TECHNICAL NOTE www.rsc.org/loc | Lab on a Chip

Tuneable hydrophoretic separation using elastic deformation ofpoly(dimethylsiloxane)†

Sungyoung Choi and Je-Kyun Park*

Received 13th November 2008, Accepted 2nd March 2009

First published as an Advance Article on the web 13th March 2009

DOI: 10.1039/b820364d

This paper demonstrates a method for tuning elastomeric microchannels for hydrophoretic separation

made in poly(dimethylsiloxane) (PDMS). Uniform compressive strain is imposed on the elastomeric

microchannel between two acrylic substrates by fastening the bolts. The elastomeric microchannel can

change its cross-section during compression, simultaneously tuning the criterion for hydrophoretic

ordering. The change of the channel cross-section under compression is studied using a confocal

microscope and finite element method (FEM). By pressing the channel for hydrophoretic separation,

we achieved tuning of the separation criterion from 7 to 2.5 mm in particle diameter.

Microfabrication allows for engineering fluidic devices with

accurate and well-defined channel dimensions at micro- or nano-

scale. In microfabricated devices, a detailed understanding of

fluid transport has been contributed to the development of new

techniques for sorting bioparticles. Microfluidic networks,

a connected set of microchannels, have been used to separate

blood plasma and isolate leukocytes by precisely controlling the

ratio of the hydrodynamic resistances between the micro-

channels.1–3 Microfabricated filters with precise pore sizes were

designed for leukocyte enrichment, leukapheresis, and macro-

molecular separation.4–6 Particle separation in a pinched micro-

channel has been implemented by pushing and aligning particles

to the channel wall.7 Regular sieve structures with homogeneous

nano-pores have been used to continuously separate DNA and

proteins improving our understanding of their separation

processes.8–10 Such devices have great potential in cellular and

molecular separation applications, yet they lack the flexibility for

particle separation. As changing separation targets with ones of

new size, it is inevitable to modify the devices with new designs

and dimensions.

Soft-lithography represents non-conventional lithographic

techniques using elastomeric materials, mostly poly-

(dimethylsiloxane) (PDMS).11,12 The PDMS elastomer is a soft

material with Young’s modulus of approximately 750 kPa,

allowing the design and fabrication of flexible microfluidic

systems that can adjust the dimension and shape of a micro-

channel by applying mechanical loads. Adjustable nano-

channels made in PDMS were used to selectively transport

nanoparticles through dynamic modulation of the channel

cross-section.13 Tuneable separation in a pinched microchannel

has been implemented by using microvalves that modulate the

ratio of the hydrodynamic resistances between outlet chan-

nels.14 In a micropost array for deterministic lateral

Department of Bio and Brain Engineering, College of Life Science andBioengineering, KAIST, 335 Gwahangno, Yuseong-gu, Daejeon, 305-701,Republic of Korea. E-mail: jekyun@kaist.ac.kr; Fax: +82-42-350-4310;Tel: +82-42-350-4315

† Electronic supplementary information (ESI) available: Figs. S1 and S2.See DOI: 10.1039/b820364d

1962 | Lab Chip, 2009, 9, 1962–1965

displacement, a critical particle size for separation has been

tuned by stretching the device.15

Recently, micron- and submicron-scale obstacles have been

used to separate microparticles, cells, and macromolecules by

hydrophoresis which is the movement of suspended particles

under the influence of the pressure field induced by the obsta-

cles.16–18 This technique takes advantage of a sheathless method,

passive operation, and single channel. However, the micro-

fabricated obstacles for hydrophoresis also have the aforemen-

tioned flexibility problem. To assume hydrophoretic ordering,

the diameter of particles should be more than the half of the

obstacle gap (Fig. 1).18 As changing separation targets with ones

Fig. 1 Size-tuneable microfluidic device for hydrophoretic separation.

The elastomeric channel consists of an anisotropic obstacle array to

induce hydrophoretic ordering of particles. A fluid flow is applied along

the y-axis. (A) Particles (d in diameter) smaller than h1/2 follow the

rotational flow without hydrophoretic ordering. (B) By compression of

the device (H in thickness) into H�DH, the obstacle gap can be adjusted

to h2 < 2d where the particles are in hydrophoretic ordering.

This journal is ª The Royal Society of Chemistry 2009

Fig. 2 Confocal cross-section images of the microchannel for hydro-

phoretic separation. (A) Schematic of the channel geometry filled with

fluorescein isothiocyanate (FITC). (B) Cross-section a0a (upper) and b0b

(lower) without compression. (C) Cross-section a0a (upper) and b0b

(lower) under compression. Applied compressive strain makes the cross-

sectional area decrease and lowers the channel height (lateral and vertical

scale bars¼ 10 mm). The blurring at the bottom boundaries of the images

in panel C might be due to the scattering at the surface of the acrylic

substrate. The dashed lines are drawn to clarify the top and bottom

boundaries of the channel.

of new size, the devices have to be newly designed and fabricated

with appropriate dimensions. To overcome this limitation we

demonstrate a tuneable elastomeric microchannel with aniso-

tropic obstacles whose gap can be adjustable by applying

compressive forces. We present a compression system and

compression conditions for tuning channel dimension and

geometry to separate target particles of many different sizes.

