You are on page 1of 14

Chemical Engineering Science 58 (2003) 3793 – 3806

www.elsevier.com/locate/ces

E ects of surface-active agents on mass transfer of a solute into single


buoyancy driven drops in solvent extraction systems
Xiaojin Lia;1 , Zai-Sha Maoa;∗ , Weiyang Feib
a Institute of Process Engineering, Chinese Academy of Sciences, Beijing 100080, China
b State Key Lab of Chemical Engineering, Tsinghua University, Beijing 100084, China
Received 25 June 2002; received in revised form 18 November 2002; accepted 25 February 2003

Abstract

Numerical simulation was made of transient mass transfer to a surfactant contaminated buoyancy-driven drop controlled by appreciable
resistance in both liquid phases. For this purpose, the momentum equations were formulated and solved in a boundary-7tted orthogonal
coordinate system. On the basis of resolved hydrodynamics of the contaminated drop, the transient mass transfer was formulated and
solved in the same coordinate system. In order to check the applicability of the numerical scheme, single drop extraction experiments
were conducted in a totally closed droplet 7le column with the terminal e ect e;ciently eliminated. The MIBK–acetic acid–water system
was used with small quantities of SDS (sodium dodecyl sulphate), Triton X-100, or Tween 80 introduced into the continuous phase. For
these experimental cases, the @ow 7eld and the drag coe;cient of a contaminated drop were simulated 7rst. The numerical prediction
of the drag coe;cient is found in good agreement with the corresponding experimental data. It illustrates that the behavior of a drop
approaches that of a rigid sphere and that about 100 times higher bulk concentration of SDS than that of Triton X-100 is required for
the same extent contamination of a MIBK drop of the same size. Then the information of the @ow 7eld of a contaminated MIBK drop
was used in simulating the transient mass transfer of solute into the drop. The resulted extraction fraction and overall mass transfer
coe;cient are in reasonable coincidence with the experimental data. Both numerical results and experimental data show that overall mass
transfer coe;cient of a heavily contaminated drop is only about one third of that in the pure system. This can be explained well by the
distribution of the local Sherwood number, which drops down abruptly along the rear stagnant surface. Also the interfacial resistance of
adsorbed surfactant was incorporated in the mass transfer model and then estimated by the least square 7tting the simulation with data.
The numerical results also show that Tween 80 presents obvious interfacial resistance on the acetic acid di using across the interface,
whereas SDS and Triton X-100 show no interfacial resistance. It is suggested that the numerical simulation can be resorted in some
solvent extraction systems containing surfactants to conduct numerical experiments and parametric study.
? 2003 Elsevier Ltd. All rights reserved.

Keywords: Surfactant; Drag coe;cient; Solvent extraction; Single drop; Numerical simulation; Mass transfer experiment

1. Introduction of surface-active contaminants may have profound e ect on


the behavior of drops and bubbles. A contaminant can elim-
Mass transfer to and from drops moving in a continuous inate internal circulation, signi7cantly increase the drag and
phase is common to many industrial processes, especially reduce mass transfer rate. The study of mass transfer of a
for the unit operation of liquid–liquid extraction. In most in- solute to (or from) a single drop with adsorbed surfactant
dustrial extraction equipment, unavoidable trace quantities on the interface provides an essential base for better under-
standing of liquid extraction process and for scienti7c design
of the equipment. Many theoretical and experimental studies
have been devoted to analyzing the e ect of surface-active
∗ Corresponding author. Tel: +86-10-6255-4558; agents on the single-drop mass transfer. However, these em-
fax: +86-10-6256-1822. pirical and idealized theoretical equations do not provide in
E-mail addresses: xiaojin@fe.up.pt (X. Li), zsmao@home.ipe.ac.cn general reliable predictions in agreement with experimental
(Z.-S. Mao).
1 Present address: Laboratory of Separation and Reaction Engineering, data, since it is a di;cult and complicated job to account for
Department of Chemical Engineering, Faculty of Engineering, Porto Uni- the roles played by surface-active agents in a quantitative
versity, 4200-465 Porto, Portugal. sense.

0009-2509/$ - see front matter ? 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0009-2509(03)00237-9
3794 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

Decrease in mass transfer rate observed for these sin- continuous phase to single drops in several extraction sys-
gle drop systems contaminated by surface-active agents tems. Li, Mao, Chen, and Fei (2002) studied the process
is attributed mainly to two mechanisms: hydrodynamic of transient mass transfer to a single drop with appre-
and physico-chemical ones. The hydrodynamic e ect in- ciable resistance in both liquid phases numerically. For
cludes reduction of internal circulation velocities to near surfactant-containing systems, Takemura and Yabe (1999),
zero even for large drops, decrease of the terminal veloc- Ponoth and McLaughlin (2000) reported the numerical
ity and limitation to interface convection due to adverse simulation of mass transfer to rising bubbles. In their work,
interfacial tension gradient. Hydrodynamic information on however, the mass transfer inside the bubble was not ac-
a single drop whether or not contaminated by surfactant counted for and the contribution of internal circulation was
is always a prerequisite for estimation of interfacial mass ignored. No numerical work seems to be available on tran-
transfer. Since Frumkin and Levich introduced the concept sient mass transfer to a contaminated drop moving steadily
of interfacial tension gradient along the surface of the drop with mass transfer resistance in both phases.
(Levich, 1962), many theoretical works (Gri;th, 1962; The e ect of surfactant on mass transfer coe;cient of sin-
Ruckenstein, 1964; Schechter & Farley, 1963; Lochiel, gle drops has been widely studied experimentally (Garner &
1965; Clift, Grace, & Weber, 1978; Brunn & Isemin, 1984; Hale, 1953; Thorsen & Terjesen, 1962; Huang & Kintner,
Quintana, 1990; Ramirez & Davis, 1999) adopted this con- 1969; Mekasut, Molinier, & Angelino, 1979; Slater, Baird
cept to examine the e ect of surfactant on transfer rate of & Liang, 1988; Temos, Pratt, & Stevens, 1996). In these
single bubbles or drops in steady motion driven by buoy- experimental studies, the droplet 7le column and solvent ex-
ancy. Gri;th (1962) has used a modi7cation of Savic’s traction systems with mass transfer resistance in both phases
stream function (Savic, 1953) for a contaminated drop to were mostly employed. For example, Temos et al. (1996)
estimate the mass transfer rate to or from a viscous sphere chose the typical ternary MIBK–acetic acid–water system
moving in creeping @ow. Clift et al. (1978) and Quintana which has approximately equal mass transfer resistances in
(1990) predicted the mass transfer rate to a bubble when a both phases and employed a droplet 7le column to obtain
stagnant cap developed. The physico-chemical e ect means mass transfer coe;cients separately for two phases and also
surface blocking or interaction between surfactant and so- overall mass transfer coe;cient.
lute molecules. They are usually referred to as an interfacial In this paper, the mass transfer coe;cient of single drops
resistance or barrier. Goodridge and Robb (1965) used was examined numerically and experimentally as a function
an energy barrier to characterize the e ect. In a numbers of drop size and extent of contamination by surfactant. The
of classic experiments, Langmuir and Schaefer (1943), @ow 7eld for a single drop a ected by surfactant was sim-
Archer and La Mer (1955) quanti7ed the extent of mass ulated numerically and then used to calculate the transient
transfer reduction arising from the interfacial steric hin- mass transfer to single drops with appreciable resistance in
drance as a function of surfactant 7lm thickness and alkyl both phases. The process of transient mass transfer to a sin-
chain length. gle drop with resistance in both phases was mathematically
Numerical simulation of mass transfer requires the hy- formulated and numerically solved in the boundary-7tted
drodynamic detail on a single drop moving steadily in orthogonal coordinate system. Moreover, the e ect of in-
an in7nite immiscible @uid medium. Ryskin and Leal terfacial resistance of adsorbed surfactant on mass trans-
(1984) simulated the steady buoyancy-driven motion of fer was also taken into account through suitable boundary
deformable bubbles at intermediate Reynolds numbers by conditions. A series of experiments were conducted in the
the stream function–vorticity formulation in an orthogonal MIBK–acetic acid–water system containing known amount
boundary-7tted coordinate system. Dandy and Leal (1989) of SDS or Triton X-100 in the continuous phase. The ex-
extended this technique to the case of liquid drops with con- perimental cases were simulated successfully to verify the
siderable deformation. Later, many researchers (Leppinen, validity of numerical procedure and reasonable agreement
Renksizbulut, & Heywood, 1996; Bel Fdhila & Duineveld, was obtained.
1996; McLaughlin, 1996; Cuenot, Magnaudet, & Spennato,
1997) successfully simulated the motion of a single bub-
ble contaminated by surface-active agents. Cuenot et al. 2. Formulation of mass transfer to single drop
(1997) developed a successful technical route for numeri-
cal simulation of @uid @ow associated with single bubbles The typical situation of mass transfer to a single drop is
contaminated by impurity. Li and Mao (2001) evaluated that a deformable drop rises or falls steadily at its terminal
numerically the e ects of surfactant on the motion of a velocity U in an immiscible liquid medium. The following
single drop driven by buoyancy in an in7nite immiscible assumptions are considered reasonable (Mao et al., 2001; Li
liquid medium. et al., 2002):
With the hydrodynamic information of a single drop, nu-
merical simulation of steady and transient mass transfer of a (1) Both liquids are viscous, incompressible and New-
single bubble or drop has been carried out. Mao, Li, & Chen tonian, the @ow in both phases is axisymmetric and
(2001) simulated the external mass transfer from the laminar;
X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806 3795

