Controllable Formation of Aromatic Nanoparticles in A Three-Dimensional Hydrodynamic Flow Focusing Microfluidic Device

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

RSC Advances

View Article Online


PAPER View Journal | View Issue
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

Controllable formation of aromatic nanoparticles in a


three-dimensional hydrodynamic flow focusing
Cite this: RSC Advances, 2013, 3,
17762 microfluidic device3
Liguo Jiang,a Weiping Wang,{b Ying Chauab and Shuhuai Yao*ac

Herein we investigate the formation of aromatic organic nanoparticles in a three-dimensional (3D)


hydrodynamic flow focusing microfluidic device. We demonstrate a microfluidic based solvent/non-
solvent exchange technique that enables controllable formation of aromatic nanoparticles with tunable
size and size distribution. The 3D focusing is achieved by hydrodynamically focusing the sample stream
with sheathed streams in both horizontal and vertical directions, and solvent/non-solvent exchange
happens between the solvent contained sample stream and non-solvent contained horizontally sheathed
streams. The 3D focusing effect was visualized using florescence confocal microscopy and the dynamics of
solvent (DMF) depletion in the focused stream was calculated using a finite element computation software
package COMSOL Multiphysics. By analyzing the results of self-assembled aromatic nanoparticles in the
microfluidic device, we find that the speed of DMF depletion strongly influences the size and size
Received 26th April 2013,
Accepted 24th July 2013
distribution of self-assembled aromatic nanoparticles, and a rapid depletion of DMF is critical for achieving
small aromatic nanoparticles with narrow size distribution. This work suggests that our 3D hydrodynamic
DOI: 10.1039/c3ra42071j
flow focusing microfluidic device with the ability to precisely control the convective-diffusion process, and
www.rsc.org/advances continuously vary the fluid flow conditions is promising for studying the formation of nanomaterials.

1 Introduction Recently we designed a tripodal molecule (FTAEA) by


coupling 9-fluorenylmethoxycarbonyl (Fmoc) groups with
Aromatic interactions, consisting of van der Waals, hydro- three legs of tris(2-aminoethyl)amine (TAEA) to fabricate
phobic and electrostatic forces1,2 are ubiquitous in chemical nanoparticles (NPs) by self-assembly.16,17 This idea was
and biological systems and play an important role in inspired by clathrin molecules which can assemble into a
stabilizing the structures of DNA, RNA and proteins.3–5 They polyhedral lattice surrounding the membrane vesicle for
also play a significant role in the self-assembly of amyloid trafficking in cells,18 and fully employs the concept of p–p
fibrils.6 And the interactions with aromatic rings are key interactions mediated molecular self-assembly. Such aromatic
processes in chemical and biological recognition.7 Although NPs self-assembling from small molecules can be functiona-
aromatic interactions are theoretically not fully understood,8,9 lized in a versatile manner for targeted drug delivery.17
the utilization of aromatic residues to mediate molecular self- For nanoparticle drug delivery systems, rapid clearance of
assembly to form well-ordered novel nano-structure materials circulating NPs is a critical issue. The size of NPs has been
has received notable achievements.10–15 It is assumed that the demonstrated to be one of the key factors affecting the
p stacking interactions between aromatic moieties provide biodistribution and clearance of the circulating system.19 This
energetic contribution as well as order and directionality for is due to the fact that particle size not only dictates their
the formation of well-ordered nanostructures.10,11 transport and adhesion in blood vessels but also mediates the
cellular response.20,21 From this point of view, controllable
formation of stable NPs with tunable size is critical for the
a
Division of Biomedical Engineering, Bioengineering Graduate Program, Hong Kong
application of NPs as drug delivery carriers. Based on the
University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China
b
Department of Chemical and Biomolecular Engineering, Hong Kong University of classical nucleation and growth theory,22,23 the formation of
Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China NPs in solution is engaged in two processes: nucleation and
c
Department of Mechanical Engineering, Hong Kong University of Science and growth. And the kinetics of two processes is governed by the
Technology, Clear Water Bay, Kowloon, Hong Kong, China. E-mail: meshyao@ust.hk saturation conditions of the solution, which is determined by
3 Electronic supplementary information (ESI) available. See DOI: 10.1039/
the monomer concentration and the solvent/non-solvent
c3ra42071j
{ Current address: The David H. Koch Institute for Integrative Cancer Research, (dimethylformamide (DMF) is the solvent and water is the
Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA non-solvent fluid in our system) ratio in solution. When the

17762 | RSC Adv., 2013, 3, 17762–17769 This journal is ß The Royal Society of Chemistry 2013
View Article Online

