Pumping-out photo-surfactants from an air-water interface using light

Article abstract of DOI:10.1039/c1sm05378g

We study the adsorption dynamics of an azobenzene-based photo-responsive charged surfactant to investigate how photo-stimulation impacts the dynamics at an air-water interface. The hydrophobic tail of this photo-responsive surfactant photo-converts reversibly from a cis to a trans conformation when the wavelength switches from UV to blue. This change in conformation results in a decrease of the surface tension. Using a kinetically limited model of adsorption, including the electrostatics effects and the competition between the two photo-isomers, we reproduce the dynamics of adsorption of AzoTAB measured experimentally. We find that the cis isomer adsorbs 10 times faster than the trans isomer but the cis conformation also desorbs 300 times faster. As a result, within a few seconds a non-stimulated interface becomes composed of almost 100% trans isomers. We then focus on the competition between the photo-conversion and the adsorption at the interface. Indeed when the interface is stimulated, part of the adsorbed trans isomers rapidly convert to cis. As the latter quickly desorbs, the surface coverage decreases: the light induces a "pumping-out" of the interface. The photo-stimulated interface reaches a stationary state where a vertical gradient of composition is established below the surface. Finally, this study highlights a new way to stimulate a photo-responsive interface: for a solution prepared under blue light, instead of photo-converting the bulk composition by stimulating under UV (which can be quite slow for high absorbance solutions), one can tune and reach a stabilized value of the surface tension in few seconds by stimulating the interface with a blue light with high enough intensity. The Royal Society of Chemistry 2011.

Full text of DOI:10.1039/c1sm05378g

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links <
C
Soft Matter
Cite this: Soft Matter, 2011, 7, 7866
www.rsc.org/softmatter
PAPER
Pumping-out photo-surfactants from an air–water interface using light†
a
a
b
c
a
a
E. Chevallier, A. Mamane, H. A. Stone, C. Tribet, F. Lequeux and C. Monteux*
Received 4th March 2011, Accepted 13th June 2011
DOI: 10.1039/c1sm05378g
We study the adsorption dynamics of an azobenzene-based photo-responsive charged surfactant to
investigate how photo-stimulation impacts the dynamics at an air–water interface. The hydrophobic
tail of this photo-responsive surfactant photo-converts reversibly from a cis to a trans conformation
when the wavelength switches from UV to blue. This change in conformation results in a decrease of the
surface tension. Using a kinetically limited model of adsorption, including the electrostatics effects and
the competition between the two photo-isomers, we reproduce the dynamics of adsorption of AzoTAB
measured experimentally. We find that the cis isomer adsorbs 10 times faster than the trans isomer but
the cis conformation also desorbs 300 times faster. As a result, within a few seconds a non-stimulated
interface becomes composed of almost 100% trans isomers. We then focus on the competition between
the photo-conversion and the adsorption at the interface. Indeed when the interface is stimulated, part
of the adsorbed trans isomers rapidly convert to cis. As the latter quickly desorbs, the surface coverage
decreases: the light induces a ‘‘pumping-out’’ of the interface. The photo-stimulated interface reaches
a stationary state where a vertical gradient of composition is established below the surface. Finally, this
study highlights a new way to stimulate a photo-responsive interface: for a solution prepared under
blue light, instead of photo-converting the bulk composition by stimulating under UV (which can be
quite slow for high absorbance solutions), one can tune and reach a stabilized value of the surface
tension in few seconds by stimulating the interface with a blue light with high enough intensity.
1
. Introduction
responsive systems based on the conventional pH, salt, or
temperature triggers remain typically close to equilibrium.
In contrast, more rapid switching should be obtainable by
using light as a trigger, which in addition could be applied to
form non-equilibrium patterns at short length scales. Amphi-
philic molecules bearing azobenzene photochromes are now
successfully applied to photo-switch the equilibrium surface
tension, as well as reversible transitions between monomers/
Adsorption dynamics of amphiphilic molecules at liquid inter-
1
faces controls many properties of liquids such as spreading,
2
3
coating or foaming, which are involved in industrial applica-
tions such as ink-jet printing, extraction of minerals by flotation
or any industrial process where fluid interfaces are created
rapidly. Recently, stimuli–responsive molecules have emerged as
a powerful tool to control interfacial properties. For instance,
neutral surfactants containing an ethylene-oxide-based polar
head can be used for pH-controlled or temperature-controlled
phase separation of emulsions and micro-emulsions or in situ
6–9
micelles/vesicles assemblies of surfactants in solutions, emul-
10–12
sions or foam stability.
13
Recently, Shang et al. optimized the photovariation of
surface tension by tailoring neutral surfactants, containing one
4,5
inversions of emulsions type. Any variation of these control
parameters brings, in principle, the interface to a non-equilib-
rium state, at least transiently. However, the time-scale of the
variation usually achieved by pH or temperature shifts compares
to the diffusion times of amphiphilic molecules. In turn,
azobenzene group in their tail. They obtained typical increases by
ꢀ1
up to 14 mN m of surface tension upon exposure to UV light,
and conversion of the population of the surfactants from
13
predominantly trans to predominantly cis conformations. Most
of the investigations on azobenzene-containing amphiphiles
focused however on static properties, such as equilibrium surface
14,15
1
3
tension, critical micellar concentration,
or more generally
6
,8,16,17
a
photo-switches between equilibrium assemblies,
formation
Laboratoire de Sciences, et Ing eꢀ nierie de la Mati eꢁ re molle (PPMD -
SIMM), CNRS-UMR 7615, ESPCI, 10 rue Vauquelin, 75005 Paris,
of complexes, contact angle, etc.
