Journal of Colloid and Interface Science 305 (2007) 175–182

www.elsevier.com/locate/jcis

Transport of ions through the oil phase of W1 /O/W2 double emulsions

Jing Cheng a , Jian-Feng Chen a,b , Min Zhao a , Qing Luo b , Li-Xiong Wen a,∗ ,
Kyriakos D. Papadopoulos c
a Key Lab for Nanomaterials, Ministry of Education, Beijing University of Chemical Technology, Beijing 100029, China
b Research Center of the Ministry of Education for High Gravity Engineering & Technology, Beijing University of Chemical Technology, Beijing 100029, China
c Department of Chemical & Biomolecular Engineering, Tulane University, New Orleans, LA 70118, USA

Received 12 May 2006; accepted 20 September 2006

Available online 24 October 2006

Abstract
Using a capillary video microscopy technique, the ion transport at liquid–liquid interfaces and through a surfactant-containing emulsion liquid
membrane was visually studied by preparing a double emulsion globule within the confined space of a thin-walled, transparent, cylindrical
microtube. NaCl and AgNO3 were selected as the model reactants and were prepared to form a NaCl/AgNO3 pair across the oil film. By observing
and measuring the formed AgCl deposition, it was found that both Cl− and Ag+ could transport through a thick oil film and Ag+ was transported
faster than Cl− . Interestingly, the ion transport was significantly retarded when the oil film became extremely thin (<1 µm). The results suggested
that the transport of ions mainly depends on the “reverse micelle transport” mechanism, in which reverse micelles with entrapped ions and water
molecules can be formed in a thick oil film and their construction will get impeded if the oil film becomes extremely thin, leading to different ion
transport rates in these two cases. The direction of ion transport depends on the direction of the osmotic pressure gradient across the oil film and
the ion transport is independent of the oil film thickness in the investigated thick range. Ions with smaller Pauling radii are more easily entrapped
into the formed reverse micelles and therefore will be transported faster through the oil film than bigger ions. Oil-soluble surfactants facilitate
ion transport; however, too much surfactant in the oil film will slow down the ion migration. In addition, this study showed no support for the
“molecular diffusion” mechanism of ion transport through oils.
© 2006 Elsevier Inc. All rights reserved.

