PHYSICAL REVIEW E 86, 051602 (2012)

Droplet hydrodynamics during lysozyme protein crystallization

T. Pradhan, M. Asfer, and P. K. Panigrahi
Department of Mechanical Engineering, Indian Institute of Technology, Kanpur, India
(Received 3 July 2012; revised manuscript received 28 August 2012; published 7 November 2012)
Various experimental studies in zero gravity have been reported in the literature for generation of superior
quality crystals due to the importance of convective transport on protein crystal quality. However, limited
experimental and numerical studies are available in the literature for a complete characterization of transport
phenomena during the protein crystal growth process. The present investigation reports experimental results on
convective motion inside the droplet during protein crystallization by the vapor diffusion method. Lysozyme
crystals are grown using a sitting drop method and micro-particle image velocimetry is used for investigating
the detailed hydrodynamics inside the droplet. Dynamic evolution of the flow field for the complete crystal
growth process, i.e., during the prenucleation, nucleation, and postnucleation stage, is reported. Various flow
transitions are observed during the complete cycle of the protein crystal growth process. Significant Marangoni
convection is observed during the prenucleation period followed by buoyancy-driven convection during the
postnucleation period. The Marangoni convection flow patterns observed during the prenucleation stage match
those of evaporating droplets. The postnucleation convection patterns are similar to those of ethanol-water mixture
evaporation with high ethanol concentration.

DOI: 10.1103/PhysRevE.86.051602 PACS number(s): 81.10.−h

I. INTRODUCTION Droplet evaporation is one of the primary transport mecha-

Molecules in protein crystals are held together by a
weak force of attraction [1]. Therefore, protein crystals are
susceptible to small stresses. Stress on crystal surface due to
fluid flow may cease crystal growth by removing the molecules
from the surface. The random fluid motion may also deteriorate
the crystal quality. To reduce the buoyancy-driven flow, many
experiments have been carried out in space stations [2].
Crystals grown in a microgravity environment are superior to
crystals grown in the presence of gravity in terms of crystal size
and diffraction characteristics [2]. Microgravity conditions
reduce buoyancy-driven turbulent flows in the bulk solution
from which a crystal emerges and so are thought to promote
crystal nucleation and ideal growth [3].
Lin et al. [4] carried out numerical simulation of convection
around a crystal placed inside a crystal growth chamber. They
observed convection due to depletion of solute concentration
near the crystal, leading to generation of an upward plume.
Savino and Monti [5] reported numerical simulation results
for prenucleation and postnucleation phases of the protein
crystal growth process. Marangoni convection due to the
concentration gradient was observed during the prenucleation
stage. Kawaji et al. [6] studied convection inside the droplet
using the hanging drop method of protein crystallization using
particle tracking velocimetry having a velocity resolution
well below 0.2 μm/s. They did not observe any detectable
displacement of tracer particles, indicating the absence of
any convection inside the droplet at typical protein crystal
growth conditions. Lappa [7] compared the growth and fluid
dynamics of protein crystals with those of soft organic
tissues using numerical simulation. Srivastava et al. [8]
reported the concentration field around a growing Potassium
dihydrogen phosphate (KDP) crystal using laser schlieren
tomography. Gupta et al. [9] reported the transport phe-
nomena during the lysozyme protein crystal growth process
in a hanging drop configuration using the color schlieren
technique.
nisms during protein crystal growth by vapor diffusion. Exper-
imental studies on the hydrodynamics of droplet evaporation
during the protein crystal growth process are not available
in the literature. However, several studies on the general
droplet evaporation process are available and some of these
studies with relevance to the protein crystal growth process
are reviewed below.
Deegan et al. [10] reported the drying of dilute colloidal
suspensions of polystyrene microspheres in water for three dif-
ferent configurations: (a) normal evaporation, (b) evaporation
of the droplet surrounded by a bath of water approximating
a spatially uniform evaporation condition, and (c) a droplet
inside a chamber with a small hole above its center for
approximating center-enhanced evaporation conditions. The
deposit profile was observed to be centrally dominated for the
center-enhanced evaporation case (c), in contrast to that of
case (a) and case (b), indicating inward flow near the substrate
surface for case (c). Fischer [11] examined the deposition
pattern of colloidal particles from the droplet using lubrication
theory for various evaporation models: (a) edge-enhanced
evaporation, (b) constant evaporation and, (c) center-enhanced
evaporation. The circulation direction, i.e., inward or outward
flow near the substrate surface, was observed to relate to the
mode of evaporation. The liquid flows inward to the droplet
center for the center-enhanced evaporation case and outward
from the droplet center for either edge-enhanced or constant
evaporation conditions.
Kang et al. [12] reported the velocity field inside an evapo-
rating ethanol droplet on a hydrophobic substrate for different
ethanol concentrations using particle image velocimetry (PIV).
They observed inward flow near the substrate surface and an
upward plumelike flow near the droplet center region. The
regular flow field transformed to an irregular flow pattern
with increase in ethanol concentration. This was attributed
to buoyancy-induced convection due to the unstable density
distribution from the faster evaporation of ethanol at the surface