France. E-mail: cecile.monteux@espci.fr
b
Princeton Univ, Dept Mech & Aerosp Engn, Princeton, NJ, 08544, USA
An application of a non-equilibrium state generated by light
was illustrated by Diguet et al. who manipulated droplets of oil at
the water–air surface in the presence of a cationic azobenzene
c
P o^ le Chimie Biophysique, UMR 8640, D eꢀ partement de Chimie, Ecole
Normale Sup eꢀ rieure, 24 rue Lhomond, 75005 Paris, France
18
derivative. In this case, the water–oil interface was exposed to
a pattern of alternating UV and blue stripes to produce
†
1
Electronic supplementary information (ESI) available. See DOI:
0.1039/c1sm05378g
7
866 | Soft Matter, 2011, 7, 7866–7874
This journal is ª The Royal Society of Chemistry 2011

View Article Online
a gradient of concentration of cis and trans isomers. It was
argued that the gradient of surface tension thus created, induced
18
Marangoni flows of liquids and caused the motion of the drop.
To understand the dynamical response of such azosurfactant
solutions, one of the difficulties lies in the fact that the bulk
solution always contains a mixture of cis and trans molecules,
whose composition depends on the illumination conditions.
Hence both the dynamical and equilibrium surface tensions
result from the competition of the two conformations at the
interface. Kinetics of adsorption were however studied by Cic-
15
ciarelli et al., who analyzed the competitions between cis and
trans isomers for the air–water surface, specifically at long time
scales of several hours and for micellar solutions. They also
applied pulses of light and measured transient variations of
surface pressure that were matching well to switches between the
15,19
isotherms of cis-rich and trans-rich interfaces.
Here we aim at characterizing the short-time dynamics of the
two competing photo-surfactants at the air–water interface
and we highlight new interesting possibilities to stimulate an
interface and tune its surface tension. We chose a soluble
cationic azobenzene surfactant, an AzoTAB, which has been
tailored to trigger ample photo-response of the surface tension,
and to have a reasonably high CMC so as to study the
adsorption dynamics of non-micellar solutions. We studied the
kinetics of adsorption of the azoTAB by maximum bubble
pressure tensiometry and we find that a competition between
cis and trans occurs on very short times, i.e. shorter than a few
seconds. Using a kinetically limited model taking into account
the electrostatic barrier to adsorption, we obtain the constants
of adsorption and desorption of both cis and trans isomers. We
find that the cis isomers adsorb 10 times faster than the trans
isomers but desorb 300 times faster. As a result, the surface
becomes very quickly—in a few seconds—enriched with trans
isomers. Unexpectedly, we show that upon illumination—
either blue or UV—such a trans-rich interface is partly con-
verted to the cis conformation, which quickly desorbs,
inducing a rapid decrease of the total surface excess and then
a rapid rise of the surface tension. We show that this ‘‘pump-
ing-out’’ of the surfactants from the interface can be main-
tained in a stationary regime as long as the interface is
illuminated irrespective of the wavelength.
Fig. 1 Photo-conversion from the trans to the cis isomer of the azoTAB
surfactant studied. UV spectrophotometric measurements showed that
when the solutions are prepared under UV light (l ¼ 365 nm) the solution
contains 16% of trans isomers as compared with 66% of trans isomers
under monochromatic blue light (l ¼ 436 nm).
The kinetics of the photo-conversion was shown to be first-
13
order kinetics with a characteristic time that is inversely
proportional to the light intensity. Hence the time evolution of
the composition in a dilute solution is dictated by the following
expression:
dntrans
dncis
¼
ꢀantrans þ bncis ¼ ꢀ dt
(1a)
dt
3transðlÞI ftrans/cisl
0
where aðlÞ ¼
;
hNAc
l
3
cisðlÞI fcis/transl
0
bðlÞ ¼
(1b)
hN c
A
l
i
where n is the number of ‘‘i’’ molecules, and a(l) and b(l) are
kinetic coefficients at the wavelength of stimulation l of photo-
conversion from trans to cis and cis to trans, respectively. The
latter can be expressed as a function of the quantum yield of
conversion, ftrans/cis or fcis/trans, as well as the molar extinction
coefficient at the stimulating wavelength 3trans(lstim)or 3cis(lstim),
ꢀ2
the light intensity I
0
(in W m ). In eqn (1) c
l
is the velocity of
light and h is Planck’s constant. The determination of a and b is
explained in the ESI, S1†.
2
. Materials and experimental methods
2
.1 Exposure to light
The photo-responsive surfactant, further called AzoTAB is
a cationic surfactant with a trimethylammonium bromide head
group. Its complete formula is (4-butylphenyl)-2-(4-trimethy-
lammoniumpropoxyphenyl) diazene (Fig. 1), whose synthesis is
described in the ESI, S1†.
Photo-conversion of the surfactant in aqueous solutions were
achieved by exposure to UV or visible lights emitted by either
a UV neon lamp (for CCM, Bioblock) with a large emission
ꢀ2
spectrum (intensity of 0.4 mW cm maximum at 365 nm) or
The most thermodynamically stable conformation, the trans
conformation is photo-converted to a cis conformation upon
illumination under blue light (about 436 nm). The molecules
revert back to trans either by illuminating under an ultra-violet
light (about 365 nm) or by maintaining the solution in the dark
for several days. The photo-isomerization is reversible although
a 100% conversion into cis or trans cannot be achieved upon
illumination so the studied solutions are always a mixture of the
two conformations.
a Oriel 250 W mercury arc lamp equipped with an interference
bandpass filter at l ¼ 436 nm ꢁ 10 nm (blue). The intensity is of
ꢀ2
the order of 2 mW cm .