(2) The physical properties of both @uids are constant and With subscripts 1 inserted into Eq. (2) to index the external
the in@uence of mass transfer on physical properties is domain and Eq. (3) incorporated, we have
negligible;  
@c1 1 @ 1 @c1 @ 1 @c1
(3) The surfactant is only soluble in the continuous phase − −
and not soluble in the droplet, and the surfactant has no @t y1 h 1 h
1 @
1 @ 1 @ 1 @
1
e ect on the bulk physical properties, such as viscosity,   
D1 @ h
1 y1 @c1
density and so on. Due to the constancy of physical =
properties of both liquid phases and the interface, the y1 h 1 h
1 @ 1 h 1 @ 1
solution of mass transfer can be decoupled from the  
problem of @uid @ow. @ h 1 y1 @c1
+ : (4)
@
1 h
1 @
1
A numerical approach on the e ect of surfactant on the
motion of a buoyancy-driven drop at intermediate Reynolds In the present numerical procedure, the governing equations
numbers has been described in detail in our previous work are non-dimensionalized and the following non-dimensional
(Li & Mao, 2001). In the following, the numerical proce- variables and groups are de7ned for this purpose:
dure for mass transfer to a single contaminated drop with
c1 1 h 1
reasonable resistance in both phases is brie@y described and C1 = ; 1 = ; H 1 = ;
c1∞ R2 U R
the contribution of the interfacial resistance on mass transfer
is also estimated numerically. The simulation is focused on h
1 y1 2RU
the mass transfer of small solute molecules across the drop H
1 = ; Y1 = ; Pe1 = :
R R D1
surface, while that of larger surfactant molecules from the
bulk onto the interface has been incorporated into the nu- Among them Pe is the Peclet number, symbolizing the rel-
merical simulation of @uid @ow of the contaminated drops. ative strength of convection to molecular di usion. Finally,
the non-dimensional convective di usion equation for the
external region is
2.1. Transient convective di4usion equation
    
Pe1 @C1 Pe1 @ @1
In general, the transient mass transfer of a solute to or H 1 H
1 Y1 + − C1
2 @ 2 @ 1 @
1
from a single drop is governed by the convective di usion
equation in the general form,  
@ @1
@c + C1
+ u · ∇c = D∇2 c (1) @
1 @ 1
@t     
@ @C1 @ Y1 @C1
the solution of which is relied on the resolved @uid @ow in = f 1 Y1 + ; (5)
the drop and the continuous phase as addressed by Li and @ 1 @ 1 @
1 f1 @
1
Mao (2001). For an axisymmetric drop moving in an in7- where the distortion function is de7ned as the ratio of two
nite immiscible stagnant liquid, the motion and mass transfer scale factors:
may be formulated in a two-dimensional orthogonal curvi-
linear coordinate system. The external and internal domains h
1 ( ;
) H
1 ( ;
)
f1 ( ;
) = = ;
of a drop in the physical plane (x; y) may be transformed h 1 ( ;
) H 1 ( ;
)
respectively into a unit square in the computational plane
H
2 ( ;
)
( ;
), as shown in Fig. 1. This task of grid generation has f2 ( ;
) =
H 2 ( ;
)
been resolved by Ryskin and Leal (1983). In our case, the
expanded form of Eq. (1) in such an orthogonal grid reads and used in the numerical generation of orthogonal grids
@c u @c u
@c for the exterior and interior of the drop to be simulated. In
+ + this work, a large enough region for the exterior phase was
@t h @ h
@

     chosen and the 7nite domain distortion function was applied


D @ h
y @c @ h y @c (Mao & Chen, 1997):
= + : (2)
h h
y @ h @ @
h
@

f1 ( 1 ;
1 ) = (1 − 0:5 cos 
1 ):

In the left-hand ( 1 ;
1 ) coordinate system for the external
domain as sketched in Fig. 1, the velocity components along Here  is a parameter for specifying the position of the re-
curvilinear coordinate axes are related to the stream function mote boundary, and the outer radius R∗ = 20R corresponds
1 by to  = 2:5. For the interior phase, the distortion function pro-
vided by Ryskin and Leal (1984) was used:
1 @ 1 1 @ 1
u 1 = − ; u
1 = : (3)
y1 h
1 @
1 y1 h 1 @ 1 f2 ( 2 ;
2 ) =  2 (1 − 0:5 cos 
2 ):
3796 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

Fig. 1. The correspondence of exterior and interior of a deformable drop with the computational domain: (a) external domain; (b) drop interior; and
(c) unit square in ( ;
).

