On The Universality of Marangoni-Driven

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

.

On the universality of Marangoni-driven


arXiv:1712.03192v2 [physics.flu-dyn] 11 Dec 2017

spreading along liquid–liquid interfaces


Bram F. van Capelleveen1 †, Robin B. J. Koldeweij1,2 †, Detlef
Lohse1,4 and Claas Willem Visser1,3 ‡
1
Physics of Fluids group & Max Planck – Twente Center for Complex Fluid Dynamics,
Department of Science and Technology, J. M. Burgers Center for Fluid Dynamics, University of
Twente, 7500 AE Enschede, The Netherlands
2
Equipment for Additive Manufacturing, TNO, 5612 AP Eindhoven, The Netherlands
3
Max Planck Institute for Dynamics and Self-Organization, 37077, Göttingen, Germany
4
John A. Paulson School of Engineering and Applied Sciences, Wyss Institute for Biologically
Inspired Engineering, Harvard University, Boston, Massachusetts 02115, USA.

(Received xx; revised xx; accepted xx)

When two liquids of different surface tensions come into contact, the liquid with lower
surface tension spreads over the other as this configuration minimizes the total free energy.
This Marangoni-driven spreading has been studied for various geometries and surfactants,
but the dynamics of the binary geometry (drop-drop) has hardly been quantitatively
investigated, despite its relevance for droplet encapsulation applications. It is furthermore
unclear whether these different situations possess universal dynamics. Here we use laser-
induced fluorescence (LIF) to observe the dynamics of binary droplets as a function
of surface tension difference between the liquids ∆σ and viscosity η. The distance L
over which the low-surface-tension liquid has covered the high-surface-tension droplet is
measured. The early-stage spreading dynamics is described by the power-law L(t) ∼ tα
with α = 3/4, as expected based on scaling arguments for viscosity-limited spreading on
a deep bath. This exponent decreases in a later stage, when the boundary layer thickness
becomes comparable to the droplet diameter D. Spreading √ is dissipation-inhibited for
viscosities corresponding to Ohnesorge numbers Oh = η/ ρ∆σD & 0.2 where ρ is the
density. Finally, we show that our results collapse onto a single curve of rescaled distance
L̃(t̃) as a function of rescaled time t̃, which also captures previous results for different
geometries, surface tension modifiers (both solvents and surfactants), and miscibility. As
this curve spans 7 orders of magnitude, we conclude that Marangoni-induced spreading is
a universal phenomenon for most practically encountered liquid-liquid systems.

Key words: Encapsulation, droplet spreading, surface tension, impact, surfactant

1. Introduction
Pliny the Elder (AD 23-79) first noted in his natural history that oil, when poured
onto rough water, would cover and smoothen the surface. Benjamin Franklin (1774),
while under way to London, observed that boats sailing in the wake of jettisoned cooking
† B. F. van Capelleveen and R. B. J. Koldeweij contributed equally to this work and should
be considered as co-first authors
‡ Email address for correspondence: c.visser@utwente.nl
2 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser
grease would do so more smoothly. Intrigued by the phenomenon, he set out to investigate
how far oils would spread when deposited on the surfaces of ponds. Much later it was
Lord Rayleigh (1890) who postulated that this film of oil would consist of a mono-layer,
which enabled him to rather accurately identify the size of a molecule of oil. However, no
conclusive theory of spreading was formulated. Since then, the spreading of oil over water
has received theoretical (e.g. Jensen 1995), experimental (e.g. Berg 2009) and numerical
(e.g. Karapetsas et al. 2011) investigations, and is now well-understood. Furthermore,
the spreading of liquid droplets onto baths with a different surface tension has been
studied for miscible (Santiago-Rosanne et al. 2001) and immiscible (Sinz et al. 2011)
liquids and surfactant-based surface tension modifiers (Hanyak et al. 2012) as well as pure
liquids (Hernández-Sánchez et al. 2015). In all cases, different liquids and non-uniform
concentrations at the liquid-liquid interface give rise to interfacial tension gradients. The
associated surface stresses pull the liquid from the low-surface tension regions to the
regions with a higher surface tension, resulting in Marangoni flow.
The distance over which the low surface tension liquid spreads in time can be described
by a simple scaling law:

L(t) = βtα , (1.1)