RSC Advances Paper

solution is undersaturated, in which the monomer concentra-


tion is lower than the critical concentration or the solubility of
the monomer at the presented solvent/non-solvent ratio,
aromatic NP nuclei may be formed, but it rapidly dissipates
due to the high nucleation barrier.23 In such situations, either
increasing the monomer concentration or decreasing the
solvent/non-solvent ratio will create a supersaturation condi-
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

tion in solution, which leads to aromatic NP nucleation and


growth. And decreasing the solvent/non-solvent ratio can be
readily accomplished by exchanging or replacing solvent
molecules with non-solvent molecules. Thus, we expect that
controlling the solvent/non-solvent ratio is a straightforward
way to control the formation of aromatic NPs. Conventionally,
organic NPs are self-assembled by bulk nanoprecipitation
method,24 which involves dropwise addition of stock solution
in solvent into a larger volume of non-solvent and mixing
solutions by vortex. Although the nanoprecipitation method is
simple, this approach is by no means to control the
microscopic solvent/non-solvent ratio due to the uncontrol-
lable nature of chaotic and turbulent mixing, resulting in
uncontrollable size and size distribution. The other issue of
the current NP formation method comes from the conflict
between the slowly bulk mixing by vortex and the rapid self-
assembly of aromatic NPs. This conflict may hurdle the ability
to investigate the dynamic process of the NP formation and
thus limit us in understanding the fundamental of the self-
assembly mechanism.
Hydrodynamic flow focusing (HFF)25–27 micromixers, in
which the central stream is hydrodynamically focused to a
narrow jet by two side streams at a cross junction under
Fig. 1 (a) A schematic shows the concept of hydrodynamic flow focusing.
laminar flow condition (Fig. 1a), provide an alternative
Solvent/non-solvent exchange is achieved by solvent molecules diffusing into
platform for nanomaterial formation.28–33 Hydrodynamic the focused stream and non-solvent molecules diffusing out of the focused
focusing reduces the stream width (wf) to a few micrometers stream. (b) A schematic of the microfluidic device showing the concept of
even submicrometers, and consequently reduces the diffusion 3DHFF. Cross section A shows the sample stream focused in vertical direction,
path (wf), where solvent molecules diffuse out and non-solvent and cross section B shows the sample stream focused in both vertical and
horizontal direction.
molecules diffuse in. Therefore, the time for solvent extraction
from its initial solvent/non-solvent ratio to a critical ratio,
when nuclei starts to be formed and growth, can be greatly
two-level layers35 or a single layer with sequential inlets.34 In a
shortened by the diffusion based extraction in HFF.
3DHFF system, the reagents and precipitating NPs are isolated
Furthermore, the diffusion path can be controlled by the
from the channel walls, therefore aggregation and/or clogging
applied pressure or flow rate and thus the speed of the solvent
can be minimized. In addition, by constraining the sample
depletion can be varied to offer different stable reaction
stream in the center of the microchannel where flow velocity
environments for organic nanoparticle nucleation and growth.
reaches the maximum with less variation, the 3D focused
This technique has already been successfully applied to
sample stream is expected to have a uniform width and thus
control the liposome and block copolymer vesicle forma-
improved the uniformity of the solvent/non-solvent ratio.
tion,29,30 polymeric nanoparticle self-assembly,31 noisome and
In this work, we used a simple 3DHFF microfluidic device36
liposome-hydrogel hybrid nanoparticle self-assembly,32,33 and
made of two layers of PDMS for nanoparticle formation.
rubrene nanocrystal formation.28 However, most HFF systems
FTAEA NPs with tunable size and narrow size distribution were
using two-dimensional (2D) focusing, in which the reagent in
obtained by controlling the microfluidic flow conditions. The
the central stream is only focused in the horizontal plane, may
encounter several problems such as sticking of molecules on size distribution of the NPs collected at different flow
the channel walls34 (see ESI3 Fig. S1) and non-uniform solvent/ conditions was measured using dynamic light scattering
non-solvent exchange in the vertical direction. To address (DLS). Based on the classic nucleation and growth theory, we
these problems, a three-dimensional hydrodynamic flow explored how the size of self-assembled FTAEA NPs is related
focusing (3DHFF) configuration in both horizontal and vertical to the dynamics of the solvent (DMF) depletion in the focused
planes has been implemented in microfluidic systems using stream.