The non-dimensional equation for the drop phase in the of solute at the remote in7nity c1∞ is assumed to be a con-
right-hand reference frame is similar: stant, which is then used as the non-dimensionalizing factor
    
Pe2 @C2 Pe2 @ @2 for the concentration 7elds. If the interfacial resistance for
H 2 H
2 Y2 + C2 the adsorbed surface active agent is ignored, thermodynamic
2 @ 2 @ 2 @
2
equilibrium, Eq. (11), will exist for solute concentration at
  two sides of the interface. Thus, the following initial and
@ @2 boundary conditions may be adopted for mass transfer from
+ − C2
@
2 @ 2 the continuous phase to the drop (c → d):
     I:C:1 C1 ( 1 ;
1 ; 0) = 1 ( = 0); (8)
@ @C2 @ Y2 @C2
= f 2 Y2 + : (6)
@ 2 @ 2 @
2 f2 @
2
I:C:2 C2 ( 2 ;
2 ; 0) = 0 ( = 0); (9)
For the drop phase, the dimensional velocity components are
1 @ 2 1 @ 2 D1 @C1 D2 @C2
u 2 = ; u
2 = − (7) B:C:1 =− ;
y2 h
2 @
2 y2 h 2 @ 2 h 1 @ 1 h 2 @ 2

and the relevant non-dimensional variables are (@ux continuity at the drop surface;
c2 2 h
C2 = ∞ ; 2 = 2 ; H 2 = 2 ; 1 = 2 = 1); (10)
c1 R U R
h
2 y2 2RU B:C:2 C2 = mC1
H
2 = ; Y2 = ; Pe2 = :
R R D2 (extraction equilibrium at the drop surface;

1 = 2 = 1); (11)
2.2. Initial and boundary conditions
B:C:3 C1 ( 1 ;
1 ; ) = 1
For the situation of solute mass transfer with resistance re-
siding in both continuous and drop phases, the concentration (at the remote boundary; 1 = 0); (12)
X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806 3797

B:C:4 C2 continuous and 7nite and the former is calculated by


11
(at the drop center;
2 = 2 = 0); (13) C2 ( 2 ;
2 ; )h 2 h
2 y2 d 2 d
2
Q
C 2 () = 0 0  1  1 :
h h y d 2 d
2
0 0 2
2 2
@C1 @C2
B:C:5 = =0 In order to evaluate the mass transfer coe;cient with a
@
1 @
2
drop as a whole, it is necessary to calculate 7rst the total
(on the axis of symmetry;
1 =
2 = 0): (14) instant di usive @ux N , which is actually the sum of local
transfer @ux across the interface,
If there is any interfacial resistance caused by added sur- 
face active agents in a liquid–liquid extraction system, the D1 @c1 
nloc = −
solute concentration di erence, (mc1S − c2S ), will prevail at h @ 1 
1 1 =1
the interface. The resistance can be evaluated from the con-
over the drop surface:
centrations in the vicinity of the interface during mass trans- 1
fer, which is a direct indication of the barrier mechanism. On
N = 2 nloc y1 h
1 d
1
the other hand, when the drop surface is covered by the ad- 0
sorbed surfactant, the mass transfer area is simply assumed 
to be (1 − ) of interfacial area. Then the total instant dif-
1
@c1 
= −2D1 f 1 y1 d
1 : (19)
fusion @ux over the interface can be formulated as 0 @ 1  1 =1

n = ks (1 − )(mc1S − c2S ): (15) If the overall mass transfer coe;cient kod is de7ned based
on the average drop concentration (that is practically the
In this equation, the interfacial mass transfer coe;cient, ks , only available measure of solute concentration for drops
was introduced, which can be evaluated later by the least in most experimental investigations), it would be related
square 7tting. The following boundary conditions can be with the local concentration gradient at the interface by the
obtained on the local @ux conservation of the solute through conservation of mass @ux over the interface by
the surfactant 7lm: 1 
@c1 
D1 @c1 D2 @c2 kod S(mc1∞ − cQ2 ) = −2D1 f1 y 1 d
1 : (20)
=− = ks (1 − )(mc1S − c2S ): (16) 0 @ 1  1 =1
h 1 @ 1 h 2 @ 2
Since the solute concentration at the in7nity is always kept
Thus, Eq. (16) can be used to substitute Eqs. (10) and (11) to
to be C1∞ = 1 and the total surface area of a drop is
solve the convective di usion equations, and the joint e ect 1
of hydrodynamics and energy barrier on mass transfer can be
S = 2 y1 h
1 d
1 (21)
evaluated numerically. It can be shown by variable transform 0
that the characteristic value of Shod from the simulation is
then the overall mass transfer coe;cient is expressed as
the same, no matter whether the direction of mass transfer 
1 @C1 
is c → d or d → c. f y d
1
D1 0 1 1 @ 1 
1 =1
kod = − 
m − CQ 2 1
y1 h
1 d
1
2.3. Estimation of extraction fraction and mass transfer 0
coe7cient 1 
@C1 
f 1 Y 1 @ 1  d
1
D1 0 1 =1
The extraction fraction Em is a variable useful for mon- =−  (22)
R(m − CQ 2 ) 1
Y1 H
1 d
1
itoring the temporal extraction process. It is de7ned as the 0

ratio of current mass of solute extracted into the drop over and the Sherwood number is
the maximum solute extraction: 1 
@C1 
f 1 Y 1 @ 1  d
1
CQ out − CQ in 2Rkod 2 0 1 =1
Em = 2∗ 2
(17) Shod = =−  : (23)
C − CQ in
D1 m − CQ 2 1
Y1 H
1 d
1
2 2 0

in which C2∗ is the drop concentration in equilibrium with On the other hand, the overall mass transfer coe;cient kod
C1∞ , i.e. C2∗ = mC1∞ , and superscripts ‘in’ and ‘out’ indi- may be evaluated from the overall solute conservation based
cate two successive locations where the average drop con- on the drop as follows:
centration was experimentally determined, In our numerical N d CQ 2
investigation with Eq. (9) as the initial condition Em is sim- ∞ = kod (C2∗ − CQ 2 )S = V : (24)
c1 dt
ply the average concentration in the drop phase divided by
the distribution coe;cient m: Integrate the above equation gives
CQ 2 () V C ∗ − CQ out
Em = (18) kod = − ln 2∗ 2
; (25)
m S(tout − tin ) C − CQ in 2 2
3798 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