1539-3755/2012/86(5)/051602(8) 051602-1 ©2012 American Physical Society

T. PRADHAN, M. ASFER, AND P. K. PANIGRAHI PHYSICAL REVIEW E 86, 051602 (2012)

compared to the core region. Hu and Larson [13] observed Droplet containing
z protein and precipitants
different deposition patterns of polystyrene particles during
the drying of water and octane droplets on a Corning cover slip. y
Crystal Growth Chamber

Both experiment and Brownian dynamics simulation showed x

similar particle deposition profiles after drying. The octane Reservoir Solution
Objective
droplet particle deposition peaks at the center of the droplet,
in contrast to that of water, where the peak is observed at
the edge of the droplet. This difference in particle deposition
pattern was attributed to the opposite direction of Marangoni Epi-fluorescent Prism
Nd:YAG Laser

convection between the octane and water droplets.

Ristenpart et al. [14] reported circulation behavior during Synchronizer
evaporation of methanol and ethanol droplets based on the CCD Camera
particle deposition patterns and numerical simulation of the Computer
Laplace equation. They observed the direction of the circulat-
ing flow to be a function of relative thermal conductivity (KR ), (a)
i.e., the ratio between the thermal conductivity of substrate and
liquid. Radially inward flow along the substrate was observed
for KR < 1.45 and outward flow along the substrate was
observed for 1.45 < KR < 2.0.
The vapor diffusion method of protein crystallization
involves complex transport processes initiated by the con-
centration difference between the droplet and the reservoir
solution. Evaporation from the droplet to the reservoir solution
takes place due to the difference in the partial pressure of
water vapor in the crystallization chamber. The evaporation
process leads to temperature and concentration gradients,
resulting in Marangoni convection inside the droplet. The mass (b)
transfer from the droplet also causes a density difference along
the vertical direction, which influences the flow field in the
presence of gravity.
In summary, the crystal growth process involves fluid flow,
mass transfer, moving boundaries, three-dimensionality, mul-
ticomponent systems, and a combined mode of heat transfer,
i.e., conduction, convection, evaporation, and condensation.
Therefore, accurate numerical modeling is limited by various
assumptions required for simplification of the governing equa-
tions. Quantitative experimental studies face the challenge of (c)
sophisticated hardware requirements for investigation at small
scales. In addition, precise control of the experimental condi- FIG. 1. (Color online) (a) Experimental arrangement for visu-
tions is also essential due to the sensitivity of the nucleation and alization of the protein crystal growth process, (b) schematic of
protein crystal growth process. The present study reports the the crystal growth chamber and associated physical processes, and
hydrodynamics during the complete cycle, i.e., prenucleation, (c) schematic illustrating the PIV data acquisition process.
nucleation, and postnucleation growth phases, of a lysozyme
protein crystal grown by using the sitting drop method. acetate with a pH of 4.5), and 222 μL of water. A droplet of
0.3 μL containing protein (70% W/V) and reservoir solution
in a 7:3 ratio is placed on the siliconized cover slip. The initial
II. EXPERIMENTAL DETAILS
radius of the spherical droplet is about 500 μm. Fluorescent
The complete experimental setup for visualization of the particles of 2 μm in diameter (Invitrogen carboxylate modified
protein crystal growth process is shown in Fig. 1(a). The fluorospheres) is added to the droplet for visualization of the
droplet is illuminated by a 532-nm-wavelength laser beam flow field. The seeding concentration inside the droplet is kept
from a Nd:Yag laser (Newwave) passing through an epi- at 0.0166% volume fraction. There are approximately 8–12
fluorescent prism and microscope objective. The fluorecent particles per interrogation volume inside an interrogation area
particle images inside the droplet are acquired by a charge- of 64 × 64 pixels. Crystals grown at the same experimental
coupled device camera synchronized with the laser exposure conditions without and with seeding particles show similar
timing. size, growth rate, and number density distribution, indicating
The droplet is placed on a siliconized cover slip. The cover a negligible influence of seeding particle on the crystal growth
slip is kept inside the crystal growth chamber and the droplet is process.
surrounded by the reservoir solution [Fig. 1(b)]. The reservoir PIV measurements are conducted at eight horizontal X-Y
solution is prepared using 250 μL NaCl (30% W/V), 428 planes [Fig. 1(c)] inside the droplet by appropriately traversing
μL ethyl glycol (70% V/V), 100 μL buffer (0.1M sodium the microscope objective in the Z direction. The time interval

