Rapid fabrication of microchannels using microscale plasma activatedtemplating (mPLAT) generated water molds{{

Shih-hui Chao,*ab Robert Carlsonb and Deirdre R. Meldrumab

Received 14th December 2006, Accepted 7th February 2007

First published as an Advance Article on the web 5th April 2007

DOI: 10.1039/b618269k

Poly(dimethylsiloxane) (PDMS) is a common material used in fabricating microfluidic devices. The

predominant PDMS fabrication method, soft lithography, relies on photolithography for fabrication

of micropatterned molds. In this technical note, we report an alternative molding technique using

microscale PLasma Activated Templating (mPLAT). The use of photoresist in soft lithography is

replaced by patterned water droplets created using mPLAT. When liquid PDMS encapsulates

patterned water and then solidifies, the cavities occupied by water become structures such as

microchannels. Using this method, device fabrication is less time consuming, more cost efficient and

flexible, and ideal for rapid prototyping. An additional important feature of the water-molding

process is that it yields structural profiles that are difficult to achieve using photolithography.

Introduction

Present day poly(dimethylsiloxane) (PDMS) soft lithography

relies on photolithography for fabrication of solid molds with

features down to micrometre and even nanometre scales.1

Liquids have not been an option for making molds because

they do not retain shapes at macroscopic scales. However, in

this article, we demonstrate that at micrometre scales, the

shapes of water droplets can be well-controlled and well-

retained for molding polymers (e.g., PDMS) that cure below

the boiling temperature of water. We replace the photoresist

in conventional soft lithography by patterned water as an

alternative method to fabricate PDMS microchannels.

It is well-known that O2-plasma changes polymer surfaces

from hydrophobic to hydrophilic.2 For a PDMS surface made

of 10 wt% cross-linker, the contact angles of untreated and

plasma treated surfaces are 113.5u and 60u, respectively.3

We create hydrophilic patterns using microscale PLasma

Activated Templating (mPLAT), a technique that employs a

mask to block out O2 plasma exposure in certain areas to

create hydrophilic patterns on hydrophobic plastic surfaces.

We have previously reported using mPLAT with microfabri-

cated PDMS stamps as masks to fabricate metal and carbon

electrodes4 and other groups have used similar techniques to

protein patterns5 on the surfaces of various polymers.

Experimental

The principle of making mPLAT-based water molds to fabri-

cate PDMS microchannels is straightfoward. Hydrophilic

patterns are first created on PDMS substrates by spatially

selective plasma exposure using a mask. After plasma treat-

ment, the substrate is briefly dipped into water, leaving hydro-

philic areas covered and hydrophobic areas exposed to air.

Uncured PDMS is then carefully poured on the substrate.

Submersing the patterned droplets in uncured PDMS does not

alter the pattern because oil-based PDMS is immiscible in

water. The hydrophilic pattern retains water droplets until the

PDMS cures. The cavities occupied by water become struc-

tures such as microchannels. Using this approach, device

fabrication does not require photolithography. The details of

the fabrication process are demonstrated in Fig. 1. We

used a 5 vol% glycerol (M778-07, Mallinckrodt Baker, Inc.,

Phillipsburg, NJ) solution to reduce the evaporation rate. The

aMicroscale Life Sciences Center, University of Washington, Seattle,USA. E-mail: joechao@u.washington.edu; Fax: +1-206-221-5264;Tel: +1-206-685-6885bDepartment of Electrical Engineering, University of Washington,Seattle, USA{ The HTML version of this article has been enhanced with colourimages.{ Electronic supplementary information (ESI) available: Details ofmodularized fabrication and Fig. S1. See DOI: 10.1039/b618269k

Fig. 1 mPLAT molding: (a) Adhere a mask to a flat PDMS slab; (b)

Expose to oxygen plasma; (c) Remove the mask, dip-coat the slab in

glycerol solution. Droplets only form in the exposed region; (d) Pour

uncured PDMS on the slab without disturbing the droplet; (e) Cure

PDMS in a waterbath at 60 uC for two hours. The ports were punched

by an 18.5-gauge hypodermic needle to interface 1/320 PEEK tubing.

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

This journal is � The Royal Society of Chemistry 2007 Lab Chip, 2007, 7, 641–643 | 641

Publ

ishe

d on

05

Apr

il 20

07. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

7:02

:38.

View Article Online / Journal Homepage / Table of Contents for this issue

uncured PDMS is poured to the side of this solution, letting

it gently flow over the droplets. All the mPLAT processes

described in this article were done in a plasma cleaner (PDC-

32G, Harrick Plasma, Ithaca, NY) with 6.8 W RF-power. The

major requirements are minute-long O2 plasma treatments and

masks to block O2 plasma, resulting in less time consuming,

more cost efficient and flexible PDMS fabrication, ideal for

rapid prototyping as demonstrated in the following sections.

