLa in microreator without sonication), which is much
individual mass-transfer coefficient under sonication can then higher than that in conventional ultrasonic reactors (usually
be rewritten as between 10 and 110%) due to the better control of gas-liquid
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi flow field and sound field. The mass-transfer intensification
0 DL ðUS 1UA Þ can be attributed to two reasons: surface wave oscillations
kL 52 (14)
pdB increase the specific surface area, while cavitation micro-
streaming accelerates the interface renewal and thus improves
which can also be rearranged as the individual mass-transfer coefficient. Although based on
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi USMR, such mechanism would also be helpful for understand-
US 1UA
kL0 5 kL (15) ing mass-transfer processes in conventional ultrasonic equip-
US
ment. With a better understanding and control of the
Once we know the acoustic streaming velocity and bubble intensification mechanism, USMR would become an efficient
slip velocity, the mass-transfer coefficient under sonication method to enhance gas-liquid adsorption, reaction, and gas
can be calculated. Taking MR0.5 for example, the bubble slip agitation processes.
velocity calculated by Eq. 13 is around 0.02 m/s. As discussed
before (see Figure 9), the acoustic streaming velocity near the Acknowledgments
bubble surface is about 0.03 and 0.04 m/s at ultrasound power
The authors are grateful to Professor Cheng Wang (Mis-
of 30 and 40 W, respectively (and 0 m/s at power 0–20W).
souri University of Science and Technology) for his valuable
The individual mass-transfer coefficients calculated by insert-
advice in conducting the streak photography experiments
ing these values into Eq. 14 are also showed in Figure 16b.
and analyzing the bubble oscillation phenomena. Authors
The calculated results are close with the values estimated from
would like to thank Jiansheng Chu, Hengqiang Li, and Feng-
the experimental data, which verifies the rationality of Eq. 14.
jun Jiao (engineers in our research group) for their help in
However, the theoretical values are systematically higher than
manufacturing the microreactor and building the experimen-
the experimental results, which may be caused by the overesti-
tal setup. The authors acknowledge the financial supports for
mate of US.
this project from National Natural Science Foundation of
In conclusion, mass-transfer enhancement in our USMRs
China (Nos. 21225627 and 91334201).
can be attributed into two reasons, as illustrated in Figure 15.
First, the specific surface area is increased by surface wave
oscillation on the bubble. Second, the mass-transfer coefficient Notation
is improved by the cavitation microstreaming, which adds to A= oscillation amplitude of the surface waves
the bubble slip velocity and increase the surface renewal rate. a= specific surface area
a0 = specific surface area under sonication
The overall volumetric mass-transfer coefficient under ultraso- Ca = two phase capillary number defined by (Ca 5 lL jTP/rL)
nication can be expressed as cp = specific heat
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi D= depth of the main channel
US 1UA
ðkL aÞ0 5 bkL a (16) DL = diffusion coefficient in the liquid
US dB = average equivalent diameter of the bubble
f= frequency
where the length ratio (b) and streaming velocity (UA) depend jG = superficial gas velocity
on the ultrasound power, bubble shape, and environment. jL = superficial liquid velocity
kL = individual mass-transfer coefficient
kL0 = individual mass-transfer coefficient under sonication
Conclusion kLa = overall volumetric mass-transfer coefficient
(kLa)0 = overall volumetric mass-transfer coefficient under
To intensify the gas-liquid mass-transfer process and study sonication
the microscopic mechanism, three USMRs were built and LB = length of the bubble
characterized. A strong and uniform sound field with power P= ultrasound power
density of 0.07–0.3 W/mL was generated in the microreactor. Pth = acoustic power density
QG = gas flow rate
Under this ultrasound field, surface wave oscillations with dif- QL = liquid flow rate
ferent modes were excited on the slug bubble during the stable QTP = total flow rate (QTP 5 QG 1 QL)
cavitation process. A confinement effect was found that it SB = surface area of the bubble
needs more ultrasound energy to excite the surface wave oscil- t= two phase exposure (or contact) time
UA = acoustic streaming velocity
lation for bubbles confined in smaller channels. Cavitation US = bubble slip velocity
microstreaming with a velocity of 0.03–0.04 m/s was observed VB = volume of the bubble
near the oscillation bubbles. W= width of the main channel
The influences of ultrasonication on hydrodynamics of the x= distance from the bubble’s location in the channel to the
inlet
gas-liquid Taylor flow were investigated. Ultrasound oscilla- Y= admittance
tions change the rupture/expansion step time and the breakup DPtot = total frictional pressure drop of gas-liquid Taylor flow
point, resulting in a variation of the initial bubble length with DPS = pressure drop in the liquid slug
sound power. Bubble surface oscillations and cavitation DPbody = pressure drop across the bubble body

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1305
 DPcaps = excess pressure drop over the bubble’s end caps 23. Gogate PR, Sutkar VS, Pandit AB. Sonochemical reactors: important
DPcirculation = pressure drop caused by the inner recirculation in the liq- design and scale up considerations with a special emphasis on heter-
uid slug ogeneous systems. Chem Eng J. 2011;166:1066–1082.