with a spreading exponent α and a prefactor β which carries dimensions. For spreading on
a bath much deeper than the boundary layer thickness δBL (detailed below), an exponent
of α = 3/4 is found (Ruckenstein et al. 1970; Suciu et al. 1970; Huh et al. 1975; Joos
& Pintens 1977; Dipietro et al. 1978; Foda & Cox 1980; Camp & Berg 1987; Fraaije
& Cazabat 1989; Phillips 1996; Berg 2009). Related cases have also been investigated,
such as the influence of evaporation or absorption on deep-layer spreading for which the
spreading exponent decreases to α = 1/2 (Dussaud & Troian 1998; Chauhan et al. 2000;
Santiago-Rosanne et al. 2001). On the other hand, α = 1/2 is also observed for spreading
on thin layers (Ahmad & Hansen 1972; Borgas & Grotberg 1988; Gaver & Grotberg 1990;
Jensen & Grotberg 1992; Starov et al. 1997; Afsar-Siddiqui et al. 2003; Bull & Grotberg
2003; Hernández-Sánchez et al. 2015).
However, little research has been done on liquid on liquid spreading in different
geometries, most notably the case of two droplets meeting in air as depicted in figure
1a. This binary droplet geometry is of particular interest for encapsulation of droplets,
as applied in electronics (Ten Cate et al. 2014), food or vitamin encapsulation (Blanco-
Pascual et al. 2014), bio-fabrication (Duan et al. 2013), pharmacy (Yeo et al. 2003, 2004),
and the creation of complex microparticles (Hayakawa et al. 2016). Improving these
technologies is challenging (Nedovic et al. 2011), as spreading due to a difference in
surface tension in binary droplet systems is mostly studied without taking into account
the dynamics of the encapsulation front (Chen & Chen 2006; Chen 2007; Thoroddsen
et al. 2007; Planchette et al. 2010; Blanchette 2010; Focke & Bothe 2012; Nowak et al.
2016, 2017). In particular, the influence of surface tension and viscosity on the spreading
dynamics is still unknown, and it is unclear to what extent results for different geometries,
surfactants, and miscibility can be applied to binary droplets (or vice versa).
In this paper we study the spreading and subsequent encapsulation of droplets with
imbalanced surface tensions over each other in the binary drop geometry. By applying
stroboscopic imaging to this problem for the first time, we observe and measure the
Marangoni flow associated with the spreading of liquids over each other. To assess whether
the scaling laws developed for spreading on a liquid substrate are universal, we compare
our results to systems with different geometries, surfactants, and miscibility.
On the universality of Marangoni-driven spreading 3

a +LJKı /RZı
x
bx capillary
forces
1 2 Contact
U0
z
viscous
Coalescence drag dBL

} c
Time

Marangoni-driven
film spreading (this work) D0/4
L(t) x
z
back
flow
Encapsulation

Final stage

Figure 1. (Color online) Overview of binary drop spreading; (a) At t = 0 the drops touch.
Coalescence is initially dominant, opening the neck due to local curvature (blue arrows). This
regime is followed by Marangoni-driven spreading of the droplet with lower σ over the other one
(green arrows). The leading edge position is given as L(t). Ultimately, this mechanism results
in encapsulation of droplet 1. (b) Marangoni-driven spreading in the early phase, where the
Marangoni stress is balanced by the viscous boundary layer. The curved surface of a drop is
given as the x-direction; z points into the inner droplet. (c) Spreading in the late phase, where a
back-flow limits the development of the boundary layer.

2. Governing scaling laws


2.1. Dimensionless numbers & force balance
Figure 1a shows the stages of spreading for two droplets of different surface tension.
Upon contact, the small radius of curvature results in a capillary flow that rapidly increases
the contact radius at constant velocity (Paulsen et al. 2011), followed by a decrease in
velocity (Thoroddsen et al. 2007). Subsequently, the low-surface-tension droplet (2) covers
the high-surface tension droplet (1) by Marangoni-driven spreading over a distance L(t).
Understanding the dynamics of this L(t) is the focus of this work. Eventually, droplet
2 fully encapsulates droplet 1 and both interfaces obtain a spherical shape by surface
tension.
In the following, we recapitulate spreading models which we will later apply to the
drop-drop geometry. Here, D, ρ, η, and σ denote diameter, density, viscosity, and surface
tension, respectively; indices 1 and 2 refer to the droplets as shown in figure 1a. The
densities of both droplets are assumed to be equal, and the difference in surface tension
is ∆σ = σ1 − σ2 . The influence of gravity is neglected as the Bond number ρi gL2 /∆σ of
the droplet is smaller than 1. Furthermore, we assume that mass diffusion is insignificant
even though we use miscible liquids, as the ratio between advective to diffusive forces
is typically Pe = ∆σD0 /η1 φs ∼ 107 , where Pe indicates the Péclet number and φs is
the mass diffusivity. The stress balance at an interface with surface tension gradients
balanced by the viscous stress is given by:
∇s σ + τ1 + τ2 = 0, (2.1)
4 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser
where ∇s is the surface gradient operator, and τi = ηi ∂Ux /∂z denotes the viscous stresses
in the liquids. The following stress balance can be obtained:

dσ dUx dUx
∼ −η1 − η2 , (2.2)
dx dz z=0 dz z=0
with Ux the velocity of liquid 1 in the x-direction. Equation (2.2) was solved using scaling
arguments (Berg 2009), as widely validated for droplet spreading onto a liquid bath
(Joos & Pintens 1977; Foda & Cox 1980; Camp & Berg 1987). For the binary droplet
geometry we assume that the depth of the “bath” is provided by the inner drop’s radius.
Furthermore, as the thin film is free at the liquid-air surface, the shear stress within the
film is assumed to be negligible and we ignore the second viscous term in (2.2).
Two cases can be distinguished. First, if the viscous boundary layer is much smaller
than the drop’s diameter, spreading essentially occurs like onto a deep bath. The velocity
must vanish far away from the surface. Following previous studies on Marangoni-driven
spreading (Joos & Pintens 1977; Bergeron & Langevin 1996; Berg 2009), we approximate
the velocity profile in the boundary layer by an exponential function (as discussed in
detail by (Van Den Tempel & Van De Riet 1965; Lucassen 1967)):
Ux (t, x, z) = Ux (t, x, z = 0)e−z/δBL , (2.3)
where δBL = (η1 t/ρ1 )1/2 refers to the boundary layer penetration thickness as shown in
figure 1b (Blasius 1908). For miscible liquids, the surface tension gradient is assumed to
scale as dσ/dx ∼ ∆σ/L(t) (Jensen 1995). Substituting the velocity profile of equation
(2.3) in equation (2.2) provides:

