See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318207030

Analysis of fluid flow and particle transport in evaporating droplets exposed

to infrared heating

Article · October 2013

CITATION READS

1 400

1 author:

Ashish Kumar Thokchom

Shiv Nadar University
19 PUBLICATIONS 215 CITATIONS

SEE PROFILE

All content following this page was uploaded by Ashish Kumar Thokchom on 05 July 2017.

The user has requested enhancement of the downloaded file.

 International Journal of Heat and Mass Transfer 68 (2014) 67–77

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

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

Analysis of fluid flow and particle transport in evaporating droplets

exposed to infrared heating
Ashish Kumar Thokchom a, Abhishek Gupta b, Pallippadan Johny Jaijus c, Anugrah Singh a,⇑
a
Indian Institute of Technology Guwahati, Guwahati, Assam, India
b
Fluidyn Software & Consultancy Private Limited, Bangalore, Karnataka, India
c
Center for Bioengineering, Department of Mechanical Engineering, University of Canterbury, Christchurch, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: Analysis of fluid flow and particle transport inside evaporating droplets exposed to external radiation was
Received 15 May 2013 carried out by experiments and numerical simulations. In this study, we have shown that by altering the
Received in revised form 4 September 2013 free surface temperature we can modify the fluid flow profile inside the droplet and hence the deposition
Accepted 6 September 2013
pattern of solute particles on the substrate. The fluid velocity and particle concentration profiles inside
Available online 3 October 2013
the evaporating droplet were measured by Particle Image Velocimetry (PIV) technique. Experiments were
carried out on a small sessile water droplet containing dispersed polystyrene particles. To avoid problem
Keywords:
of image correction encountered in PIV measurements with 3D droplets, our experiments were per-
Evaporating droplet
Particle Image Velocimetry
formed on an equivalent disc shaped 2D drop sandwiched between two non-wetting surfaces, while
Pattern formation the base of the droplet was pinned to a wetting surface. The top surface of the droplet was heated by
Marangoni flow Infrared (IR) light. The temperature of droplet surface was measured by thermocouples. The velocity field,
Computational fluid dynamics particle concentration profile and particle deposition patterns were studied during evaporation process.
We have also performed numerical simulations by solving continuity, momentum and energy transport
equations. The computed velocity profiles resulting from buoyancy and Marangoni convection are in
qualitative agreement with the experiments.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Nakoryakov et al. [11] have studied the behavior of evaporating

A ring shaped deposition pattern rather than uniform deposi-
tion is observed when a liquid droplet containing dispersed parti-
cles dries out (due to evaporation) on a wetting solid surface. This
natural phenomenon is important in many applications such as
DNA microarray [1–3], ink-jet printing [4], diseases diagnosis [5],
thin film coatings [6] and manufacture of novel optical and elec-
tronic materials [7]. In these applications the phenomenon of the
fluid flow inside the evaporating droplet plays a key role in depo-
sition patterns. Deegan et al. [8–10] have extensively reported this
phenomenon (commonly known as coffee-ring effect) which has
received increased attention in the recent years. They explained
that a ring like deposition pattern is due to the pinning of contact
line, and the solvent lost from the free surface carries solute parti-
cles towards the edge during evaporation process. Buoyancy driven
convection or Marangoni flow due to temperature gradient within
the evaporating droplets were not considered in their works.
⇑ Corresponding author. Tel./fax: +91 361 2582259.
E-mail addresses: a.thokchom@iitg.ernet.in (A.K. Thokchom), petro.abhishek@
gmail.com (A. Gupta), jaijus.pallippadanjohny@pg.canterbury.ac.nz (P.J. Jaijus),
anugrah@iitg.ernet.in (A. Singh).
droplets on a heated surface and observed strong influence of ther-
mo-physical and geometrical parameter on the droplet evapora-
tion. The difference in evaporation characteristics between pure
and binary liquid droplet was studied by Zhang et al. [12]. Popov
[13] and Fischer [14] have studied the particle deposition theoret-
ically by considering a very thin drop and derived the expressions
for height averaged radial velocity profile inside the drop. Widjaja
and Harris [15] performed numerical simulations to study the
deposition pattern in evaporating droplets containing solute parti-
cles. They found that the particle deposition rate on the substrate is
influenced by the convection and diffusion mass transfer of the
particles in the bulk liquid. They also did not consider the role of
Marangoni flow in evaporating droplets. Girard et al. [16] numeri-
cally analyzed the evaporating droplet on a heated substrate under
microgravity condition and investigated the role of heated sub-
strate on flow patterns. Their numerical simulations revealed that
the Marangoni convection is generated within the droplet due to
temperature gradients on free surface. Recently Risenpart et al.
[17] have observed that the non-uniform evaporation also pro-
duces the ‘coffee-ring’ effect.
Many experimental studies have shown that fluid flow within
the droplet plays an important role in deposition pattern. Hu and