Keywords: Ion transport; Visualization; Double emulsion; Reverse micelle

1. Introduction Mass transfer in W1 /O/W2 double emulsions includes the

Water-in-oil-in-water (W1 /O/W2 ) double emulsions have
found many applications in various fields, including wastewa-
ter treatment, and have recently aimed for slow and sustained
release of active matter from the internal reservoir into the con-
tinuous aqueous phase [1–8]. In most cases, the matter transfer
between the interfaces through the oil film is a key to the suc-
cessful application of double emulsions. However, the problems
associated with the uncontrolled transport of materials still raise
interesting questions and remain one of the biggest challenges
to both research and industry, therefore limiting the develop-
ment of more commercial products of double emulsions.
* Corresponding author. Fax: +86 10 64434784.
E-mail address: wenlx@mail.buct.edu.cn (L.-X. Wen).
0021-9797/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcis.2006.09.055
migration of water and/or water-soluble materials through the
oil film, and they are both dominated by two major release path-
ways: (1) the “swelling–breakdown” mechanism; (2) diffusion
and/or permeation through the oil membrane [9]. Specifically,
in the latter case, Kita et al. proposed two possible mechanisms:
(1) “molecules pass through thin lamellae of surfactants which
partially form in the oil layer due to fluctuation of its thickness”;
(2) molecules diffuse across the oil layer by being incorporated
in “reverse micelles” [10]. Colinart et al. suggested one more
possible way for the water transport, which occurs via hydrated
surfactant [11].
Law et al. found that when the double emulsion was stabi-
lized by a complex of BSA (bovine serum albumin, 0.2 wt%)
and Span80 (sorbitan monooleate, 5–10 wt%) and NaCl was
entrapped in the inner aqueous phase, 60% of the NaCl was
transported from the internal aqueous droplets to the external
176 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182
aqueous phase [12]. After measuring the conductivity of the ex-
ternal aqueous phase in double emulsions, Sela et al. reported
that halides, with the exception of iodide, could be transported
from the inner aqueous phase to the outer aqueous phase [13].
However, such methods based on measuring the variation of
physical properties in the external aqueous phase cannot dis-
tinguish the release caused by the swelling–breakdown mecha-
nism from that caused by diffusion and/or permeation through
the oil film. Due to this shortage, contradictory research re-
sults have been reported. Magdassi and Garti suggested that the
migration of NaCl in double emulsions was realized through
a diffusion-controlled mechanism and not through coalescence
followed by breakdown of the double emulsion [14]. Other re-
searchers believed that water-soluble materials in the inner W1
aqueous phase of double emulsions are delivered to the outside
W2 phase exclusively by the “swelling–breakdown” mecha-
nism, instead of the diffusion/permeation mechanism [15,16].
In our previous work, it was found that the transport of NaCl
through the oil membranes in stable W1 /O/W2 double emul-
sions could not be observed when the oil films were extremely
thin [17]. It was also found that the water transport through
a very thin oil film was very quick and occurred mainly via
the hydrated surfactant mechanism, while the water migration
across a thicker oil film was significantly slower and it was
dominated by spontaneously emulsified droplets and reverse
micelle mechanisms. Water transport cannot be accomplished
by the mechanism of “molecules passing through thin lamellae
of surfactants” [17–19].
In this work, experiments were designed to visually inves-
tigate how salt ions are transported through oils and through
the oil–water interface in double emulsion systems by using the
capillary video microscopy technique, and to study the depen-
dence of the salt transport on the oil–membrane thickness and
surfactant concentration. In each experiment, a single stable
W1 /O/W2 double-emulsion globule was prepared in a micro-
capillary with a different oil thickness. Owing to their visible
reaction product, NaCl and AgNO3 were chosen as the model
reactants and were dissolved in the internal and external aque-
ous phases, respectively, to form a NaCl/AgNO3 pair across the
oil film, and the ion transport was studied by observing the
formed AgCl deposition. Benefited from the prepared single
stable double-emulsion globule, this visual study can distin-
guish and provide unambiguous evidence for other ion trans-
port mechanisms through the oil film besides the “swelling–
breakdown” mechanism. In addition, the different ion transport
rates for different ions through a thick oil membrane or through
an extremely thin oil membrane were also studied, which is es-
sential for further understanding the ion transport mechanisms
and has not been reported in previous studies.
2. Experimental
2.1. Materials
Deionized water was used as the aqueous medium, and
n-hexadecane (97% pure), as the working oil. NaCl (99.5%
pure), AgNO3 (99.8% pure), and surfactant Span80 (sorbitan
monooleate) were purchased from Beijing Chemical Reagent
Co., Ltd., China and used as received without further purifi-
cation. Using a micropipette puller (Narishige PB-7, Japan),
microcapillaries (∼100 µm i.d., ±10 µm) were fabricated from
melting-point tubes (1.1–1.2 mm i.d. × 100 mm, Corning)
and injection micropipettes with a fine tip (10–15 µm o.d.)
were shaped by pulling microtubes (0.688 mm i.d. × 78 mm,
Drummond Scientific). The external surface of the tip of the
micropipettes used to inject the internal aqueous droplet was
rendered as hydrophobic.
2.2. Experimental setup
All experiments were conducted on an Olympus IX-51 in-
verted microscope that was connected to a high-resolution mon-
itor (JVC, Japan) and a VCR (JVC, Japan) through a JVC cam-
era. The microscope was also connected to a PC loaded with
Daheng image analysis software (Daheng Co., Ltd., China) for
capturing images and measuring the droplet size and the area
of the produced AgCl precipitates. Two three-dimensional hy-
draulic micromanipulators (Narishige, Japan) were mounted on
the microscope, and a microinjection system (IM-300, Nar-
ishige, Japan) was provided to prepare the double emulsion
globules.
2.3. Preparation of W1 /O/W2 globules
To prepare the W1 /O/W2 globules, a microcapillary filled
with water (W2 ) was mounted on a self-designed capillary
holder, which was then fixed onto the microscope’s stage.
Held by the manipulator, a nontreated micropipette filled with
hexadecane containing a certain amount of Span80 (O) for sta-
bilization was inserted into the middle of the microcapillary to
inject a large oil drop there. Finally, a hydrophobically treated
micropipette, which was filled with AgNO3 or NaCl solution
(W1 ), was inserted into the oil drop from the other end of the
microcapillary to inject a small aqueous droplet inside the large
oil drop.
By careful positioning of the internal aqueous droplet during
the preparation, the W1 /O and O/W2 interfaces could be in con-
tact or could have a controllable distance in between, meaning
that the oil thickness could be adjusted in this study.
2.4. Visual study of ion transport through the oil film
After the double emulsion system was prepared, another
nontreated micropipette, which was filled with either NaCl or
AgNO3 solution according to the content in the internal aque-
ous droplet in order to form a NaCl–AgNO3 counterpart across
the oil film, was then inserted into the micropipette and was
brought to about 30 µm away from the oil film to inject some
(∼4000 µm3 ) NaCl or AgNO3 solution just outside the oil drop.
Special care was taken here not to inject any of the solution into
the oil drop. Since the NaCl and AgNO3 aqueous solutions in
the as-prepared double emulsion system were separated by the
oil film, the formation of precipitates on either side of the oil
film suggested the occurrence of ion transport through the oil
 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182 177