Experimental

We fabricated hydrophoretic devices incorporating anisotropic

microfluidic obstacles in PDMS using two-step photolithog-

raphy, similarly described in our previous work.18 The channels

are 50 mm wide and 40 mm height with 180 obstacles with an

obstacle gap of 13 mm, a thickness of 23 mm, and a pitch distance

of 27 mm. The obstacles are inclined at an angle of 80 to the right

side wall. The deviations of all dimensions were below 5.5%.

Fluorescent polystyrene beads with 1 (green), 4 (red), and 10

mm (green) diameters were purchased from Polysciences (War-

rington, PA) and Molecular Probes (Eugene, OR). The beads

were prepared in 2% pluronic F68 solution (Sigma-Aldrich, St.

Louis, MO) with an average concentration of �2.0 � 103, 7.3 �102, and 1.8 � 102 mL�1, respectively.

A two-photon laser scanning microscope (LSM510; Carl Zeiss,

Germany) with femto-second pulsed laser (Chameleon, Coherent,

Santa Clara, CA) was used to provide cross-section images of the

deformed channels. We measured the thickness of the PDMS

devices and its variation under compression using vernier calipers.

Finite-element calculations were performed to study the rela-

tionship between the uniform compressive displacement applied

to PDMS devices and channel deformation using a commercial

software (CFD-ACE+; ESI, Huntsville, AL). The 2D mesh for

finite-element calculations of PDMS channel compression has

thickness of 3.68 mm and width of 6 mm (see ESI Fig. S1†). The

microchannel of 13 mm in height and 50 mm width is defined at

the bottom of the 2D mesh. The Poisson ratio of PDMS is close

to 0.49, which can be treated as a perfectly incompressible

material.19 The Young’s modulus of PDMS is assumed to be 750

kPa.19 The right and left walls of the device are set as the free

boundary condition. The bottom wall is set as the fixed boundary

condition. Setting the constant displacement values for the top

wall as the prescribed displacement boundary condition, we

applied compressive forces to the mesh.

Results and discussion

Principle of tuneable hydrophoretic separation

A compression system is composed of bolts, nuts, and acrylic

substrates (see ESI Fig. S2†). Uniform compressive strain is

imposed on the elastomeric microchannel between two acrylic

substrates by fastening the bolts. The magnitude of compressive

strain applied to the microchannel is varied by changing

compressive displacement. Hydrophoretic ordering is deter-

mined by a kind of steric hindrance mechanism (Fig. 1).18 Steric

hindrance occurs when the rotational motion of particles is

affected by the existence of the obstacles. The particle–obstacle

interaction deflects large particles assuming hydrophoretic

ordering from their streamline and leads to equivalent flow

paths. In contrast, relatively smaller particles out of the critical

This journal is ª The Royal Society of Chemistry 2009

ordering condition just follow the rotational flows induced by the

anisotropic resistance of the obstacles. To assume hydrophoretic

ordering, the diameter of particles should be more than the half

of the obstacle gap. Conversely speaking, the diameter range for

hydrophoretic ordering can be modulated by adjusting the height

of the obstacle gap.18 For example, in the non-deformed PDMS

device, the rotational motion of particles (d in diameter) smaller

than the half of the obstacle gap (h1) cannot be affected by the

existence of the obstacles and follows the rotational streams

(Fig. 1A). The channel deformation by the compressive

displacement hinders the particles (d > h2/2) from following the

rotational flows and allows hydrophoretic ordering (Fig. 1B).

Application of compressive displacement to PDMS devices

In order to acquire cross-section images of the microchannel under

compressive strain, we filled the microchannel with fluorescein

isothiocyanate dye as contrast (Fig. 2A). The frames in Fig. 2B

are confocal micrographs of the cross-sections. When a uni-

form compressive strain (DH/H ¼ 0.16) is applied to the entire

PDMS device (H ¼ 3.68 mm), the top wall of the microchannel

deflects downward (Fig. 2C). The microfabricated channel with

a rectangular cross-section has a maximum deflection at the

middle.20 After compression, the gap height at the middle (cross-

section a0–a in Fig. 2A) was changed from 13 mm to 4.9 mm.

Lab Chip, 2009, 9, 1962–1965 | 1963

The large compressive strain of 0.16 (DH¼ 0.6 mm) applied to the

entire device results in a small channel deflection of 8.1 mm.