0017-9310/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijheatmasstransfer.2013.09.012
68 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
Nomenclature
CCD charge couple device
IR infrared light
PIV Particle Image Velocimetry
LED light emitting diode
P pressure (N/m2)
Cp specific heat of the liquid (kJ/kg °C)
u velocity in x-direction (m/s)
v velocity in y-direction (m/s)
T temperature (K)
t0 dimensionless evaporation time
J mass flux (kg/s m2)
N dimensionless particles concentration
ho initial height of the droplet
g accelaration due to gravity (m s2)
h height of droplet at any time
Larson [18–21] using the lubrication analysis of fluid flow studied
the effect of Marangoni flow inside the evaporating droplets
through finite element simulations. They observed that the ring
like deposition pattern is not only due to the pinning of contact line
but Marangoni stress driven flow also has strong influence on par-
ticle deposition. Their experimental results indicate that the parti-
cle deposition in octane droplet can be localized predominantly at
the center instead at the edge of the contact line. They believe that
a non-uniformity of temperature of the surfaces leads to difference
in surface tension on the interface which induces thermal Marang-
oni flow. This makes the fluid near the edges rise towards the top
of the droplet and plunge downward to the center of the droplet.
They observed qualitative agreement of experimental and simula-
tion results for deposition pattern. The particle concentration pro-
file showed a large peak at the edges of the water droplet when
Marangoni flow was weak or absent. When the Marangoni flow
was strong, a large peak was observed at the center of octane drop-
let. The role of surfactant on Marangoni flow was also reported by
them. Zhang and Wang [22] also observed the Marangoni flow in
their study of natural convection in an evaporating droplet. Savino
et al. [23] and Savino and Monti [24] have experimentally shown
the existence of Marangoni flow in droplets of organic fluid but
not in case of water droplet. David et al. [25] investigated experi-
mentally the role of thermal conductivities of the substrate on
the evaporation of water droplet. Girard et al. [26] measured the
temperature profile inside a water droplet placed on a heated sub-
strate by infrared camera and observed that the temperature gra-
dient within the evaporating droplet was less than 1 °C. A small
temperature difference (as small as 0.5 °C) within the droplet
was enough to generate the Marangoni flow. Lu et al. [27] have
performed numerical investigation of the fluid flow due to temper-
ature distribution inside the evaporating droplet placed on a
heated surface. Kang et al. [28] have recently visualized the flow
pattern inside the sessile droplet of NaCl solution by PIV technique
and observed similar flow pattern at different concentrations. Most
of these studies on evaporating droplets are for the case of pinned
contact line condition. Adachi et al. [29] have observed the ‘stick
slip’ motion of the contact line when there is a completion between
the friction force and surface tension and this generates a stripped
film composed of particles after the droplet dries out.
Quantitative visualization of fluid flow is of great interest in sit-
uations involving evaporating droplets. It is of great importance to
have an accurate velocity field data inside the evaporating droplet.
Besides the fluid flow, the deposition pattern is also influenced by
the nature of the substrate surface. Xu et al. [30] have shown that
nanoparticles can be self-assembled by allowing a solution of
Greek symbol
ho initial contact angle of the droplet
h contact angle of the droplet at any time
l droplet liquid viscosity (kg/m s)
X interfacial tension
r local value of surface tension (N/m)
k thermal conductivity of the liquid (W/m K)
q density of droplet liquid (kg/m3)
b thermal expansion coefficient (K1)
a thermal diffusivity (m2/s)
Subscripts
o initial condition at t = 0
i condition at any time t
nanoparticles to evaporate on a sphere-on-flat geometry. This
self-assembly of particles on a substrate is of great importance
in many micro level applications. It is a great challenge to control
this phenomenon in small droplets. Hong et al. [31,32] were able
to get ordered polymers pattern of concentric ring and punch-like
shape by interacting the polymer and the substrate. Marin et al.
[33] observed the ordered particles arrangement at the early stage
while disordered arrangement were observed at the end of the
evaporation of droplet. Wong et al. [34] used this phenomenon
to separate the biological entities in a liquid droplet with an aim
to develop low cost technologies for disease diagnostics in re-
source poor environment. Majumder et al. [35] found that the
phenomenon of evaporating droplet can be used for depositing
catalyst nanoparticles to form a single walled carbon tubes as well
as to manufacture plasmonic films of well-spaced, un-aggregated
gold nanoparticles.
Despite their importance, there are very few works on the anal-
ysis of fluid flow and the particles transport inside the evaporating
droplets due to inherent problem encountered in imaging a small
droplet. Understanding the proper mapping of fluid flow inside
the evaporating droplet can help to manipulate the particles trans-
port, so that highly ordered structures in micro-devices can be pro-
duced in large scale. In one of the earlier studies, Yarin et al. [36]
have analyzed the velocity inside the levitated evaporating droplet.
Savino et al. [23] carried out experimental measurements and
numerical simulation to study the Marangoni and buoyancy effect
on velocity field inside the hanging evaporating droplet. They ob-
served that n-octane droplet exhibit Marangoni convection
whereas pure water droplet does not. Savino and Monti [24] solved
the steady state velocity field inside the evaporating droplet by
considering a Marangoni stress on air–liquid interface. Due to the
wrong velocity mapping during the measurement of the velocity
field their experimental results did not match with the theoretical
predictions. Without a proper image correction it is very difficult to
analyze the velocity field in 3D droplets. This problem was over-
come by Kang et al. [37] who developed correlations to map the
image in the velocity plane to that of image plane. Minor et al.
[38] reported that the ray tracing algorithm by Kang et al. is valid
only for hemispherical shape and needs correction for droplet hav-
ing the shape of a sphere. Recently, Saha et al. [39] used the ray
tracing method and improved algorithm of Minor et al. [38] to cor-
rect the optical distortion during the measurement of velocity vec-
tors in a levitated evaporating droplet. Yan et al. [40] have also
measured the velocity field inside the levitated drops by tracking
the displacement of tail or front end of streamlines in the recorded
digital images. Ueno and Kochiya [41] and Pereira et al. [42] have
 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
used 3D particle tracking method to analyze the capillary flow in-
side the evaporating droplet. To overcome the problem of distor-
tion of image due to refraction of light from the droplet surface
some authors have visualized the droplet from the bottom of a
transparent substrate. However, planar 2-component PIV tech-
nique which uses thin light sheet for visualization cannot be used
with bottom viewing because, in an axi-symmetric sessile droplets
the visualization of the flow field in a plane perpendicular to the
base is required. Besides, this restricts the study to transparent
substrates only. These methods also require two CCD cameras
[43,44]. Jaijus and Singh [45] have developed a simple method
where the correction of image mapping and velocity mapping are
not required. A confined 2D droplet analog to 3D droplet is gener-
ated in between two parallel plates with narrow space. This meth-
od is simple and easy to implement. No special optical
arrangements are required and the fluid flow can be visualized
with any light source. We have used this method to obtain the
velocity field inside the evaporating droplet subjected to Infrared
light heating. Numerical simulations were also performed by solv-
ing continuity, momentum and energy transport equations inside
the heated droplets. The simulations were performed on the same
geometry and the thermal boundary conditions that prevailed in
the experiments. The experimental detail is presented in Section 2
followed by numerical simulation method in Section 3. The results
from experiments and simulations are presented in Section 4, fol-
lowed by discussion in Section 5.
2. Experimental details
2.1. Experimental setup
This section describes the experimental set-up and procedures
for PIV measurement of the internal flow in a droplet. The sche-
matic of the experimental setup is shown in Fig. 1. A single droplet
of water containing spherical polystyrene particles of mean diame-
ter 10 lm was placed on a wetting surface (microscope glass slide)
confined between two teflon coated non-wetting transparent glass
plates. The droplet acquires the shape of a disc whose base re-
mained pinned to the substrate. The two side walls (confining par-
allel glass plates) which is non-wetting to the water allows free
movement of droplet surface due to evaporation. In our experi-
ments the thickness of the disc shaped droplets were measured to
be 800 lm and the diameters varied between 4–6 mm. A white
LED light source was used to illuminate the dispersed particles. A
150 W infrared lamp was used to heat the droplet from the top.
IR light is rapidly absorbed by the surface of the water raising the
surface temperature. Since IR heating is line-of-sight method, the
position of the IR lamp could be adjusted to apply symmetric or
Fig. 1. Experimental set up to visualize the fluid flow inside the evaporating droplet.
69
asymmetric heating of the droplet surface. In our experiments the
IR light was allowed to fall on the drop surface through a small
aperture. If the center of the aperture falls on the droplet axis then
the temperature on the top left surface was expected to be similar
(and symmetric) to that on the right side. On the other hand, if the
IR light was placed on the right side, then only the top right surface
of the drop will fall in the line-of-sight and therefore, the tempera-
ture there will be higher compared to the left side. A CCD camera
(Pixelfly Hires from PCO) having a resolution of 680  512 pixels
in conjunction with a macro zoom lens (Navitar) was used to cap-
ture the images of particles. The images were recorded with a frame
rate of 19 frames per second and stored on a computer via frame
grabber card for PIV analysis. The temperatures near the top and
two edges of the droplet were also measured using a thermocouple.
2.2. Experimental procedure
The particles used were white spherical polystyrene particles
having mean diameter of 10 lm. First, these particles were dis-
persed in water with gentle mixing in a test tube. The suspension
was kept for some time to allow any heavier particles to settle
down. The top portion of the suspension was transferred into an-
other test tube. This was used as the sample for the analysis. In
the present analysis we have used 1% (Volume by Volume) of dis-
persed particles. Using a syringe a small droplet was generated
touching the wetting substrate. The droplet took the shape of a disc
sandwiched between the two parallel non-wetting surfaces as
shown in Fig. 1. The two parallel plates confining the droplet were
held in vertical position and the CCD camera was placed on a hor-
izontal rail for capturing the images to be used for PIV analysis. PIV
is a method of determining velocity fields from images of particles
moving with the flow. It measures the whole velocity fields by tak-
ing two successive images and calculating the distance of individ-
ual particles traveled within the inter-framing time. From the
known time difference and the measured displacement, the veloc-
ity is calculated. Since it is not possible to track individual particles,
a statistical analysis is performed by dividing the image into grids
of small sections called as interrogation areas. The corresponding
interrogation areas within each of the two images are then cross-
correlated. The cross-correlation determines the requirement of
displacement to shift the first area to best overlap the second area.
The knowledge of this displacement and time between the two
images gives the velocity corresponding to the interrogation spot.
This process is repeated at each grid point within the image, result-
ing in a map of velocity vectors to describe the flow. The continu-
ous images taken from the CCD camera were stored in 680  512
pixels, 12 bit gray scale format on a PC via frame grabber. Fig. 2
shows a sample image of the particles in the droplet. Fig. 2(a)
shows a raw image of droplet taken from CCD camera. The
70 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77