b = length ratio of sine wave line to its original straight line 24. Stankiewicz A. Energy matters alternative sources and forms of
k = wavelength energy for intensification of chemical and biochemical processes.
q = density Chem Eng Res Des. 2006;84(A7):511–521.
r = surface tension 25. Freitas S, Hielscher G, Merkle HP, Gander B. Continuous contact-
rG = hold up of the gas phase and contamination-free ultrasonic emulsification—a useful tool for
rL = hold up of the liquid phase pharmaceutical development and production. Ultrason Sonochem.
2006;13:76–85
26. Cadwell ML, Fogler HS. Ultrasonic gas absorption and acoustic
Literature Cited streaming observations. Chem Eng Prog Symp Series. 1971;67:124–
1. Yue J, Chen GW, Yuan Q, Luo L, Gonthier Y. Hydrodynamics and 127.
mass transfer characteristics in gas–liquid flow through a rectangular 27. Gondrexon N, Renaudin V, Boldo P, Gonthier Y, Bernis A, Petrier
microchannel. Chem Eng Sci. 2007;62(7):2096–2108. C. Degassing effect and gas-liquid transfer in a high frequency sono-
2. Yue J, Luo LA, Gonthier Y, Chen GW, Yuan Q. An experimental chemical reactor. Chem Eng J. 1997;66:16–21.
study of air-water Taylor flow and mass transfer inside square micro- 28. Kumar A, Gogate PR, Pandit AB, Delmas H, Wilhelm AM. Gas liq-
channels. Chem Eng Sci. 2009;64(16):3697–3708. uid mass transfer studies in sonochemical reactors. Ind Eng Chem
3. Al-Rawashdeh M, Zalucky J, Muller C, Nijhuis TA, Hessel V, Res. 2004;43:1812–1819.
Schouten JC. Phenylacetylene hydrogenation over 29. Laugier F, Andriantsiferana C, Wilhelm AM, Delmas H. Ultrasound
[Rh(NBD)(PPh3)2]BF4 catalyst in a numbered-up microchannels in gas-liquid systems: effects on solubility and mass transfer. Ultra-
reactor. Ind Eng Chem Res. 2013;52:11516–11526. son Sonochem. 2008;15:965–972.
4. Liedtke AK, Bornette F, Philippe R, de Bellefon C. Gas–liquid–solid 30. Herran NS, Lopez JLC, Pere JAS. Gas-liquid mass transfer in soni-
“slurry Taylor” flow: experimental evaluation through the catalytic cated bubble columns. Effect of reactor diameter and liquid height.
hydrogenation of 3-methyl-1-pentyn-3-ol, Chem Eng J. 2013;227: Ind Eng Chem Res. 2012;51:2769–2774.
174–181. 31. Ashokkuma M, Lee J, Kentish S, Grieser F. Bubbles in an acoustic
5. Leclerc A, Alame M, Schweich D, Pouteau P, Delattre C, de field: an overview. Ultrason Sonochem. 2007;14:470–475.
32. Leighton TG. Bubble population phenomena in acoustic cavitation.
Bellefon C. Gas–liquid selective oxidations with oxygen under
Ultrason Sonochem. 1995;2(2):S123–S136.
explosive conditions in a micro-structured reactor. Lab Chip. 2008;8:
33. Rivas DF, Cintas PC, Gardeniers HJGE. Merging microfluidics and
814–817.
sonochemistry: towards greener and more effcient micro-sono-reac-
6. Hessel V, Kralisch D, Kockmann N, Noel T, Wang Q. Novel pro-
tors. Chem Commun. 2012;48:10935–10947.
cess windows for enabling, accelerating, and uplifting flow chemis-
34. Leighton TG. What is ultrasound? Prog Biophys Mol Biol. 2007;93:
try. ChemSusChem. 2013;6:746–-789.