This journal is ß The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 17762–17769 | 17763
View Article Online

Paper RSC Advances

2 Experimental section Health) were used to analyze and extract the cross-sectional
images. An inverted epifluorescence microscope (Eclipse Ti,
2.1 Reagents and materials Nikon) with an objective lens (206/0.45 NA) was also used to
FTAEA dissolved in dimethylformamide (DMF) stock solution observe microfluidic mixing in the device. A mercury lamp
was prepared as described in the previous work,16 and stored with a filter cube (B-2A, Nikon) was used for fluorescence
at 220 uC prior to usage. Depending on the experimental excitation/emission. A highly sensitive CCD camera (EXi Blue,
requirements, the stock solution was dissolved in DMF–water Q-IMAGING) was mounted on the microscope to capture the
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

(75 : 25, volume ratio) mixture to make FTAEA solutions at images.


different concentrations. Poly(dimethylsiloxane) (PDMS)
(Sylgard 184 silicone elastomer) was purchased from Dow 2.4 DLS measurement
Corning. Photoresist SU-8 2025 was purchased from The nanoparticle number average diameter and distribution
MicroChem Corp. FITC-dextran-10K and fluorescein were measurements were performed on Zeta Plus (Brookhaven
purchased from Invitrogen. Instruments Corp.) with the use of the BIC Particle Sizing
Software. The collected product solution (y80 mL) was put into
2.2 Microfluidic device fabrication and implementation
a cuvette (Eppendorf UVette) and tested at 25 uC with a
Fig. 1b schematically shows the concept of the fabricated detection angle of 90u and a wavelength of 659 nm. Each
3DHFF microfluidic device which consists of two layers of sample was run for 5 times with 60 s per run. The data of each
microchannels.36 Within the microchannel network, the run were collected only when its baseline index was larger than
sample stream (FTAEA in DMF–water solution, red color) is or equal to 5.
first vertically focused by two streams of DMF–water (green
color) to form a parallel central stream (see cross section A) in
a 10 mm width central microchannel. And the laminar central
stream is horizontally focused by two water streams (blue 3 Numerical simulation section
color) from both sides, resulting in an isolated sample stream
The microfluidic flow and DMF depletion dynamics in the
(see cross section B) in a 20 mm width microchannel. The
microfluidic device was fabricated with two layers of PDMS central stream sheathed by two adjacent water streams in a
using a standard soft lithography process. For PDMS replica hydrodynamic flow focusing device were simulated using a
molding, a negative photoresist SU-8 (SU-8 2025) was spin- commercial finite element code, COMSOL MULTIPHYSICS 4.1
coated on a silicon wafer in a thickness of 40 mm, and then it (Comsol, Inc.). The fluid flow in microfluidic channels is
was photolithographically patterned and developed, yielding a governed by the incompressible steady-state Navier–Stokes
mold master of 40 mm deep microchannels. A PDMS mixture equations, and the diffusion between DMF and water is
(in a 10 : 1 weight ratio of monomer and curing agent) was governed by the steady-state convective-diffusion equation:37–41
poured over the molds, degassed in a vacuum chamber, and A
cured in an oven at 85 uC for 2 h. The resulted PDMS layers are +?V = 0 (1)
4 mm and 0.8 mm in thickness, respectively. After the PDMS A A A
replicas were removed from molds, individual device was cut, r(V ?+)V = 2+P + m+2V (2)
and inlet, outlet holes were punched by a pan head needle on A
the thick PDMS replica. The two PDMS replicas were treated V ?+C = D+2C (3)
with oxygen plasma and aligning bonded to seal microchan- A
nels. Before an experiment, the device was fixed on the chip- where V is the flow velocity, r is the fluid density, P is the
holder and reagents were preloaded in glass (Hamilton, Reno, pressure, m is the fluid dynamic viscosity, C is the concentra-
NV) and plastic (BD) syringes with polyetheretherketone tion of DMF, and D is the mutual diffusion coefficient of DMF
(PEEK, Upchurch Scientific) capillary tubing connecting the in water. In the simulation, the fluid density, dynamic
syringes to microdevice access holes. Reagents were pumped viscosity, and diffusion coefficient are a function of the DMF
into the microdevice using syringe pumps (KD Scientific) and concentration and are expressed by a fourth order polynomial
the self-assembled products were collected from the outlet. fitting density data, viscosity data, and diffusion coefficient
2.3 Confocal laser scanning microscopy and epifluorescence data at 20 uC, as reported by Volpe et al.42
microscopy To quantify the DMF depletion dynamics in the central
stream, we traced 20 streamlines across the focused central
Cross sections of 3D focused sample streams were imaged by
stream in the midplane of the microchannel. We follow the
the Zeiss laser scanning confocal microscope (LSM7 DUO)
suggestion from Hertzog et al.39 and Park et al.27 to define the
with 0.5 mm resolution in both horizontal and vertical
directions. An Argon ion laser at 488 nm wavelength was time ti of DMF depletion at each streamline in our system, of
employed as the excitation source. The excitation beam was which the clock starts when the concentration of DMF along
focused into the chip by an air immersion objective lens (406/ the streamline has decrease to the 99% of the initial
0.6 NA) to excite FITC-dextran-10K labeled sample stream and concentration in the central stream, and the average time of
the emission light was collected by the same objective and DMF depletion ,t> at x location is calculated by summing and
detected by a photomultiplier tube (PMT) detector. ZEN 2009 averaging the contribution of the depletion times from all the
analysis software (Zeiss) and ImageJ (National Institutes of streamlines:

17764 | RSC Adv., 2013, 3, 17762–17769 This journal is ß The Royal Society of Chemistry 2013
View Article Online

RSC Advances Paper

ð
x dx
ti ðxÞ~ (4)
x0 ,c~0:99c0 ui ðxÞ

P
i ti ðxÞwi ðxÞ
P
StðxÞT~ (5)
i wi ðxÞ
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

And the standard deviation of the depletion time, which


represents the uniformity of DMF concentration across the x
location, is defined as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
P
i ðti ðxÞ{St ðxÞTÞ2 wi ðxÞ Fig. 2 Cross section images of 3D focused streams at different flow conditions,
sðxÞ~ P ðti w0Þ (6)
taken at 40 mm downstream at cross section B in Fig. 1b. Flow rates in
i wi ðxÞ
experiments: Qbf1–Qbf2–QSa–QWt ml min21 (a) 1–1–0.5–20; (b) 1–1–0.5–40; (c) 1–
1–0.5–60.
where ui is the x component of the flow velocity in the ith
streamline, wi is the width of the ith streamline, and C0 is the
initial concentration of DMF. All the simulated data are
analyzed using customer-built programs in MATLAB. 4.2 Focused stream width determined by FRR
As the time for a molecule to diffuse across the focused stream
is primarily determined by the width of the focused stream, it
is necessary to investigate the governing parameters for the
4 Results and discussion
width of the focused stream. As shown in the confocal scans,
4.1 Cross-sectional images of 3D focused stream the artifact of the non-uniform profile is negligible in the
3DHFF focused stream and the stream is mostly focused in the
Due to the fact that the 2D focused stream may spread out of
horizontal direction. Therefore, we use a 2D theoretical model
the midplane (over the depth of the channel) and its speed (Fig. 1a) for predicting the width of the focused stream in a
varies from zero near the channel walls to a maximum at the rectangular channel.26,29 We combine the sample stream and
midplane, 3D focusing is preferred to achieve a nearly uniform the two vertical streams as the central stream. According to the
velocity in the focused stream and therefore nearly uniform mass conservation, the width of the focused central stream
solvent/non-solvent ratio for the sample molecules carried in can be derived as
the focused stream. To visualize the 3D focusing effect of the
sample stream, we labeled the sample stream with FITC- w0
Wf ~ (7)
dextran-10K, and obtained the cross sectional images of the cð1zFRRÞ
sample stream at different flow conditions in the 3DHFF
where w0 is the width of the outlet channel, c is the ratio of the
microdevice using confocal laser scanning microscopy. The
average velocity in the focused central stream to the average
flow rates of the sample stream and the two vertical focusing
velocity in the outlet channel, and the flow rate ratio (FRR) is
streams were set at 0.5, 1, and 1 mL min21, respectively, while
defined as the ratio of the total flow rate of the two side
the total flow rate of water streams varied from 20, 40 to 60 mL
min21, and the corresponding Reynolds number is 0.83, 1.66, streams to the flow rate of the central stream. Hence, wf is
and 2.49, respectively. The confocal scans were taken at 40 mm governed by FRR and the velocity ratio (c). The theoretical
downstream of the outlet channel at the location of cross analysis results26 show that the velocity ratio c approximately
section B, as indicated in Fig. 1b, and the results are shown in equals to 1.5 when the aspect ratio (e, height to width) of the
Fig. 2a–c. They clearly reveal several important facts. Firstly, all outlet channel is larger than 1.78 (e = 2 in our device).
three focused streams are isolated from the channel walls. Therefore, the diffusion path wf is mainly governed by the FRR
Secondly, the focused stream profile has relatively uniform in a given fluidic network. To verify the relationship between
width which ensures the nearly uniform molecular diffusion wf and FRR, we experimentally measured the midplane width
across it. Thirdly, the focused stream becomes narrower at of the focused stream in the confocal scans at different FRR
higher water flow rate conditions, thus the time for water conditions. The FITC-dextran-10K labeled sample solution
molecules to diffuse across the focused stream is shorter, remains as a focused stream at the center of the channel and
according to the Fick’s law of diffusion.43 We also took the width of the focused stream is determined when the intensity
cross section images at a fixed total flow rate and a fixed flow drops to half of maximum value in the lateral direction. As
rate of Qwt while changing the flow rate of Qbf1, Qbf2, and QSa, shown in Fig. 3, the experimental data match well with
as shown in ESI3 Fig. S2. We have found that the aspect ratios theoretical calculation by eqn (7). This indicates that FRR,
of the focused streams vary but the widths of focused streams apparently, is the dominated parameter that governs the width
are almost identical, which imply that the processes of the of the focused stream, and hence the solvent depletion
diffusion based solvent depletion in these conditions are dynamics in HFF system. Thus for all experiments and
similar. simulations presented in this article, we set the flow rate of