051602-2
DROPLET HYDRODYNAMICS DURING LYSOZYME PROTEIN . . . PHYSICAL REVIEW E 86, 051602 (2012)

600 (µm/s)
2.6
400 2.2
1.8

200 1.4
1.0

y (µm)
0.6
0
0.2

-200
(a)
-400

-600
-500 0 500
x (µm)
(a)
Crystals 600 (µm/s)
2.6
400 2.2
1.8
(b) 1.4
200
y (µm)
1.0
0.6
0
0.2

-200

-400

-600
-500 0 500
(c) x (µm)
(b)
FIG. 2. Droplet image with the growing crystal at different times (µm/s)
after initiating the crystal growth process: (a) 70 min, (b) 100 min, 2.6
and (c) 250 min. 2.2
1.8

between two laser pulses is set between 0.5 to 2.5 s depending 1.4
y (µm)

1.0
on the strength of convection velocity. PIV evaluation is carried
0.6
out using an adaptive cross-correlation algorithm in a multigrid 0.2
setting from 512 × 512 pixels to 64 × 64 pixels interrogation
area. A Thermosensorik camera (Insb 640 SM/M) is used for
IR imaging of the droplet.

III. RESULTS AND DISCUSSION

Small detectable protein crystals are observed by optical
microscope after 60 min of initiating the crystal growth
process. Figure 2 shows the bright field image of the protein
drop after 70, 100, and 250 min of initiating the crystal growth
process. Two crystals are seen at the bottom right corner of
the droplet, which grow with time. Crystal images acquired at
later times did not show any more crystals at other locations.
Therefore, the transport processes inside the droplet discussed
in the following section can be clearly related to these growing
crystals. The hydrodynamics inside the droplet during the
prenucleation and postnucleation period is of interest to the
present study and will be discussed in the following sections.
Figure 3 shows the velocity vector field inside the droplet
for various transverse, X-Y planes corresponding to different
distances from the cover slip surface (Z = 80, 120, and
x (µm)
(c)
FIG. 3. (Color online) Velocity vector field at different X-Y
planes during the prenucleation stage of the crystal growth process,
i.e., after 15 min of initiating the crystal growth process at various
Z locations from the cover slip surface: (a) Z = 80 μm, (b) Z =
120 μm, and (c) Z = 160 μm. (See Supplemental Material [15] for
the visualization movie during the prenucleation stage of the crystal
growth process.)
160 μm) after 15 min of initiating the crystal growth process,
which corresponds to the prenucleation stage of the crystal
growth process. There is inward flow toward the center of the
droplet near the substrate surface and outward flow from the

051602-3
T. PRADHAN, M. ASFER, AND P. K. PANIGRAHI PHYSICAL REVIEW E 86, 051602 (2012)