Fig. 2. (a) Raw image of the droplet taken from the CCD camera, (b) Cleaned image for PIV analysis.

particles are represented as white spots in these images. The area ever, in this work we have simplified our simulations by carrying out
of interest is the region enclosed between free surface (top) and so- steady state solution of continuity, momentum and energy equations
lid surface (bottom). Due to light reflection from the top and bot- on a fixed grid since the objective of present simulation was to see if
tom surfaces the image shows thick white regions there. The the fluid flow profile resulting from buoyancy and Marangoni convec-
boundary of the area of interest was found by locating the free sur- tion can be predicted qualitatively from the numerical simulation.
face and solid surface and the cleaned image which was subjected Therefore our steady state simulations can be considered at best only
to PIV analysis is shown in Fig. 2(b). To analyze the velocity from qualitative. We would like to mention that in a confined disc shaped
images, we used two-frame cross correlation method [44] which drop attached to the substrate the only region through which evapo-
gives only a single displacement peak and hence is ideal for flows ration takes place is the thin circular strip in contact with the air.
that have recirculation patterns [45]. The best overlapping pixels Hence, the area of exposure for evaporation is very small and the mass
area was estimated by varying interrogation window sizes from transfer during the evaporation will be low. Therefore, the steady
64  64 to 128  128 pixels. state simulation (without considering mass transfer) is expected to
To evaluate the concentration of solute particles each image is give us the qualitative information about the velocity profile. The
divided into 10 bins of equal width as shown in Fig. 3. The number schematic of the computational domain is shown in Fig. 4. The system
of particles in each bin of the droplet is calculated by a particle is a 2-D sessile droplet of incompressible fluid (water) of constant vis-
tracking program. In the grey scale format of the image each pixel cosity and having the shape of a circular arc resting on a flat surface. A
is assigned an intensity whose value lies between 0 and 255; Cartesian co-ordinate system is used with local radial coordinate x
where 0 represents complete black pixel and 255 represents com- and axial co-ordinate y.
plete white pixel. The average particle size was measured to be 2 The steady state governing equations for mass, momentum and
pixels. Therefore any two adjacent pixels whose intensity value lies energy transport inside the droplet are given below.
in between 240 and 255 were counted as one particle. All the par- Continuity equation:
ticles in the bins were counted. In this procedure a small error in
@ðquÞ @ðqv Þ
particle count is expected due to overlapping particles but it is þ ¼0 ð1Þ
not expected to alter the particle count qualitatively. In a confined @x @y
disc shaped drop attached to the bottom substrate, the only region Momentum equations:
through which evaporation takes place is the thin circular strip in !
 
contact with the air. If the thickness of the disc is increased the @u @u @P @2u @2u
area of the strip and hence the rate of evaporation will also in- q u þv ¼ þl þ ð2Þ
@x @y @x @x2 @y2
crease. However, the thickness of the disc should be small enough
so that there are no out of plane motion of the dispersed particles.   !
If the thickness of the disc is large then there can be motion of par- @v @v @P @2v @2v
q u þv ¼ þl þ þ qbðT  T 0 Þg y ð3Þ
ticles perpendicular to the image plane (if there is temperature dif- @x @y @y @x2 @y2
ference between the two confining glass plates).
Energy equation:
  !
3. Numerical simulations @T @T @2T @2T
qC p u þ v ¼k þ ð4Þ
@x @y @x2 @y2
In an evaporating droplet with pinned contact line the droplet con-
tinuously looses mass, as a result the contact angle and height of the In the above equations, u and v are the x and y component of veloc-
droplet also decreases with time. The transient numerical simulation ity of fluid having density q and viscosity l. Cp and k are the specific
would require solution of momentum, energy and mass transport
equations. Such simulations can be performed on moving grids. How-

Fig. 3. Schematic representation of bins in droplets to calculate particle Fig. 4. Computational domain of the droplet placed on the solid surface in Cartesian
concentration. coordinate system.
 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
heat and thermal conductivity of the water respectively, and in y-
momentum equation, b denotes the thermal expansion coefficient
of water. The boundary conditions considered in the simulation
are given below:
1. The temperature of the solid as well as free surface of the drop-
let are specified in the form of linear profile obtained from the
experimental values of temperatures at the apex as well as left
and right edges.
2. The no-slip boundary condition is applied on the bottom solid
surface.
3. At the liquid–gas interface, a tangential Marangoni stress due to
surface temperature variation is considered as follows:
    
