Mass transfer characteristics of bubbly flow in microchannels

Lu Yang, Jing Tan, Kai Wang, Guangsheng Luo n

The State Key Laboratory of Chemical Engineering, Department of Chemical Engineering, Tsinghua University, Beijing 100084, China

H I G H L I G H T S

� The mass transfer characteristics of bubbly flow were first studied.� The mass transfer coefficients were determined by an online measurement method.� The mass transfer coefficients of bubble forming and flowing stages were studied.� The effect factors on mass transfer coefficient were discussed.� The empirical correlations were established to predict mass transfer coefficients.

a r t i c l e i n f o

Article history:Received 27 October 2013Received in revised form2 February 2014Accepted 4 February 2014Available online 11 February 2014

Keywords:Microfluidic deviceBubbly flowAbsorptionMass transferMicrostructureTransport process

a b s t r a c t

The mass transfer characteristics of bubbly flows in co-flowing and T-junction microchannels aredescribed in this study. A CO2–N2 gas mixture and a monoethanolamine–ethylene glycol solution wereused as the gas and liquid phases, respectively, to generate gas/liquid bubbly microflows. The bubbledimensions changed obviously during the bubble-forming and bubble-flowing stages. The overall masstransfer coefficients (KL) at these stages were determined by developing an online measurement methodthrough which time-dependent changes in bubble volume were analyzed. Investigation of the effects ofphase flow rate and concentration of the two phases on KL showed that KL was in the range of 1�10�4–

4.9�10�3 m/s. In addition, empirical correlations were established to predict KL for different masstransfer stages.

& 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Microfluidic devices, which have undergone rapid developmentover the last two decades, exhibit many attractive advantages ascompared with conventional large-scale devices (Gunther andJensen, 2006; Kobayashi et al., 2006; Mae, 2007). Microfluidic deviceshave been widely applied in various research areas, includingchemical synthesis, separation, material preparation, and biologicalengineering (Huebner et al., 2008; Kobayashi et al., 2004; Seemannet al., 2012; Wang et al., 2011). Gas/liquid systems play importantroles in scientific and industrial fields, and a number of gas/liquidreaction processes and separation processes have been successfullyrealized in microfluidic devices (Leclerc et al., 2008; Tan et al., 2011,2012). In general, adequate reduction of the mass transfer limitationto greatly enhance the process is desired. Therefore, investigating themass transfer characteristics in various flow regimes and microfluidicdevices is very essential.

In recent years, several experimental investigations, in which themass transfer characteristics of gas/liquid systems were examined inmicrofluidic devices, have been published. Yue et al. (2009) studiedthe mass transfer performance of air/water Taylor flow in Y-shapedmicrochannels. H.J. Su et al., 2010 investigated the mass transferprocess of H2S absorption by methyldiethanolamine (MDEA) solutionin a cross-flow T-junction microchannel. Aoki et al. (2011) studied themass transfer coefficient of slug flow in miniaturized channels andestablished a correlation between the Sherwood number and thePeclet number. Roudet et al. (2011) examined the process of oxygenmass transfer from gas phase to water phase in straight and mean-dering millimetric channels for both Taylor flow and slug-annularflow regimes. In general, the mass transfer coefficients in theprevious studies are usually determined by analyzing the soluteconcentration of samples collected at the outlet of microfluidicdevices. However, the sample collection time and phase separationtime are often longer than the fluid residence time in microfluidicdevices, which might result in inaccurate characterization of masstransfer (Kashid et al., 2011a). A simple approach to overcome thisproblem in gas/liquid systems is to establish an online measurementmethod that determines the quantity of mass transfer and the solute

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ces

Chemical Engineering Science

http://dx.doi.org/10.1016/j.ces.2014.02.0040009-2509 & 2014 Elsevier Ltd. All rights reserved.

n Corresponding author. Tel./fax: þ86 10 62783870.E-mail address: gsluo@tsinghua.edu.cn (G. Luo).

Chemical Engineering Science 109 (2014) 306–314

concentration by analyzing the volume changes in the gas slugs. Byusing this approach, Abolhasani et al. (2012) obtained the volumetricmass transfer coefficient and Henry's constant for CO2 dissolving inphysical solvents. Li et al. (2012) investigated the process of CO2

chemical absorption and determined the reaction rate constant. Inour group, Tan et al. (2012a) also used this approach and obtainedthe overall mass transfer coefficient of segmented flow with CO2

chemical absorption.The aforementioned studies focused mainly on the mass transfer