where of acetic acid between two phases is close to unity, so the


4 3 resistance to mass transfer lies in the both phases.
V ≈ R ; S ≈ 4R2 : To minimize the experimental error due to variable
3
terminal e ect occurring at the main interface at the
The surface area is approximate due to the drop deformation. collecting funnel, a totally closed experimental column
If the drop assumes spherical shape or nearly spherical, the (?50 mm; 100 cm high) was designed as sketched in Fig. 2
above equation becomes (Li, Mao, & Chen, 2001). The key point is that the new
R C ∗ − CQ out extraction column vents to the atmosphere only through the
kod = − ln 2∗ 2
: (26)
3(tout − tin ) C − CQ in drop phase collecting tubing. Thus, when a drop is forced
2 2
into the column by a precision injection pump, equal amount
When the time is dimensioned by R=U , Eq. (26) reads of dispersed phase must be displaced out of the column.
U C ∗ − CQ out This guarantees that the interface can be easily controlled at
kod = − ln 2∗ 2
: (27) the conic base of the collecting funnel with minimized mass
3(out − in ) C − CQ in
2 2
transfer to the drop phase after coalescence. For this rising
2.4. Numerical solution of mass transfer equation drops, experiments were carried out in the following proce-
dure. First, the coalescent interface was adjusted at the base
To solve PDEs (5) and (6) numerically, the control vol- of the conic collecting funnel for collecting drops. Then,
ume formulation described by Patankar (1980) was adopted single drops with speci7ed concentration of solute were al-
with satisfactory results, especially when the convective lowed to form and released from the nozzle. Droplet rising
mass transfer played more important role as Pe increased. velocity was measured with a stopwatch between two points
In this work, the convective term was discretized by the spaced 30 –40 cm apart, at least 20 –30 cm from the nozzle
power-law scheme. More details will be found in Mao et al. tip to guarantee the drops reaching its terminal velocity. The
(2001) and Li et al. (2002). Numerical tests were conducted coalesced drops @owed automatically out of the column via
to con7rm that the mesh size and time step showed little ef- the collecting funnel and samples were collected for chemi-
fect on the numerical solution, and it was determined that the cal analysis of the acetic acid concentration. Repeat experi-
computation conducted on a 81 × 81 grid with S 6 0:01 ments at several di erent drop collecting locations, then the
became su;ciently accurate. contribution in the stages of drop formation and accelerated
motion can be eliminated by using the solute concentration
at the 7rst location as the time origin 7r calculating kod , and
3. Single drop extraction experiments with surfactant in the mass transfer during the steady state motion was quanti-
the continuous phase 7ed. After the experimental runs with pure water completed,
repeat the new runs with speci7ed concentration of SDS,
3.1. Experimental apparatus and procedure Triton X-100 or Tween 80 solution. Thus, experiments with
di erent surfactant concentrations provided the information
In order to check the applicability of the developed on e ect of surface-active agents on the motion and mass
numerical scheme, transfer of acetic acid from aque- transfer rate.
ous continuous phase to rising MIBK drops was con-
ducted in a droplet 7le column, and the surfactants used 3.2. Drag coe7cient
were sodium dodecyl sulphate (SDS), Triton X-100
and Tween 80. The corresponding physical parame- Table 2 gives the terminal velocities of MIBK drops mov-
ters for the surfactants are: the maximum adsorption, ing in pure water or in SDS solution. The drag coe;cient
∞
= 1:4 × 10−6 mol m−2 , the Langmuir-von Szyszkowski can be calculated by
constant, a = 7:5 × 10−2 mol m−3 for SDS (Bel Fdhila  
4 dg
& Duineveld, 1996) and ∞ 
= 2:91 × 10−6 mol m−2 , CD = (1 − ,); (28)
 −4 −3 3 U2
a = 6:6 × 10 mol m for Triton X-100 separately (Lin,
McKeigue, & Maldarelli, 1990). The physical properties where g is gravitational acceleration, and , the density ratio
for MIBK–acetic acid–water system are listed in Table 1 of the drop to the continuous phase. The drag coe;cient
(Misek, 1978). In this system, the distribution coe;cient from the experiments is plotted as a function of Reynolds

Table 1
Physical properties of MIBK–acetic acid–water system (20◦ C)

Disp. Solute Cont. Trans. (d × 103 (c × 103 )d × 103 )c × 103 Dd × 109 Dc × 109 Cd0 Cc0 * × 103 m
phase phase direct. (kg=m3 ) (kg=m3 ) (Pa s) (Pa s) (m2 =s) (m2 =s) (wt%) (wt%) (N/m)

MIBK HAc water c→d 0.8155 1.004 0.60 1.221 1.44 0.91 0 2.25 8.05 0.64
X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806 3799

Fig. 2. Sketch of the experimental setup.

Table 2
In@uence of SDS on the mass transfer of MIBK(d) –acetic acid(s) –water(c) system (solute transfer direction:c → d; d = 1:56 mm; c1∞ = 2:25 wt%; cd0 =
0 wt%)

Serial no. Position Distance t cd; i E kod × 105 St ∗ × 105


kod
(cm) (s) (wt%) (m/s) (s) (m/s)

MS5-0 cSDS = 0 mol= m3 U = 8:33 cm= s


1 10 1.3 0.61 0 — 0 —
2 30 3.8 1.00 0.470 6.88 2.5 6.88
3 50 6.2 1.12 0.614 5.16 4.9 3.45
4 70 8.5 1.21 0.723 4.63 7.2 3.58
5 90 11.0 1.26 0.783 4.13 9.7 2.66
MS5-1 cSDS = 1:26 × 10−3 mol= m3 U = 8:00 cm= s
1 10 1.4 0.63 0 — 0 —
2 30 4.1 0.93 0.375 4.81 2.7 4.81
3 50 6.7 1.03 0.496 3.54 5.3 2.23
4 70 9.3 1.11 0.598 3.11 7.9 2.25
5 90 11.6 1.17 0.666 2.85 10.2 2.01
MS5-2 cSDS = 2:52 × 10−3 mol= m3 U = 7:54 cm= s
1 10 1.4 0.66 0 0 —
2 30 4.1 0.92 0.333 3.47 2.7 3.47
3 50 6.8 1.03 0.474 2.75 5.4 1.86
4 70 9.4 1.08 0.538 2.21 8.0 1.51
5 90 12.2 1.14 0.615 2.04 10.8 1.28
MS5-3 cSDS = 5:04 × 10−3 mol= m3 U = 6:77 cm= s
1 10 1.6 0.68 0 0 —
2 30 4.6 0.89 0.2766 2.48 3.0 2.484
3 50 7.5 0.99 0.408 2.01 5.9 0.98
4 70 10.7 1.05 0.487 1.71 9.1 1.06
5 90 13.8 1.1 0.552 1.55 12.2 0.95
3800 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

 
2.0 ∗ d c∗ − cd; i+1
kod =− ln d ∗ (33)
rigid sphere correlation
6(ti+1 − ti ) cd − cd; i
fluid sphere correlation
1.5
pure system
completely contaminated
where cd∗ (wt%) represents the drop concentration in equi-
librium with the continuous concentration (=mc1∞ ); cd; 1 the
average drop concentration at the 7rst location.
CD