membrane and the formation rate of the precipitate could re-

veal the ion transport rate qualitatively.

3. Results and discussion

3.1. Microscopic study of ion transport

In the first experiment, the internal droplet W1 contained

2.5 M AgNO3 , the oil film (O) consisted of hexadecane with
0.05 M Span80, and a small amount (∼4000 µm3 ) of 2.5 M
NaCl solution was injected outside the oil drop into the exter-
nal water phase W2 from the right side of the microcapillary.
The W1 droplets were micromanipulated so that they did not
touch the external O/W2 interface. As shown in Fig. 1, two W1
droplets were injected into the oil globule and NaCl solution
was added from the right side of the oil drop. The oil layer
thickness was maintained at ∼30 µm during the experiment.
Some particles started to form in the right W1 droplet 20 s after
the addition of NaCl and more precipitates were produced with
time. Through image analysis, it was found that the area of the
formed precipitates grew to 19.84 µm2 at 25 s and 52.16 µm2 at
50 s after the addition of NaCl, as shown in Table 1. Such for-
mation of precipitates could not be observed if no NaCl solution
was added, demonstrating that it was the result of Cl− transport
from W2 into W1 (which contained 2.5 M AgNO3 ) and thus
formation of AgCl precipitate. However, no particle formation
was noticed on the O/W2 interface, indicating zero or negligible
transport of Ag+ from W1 into W2 under such conditions.
To investigate whether Ag+ could transport through the oil
film or not, the contents of W1 and W2 of the above experiments
were exchanged without varying the concentrations; that is, W1
contained 2.5 M NaCl, O consisted of hexadecane with 0.05 M
Span80, and a small amount (∼4000 µm3 ) of 2.5 M AgNO3 so-
lution was injected outside the oil drop into W2 . As shown in
Fig. 2, AgCl precipitate began to be formed in a shorter time in
W1 after the injection of AgNO3 and more AgCl particles were
produced after the same time as compared to those produced
when AgNO3 was initially prepared in W1 . It was found that
462.56 µm2 precipitate formed at 25 s and 1354.32 µm2 pre-
cipitate formed at 50 s, as shown in Table 1. This suggests that
Ag+ not only could transport through the oil film but also had a
higher transport rate than that of Cl− transport. Similarly to the
first experiment, no ion transport from W1 into W2 under these
conditions could be observed.
In the above two experiments, precipitate was observed
when W1 droplets did not come into visual contact with the
O/W2 interface. In the instances where there was contact, the
formation of AgCl precipitate on either side of the oil film could
hardly be observed, indicating very slow ion transport when
the oil film was very thin. The resolution of the optical micro-
Fig. 1. Transport of Cl− through the oil film in double emulsion globules.
scope is about 1 µm. When the two interfaces were in visual
W1 = 2.5 M AgNO3 solution; O = hexadecane + 0.05 M Span80; W2 =
contact, the oil film thickness between them was immeasurable water + NaCl.
under the microscope, indicating a thickness below 1 µm. Fig. 3
shows the ion transport when the W1 droplet was in visual con- after 125 s, the area of which was only 1.62 µm2 , and there
tact with the O/W2 interface, with W1 containing 2.5 M NaCl, was no noticeable growth of these particles as the experiment
a small amount of 2.5 M AgNO3 solution injected outside the continued. Similar results were obtained when the Span80 con-
oil drop into W2 , and the oil film (O) consisting of hexadecane centration increased to 0.1 M, but no deposition was observed
with 0.05 M Span80. Only two tiny particles appeared in W1 if the Span80 concentration was 0.01 M (Table 1). This minute
178 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182