∆σ Ux (t, x, z = 0)
∼ η1 . (2.4)
L(t) δBL
Together with the formulation for the boundary layer and describing the velocity of the
spreading front by U (t, x, z = 0) ∼ L̇(t), equation (2.4) results in the following differential
equation:

r
∆σ η1 t
L(t)L̇(t) ∼ . (2.5)
η1 ρ1
Integration of this equation yields the scaling law for the spreading distance:

∆σ 1/2
L(t) ∼ 1/4
t3/4 . (2.6)
(ρ1 η1 )
Second, if the viscous boundary layer thickness becomes of the order of the half radius
(see figure 1c), the exponential velocity decay is expected to break down. Although the
actual flow geometry is 3-dimensional and may be complex (Blanchette & Bigioni 2009),
we assume a zero velocity boundary condition at z = D1 /4, as this value would provide a
coarse estimate for both the boundary layer and the back-flow along the droplet’s axis.
Furthermore, we assume that this flow is described by a Couette flow, as has previously
been done for the spreading of a droplet on a bath (Ahmad & Hansen 1972):
 
4z
Ux (t, x, z) = Ux (t, x, z = 0) 1 − . (2.7)
D
With this flow profile, the penetration depth and the scaling for the surface tension
On the universality of Marangoni-driven spreading 5
a (Top view) b c d Analysis
overlay

Camera 1 2 1 2 1 2
Microscope
Objective Frame 2
Frame 1
¨t.0.5ȝs %Tcap)
Pulse #1
Dichotous
e (Frontal view)
Syringe Pump Trigger system
Mirror
g
Laser Beam
Expander
Dn
1 2 1 2

Antiwetting
Drop Needles 1 2 10x
Trigger
Diffuser D0
x LIF x
Pulse #2
z y
Pulse Generator

Figure 2. (Color online) (a) Top view of the set-up (schematic). Droplets 1 and 2 are illuminated
by pulse #1, as shown by the blue arrows. Fluorescent light is emitted by droplet 1 and passes
the dichroic mirror, as indicated by the red arrow. An example fluorescent image is shown in
(b). After ∆t = 0.5µm, the droplets are also illuminated by pulse #2, resulting in a bright-field
image as shown in (c). A processed overlay of both images is shown in (d). The entire image
capturing sequence is repeated with different delays after droplet contact, resulting in time series
of the spreading front. Figure (e) shows the setup from the side-view perspective of the camera.
(Inset) Schematic drawing of the dedicated trigger system. The conducting droplets act as a
switch. The resulting signal on contact is amplified and fed to the pulse generator controlling the
timing system.

gradient and velocity, equation (2.2) reduces to:


 1/2
∆σ
L(t) ∼ h t1/2 . (2.8)
η1
As stated above, in both cases the spreading distance is described by L(t) ∼ βtα .
The validity of (2.6) and (2.8) for the droplet on droplet geometry will be investigated
experimentally by independently changing ∆σ, η1 , or η2 .

3. Experimental set-up and materials


3.1. Experimental set-up and measuring techniques
To create the binary drop system, two droplets are suspended from a needle. Both
droplets are grown by use of the same syringe pump (Harvard PHD 2000) with identical
syringes. The growth-rate is varied for optimal droplet generation, while maintaining
the dripping regime (i.e. W e = ρU 2 L/σ < 1). The size of the droplets can be varied by
adjusting the distance between the needles, as the moment of contact corresponds to the
start of coalescence and the subsequent spreading process. Images obtained from a dual
6 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser
frame camera (PCO Sensicam qe) are used to to calculate the initial droplet diameter D0
and track the spreading front. Teflon coated needles (Hamilton Company) are used to
prevent spurious wetting.
Due to the limited frame-rate of the dual frame camera, all image sequences are taken
with stroboscopic illumination. A schematic representation of the lighting system is given
in figure 2a. A pulsed laser (Litron Nano S PIV 400 mJ with a wavelength of 532 nm)
is triggered to illuminate the droplet pair from the front. A fluorescent inner phase
(fluorescein, emission at 525 nm) is used in the droplet with the highest surface tension,
while a 15 vol% ink (Brother LC-800) solution is used as low σ base. Using this opaque
liquid with known optical extinction length (shown in the supplementary material) allows
visualization of the encapsulation by the spreading layer as a growing black region as
displayed in figure 2b.
Figure 2c shows the second frame from the dual shutter camera. To produce broadband
incoherent illumination while keeping the short exposure time (van der Bos et al. 2011),
this bright field image is illuminated by a fluorescent diffuser (Part number 1108417,
LaVision) that, in turn, is exposed to a second pulsed laser (Evergreen 600 mJ with a
wavelength of 532 nm). Varying the delay between the trigger signal and laser pulse leads
to an image sequence of the spreading. Both camera frames are taken within 0.5 µs of
each other, a short time over which no significant motion occurs. In order to curtail errors,
all measurements are conducted three times.
To ensure a reproducible starting time t = 0 for each experiment, the droplets are
used to close a home-developed electrical circuit and thereby initiate the electrical trigger
as shown in Figure 2e. Conductivity is achieved by adding 1 vol% NaCl solution to all
samples and applying a constant electrical potential between the needles. NaCl greatly
increases the conductivity of the bulk liquid, while retaining the surface tension for low
voltages (Schmid et al. 1962).