1.0

0.5
4. Numerical simulation for the experimental cases

4.1. Numerical procedure


0.0
40 80 120 160 200 In order to simulate the transient mass transfer of a con-
Re taminated drop, the @ow inside and outside the drop is 7rst
simulated with speci7ed values of 9 fundamental parame-
Fig. 3. Comparison between correlations and experimental values of the
ters in Table 3. Note that the dimensionless parameters of
drag coe;cient for MIBK drops in pure and completely contaminated
(cSDS = 0:01 mol=m3 ) systems. Re, We, Bi, Pe∗ , Pec∗ , Pes∗ are designed based on the ris-
ing velocity in the pure system as their initial values and
should be recalculated according to the reduced terminal
numbers and compared with the empirical correlations as velocity in each later iteration. When the numerical simu-
shown in Fig. 3. For rigid spheres the correlation recom- lation for the @ow 7eld of a contaminated drop converges,
mended by Schiller and Nauman (Clift et al., 1978) was store the node coordinates in two domains and the values of
applied: distortion functions f1 ( ;
) and f2 ( ;
), stream function
24 and vorticity ! on those nodes for the purpose of being
CD; rigid = (1 + 0:15Re0:687 ): (29) used in the numerical simulation of transient mass transfer
Re
to the same drop. Second stage is to simulate the transient
Harper and Moore’s equation (recommended by Clift et al., process of mass transfer of the contaminated drop with the
1978) was used for @uid spheres, relevant physical parameters and the proper boundary condi-
 
48 3. (2 + 3.)2 tions. For the MIBK–acetic acid–water system, the physical
CD; @uid = 1+ + √ (B1 + B2 ln Re) ; properties, @ow conditions and parameters for mass trans-
Re 2 Re
(30) fer are tabulated in Table 4, in which the Peclet numbers of
solute, Pec and Pes , are based on the measured rising ve-
where the value of B1 and B2 can be referred to Clift et al. locity. The mass transfer timer starts as soon as the average
(1978). From Fig. 3, it is observed that experimental val- drop concentration matches the measurement values at the
ues for MIBK drops in pure water are in good agreement 7rst location (cd; 1 ), and this timing convention is also ob-
with the predictions for @uid sphere, whereas the values in served in later presenting the relation of Em and kod against
completely contaminated water (cSDS = 0:01 mol=m3 ) are in the mass transfer time.
reasonable coincidence with the prediction for rigid spheres.
This shows the hydrodynamic behavior of a completely con- 4.2. Numerical simulation of drag coe7cient a4ected by
taminated drop is similar to that of a rigid sphere. contamination

3.3. Mass transfer coe7cient According to the numerical scheme developed by Li and
Mao (2001), the @ow 7eld inside and outside a contaminated
From typical experimental runs of single drops contam- drop was simulated. Fig. 4 illustrates the e ect of SDS on the
inated by di erent concentrations of SDS solution with @ow 7eld of single MIBK drops with equivalent diameter of
certain diameter (d=1:56 mm) and rising distance, the aver- 1:56 mm. In a pure system (MS5-0), no recirculating wake
age solute concentrations in MIBK drops were also listed in appears behind the drop. However, when the drop is con-
Table 2. For the solute transfer direction c → d, the ex- taminated by SDS (MS5-2 and MS5-3), a detached recircu-
traction fraction, Em , the overall mass transfer coe;cient, lating wake is formed. And the wake is lengthened and gets

kod , and the temporal mass transfer coe;cient, kod , can be progressively closer to the drop rear surface when the SDS
obtained on Eqs. (31)–(33): concentration continues to increase. From the @ow 7eld and
cd; i − cd; 1 the almost attached circulating wake (MS5-3), it is demon-
Em = (31) strated that the behavior of a heavily contaminated drop
cd∗ − cd; 1
 ∗  approaches that of a rigid sphere. Meanwhile, developed in-
d cd − cd; i ternal circulation is slightly weakened as SDS concentration
kod = − ln ∗ (32)
6(ti − t1 ) cd − cd; 1 increases, and the vortex center is gradually shifted to the
X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806 3801

Table 3
Starting parameters used for the numerical simulation of motion for MIBK drops contaminated by SDS (. = 0:4914; , = 0:8123)

Serial no. K Bi Pe∗ × 10−5 Pec∗ × 10−5 Pes∗ × 10−5 Re We


MS5-0 0.0 0.0285 — 1.30 1.30 106.9 1.35
MS5-1 0.0168 2.33
MS5-2 0.0336 1.17
MS5-3 0.0672 0.778

MS7-0 0.0 0.027 — 2.19 2.19 180.0 3.033


MS7-1 0.0336 1.56
MS7-2 0.0672 1.04
MS7-3 0.1008 0.779
Note: MS5 series: d = 1:56 mm; MS7 series: d = 1:97 mm.

Table 4
Parameters used for numerical simulation of mass transfer for MIBK drops (solute transfer direction: c → d) (m = 0:64; Dc = 1:44 × 10−9 m2 =s;
Dd = 0:91 × 10−9 m2 =s; cd0 = 0; cc0 = 2:25%)

Serial no. d U t Ped × 10−5 Pec × 10−5 cd; 1


(cm) (cm/s) (s) (wt%)

MS5-0 0.156 8.33 19.2 0.902 1.428 0.61


MS5-1 0.156 8.00 20.0 0.867 1.371 0.63
MS5-2 0.156 7.55 21.2 0.818 1.294 0.66
MS5-3 0.156 6.77 23.6 0.734 1.162 0.68

MS7-0 0.197 11.11 14.4 1.520 2.405 0.57


MS7-1 0.197 10.53 15.2 1.441 2.280 0.55
MS7-2 0.197 9.75 16.4 1.334 2.111 0.62
MS7-3 0.197 8.89 18.0 1.216 1.925 0.61

has been noted that the stagnant cap model is only a 7rst
approximation to the real situation of continuous change of
MS5-0
physical parameters on the contaminated drop surface (Li
& Mao, 2001).
For the experimental cases, the numerical drag coe;cients
of single MIBK drops whether or not contaminated by SDS
are compared with the experimental data in Fig. 5. For the
same diameter drop, it is obvious that the drag coe;cient
MS5-2 increases gradually with the increment of the SDS concen-
tration as the drop progressively behave more like a rigid
sphere. It is also seen from Fig. 5 that a much higher SDS
concentration is needed for a larger drop to reach such an
asymptotic state. It is found that the numerical values are in
good agreement with the corresponding experimental data
(in Fig. 6) and the average relative error is +8.5% on the ba-
MS5-3 sis of the measurements. The margin is probably due to the
overestimation of deformation on the contribution of drag
coe;cient. The simulation shows again that the numerical
route depicted by Li and Mao (2001) is valid in evaluating
Fig. 4. In@uence of SDS on the @ow structure around and inside the drop the e ect of surfactant on motion of single drops.
(d = 1:56 mm).
Fig. 7 presents the rise velocity of a MIBK drop with
equivalent diameter of 1:56 mm versus the bulk concentra-
drop nose, suggesting the persistent mobility of the inter- tion for the two surfactants, Triton X-100 in (a) and SDS
face in the nose region. This seems to demonstrate the oc- in (b). For both cases, the numerical velocities of a MIBK
currence of a typical case of stagnant cap. Nevertheless, it drop whether or not a ected by contaminants are in good
3802 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