Table 1
Ion transport through an oil film under different conditions
No.a Span80 No contact between W1 droplet and O/W2 interface W1 droplet contacted with O/W2 interface
conc. Average time Area of formed deposition Average time Average time Area of formed Average time
(M) needed for the elapsed before needed for the deposition elapsed before
After 25 s After 50 s
first observable rupture of first observable after 5 min rupture of
(µm2 ) (µm2 )
deposition oil film deposition (µm2 ) oil film
1 0.01 42 s − 14.82 95 s Not found Film ruptured 76 s
2 0.05 18 s 19.84 52.16 Nb 138 s 1.32 382 s
3 0.1 28 s − 20.3 Nb 278 s 1.14 565 s
4 0.01 22 s 36.79 198.67 95 s Not found Film ruptured 86 s
5 0.05 12 s 462.56 1354.32 Nb 85 s 2.15 358 s
6 0.1 20 s 82.32 386.56 Nb 126 s 1.73 526 s
a 1, 2, 3—Cl− transport (W = 2.5 M AgNO solution; O = hexadecane + Span80; W = water + NaCl). 4, 5, 6—Ag+ transport (W = 2.5 M NaCl solution;
1 3 2 1
O = hexadecane + Span80; W2 = water + AgNO3 ).
b N—no rupture of the oil film was found in 2 h.