3–83.
7. de Mas N, Gunther A, Schmidt MA, Jensen KF. Microfabricated
35. Hubner S, Kressirer S, Kralisch D, Bludszuweit-Philipp C, Lukow
multiphase reactors for the selective direct fluorination of aromatics.
K, Janich I, Schilling A, Hieronymus H, Liebner C, Jahnisch K.
Ind Eng Chem Res. 2003;42:698–710. Ultrasound and microstructures-a promising combination? ChemSu-
8. Hartman RL, Jensen KF. Microchemical systems for continuous-flow sChem. 2012;5:279–288.
synthesis. Lab Chip. 2009;9:2495–2507. 36. Stankiewicz A. On the applications of alternative energy forms and
9. Sobieszuk P, Aubin J, Pohorecki R. Hydrodynamics and mass trans- transfer mechanisms in microprocessing systems. Ind Eng Chem Res.
fer in gas-liquid flows in microreactors. Chem Eng Technol. 2012; 2007;46:4232–4235.
35(8):1346–1358. 37. Yue J, Luo LA, Gonthier Y, Chen GW, Yuan Q. An experimental
10. Kashid MN, Renken A, Kiwi-Minsker L. Gas-liquid and liquid- investigation of gas-liquid two-phase flow in single microchannel
liquid mass transfer in microstructured reactors. Chem Eng Sci. contactors. Chem Eng Sci. 2008;63(16):4189–4202.
2011;66:3876–3897. 38. Iida Y, Tuziuti T, Yasui K, Towata A, Kozuka T. Bubble motions
11. Zaloha P, Kristal J, Jiricny V, Volkel N, Xuereb C, Aubin J. Charac- confined in a microspace observed with stroboscopic technique.
teristics of liquid slugs in gas-liquid Taylor flow in microchannels. Ultrason Sonochem. 2007;14:621–626.
Chem Eng Sci. 2012;68:640–649. 39. Hashmi A, Yu G, Reilly-Collette M, Heiman G, Xu J. Oscillating
12. Sobieszuk P, Pohorecki R, Cyganski P, Grzelka J. Determination of bubbles: a versatile tool for lab on a chip applications. Lab Chip.
the interfacial area and mass transfer coefficients in the Taylor gas– 2012;12:4216–4227.
liquid flow in a microchannel. Chem Eng Sci. 2011;66(23):6048– 40. Yao CQ, Dong ZY, Zhao YC, Chen GW. The effect of system pres-
6056 sure on gas-liquid slug flow in a microchannel. AIChE J. 2014;60:
13. Nieves-Remacha MJ, Kulkarni AA, Jensen KF. Gas-liquid flow and 1132–1142.
mass transfer in an advanced-flow reactor. Ind Eng Chem Res. 2013; 41. Lin SY. The Mechanism and Design of Ultrasound Transducer.
52:8996–9010. Beijing: Science Press, 2004.
14. Kockmann N. Transport Phenomena in Micro Process Engineering. 42. Mathieson A, Cardoni A, Cerisola N, Lucas M. The influence of pie-
Berlin: Springer, 2007. zoceramic stack location on nonlinear behavior of Langevin trans-
15. Tabeling P. A brief introduction to slippage, droplets and mixing in ducers. IEEE Trans Ultrason Ferroelectr Freq Control. 2013;60:
microfluidic systems. Lab Chip. 2009;9:2428–2436. 1126–1133.
16. Kashid MN, Renken A, Kiwi-Minsker L. Microstructured Devices 43. Dual J, Hahn P, Leibacher I, Moller D, Schwarz T. Acoustofluidics
for Chemical Processing. Weinheim: Wiley-VCH, 2015. 6: experimental characterization of ultrasonic particle manipulation
17. Hessel V, Lowe H, Schonfeld F. Micromixers—a review on passive devices. Lab Chip. 2012;12:852–862.
and active mixing principles. Chem Eng Sci. 2005;60(8-9):2479– 44. Loning JM, Horst C, Hoffmann U. Investigations on the energy con-
2501. version in sonochemical processes. Ultrason Sonochem. 2002;9:169–
18. Nguyen NT, Wu ZG. Micromixers—a review. J Micromech Micro- 179.
eng. 2005;15(2):R1–R16. 45. Macedo RG, Verhaagen B, Rivas DF, Gardeniers JGE, van der Sluis
19. Gunther A, Khan SA, Thalmann M, Trachsel F, Jensen KF. Trans- LWM, Wesselink PR, Versluis M. Sonochemical and high-speed
port and reaction in microscale segmented gas–liquid flow. Lab optical characterization of cavitation generated by an ultrasonically
Chip. 2004;4:278–286. oscillating dental file in root canal models. Ultrason Sonochem.