2
.2. UV-vis spectrophotometry
Stationary state of photo-conversion. The solutions of AzoTAB
contain a mixture of cis and trans isomers that depend on the
wavelength of the light to which they are exposed (unless the
This journal is ª The Royal Society of Chemistry 2011
Soft Matter, 2011, 7, 7866–7874 | 7867

View Article Online
Table 1 Kinetic coefficients of photo-conversion and composition of the
solutions for the blue and UV monochromatic sources and the UV1 large
spectrum source
2.3. Surface tension measurements
The long time surface tension measurements were performed
with a classical pendant drop tensiometer (Teclis, France). The
image of the drop is recorded using a camera and the profile is
fitted to the solution of the Laplace equation to obtain the
surface tension. The solutions are illuminated for 30 minutes
before the surface tension measurement but not during surface
tension measurements. Special care was then taken to keep the
experimental set-up in the dark. A red optical filter was mounted
on the background light to prevent photo-conversion of the
surfactants during the measurements. In practice, the apparent
surface tension decreases with time and should reach at long
incubation time the equilibrium plateau value. At time 100
seconds, we found that the rate of decrease of the surface tension
UV1 large
spectrum source
Stimulus
Blue 436 nm
ꢀ
1
a/s for 3 mW cm
ꢀ2
ꢀ2
0.0276
0.0552
66
0.2071
0.0394
16
ꢀ
1
b/s for 3 mW cm
%
%
trans
cis
34
84
solution are incubated in the dark for several days to return to
the more stable 100% trans form). The difference between the
UV/visible spectra of the cis and trans isomers enabled us to
determine the composition of their mixtures, and the kinetic
coefficients of their photo-isomerization in solution, which are
important parameters in the following investigation of the
surface response to light. To measure the molar fractions of cis
and trans AzoTAB in the various conditions of illumination, we
resorted to absorbance measurements in the UV/visible (Hewlett-
Packard 8453 spectrophotometer). During the measurements,
the samples were constantly irradiated with the light source in
order to reach the photo-stationary state. By comparison of
measurements with and without exposure, we checked that the
light flashes briefly produced (ca. 1 s) by the spectrophotometer
during measurements did not perturb this stationary state. As
described in the ESI (S1†), the spectra of the trans conformation
and of intermediate states (mixtures of cis/trans) were fitted as
sums of the cis and the trans spectra to obtain the stationary cis–
trans composition of the samples reported in Table 1.
ꢀ1
was smaller than 0.1 mN m per minute. Somewhat arbitrarily
to avoid long incubation that possibly exposes samples to
residual ambient lights) we report the values measured at time
(
1
00 s in isotherms given in Fig. 2.
Short time dynamic surface tensions were measured using
21,3
a maximum bubble pressure apparatus
(BPA-1S-Sinterface)
without any photo-stimulation. The method is based on the
measurement of the maximum pressure in a bubble growing at
the tip of an immersed capillary. When a bubble is produced at
the tip of a capillary, its radius of curvature first decreases up to
reaching those of the capillary, when an hemispheric bubble is
formed. Beyond that point, the bubble grows and its radius
decreases accordingly. The maximal Laplace’s pressure is thus
corresponding to a bubble radius equal to the capillary’s one. By
varying the flow-rate, one can vary the age of the interface at
which the surface tension is measured, therefore enabling to
measure dynamic surface tensions. However with such tech-
nique, the interface expands during the experiment, hence results
are different from those obtained from measurements performed
Dynamics of photo-conversion. To determine the kinetic coef-
ficients of the photo-conversion, ‘‘a’’ (for trans to cis) and ‘‘b’’ (cis
22
13
with immobile bubbles as pointed out by Christov et al. In
to trans) we used Shang’s method, by measurement of pseudo-
first order kinetics of the variation of the absorbance of dilute
solutions of AzoTAB, exposed to varying intensities of illumi-
nation under UV or blue lights. We checked that the kinetic
constants varied in proportion to the intensities of light. Exper-
imentally the value of intensity is however an average over the
Christov’s article, the effect of the surface dilatation is taken into
account by normalisation of the time called age of bubble, tage
,
into an universal time, t , corresponding to the non-deformed
u
2
2
interface: t
u
¼ tage/Z , where the factor Z can be determined by
performing measurements of reference solutions, e.g. sodium
dodecylsulfate. Using Christov’s method, we found Z ¼ 9.5, and
times given in figures will accordingly correspond to bubble life
2
surface of our powermeter (ca. 1 cm ) which may differ slightly
from the exact value of intensity passing through the smaller
surface of the sample probed (for instance because of heteroge-
neities of the beam), leading to wrong values of a and b.
To bypass this sensitivity, we took the precise value of the ratio
b/a deduced from the stationary composition measurement as:
2
2
times divided by Z . The value of Z does not affect the ratio of
the desorption constants of cis and trans isomers, which is the
main finding of this article.
3
. Static and dynamic surface tension measurements
b
a
ctrans
¼
(2)
ccis
3.1 Static surface tension
a and b could then be calculated from eqn. (3):
aþb
3cisFcis/trans þ 3trans
F
trans/cis
As cis and trans isomers have different conformations, the area
per adsorbed molecule is expected to change with the cis–trans
composition of the solution. This parameter is obtained by fitting
the surface tension isotherms, as shown in Fig. 2. The isotherms
correspond to solutions prepared under blue and UV illumina-
tion that have different compositions as summarized in Table 1.
These cis-rich solutions exhibit a higher surface tension than the
solutions containing a majority of trans isomers. At fixed total
concentration of AzoTAB, we observe that increasing the frac-
tion of cis isomers increases the surface tension by up to 20 mN
¼
A
I l=hN cl
(3)
We use a quantum yield from cis to trans, Fcis/trans of the order
of 0.5, which is a common value for azobenzene molecules given
in ref. 20. In addition, the ratio Ftrans–cis/Fcis–trans of approxi-
13
mately 1.3 as determined by Shang et al. matched with our
data. Finally, the estimate of a and b and the cis–trans compo-
sitions of the solutions are summarized in Table 1.
7
868 | Soft Matter, 2011, 7, 7866–7874
This journal is ª The Royal Society of Chemistry 2011

View Article Online
Fig. 2 Surface tension isotherms for solutions prepared under different
types of illumination conditions. The squares correspond to solutions
prepared in blue light containing 66% of trans isomers, the circles are for
solutions prepared under UV and contain 16% trans.