Fig. 2. Transport of Ag+ through the oil film in double emulsion globules. Fig. 3. Ion transport through the oil film when the W1 droplet and O/W2 in-
W1 = 2.5 M NaCl solution; O = hexadecane + 0.05 M Span80; W2 = wa- terface were in visual contact. W1 = 2.5 M NaCl solution; W2 = water +
ter + AgNO3 . AgNO3 ; O = hexadecane + 0.05 M Span80.
 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182
ion transport between the contacted W1 droplet and the O/W2
interface was further confirmed by additional experiments when
the contents of W1 and W2 were exchanged.
Effects of the oil-soluble surfactant on the ion transport were
also investigated and the results are shown in Table 1. It indi-
cates that the time needed for the first observable precipitate
decreased with increasing Span80 concentration when the sur-
factant concentration was low, while this time increased with in-
creasing Span80 concentration in the high range. It also shows
that increasing Span80 concentration would produce more de-
position in the same period of time, but this trend was reversed
if there was too much Span80 in the oil. When the two inter-
faces were in visual contact, deposition could hardly be ob-
served over a relatively long period of time. However, the Cl−
transport was apparently slower than the Ag+ transport. These
observations demonstrated that the ion transport is facilitated
by the oil-soluble surfactant when the surfactant concentration
is in the low range, while the transport is slowed down by an
extra amount of surfactant.
The ion transport distinction between visually contacting
and noncontacting W1 droplet and O/W2 interface was signifi-
cantly different from the water transport in such two cases. As
reported in our previous work [18], the water transport was very
slow and independent of the oil layer’s thickness in the absence
of visual contact between W1 and O/W2 interface, while it
became significantly faster when W1 was in visual contact
with W2 . This can be explained by water transport between vi-
sually contacting W1 and W2 being realized mainly by the dif-
fusion of hydrated surfactants, whereas with no contact, spon-
taneously emulsified droplets and reverse micelles are responsi-
ble for the water migration. Owing to the remarkable difference
between the ion transport and the water migration, ion transport
through the oil film must be dominated by different mecha-
nisms.
3.2. Mechanisms of ion transport through the oil film
According to the above discussion, both Cl− and Ag+ could
transport through the oil film between visually noncontacting
W1 /O and O/W2 interfaces and the transport rate of Ag+ was
higher than that of Cl− , while the ion transport was significantly
impeded if the two interfaces were in visual contact. Besides the
“swelling–breakdown mechanism,” which did not occur at all
in this work, the other two proposed mechanisms for ion and
water transport through the oil film are (1) “molecules passing
through thin lamellae of surfactants which partially form in the
oil layer due to fluctuation of its thickness”; (2) “diffusion of
ions across the oil layer by being incorporated in reverse mi-
celles” [9]. The observations in this study showed that the first
mechanism is highly unlikely; the retarded ion transport be-
tween visually contacting W1 /O and O/W2 interfaces indicates
that ions cannot diffuse across the thin oil membrane directly.
The previously reported higher water transport rate through thin
oil films than through thick oil films is attributed to the hydrated
surfactants mechanism rather than to direct molecular diffusion
[18]. Justifications for this conclusion include the following:
(1) If water molecules could pass through the thin lamellae of
179
surfactants directly when W1 was in contact with the O/W2 in-
terface, water-soluble molecules should have been able to pass
through the oil film too, along with the water molecules, at a
much higher transport rate than through a thick oil film. How-
ever, the transport of both AgNO3 and NaCl was significantly
retarded when the oil film became very thin. (2) If it is true
that water molecules may diffuse through the thin oil film di-
rectly, the driving force will be the osmotic pressure gradient
across the film and the water transport rate should change if the
osmotic pressure gradient changed. However, the water trans-
port rate was found to be independent of the osmotic pressure
gradient in a wide range [17]. (3) Previous study showed that
the water-transport rate increases near-linearly with increasing
Span80 concentration in the oil phase over a significant range
of surfactant concentration [19], providing a sound base for the
hydrated surfactants mechanism rather than the direct molecu-
lar diffusion. Therefore, this study, combined with the previous
ones [17–19], shows no support for the direct molecular diffu-
sion mechanism of water and water-soluble materials through
the thin lamellae of surfactants when the oil layer is extremely
thin.
The transport of ions through the oil film between visually
noncontacting W1 /O and O/W2 interfaces must occur through
the reverse micelle mechanism, which is consistent with the wa-
ter transport. In this mechanism, reverse micelles are formed at
one interface with ions and water molecules are entrapped from
the aqueous phase and diffuse across the oil film to the other
interface, where the entrapped ions and water molecules are re-
leased into the other aqueous phase, as illustrated by Fig. 4a.
In the aqueous phase, ions exist in the hydrated state, that is,
cations are surrounded by water molecules with oxygen atoms
approaching them, while water molecules surround the anions
with hydrogen atoms approaching them [20]. The ionic hydra-
tion leads to the formation of a water shell surrounding the
ions and this water shell would move to the oil/water interface
with ions entrapped. Since there are activated atomic hydrogen
and activated atomic oxygen over the hydrophilic polar head
of Span80 [21], the surfactant molecules will be bridged with
the water molecules through hydrogen bonds when the water
shell gets very close to the interface, thus forming a reverse mi-
celle at the O/W interface with entrapped water molecules and
ions. Whether the reverse micelles diffuse across the oil film or
not depends on the direction of the osmotic pressure gradient,
because the majority content in the reverse micelles is water.
That was also confirmed by the observed slight swelling of the
internal W1 droplets in the above experiments in which ions
were found to transport from W2 to W1 . In the process of ionic
hydration, the ionic Pauling radius is a key factor that domi-
nates the strength of the ionic hydration [20]. For both cations
and anions, with the increase of ionic Pauling radius, the ionic
hydration interaction becomes weaker. As reported previously,
the Pauling radii of Ag+ and Cl− are 1.26 and 1.81 Å, respec-
tively [22]. Therefore, Ag+ can be hydrated more easily than
Cl− . In addition, the smaller Ag+ may form a smaller water
shell around itself, thus leading to easier entrapment of Ag+
into the reverse micelles than of Cl− . That can explain the ob-
served quicker transport of Ag+ across the oil film than of Cl− .
180 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182