dr dr dT
¼ :
dx dT dx
A constant value of temperature coefficient of surface tension or/
oT = 0.0001657 N m1 °C1 (as reported by Hu and Larson [19])
is taken at air–water interface to include the effect of Marangoni
stress. The equations of continuity, momentum and energy were
solved by ANSYS Fluent software. The simulation was carried out
assuming the quasi-steady state which neglected the decreasing
height of the droplet due to evaporation. Since the particle concen-
tration is very low, we neglected the effect of particle on the fluid
flow inside the droplet and properties of the solvent. In the steady
state analysis, second order upwind scheme for momentum and en-
ergy equation was used and SIMPLE (Semi-Implicit Method for
Pressure Linked Equations) algorithm was used for the pressure–
velocity coupling. The temperature was assumed to vary linearly
over the free surface and the bottom surface. To confirm the grid
independent solution we have conducted simulations on two differ-
ent grids having 42076 (grid A) and 74392 (grid B) quadrilateral
cells. The results from the two grids were found to be similar con-
firming the grid independent solution. In the following sections
the numerical results presented are from simulations performed
on grid B. The results obtained from simulations were analyzed to
understand the velocity filed and recirculation pattern.
4. Results
4.1. Analysis of height and contact angle of the droplet
As mentioned in previous sections, except for the rate of evap-
oration the experiments with disc shaped 2D drops have similar
evaporation dynamic as the 3D droplets [45,46]. In our experi-
ments the droplet rests on a wetting and rough substrate where
the contact line gets pinned. Since the droplet was confined be-
tween two parallel plates which were non wetting to water, the
top surface was free to move. As the evaporation progresses there
is fluid volume loss in the droplet which makes the contact angle
Fig. 5. Time evolution of (a) contact angle of the droplet, and (b) height of the droplet.
71
and height of the drop decrease continuously. The images of the
droplets were recorded using a CCD camera which was placed on
a horizontal rail. From these images we determined the contact an-
gle and height of the droplet. Fig. 5(a) shows the plot of contact an-
gle with time for the three cases of infrared heating of the droplet
free surface. In this plot the contact h is normalized by initial con-
tact angle ho and evaporation time t is normalized by the total dry-
ing time td. The initial angles measured for these three cases: top,
left and right heating, were 79°, 69° and 61° respectively. It can be
observed that the contact angle decreases almost linearly with
time. Similarly, Fig. 5(b) shows that the height of the droplet (nor-
malized by the initial height) also decreases linearly with time. The
initial height measured for the three cases, top, left and right heat-
ing were 2.27 mm, 1.95 mm and 1.39 mm respectively. The linear
decrease in contact angle and height of the drop with time is also
ð5Þ
reported for 3D droplets [41]. This also shows that the lateral non-
wetting surfaces are not changing the droplet dynamics
significantly.
4.2. Velocity and temperature profiles
We first report the results for the case of symmetric heating of
the droplet when the IR light was placed on the top. First, a small
drop of water containing polystyrene particles was released near
the bottom wetting substrate from a syringe through the gap be-
tween the two parallel glass plates. As the droplet touched the
wetting surface it spread in the form of a half-disc between the
two non-wetting surfaces. The IR lamp was switched on, and there-
after, the CCD camera started grabbing the images. It was observed
that within few second (3–4 s) of turning on the IR lamp the fluid
motion started. The motion appeared to be slow and streamlined
and we did not observe any churning of the flow with IR heating
since the temperature rise was small. A part of the IR light falling
over the droplet surface is absorbed by the liquid raising its tem-
perature. The particles dispersed in the droplet can also increase
the absorption coefficient, but in the present case of very dilute
system the influence of particles on absorption is not expected to
be large. To initiate the motion due to Marangoni stress, heating
of complete droplet is not necessary and it is the gradient in sur-
face temperature of the droplet which initiates the motion. Since
the IR light is rapidly absorbed by the surface of the drop raising
its temperature, the Marangoni convection is initiated instantly.
The flow profile in the case of symmetric heating with IR light
was observed to be qualitatively similar to that of velocity profile
of evaporating drop under ambient conditions (without IR heat-
ing). However the velocity magnitude in the later case was much
smaller. To establish the strength of the Marangoni convection,
the measurement of droplet surface temperature profile is needed.
IR thermography has been used by Lu et al. [27] and Saha et al. [39]
to measure the surface temperature of 3D evaporating droplets.
72 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
Since the drop size in our experiments was quite small, a very high
resolution IR camera was required to measure the surface temper-
ature profile. Moreover, the droplet was sandwiched between two
glass plates; therefore the IR camera (if placed normal to the drop-
let plane) could have measured the temperature of the top surface
of the glass plate and not the actual temperature inside the liquid.
It was possible to get free surface temperature profile by taking the
IR images from the top view (above the gap between the two glass
plates) but curvature of the drop can make the measurement erro-
neous unless a high resolution camera with proper image correc-
tion is used. Due to unavailability of high resolution IR camera in
our lab we have measured the temperatures at the apex and two
edges of the droplet using thermocouples and the profile was esti-
mated by linear fit. We expect this to be good approximation since
the temperature difference was not large. Fig. 6 shows the experi-
mental velocity vectors superimposed on the velocity contours at
four different time instants of evaporation. The contact angle and
height of the droplet shown in Fig. 6(a) was 69.29° and 2.2 mm
respectively. At this condition the temperature of the apex was
measured to be 50.25 °C and those of left and right corners were
49.62 °C and 49.60 °C respectively. Although all the precautions
were taken to place the infrared lamp exactly above the middle
surface of the droplet but still the temperature difference between
the apex and two edges are found to be slightly different. Since the
temperature profile was nearly symmetric we observe two counter
rotating convection cells. As the evaporation progressed the height
and contact angle of the droplet was found to decrease. Fig. 6(b)–
(d) shows the velocity vector and velocity contour when the height
of the droplet reduced to 2 mm, 1.7 mm and 1 mm respectively.
The corresponding values of contact angle were measured to be
62.54°, 35.59° and 19.23° respectively. It can be observed that as
the height of the droplet decreases the difference in temperature
of apex and edges decreases. But, in all the cases the temperature
measured at the edges was lower than the apex of the droplet. This
temperature difference between the edge and apex region will cre-
ate a surface tension gradient on the surface of the droplet. The
surface tension will be more at the edge than the apex of the drop-
let. The gradient in surface tension will set up Marangoni flow from
the apex towards the edge along the surface of the droplet. This
gives rise to two counter rotating vortices. Due to smallness of
the droplet the effect of buoyancy driven convection is expected
to be small. Since the temperature of the fluid near bottom sub-
strate was lower than the top surface even the buoyancy driven
Fig. 6. Experimental velocity vectors superimposed over velocity contour colored by the magnitude for the case of symmetric heating by IR lamp on the top of the droplet at
four different time instants after the start of evaporation. The heights of the droplets were (a) 2.2 mm (b) 2 mm (c) 1.7 mm and (d) 1 mm. The contact angles were 69.29°,
62.54°, 35.59° and 19.23° respectively. The aspect ratio of these images has been changed for the clarity of the velocity vectors.
convection will cause the fluid to move in clockwise direction in
the right half and counter-clockwise in the left half. Thus two sym-
metric convection cells were formed inside the droplet and the
particles follow the fluid motion. This process causes some of the
particles to get deposited near the edge and become concentrated
with time. We would like to mention that, when the contact angle
decreased to low values, there were few dispersed particles (since
most of the particles are already deposited) making the finding of
correlation peaks erroneous and hence it was not possible to get
enough velocity vectors at the current concentration of particles.
In our experiments we could not get correct velocity vectors for
droplet contact angle below 17°. Hu and Larson [20] have obtained
analytical and numerical solution of Marangoni flow field in an
evaporating sessile droplet. They observed that thermal field in
the droplet can change significantly with evaporation time and it
can also reverse the flow direction. They also observed that at long-
er times the faster rate of evaporation at the droplet edge makes it
cooler there and there is reversal of recirculation direction below
the contact angle of 14°. Unfortunately we could not verify this
in our experiments.
The numerical simulations were carried out for steady state
conditions using the boundary temperature profile obtained from
the experiments. The top left boundary was assigned linearly vary-
ing temperature profile between the apex temperature and left
edge temperature, whereas the top right surface was assigned lin-
early varying temperature between the apex temperature and
right edge temperature. The bottom surface was assigned temper-
ature profile which varied linearly between the left edge and right
edge temperatures. No-slip condition was assigned at the bottom
surface, whereas Marangoni stress condition as discussed in Eq.
(5) was assigned to the top surface. Fig. 7(a) and (b) shows the
velocity vectors superimposed over the velocity contours obtained
from the simulations for the same geometry of the droplet that was
reported in Fig. 6(a) and (c) respectively. The corresponding tem-
perature contours obtained from simulations are plotted in
Fig. 7(c) and (d) respectively. As was the case in the experiments
the simulations also show two counter-rotating cells. The direc-
tions of these cells are same as was reported in the experiments.
However, we observed that the magnitude of velocity in simula-
tions was considerably higher compared to the experiments. The
main reason for this quantitative difference is that in the steady
state simulations we have not considered momentum transport
resulting from evaporation flux since our objective was to see the
 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77 73