This journal is ß The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 17762–17769 | 17765
View Article Online

Paper RSC Advances

images (Fig. 4e–h). In the experiments, the FITC labeled water


streams were pumped into the two side inlets, and DMF/water
solvent streams were pumped into the device as the central
stream. Both simulated and experimentally captured images at
the midplane show a reasonable coherence at the same FRR
conditions, indicating the accuracy of our simulation results.
When changing the FRR condition, we observed that the
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

focused central stream reached the steady-state flow within


one minute and the focused central stream was very stable.
Fig. 4 clearly illustrates the speed of DMF depletion at
different FRR conditions along the channel. At low focusing
conditions, the intensity of the focused stream slightly
changes along the channel, indicating the slowly exchanging
of DMF and water molecules across the relatively wide jet. In
contrast, the intensity of the focused stream rapidly changes at
high focusing conditions due to the fast extraction of DMF
Fig. 3 3D focused stream width versus FRR. Symbols represent the experimen- molecules out of the narrow focused jet. At high FRR
tally measured data and the solid line is the theoretical prediction by eqn (7) (c = conditions, another important phenomenon was observed in
1.5). experiments that two slightly brighter belts (Fig. 4g and h)
appear around the central stream probably due to the
changing of the fluorescence yield when microfluidic environ-
the two vertical focusing streams and the sample stream at 1, 1 mental polarity and viscosity change upon the curved
and 0.5 mL min21, respectively, which resulted in a 2.5 mL interfaces.44 We believe that the curved interface is caused
min21 of the central stream. We varied the flow rate in the two by the larger flow velocity gradients near the channel top and
side channels to vary the value of FRR. bottom walls.
As the formation of NPs is determined by the microenviron-
4.3 Numerical analysis of mixing performance ment of reagent molecules and this microenvironment is
Under laminar flow condition and along with flow boundary mediated by the diffusion based extraction of DMF in the
and concentration boundary conditions in microfluidic HFF stream, to carry out controllable formation of NPs, we
system, the flow and concentration fields are governed by the investigated the dynamics of DMF depletion in the hydro-
steady-state Navier–Stokes equations and convective-diffusion dynamic focused stream as described in the simulation
equation. Use of 2D flow simulation in our analysis, that section. Fig. 5a shows the depletion of DMF in the central
greatly reduces computational time, is justified for the stream with the time defined in eqn (5) at different FRR
uniform width of the 3D focused stream. Since the aspect conditions. The results clearly reveal that, at low FRR
ratio of the focused stream is more than 10, the DMF depletion conditions, the volume fraction of DMF in the focused stream
dynamics at the midplane can be approximated using a 2D depletes gently. In contrast, at high FRR conditions, the
model (Fig. 4a–d). Meanwhile, the 2D simulation results can volume fraction of DMF deplete rapidly within sub-millise-
be qualitatively validated by the epifluorescence microscopic cond timescale. In addition, at different FRR conditions, the

Fig. 4 Comparison of numerically simulated DMF depletion (a–d) and experimentally observed DMF depletion (e–h) in the focused stream at different FRR conditions.
The width of the focused central stream decreases as the FRR increases, which results in faster depletion of DMF in the focused stream.

17766 | RSC Adv., 2013, 3, 17762–17769 This journal is ß The Royal Society of Chemistry 2013
View Article Online

RSC Advances Paper

is very uniform to provide homogeneous reaction environ-


ments.