(a)
(b)
Fig. 4. Schematic illustration of ion transport in double emulsions by being
incorporated in “reverse micelles.” (a) The easy formation of reverse micelles
in a thick oil film; (b) the retarded formation of reverse micelles in a thin oil
film.
In the cases where W1 and W2 were visually noncontacting,
although the thicknesses of the oil film were not exactly the
same in different experiments, it was expected that the effects
of oil film thickness on the ion transport might be negligible.
In fact, no noticeable difference was observed in the repeating
experiments, even though the oil film thicknesses were differ-
ent (30–120 µm), as shown in Table 2. The reason was that the
ions were transported by being entrapped in reverse micelles
that had the majority of the content as water molecules; and as
reported previously [18], in the absence of visual contact the
water transport rates through the oil layer are almost indepen-
dent of the layer’s thickness. In the investigated range of oil
film thickness, the mass migration across the oil film is proba-
bly dominated by the rate of some interfacial processes such as
the formation or the breakdown of reverse micelles, rather than
the diffusion process in the oil film.
The diffusion rate of reverse micelles in the oil phase was
also calculated approximately using the Stokes–Einstein equa-
tion
kB T
D= .
6πμa
The viscosity, μ, of the oil phase at different Span80 concen-
trations was measured by a viscosimeter (Rhoestress rs150,
German Haake), showing a value of 0.27, 0.3, and 0.42 mPa s
at Span80 concentrations of 0.01, 0.05, and 0.1 M, respec-
tively. The radius, a, of the reverse micelle is about 3 nm,
which was measured by laser light scattering (200-SM, Blook
Haven, America). The diffusion time, td , was then calculated as
td = (L2 /D) (L is the oil film thickness). As discussed above,
the experimental average time needed for the first observable
deposition was almost independent of the oil film thickness in
the investigated range, indicating that the ion transport through
the oil film is not a diffusion-controlling process. Therefore, the
diffusion time of reverse micelles through the oil film should be
shorter than the time needed for the first observable deposition,
and this was verified by the calculated diffusion time at thin oil
films, as shown in Table 2. However, the calculated diffusion
time for thick oil films was longer than the experimental time
needed for the first observable deposition, demonstrating some
discrepancy between the calculation and the experiments. The
reason may be that the Stokes–Einstein equation is mostly used
for calculating diffusion without osmotic pressure, and the dif-
fusion should be much faster when driven by osmotic pressure.
More suitable or adjusted models for the computation should
be developed in the future. Nevertheless, the calculated diffu-
sion time, even by the Stokes–Einstein equation, is still on the
same time scale as the experimental time.
Though reverse micelles can be readily formed in a thick oil
film, as illustrated in Fig. 4a, their formation must be greatly
hindered when the oil film becomes extremely thin (Fig. 4b).
One reason may be that the intertwisting between the hydropho-
bic tails of the two adsorbed surfactant layers in the thin oil
film can impede the formation of reverse micelles; another rea-
son is that when the oil film is very thin, most of the surfactant