Fig. 7. Velocity vector and contour map obtained from numerical simulation for symmetric heating: (a) h = 2.2 mm, h = 69.29° (b) h = 1.7 mm, h = 35.59°. The figures (c) and
(d) show the corresponding temperature contour inside the droplet obtained from numerical simulation. The aspect ratio of these images has been changed for the clarity of
the vectors and contours.

Fig. 8. Experimental velocity vectors superimposed over velocity contour colored by the magnitude at four different time instants for the case of asymmetric heating by IR
lamp on the left of the droplet. The height of the droplet was (a) 1.8 mm (b) 1.4 mm (c) 1 mm and (d) 0.9 mm. The corresponding contact angles were 57.48°, 34.35°, 18.35°
and 17.28° respectively. The aspect ratio of these images has been changed for the clarity of the velocity vectors.

qualitative nature of flow inside the droplet. Nonetheless, we ob-
serve good qualitative agreement between the experimental and
simulation results.
In the next experiment we placed the lamp on the left side of
the droplet. This gives higher temperature on the left compared
to the right. Fig. 8 shows the experimentally measured velocity
vectors superimposed over the velocity contours colored by the
velocity magnitude at four different time instants. The heights of
the droplet in these cases were 1.8 mm, 1.4 mm, 1 mm and
0.9 mm. The corresponding values of contact angles were 57.48°,
34.35°, 18.35° and 17.28° respectively. We would like to mention
that the aspect ratio of the plots in these images has been properly
adjusted for the clarity of the velocity vectors. The temperatures
measured at three places near the air–water interface are also
shown in the images. The temperature of the left edge was mea-
sured to be higher than that of apex and right edge. In all the four
cases we observe that the temperature decreases from the left edge
to the right edge and the apex temperature falls between these two
values. The qualitative behavior of temperature profile remains the
same at all the times for which we have shown the profiles. This
temperature gradient will result higher surface tension at the right
edge than the left edge. The gradient in surface tension will set up
Marangoni flow from the left edge towards the right edge along the
surface of the droplet. The buoyancy driven convection is also ex-
pected to cause the fluid flow in clock-wise direction. Unlike the
previous case we observed only one convection cell. The velocity
and temperature profiles obtained from steady state simulations
are shown in Fig. 9. The velocity profiles for the two different
heights of the droplets (1.8 mm and 0.9 mm) are shown in
Fig. 9(a) and (b) which also confirm a single convection cell with
clockwise rotation. The corresponding temperature profiles are
shown in the Fig. 9(c) and (d) respectively. Both velocity and tem-
perature profiles are observed to be asymmetric. Higher tempera-
ture is observed in the left half compared to the right half of the
droplet. Again we observe only a qualitative agreement with the
experiments and the velocity magnitude from the simulations
are found to be much higher for the reasons mentioned previously.
Finally we have conducted experiments in which the IR lamp
was placed on the right side of the droplet. This arrangement re-
sulted into droplet temperature profile which showed higher tem-
perature near the right compared to the left. The experimentally
measured velocity vectors and velocity contours are shown in
Fig. 10 for four different time instants corresponding to the droplet
height of 1.4 mm, 1.2 mm, 1 mm and 0.8 mm. The contact angles
74 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77

Fig. 9. Velocity vector and contour map obtained from numerical simulation for asymmetric heating of the drop by placing the IR light on the left side: (a) h = 1.8 mm,
h = 57.48° (b) h = 0.9 mm, h = 17.28°. The figures (c) and (d) show the corresponding simulation temperature contour inside the droplet. The aspect ratio of these images has
been changed for the clarity of the vectors and contours.

Fig. 10. Experimental velocity vectors superimposed over velocity contour colored by the magnitude at four different time instants for the case of asymmetric heating by
keeping IR lamp on the right. The height of the droplets were (a) 1.4 mm (b) h = 1.2 mm (c) 1 mm and (d) h = 0.8 mm. The contact angles were 60.21°, 36.25°, 25.67° and
18.34° respectively. The aspect ratio of these images has been changed for the clarity of the velocity vectors.