3.2. Materials
D0 is taken as the diameter of the droplet to be encapsulated. In a typical experiment
the pump-rate is 5 ml min−1 , resulting in the release of droplets with a diameter of
2.5 ± 0.5mm. With spreading events lasting for 5–10 ms, the increase of D0 is ∼ 0.1%
negligible throughout a measurement.
To achieve surface tension gradients, mixtures of milli-Q water, ethanol, glycerol and
Brother LC-800 ink are prepared. This resulted in a small viscosity variability of 1 to 2
mPa s, which has an influence of maximal 6% on the prefactor β as estimated based on
figure 4. For changing the viscosity ratio between the droplets, we adapt η1 for η1 /η2 > 1,
while η2 is adapted for η1 /η2 < 1. In both cases, the viscosity of the other liquid is
maintained at 1mPa s. The ranges of material parameters are listed in table 1. Liquid
parameter values are taken from literature or obtained experimentally when mixtures
include ink. The measured properties of the used liquids are shown in the supplementary
material.

3.3. Data acquisition & Analysis


An example image sequence is shown in figure 3. The sequences in figures 3a and 3b
show the significant dynamics of both the spreading film and the droplets’ contours,
respectively. The images in (a) and (b) were captured with a delay of 0.5 µs, but virtually
no motion occurs between
p these images as this delay is much smaller than the capillary
time scale (Tcap = ρR03 /σ ≈ 5 ms). The overlay in figure 3c confirms their excellent
spatial collapse.
On the universality of Marangoni-driven spreading 7

Droplet 1 Droplet 2 Droplets


Surface tension Viscosity Surface tension Viscosity Initial diameter
(mN m−1 ) (mPa s) (mN m−1 ) (mPa s) (mm)
Range 46.2 to 72.8 1 to 100 22.9 to 46.2 1 to 100 2.00 to 2.92
Error 5% 2% 5% 2% 0.1%
Table 1. Range of the employed material parameters. Errors are given in relative form. Mixtures
of Milli-Q water, ethanol, glycerol, and Brother LC-800 ink were used to create these liquids.
Literature values are taken for those mixtures not including ink (Vazquez et al. 1995; Alkindi
et al. 2008; Khattab et al. 2012; Takamura et al. 2012). All data for mixtures containing ink
obtained from measurements performed at room temperature.

The analysis is performed as follows. The area through which less than 5% intensity is
transmitted is compared to uncovered areas of the drop, and classified as being covered by
the film. Based on the optical penetration depth measurements, this corresponds to a film
thickness of approximately 63 µm. An example of the resulting image analysis sequence
is given in figure 3d. We trace the spreading along the bottom of the droplet pair to
find the 3-dimensional spreading contour. By overlaying the resulting binary images, we
observe the spreading front of Marangoni-induced spreading in the binary drop system,
represented by the red lines in figure 2d and in figure 3d.

4. Results
4.1. Spreading as a function of time and viscosity
Based on the time-resolved spreading dynamics as shown in figure 3, we determine the
position of the spreading front L(t) as a function of time in figure 4. For low viscosities
(η1 ≈ η2 6 20), the data is consistent with power-law behavior with a scaling exponent
α = 3/4. Increasing the viscosity to η1 ≈ η2 = 50mPa s leads to a significant decrease
in spreading rate, and increasing the viscosity even further prevents the spreading. The
Ohnesorge number
p
Oh = η/ ρ∆σD (4.1)
for this case is Oh ≈ 0.2, i.e. spreading seems to be inhibited when global viscous forces
become comparable to surface tension forces. This suggests a transition from Marangoni
spreading to coalescence of droplets with equal surface tensions for Oh ∼ O(1).

4.2. Spreading as a function of ∆σ


Figure 5a shows the position of the leading edge of the spreading front as a function
of the surface tension difference, which was varied from ∆σ = 0.4mN m−1 to ∆σ = 32.1
mN m−1 . The spreading rate can be described with power-law behavior L ∼ tα , with the
exception of very low ∆σ = 0.4 mN m−1 for which spreading is inhibited. This threshold
occurs at Oh ≈ 0.15, i.e. at a similar value as the suppression threshold for more viscous
droplets (see previous paragraph). For ∆σ > 10mN m−1 , the expected value for initial
spreading (α = 3/4) is observed within error, as shown in figure 5b. The transition
between these regimes is remarkable: rather than a steady increase of α towards 3/4,
an abrupt increase is observed to α ≈ 0.5 at ∆σ ≈ 1mN m−1 , followed by a gradual
increase to 3/4. The origin of this intermediate value could lie in evaporation of the
droplet (Dussaud & Troian 1998; Chauhan et al. 2000; Santiago-Rosanne et al. 2001),
8 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser

g
0.1 1.1 2.1 3.1 4.1 ms
b

c Front

D0

Capillary
d Waves

L(t)

Time

Figure 3. (Color online) Example of image analysis; (a) Sequence of images obtained with setup
of figure 2. Spreading is from right to left. The numbers indicate the time after first contact
occurred, Tcap ≈5 ms; (b) Subsequently taken iLIF images, ∆t = 0.5µs; (c) Overlay of (a) and
(b) showing temporal accuracy and visualizing the spreading of the right droplet over the left.
Capillary waves and spreading front are readily observed; (d) Analysis result after image sequence
(a) is converted to binary form with an intensity threshold of 5%. The singular red dot indicates
the x-position of merging, while the red line indicates the spreading distance of the low-surface
tension solution. The scale-bar corresponds to 1 mm.

or an intermediate regime around ∆σ ≈ 1mN m−1 in which tangential gradients in the


surface tension affect the spreading, for which α = 3/8 was proposed (Jensen 1995).