Table 2
Comparison between the calculated diffusion time of reverse micelles and the experimental average time needed for the first observable depositiona
Span80 The calculated diffusion Average time needed for the first Average time needed for the first
conc. time observable deposition (Ag+ )b observable deposition (Cl− )c
(M) L ≈ 30 µm L ≈ 60 µm L ≈ 120 µm L ≈ 30 µm L ≈ 60 µm L ≈ 120 µm L ≈ 30 µm L ≈ 60 µm L ≈ 120 µm
0.01 3.3 s 13.4 s 52.8 s 22 s 22 s 21 s 42 s 41 s 42 s
0.05 3.7 s 14.9 s 59.5 s 12 s 12 s 13 s 18 s 18 s 18 s
0.1 5.2 s 20.7 s 83.2 s 20 s 19 s 20 s 28 s 29 s 28 s
a L—the oil film thickness.
b W = 2.5 M NaCl solution; O = hexadecane + Span80; W = water + AgNO .
1 2 3
c W = 2.5 M AgNO solution; O = hexadecane + Span80; W = water + NaCl.
1 3 2
 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182
(a)
Fig. 5. Ion transport between two aqueous drops suspended in oil. (a) W and W drops were in no visual contact; (b) W and W drops were in visual contact. W =
1.0 M NaCl solution; W = 2.5 M AgNO3 solution; O = hexadecane + 0.05 M Span80.
molecules are more likely to be involved in the easier hydra-
tion process that can only transport water molecules quickly
through the oil film [11,17]. Since ions cannot diffuse through
the oil film directly, as discussed above, this retarded formation
of reverse micelles in the thin oil film results in a remarkably
slower ion transport than that when the two interfaces are at
noncontact.
The formation of reverse micelles is also affected by the oil-
soluble surfactant. More surfactant molecules in the oil lead to
more reverse micelles, hence facilitating ion transport. How-
ever, further increase of the surfactant concentration at a high
level slowed down the ion transport across the oil film, as shown
in Table 1. Although the viscosity of the oil phase increases
at high oil-soluble surfactant concentrations, which will hinder
the diffusion of reverse micelles across the oil film, it is not
the primary cause for the retarded ion transport in the investi-
gated range of the oil film thickness (30–120 µm), because the
variation of oil film thickness at high surfactant concentrations
still had negligible effects on the ion transport across the oil
membrane, as shown in Table 2. Overconcentrated oil-soluble
surfactant may have some negative influence on the formation
or the breakdown of reverse micelles at the interfaces and hence
slow down the ion migration; however, the mechanisms behind
this are not clear at present.
Another observed phenomenon in this study is that neither
Ag+ nor Cl− could transport noticeably through the oil film
from W1 into W2 . In the preparation of W1 /O/W2 , W1 always
contained 2.5 M NaCl or AgNO3 solution. Though concen-
trated AgNO3 or NaCl solution was injected into W2 to form
181
(b)
the NaCl–AgNO3 pair across the oil film, the amount of the in-
jection was small (∼4000 µm3 ) and it was injected about 30 µm
away from the oil film, thus getting diluted by the W2 water
reservoir. Therefore, W1 always had a higher osmotic pressure
than W2 , leading to the migration of reverse micelles from the
O/W2 interface to the W1 /O interface, that is, the transport of
the entrapped ions from W2 into W1 . This was also verified by
further experiments on transport between two aqueous drops
suspended in oil, as shown in Fig. 5 and Table 3, in which two
aqueous drops (W and W ) contained NaCl and AgNO3 , respec-
tively, were prepared in the continuous oil phase with 0.05 M
Span80. Precipitates were always produced in the aqueous drop
with higher salt concentration no matter whether the two drops
were in visual contact or not. If the two drops had the same
salt concentrations, no noticeable precipitates were observed in
either of them.
Though reverse micelles can be formed at both W1 /O and
O/W2 interfaces, they may be formed more easily at the in-
terface with lower salinity, diffuse across the oil film, and get
dehydrated more easily at the interface with higher salinity
and release their content at that interface. This may contribute
partly to the unidirectional ion migration, that is, ions only be-
ing transported to the aqueous phase with higher salinity. The
osmotic pressure will also drive a net flux in the direction of
higher salinity until the system equilibrium is reached, as stated
by the conventional diffusion and transport theories. However,
the above-mentioned two factors cannot fully explain the phe-
nomena observed in this study, such as the unidirectional ion
migration under unbalanced osmotic pressure and the negligible
182 J. Cheng et al. / Journal of Colloid and Interface Science 305 (2007) 175–182

Table 3
Ion transport between two aqueous drops suspended in oil at different conditionsa
NaCl AgNO3 No contact between W and W W contacted with W
conc. in W conc. in W Depositions Depositions Depositions Depositions
(M) (M) produced in W produced in W produced in W produced in W
2.5 3.0 No Yes No Yes
2.5 1.0 Yes No Yes No
3.0 2.5 Yes No Yes No
1.0 2.5 No Yes No Yes
2.5 2.5 Hardly observable Hardly observable Hardly observable Hardly observable
1.0 1.0 Hardly observable Hardly observable Hardly observable Hardly observable
a W = NaCl solution; W = AgNO solution; O = hexadecane + 0.05 M Span80.
3