were 60.21°, 36.25°, 25.67° and 18.34° respectively. In all the four
cases the highest temperature was found near the right edge which
is closer to the IR source. The left edge had the lowest temperature
while the temperature of the apex was in between left and right
values. Therefore, an asymmetric temperature gradient along the
surface of the droplet existed. This asymmetric temperature gradi-
ent will result into a surface tension gradient where the left side
will be at higher surface tension compared to the right side. This
will cause the liquid to be pulled from the right side to the left side
along the free surface. As expected we observe only one convection
cell in counter-clockwise direction. The center of this cell is also
shifted towards the right side. Fig. 11(a) shows the velocity profile
from the simulations corresponding to the geometry of Fig. 10(a),
whereas Fig. 11(b) shows the velocity profile for the droplet in
Fig. 10(c). The corresponding temperature profiles from the simu-
lations are shown in Fig. 11(c) and (d) respectively. The right half of
the droplet is at higher temperature compared to the left half. The
velocity profiles are observed to be in qualitative agreement. It is
expected that unsteady simulations in which the computations
are performed on a moving grid with proper evaporation flux at
the surface of the droplet would perhaps provide quantitative
agreement as well.
4.3. Concentration profile and pattern formation
Though, there have been some studies on velocity profile in
evaporating droplets, very few have reported evolution of concen-
tration profiles. In this section we present the dynamics of particle
concentration for three different cases of evaporation for which we
reported the velocity profiles in the earlier section. The images were
divided into 10 bins of equal width and number of particles in each
bin was counted. It was observed that initially the particle distribu-
tion was uniform. With the progress of evaporation the resulting
convection transported the particles from the bulk towards the
edges. Fig. 12 shows the number of particles in a bin at a given time
(N) scaled by the number of particles initially present in that bin
(N0) for all the ten bins at various times. The time shown in the plots
are normalized by the total evaporation time for drying of a droplet.
The dynamics of particle deposition for the case of symmetric heat-
ing of the droplet is shown in Fig. 12(a). It can be observed that the
number of particles in the central bins (3–8) decreases with time,
whereas the numbers of particles in the bins near the left edge (1
and 2) and that of right edge (9 and 10) increases with time.
Fig. 12(b) shows the number of particles in various bins for the case
of left side heating of the droplet. In this case, the corresponding
 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
Fig. 11. Velocity vector and contour map obtained from numerical simulation for asymmetric heating of the drop by placing the IR light on the right side: (a) h = 1.4 mm,
h = 60.21° (b) h = 1 mm, h = 25.67°. The figures (c) and (d) show the corresponding simulation temperature contour inside the droplet. The aspect ratio of these images has
been changed for the clarity of the vectors and contours.
velocity profile which was shown in Fig. 7 predicted concentration
of particles towards the right edge. As expected the number of par-
ticles in the bins near the right edge (8, 9 and 10) was found to in-
crease with time, whereas other bins (1–7) showed decrease in
particle count with time. The particle dynamics for the case of right
side heating shows opposite behavior which is clear from Fig. 12(c).
As the evaporation progresses the particles get concentrated in the
left bins (1–3) whereas in all other bins there is decrease of particle
count with time.
The observation of the particle concentration dynamics is fur-
ther clarified by careful observation of the droplet images towards
the end of evaporation process. Fig. 13(a) shows the image of the
droplet and particles after 10 min of symmetric heating from the
top. The bulk of the region is devoid of particles, whereas two
white patches can be clearly seen near the left and right edges.
The corresponding images for the case of left and right heating
are shown in Fig. 13(b) and (c) respectively. The images in
Fig. 13(b) and (c) correspond to 18 and 9 min of evaporation time
respectively. These images confirm that when the fluid flow is in
clockwise direction (for the case of left heating), the particles are
deposited towards the right edge. An opposite behavior is observed
when the right heating produced convection cells moving in coun-
ter-clockwise direction.
5. Discussion
The applications of particle deposition by droplet drying are
numerous. In some cases it is desired to have uniform films of par-
ticles where the ‘coffee-ring’ effect is undesirable (as it brings the
particles to the contact line by the convective macroscopic flow).
In several other applications this phenomena is exploited for depo-
sition of particles in a well spaced patterns. Many previous studies
on ‘coffee-ring’ effect did not consider role of recirculation internal
flow in deposition pattern. Recent flow visualization studies have
reported that droplet internal flow shows strong recirculation
resulting from Marangoni convection. Most of the research work
on this topic has focused on droplet evaporation and particle depo-
sition under ambient conditions and very little attention has been
given to the control of the internal flow. It is well known that prop-
erly controlled and engineered Marangoni flow can be useful in
enhancing heat transfer, inkjet printing and nanoparticle deposi-
tion [35]. The manipulation of the internal flow during drying
can allow to control and redirect evaporation driven deposition
and assembly of small particles. Hu and Larson [21] reported that
75
Marangoni flow is not observed in water-based suspensions be-
cause of the combination of lower volatility, higher heat capacity
(compared to organic liquids) and because of the presence of sur-
face active agents on the surface. Since such suspensions are used
in many applications, therefore it is highly desired to devise a sim-
ple method for inducing Marangoni flow in aqueous droplets. Gir-
ard et al. [26] have shown that the Marangoni flow contribution to
the evaporation rate of small heated water droplets resting on hot
substrate is negligible. Therefore, it is even more useful to control
the strength and direction of recirculation flow. Marangoni flow
can be generated by the presence of surface active agents on the
droplet surface that during drying develops a concentration gradi-
ent along the free surface leading to a gradient in surface tension. It
can also be generated by the temperature gradient on the surface
since surface tension is a decreasing function of temperature.
Non uniform evaporation at the free surface can also generate tem-
perature gradient and hence Marangoni flow, but they are promi-
nent only in high volatile solvents such as octane, ethanol or
pentane and difficult to generate in aqueous suspensions. In this
paper we have shown that Marangoni flow can be induced easily
in evaporating water droplets by exposing them to thermal radia-
tion. The Marangoni recirculation arising from thermal surface ten-
sion gradient during drying allows significant flow of particles
from the bulk to the pinned edges leading to particle deposition.
To avoid the complexity involved with image and velocity cor-