1.5 0.10
1.24 mm
1.56 mm 0.09
1.79 mm
1.97 mm 0.08

U, m/s
1.0 0.07

0.06 measured
D

simulated
C

0.05

0.5
0.04
1E-7 1E-6 1E-5 1E-4

(a) c1surf, mol/m 3

0.10
0.0
0 2 4 6 8 10 0.09
cSDS × 10 , mol/m
-3 3
0.08

Fig. 5. The e ect of SDS concentration in the bulk phase on the drag

U, m/s
0.07
coe;cient of MIBK drops with equivalent diameter varied from 1.24 to
1:97 mm. Solid symbols stand for the experimental data and lines for the 0.06 measured
simulated results.
simulated
0.05

1.6 0.04
1E-5 1E-4 1E-3 0.01

(b) c1surf, mol/m3


1.2
Fig. 7. The rise velocity of MIBK droplets with an equivalent d=1:56 mm
CD (simulated)

versus the surfactant concentration in the bulk: (a) Triton X-100, (b) SDS.

0.8
SDS should be 110 times higher than that of Triton X-100
in order to get the same extent of drop motion retardation.
0.4
4.3. The hydrodynamic e4ect on transient mass transfer

0.0 In this section, the process of transient mass transfer to


0.0 0.4 0.8 1.2 1.6 a MIBK drop contaminated by SDS was simulated. The
comparison between the results predicted by numerical
CD (experimental)
simulation and the experimental data is shown in Fig. 8.
Fig. 6. Parity plot of the simulated and experimental drag coe;cients of It is observed that the numerical simulation is quite suc-
MIBK drops contaminated by SDS in the bulk phase. cessful, and both the extraction fraction and the overall
mass transfer coe;cient decrease with the increase of SDS
bulk concentration. When SDS bulk concentration attains
agreement with the corresponding experimental values. 5:04 × 10−3 mol=m3 (MS5-3), the value of kod is only about
Comparing (a) and (b), it is found that for the same extent one third of that in the pure system. This can be explained
of terminal velocity decrease, the bulk concentration of SDS by the distribution of interfacial velocity and the local
is about 100 times higher than that of Triton X-100. Li and mass transfer coe;cient at t = 0 along the drop surface,
Mao (2001) veri7ed numerically that the drag coe;cients which are plotted against  in Figs. 9 and 10. It is observed
of a single contaminated drop is determined only by the di- that the local Sherwood number at the nose ( = 0◦ ) is,
mensionless parameters of K when Bi  1. For the experi- in general, much higher than that in the rear or wake region
mental case, the Biot numbers listed in Table 3 is much less ( approaching 180◦ ). When the MIBK is contaminated
than unity, so the extent of contamination is only depended heavily, it is found from the pro7les of u
that the interfacial
on the value of K. It is noticed that the value of a for surface becomes stagnant at the rear stagnant region and
SDS is about 110 times higher than that for Triton X-100. is much less than that shown in Fig. 9. Correspondingly,
According to the de7nition of K, the bulk concentration of the value of Shloc drops down abruptly along the stagnant
X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806 3803

50
0.8 Sh loc
40 uη X20
Em, dimensionless

duη /dξX4
0.6
30

0.4
20

Numerical Experimental
0.2 MS5-0 MS5-0 10
MS5-2 MS5-2
MS5-3 MS5-3
0
0.0
0 2 4 6 8 10 12 14
(a) -10
t, s 0 20 40 60 80 100 120 140 160 180
, degree
8 Numerical Experimental
MS5-0 MS5-0
7 MS5-2 MS5-2
Fig. 10. Pro7les of interfacial shear, velocity and local Sherwood num-
MS5-3 MS5-3 bers (t = 0) along the heavily contaminated drop surface (MS5-3,
kodX105, m/s

6
d = 1:56 mm; cSDS = 5:04 × 10−3 mol=m3 ).
5

4
18

3 Numerical Experimental
16
MS7-0 MS7-0
2 14 MS7-2 MS7-2
MS7-3 MS7-3
1 12
0 2 4 6 8 10 12 14
kodx10 , m/s

10
(b) t, s
5

Fig. 8. Contrast of simulation with experiment on the relationships of (a) 6


extraction fraction Em , and (b) overall mass transfer coe;cient kod with
4
mass transfer time (d = 1:56 mm).
2

0
0 2 4 6 8 10
50
t, s
Shloc
40 uηX20 Fig. 11. Contrast of simulation with experiment on the relationships of
duη/dξX4 overall mass transfer coe;cient kod with mass transfer time (d=1:97 mm).
30

20
on the interface. We also simulated the overall mass transfer
10 coe;cient for the MIBK drops with equivalent diameter of
1:97 mm and compared with the experimental data as shown
0 in Fig. 11. The same phenomena can also be observed. It
seems to suggest that the adsorbed SDS has no resistance to
-10 acetic acid transferred across the interface.
0 20 40 60 80 100 120 140 160 180
, degree
4.4. The e4ect of interfacial resistance on transient mass
Fig. 9. Pro7les of interfacial shear, surface velocity and local Sherwood transfer
numbers (t = 0) along the clean drop surface (MS5-0, d = 1:56 mm).
To identify possible interfacial resistance to mass transfer
by adsorbed surfactant layer, the mass transfer to a Tween
surface (Fig. 10) and assumes much lower values than that 80 contaminated MIBK drop was simulated numerically,
in the pure system (Fig. 9). and the hydrodynamic and energy barrier contribution on
From Fig. 8(b), it is also observed that for the pure sys- mass transfer were taken into account altogether. First, only
tem, the discrepancy of kod between the simulation and ex- the contribution of the hydrodynamic factor on mass trans-
perimental data is about 25%, and the discrepancy remains fer was accounted numerically and the results are shown in
roughly the same no matter the MIBK drops are slightly or Figs. 12 and 13 (solid line). It is found for the clean extrac-
heavily contaminated. So it seems that the discrepancy of tion system that the numerical results is about 12% higher
kod is not dependent only on the amount of SDS absorption than the experimental data (Fig. 12), whereas, about 30%
3804 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

1.0 1.0

without interfacial resistance


experimental
0.8 0.8 k s=52.4X10 -5 m/s

0.6 0.6
Em

Em
0.4 0.4

ks=infinite
experimental
0.2 -5 0.2
ks=497X10 m/s

0.0 0.0
0 2 4 6 8 10 0 2 4 6 8 10
(a) t, s (a) t, s

12 7

experimental without interfacial resistance


10 experimental
ks=infinite 6
-5
-5 ks=52.4X10 m/s
ks=497X10 m/s
8
kodX10 , m/s

kod X10 , m/s


6
5

5
4

4
3
2

2
0
0 1 2 3 4 5 6 7 8 9
0 1 2 3 4 5 6 7 8 9 10
(b) t, s (b) t, s

Fig. 12. Comparison of the simulation results with interfacial resistance Fig. 13. Comparison of the simulation results with interfacial resistance
taken into account or not with experimental data of (a) extraction fraction taken into account or not with experimental data of (a) extraction fraction
Em and (b) overall mass transfer coe;cient kod with mass transfer time Em and (b) overall mass transfer coe;cient kod with mass transfer time
(d = 1:79 mm; clean MIBK-acetic acid–water system). (d = 1:79 mm; cTween 80 = 0:005 mol m−3 , heavily contaminated drop).