4.3. Variation in viscosity ratio


The effect of changing the viscosity ratio η1 /η2 between the droplets is shown in figure
6a. Increasing the viscosity of either liquid results in a decrease in spreading over the
entire temporal domain, as expected based on the viscosity-dependent prefactor provided
in (2.6). Although figure 6a suggests that the spreading exponents are relatively close
to 3/4, their quantification resulted in scattered data as shown in figure 6b. Hence, no
definitive trend could be established. Spreading is virtually inhibited for η1 /η2 = 0.01,
again corresponding to Oh ∼ O(1) with an Ohnesorge number based on the more viscous
liquid. This leads to an added selection criterium for the outer droplet, compared to those
used in (Yeo et al. 2003).
For η1 /η2 = 90.50, spreading is reduced at later times t & 6ms. We suspect that the
On the universality of Marangoni-driven spreading 9

3
(mm)

4 4
3

0.5

0.3

0.2

(ms)

Figure 4. Time evolution of the leading edge position L(t) as a function of the viscosities ηi,j ,
with 18.9 6 ∆σ 6 23.2mN m−1 . A dependence of the viscosity on the spreading is apparent, as
also shown in equation (2.6). Dotted and dotted-dashed lines have slope 3/4 and are placed to
guide the eye.

boundary layer hits the back-flow at that moment, as shown in figure 1c. To assess this
hypothesis, we calculate the relative boundary layer thickness (for 50% of the surface
velocity) at t = 6ms, providing δBL = (η1 t/ρ1 )1/2 = 0.7mm, which is indeed close to
D0 /4 = 0.62mm ±0.12mm. At this moment, a regime change to thin film behavior
(equation (2.8)) may occur. However, as α could not be reliably established for the
applicable subset of measurements, this limiting behavior remains to be investigated.

4.4. Spreading universality


As a final step, we assess the universality of Marangoni-driven spreading. Based on
the inner phase liquid variables and the spreading parameter, time- and length-scales are
rescaled as (Huh et al. 1975):
L
L̃ = η12
(4.2)
∆σρ1
t
t̃ = η13
(4.3)
∆σ 2 ρ1

The resulting scaling is given in figure 7. Remarkably, all measurements follow a line
with slope 3/4, even though not all individual time series collapse well onto it. A deviation
is only observed for data with small η1 /η2 , as the outer droplet’s viscosity is not included
10 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser

(a) Dv (b)

1
4 2
3 3/4
1/2
1
(N m-1)
= 32.1 mN/m = 14.4 mN/m
= 26.1 mN/m = 12.3 mN/m
= 25.8 mN/m = 9.1 mN/m
= 23.2 mN/m = 8.0 mN/m
= 21.8 mN/m = 6.1 mN/m
= 20.4 mN/m = 3.9 mN/m
= 20.1 mN/m = 1.9 mN/m
= 15.5 mN/m = 0.4 mN/m

Figure 5. (Color online) (a) Time evolution of L(t) for various ∆σ. The dashed-dotted line has
slope 3/4, corresponding to equation (2.6), and the dashed line has slope 1/2, corresponding to
spreading with influence of absorption or evaporation (Dussaud & Troian 1998; Chauhan et al.
2000; Santiago-Rosanne et al. 2001). (b) α as a function of ∆σ obtained from best fit to lines in
(a). An upwards trend from α = 0, to an apparent plateau for α = 0.75 is visible.

(a) (b)

4
3
(mm)

1
1 / 2 = 90.50
1 1
/ 2 = 9.28
/ 2 = 72.19 1
/ 2 = 0.98
1
/ 2 = 45.45 / = 0.46
1 1 2
/ 2 = 37.54 1
/ 2 = 0.10
1
/ = 18.28 1
/ 2 = 0.05
1 2
/ = 16.15 1
/ 2 = 0.01
1 2

(ms)

Figure 6. (Color online) Time evolution of the spreading edge L(t) for various viscosity rates
η1 /η2 , with 19.8 6 ∆σ 6 28.3mN m−1 and corresponding spreading exponents α; (a) No unique
value of the spreading exponent α is readily observed. The dotted line has slope 1 and is drawn to
guide the eye. The dashed line has slope 3/4. The spreading prefactor β in L(t) = βtα decreases
with deviation from η1 /η2 = 1. One of the ηi was kept at 1mPa s to create the ratios shown; (b)
α as a function of η1 /η2 .

in the rescaled variables L̃ and t̃. Attempts to accommodate this parameter did not result
in their collapse to the master curve, possibly due to disregarding the viscosity of the
outer liquid in equation (2.2).
Figure 7 also includes literature data for droplet spreading on a bath, for surfactant-
On the universality of Marangoni-driven spreading 11