performance of gas slug flow, but it is a challenge to determinewhether mass transfer occurs in the liquid film between the gas slugand the channel wall (Roudet et al., 2011; Sobieszuk et al., 2011).However, few studies have reported the mass transfer performance ofbubbly flow in microfluidic devices. Considering the advantages ofbubbly flow, such as its well-defined specific interfacial area andgreater stability, here we developed an online measurement approachand studied the mass transfer characteristics of bubbly flow with CO2

chemical absorption for the first time. Moreover, owing to the greatimpact of mass transfer stages on process control and equipmentdesign, distinguishing different stages of mass transfer is considered tobe important (Aoki et al., 2011; Shao et al., 2012a; Wang et al., 2009;Xu et al., 2012). In the case of liquid/liquid systems, Xu et al. (2008)investigated the mass transfer rules in the droplet-forming anddroplet-flowing stages. For gas/liquid systems, Tan et al. (2012b)studied the mass transfer performance of segmented flow duringthe forming stage in T-junction microchannels with different contactangles. In this study, we calculated the overall mass transfer coefficient(KL) at the bubble-forming and bubble-flowing stages. In addition, theeffects of phase flow rate and concentration of the two phases on themass transfer coefficients were also clarified, and correlation equationswere developed to predict the mass transfer coefficients in differentstages.

2. Experiments and methodology

2.1. Microfluidic devices

Fig. 1 shows schematic diagrams of the microfluidic chips usedin this study. Two different microfluidic devices – a co-flowingmicrochannel and a T-junction microchannel with embeddedcapillaries – were designed to generate monodispersed micro-bubbles. These devices were fabricated on polymethyl methacry-late (PMMA) plates by precision milling and sealed by a high-pressure thermal sealing machine (A274, Techson). The micro-channels have a square cross section with an area of A¼0.36 mm2.Glass capillaries with a tapered tip were used as the dispersedphase inlet channel to prevent large-area contact of gas and liquidphases before bubble formation. The capillaries were pulled by aflaming micropipette puller (Sutter Instrument Company) to formcuspate tips with a diameter of 45 μm. The meandering channelslocated downstream provided sufficient time to accomplish themass transfer process.

2.2. Operation and observation

A high-speed CMOS camera (DK-2740, Dantec Dynamics) andmicroscope (BXFM, Olympus) were used to record the bubble-forming and bubble-flowing stages with frame frequencies of2000 and 1000 fps, respectively. Both gas and liquid phases weredelivered and controlled by syringe pumps (LSP01-1A, Longer).The gas flow rate (QG) was controlled at 75 μL/min or 100 μL/min,and the liquid flow rates (QL) varied from 225 μL/min to 700 μL/min.The corresponding uG (QG/A) was in the range of 3.5–4.6 mm/s, whileuL (QL/A) ranged from 10 mm/s to 32 mm/s. As a result, the averageresidence time of two-phase fluids varied from 2.4 s to 6.5 s.

The Reynolds number ReG was in the range of 0.2–0.3, and ReLranged from 0.4 to 1.4. In our experiments, the polydispersity indexfor microbubbles was less than 3%, indicating that the microbubbleswere highly uniform (Xu et al., 2006).

2.3. Experimental system

In this study, the experimental system comprised a CO2–N2 gasmixture and a MEA–ethylene glycol solution (Sada et al., 1985). Thegas (purity 99.99%) was purchased from Beijing Huayuan Company,and the analytically pure MEA and ethylene glycol (Z99.0%) wereprovided by Beijing Chemical Plant. Different CO2 volume fractionsof CO2–N2 gas mixture (17 vol%, 37 vol%, and 56 vol%) and differentMEA concentrations' solutions (5 wt% and 15 wt%) were preparedbefore the experiments. The viscosity was measured with anUbbelohde viscometer for all liquid phases, and the surface tensionof each system was measured with a pendant drop interfacialtension meter (OCAH200, DataPhysics Instruments GmbH). Theviscosity of each gas mixture and the diffusion coefficient of CO2

in each solution were calculated by the mixing method (Dubois andThomas, 2009; Yue et al., 2004). The physical properties of thedifferent experimental systems are listed in Table 1.

2.4. Methodology

In this study, an online measurement method through whichtime-dependent changes in bubble volume were analyzed wasdeveloped to determine the mass transfer coefficients at thebubble-forming and bubble-flowing stages. This method wasbased on the following premises:

(1) The CO2–N2 gas mixture can be taken as ideal gas.(2) N2 is considered as an inert component, and its solubility in

liquid solutions is negligible. The evaporation of liquid solu-tions can be ignored as well.

Fig. 1. Schematic diagrams of microfluidic devices: (a) co-flowing microchanneland (b) T-junction microchannel.

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314 307

(3) The physical properties of the liquid phase change slightlywith the absorption of CO2.

(4) The reaction between CO2 and MEA is fast and mainly takesplace within the liquid film near the gas–liquid interface. Theconcentration of CO2 in the liquid bulk phase is approximatelyequal to zero.