droplet center near the top region of the droplet. It may be

noted that this observation is contrary to the observation in o
Ref. [6], where no convective motion was observed during the 25 25.2 25.5 25.6 25.8 26 26.2 ( C)
protein crystallization in the hanging drop configuration. The
conductivity ratio (KR ) for the present study is about 1.33 and
the flow patterns observed in the present study (Supplemental
Material [15] and Fig. 3) is similar to that of methanol and
ethanol droplet evaporation in Ref. [14] for a similar value of
KR . Based on the drying pattern of octane droplets Hu and
Larson [13] also predicted inward Marangoni flow near the
substrate surface.
The visualization images of the protein droplet at different
instants of time indicate a pinning behavior of the droplet due
to the hydrophobic nature of the cover slip. The influence of
solute diffusion on the droplet convection pattern during the
prenucleation stage can be assumed to be negligible. The mass
transfer from the droplet surface due to the partial pressure
difference between the ambient near the droplet surface and
that of the reservoir solution is the primary driving force for
generation of the flow field inside the droplet. Calculation
of relevant dimensionless numbers is expected to provide (a)
insight into the physical mechanism responsible for detailed
hydrodynamics inside the droplet. The physical properties of
the protein droplet has been assumed to be same as that of
water for calculation of these dimensionless numbers. The
o
other experimental parameters used for the calculation are as 24 24.4 24.8 25.2 25.6 26 ( C)
follows: radius of the droplet, R = 0.5 mm, and height of the
droplet, ho = 0.5 mm. The temperature difference is assumed
to be T = 1.0 ◦ C based on a sample IR thermography
measurement of the droplet (see Fig. 4).
The capillary number (− (δσ/δT σ
)T
), which is the ratio of
viscous force to surface tension force, is equal to 2.2 ×
10−3 . The Bond number (ρgRho /σ ), which is the ratio of
gravitational force to surface tension force, is equal to 0.3 ×
10−3 . Smaller values of capillary number and Bond number
indicate the dominant role played by surface tension, which
is also responsible for the spherical shape of the droplet.
The thermal Grashoff number (gβT R 3 /ν 2 ), which is the
ratio of buoyancy force to viscous force, is equal to 0.238,
indicating the negligible role played by thermal buoyancy. The
thermal Marangoni number ( −δσ/δT μα
)RT
), which is the ratio
of surface tension gradient force to viscous force, is equal to
62 840, indicating the greater role played by the surface tension
gradient force. Overall, a higher value of Marangoni number (b)
indicates the dominant role played by Marangoni convection
inside the droplet (Fig. 3). FIG. 4. (Color) Sample IR images of the droplet during the
The net change in surface tension gradient depends on prenucleation stage of the protein crystal growth process at different
the relative contribution of the temperature gradient and times: (a) t = 2 min and (b) t = 32 min.
concentration gradient. The local change in surface tension
on the droplet surface can be expressed as on Marangoni stress depends on the local values of the
concentration gradient and the temperature gradient.
∂σ ∂σ
σ = C + T . (1) The convection patterns in Fig. 3 are similar to drying
∂C ∂T of colloidal suspensions with center-dominated evaporation
Hence, the ratio between the strength of solutal Marangoni [10,11]. A sample IR image of a protein droplet during
convection to that of thermal Marangoni convection is the prenucleation stage is shown in Fig. 4. The lower
δσ δσ δσ
proportional to δC / δT . The magnitude of δC is about temperature at the central region of the droplet in comparison to
−4 −4
−2 × 10 m /s and δT is about −2.2 × 10 N/(m K)
3 2 δσ
the edge region indicates center-dominated evaporation of
for lysozyme. Therefore, it is expected that both solutal the droplet during the protein crystal growth process. This
and thermal Marangoni convection have the same order of can possibly be attributed to higher mass transfer at the top
magnitude. Hence, the relative importance of both mechanisms region of the droplet. The viscous resistance offered by the

051602-4
DROPLET HYDRODYNAMICS DURING LYSOZYME PROTEIN . . . PHYSICAL REVIEW E 86, 051602 (2012)

cover slip surface to the motion of the water vapor from 600 (µm/s)
the droplet to the surrounding reservoir may be responsible 0.25
for the center-dominated evaporation. The similarity in the 400 0.22
convection pattern between the present study and [10,11] with 0.19
central-dominated evaporation indicates the dominant role 200 0.16

y (µm)
played by the evaporation mode in the droplet hydrodynamics. 0.13
0 0.10
Figure 5 shows the velocity vector field after 60 min of
0.07
initiating the crystal growth process, which corresponds to the 0.04
nucleation stage of the crystal growth process. The magnitude -200