4.4 Dependence of FTAEA NP size and size distribution on FRR


We investigated the formation of FTAEA NPs in terms of the
size and size distribution by controlling the depletion of DMF
mediated by different FRR conditions in the 3D HFF
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

microfluidic device. Fig. 6 shows the results of self-assembled


FTAEA NPs at different FRR conditions with the sample stream
containing an FTAEA initial concentration of 20 mM in DMF–
water (75 : 25, volume ratio). For FRR , 20, the size and size
distribution of self-assembled FTAEA NPs are highly sensitive
to the FRR condition. As shown in Fig. 6a, at FRR = 8, relatively
larger FTAEA NPs with a mean size of y140 nm and a wide
size distribution are obtained, while increasing FRR results in
FTAEA NPs with smaller mean size and an improved size
distribution. For FRR ¢ 20, the size and size distribution of
self-assembled FTAEA NPs remain nearly unchanged for
different FRR conditions (Fig. 6b–d), all showing a mean size
of y40 nm and a similar narrow size distribution. For
comparison, aromatic NPs were formed by conventional bulk
nanoprecipitation, and the results are presented in Fig. 6e and
f. The testing samples have the same finial FTAEA concentra-
Fig. 5 (a) Simulation results show the depletion of the DMF along the focused tion and the DMF–water ratios correspond to FRR = 12 (Fig. 6e)
central stream at different FRR conditions, and the depletion timescale is on the and FRR = 32 (Fig. 6f). Slow and uncontrollable exchange of
order of submillisecond to millisecond. (b) Simulation results show the standard solvent/non-solvent in bulk nanoprecipitation results in
deviations of the depletion time along the outlet channel. Small values in the
aromatic NPs with a relatively large size and a broad size
standard deviation in time ensure uniform DMF concentrations across the
focused stream.
distribution. A similar trend of NP size and size distribution
changing with different FRR conditions were obtained in the
3DHFF microfluidic device when the sample stream contains
an FTAEA initial concentration of 10 mM (see ESI3 Fig. S3).
values of standard deviations of the depletion time defined in
Combining the theory of nucleation and growth in solutions
eqn (6) are shown in Fig. 5b varying from 24 to 42 ms, which
and the numerical analysis of DMF depletion dynamics
means that the DMF concentration within the focused stream
presented in Fig. 4 and 5, we try to explain the results of

Fig. 6 DLS measurements of self-assembled FTAEA NPs at different FRR conditions with the sample stream containing an FTAEA initial concentration of 20 mM. (a)
For FRR , 20, the mean size and size distribution of NPs are highly dependent on FRR. (b–d) For FRR ¢ 20, the mean size and size distribution of NPs are similar. (e) &
(f) Aromatic NPs formed by conventional bulk nanoprecipitation show a wide size distribution.

This journal is ß The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 17762–17769 | 17767
View Article Online

Paper RSC Advances

when the value of FRR is larger than 20, smaller FTAEA NPs
with similar mean size and size distribution are obtained.

5 Conclusions
In this paper, we demonstrate controllable formation of FTAEA
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