(5) The pressure in the gas bubbles does not significantly changeduring the bubble-flowing stage and can be regarded as aconstant.

Premise (4) is validated by the CO2–MEA reaction kinetics. MEAis a primary alkanolamine. When CO2 is absorbed by MEA, thereaction between them proceeds through the formation of azwitterion as an intermediate

CO2þRNH2 ⟷k1 ;k� 1RNþH2COO

� ð1ÞThis zwitterion undergoes deprotonating by a base, resulting incarbonate formation (Vaidya and Kenig, 2007). Usually in non-aqueous solvents, only the amine is regarded as a deprotonatingbase (Versteeg et al., 1996):

RNþH2COO� þRNH2⟹

kbRNHCOO� þRNHþ

3 ð2ÞFrom this mechanism, the overall reaction rate equation can bederived from the assumption of a quasi-steady state condition forthe zwitterion concentration

rCO2 ¼CCO2CMEA

1=k1þððk�1Þ=ðk1kbCMEAÞÞð3Þ

Because the absorption proceeded at high MEA concentrations inthe experiments, the MEA concentrations were regarded as con-stants. Therefore, the reaction process can be considered as apseudo-first-order reaction with respect to CO2. The reaction rateis given as

rCO2 ¼ k0CCO2 ð4ÞTherefore, the Hatta number is defined as (Alvarezfuster et al.,1981)

Ha¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffik0DCO2

pkL0

ð5Þ

Here, kL0 is the liquid-side mass transfer coefficient of CO2 insolvent without chemical absorption. Under experimental condi-tions, the values of Ha are greater than 3 (Park et al., 2005),indicating that the reaction is fast and mainly takes place withinthe liquid film (Kashid et al., 2011a). Therefore, Premise (4) issatisfied.

Premise (5) is explained by the fluid pressure drop in themicrochannel. According to the homogenous flow model sug-gested by Yue et al. (2004), the two-phase flow pressure drop isexpressed as

ΔpΔL

� �TP

¼ f1DH

G2

2ρTPð6Þ

Here DH and G are the hydraulic diameter of the microchannel andmass velocity of the fluid, respectively. The pressure drop of the

total channel path was less than 4 kPa, and is negligible incomparison with the channel outlet pressure of 1 atm. In addition,the volumes of the N2 bubble at the middle and exit of themicrochannel were also examined under the experimental condi-tions. The relative variation of bubble volumes ΔVN2/VN2 was lessthan 3%, illustrating that the pressure in the gas bubbles does notsignificantly change during the bubble-flowing stage. Therefore,premise (5) is also satisfied.

2.4.1. Mass transfer during bubble-flowing stageFig. 2(a) shows some typical images of the mass transfer

process during the bubble-flowing stage. The size of the sphericalbubbles decreased gradually covering a range from 86% to 54% ofthe microchannel width owing to CO2 mass transfer. In thisprocess, the mass transfer flux is expressed as

NCO2 ¼ �dnCO2

Adt¼ KLðCn

CO2�CCO2 Þ ð7Þ

Here, A is the surface area of the gas bubble. Cn

CO2, which is the

equilibrium concentration of CO2 in the liquid phase correspond-ing to the CO2 concentration in the gas phase, is obtained byHenry's law. According to premise (4), the CO2 concentration inthe liquid bulk phase CCO2 is approximately equal to zero. Thus,Eq. (7) can be rewritten as

NCO2 ¼ KLCn

CO2¼ KLHpCO2

ð8Þ

The solubility of CO2 in the MEA solution is assumed to beidentical to that in pure solvent. Here, H is the Henry constant ofCO2 in ethylene glycol obtained from Gui et al. (2011). pCO2

, whichis the partial pressure of CO2 in the gas bubble, is related to thetotal bubble pressure p as follows:

pCO2¼ p

VCO2

V¼ p

V�VN2

Vð9Þ

Here, V and VN2 represent the bubble volume and N2 partialvolume, respectively. In the experiments, the bubble stoppeddecreasing in size when it passed through 25 mm from thecapillary tip. Because the distance between the microchanneloutlet and capillary tip is almost 90 mm, the mass transfer processis likely complete near the channel outlet. As a result, the bubblevolume at the microchannel outlet is equal to the N2 partialvolume.

On the basis of the ideal gas law, the mass transfer flux can alsobe expressed as

NCO2 ¼ �dnCO2

Adt¼ � p

ARTdVdt

ð10Þ

The following equation is obtained by combining Eqs. (8), (9),and (10):

� pART

dVdt

¼ KLHpV�VN2

Vð11Þ

Integration of Eq. (11) yields

�Z V t

V i

VV�VN2

dV ¼ KLHRTZ t

0A dt ð12Þ

Table 1Physical properties at 28 1C.