rection in PIV study of 3D droplets we have considered a simple
case of 2D droplet. It is true that 2D problem is not a complete ana-
logue to that of the 3D droplet but in those cases where the droplet
is symmetric (which is true for smaller drops) and the internal flow
is not chaotic, the evaporation dynamics is qualitative similar. If we
imagine a plane (perpendicular to the substrate) that cuts through
the 3D drop with azimuthal symmetry, the velocity profile on this
plane will be qualitatively similar to our 2D system. Jaijus and
Singh [45] have compared the time evolution of contact angle for
2D disc shaped droplet with that of 3D droplet experiments and
observed that the evaporation dynamics is qualitative similar ex-
cept that rate of evaporation is much slower.
The system considered under present experimental investiga-
tion is transient and resembles the 3D evaporating droplet, though
the rate of evaporation in confined system is much less since the
evaporation takes place only from the top surface (thin circular
strip). Our experiments were conducted in an environment where
Marangoni flow dominated over other convections. During the
droplet evaporation buoyancy, viscous and surface tension forces
play important role. The relative contribution between natural
76 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
Fig. 12. Time evolution of the particle count N (scaled by the initial count N0) in
various bins of the droplet subjected to: (a) IR Light on the top (b) IR light on the left
of the droplet, and (c) IR light on the right of the droplet. The time shown in the
plots is normalized by the total evaporation time.
convection and Marangoni convection is given by the Bond number
(Bo) which is the ratio of buoyancy force to surface tension force.
dr
Bo ¼ bg qL2 =
dT
In the above equation, b is the temperature coefficient of thermal
expansion of water, g is gravitational acceleration, q is water den-
sity, r is surface tension and T is temperature. L is the characteristic
length scale which we have taken as diameter of the droplet. When
Bo  1, internal flow is dominated by the Marangoni convection.
The non-dimensional number used to characterize the strength
of natural convection is the Grahshoff number (Gr) which is the ra-
tio of buoyancy force to viscous force and is given by,
bgL2 DT
Gr ¼
m2
Fig. 13. Images of the droplet showing deposition of the particles near the edges.
(a) Symmetric heating from the top, (b) asymmetric heating on the left and (c)
asymmetric heating on the right side of the droplet.
In the above equation DT is the temperature difference between the
droplet and ambient and t is the kinematic viscosity of water. The
natural convection can be ignored when Gr < 2400 [27]. For a pure
liquid droplet, a thermal Marangoni number (Ma), defined as ratio
of surface tension gradient induced by the temperature gradient
to viscous force is utilized to describe the intensity of Marangoni
convection.
dr
dT
DTL
Ma ¼ ð8Þ
la
In the above equation, l is the dynamic viscosity of liquid and a is
the thermal diffusivity. Marangoni convection can be neglected
when Ma is less than 100.
The value of Bond numbers (Bo) in our experiments with top,
left and right heating cases were 0.01670, 0.02664 and 0.01914
respectively. The corresponding values of Grahshoff number (Gr)
were computed to be 39.3, 44.5 and 11.9 respectively. Finally the
Marangoni numbers (Ma) for the three cases were 14400, 18200
and 15400 respectively. These values clearly indicate that all our
experiments were strongly dominated by Marangoni convection.
This Marangoni convection causes the suspended particles to move
along with the fluid path lines. In our PIV experiments the sus-
pended particles also act as tracer particles whose images are ana-
lyzed for cross correlation. For high accuracy of PIV data the
particle response time should be faster than the time scale of the
mean flow. The ratio of the characteristic time of a particle to a
characteristic time of the flow is given by the Stokes number (St).
qp d2 U
St ¼ ð9Þ
18lL
In the above equation qp is the density of particle, d is particle
ð6Þ
diameter, U is characteristic velocity of flow, l is dynamic viscosity
of fluid and L is the length scale of the droplet. For St  1, particle fol-
lows fluid streamlines closely without affecting the fluid velocity.
The maximum fluid velocity observed in our experiments was
15 mm/s. For 10 lm diameter polystyrene particles (den-
sity  1.05 gm/cc) dispersed in 5 mm diameter water droplet (den-
sity  1 gm/cc, viscosity  1 cp) the Stokes number (St) is O(103).
Therefore we do not expect two-way coupling and the particles
are expected to simply move with the fluid flow. We would also like
to mention that the particle density (1.05 gm/cc) was slightly higher
than the fluid density (1 gm/cc) which will induce small settling
ð7Þ
velocity. However, the size of the particle was much smaller
 A.K. Thokchom et al. / International Journal of Heat and Mass Transfer 68 (2014) 67–77
compared to the drop size, and hence the settling velocity was much
smaller. The typical settling velocity in our experiments was com-
puted to be about 1 lm/s. In the case of Marangoni flow (due to recir-
culation) the time scale of settling will be substantially longer.
Therefore, it is the direction and strength of flow near the substrate
which influences the particle deposition. While the suspended par-
ticles visually confirm recirculation in the drop, the internal motion
of the fluid in the drop has a substantial effect in shaping the unifor-
mity or non-uniformity of deposition. The particles will get depos-
ited to those places where the velocity vector is directed towards
the substrate. These facts are quite obvious from our experiments.
The objective of numerical simulations was to see if the fluid flow
profile resulting from buoyancy and Marangoni convection can be pre-
dicted qualitatively from the numerical simulation. The steady state
simulations reported in this work have considered only momentum
and energy equation on fixed grid. Therefore the results were only
qualitative. In order to get quantitative measurement of fluid velocity
during evaporation the numerical simulation should also have been
performed for transient case where the drop height changes continu-
ously with time due to evaporation. Such simulations can be per-
formed on moving grids and this is the topic of our future work.
6. Conclusion
The fluid flow and particle concentration dynamics inside the
evaporating water droplet containing polystyrene particles which
is subjected to symmetric and asymmetric heating from the top
using IR light was studied via experiments and numerical simula-
tions. To avoid the lens effect encountered in imaging the 3D drop-
lets we have conducted the experiment on a disc shaped droplet.
Since the object plane was parallel to the image plane no correc-
tion in velocity mapping in PIV analysis is required. The tempera-
tures at the apex and edges were measured in the experiments
which were used to assign the temperature boundary condition
in numerical simulations. The numerical simulations were carried
out for steady state condition for the same geometry which was ta-
ken in the experiments. The convection pattern resulting from
Marangoni stress depends on the temperature distribution on the
surface of the droplet. While two counter-rotating but symmetric
vortices were observed in case of top heating; the asymmetric
heating produced only one. The velocity profiles computed from