The requirements of masks for mPLAT channel fabrication

include the ability to isolate O2 plasma, ease of machining, ease

of peeling-off without the risk of damaging both masks and

PDMS surfaces, availability, and, secondarily, reusability. We

chose overhead transparencies to meet these requirements. For

simple channel structures with low resolution requirements,

the mask patterns were manually cut by razorblades; for

channels of complicated structures and/or with high resolution

requirements, we micromachined the transparencies using a

355 nm Nd:YAG laser (Electro Scientific Industries model

4440, Portland, OR).

Channel structures fabricated using the mPLAT water

molding method can be incorporated into multilayer PDMS

structures in the following ways: (1) The mPLAT-based layers

can be bound with other PDMS layers fabricated using either

conventional soft lithography or the mPLAT method; and (2)

The cured PDMS structures can be mPLAT-treated so water

molds can be made on the newly cured layers. A multilayer

pneumatic valve has been made for demonstrative purposes,

and will be described in the Results and Discussion section.

A microfluidic network can be made by multiple exposures

of patterns of individual masks. It has been observed that the

surface exposed to O2 plasma earlier expressed lower levels of

hydrophilicity than the ones exposed later.5 We found the

physical contact between a mPLAT-treated PDMS surface and

a mask surface degrades hydrophilicity of the formerly treated

surface. This degradation is not desirable for water molding

because the channel patterns cannot be filled with water

properly. Our solution is to use water as a protection layer

between treated PDMS surface and masks during the later

exposures. Therefore, following the first exposure, the PDMS

slab was dipped and pulled from water. After pulling out from

the solution, the mask was gently attached to the wet PDMS

slab, so the hydrophilic parts of the surface were separated

from the mask by the glycerol solution. Then the later plasma

exposure procedure duplicates that of the first exposure in

Fig. 1. Multiple mPLAT exposure allows for modularized

fabrication, demonstrated in the ESI.{

Results and discussion

To understand the resolution limits of mPLAT microstructure

fabrication, we laser-cut 20 mm-long slits of widths ranging

from 100 mm to 1 mm on an overhead transparency (Fig. 2a).

Two 1/80 holes were punched at the two ends of all slits to

create tubing interface for the PDMS channels. To test the

production yield rate as a function of channel width, eight

PDMS chips (e.g., Fig. 2b) were made using the mask (Fig. 2a)

and the same batch of PDMS of 10 wt% cross-linker (Dow

Sylgard 184) under the same conditions. A channel would be

identified as successful if liquid could flow from one end to the

other. The yield rates of the channels show almost 100%

success for widths larger than 400 mm, with the yield decreas-

ing at smaller scales (Fig. 2c). Only two of the eight 100 mm

wide channels were connected throughout the channel length.

Hydrophobicity is a molecular property of native PDMS

surfaces. Ideally, the smallest achievable channel width should

be much smaller than 100 mm. However, the edges of our

laser-cut masks are quite rough at a length scale of around

100 microns (Fig. 2d), which introduces complicated boundary

conditions into the hydrophilic pattern. Where the edge of the

pattern is rough, it appears it is energetically favorable to

pinch off the water pattern and thereby truncate the resulting

channel—a primary reason responsible for the decrease in

yield for narrower channels. Ref. 5 reported the smallest

plasma generated feature of 20 mm with good hydrophilic/

hydrophobic contrast, implying the potential of smaller

mPLAT channels. We believe masks with smoother edges will

enable production of small water features with higher yields.

The typical profile of the channels is rounded with an aspect

ratio around 5–6% (Fig. 2e–f). The edge of the water pattern is

pinned at a boundary of hydrophilic and hydrophobic regions.

Structures of different heights can be easily made in one layer

by tuning the width of the water feature (Fig. 2e).

An important advantage of mPLAT microchannel fabrica-

tion is the ease of prototyping for lab-on-a-chip devices,

especially those with simple channel structures and low

resolution requirements. A demonstrative example for fast

prototyping was made by using a transparency mask with the

pattern shown in Fig. 3b manually cut by a razorblade and a

1/80 hole punch. The pattern was similar to those used in

on-chip capillary electrophoresis.6 Fig. 3c and d show the

produced channels filled with food coloring. The fabrication

time required for this chip device is about 2 hours, almost

Fig. 2 (a) Laser machined mask with slits from 100 mm to 1 mm

widths. (b) Resulting channels filled with food coloring. (c) Yield rates

versus channel width. (d) Close-up of the mask with a 100 mm slit. (e, f)

Cross-section micrographs of channels of 1 mm and 750 mm nominal

widths, respectively. The actual widths and maximum heights are

950 mm and 60 mm for the channel in (e), and 760 mm and 40 mm in (f).

642 | Lab Chip, 2007, 7, 641–643 This journal is � The Royal Society of Chemistry 2007

Publ

ishe

d on

05

Apr

il 20

07. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

7:02

:38.