Fig. 3 Dynamic surface tension of AzoTAB solutions in 1 mM NaCl.
Interface squares represent solutions containing 0.23 mM trans isomers in
the presence of 0.1 mM cis isomers (filled squares) or 1.17 mM cis-isomers
(open squares). The circles represent solutions containing 0.92 mM of
trans isomers in the presence of 0.48 mM (filled circle) or 4.58 mM cis
isomers (open circles). Filled symbols are for solutions exposed to blue
light and open circles are for solutions prepared under UV light. The time
ꢀ1
m . To estimate the surface excess, G
N
, we use the Langmuir–
von Szyszkowski isotherm to fit our data:
2
scale given by the apparatus, tage, has been divided by a factor Z as
suggested by Cristov et al. to account for surface deformation during the
p ¼ nRTG
N
ln(1 + K
L
c)
(4)
experiment.
where p is the surface pressure, K
L
is the Langmuir–von Szysz-
short time dynamics is the same for both curves but at longer
times, we measure lower surface tensions for the solution con-
taining a higher concentration in trans.
kowski constant and n ¼ 2 for charged 1 : 1 surfactants. Alter-
native isotherms can be proposed to account for interactions
between surfactants such as a Frumkin isotherm. However, we
preferred to limit here the number of adjustable parameters by
choosing the Langmuir isotherm. Unexpectedly, we obtain the
same value of GN for all three isotherms within the experimental
We propose the following qualitative scenario to interpret these
results: at short time both surfactants adsorb roughly at the same
rate. However at longer time scales, the concentration of trans
isomers controls the interfacial tension, hence the trans isomers
would adsorb preferentially at the interface. Such preferential
ꢀ
6
ꢀ2
error, GN ¼ 8.10 mol m ꢁ15% corresponding to a molecular
ꢂ
2
area of u
N
¼ 20 A per molecule. The measured molecular areas
13
adsorption has been observed by Shang et al., although at much
longer time scales of the order of several hours. The fact that the
competition between both surfactants occurs over very short time
scales is critical to the opportunity to trigger rapid variations of
interfacial properties. In the next section we quantify more
precisely the interfacial dynamics of the two isomers.
are lower than the one measured for C
n
TAB (n ¼ 10, 12, 14, 16)
2
3
24
by surface tension or neutron reflectometry for which u
N
¼
ꢂ
2
5 A per molecule is measured. It is likely that p-stacking
4
between the azobenzene groups of the surfactant makes the
25
molecular area decrease. Putting all of these remarks together,
we conclude that in our samples, the trans isomers should be
predominant at the air–water interface irrespective of the pres-
ence of the marked excess of cis isomers in the bulk solution.
4. Modeling the competitive adsorption of cis and
trans isomers
3
.2 Short time dynamic surface tension
To predict the exchanges rates and fluxes of surfactants under
illumination, and their roles on the photo-control of interfacial
tension, we propose an elementary model of the adsorption/
desorption process. Our goal is to fit the dynamic surface
tensions to determine the orders of magnitude of the parameters
which control the interfacial properties, specifically the kinetic
coefficients of adsorption and desorption. Then, we will use these
results to predict in which conditions the photo-stimulation can
control the interfacial properties, for instance, fluxes of mole-
cules to/from the interface.
Another parameter that may be influenced by the composition of
the sample is the dynamics of adsorption of surfactants at the
interface. In Fig. 3 we compare the dynamic surface tension for
four samples. The two curves represented in squares (respectively,
the circles) correspond to solutions having the same concentration
of trans, e.g. 0.23 mM (respectively, 0.92 mM)—but different
concentrations of cis isomers. At times shorter than 0.1 s and for
a fixed concentration of trans form, the surface tension is lower for
the solution containing a majority of cis isomers and equal trans
concentration. In contrast, at longer times (1 s) the surface
tensions of this pair of samples converge to the same value.
Another interesting result is obtained when comparing solutions
that contain similar total surfactant concentrations but different
amounts of trans isomers. As shown in Fig. 3, open squares and
filled circles correspond to a total concentration of 1.4 mM and
respective trans concentrations of 0.23 and 0.92 mM. The very
4
.1 Modeling the interface to deduce the constants of
desorption
To describe the adsorption dynamics of the surfactants we use
a model where the dynamics is limited by the adsorption kinetics
and not by the diffusion. To describe the adsorption dynamics of
This journal is ª The Royal Society of Chemistry 2011
Soft Matter, 2011, 7, 7866–7874 | 7869

View Article Online
26
charged surfactants, Bonfillon and Langevin developed a kinet-
ically–limited model, where the energy barrier of adsorption is of
electrostatic origin and increases with time. Indeed when the
charged surfactants adsorb to the interface, the surface potential
of the interface increases, which makes further adsorption of the
surfactants more and more difficult. For such a model, the vari-
ation of the surface excess concentration with time can be written:
dG
dt
ꢀ
ꢁ
¼
kadscsub 1 ꢀ u GðtÞ ꢀ k u GðtÞ
(5)
N
des
N
where G is the surface excess at a given time, kdes is the constant
to desorption, csub the surfactant concentration in the subsurface
27
as defined by Joos and Fainerman, and uN the surface area per
molecule of surfactant. The constant of adsorption kads can be
written:
ꢀ
2arcshðGlB= 2kÞ
0
ads
Fig. 4 Overview of the fluxes described in the model. Surfactants are
photo-converted with kinetics constants a and b. Trans and cis confor-
mations are characterized by adsorption constants, kads and kdes, and the
bulk diffusion coefficient, D, to the interface.
kads ¼ k
e
(6)
ꢀ
1
where G is the surface excess, l
B
is the Bjerrum length, k is the
0
Debye–Huckel length and kads is the constant of adsorption for
ꢀ
18
ꢀ2
a bare surface. To fit our data, we will use l
B
/k ¼ 1.5 ꢂ 10
m
ꢀ17
ꢀ2
instead of the theoretical value of l
B
/k ¼ 5.8 ꢂ 10 m . This
surfactant, which is assumed to be the same for the two confor-
2
ꢀ
10
mations—we will take D ¼ 9.10 m s , the value obtained for
C TAB by Ribeiro et al., 2004. Moreover we assume that cis–
ꢀ1
type of correction has already been described in ref. 28. It may be
attributed to the fact that the effective potential seen by the
surfactants molecules close to the interface is lower than the
theoretical potential at the surface.