Density, ρ (g/mL)Surface tension,γ (mN/m)

Viscosity, μ (mPa s)Diffusion coefficientof CO2 (m2/s)

17 vol% CO2–N2 1.27�10�3 1.71�10�2

37 vol% CO2–N2 1.39�10�3 1.58�10�2

56 vol% CO2–N2 1.50�10�3 1.49�10�2

5% MEA solution 1.105 50.5 15.6 0.185�10�9

15% MEA solution 1.096 50.3 17.4 0.170�10�9

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314308

Here, Vi is determined at the moment the bubble pinches off fromthe capillary tip; this moment corresponds to the beginning of thebubble-flowing stage. Vt and t denote the bubble volume andmoving time, respectively, when the bubble-flowing distancereaches 2.5 mm. By determining how the bubble volume changeswith time in the flowing stage, the overall mass transfer coefficientKL can be simply obtained according to Eq. (12).

2.4.2. Mass transfer during bubble-forming stageFig. 2(b) is a typical image of the mass transfer process during

the bubble-forming stage. It shows that the bubbles extend in axialdirection until they pinch off from the capillary tip and then shrinkto sphere. In this process, the mass transfer proceeded simulta-neously with bubble generation. Eqs. (7) and (8) remain valid, andthe following equation is obtained by integrating Eq. (8):Z tg

0AKLHpCO2

dt ¼ �ΔnCO2 ð13Þ

Here, tg is the bubble formation time and ΔnCO2 is expressed as

�ΔnCO2 ¼poðVN2=φ�V iÞ

RTð14Þ

where φ is the volume fraction of N2 in the initial gas mixture andpo is the channel outlet pressure. During bubble formation, thepartial pressure of CO2 in the gas bubble varied with time and it isgiven as

pCO2¼ pg

V�QGtφV

ð15Þ

The pressure of the dispersed phase shows no obvious changes formicrofluidics (Abate et al., 2012). Here, pg, which is considered asthe average pressure inside the gas bubble, is related to po basedon the mass balance

poVN2=φ¼ pgQGtg ð16ÞCombining Eqs. (13)–(16) yields

KLHZ tg

0AV�QGtφ

Vdt ¼ ðVN2=φ�V iÞQGtg

VN2=φRTð17Þ

According to Eq. (17), the overall mass transfer coefficient KL issimply calculated by recording the increase in the bubble dimen-sions at the bubble-forming stage.

Moreover, the contribution of mass transfer during the bubble-forming stage to the total mass transfer process is estimated as

η¼ VN2=φ�V i

VN2=φ�VN2

ð18Þ

3. Results and discussion

3.1. Mass transfer characteristics of bubble-flowing stage

In general, KL is primarily affected by the diffusion of thesubstance and the boundary layer thickness, and apparentlyimproved by the flow rate and chemical reaction (Y.H. Su et al.,2010). During the bubble-flowing stage, the flow field conditionaround or inside the bubble and the concentration of substancesare similar, especially in the case of chemical absorption. In this

Fig. 2. Typical mass transfer processes of bubbly flow in T-junction and co-flowing microchannels: (a) bubble-flowing stage and (b) bubble-forming stage. (QG¼75 μL/min,56% CO2; QL¼300 μL/min, 15% MEA.).

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314 309

section, the effects on the mass transfer characteristics aredescribed in terms of two aspects: the flow condition andchemical reaction.

3.1.1. Effects of phase flow ratesFig. 3 shows the effects of gas and liquid phase flow rates on the

mass transfer coefficient. The mass transfer coefficient increasesslightly as the two phase flow ratio increases; it shows little changewith variation in the gas flow rate. These results are explained by theobservation that in the liquid plug, the internal circulation flowderived from the interaction between the channel wall and contin-uous phase is intensified at a high liquid flow rate. Because theinternal circulation reduces the hydrodynamic boundary layer, theconcentration near the interface is renewed by the flow, resulting inthe enhancement of liquid-side mass transfer (Dessimoz et al., 2008).In contrast, the flow field conditions around or inside the bubblewere almost unaffected by the gas flow rate.

3.1.2. Effects of substance concentrationFig. 3 also shows the effects of MEA concentration on the mass

transfer coefficient. The overall mass transfer coefficient increaseswith the absorbent concentration. This is because the liquid-sidemass transfer coefficient was enhanced owing to the high-intensity chemical reaction. A well-known fact is that whenHatta number is greater than 3, the value of the enhancementfactor E, which represents the intensification effect of the chemicalreaction on the mass transfer process and is defined as kL/kL0, is

approximately equal to Ha (Danckwerts, 1970). According to Eqs.(3)–(5), the liquid-side mass transfer coefficient kL is proportionalto the MEA concentration. In other words, a high absorbentconcentration enhanced the liquid-side mass transfer process bypromoting chemical absorption.