20. Tan J, Lu YC, Xu JH, Luo GS. Mass transfer performance of gas- 2014;21:324–335.
liquid segmented flow in microchannels. Chem Eng J. 2012;181: 46. Lauterborn W, Kurz T. Physics of bubble oscillations. Rep Prog
229–235. Phys. 2010;73:106501(88pp).
21. Dong ZY, Yao CQ, Zhang XL, Xu J, Chen GW, Zhao YC, Yuan Q. 47. Ashokkumar M. The characterization of acoustic cavitation bub-
A high-power ultrasonic microreactor and its application in gas- bles—an overview. Ultrason Sonochem. 2011;18:864–872.
liquid mass transfer intensification. Lab Chip. 2015;15:1145–1152. 48. Birkin PR, Offin DG, Vian CJB, Leighton TG, Maksimov AO.
22. Mason TJ, Peters D. Applied Sonochemistry: Uses of Power Ultra- Investigation of non-inertial cavitation produced by an ultrasonic
sound in Chemistry and Processing. Weinheim: Wiley-VCH, 2002. horn. J Acoust Soc Am. 2011;130(5):3297–3307.

1306 DOI 10.1002/aic Published on behalf of the AIChE April 2016 Vol. 62, No. 4 AIChE Journal
49. Offin DG, Birkin PR, Leighton TG. Electrodeposition of copper in 69. Garstecki P, Fuerstman MJ, Stone HA, Whitesides GM. Formation
the presence of an acoustically excited gas bubble. Electrochem of droplets and bubbles in a microfluidic T-junction-scaling and
Commun. 2007;9:1062–1068. mechanism of break-up. Lab Chip. 2006;6(3):437–446.
50. Prabowo F, Ohl CD. Surface oscillation and jetting from surface 70. van Steijn V, Kleijn CR, Kreutzer MT. Predictive model for the size
attached acoustic driven bubbles. Ultrason Sonochem. 2011;18:431–435. of bubbles and droplets created in microfluidic T-junctions. Lab
51. Dangla R, Poulain C. When sound slows down bubbles. Phys Fluids. Chip. 2010;10(19):2513–2518.
2010;22:041703. 71. Kreutzer MT, Kapteijn F, Moulijn JA, Heiszwolf JJ. Multiphase
52. Elder SA. Cavitaion microstreaming. J Acoust Soc Am. 1959;31:54– monolith reactors: chemical reaction engineering of segmented flow
64. in microchannels. Chem Eng Sci. 2005;60(22):5895–5916.
53. Tandiono T, Ohl SW, Ow DSW, Klaseboer E, Wong VV, Camattari 72. Yue J, Rebrov EV, Schouten JC. Gas-liquid-liquid three-phase flow
A, Ohl CD. Creation of cavitation activity in a microfluidic device pattern and pressure drop in a microfluidic chip: similarities with
through acoustically driven capillary waves. Lab Chip. 2010;10:
gas-liquid/liquid-liquid flows. Lab Chip. 2014;14(9):1632–1649.
1848–1855.
73. Fuerstman MJ, Lai A, Thurlow ME, Shevkoplyas SS, Stone HA,
54. Eggers J. Nonlinear dynamics and breakup of free-surface flows. Rev
Whitesides GM. The pressure drop along rectangular microchannels
Mod Phys. 1997;69:865–929.
55. Xu J, Attinger D. Control and ultrasonic actuation of a gas–liquid containing bubbles. Lab Chip. 2007;7:1479–1489.
interface in a microfluidic chip. J Micromech Microeng. 2007;17: 74. Jovanović J, Zhou W, Rebrov EV, Nijhuis TA, Hessel V, Schouten
609–616. JC. Liquid–liquid slug flow: hydrodynamics and pressure drop.