30
1
6
trans conversion is similar at the interface and in the bulk.
Writing the equality of the fluxes at the interface, we obtain the
We then combine eqn (5) and (6) and add terms corresponding
to the exchange of cis and trans because of photo-conversion,
whose kinetic coefficients a and b where defined in Section 2. We
relation between the subsurface and the bulk solution:
8
>
:
dGtrans
dt
vctrans
>
>
<
þ aGtrans ꢀ bGcis ¼ D
vx
29
extend the model to a mixture of surfactants by considering
that the adsorption flux is proportional to the available area at
the interface and the desorption flux is proportional to the
(9)
>
dGcis
vccis
þ bGcis ꢀ aGtrans ¼ D vx
dt
surface excess of the surfactant likely to desorb. We obtain
cis
From their model Bonfillon and Langevin derive the corre-
26
the following set of equations where Gcis (resp. Gtrans), u
trans
N
sponding surface pressure isotherm:
ꢃ
cis
cis
0cis
0trans
ꢄ
(
resp. u
N
), csub (resp. csub ), kads (resp. kads
) are the
nk TG
B
G
N
surface excess, surface molecular area for a saturated interface,
concentration in the subsurface, adsorption constant, and
desorption constant of the cis isomers (respectively, trans
isomers).
PðGÞ ¼ ꢀ
ln 1 ꢀ G
þ nk
B
T
(10)
2
N
According to Bonfillon and Langevin, this expression of the
isotherm is valid only for high surface excess concentrations—
8
>
:
ðGtransþGcisÞlB
>
dGtrans
ꢀ2arcsh
e
ꢀ
ꢁ
>
<
0trans
¼ k ads
2k
trans
sub
trans
cis
N
trans trans
N
c
1 ꢀ uN
G
trans ꢀ u Gcis ꢀ kdes
u G
trans ꢀ aGtrans þ bGcis
dt
(
7)
ðGtransþGcisÞlB
2k
>
ꢀ2arcsh
ꢀ
ꢁ
dGcis
0cis
ads
cis
sub
trans
cis
N
cis cis
des
¼
k
e
c
1 ꢀ uN
G
trans ꢀ u Gcis ꢀ k u Gcis þ aGtrans ꢀ bGcis
N
dt
The different fluxes accounted for in the simulation are
summarized in Fig. 4. The diffusive fluxes from solution to the
surface and the photo-conversion in the bulk yield the following
boundary conditions:
Gl /k [ 1. Again, we extend this model to the case of a mixture
b
30
of surfactants:
tot
TGN
nk
B
ꢀ
ꢁ
trans
N
cis
N
P ¼ ꢀ
ln 1 ꢀ u
G
trans þ u Gcis þ nk
B
TðGtransþGcisÞ
(11)
2
8
>
:
ꢂ
2
v c
vctrans
vt
>
trans
>
<
¼ D
ꢀ actrans þ bccis
vx2
ꢂ
(8)
From eqn (7–9) and (11), the dynamic surface tension can be
predicted numerically and fitted to the experimental data.
One should bear in mind that high degree of sophistication in
the model would mask the main purpose of this work. The
ambition of this model is to define which parameters and which
2
>
vccis
v ccis
¼ D vx2 ꢀ bccis þ actrans
vt
th
Here c
i
(x, t) is the concentration of the i surfactant at a distance x
from the interface, and D is the diffusive coefficient of the
7
870 | Soft Matter, 2011, 7, 7866–7874
This journal is ª The Royal Society of Chemistry 2011

View Article Online
conditions enable photo-control of non-equilibrium surface
states and the ability to tune surface properties. Indeed, this
system raises many theoretically complex issues to account for
the cationic nature of the surfactants and the electrostatic
31–34
effects.
Moreover the competitive adsorption and surface
interaction in mixtures of surfactants is also a complex
27,29,35–37
phenomenon.
Finally we aim at reducing as much as
possible the number of adjustable parameters in order to obtain
reasonably independent fitted values. This elementary model
0
trans
0cis
trans
cis
reduces the adjustable parameters to four: kads , kads , kdes , kdes
.
We determine these four parameters by fitting the numerical
simulation to the data of dynamics of adsorption obtained
without photostimulation, as shown in Fig. 3, beginning with a ¼
b ¼ 0. We use the set of data corresponding to the trans
concentration equal to 0.92 mM (filled and open circles). We do
not use the other set of data because it corresponds to low surface
coverages for which the model of Bonfillon and Langevin does
not apply. Moreover we make the assumption that the value of
the area per molecule of the cis isomers is 1.6 times the surface
area of the trans surfactants, based on the shape of the
Fig. 5 Temporal evolution of the composition of the interface from
a solution prepared under blue light and stimulated under UV. The
simulation is done for a concentration of 1.4 mM in surfactant and for an
intensity I ¼ 100 mW cmꢀ
2
.
13
molecules.
UV reflects the lower affinity of the cis for the interface. Finally,
the UV stimulation results in the decrease of the total surface
excess with time.
The fitting procedure, which can be found in the ESI (S2†),
allows us to determine independently these four parameters.