NPs by nanoprecipitation enabled by solvent/non-solvent


exchange in a 3D HFF microfluidic device. From the cross-
sectional view, the focused sample stream is isolated from the
channel walls by the sheath flow in both vertical and
horizontal directions, thereby preventing surface sticking
and channel clogging problems. The stable, uniformly focused
stream ensures a homogenous reaction environment.
Numerical simulation results show that the dynamics of
DMF depletion can be mediated by the flow rate in the
microchannels. Based on the self-assembled NP size distribu-
Fig. 7 The mean size of self-assembled NPs versus FRR for two different initial tion and the classical nucleation and growth theory, we
FTAEA concentrations. The trends of FRR conditions controlling the NP speculate that rapid depletion of DMF is critical to achieve
formation in the nucleation and growth progress are clearly illustrated. small and uniform aromatic NPs. At low water-to-sample flow
rate ratio, gentle depletion of DMF in the focused central
stream leads to a lot of solubilized FTAEA molecules for nuclei
self-assembled FTAEA NPs in the 3D HFF microfluidic device. growing, resulting in larger FTAEA NPs. In contrast, at high
Fig. 7 plots the mean size of self-assembled NPs versus the FRR flow rate ratio, rapid depletion of DMF induces the formation
conditions. The dashed lines and dotted lines illustrate the of a larger number of FTAEA nuclei. Thus, few free FTAEA
trends of the FRR conditions controlling the NP formation at molecules are left for nuclei growth, resulting in smaller
two different FTAEA initial concentrations. Apparently, a FTAEA NPs with narrower size distribution. The formation of
higher initial concentration of FTAEA results larger aromatic aromatic NPs with a desired size can be achieved using the
NPs at the same FRR condition. One possibility is that in a
demonstrated 3DHFF microfluidic system, which benefits
micrometer or submicrometer confined space the number of
biomedical applications with the NPs used as drug delivery
nuclei are likely the same at a certain FRR condition.
carriers. The developed 3DHFF microfluidic system is com-
Therefore, for higher concentration with more monomer
pletely applicable for fast self-assembly of nanomaterials with
molecules in the focused stream, larger NPs were formed.
or without sticking problem on the surface wall, for the solvent
The assumption of the same number of nuclei is quantitatively
depletes rapidly in the narrowly focused stream, 3D focused
verified by the fact that the volume (pR3/6, R is the mean size of
NPs) of NPs formed with FTAEA initial concentration of 20 mM stream is isolated from surface walls, and the uniform width of
is approximately two times larger than that formed with the 3D focused stream ensures a relatively homogenous
FTAEA initial concentration of 10 mM when the FRR condition reaction environment for the formation of nanomaterials.
is the same. Furthermore, the dependence of self-assembled
NPs on FRR is also obvious. At low FRR conditions, the side
sheath flow forms a relatively wide focused sample stream, Acknowledgements
across which the exchange of solvent molecules slowly takes
place. In the solutions of the focused sample stream, the DMF We thank the Nanoelectronics Fabrication Facility, Biosciences
fraction is still high, where less FTAEA nuclei are formed and Central Research Facility, Materials Characterization and
abundant FTAEA molecules remain solubilized. Thereby, the Preparations Facility, Advanced Engineering Material Facility
nuclei of FTAEA NPs are growing in size due to the aromatic and Professor Baoling Huang’s group at HKUST for their
stacking interaction of free FTAEA molecules on the nuclei technical support. This work was supported by HKUST Research
during gentle depletion of DMF. Therefore, the size of FTAEA Project Competition Grant (RPC10EG30).
NPs highly depends on the FRR condition. In contrast, at high
FRR conditions, rapid extraction of DMF molecules takes place
by diffusion of water molecules across a tightly focused References
stream. As a result, a steep depletion of DMF occurs in the
1 C. A. Hunter and J. K. M. Sanders, J. Am. Chem. Soc., 1990,
focused central stream along the downstream which provides
112, 5525–5534.
nearly homogeneous microenvironment for FTAEA NP forma-
2 M. L. Waters, Curr. Opin. Chem. Biol., 2002, 6, 736–741.
tion. The steep depletion of DMF induces a larger number of
3 M. E. W. Saenger, Principles of Nucleic Acid Structure,
FTAEA nuclei formed, leading to less dissolved FTAEA free
Springer Advanced Texts in Chemistry, 1984.
molecules in the focused stream for NP growth. As a result, 4 S. Burley and G. Petsko, Science, 1985, 229, 23–28.

17768 | RSC Adv., 2013, 3, 17762–17769 This journal is ß The Royal Society of Chemistry 2013
View Article Online

RSC Advances Paper

5 C. A. Hunter, J. Singh and J. M. Thornton, J. Mol. Biol., 27 H. Y. Park, X. Qiu, E. Rhoades, J. Korlach, L. W. Kwok, W.
1991, 218, 837–846. R. Zipfel, W. W. Webb and L. Pollack, Anal. Chem., 2006,
6 E. GAZIT, FASEB J., 2002, 16, 77–83. 78, 4465–4473.
7 E. A. Meyer, R. K. Castellano and F. Diederich, Angew. 28 V. Génot, S. Desportes, C. Croushore, J.-P. Lefèvre, R.
Chem., Int. Ed., 2003, 42, 1210–1250. B. Pansu, J. A. Delaire and P. R. von Rohr, Chem. Eng. J.,
8 M. O. Sinnokrot and C. D. Sherrill, J. Phys. Chem. A, 2006, 2010, 161, 234–239.
110, 10656–10668. 29 A. Jahn, W. N. Vreeland, M. Gaitan and L. E. Locascio, J.
9 S. Grimme, Angew. Chem., Int. Ed., 2008, 47, 3430–3434. Am. Chem. Soc., 2004, 126, 2674–2675.
Published on 25 July 2013. Downloaded by Institute of Macromolecular Chemistry on 06/11/2017 15:58:36.