56. Gao SQ, Liu HP. Capillary Mechanics. Beijing: Science Press, 2010. Chem Eng Sci. 2011;66:42–54.
57. Nyborg WL. Acoustic streaming. In: Mason WP, editor. Physical 75. Bretherton FP. The motion of long bubbles in tubes. J Fluid Mech.
Acoustics IIB. New York: Academic Press, 1965. 1961;10(02):166–188.
58. Nyborg WL. Acoustic streaming. In: Hamilton MF, Blackstock DT, 76. Wong H, Radke CJ, Morris S. The motion of long bubbles in polyg-
editors. Nonlinear Acoustics. New York: Academic Press, 1998. onal capillaries. Part 2. Drag, fluid pressure and fluid-flow. J Fluid
59. Wiklund M, Green R, Ohlin M. Acoustofluidics 14: applications of Mech. 1995;292:95–110.
acoustic streaming in microfluidic devices. Lab Chip. 2012;12:2438– 77. Yao CQ, Dong ZY, Zhao YC, Chen GW. An online method to mea-
2451. sure mass transfer of slug flow in a microchannel. Chem Eng Sci.
60. Wang C, Jalikop SV, Hilgenfeldt S. Efficient manipulation of micro- 2014;112:15–24.
particles in bubble streaming flows. Biomicrofluidics. 2012;6:012801. 78. Yao CQ, Dong ZY, Zhao YC, Chen GW. Gas-liquid flow and mass
61. Wang C, Rallabandi B, Hilgenfeldt S. Frequency dependence and transfer in a microchannel under elevated pressures. Chem Eng Sci.
frequency control of microbubble streaming flows. Phys Fluids. 2015;123:137–145.
2013;25:022002. 79. Yao CQ, Zhao YC, Ye CB, Dang MH, Dong ZY, Chen GW. Char-
62. Tho P, Manasseh R, Ooi A. Cavitation microstreaming patterns in sin- acteristics of slug flow with inertial effects in a rectangular micro-
gle and multiple bubble systems. J Fluid Mech. 2007;576:191–233. channel. Chem Eng Sci. 2013;95:246–256.
63. Ryu K, Chung SK, Cho SK. Micropumping by an acoustically 80. Higbie R. The rate of absorption of a pure gas into a still liquid dur-
excited oscillating bubble for automated implantable microfluidic ing short periods of exposure. Trans AIChE. 1935;31:365–389.
devices. J Lab Autom. 2010;15:163–171.
81. van Baten JM, Krishna R. CFD simulations of mass transfer from
64. Tovar AR, Patel MV, Lee AP. Lateral air cavities for microfluidic
Taylor bubbles rising in circular capillaries. Chem Eng Sci. 2004;59:
pumping with the use of acoustic energy. Microfluid Nanofluid.
2535–2545.
2011;10:1269–1278.
65. Ahmed D, Mao X, Shi J, Juluri BK, Huang TJ. A millisecond micro- 82. Calderbank PH, Lochiel AC. Mass transfer coefficients, velocities
mixer via single-bubble-based acoustic streaming. Lab Chip. 2009;9: and shapes of carbon dioxide bubbles in free rise through distilled
2738–2741. water. Chem Eng Sci. 1964;19:485–503.
66. Ozcelik A, Ahmed D, Xie Y, Nama N, Qu Z, Nawaz AA, Huang 83. Calderbank PH, Korchinski IJO. Circulation in liquid drops (a heat
TJ. An acoustofluidic micromixer via bubble inception and cavitation transfer study). Chem Eng Sci. 1956;6:65–78.
from microchannel sidewalls. Anal Chem. 2014;86:5083–5088. 84. Lapidus L, Elgin JC. Mechanics of vertical-moving fluidized sys-
67. Startrails (Version 2.3) is a free software developed by Achim Schal- tems. AIChE J. 1957;3:63–68.
ler in 2013. Available at: http://www.startrails.de/html/software.html. 85. Lockett MJ, Kirkpatrick RD. Ideal bubbly flow and actual flow in
68. Dang MH, Yue J, Chen GW, Yuan Q. Formation characteristics of bubble columns. Trans Inst Chem Eng. 1975;53:267–273.
Taylor bubbles in a microchannel with a converging shape mixing
junction. Chem Eng J. 2013;223:99–109. Manuscript received July 24, 2015, and revision received Sep. 26, 2015.

AIChE Journal April 2016 Vol. 62, No. 4 Published on behalf of the AIChE DOI 10.1002/aic 1307