higher for the Tween 80 contaminated system (Fig. 13). Heideger (1970) evaluated the resistance from direct mea-
Compared with the simulation results of the SDS contam- surement of concentration in the vicinity of the interface and
inated system, it is deduced that Tween 80 molecules ad- found that the transfer of n- and iso-butyl alcohols from wa-
sorbed on the interface has evident mass transfer resistance ter to carbon tetrachloride indicated large interfacial resis-
(1=ks ) on acetic acid, which can be estimated by the least tance (500 –2500 s=cm) due to the presence of surface active
square 7tting to the experimental data. To make the numeri- agents. Although the interfacial resistance has been identi-
cal results 7t with the experimental data, the drop surface of 7ed in this work, more e orts are required to shed light to
a pure system is also assumed to o er interfacial resistance the whole picture of the nature of interfacial resistance from
(1=ks ) on the solute. For the pure system, it was obtained that the adsorbed surfactant molecules.
when ks = 497 × 10−5 m s−1 , the simulation results of the
extraction fraction and the overall mass transfer coe;cient
(dashed line in Fig. 12) is best 7tted with the experimen- 5. Conclusions
tal data. Whereas, the optimal value of ks for the Tween 80
contaminated system equals 52:4 × 10−5 m s−1 , when the Numerical simulation of transient mass transfer to a single
numerical results are best 7tted with the experimental data drop with comparable resistance in both phases was mathe-
(dashed line in Fig. 13). It can be concluded that interfacial matically formulated. In a boundary-7tted orthogonal coor-
resistance in a surfactant monolayer 7lm increases ten times dinate system, the general governing equation for transient
higher than in a clean 7lm, due to physico-chemical inter- mass transfer was expanded and discretized with the con-
action between surface active agent and solute molecules. trol volume formulation. Under suitable initial and bound-
Quantitative comparison of the simulation results of inter- ary conditions, the convective di usion of solute was solved
facial resistance with literature values is di;cult, since no on the basis of numerical simulation of @ow 7elds inside
value for the present solvent extraction system has been and around the drop. The e ect of retarded drop motion and
reported, whereas similar values of interfacial resistance interfacial resistance on the rate of mass transfer due to sur-
for other liquid–liquid systems are available. Mudge and factants were evaluated numerically.
X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806 3805

A series of single drop extraction experiment were con- Dc∗ molecular di usivity of surfactant in contin-
ducted in a totally-closed droplet-7le column with the drop uous phase, m2 =s
phase collecting tubing only vented to the atmosphere. Ds∗ molecular di usivity of surfactant at inter-
MIBK–acetic acid–water system was used in this work face, m2 =s
with small quantities of SDS, Triton X-100 or Tween Em extraction fraction
80 introduced into the continuous phase. The e ect of f( ;
) distortion function
surfactants on terminal velocity, extraction fraction was h ; h
scaling factor, m
obtained experimentally. It is found that surface-active H ; H
dimensionless scaling factor
agents have great e ects on the motion and mass transfer k mass transfer coe;cient, m=s
rate of MIBK drops, and that the rising velocity and mass K dimensionless concentration, 3c1surf =2
fraction decrease gradually with the increase of surfactant m distribution coe;cient
concentration. n molar @ux, mol=m2 s
For the experimental cases, the motion of a contaminated N total @ux, mol=s
drop was simulated 7rst. The numerical results show that the Pe∗ Peclet number at sub-surface, U∞ =Dc∗ c1surf
predicted drag coe;cient is in good agreement with the ex- Pec∗ Peclet number of surfactant in continuous
perimental data. It is also found from the @ow structure that phase, 2UR=Dc∗
the behavior of a heavily contaminated drop approaches that Pes∗ Peclet number of surfactant at interface,
of a corresponding rigid sphere. The transient mass transfer 2UR=Ds∗
of the contaminated drop was also simulated successfully Pec Peclet number of solute in continuous phase,
with relevant physical parameters based on the simulated 2RU=Dc
@ow 7eld. The numerical predictions of the extraction frac- Ped Peclet number of solute in dispersed phase,
tion and overall mass transfer coe;cient are in reasonable 2RU=Dd
coincidence with the experimental data. The numerical pre- R volume-equivalent drop radius, m
diction of the overall mass transfer coe;cient of a heavily Re Reynolds number, 2RU(=)
contaminated drop is only about one third of that in pure Sc Schmidt number, )=(D
systems. This is well explained by the fact that the local Sh Sherwood number, 2Rk=D
Sherwood number drops down abruptly along the stagnant t time, s
surface at the drop rear. u velocity component, m=s
Moreover, the contribution of @ow and the interfacial re- U terminal velocity, m=s
sistance on mass transfer for the contaminated cases were V drop volume, m3
simulated numerically. The numerical simulation identi7ed We Weber number, 2RU 2 (=*
signi7cant interfacial resistance of adsorbed Tween 80 to x; y coordinate in physical plane, m
the acetic acid transfer across the interface, whereas SDS X; Y dimensionless coordinate (x=R; y=R)
and Triton X-100 present negligible resistance. It seems that
the numerical simulation can be used to predict the motion Greek letters
and mass transfer of a single drop, no matter whether it is
contaminated by surfactant. 2 desorption rate constant, s−1
3 adsorption rate constant, m3 =mol=s
 dimensionless surface adsorption
∞ maximum surface adsorption, mol=m2
Notation  domain adjusting factor
a Langmuir–von Szyszkowski constant, , relative density, (2 =(1
mol=m3  dimensionless time, tU=R
Bi Biot number, 2R2=U . relative viscosity, )2 =)1
c concentration of solute, mol=m3 or wt% ) viscosity, Pa s
cd concentration of solute in dispersed phase, ;
coordinate in computational plane,
mol=m3 or wt% 0 6 ;
6 1
c1surf concentration of surfactant in continuous ( density, kg=m3
phase, mol=m3 * interfacial tension, N=m1
CD total drag coe;cient, dimensionless stream function, m3 =s
C dimensionless concentration of solute  stream function, dimensionless
d volume-equivalent drop diameter, m ! vorticity, s−1
Dc molecular di usivity of solute in continuous Superscripts
phase, m2 =s
Dd molecular di usivity of solute in dispersed 0 previous time step
phase, m2 =s ∗ in equilibrium with other phase
3806 X. Li et al. / Chemical Engineering Science 58 (2003) 3793 – 3806