10 M. Reches and E. Gazit, Science, 2003, 300, 625–627. 30 J. Thiele, D. Steinhauser, T. Pfohl and S. Förster, Langmuir,
11 M. Reches and E. Gazit, Nano Lett., 2004, 4, 581–585. 2010, 26, 6860–6863.
12 Y. Zhang, H. Gu, Z. Yang and B. Xu, J. Am. Chem. Soc., 2003, 31 R. Karnik, F. Gu, P. Basto, C. Cannizzaro, L. Dean, W. Kyei-
125, 13680–13681. Manu, R. Langer and O. C. Farokhzad, Nano Lett., 2008, 8,
13 V. Jayawarna, M. Ali, T. A. Jowitt, A. F. Miller, A. Saiani, J. 2906–2912.
E. Gough and R. V. Ulijn, Adv. Mater., 2006, 18, 611–614. 32 C. T. Lo, A. Jahn, L. E. Locascio and W. N. Vreeland,
14 R. Orbach, L. Adler-Abramovich, S. Zigerson, I. Mironi- Langmuir, 2010, 26, 8559–8566.
Harpaz, D. Seliktar and E. Gazit, Biomacromolecules, 2009, 33 J. S. Hong, S. M. Stavis, S. H. DePaoli Lacerda, L.
10, 2646–2651. E. Locascio, S. R. Raghavan and M. Gaitan, Langmuir,
15 A. M. Smith, R. J. Williams, C. Tang, P. Coppo, R. 2010, 26, 11581–11588.
F. Collins, M. L. Turner, A. Saiani and R. V. Ulijn, Adv. 34 M. Rhee, P. M. Valencia, M. I. Rodriguez, R. Langer, O.
Mater., 2008, 20, 37–41. C. Farokhzad and R. Karnik, Adv. Mater., 2011, 23,
16 W. Wang and Y. Chau, Chem. Commun., 2011, 47, H79–H83.
10224–10226. 35 C. Simonnet and A. Groisman, Appl. Phys. Lett., 2005, 87,
17 W. Wang and Y. Chau, Chem. Mater., 2012, 24, 946–953. 114104.
18 T. Kirchhausen, Annu. Rev. Biochem., 2000, 69, 699–727. 36 C. Chih-Chang, H. Zhi-Xiong and Y. Ruey-Jen, J. Micromech.
19 F. Alexis, E. Pridgen, L. K. Molnar and O. C. Farokhzad, Microeng., 2007, 17, 1479.
Mol. Pharmaceutics, 2008, 5, 505–515. 37 A. Jahn, S. M. Stavis, J. S. Hong, W. N. Vreeland, D.
20 S. Mitragotri and J. Lahann, Nat. Mater., 2009, 8, 15–23. L. DeVoe and M. Gaitan, ACS Nano, 2010, 4, 2077–2087.
21 W. Jiang, Y. S. KimBetty, J. T. Rutka and C. 38 Z. Wu and N.-T. Nguyen, Sens. Actuators, B, 2005, 107,
W. ChanWarren, Nat. Nanotechnol., 2008, 3, 145–150. 965–974.
22 M. Perez, M. Dumont and D. Acevedo-Reyes, Acta Mater., 39 D. E. Hertzog, X. Michalet, M. Jäger, X. Kong, J. G. Santiago,
2008, 56, 2119–2132. S. Weiss and O. Bakajin, Anal. Chem., 2004, 76, 7169–7178.
23 D. Kashchiev and G. M. van Rosmalen, Cryst. Res. Technol., 40 D. E. Hertzog, B. Ivorra, B. Mohammadi, O. Bakajin and J.
2003, 38, 555–574. G. Santiago, Anal. Chem., 2006, 78, 4299–4306.
24 H. Fessi, F. Puisieux, J. P. Devissaguet, N. Ammoury and 41 S. Yao and O. Bakajin, Anal. Chem., 2007, 79, 5753–5759.
S. Benita, Int. J. Pharm., 1989, 55, R1–R4. 42 C. D. Volpe, G. Guarino, R. Sartorio and V. Vitagliano, J.
25 J. B. Knight, A. Vishwanath, J. P. Brody and R. H. Austin, Chem. Eng. Data, 1986, 31, 37–40.
Phys. Rev. Lett., 1998, 80, 3863–3866. 43 A. Fick, Ann. Phys. Chem., 1855, 170, 59–86.
26 L. Gwo-Bin, C. Chih-Chang, H. Sung-Bin and Y. Ruey-Jen, J. 44 D. Magde, G. E. Rojas and P. G. Seybold, Photochem.
Micromech. Microeng., 2006, 16, 1024. Photobiol., 1999, 70, 737–744.

This journal is ß The Royal Society of Chemistry 2013 RSC Adv., 2013, 3, 17762–17769 | 17769

You might also like