Fig. 4 shows the effects of the initial CO2 volume fraction on themass transfer coefficient. The overall mass transfer coefficientshows no obvious change with the initial CO2 concentration,especially for the higher concentrations.

In this section, the effects of phase flow rate and substanceconcentration on the overall mass transfer coefficient during thebubble-flowing stage are discussed. The mass transfer coefficientsare mainly affected by the phase flow ratio and the absorbentconcentration. Therefore, the mass transfer process tends to becontrolled more by the liquid phase resistance.

3.2. Mass transfer characteristics of bubble-forming stage

Similarly, the effects on the mass transfer characteristics of thebubble-forming stage were investigated. In addition, the contribu-tion of mass transfer during the bubble-forming stage to the totalmass transfer process was also studied.

3.2.1. Effects of phase flow ratesFig. 5 shows the effects of gas and liquid phase flow rates

on the mass transfer coefficient. The mass transfer coefficient

Fig. 3. Effects of phase flow rate and absorbent concentration on the mass transfercoefficient at bubble-flowing stage: (a) co-flowing microchannel and (b) T-junctionmicrochannel.

Fig. 4. Effects of initial CO2 volume fraction on the mass transfer coefficient duringbubble-flowing stage: (a) co-flowing microchannel and (b) T-junction microchannel.

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314310

increases slightly with the two phase flow ratio. However, KL alsoincreases with the gas flow rate. This trend is different from thatduring the bubble-flowing stage. This is because the vortex flowpatterns in the gas bubble are not only derived from the relativeshearing motion between the two phases, but also from theinjection of the gas phase (Shao et al., 2012b; Xu et al., 2008).

3.2.2. Effects of substance concentrationFig. 5 also shows the effects of MEA concentration on the mass

transfer coefficient. The overall mass transfer coefficient shows noobvious change by varying the absorbent concentration. Fig. 6shows the effects of the initial CO2 volume fraction on the masstransfer coefficient. The overall mass transfer coefficient increasesmarkedly with the initial CO2 volume fraction. For a high-concentration absorption process, the effects of gas phase con-centration must be taken into account (kG�p/pN2mD/RTδ). Here, ahigh CO2 volume fraction reduced the mass transfer resistance inthe gas phase (Sobieszuk and Pohorecki, 2010).The investigationsabove on the effects of the phase flow rate and substanceconcentration on the overall mass transfer coefficient during thebubble-forming stage show that the mass transfer coefficients aremainly affected by the phase flow ratio, gas flow rate, andsubstance concentration of the gas phase. This differs from thebubble-flowing stage in that the mass transfer process tends to becontrolled more by the gas phase resistance.

3.2.3. Mass transfer contributions during bubble-forming stageThe contributions of mass transfer during the bubble-forming

stage to the total mass transfer process are listed in Table 2.Remarkably, the bubble formation time in the experiments wasless than 5 ms, and the bubble flow time was at least 20 times ofits formation time. However, even within such a short time, themass transfer contribution ratio η still exceeded 25%. Therefore,the contribution of mass transfer during the bubble-forming stagewas reasonably large given the extremely short mass transfer time.This characteristic, which is different from that of the traditionalmass transfer process of gas/liquid systems, may be caused by theflow conditions and microscaled bubble size. Table 2 also showsthat the mass transfer coefficient in the microchannels is nearly anorder of magnitude greater than that in a traditional contactor. Inparticular, owing to the small bubble size and large specific surface

Fig. 5. Effects of phase flow rate and absorbent concentration on the mass transfercoefficient at bubble-forming stage: (a) co-flowing microchannel and (b) T-junctionmicrochannel.

Fig. 6. Effects of initial CO2 volume fraction on the mass transfer coefficient duringbubble-forming stage: (a) co-flowing microchannel and (b) T-junction microchannel.

Table 2Comparison of mass transfer characteristics (mass transfer coefficient of bubble-flowing stage and mass transfer contribution of bubble-forming stage) for bubblyflow between microchannel and traditional gas/liquid contactor.

dav(mm)

KL

(10�4 m/s)KLa(s�1)

η(%)

Co-flowing microchannel 0.39–0.52 13–49 15–70 26–50T-junction microchannel 0.38–0.51 18–49 17–77 25–52Bubble column

(Charpentier, 1981)1–3 1–4 5�10�3–0.24 o10

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314 311

area, the volumetric mass transfer coefficient in the microchannelsis almost two orders of magnitude higher than that in a macro-scaled device.