= 26.1 mN m-1
1
/ 2 = 90.50 h1=h2= 10.31 mPa s
= 25.8 mN m-1 h1=h2= 20.30 mPa s
1/ 2 = 72.19
= 23.2 mN m-1 1
/ 2 = 45.45 h1=h2= 50.47 mPa s
= 21.8 mN m-1 1
/ 2 = 37.54
/ 2 = 18.28 Increasing h1=h2
= 20.4 mN m-1 1

= 20.1 mN m-1 1
/ 2 = 16.15

= 15.5 mN m-1 1
/ 2 = 9.28
/ 2 = 0.46
= 14.4 mN m-1 1

1
/ 2 = 0.19
= 12.3 mN m-1
1 2
/ = 0.10
= 9.1 mN m-1
1 2
/ = 0.05
= 8.0 mN m-1
= 6.1 mN m-1 1 2
/ = 0.01 h1/h2 .
= 3.9 mN m-1 Decreasing h1/h2

Surfactants
{
= 1.9 mN m-1

{{
Joos & v.Hunsel (1984)
Berg (2009) DTAB
Decreasing Dv Berg (2009) SDS

Bath geometry
{
˜L=

Huh et al.(1975) 971 cs on water

Immiscible
h1.h2.50 mPa s
Huh et al.(1975) 9 cs on water

liquids
Huh et al.(1975) 48 cs on water
Camp & Berg (1987)

Kim et al.(2017) IPA on 1 cs


4 Kim et al.(2017) IPA on 2 cs

Miscible
liquids
Kim et al.(2017) IPA on 5 cs

3 Kim et al.(2017) methanol on 1 cs


Kim et al.(2017) methanol on 2 cs
Kim et al.(2017) methanol on 5 cs

˜t =

Figure 7. Rescaled spreading distance as a function of rescaled time. All our measurements
collapse onto a power-law with a slope of 3/4 (some of the data points are not visible due to the
many measurements included), with the exception of η1 /η2 < 1 for which spreading is reduced
(as indicated by the arrow). Values from literature of different geometry and for both immiscible
liquids and surfactant-based surface tension modifiers are reasonably described by the same
curve.

driven (Joos & Van Hunsel 1984; Berg 2009), immiscible (Huh et al. 1975; Camp & Berg
1987), and miscible liquids (Kim et al. 2017). These measurements seem to follow the
same scaling law, both for surfactant-driven spreading and pure liquids that possess a
different surface tension. The spreading dynamics also collapse for immiscible and miscible
liquids, which was expected for the small Péclet numbers used (see section 2.1). As the
dimensionless parameters are reasonably described by a single power law, over 7 orders of
magnitude, figure 7 provides a universal engineering tool for Marangoni-driven spreading
in various systems.

5. Conclusions & Outlook


In summary, we have investigated the dynamics of Marangoni-driven spreading of
droplets over droplets. Using a dual-frame camera and a robust trigger mechanism, we
obtained reproducible stroboscopic image sequences of spreading distance L(t). The results
were compared to power-law behavior L(t) = βtα , as predicted by existing theories for
spreading in different geometries.
The following conclusions are drawn: First, for surface tension differences ∆σ &
10 mN m−1 , Marangoni-driven spreading is described by L(t) ∼ t3/4 within error, as
predicted for spreading on deep layers. An equal change in the viscosity of both liquids
12 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser
reduces the prefactor, whereas the scaling exponent α = 3/4 is maintained. Second,
Marangoni-driven spreading is suppressed for Oh & 0.2, i.e. for cases where viscous forces
become globally comparable to capillary effects. Third, the influence of viscosity changes
in only one of the liquids proved challenging to quantify within the effective power law (eq.
(1.1)), and did not provide reliable values for α and β. For high viscosity inner droplets, a
sharp decrease in the spreading exponent to α ≈ 0.5 is observed after a few milliseconds.
This decrease coincides with a boundary layer thickness of δBL ∼ D0 /4, suggesting a
transition from spreading over a deep pool to spreading over thin films (eq. (2.8)).
Universal behavior is observed for Marangoni-driven spreading in general, as shown
in figure 7. Here, different geometries, miscibility, and driving mechanisms (surfactant
systems or pure liquids) collapse onto a single curve. Remarkably, even measurements
with different individual exponents are congruent to this curve. This universal behavior
may be readily exploited to engineer spreading and encapsulation in a wide range of
applications.

Acknowledgments
We gratefully acknowledge insightful discussions with S. Wildeman, S. Karpitschka,
J.H. Snoeijer and T. Kamperman, and we thank E. Zijlma for his contributions to the
dedicated electronics.