We finally obtain the values of the adsorption and
As the dynamics is governed by the rate of photo-conversion in
the bulk solution, the time response for concentrated solutions
exhibiting high absorbance will be considerably slower than what
is predicted by our model (at low absorbance, as assumed in the
model, the intensity does not decrease with increasing penetra-
tion depth in solution, which corresponds to the fastest conver-
sion in solution). Such high absorbances are easily reached in
0
trans
ꢀ2
desorption constants of both conformations—kads
ꢀ1
, ꢃ 3.5 10
ꢀ
1
trans
ꢀ2
m s , kdes
cis
¼ 5.5 10–5 mol m s . Moreover we find
ads
0cis
k
k
des
trans
des
that
¼ 300 and
¼ 10.
0trans
ads
k
k
As a conclusion, from the fits of our data, we find that the cis
conformation adsorbs 10 times faster than the trans at the
interface but they desorb 300 times faster. This results in a rapid
enrichment of the interface with the trans isomers. The model
fairly describes how at very short time scale the dynamic surface
tension was highly sensitive to the presence of the cis isomers,
though at longer time scale and at equilibrium, surface tension
was dominated by trans isomers. It is confirmed also that the
surface area fitted from the isotherms of Fig. 2 is actually that of
the trans form of AzoTAB.
4
ꢀ1
ꢀ1
practice (extinction coefficient of azo ca. 10 L mol cm ) and
high intensities have to be used to accelerate the dynamics.
However in confined geometries, such as thin-liquid films or
foams where the thickness to photo-convert is as thin as tens of
microns, the absorbance is smaller than 1. In this case, our model
is valid and predicts the time scales of the interfacial response.
Stimulating with blue light a blue-adapted solution. Let us now
focus on a situation where the time of photo-conversion of the
bulk will no longer be limiting for tuning the surface tension. We
investigate here the case of a solution previously prepared under
blue light and then exposed to blue light. In that case the photo-
conversion of the bulk solution is stationary and the limiting step
will be either the photo-conversion of the interface or the dynamics
of exchange of surfactants at the interface.
In the next section, we will use the model to make predictions
about the dynamical response of the interface to a photo-
stimulation.
4
.2. Predicting the interfacial response when the interface is
stimulated with light
Stimulating with UV light a blue-adapted solution. Typical
experiments using photoswitchable molecules usually involve
stimulation of a blue-adapted sample by exposure under UV
light (or the reciprocal UV-adapted state switched under blue
Under blue light, the interface has to be composed of 66%
trans isomers but according to our model at equilibrium the
fraction of trans isomers at the interface should be much higher
because the trans isomers adsorb preferentially at the interface.
Therefore the photoconversion induces an excess of cis isomers
at the interface in comparison to the equilibrium situation at the
interface without illumination. The excess of cis isomers at the
interface induces a cascade of fluxes as illustrated in Fig. 6. The
cis isomers in excess created at the interface by the stimulation
desorb, then they photo-convert in the bulk into trans isomers
and the trans in excess re-adsorb at the interface. As a conclusion,
illumination induces an increase of the ratio of the cis isomers at
the interface, which have a low affinity for the interface. As
a result, the total surface excess decreases. In other words the
15,18
light).
concerning the time evolution of Gtrans, Gcis as well as cbulk
the blue-adapted sample is switched from blue to UV light. As
In Fig. 5 we report the results predicted by our model
trans
as
bulk
can be seen, both ctrans
and Gtrans decrease with time. Inter-
bulk
estingly, the time evolution of ctrans
and Gtrans is similar. So,
under these conditions the limiting step appears to be the cis–
trans photo-conversion and an equilibrium is continuously
established between the bulk and the surface. The small increase
of Gcis with time, and the fact the Gcis/Gtrans does not reach the
ratio corresponding to the photo-stationary state in bulk under
This journal is ª The Royal Society of Chemistry 2011
Soft Matter, 2011, 7, 7866–7874 | 7871

View Article Online
which modifies significantly the interfacial composition. In this
case the characteristic time of photo-conversion at the interface,
sint-f, is shorter than the characteristic time, sint-eq, that is needed
to reach the adsorption equilibrium without illumination. From
the experimental data shown in Fig. 3, we find that sint-eq is of the
ꢀ1
order of 1 second whereas sint-f, is equal to (a + b) and scales as
ꢀ1
I . To obtain a significant modification of the interface by the
blue light, the condition is that sint-eq [ sint-f, which gives
ꢀ2
a cross-over intensity I ¼ 60 mW cm , indicated by an arrow in
Fig. 7. Such intensities can easily be obtained using a standard
microscope objective to focus the beam emitted from sources
such as common LEDs. Moreover it should be noted that the
threshold intensity can be lowered by choosing a photo-surfac-
tant which a slower adsorption dynamics, i.e. with a higher sint-eq
.
Our work shows that for practical applications, one can expect
to tune the surface tension by controlling the intensity thanks to
the ‘‘pumping-out’’ phenomenon. Compared to the slow photo-
conversion from blue to UV in the bulk for solutions of high
absorbances, fast equilibration of the interface—of the order of
Fig. 6 Schematic cycle describing fluxes and photo-conversions occur-
ring at the interface in the stationary regime. Trans isomers from the bulk
continuously adsorb at the interface where they are converted into cis by
light. The cis isomers desorb into the bulk, where they convert to the trans
isomers again.
sint-eq ¼ 1 second—will be obtained even for high concentrations
when stimulating with blue light a blue-adapted solution (or
stimulating with UV light a UV-adapted solution) because the
bulk solution has already established the composition dictated by
the illumination conditions. Moreover we show here for the first
time that using this azoTAB photosurfactant, only one wave-
length is needed, either UV or blue, to provoke a strong and fast
variation of the surface tension. Indeed, to the best of our
knowledge previously reported studies using photosurfactants
usually involve a switch from one wave-length to another.
light stimulation induces a ‘‘pumping-out’’ of the surfactants
from the interface. Actually this ‘‘pumping-out’’ phenomenon
appears as soon as the interface is stimulated, whatever the
wavelength used for the stimulation and the preparation of the
sample. Such fluxes also induce vertical gradients of the cis/trans
composition that will be explained later.