3.3. Empirical correlations for predicting mass transfer coefficients

Classic correlations have been established to predict the convec-tive mass transfer processes under laminar flow conditions. Severaldimensionless numbers, such as Sh, Sc, and Re, are used to describethe effects of the physical properties and flow conditions. During thebubble-flowing stage, our results could be fitted with the same formSh–ReG

αReLβScγ. Here, the Sherwood number Sh is defined as KLdh/DHa,

and it has eliminated the effect of chemical reaction by removing theenhancement factor. The regression parameters are listed in Table 3,and Fig. 7 shows that the predicted values agree well with theexperimental results.

In the case of the bubble-forming stage, based on the previousanalysis, KL is proposed to be proportional to CacΦ. Here the capillary

number represents the dominant force during microscaled bubblegeneration and determines the bubble breakup (Fu et al., 2010). Thegas phase volumetric flow fraction Φ represents the effects of gasflow conditions. Here, KL

0 is defined as KL/(1�pN2/p), and is specifi-cally introduced to eliminate the effect of gas phase concentration. Inthe same way, the results were fitted with the regression parameterslisted in Table 3. Fig. 8 shows that the predicted values agree wellwith the experimental results.

Moreover, Table 3 reveals that during the bubble-flowing stage,there are few differences in the mass transfer characteristicsbetween the co-flowing and T-junction microchannels. In contrast,during the bubble-forming stage, marked differences between thetwo microchannels are evident by comparing the exponentialcoefficients of each dimensionless number. Owing to the impor-tance of the flow field in the bubble generation process, thedispersion structure of the microchannel can significantly influ-ence the mass transfer characteristics (Kashid et al., 2011b).

4. Conclusion

In this study, the mass transfer characteristics of bubbly flow inmicrochannels were systematically investigated. An online mea-surement method was successfully developed to obtain the overallmass transfer coefficient during the bubble-forming and bubble-flowing stages. On the basis of this method, the effects of phase flowrate and substance concentration on the mass transfer coefficient

Table 3Empirical correlation parameters for predicting mass transfer coefficients.

Flowing stage Forming stage

Co-flowing microchannel Sh¼C0ReG0.33ReL

0.50Sc0.33 KL0 ¼K0Φ

0.74Cac1.07

K0¼1.40 m/sT-junction microchannel C0¼9.6 KL

0 ¼K0Φ0.62Cac

1.09

K0¼1.27 m/s

Fig. 7. Prediction of overall mass transfer coefficients during bubble-flowing stage:(a) co-flowing microchannel and (b) T-junction microchannel.

Fig. 8. Prediction of overall mass transfer coefficients during bubble-forming stage:(a) co-flowing microchannel and (b) T-junction microchannel.

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314312

were investigated. The results showed that the mass transfercoefficient of the bubble-flowing stage in microchannels was nearlyan order of magnitude higher than that in a traditional contactor.The contribution of mass transfer at the bubble-forming stage wasreasonably large considering the extremely short mass transfertime. Moreover, empirical correlations were established to predictthe overall mass transfer coefficients, and comparisons betweendifferent bubble-generation methods were carried out as well. Thetheoretical analysis and prediction of the mass transfer coefficientwill be considered in our further work.

Nomenclature

KL overall mass transfer coefficient (m/s)KLa volumetric mass transfer coefficient (s�1)CCO2 CO2 concentration in liquid bulk phase (mol/L)CCO2n equilibrium concentration of CO2 in liquid phase (mol/L)kL0 liquid-side mass transfer coefficient of physical

absorption (m/s)kL liquid-side mass transfer coefficient (m/s)kG gas-side mass transfer coefficient (mol/ m2 � s �Pa)A surface area of gas bubble (m2)H Henry constant (mol/m3 Pa)dh hydraulic diameter of microchannel (mm)D diffusion coefficient of CO2 in liquid solution (m2/s)dav average bubble diameter (mm)u superficial velocity of gas or liquid phase (mm/s)

Dimensionless numbers

Ha Hatta number, Ha¼(k0DCO2)1/2/kL0E mass transfer enhancement factor, E¼kL/kL0Sh Sherwood number, Sh¼KLdh/DHaReL Reynolds number of liquid phase, ReL¼ρLuLdh/μL

ReG Reynolds number of gas phase, ReG¼ρGuGdh/μG

Sc Schmidt number, Sc¼μ/ρDCac capillary number of continuous phase, Cac¼μuc/γ

Greek letters

φ volume fraction of N2 in initial gas mixtureη mass transfer contribution ratio of bubble-forming stageδ liquid film thickness (mm)ρ density of gas or liquid phase (g/mL)μ viscosity of gas or liquid phase (mPa � s)γ surface tension (mN/m)Φ gas phase volumetric flow fraction, Φ¼QG/(QGþQL)

Acknowledgment

We would like to acknowledge the National Natural ScienceFoundation of China (21036002, 91334201) and SRFDP(20090002110070) for supporting this study.