REFERENCES
Afsar-Siddiqui, Abia B., Luckham, Paul F. & Matar, Omar K. 2003 The spreading of
surfactant solutions on thin liquid films. Adv. Colloid Interface Sci. 106 (1-3), 183–236.
Ahmad, J. & Hansen, R. S. 1972 A simple quantitative treatment of the spreading of monolayers
on thin liquid films. J. Colloid Interface Sci. 38 (3), 601–604.
Alkindi, A. S., Al-Wahaibi, Y. M. & Muggeridge, A. H. 2008 Physical porperties (densisy,
excess molar volume, viscosity, surface tension and refractive index) of ethanol + glycerol.
J. Chem. Eng. Data 53, 2793–2796.
Berg, S. 2009 Marangoni-driven spreading along liquid-liquid interfaces. Phys. Fluids 21 (3).
Bergeron, V. & Langevin, D. 1996 Monolayer Spreading of Polydimethylsiloxane Oil on
Surfactant Solutions. Physical Review Letters 76 (17), 3152–3155.
Blanchette, F. 2010 Simulation of mixing within drops due to surface tension variations. Phys.
Rev. Lett. 105 (7), 13–16.
Blanchette, F. & Bigioni, T. P. 2009 Dynamics of drop coalescence at fluid interfaces. J.
Fluid Mech. 620, 333.
Blanco-Pascual, N., Koldeweij, R. B J, Stevens, R. S A, Montero, M. P., Gómez-
Guillén, M. C. & Cate, A. T Ten 2014 Peptide Microencapsulation by CoreShell
Printing Technology for Edible Film Application. Food Bioprocess Technol. 7 (9), 2472–
2483.
Blasius, H. 1908 Grenzschichten in Flüssigkeiten mit kleiner Reibung. Z. Math. Phys. 56, 1–37.
Borgas, Michael S. & Grotberg, James B. 1988 Monolayer flow on a thin film. J. Fluid
Mech. 193 (1988), 151.
van der Bos, A., Zijlstra, A., Gelderblom, E. & Versluis, M. 2011 ILIF: Illumination by
laser-induced fluorescence for single flash imaging on a nanoseconds timescale. Exp. Fluids
51 (5), 1283–1289.
Bull, J. L. & Grotberg, J. B. 2003 Surfactant spreading on thin viscous films: Film thickness
evolution and periodic wall stretch. Exp. Fluids 34 (1), 1–15.
Camp, D. W. & Berg, J. C. 1987 The spreading of oil on water in the surface tension regime.
J. Fluid Mech. 184, 445–462.
Chauhan, A., Svitova, T. F. & Radke, C. J. 2000 A Sorption-Kinetic Model for Surfactant-
Driven Spreading of Aqueous Drops on Insoluble Liquid Substrates. J. Colloid Interface
Sci. 222 (2), 221–232.
On the universality of Marangoni-driven spreading 13
Chen, R. 2007 Dieseldiesel and dieselethanol drop collisions. Appl. Therm. Eng. 27 (2-3),
604–610.
Chen, R. & Chen, C. 2006 Collision between immiscible drops with large surface tension
difference: diesel oil and water. Exp. Fluids 41 (3), 453–461.
Dipietro, N. D., Huh, C. & Cox, R. G. 1978 The hydrodynamics of the spreading of one
liquid on the surface of another. J. Fluid Mech. 84 (03), 529–549.
Duan, B., Hockaday, L. A., Kang, K. H. & Butcher, J. T. 2013 3D Bioprinting of
heterogeneous aortic valve conduits with alginate/gelatin hydrogels. J. Biomed. Mater.
Res. - Part A 101 A (5), 1255–1264.
Dussaud, Anne D. & Troian, Sandra M. 1998 Dynamics of spontaneous spreading with
evaporation on a deep fluid layer. Phys. Fluids 10 (1), 23.
Focke, C. & Bothe, D. 2012 Direct numerical simulation of binary off-center collisions of shear
thinning droplets at high Weber numbers. Phys. Fluids 24 (7), 073105.
Foda, M. & Cox, R. G. 1980 The spreading of thin liquid films on a water-air interface. J.
Fluid Mech. 101 (1), 33.
Fraaije, J. G. E. M. & Cazabat, A. M. 1989 Dynamics of spreading on a liquid substrate. J.
Colloid Interface Sci. 133 (2), 452–460.
Franklin, B., Brownrigg, W. & Farish, M. 1774 Of the Stilling of Waves by means of Oil.
Philos. Trans. R. Soc. London 64 (0), 445–460.
Gaver, Donald P. & Grotberg, James B. 1990 The dynamics of a localized surfactant on a
thin film. J. Fluid Mech. 213, 127.
Hanyak, M., Sinz, D. K. N. & Darhuber, A. A. 2012 Soluble surfactant spreading on spatially
confined thin liquid films. Soft Matter 8 (29), 7660.
Hayakawa, Masayuki, Onoe, Hiroaki, Nagai, Ken H. & Takinoue, Masahiro 2016
Complex-shaped three-dimensional multi-compartmental microparticles generated by
diffusional and Marangoni microflows in centrifugally discharged droplets. Scientific Reports
6 (February), 20793.
Hernández-Sánchez, J. F., Eddi, A. & Snoeijer, J. H. 2015 Marangoni spreading due to a
localized alcohol supply on a thin water film. Phys. Fluids 27 (3), 032003.
Huh, C., Inoue, M. & Mason, S. G. 1975 Uni-directional spreading of one liquid on the surface
of another. Can. J. Chem. Eng. 53 (4), 367–371.
Jensen, O. E. 1995 The spreading of insoluble surfactant at the free surface of a deep fluid
layer. J. Fluid Mech. 293 (-1), 349.
Jensen, O. E. & Grotberg, J. B. 1992 Insoluble surfactant spreading on a thin viscous film:
shock evolution and film rupture. J. Fluid Mech. 240, 259.
Joos, P. & Pintens, J. 1977 Spreading kinetics of liquids on liquids. J. Colloid Interface Sci.
60 (3), 507–513.
Joos, P. & Van Hunsel, J. 1984 Spreading of aqueous surfactant solutions on organic liquids.
J. Colloid Interface Sci. 106 (1), 161–167.
Karapetsas, G., Craster, R. V. & Matar, O. K. 2011 Surfactant-driven dynamics of liquid
lenses. Phys. Fluids 23 (12).
Khattab, I. S., Bandarkar, F., Fakhree, M. A. A. & Jouyban, A. 2012 Density, viscosity,
and surface tension of water+ethanol mixtures from 293 to 323K. Korean J. Chem. Eng.
29 (6), 812–817.
Kim, H., Muller, K., Shardt, O., Afkhami, S. & Stone, H. A. 2017 Solutal Marangoni
flows of miscible liquids drive transport without surface contamination. Nature Physics
advance online publication, 1–7.
Lord Rayleigh 1890 Measurements of the amount of oil necessary in order to check the motions
of camphor upon water. Proc. R. Soc. London 47, 364–367.
Lucassen, J. 1967 Longitudinal Capillary Waves. Transactions of the Faraday Society pp.
2221–2229.
Nedovic, V., Kalusevic, A., Manojlovic, V., Levic, S. & Bugarski, B. 2011 An overview
of encapsulation technologies for food applications. Procedia Food Sci. 1, 1806–1815.
Nowak, E., Kovalchuk, N. M., Che, Z. & Simmons, M. J. H. 2016 Effect of surfactant
concentration and viscosity of outer phase during the coalescence of a surfactant-laden
drop with a surfactant-free drop. Colloids Surfaces A Physicochem. Eng. Asp. pp. 1–8.
Nowak, E., Xie, Z., Kovalchuk, N. M., Matar, O. K. & Simmons, M. J. H. 2017 Bulk
14 B. F. van Capelleveen, R. B. J. Koldeweij, D. Lohse and C. W. Visser
advection and interfacial flows in the binary coalescence of surfactant-laden and surfactant-
free drops. Soft Matter 13 (26), 4616–4628.
Paulsen, J. D., Burton, J. C. & Nagel, S. R. 2011 Viscous to inertial crossover in liquid
drop coalescence. Phys. Rev. Lett. 106 (11), 1–4, arXiv: 1012.1298.
Phillips, W. R C 1996 On the spreading radius of surface tension driven oil on deep water.
Appl. Sci. Res. 57 (1), 67–80.
Planchette, C., Lorenceau, E. & Brenn, G. 2010 Liquid encapsulation by binary collisions
of immiscible liquid drops. Colloids Surfaces A Physicochem. Eng. Asp. 365 (1-3), 89–94.
Pliny the Elder & Rackham, H. 1938 Natural history. Harvard University Press.
Ruckenstein, E., Smigelschi, O. & Suciu, D. G. 1970 A steady dissolving drop method for
studying the pure Marangoni effect. Chem. Eng. Sci. 25 (8), 1249–1254.
Santiago-Rosanne, M., Vignes-Adler, M. & Velarde, M. G. 2001 On the Spreading of
Partially Miscible Liquids. J. Colloid Interface Sci. 234 (2), 375–383.
Schmid, G M, Hurd, R M & Snavely, E S 1962 Effects of Electrostatic Fields on the Surface
Tension. J. Electrochem. Soc. 109 (9), 852–858.
Sinz, D. K. N., Hanyak, M. & Darhuber, A. A. 2011 Immiscible surfactant droplets on
thin liquid films: Spreading dynamics, subphase expulsion and oscillatory instabilities. J.
Colloid Interface Sci. 364 (2), 519–529.
Starov, Victor M., de Ryck, Alain & Velarde, Manuel G. 1997 On the Spreading of an
Insoluble Surfactant over a Thin Viscous Liquid Layer. J. Colloid Interface Sci. 190 (1),
104–113.
Suciu, D. G., Smigelschi, O. & Ruckenstein, E. 1970 The spreading of liquids on liquids. J.
Colloid Interface Sci. 33 (4), 520–528.
Takamura, K., Fischer, H. & Morrow, N. R. 2012 Physical properties of aqueous glycerol
solutions. J. Pet. Sci. Eng. 98-99, 50–60.
Ten Cate, A. T., Gaspar, C. H., Virtanen, H. L. K., Stevens, R. S. A., Koldeweij,
R. B. J., Olkkonen, J. T., Rentrop, C. H. A. & Smolander, M. H. 2014 Printed
electronic switch on flexible substrates using printed microcapsules. J. Mater. Sci. 49 (17),
5831–5837.
Thoroddsen, S. T., Qian, B., Etoh, T. G. & Takehara, K. 2007 The initial coalescence of
miscible drops. Phys. Fluids 19 (7), 072110.
Van Den Tempel, M. & Van De Riet, R. P. 1965 Damping of waves by surface-active
materials. The Journal of Chemical Physics 42 (8), 2769–2777.
Vazquez, G., Alvarez, E. & Navaza, J. M. 1995 Tension of Alcohol + Water from 20 to 50
deg. J. Chem. Eng. Data 40 (3), 611–614.
Yeo, Yoon, Basaran, Osman A. & Park, Kinam 2003 A new process for making reservoir-
type microcapsules using ink-jet technology and interfacial phase separation. Journal of
Controlled Release 93 (2), 161–173.
Yeo, Yoon, Chen, Alvin U., Basaran, Osman A. & Park, Kinam 2004 Solvent exchange
method: A novel microencapsulation technique using dual microdispensers. Pharmaceutical
Research 21 (8), 1419–1427.

You might also like