in 7rst measurement location extraction systems with resistance in both phases. Chinese Journal of
Chemical Engineering, 10, 1–14.
out second measurement location
Li, X. J., & Mao, Z. -S. (2001). The e ect of surfactant on the motion
S drop surface of a buoyancy-driven drop at intermediate Reynolds numbers: A
∞ maximum value numerical approach. Journal of Colloid and Interface Science, 240,

average 307–322.
Lin, S. Y., McKeigue, K., & Maldarelli, C. (1990). Di usion-controlled
Subscripts surfactant adsorption studied by pendant drop digitization. AIChE
Journal, 36(12), 1785–1795.
1; d drop or dispersed phase Lochiel, A. C. (1965). The in@uence of surfactant on mass transfer
around spheres. Canada Journal of Chemical Engineering, 43(1),
2; c continuous phase 40–44.
o overall Mao, Z. -S., & Chen, J. Y. (1997). Numerical solution of viscous @ow past
a solid sphere with the control volume formulation. Chinese Journal
of Chemical Engineering, 5(2), 105–116.
Acknowledgements Mao, Z. -S., Li, T. W., & Chen, J. Y. (2001). Numerical simulation
of steady and transient mass transfer to a single drop dominated by
The 7nancial supported from the National Natural Sci- external resistance. International Journal of Heat and Mass Transfer,
ence Foundation of China (Nos. 20236050, 29836130) and 44, 1235–1247.
SINOPEC is gratefully acknowledged. McLaughlin, J. B. (1996). Numerical simulation of bubble motion in
water. Journal of Colloid and Interface Science, 184, 614–625.
Mekasut, L., Molinier, J., & Angelino, H. (1979). E ects of surfactants
References on mass transfer inside drops. Chemical Engineering Science, 34,
217–224.
Archer, R. J., & La Mer, V. (1955). The rate of evaporation through Misek, T. (Ed.). (1978). Recommended systems for liquid extraction
fatty acid monolayers. Journal of Chemical Physics, 59, 200–208. studies. Rugby, UK: EPCE Publication Series, Inst. Chem. Engrs.
Bel Fdhila, R., & Duineveld, P. C. (1996). The e ect of surfactant on Mudge, L. K., & Heideger, W. J. (1970). The E ect of surface active
the rise of a spherical bubble at high Reynolds and Peclet numbers. agents on liquid-liquid mass transfer rates. AIChE Journal, 16(4),
Physics Fluids, 8(2), 310–321. 602–608.
Brunn, P. O., & Isemin, D. (1984). Dimensionless heat-mass transfer Patankar, S. V. (1980). Numerical heat transfer and Cuid Cow. New
coe;cient for forced convection around a sphere: A general low York: Wiley.
Reynolds number correlation. International Journal of Heat and Mass Ponoth, S. S., & McLaughlin, J. B. (2000). Numerical simulation of
Transfer, 27, 2339–2345. mass transfer for bubbles in water. Chemical Engineering Science, 55,
Clift, R., Grace, J. R., & Weber, M. E. (1978). Bubbles, drops, and 1237–1255.
particles. New York: Academic Press. Quintana, G. C. (1990). The e ect of surface blocking on mass transfer
Cuenot, B., Magnaudet, J., & Spennato, B. (1997). The e ects of slightly from a stagnant drop. International Journal of Heat and Mass
soluble surfactant on the @ow around a spherical bubble. Journal of Transfer, 33, 2631–2640.
Fluid Mechanics, 339, 25–53. Ramirez, J. A., & Davis, R. H. (1999). Mass transfer to a
Dandy, D. S., & Leal, L. G. (1989). Buoyancy-driven motion of a surfactant-covered bubble or drop. AIChE Journal, 45(6), 1355–1358.
deformable drop through a quiescent liquid at intermediate Reynolds Ruckenstein, E. (1964). On mass transfer in the continuous phase
numbers. Journal of Fluid Mechanics, 208, 161–192. from spherical bubble or drops. Chemical Engineering Science, 19,
Garner, F. H., & Hale, A. R. (1953). The e ect of surface active 131–146.
agents in liquid extraction processes. Chemical Engineering Science, 2, Ryskin, G., & Leal, L. G. (1983). Orthogonal mapping. Journal of
157–163. Computed Physics, 50, 71–100.
Goodridge, F., & Robb, I. D. (1965). Mechanism of interfacial Ryskin, G., & Leal, L. G. (1984). Numerical solution of free-boundary
resistance in gas adsorption. Industrial and Engineering Chemistry problems in @uid mechanics. Part 2. Buoyancy-driven motion of a gas
Fundamentals, 4, 49–55. bubble through a quiescent liquid. Journal of Fluid Mechanics, 148,
Gri;th, R. M. (1962). The e ect of surfactants on the terminal velocity 19–35.
of drops & bubbles. Chemical Engineering Science, 17, 1057–1070. Savic, P. (1953). Circulation and distortion of liquid drops falling through
Huang, W. S., & Kintner, R. C. (1969). E ects of surfactants on mass a viscous medium. Nat. Res. Counc. Can. Div. Mech. Eng. Rep. MT-22.
transfer inside drops. AIChE Journal, 15(5), 735–744. Schechter, R. S., & Farley, R. W. (1963). Interfacial tension gradients and
Langmuir, I., & Schaefer, V. J. (1943). Rates of evaporation of droplet behavior. Canada Journal of Chemical Engineering, 41(1),
water through compressed monolayers on water. Journal of Franklin 103–107.
Institute, 235, 119–162. Slater, M. J., Baird, M. J. I., & Liang, T. -B. (1988). Drop phase mass
Leppinen, D. M., Renksizbulut, M., & Heywood, R. J. (1996). The e ects transfer coe;cients for liquid–liquid systems and the in@uence of
of surfactants on droplet behaviour at intermediate Reynolds numbers— packings. Chemical Engineering Science, 43(2), 233–245.
The numerical model and steady-state results. Chemical Engineering Takemura, F., & Yabe, A. (1999). Rising speed and dissolution rate of
Science, 51(3), 479–489. a carbon dioxide bubble in slightly contaminated bubble. Journal of
Levich, V. G. (1962). Physico-chemical hydrodynamics. Englewood Fluid Mechanics, 378, 319–334.
Cli s, NJ: Prentice-Hall. Temos, J., Pratt, H. R. C., & Stevens, G. W. (1996). Mass transfer
Li, T. W., Mao, Z. -S., & Chen, J. Y. (2001). Terminal e ect of to freely-moving drops. Chemical Engineering Science, 51(1),
drop coalescence on single drop mass transfer measurements and its 27–36.
minimization. Chinese Journal of Chemical Engineering, 9, 204–207. Thorsen, G., & Terjesen, S. G. (1962). On the mechanism of mass
Li, T. W., Mao, Z. -S., Chen, J. Y., & Fei, W. Y. (2002). Experimental transfer in liquid–liquid extraction. Chemical Engineering Science, 17,
and numerical investigations of single drop mass transfer in solvent 137–148.

You might also like