Appendix A. Supplementary material

Supplementary data associated with this article can be found inthe online version at http://dx.doi.org/10.1016/j.ces.2014.02.004.

References

Abate, A.R., Mary, P., van Steijn, V., Weitz, D.A., 2012. Experimental validation ofplugging during drop formation in a T-junction. Lab Chip 12 (8), 1516–1521.

Abolhasani, M., Singh, M., Kumacheva, E., Gunther, A., 2012. Automated micro-fluidic platform for studies of carbon dioxide dissolution and solubility inphysical solvents. Lab Chip 12 (9), 1611–1618.

Alvarezfuster, C., Midoux, N., Laurent, A., Charpentier, J.C., 1981. Chemical-kineticsof the reaction of CO2 with amines in pseudo m-nth order conditions in polarand viscous organic solutions. Chem. Eng. Sci. 36 (9), 1513–1518.

Aoki, N., Tanigawa, S., Mae, K., 2011. Design and operation of gas–liquid slug flow inminiaturized channels for rapid mass transfer. Chem. Eng. Sci. 66 (24),6536–6543.

Charpentier, J.C., 1981. Mass transfer rates in gas–liquid absorbers and reactors,Advances in Chemical Engineering. Academic Press, New York, pp. 1–133

Danckwerts, P.V., 1970. Gas–Liquid Reactions. McGraw-Hill, New York.Dessimoz, A.L., Cavin, L., Renken, A., Kiwi-Minsker, L., 2008. Liquid–liquid two-

phase flow patterns and mass transfer characteristics in rectangular glassmicroreactors. Chem. Eng. Sci. 63 (16), 4035–4044.

Dubois, L., Thomas, D., 2009. CO2 absorption into aqueous solutions of mono-ethanolamine, methyldiethanolamine, piperazine and their blends. Chem. Eng.Technol. 32 (5), 710–718.

Fu, T.T., Ma, Y.G., Funfschilling, D., Zhu, C.Y., Li, H.Z., 2010. Squeezing-to-drippingtransition for bubble formation in a microfluidic T-junction. Chem. Eng. Sci. 65(12), 3739–3748.

Gui, X., Tang, Z., Fei, W., 2011. Solubility of CO2 in alcohols, glycols, ethers, andketones at high pressures from (288.15 to 318.15) k. J. Chem. Eng. Data 56 (5),2420–2429.

Gunther, A., Jensen, K.F., 2006. Multiphase microfluidics: from flow characteristicsto chemical and materials synthesis. Lab Chip 6 (12), 1487–1503.

Huebner, A., et al., 2008. Microdroplets: a sea of applications? Lab Chip 8 (8),1244–1254.

Kashid, M.N., Renken, A., Kiwi-Minsker, L., 2011a. Gas-liquid and liquid–liquid masstransfer in microstructured reactors. Chem. Eng. Sci. 66 (17), 3876–3897.

Kashid, M.N., Renken, A., Kiwi-Minsker, L., 2011b. Influence of flow regime on masstransfer in different types of microchannels. Ind. Eng. Chem. Res. 50 (11),6906–6914.

Kobayashi, J., et al., 2004. A microfluidic device for conducting gas–liquid–solidhydrogenation reactions. Science 304 (5675), 1305–1308.

Kobayashi, J., Mori, Y., Kobayashi, S., 2006. Multiphase organic synthesis inmicrochannel reactors. Chem.—Asian J. 1 (1–2), 22–35.

Leclerc, A., et al., 2008. Gas–liquid selective oxidations with oxygen under explosiveconditions in a micro-structured reactor. Lab Chip 8 (5), 814–817.

Li, W., et al., 2012. Microfluidic study of fast gas–liquid reactions. J. Am. Chem. Soc.134 (6), 3127–3132.

Mae, K., 2007. Advanced chemical processing using microspace. Chem. Eng. Sci. 62(18–20SI), 4842–4851.

Park, S.W., Lee, J.W., Choi, B.S., Lee, J.W., 2005. Kinetics of absorption of carbondioxide in monoethanolamine solutions of polar organic solvents. J. Ind. Eng.Chem. 11 (2), 202–209.

Roudet, M., Loubiere, K., Gourdon, C., Cabassud, M., 2011. Hydrodynamic and masstransfer in inertial gas–liquid flow regimes through straight and meanderingmillimetric square channels. Chem. Eng. Sci. 66 (13), 2974–2990.