ion migration between the two aqueous phases with the same
salinity. Further investigations on the mechanisms are neces-
sary in future study.
One difference between the experiments on two aqueous
droplets and those on double emulsion globules is that precip-
itates were easily formed in the instances of contacting W and
W drops in W/O emulsions, while the formation of AgCl pre-
cipitates could hardly be observed when the W1 droplet was in
visual contact with the O/W2 interface in double emulsion glob-
ules. This is because the contacting area between the two con-
vex interfaces of W and W drops is very small and the larger
noncontacting areas are separated by a thicker oil film that can
transport ions (Fig. 5b), while the contacting area between the
convex W1 /O interface and the concave O/W2 interface is so big
that it leaves only a slight part of the interfaces without contact
(Fig. 3).
4. Conclusions
Using a capillary video microscopy technique and preparing
a W1 /O/W2 double emulsion system, the transport of different
salts through liquid–liquid interfaces and oil films with differ-
ent thickness was visually investigated. It was found for the
first time that ions can transport through thick oil films; how-
ever, the transport gets retarded if the oil film is extremely
thin (<1 µm). The ions always migrate to the aqueous phase
with higher salinity and the ion transport rate is independent
of the oil film thickness in the investigated range for thick oil
membranes. The transport of ions mainly depends on the “re-
verse micelle transport” mechanism instead of direct diffusion,
in which reverse micelles with ions and water molecules en-
trapped in their cores can be formed in a thick oil film, whereas
their formation is impeded if the oil film becomes extremely
thin, leading to the different ion transport rates in these two
cases. Ions with smaller Pauling radius are transported faster
than ions with larger Pauling radius, due to their easier entrap-
ment into the formed reverse micelles. Oil-soluble surfactant
facilitates the ion transport; however, too much surfactant in the
oil film will slow down the ion migration.
Acknowledgments
The authors gratefully acknowledge the financial support
provided by the NSF of China (Nos. 20506001, 20325621),
the Program for New Century Excellent Talents in University
(No. NCET-04-0123), the Talent Training Program of Beijing
City (No. H020821270120), the Research Fund of Multi-Phase
Reaction Laboratory, Institute of Process Engineering (CAS,
No. 2003-1), and the SRF for ROCS, SEM.
Supplementary material
Supporting materials Table S of effects of the oil film thick-
ness on the ion transport between two aqueous drops sus-
pended in oil. This material is available free of charge via
the Internet at http://www.elsevier.com or please visit doi:
10.1016/j.jcis.2006.09.055.
References
[1] N. Garti, Colloids Surf. A 123–124 (1997) 233.
[2] S. Matsumoto, Y. Kita, D. Yonezawa, J. Colloid Interface Sci. 57 (1976)
353.
[3] E. Dickinson, J. Evison, R.K. Owusu, Food Hydrocolloids 5 (1991) 481.
[4] T.F. Tadros, J. Cosmet. Sci. 14 (1992) 93.
[5] A.T. Florence, Chem. Ind. 20 (1993) 1000.
[6] B. Raghuraman, N. Tirmizi, J. Wiencek, Environ. Sci. Technol. 28 (1994)
1090.
[7] F. Valenzuela, C. Fonseca, C. Basualto, O. Correa, C. Tapia, J. Sapag,
Miner. Eng. 18 (2005) 33.
[8] A. Benichou, A. Aserin, N. Garti, Adv. Colloid Interface Sci. 108–109
(2004) 29.
[9] K. Pays, J. Giermanska-Khan, B. Pouligny, J. Bibete, J. Controlled Re-
lease 79 (2002) 193.
[10] Y. Kita, S. Matsumoto, D. Yonezawa, J. Colloid Interface Sci. 62 (1977)
87.
[11] P. Colinart, S. Delepine, G. Trouve, H. Renon, J. Membr. Sci. 20 (1984)
167.
[12] T.K. Law, A.T. Florence, T.L. Whateley, J. Pharm. Pharmacol. 36 (1984)
2.
[13] Y. Sela, S. Magdassi, N. Garti, J. Controlled Release 33 (1995) 1.
[14] S. Magdassi, N. Garti, Colloids Surf. A 12 (1984) 367.
[15] N. Jager-Lezer, I. Terrisse, F. Bruneau, S. Tokgoz, L. Ferreira, D. Clausse,
M. Seiller, J.L. Grossiord, J. Controlled Release 45 (1997) 1.
[16] S. Raynal, I. Pezron, L. Potier, D. Clausse, J.L. Grossiord, M. Seiller, Col-
loids Surf. A 91 (1994) 19.
[17] L. Wen, K.D. Papadopoulos, J. Colloid Interface Sci. 235 (2001) 398.
[18] L. Wen, K.D. Papadopoulos, Colloids Surf. A 174 (2000) 159.
[19] L. Wen, K.D. Papadopoulos, Langmuir 16 (2000) 7612.
[20] J. Zhou, X. Lu, Y. Wang, J. Shi, Fluid Phase Equilib. 194 (2002) 257.
[21] F.O. Opawale, D.J. Burgess, J. Colloid Interface Sci. 197 (1998) 142.
[22] D. Wang, J. Colloid Interface Sci. 247 (2002) 389.