simulations were found to be in qualitative agreement with the
experimental observations. The direction of fluid motion strongly
affects the particle deposition towards the edges. The dynamics
of particle concentration supported these observations.
References
[1] D. Vincent, B. Jérôme, S. Eliane, Droplet evaporation study applied to DNA chip
manufacturing, Langmuir 21 (2005) 9130–9136.
[2] F. Xiaohua, L. Bingquan, P. Eric, S.S. Young, A. Vladimir Samuilov, C. Yong, C.S.
Jonathan, Y.S. Chwen, H.M. Rafailovich, Drying of DNA droplets, Langmuir 22
(2006) 6308–6312.
[3] J. Jing, J. Reed, J. Huang, X. Hu, V. Clarke, J. Edington, D. Housman, T.S.
Anantharaman, E.J. Huff, B. Mishra, B. Porter, A. Shenker, E. Wolfson, C. Hiort, R.
Kantor, C. Aston, D.C. Schwartz, Automated High Resolution Optical Mapping
using Arrayed Fluid-Fixed DNA Molecules, Proc. Natl. Acad. Sci. USA 95 (1980)
8046–8051.
[4] D. Brutin, B. Sobac, B. Loquet, J. Sampol, Pattern formation in drying drops of
blood, J. Fluid Mech. 667 (2011) 85–95.
[5] N. Chakrapani, B. Wei, A. Carrillo, P.M. Ajayan, R.S. Kane, Capillarity-driven
assembly of two-dimensional cellular carbon nanotube foams, Proc. Natl. Acad.
Sci. USA 101 (2004) 4009–4012.
[6] Y.C. Hu, Q. Zhou, C.C. Dong, L.S. Cui, Micro-flow induced peculiar surface profile
of film from dried droplet of water-poly (ethylene oxide) solution, Key Mat.
Eng. 531–532 (2013) 358–361.
[7] D. Zhang, Y. Xie, M.F. Mrozek, C. Ortiz, V.J. Davisson, B.D. Amotz, Raman
detection of proteomic analytes, Anal. Chem. 75 (2003) 5703–5709.
[8] R. Deegan, O. Bakajin, T.F. Dupont, G. Huber, S.R. Nagel, T.A. Witten, Capillary
flow as the cause of ring stains from dried liquid drops, Nature 389 (1997)
827–829.
View publication stats
77
[9] R. Deegan, Pattern formation in drying drops, Phys. Rev. E 61 (2000) 475–485.
[10] R. Deegan, O. Bakajin, T.F. Dupont, G. Huber, S.R. Nagel, T.A. Witten, Contact
line deposit in an evaporating drop, Phys. Rev. E 62 (2000) 756–765.
[11] V.E. Nakoryakov, S.Ya. Misyura, S.L. Elistratov, The behavior of water droplets
on the heated surface, Int. J. Heat Mass Transfer 55 (2012) 6609–6617.
[12] N. Zhang, B.X. Wang, Y. Xu, Thermal instability of evaporating drops on a flat plate
and its effects on evaporation rate, Int. J. Heat Mass Transfer 30 (3) (1987) 469–478.
[13] Y.O. Popov, Evaporative deposition patterns, spatial dimension of the deposit,
Phys. Rev. E 71 (2005) 036313-1–036313-17.
[14] B.J. Fischer, Particle convection in an evaporating colloidal droplet, Langmuir
18 (2002) 60–67.
[15] E. Widjaja, M.T. Harris, Particle deposition study during sessile drop
evaporation, AIChE J. 54 (2008) 2250–2260.
[16] F. Girard, M. Antoni, S. Faure, A. Steinchen, Evaporation and Marangoni driven
convection in small heated water droplets, Langmuir 22 (2006) 11085–11091.
[17] W.D. Ristenpart, P.G. Kim, C. Domingues, J. Wan, H.A. Stone, Influence of
substrate conductivity on circulation reversal in evaporating drops, Phys. Rev.
Lett. 99 (2007) 234502.
[18] H. Hu, R.G. Larson, Evaporation of a sessile droplet on a substrate, J. Phys.
Chem. B 106 (2002) 1334–1344.
[19] H. Hu, R.G. Larson, Analysis of the microfluid flow in an evaporating sessile
droplet, Langmuir 21 (2005) 3963–3971.
[20] H. Hu, R.G. Larson, Analysis of the effects of Marangoni stresses on the micro
flow in an evaporating sessile droplet, Langmuir 21 (2005) 3972–3980.
[21] H. Hu, R.G. Larson, Marangoni effect reverses coffee-ring depositions, J. Phys.
Chem. B 110 (2006) 7090–7094.
[22] N. Zhang, W.J. Wang, Natural convection in evaporating minute drops, J. Heat
Transfer 62 (1982) 756–765.
[23] R. Savino, D. Paterna, N. Favaloro, Buoyancy and Marangoni effects in an
evaporating drop, J. Thermophys. Heat Transfer 4 (2002) 562–574.
[24] R. Savino, R. Monti, Buoyancy and surface-tension driven convection in
hanging drop protein crystallizer, J. Cryst. Growth 165 (1996) 308–318.
[25] S. David, K. Sefiane, L. Tadrist, Experimental investigation of the effect of
thermal properties of the substrate in the wetting and evaporation of sessile
drops, Colloids Surf. A: Physicochem. Eng. Aspects 298 (2007) 108–114.
[26] F. Girard, M. Antoni, K. Sefiane, On the effect of Marangoni flow on evaporation
rates of water drops, Lagmuir 24 (2008) 9207–9210.
[27] G. Lu, Y.Y. Duan, X.D. Wang, D. Lee, Internal flow in evaporating droplet on
heated solid surface, Int. J. Heat Mass Transfer 54 (2011) 4437–4447.
[28] K.H. Kang, H.C. Lim, H.W. Lee, S.J. Lee, Evaporation-induced saline Rayleigh
convection inside a colloidal droplet, Phys. Fluid 25 (2013) 042001.
[29] E. Adachi, A.S. Dimitro, K. Nagayama, Stripe patterns formed on a glass surface
during droplet evaporation, Langmuir 11 (1995) 1057–1060.
[30] J. Xu, J. Xia, Z. Lin, Evaporation-induced self-assembly of nanoparticles from
sphere-on-flat geometry, Angew. Chem. Int. Ed. 46 (2007) 1860–1863.
[31] S.W. Hong, J. Xia, Z. Lin, Spontaneous formation of mesoscale polymer patterns
in an evaporating bound solution, Adv. Mater. 19 (2007) 1413–1417.
[32] S.W. ong, M. Byun, Z. Lin, Robust self-assembly of highly ordered complex
structures by controlled evaporation of confined microfluidics, Angew. Chem.
Int. Ed. 48 (2009) 512–516.
[33] A.G. Marin, H. Gelderblom, D. Lohse, J.H. Snoeijer, Order-to-disorder transition
in ring-shaped colloidal stains, Phys. Rev. Lett. 107 (2011). 085502(1–4).
[34] T.S. Wong, T.H. Chen, X. Shen, C.M. Ho, Nanochromatography driven by the
coffee ring effect, Anal. Chem. 83 (2011) 1871–1873.
[35] M. Majumder, C.S. Rendall, J.A. Eukel, J.Y.L. Wang, N. Behabtu, C.L. Pint, T.Y. Liu,
A.W. Orbaek, F. Mirri, J. Nam, A.R. Barron, R.H. Hauge, H.K. Schmidt, M.
Pasquali, Overcoming the ‘‘coffee-stain’’ effect by compositional Marangoni-
flow-assisted drop-drying, J. Phys. Chem. B 116 (2012) 6536–6542.
[36] A.L. Yarin, B. Brenn, O. Kastner, D. Rensink, C. Tropea, Evaporation of
acoustically levitated droplets, J. Fluid Mech. 399 (1999) 151–204.
[37] K.H. Kang, S.J. Lee, C.M. Lee, S. Kang, Quantitative visualization of flow inside
an evaporating droplet using the ray tracing method, Meas. Sci. Technol. 15
(2004) 1104–1112.
[38] G. Minor, P. Oshkai, N. Djilali, Optical distortion correction for liquid droplet
visualization using the ray tracing method: further considerations, Meas. Sci.
Technol. 18 (2007) 23–28.
[39] A. Saha, S. Basu, C. Suryanarayana, R. Kumar, Experimental analysis of thermo-
physical processes in acoustically levitated heated droplets, Int. J. Heat Mass
Transfer 53 (2010) 5663–5674.
[40] Z.L. Yan, W.J. Xie, B. Wei, Vortex flow in acoustically levitated drops, Phys. Lett.
A 375 (2011) 3306–3309.
[41] I. Ueno, K. Kochiya, Effect of evaporation and solute capillary-driven flow upon
motion and resultant deposition of suspended particles in volatile droplet on
solid substrate, Adv. Space Res. 41 (2008) 2089–2093.
[42] F. Pereira, J. Lu, G.E. Castano, M. Gharib, Microscale 3D flow mapping with
lDDPIV, Exp. Fluids 42 (2007) 589–599.
[43] W. Thielicke, E.J. Stamhuis. Time-resolved digital particle image velocimetry
tool for matlab, 2005.
[44] R.J. Adrian, R.D. Keane, Y. Zhang, Super resolution particle imaging
velocimetry, Meas. Sci. Technol. 6 (1995) 754–768.
[45] P.J. Jaijus, A. Singh, Flow visualization and solute transport in evaporating
droplets, AIChE J. 56 (2010) 1674–1683.
[46] K. Uno, K. hayashi, T. Hayashi, K. Ito, H. Kitano, Particle adsorption in
evaporating droplets of polymer latex dispersions on hydrophilic and
hydrophobic surfaces, Colloid Polym. Sci. 276 (1998) 810–815.