of the maximum velocity at 60 min has dropped in comparison

-400
to the earlier time, t = 15 min, indicating the drop in rate
of evaporation with time. The vector field shows asymmetric
-600
behavior, in contrast to the symmetric flow behavior observed -500 0 500
at earlier times in Fig. 3. This may be attributed to the initiation x (µm)
of the nucleation process at the bottom right-hand corner of the
(a)
droplet. The flow direction in Fig. 5(a) near the substrate sur-
face (Z = 80 μm) is similar to that in Fig. 3(a) for all quadrants 600 (µm/s)
except the fourth quadrant. The direction of convective motion
0.25
at Z = 120 μm [Fig. 5(b)] is completely opposite to that in 400 0.22
Fig. 3(b). Figure 5(c) shows the reversal of flow direction in 0.19
the fourth quadrant compared to that in Fig. 3(c) at an earlier 200 0.16
time of crystal growth. It may be remembered that Fig. 2 y (µm) 0.13
0.10
has shown the appearance of two small crystals in the fourth 0
0.07
quadrant after 70 min of initiating the crystal growth process.
0.04
Hence, this reversal of flow pattern in the fourth quadrant of the -200
droplet may be attributed to the nucleation process occurring
at an earlier time, i.e., at about 60 min. -400

Figure 6 shows the velocity vector field from zoomed parti-

-600
cle field images acquired using 10× objectives at a later time, -500 0 500
t = 110 min. This figure corresponds to the fourth quadrant
x (µm)
of the droplet with growing crystals (see Fig. 2). Inward flow
(b)
toward the crystal in the near-substrate region and outward
flow away from the crystal near the top surface of the crystal 600 (µm/s)
are observed in the zoomed vector field. The protein molecules 0.25
moving toward the crystal are deposited on its surface, leading 400 0.22
to a decrease in the fluid density. This low-density fluid rises 0.19
like a plume near the crystal surface due to the buoyancy effect. 200 0.16
y (µm)

Therefore, there is a net outward flow in the upper region of 0.13

0 0.10
the crystal to maintain the continuity of flow. The significant
0.07
variation of the velocity magnitude in the X-Y measurement 0.04
plane indicates a high degree of flow complexity with greater -200

three-dimensionality. Figure 6(b) shows a vortical structure,

indicating the presence of a rotational flow field near the
crystal. It may be noted that Lappa [7] also observed circulation
bubbles near the growing crystal from numerical simulation.
However, a single vortex structure is observed in the present
study, in contrast to the two vortices observed in Ref. [7], due
to the presence of a neighboring crystal in the present con-
figuration. Overall, the heavier fluid with higher concentration
transports the protein molecules toward the crystal, which sub-
sequently gets deposited on the crystal surface. This leads to a
drop in the local concentration and density of the fluid near the
crystal surface, which rises like a plume away from the crystal.
This flow behavior near the growing crystal (Supplemental
Material [15] and Fig. 6) is similar to the numerical prediction
of Ref. [5] for protein crystal growth and the experimental
observation in Ref. [8] for KDP crystal growth.
All the measurements in the eight X-Y planes are acquired
in a time period of 4–5 min. The velocity field at each X-Y
measurement plane shows negligible variation during the total
-400
-600
-500 0 500
x (µm)
(c)
FIG. 5. (Color online) Velocity vector field at different X-Y
planes during the nucleation phase, i.e., 60 min after initiating the
crystal growth process for various distances from the cover slip
surface: (a) Z = 80 μm, (b) Z = 120 μm, and (c) Z = 160 μm.
(See Supplemental Material [15] for the visualization movie during
the postnucleation stage of the crystal growth process.)
time period of measurement in all X-Y planes. Therefore, these
measurements may be assumed to be instantaneous in nature.
This facilitates the use of the steady continuity equation for

051602-5
T. PRADHAN, M. ASFER, AND P. K. PANIGRAHI PHYSICAL REVIEW E 86, 051602 (2012)

(µm/s) (µm/s)
0.75 400 0.00 0.37 0.74 1.11 1.49 1.86 2.23 2.60
0.65

Z (µm)
0.55
0.45
0.35 200
y (µm)

0.25
0.15
0.0 5
0
-400 -200 0 200 400
X (µm)
(a)

(µm/s)
400 0.00 0.11 0.21 0.32 0.43 0.54 0.64 0.75
x (µm)

Z (µm)
(a)
(µm/s) 200
0.75
0.65
0.55
0.45 0
-400 -200 0 200 400
0.35
X (µm)
y (µm)

0.25
0.15 (b)
0.05

(µm/s)
400 0.00 0.04 0.07 0.11 0.14 0.18 0.21 0.25
Z (µm)

200

x (µm)
(b) 0
-400 -200 0 200 400
(µm/s)
0.75
X (µm)
0.65 (c)
0.55
0.45
0.35
FIG. 7. (Color online) Reconstructed velocity field in the X-Z
plane passing through the droplet center (Y = 0) at different times:
y (µm)