Since it is difficult to get analytical results for this structural

mechanics problem, we performed simulations using finite

element method (FEM) and a scaling approximation of the

relationship between the uniform compressive displacement of

the PDMS device and channel deformation. A uniform

compressive displacement is applied over the top wall of the two-

dimensional geometry for FEM simulations (see ESI Fig. S1†).

Rectangular-shaped channels are not changed uniformly under

the uniform compressive displacement (Fig. 3A). The sidewall

deformation of the channels is negligible compared to the extent

of the top wall deformation. In flow channels with a rounded

cross section, the compressive load can be transferred to the

channel edges and cause the uniform deformation of the chan-

nels.21 The FEM results show a linear relationship between the

compressive displacement of the PDMS device and channel

height at the middle, satisfying Hooke’s law (Fig. 3). A relation

describing the maximum variation of a microchannel under

uniform stress follows from Gervais et al.20

Dhmax ¼ cws

E(1)

where Dhmax is the height decrease at mid-width of the channel

under deformation, w is the channel width, s is an applied stress,

E is the Young’s modulus of PDMS, and c is a fit parameter.

Fig. 3 (A) Finite element method (FEM) solutions for compression of

the elastomeric device under prescribed displacements of 0.3, 0.575, and

0.75 mm along the z-axis. A 2D finite-element model as a PDMS device

has thickness (H) of 3.68 mm and width of 6 mm. The microchannel of 13

mm in height and 50 mm in width is defined at the bottom of the 2D mesh.

(B) Maximum channel height vs. compression displacement of PDMS

devices with thickness of 3.68 and 4.68 mm, respectively (:, -: FEM

results). The dashed lines are fitted from eqn. (2) with fit parameters (c) of

1.13 (H ¼ 3.68 mm) and 0.69 (H ¼ 4.68 mm).

1964 | Lab Chip, 2009, 9, 1962–1965

Since the stress–modulus ratio means strain, eqn. (1) can be

rewritten as:

Dhmax ¼ cwDH

H(2)

where DH is the thickness decrease of the entire PDMS device

and H is the thickness of the device. In this scaling approxima-

tion, the channel deformation is proportional to the channel

width and the thickness of the PDMS device rather than the

channel height (Fig. 3B). The former two parameters can be

further used to adjust the channel height for tuneable hydro-

phoretic separation.

Tuneable separation of microparticles

We demonstrated tuneable separation of microparticles with

controlled obstacle gap (Fig. 4). Without compressive displace-

ment, the obstacle gap of 13 mm assures hydrophoretic ordering

of particles more than 7 mm in diameter. Therefore, the 10 mm-

sized beads become focused into the channel middle by hydro-

phoretic ordering, while 4 mm beads are evenly distributed and

separated from the 10 mm beads under a flow rate of 1 mL min�1

(Fig. 4A, C). As mentioned, the hydrophoretic particle motion is

governed by the physical (steric) barrier of the anisotropic

microfluidic obstacles. The particle–obstacle interaction deflects

particles assuming hydrophoretic ordering and makes them

diffuse out of the rotational streamlines (Fig. 1B). At that time,

smaller particles within the critical ordering condition can

approach closer to the top wall of the channel and be exposed to

Fig. 4 Hydrophoretic separation. (A) Fluorescence image showing

separation of 10 (green) and 4 (red) mm beads before compression. The 10

mm beads are in hydrophoretic ordering, whereas the 4 mm beads remain

unfocused. (B) Fluorescence image showing separation of 4 (red) and 1

(green) mm beads after compression. By lowering the obstacle gap, the 4

mm beads are focused to the right sidewall. In contrast, the 1 mm beads are

evenly distributed and separated from the 4 mm beads. (C, D) Fluores-

cence intensities for bead streams in (A) and (B), respectively.

This journal is ª The Royal Society of Chemistry 2009

much higher transverse flows. This steric hindrance mechanism

enables hydrophoretic size separation. The 10 mm-sized beads

smaller than the obstacle gap of 13 mm will be positioned far from

the sidewall, near the channel middle. In addition, although

particles are out of the critical ordering condition, they are still

influenced by the particle–obstacle interaction. Accordingly, the

4 mm-sized beads can follow the steric hindrance mechanism even

though their distribution of the channel width is much larger

than that of particles within the critical ordering condition.