Sada, E., Kumazawa, H., Han, Z.Q., Matsuyama, H., 1985. Chemical-kinetics of thereaction of carbon-dioxide with ethanolamines in nonaqueous solvents. AIChEJ. 31 (8), 1297–1303.

Seemann, R., Brinkmann, M., Pfohl, T., Herminghaus, S., 2012. Droplet basedmicrofluidics. Rep. Prog. Phys. 75, 0166011.

Shao, H.W., Lu, Y.C., Wang, K., Luo, G.S., 2012a. Liquid–liquid flow and mass transfercharacteristics in micro-sieve array device with dual-sized pores. Chem. Eng. J.193, 96–101.

Shao, H.W., Lu, Y.C., Wang, K., Luo, G.S., 2012b. An experimental study of liquid–liquid microflow pattern maps accompanied with mass transfer. Chin. J. Chem.Eng. 20 (1), 18–26.

Sobieszuk, P., Pohorecki, R., 2010. Gas-side mass transfer coefficients in a falling filmmicroreactor. Chem. Eng. Process. 49 (8), 820–824.

Sobieszuk, P., Pohorecki, R., Cyganski, P., Grzelka, J., 2011. Determination of theinterfacial area and mass transfer coefficients in the Taylor gas–liquid flow in amicrochannel. Chem. Eng. Sci. 66 (23), 6048–6056.

Su, H.J., et al., 2010. Mass transfer characteristics of H2S absorption from gaseousmixture into methyldiethanolamine solution in a T-junction microchannel. Sep.Purif. Technol. 72 (3), 326–334.

Su, Y.H., Zhao, Y.C., Chen, G.W., Yuan, Q.A., 2010. Liquid–liquid two-phase flow andmass transfer characteristics in packed microchannels. Chem. Eng. Sci. 65 (13),3947–3956.

Tan, J., Lu, Y.C., Xu, J.H., Luo, G.S., 2012a. Mass transfer performance of gas–iquidsegmented flow in microchannels. Chem. Eng. J. 181, 229–235.

Tan, J., Lu, Y.C., Xu, J.H., Luo, G.S., 2012b. Mass transfer characteristic in theformation stage of gas–liquid segmented flow in microchannel. Chem. Eng. J.185, 314–320.

Tan, J., Shao, H.W., Xu, J.H., Lu, Y.C., Luo, G.S., 2011. Development of a membranedispersion micro-absorber for CO2 capture. J. Membr. Sci. 385 (1–2), 123–131.

Tan, J., Zhang, J.S., Lu, Y.C., Xu, J.H., Luo, G.S., 2012. Process intensification of catalytichydrogenation of ethylanthraquinone with gas–liquid microdispersion. AIChE J.58 (5), 1326–1335.

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314 313

Vaidya, P.D., Kenig, E.Y., 2007. CO2-alkanolamine reaction kinetics: a review ofrecent studies. Chem. Eng. Technol. 30 (11), 1467–1474.

Versteeg, G.F., Van Dijck, L., Van Swaaij, W., 1996. On the kinetics between CO2 andalkanolamines both in aqueous and non-aqueous solutions. An overview.Chem. Eng. Commun. 144, 113–158.

Wang, J.T., Wang, J., Han, J.J., 2011. Fabrication of advanced particles and particle-based materials assisted by droplet-based microfluidics. Small 7 (13),1728–1754.

Wang, K., Lu, Y.C., Xu, J.H., Luo, G.S., 2009. Determination of dynamic interfacialtension and its effect on droplet formation in the t-shaped microdispersionprocess. Langmuir 25 (4), 2153–2158.

Xu, J.H., Dong, P.F., Zhao, H., Tostado, C.P., Luo, G.S., 2012. The dynamic effects ofsurfactants on droplet formation in coaxial microfluidic devices. Langmuir 28(25), 9250–9258.

Xu, J.H., Li, S., Chen, G.G., Luo, G.S., 2006. Formation of monodisperse microbubblesin a microfluidic device. AIChE J. 52 (6), 2254–2259.

Xu, J.H., Tan, J., Li, S.W., Luo, G.S., 2008. Enhancement of mass transfer performanceof liquid–liquid system by droplet flow in microchannels. Chem. Eng. J. 141 (1-3), 242–249.

Yue, J., Chen, G.W., Yuan, Q., 2004. Pressure drops of single and two-phase flowsthrough T-type microchannel mixers. Chem. Eng. J. 102 (1), 11–24.

Yue, J., Luo, L.G.., Gonthier, Y., Chen, G.W., Yuan, Q., 2009. An experimental study ofair-water Taylor flow and mass transfer inside square microchannels. Chem.Eng. Sci. 64 (16), 3697–3708.

L. Yang et al. / Chemical Engineering Science 109 (2014) 306–314314