0.25
0.15 (a) t = 15 min, (b) t = 45 min, and (c) t = 75 min.
0.05

where u, v, and w are the components of velocity in the x,

y, and z directions, respectively. The discretized form of the
above equation can be written as
ui+1,j,k − ui,j,k vi,j +1,k − vi,j,k wi,j,k+1 − wi,j,k
+ + = 0,
x y z
(3)
x (µm)
(c) where the intervals x and y are the spatial distances
between neighboring locations of the velocity measurement in
FIG. 6. (Color online) Velocity vector field at different X-Y the x and y directions, respectively, and z is the distance be-
planes during the postnucleation stage using zoomed particle images tween the consecutive x-y measurement planes. The values of
acquired using the 10× objective after 110 min of initiating the crystal
x, y, and z for the present experiment are equal to 42.27,
growth process for various distances from the cover slip surface:
42.93, and 40 μm, respectively. The indices i, j , and k corre-
(a) Z = 48 μm, (b) Z = 80 μm, and (c) Z = 112 μm. (See
spond to the grid points for velocity measurement in the x, y,
Supplemental Material [15] for the visualization movie around the
growing crystal.)
and z directions, respectively. Assuming the no-slip boundary
condition at the cover slip surface and using the u,v velocity
incompressible fluid flow given as measurements at different x-y planes, one can calculate the
third component of velocity, w, using the above equation.
∂u ∂v ∂w
+ + = 0, (2) The reconstructed velocity field results are reported in
∂x ∂y ∂z Fig. 7. There are two regular circulation patterns about the

051602-6
DROPLET HYDRODYNAMICS DURING LYSOZYME PROTEIN . . . PHYSICAL REVIEW E 86, 051602 (2012)

3 stage, i.e., till time t = 45 min [Fig. 8(a)]. There is a 61%

reduction in the maximum u velocity from 2.6 to 0.6 μm/s
2
during this time period. A small asymmetry in the velocity
profile is observed at time, t = 60 min, corresponding to the
time of nucleation. The maximum u velocity drops further
1 to about 0.2 μm/s at time t = 75 min and the profile
u (µm/s)

shows complete asymmetry. Figure 8(b) shows symmetric

0 behavior of the v-velocity profile about the droplet center in
the prenucleation phase, i.e., till time t = 45 min. There is also
-1 a drop in v-velocity magnitude with increase in time till t = 45
15 Minutes min. Subsequently, the v-velocity profile shows an asymmetric
30 Minutes distribution. There is a subsequent increase in the v-velocity
-2 45 Minutes
60 Minutes magnitude at times t = 60 and 75 min. This may be attributed
75 Minutes to the buoyancy-driven flow due to the crystal growth process
-3 during the postnucleation period. The v-velocity magnitude is
-500 0 500
x (µm) higher than the u-velocity magnitude at times t = 60 and 75
(a) min. The maximum velocity during the prenucleation stage
is about 10 times higher than that during the postnucleation
0
stage. The flow circulation strength during the prenucleation
stage is related to the rate of evaporation, which influences the
time required to reach the supersaturation level for initiating
-0.1 the nucleation of the protein crystal. The convective motion
during the postnucleation stage is related to the buoyant force
v (µm/s)

caused by the crystal growth process.