In Fig. 7, we present the total surface excess and cis compo-
sition at the interface as predicted by our model. We find that
when the intensity of the light increases, the cis ratio at the
interface increases and the total surface excess decreases. This
result illustrates the competition between photo-conversion of
adsorbed trans into cis and desorption of the cis isomers. For low
intensities, the composition at the interface is weakly modified by
the illumination. On the contrary, for higher intensities the trans
to cis photo-conversion becomes faster so the desorption flux
from the interface results in a decrease of the total surface excess,
Creating stationary vertical and horizontal gradients of surfac-
tants. As illustrated in Fig. 8, horizontal—lateral—concentration
gradients can be created at the interface. We spread particles of
talc at the interface of a solution of azosurfactants prepared
under blue light. The solution is then placed under a microscope
in the dark. Then either UV or blue light is passed through the
objective of the microscope, which enables us to focus a spot of
2
light at the interface over 0.01 mm . It can be seen that when the
shutter opens, for both wavelengths, the particles rapidly
concentrate under the light spot in a few seconds. Under the spot,
both UV and blue light induce a local increase of the surface
tension whereas outside the spot, the surface tension remains
low. This heterogeneity in the surface tension then drives
a Marangoni flow toward the spot of light, which drags particles
under the spot. As explained previously this gradient of surface
tension can be obtained for both wavelengths. The intensity we
obtain with the UV source is lower than with the blue source,
which explains why the observed velocity of the particles is lower
with the UV light.
As previously stated, this stationary flux remains as long as the
solution is illuminated and it only requires one wavelength. This
stationary Marangoni flow could be used to collect colloids in
a continuous manner or to control the assembly of colloids using
patterned masks to guide light.
Fig. 7 Total surface excess (bold) and cis/trans ratio at the interface
Along with the horizontal gradient of surface tension,
a stationary gradient of cis/trans composition appears vertically
in the solution due to the stimulation of the interface. Indeed, due
to the rapid desorption of cis at the interface, the composition
below the surface is more enriched in cis than the bulk. Let us
(
dashed) as a function of light intensity when a solution prepared under
blue light is stimulated with blue light. The arrow represents the threshold
intensity above which the characteristic time of photoconversion of
adsorbed trans isomers, sint-f, becomes shorter than sint-eq, the time
needed for the interface to be in equilibrium with the bulk composition.
7
872 | Soft Matter, 2011, 7, 7866–7874
This journal is ª The Royal Society of Chemistry 2011

View Article Online
Fig. 8 Marangoni effect—a solution containing the photosurfactants is placed under a microscope and talc particles are spread at the surface. Blue or
ꢀ
2
ꢀ2
UV light is then passed through the objective of the microscope to create a light spot (ca. Iblue ꢃ 100 mW cm
I
UV ꢃ20 mW cm ). The total
concentration in surfactants is 1.15 mM. Outside the spot a majority of trans are adsorbed. Inside the spot, the surface tension rapidly increases because
trans isomers are converted to cis isomers, which rapidly desorb. This surface tension gradient results in a Marangoni flow from the outside to the inside
of the spot and gathers colloidal particles toward the spot. (a) Photograph of the fast aggregation of the talc particles under the UV or the blue spot. (b)
Schematic diagram representing the mechanism causing the aggregation of the particles. The open arrows represent the photoconversion reaction and
the filled arrows represent the transfer of molecules via Marangoni flows and adsorption/desorption fluxes.
consider this vertical gradient from the equations of the model.
From eqn (8), in a stationary regime, we can write
surfactants results in a rapid decrease of the surface excess and
rapid increase of the surface tension for both wavelengths. A
cascade of fluxes is generated below the surface and a stationary
vertical gradient of concentration of tens to hundreds to microns
exists below the surface.
8
2
d c
dz2
>
tot
>
<
D
D
¼ 0
bctot
with cN
¼
trans
(12)
>
2
a þ b
>
d ctrans
:
N
trans
These results highlight a new way to stimulate an interface
with adsorbed photo-surfactants. When switching from blue to
UV light the dynamical surface tension is limited by the photo-
conversion of the bulk solutions, which is very slow for solutions
of large absorbance. On the contrary the stimulation of a blue-
adapted solution with blue light enables a much faster equili-
bration of surface tension. With the ‘‘pumping-out’’ effect, a high
enough intensity triggers a significant decrease of the surface
excess and a rise of the surface tension. So, one can tune the
surface tension by simply controlling the intensity of a single light
source. Finally we show experimentally that only one spot of
light is sufficient to generate a horizontal stationary gradient of
surface tension and thus to drive Marangoni flows.
¼ ða þ bÞctrans þ c
dz2
and ctot ¼ ccis + ctrans
;
Solving eqn (12) gives the concentration profile of each isomer:
rffiffiffiffiffiffiffiffiffiffiffi
D
z=l
ctransðzÞ ¼ Dceꢀ þ cN where l ¼
ꢀ1=2
f I
(13)
trans
a þ b
As a conclusion the stimulation of the interface generates
a gradient of the cis–trans composition over a typical length
scale, l, which varies with the light intensity with a power of ꢀ1/2,
while the total concentration of surfactant c (z) is constant.
tot
This composition gradient spans over 200 mm to 20 mm for
ꢀ2
ꢀ2
intensities between 1 mW cm and 100 mW cm . The amplitude
of the gradient Dc can be given by boundary conditions at the
interface by writing the stationary regime of the equations at the
interface.
Acknowledgements
The authors thank J. Ruchmann for performing the preliminary
measurements of dynamic surface tensions.