When deformed by uniform compressive strain (DH/H) of 0.16,

the obstacle gap at the middle is tuned from 13 to 4.9 mm in the

PDMS device with thickness (H) of 3.68 mm, simultaneously

changing the trajectory of the 4 mm beads. After the channel

compression, the obstacle gap of 4.9 mm assures hydrophoretic

ordering of particles more than 2.5 mm in diameter. Thus the

4 mm beads focused into the right sidewall are separated from the

1 mm beads that are evenly distributed without hydrophoretic

ordering under a flow rate of 0.4 mL min�1 (Fig. 4B, D). As

mentioned, the height of the obstacle gap (h) is an important

design parameter that determines whether particles (d in diam-

eter) assume hydrophoretic self-ordering in a narrow distribution

of the channel width. Experimentally, we have found that the

obstacles of h # 2d hinder the rotational flows of the particles

and leads to hydrophoretic self-ordering.18 Therefore, the 1 mm-

sized beads out of the critical ordering condition are relatively

free from the particle-obstacle interaction and are positioned in

the wide distribution of the channel width without hydrophoretic

ordering. These results show that this system is capable of sorting

target particles of many different sizes without any modification

of channel dimension or design.

Conclusions

We have described a method of tuning elastomeric micro-

channels made in PDMS using a compression system with bolts,

nuts, and acrylic substrates. This method makes it possible to

adjust the height of the obstacle gap for hydrophoretic separa-

tion. The elastomeric microchannel can change its cross-section

during compression, simultaneously tuning the criterion for

hydrophoretic ordering. The channel deformation can be further

modulated by changing the channel width and the device thick-

ness. Under compressive stress, the microchannels with the same

height but different widths can be deformed into many different

heights by the linear relationship between the channel deforma-

tion and channel width. The partial channel pressing can also

change the microchannels into many different heights that enable

This journal is ª The Royal Society of Chemistry 2009

sequential separation in a hydrophoretic device. By these parallel

operations, the elastomeric microchannel for hydrophoretic

separation can extend its sorting range. We also performed FEM

simulations and a scaling approximation of the relationship

between the uniform compressive displacement of the PDMS

device and channel deformation for various compressive strain

values. These results will be helpful for tuning a hydrophoretic

device or probing different strain applications.

Acknowledgements

This research was supported by the the Korea Science and

Engineering Foundation (KOSEF) NRL Program (R0A-2008-

000-20109-0) and by the Nano/Bio Science and Technology

Program (2005-01291) funded by the Korea government

(MEST). The authors thank the Chung Moon Soul Center for

BioInformation and BioElectronics, KAIST.

References

1 S. Yang, A. €Undar and J. D. Zahn, Lab Chip, 2006, 6, 871.2 S. S. Shevkoplyas, T. Yoshida, L. L. Munn and M. W. Bitensky, Anal.

Chem., 2005, 77, 933.3 M. Yamada and M. Seki, Lab Chip, 2005, 5, 1233.4 V. VanDelinder and A. Groisman, Anal. Chem., 2007, 79, 2023.5 P. Sethu, A. Sin and M. Toner, Lab Chip, 2006, 6, 83.6 C. C. Striemer, T. R. Gaborski, J. L. McGrath and P. M. Fauchet,

Nature, 2007, 445, 749.7 M. Yamada, M. Nakashima and M. Seki, Anal. Chem., 2004, 76,

5465.8 J. Han and H. G. Craighead, Science, 2000, 288, 1026.9 J. Han and H. G. Craighead, Anal. Chem., 2002, 74, 394.

10 J. Fu, R. B. Schoch, A. L. Stevens, S. R. Tannenbaum and J. Han,Nat. Nanotechnol., 2007, 2, 121.

11 Y. Xia and G. M. Whitesides, Annu. Rev. Mater. Sci., 1998, 28, 153.12 D. C. Duffy, J. C. McDonald, O. J. A. Schueller and

G. M. Whitesides, Anal. Chem., 1998, 70, 4974.13 D. Huh, K. L. Mills, X. Zhu, M. A. Burns, M. D. Thouless and

S. Takayama, Nat. Mater., 2007, 6, 424.14 Y. Sai, M. Yamada, M. Yasuda and M. Seki, J. Chromatogr., A, 2006,

1127, 214.15 J. P. Beech and J. O. Tegenfeldt, Lab Chip, 2008, 8, 657.16 S. Choi and J.-K. Park, Lab Chip, 2007, 7, 890.17 S. Choi, S. Song, C. Choi and J.-K. Park, Lab Chip, 2007, 7, 1532.18 S. Choi, S. Song, C. Choi and J.-K. Park, Anal. Chem., 2009, 81, 50.19 D. Armani, C. Liu and N. Aluru, Proc. IEEE Micro Electro

Mechanical Systems, Orlando, FL, USA, 1999, 222–227.20 T. Gervais, J. El-Ali, A. G€unther and K. F. Jensen, Lab Chip, 2006, 6,

500.21 M. A. Unger, H.-P. Chou, T. Thorsen, A. Scherer and S. R. Quake,

Science, 2000, 288, 113.

Lab Chip, 2009, 9, 1962–1965 | 1965