-0.2
15 Minutes
30 Minutes
-0.3 45 Minutes
60 Minutes
75 Minutes
-500 0 500
x (µm)
(b)
FIG. 8. Velocity profile at Z = 80 μm from the cover slip surface
at different times after initiation of the crystal growth process: (a) u
velocity and (b) v velocity.
centerline of the droplet during the initial period, i.e., 15 min
after initiation of the crystal growth process [Fig. 7(a)]. The
strength of the circulation is reduced at later time, i.e., 45
min after initiating the crystal growth process [Fig. 7(b)]. No
clear circulation pattern is observed at 75 min after initiating
the crystal growth process [Fig. 7(c)]. There is a shift in
the overall flow structure toward the right-hand side of the
droplet, i.e., in the direction of the growing crystal. The
asymmetric flow pattern observed for higher concentration
ethanol drops in Ref. [12] is similar to the postnucleation flow
pattern observed here [Fig. 7(c)], indicating the influence of
buoyancy-dominated flow.
The dynamic evolution of u-velocity (y component of
velocity) and v-velocity (z component of velocity) profiles
can provide a quantitative representation and insight into the
detailed hydrodynamics during the crystal growth process.
Figure 8 shows the u-velocity and v-velocity profiles along
the diameter of the droplet at X = 0 and Z = 80 μm for dif-
ferent time instants. The u-velocity distribution shows perfect
symmetry about the droplet center during the prenucleation
IV. CONCLUSION
The convective flow field measured inside a droplet dur-
ing the lysozyme protein crystal growth process by using
the sitting drop vapor diffusion method has been reported.
Microscale particle image velocimetry is used both for
visualization and quantitative characterization of the flow
field. The dynamic evolution of the complete flow field
during the prenucleation phase, nucleation phase, and post-
nucleation phase of the protein crystal growth process is
reported.
There is an inward flow toward the droplet center from
the edge region close to the cover-slip surface during the
prenucleation stage. Simultaneous flow from the top of the
droplet surface toward the pinned contact line along the droplet
surface generates a recirculating flow pattern with a plumelike
flow from the droplet center near the cover-slip surface toward
the top of the droplet. The direction of Marangoni convection is
similar to that of a droplet with central-dominated evaporation.
The strength of Marangoni convection decreases with time
till the nucleation phase of the protein crystal. Subsequently,
buoyancy-dominated flow patterns develop around the crystal,
leading to an asymmetric complex flow field inside the droplet.
The magnitude of the maximum velocity in the buoyant flow
regime is about 1/10 of that during the Marangoni-convection-
dominated flow regime. The overall flow field in the postnucle-
ation regime is due to the combined influence of the Marangoni
force and the buoyancy force, which is also related to the
location of the nucleation site. Overall, the droplet hydrody-
namics during the protein crystal growth process shows a flow
transition from evaporating-droplet-type motion [10,11,13,14]
during the prenucleation stage to a plume-type motion during
the postnucleation stage [5,8,12]. The observation from the
present study is contrary to that of Ref. [6], where no convec-
tive motion was observed during the crystal growth process.

051602-7
T. PRADHAN, M. ASFER, AND P. K. PANIGRAHI
ACKNOWLEDGMENTS
The authors acknowledge the Department of Science
and Technology, Government of India, for the finan-
[1] A. Ducruix and R. Giege, Crystallization of Nucleic Acids and
Proteins (Oxford University Press, New York, 1999).
[2] R. K. Strong, B. L. Stoddard, and A. Arrott, J. Cryst. Growth
119, 200 (1992).
[3] J. R. Helliwell and N. E. Chayen, Nature 448, 658 (2007).
[4] H. Lin, F. Rosenberger, J. I. D. Alexander, and A. Nadarajah,
J. Cryst. Growth 151, 153 (1995).
[5] R. Savino and R. Monti, J. Cryst. Growth 165, 308 (1996).
[6] M. Kawaji, O. Gamache, D. H. Hwang, N. Ichikawa, J. Viola,
and J. Sygusch, J. Cryst. Growth 258, 420 (2003).
[7] M. Lappa, J. Theor. Biol. 224, 225 (2003).
[8] A. Srivastava, K. Muralidhar, and P. K. Panigrahi, Appl. Opt.
44, 5381 (2005).
051602-8
PHYSICAL REVIEW E 86, 051602 (2012)
cial support, Mr. Ayodhya Prasad Prajapati for data
presentation, and Mr. Maheshwar Dewangan for IR
imaging.
[9] A. S. Gupta, P. K. Panigrahi, K. Muralidhar, and R. Gupta,
J. Cryst. Growth 312, 817 (2010).
[10] R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S. R. Nagel,
and T. A. Witten, Phys. Rev. E 62, 756 (2000).
[11] B. J. Fischer, Langmuir 18, 60 (2002).
[12] K. H. Kang, S. J. Lee, C. M. Lee, and I. S. King, Meas. Sci.
Technol. 15, 1104 (2004).
[13] H. Hu and R. G. Larson, J. Phys. Chem. B Lett. 110, 7090
(2006).
[14] W. D. Ristenpart, P. G. Kim, C. Domingues, J. Wan, and H. A.
Stone, Phys. Rev. Lett. 99, 234502 (2007).
[15] See Supplemental Material at http://link.aps.org/supplemental/
10.1103/PhysRevE.86.051602 for movies.