View Article Online

entirely consumed by curing the PDMS. Comparing the close-

up views of the channel intersections in Fig. 3b and Fig. 3d, the

sharp corners of the masks were rounded due to the surface

tension of water. The actual widths of channels were smaller

than the widths of the mask patterns, possibly depending on

the distribution of the oxygen plasma.

A demonstrative example seen in Fig. 4 is a multilayer

pneumatic valve similar to the one introduced by the Quake

group.7 Two channel layers are needed for such valves: to form

a valve seat that can be closed completely, the liquid layer

requires a rounded channel profile (Fig. 4b), whereas the

control channel layer does not (Fig. 4c). To fabricate rounded

channels in the conventional way, molds made of a meltable

photoresist require an extra reflowing process. The rounded

profile of water features fabricated using mPLAT is desirable

for pneumatic valve applications (Fig. 4b). The mPLAT-based

liquid channel layer was fabricated by following the procedure

in Fig. 1. The substrate is a 30 mm PDMS membrane spun on

a piece of transparency. The pneumatic control layer was

fabricated using a photolithography-based mold. The surfaces

of the two layers were treated by O2 plasma for 15 s before

bonding to form a permanent seal. Fig. 4d and e show the

on/off operation of the assembled valve.

Conclusions

We demonstrate, for the first time, the use of water molds to

fabricate PDMS microchannels. The fabrication process is free

from the expensive operation for photolithography. The unique

features such as multiple exposures, rounded channel profiles,

and different channel heights in the same layer are difficult or

impossible to achieve using photolithography. The capital

investments (a plasma cleaner and a water bath) and the cost

of chemicals (water) are much lower than those of traditional

methods. Researchers without access to photolithography faci-

lities can produce PDMS devices from scratch in two hours.

The cost of laser-cut masks is a major part of the total

production cost. Although we cut the masks in house, they can

be made by laser machining foundries with cost ranging around

$500 (similar to the cost for standard photolithography masks),

depending on the cut patterns. Masks made of overhead trans-

parencies are highly reusable, no degradation has been observed.

Important issues of this method such as the control of

channel aspect ratio, ultimate resolution limitation, and the

possibility of high-density patterns, are to be investigated in

the future. Possible candidates to control channel aspect ratio

include loading water using condensation,8 motorized liquid

dispensers, and/or through the control of water mixture

concentration. As to resolution, our study shows that the

quality and resolution of channels are mainly affected by the

mPLAT masks. Other precise cutting tools are sought after for

well-defined mask edges. Mask materials that are more suited

for laser cutting (e.g., polyimide films) are also under

consideration. The use of mPLAT water molds already

promises more rapid, less expensive lab-on-a-chip device

prototyping than traditional methods.

Acknowledgements

We gratefully acknowledge the support of this research by the

NIH National Human Genome Research Institute, Centers of

Excellence in Genomic Science: Microscale Life Sciences

Center. We appreciate Professor R. Bruce Darling for

providing the machining laser, Dr John Koschwanez for

providing the SU-8 mold for pneumatic control channels and

Dr Barry Lutz for helpful discussions.

References

1 S. R. Quake and A. Scherer, Science, 2000, 290, 1536–1540.2 E. M. Liston, J. Adhesion, 1989, 30, 199–218.3 A. Mata, A. J. Fleischman and S. Roy, Biomed. Microdevices, 2005,

7, 281–293.4 R. Carlson, J. Koschwanez and D. Meldrum, microTAS, Transducer

Research Foundation, San Diego, CA, USA, 2005, pp. 672–674.5 B. A. Langowski and K. E. Uhrich, Langmuir, 2005, 21, 10509–10514.6 D. C. Duffy, J. C. McDonald, O. J. A. Schueller and

G. M. Whitesides, Anal. Chem., 1998, 70, 4974–4984.7 M. A. Unger, H. P. Chou, T. Thorsen, A. Scherer and S. R. Quake,

Science, 2000, 288, 113–116.8 H. Gau, S. Herminghaus, P. Lenz and R. Lipowsky, Science, 1999,

283, 46–49.

Fig. 3 (a) Manually cut mask from overhead transparency. (b) Close-

up view near the double-T section. (c) Resulting PDMS channels filled

with food coloring. (d) Close-up of the channels.

Fig. 4 (a) A pneumatic valve on a multilayer PDMS chip consisted of

a mPLAT-based liquid channel layer (top), made by the procedures in

Fig. 1 on a 30 mm-thick PDMS membrane, and a photolithography-

based pneumatic control layer (bottom, not to scale). (b, c) Cross-

section views of the two layers (not to scale). (d) The micrograph of the

channels when no pressure was applied on the control channel. The

liquid channel was filled with food coloring. (e) Air pressure was

applied using a syringe pump through the control channel, deflecting

the membrane and closing the liquid channel.

This journal is � The Royal Society of Chemistry 2007 Lab Chip, 2007, 7, 641–643 | 643

Publ

ishe

d on

05

Apr

il 20

07. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

7:02

:38.

View Article Online