5
. Conclusion
References
An elementary model describing the barrier to adsorption and
the competition between the two isomers enabled us to quantify
the difference of affinity of the two conformations for the
interface. The trans isomer exhibits a higher affinity for the
interface than the cis isomer. When the interface is exposed to
light, whatever the wavelength, cis conformations are introduced
in the bulk and/or at the interface which in turn leads to signif-
icant desorption of the azoTAB. This ‘‘pumping-out’’ of the
1 X. Zhang, J. Colloid Interface Sci., 1997, 187, 166–178.
2
B. Scheid, J. Delacotte, B. Dollet, E. Rio, F. Restagno, E. A. van
Nierop, I. Cantat, D. Langevin and H. A. Stone, Europhys. Lett.,
2
010, 90, 24002.
3 R. Miller, P. Joos and V. B. Fainerman, Adv. Colloid Interface Sci.,
1994, 49, 249–302.
4
S. Khoukh, P. Perrin, F. Bes de Berc and C. Tribet, ChemPhysChem,
005, 6, 2009–2012.
R. F. Tabor, R. J. Oakley, J. Eastoe, C. F. J. Faul, I. Grillo and
2
5
R. K. Heenan, Soft Matter, 2009, 5, 78–80.
This journal is ª The Royal Society of Chemistry 2011
Soft Matter, 2011, 7, 7866–7874 | 7873

View Article Online
6
7
J. Eastoe and A. Vesperinas, Soft Matter, 2005, 1, 338.
J. Ruchmann, S. C. Sebai and C. Tribet, Macromolecules, 2011, 44,
22 N. C. Christov, K. D. Danov, P. A. Kralchevsky,
K. P. Ananthapadmanabhan and A. Lips, Langmuir, 2006, 22,
7528–7542.
23 T. Gil ꢀa nyi, I. Varga, C. Stubenrauch and R. M ꢀe sz ꢀa ros, J. Colloid
Interface Sci., 2008, 317, 395–401.
24 E. A. Simister, R. K. Thomas, J. Penfold, R. Aveyard, B. P. Binks,
P. Cooper, P. D. I. Fletcher, J. R. Lu and A. Sokolowski, J. Phys.
Chem., 1992, 96, 1383–1388.
6
04–611.
X. Tong, G. Wang, A. Soldera and Y. Zhao, J. Phys. Chem. B, 2005,
09, 20281–20287.
8
9
1
C. Alvarez-Lorenzo, L. Bromberg and A. Concheiro, Photochem.
Photobiol., 2009, 85, 848–860.
0 S. Khoukh, R. Oda, T. Labrot, P. Perrin and C. Tribet, Langmuir,
1
2
007, 23, 94–104.
25 K. Yim and G. Fuller, Phys. Rev. E: Stat. Phys., Plasmas, Fluids,
Relat. Interdiscip. Top., 2003, 67, 041601.
26 A. Bonfillon and D. Langevin, J. Colloid Interface Sci., 1994, 168,
497–504.
1
1
1
1 A. Salonen, D. Langevin and P. Perrin, Soft Matter, 2010, 6, 5308.
2 E. Chevallier, F. Lequeux, C. Tribet, C. Monteux, submitted.
3 T. Shang, K. A. Smith and T. A. Hatton, Langmuir, 2003, 19, 10764–
1
0773.
27 P. Joos and V. B. Fainerman, Dynamic Surface Phenomena, VSP,
1999.
28 H. Ritacco, D. Langevin, H. Diamant and D. Andelman, Langmuir,
2011, 27, 1009.
1
1
1
1
1
1
2
2
4 T. Hayashita, T. Kurosawa, T. Miyata, K. Tanaka and M. Igawa,
Colloid Polym. Sci., 1994, 272, 1611–1619.
5 B. A. Cicciarelli, T. A. Hatton and K. A. Smith, Langmuir, 2007, 23,
4
753–4764.
29 V. Fainerman, Colloids Surf., A, 1998, 143, 141–165.
30 A. C. F. Ribeiro, V. M. M. Lobo, A. J. M. Valente, E. F. G. Azevedo,
M. G. Miguel and H. D. Burrows, Colloid Polym. Sci., 2004, 283, 277–
283.
31 H. Diamant and D. Andelman, J. Phys. Chem., 1996, 100, 13732–
13742.
32 P. Vlahovska, J. Colloid Interface Sci., 1997, 192, 194–206.
33 C. A. MacLeod and C. J. Radke, Langmuir, 1994, 10, 3555–3566.
34 S. Datwani, J. Colloid Interface Sci., 1999, 219, 282–297.
35 M. Mulqueen and D. Blankschtein, Langmuir, 1999, 15, 8832–8848.
36 V. B. Fainerman, E. V. Aksenenko, J. T. Petkov and R. Miller, Soft
Matter, 2010, 6, 4694–4700.
6 M. Bonini, D. Berti, J. M. Di Meglio, M. Almgren, J. Teixeira and
P. Baglioni, Soft Matter, 2005, 1, 444.
7 C. T. Lee, K. A. Smith and T. A. Hatton, Langmuir, 2009, 25, 13784–
13794.
8 A. Diguet, R. Guillermic, N. Magome, A. Saint-Jalmes, Y. Chen,
K. Yoshikawa and D. Baigl, Angew. Chem., 2009, 121, 9445–9448.
9 B. A. Cicciarelli, J. A. Elia, T. A. Hatton and K. A. Smith, Langmuir,
2007, 23, 8323–8330.
0 H. Rau, Azo Compounds, in Photochromism Molecules and Systems,
ed. H. D u€ rr, Elsevier., 2003.
1 V. B. Fainerman, R. Miller and P. Joos, Colloid Polym. Sci., 1994,
272, 731–739.
37 J. Lu, Adv. Colloid Interface Sci., 2000, 84, 143–304.
7
874 | Soft Matter, 2011, 7, 7866–7874
This journal is ª The Royal Society of Chemistry 2011