In situ fabrication of macroporous polymer networks within microfluidicdevices by living radical photopolymerization and leaching

Helen M. Simms,a Christopher M. Brotherton,a Brian T. Good,a Robert H. Davis,a Kristi S. Ansethab andChristopher N. Bowman*ac

Received 16th August 2004, Accepted 9th November 2004

First published as an Advance Article on the web 8th December 2004

DOI: 10.1039/b412589d

Novel fabrication techniques and polymer systems are being explored to enable mass production

of low cost microfluidic devices. In this contribution we discuss a new fabrication scheme for

making microfluidic devices containing porous polymer components in situ. Contact lithography,

a living radical photopolymer (LRPP) system and salt leaching were used to fabricate multilayer

microfluidic devices rapidly with various channel geometries and covalently attached porous

polymer plugs made of various photopolymerizable substrates. LRPP systems offer the

advantages of covalent attachment of microfluidic device layers and facile surface modification

via grafting. Several applications of the porous plugs are also explored, including a static mixer, a

high surface area-to-volume reactor and a rapidly responding hydrogel valve. Quantitative and

qualitative data show an increase in mixing of a fluorescein and a water stream for channels

containing porous plugs relative to channels with no porous plugs. Confocal laser scanning

microscopy images demonstrate the ability to graft a functional material onto porous plug

surfaces. A reaction was carried out on the grafted pore surfaces, which resulted in fluorescent

labelling of the grafted material throughout the pores of the plug. Homogenous fluorescence

throughout the depth of the porous plug and along pore surfaces indicated that the porous plugs

were surface modified by grafting and that reactions can be carried out on the pore surfaces.

Finally, porous hydrogel valves were fabricated which swelled in response to contact with various

pH solutions. Results indicate that a porous hydrogel valve will swell and close more rapidly than

other valve geometries made with the same polymer formulation. The LRPP-salt leaching method

provides a means for rapidly incorporating porous polymer components into microfluidic devices,

which can be utilized for a variety of pertinent applications upon appropriate selection of porous

plug materials and surface treatments.

Introduction

The world of microfluidic devices and applications is rapidly

proliferating as improved technology enables fabrication of

smaller analytical devices that have benefits of reduced reagent

consumption and analysis time. Most commercially available

microfluidic devices are fabricated from glass using micro-

machining techniques. Recently, significant efforts have

focused on the development of polymeric substrates and

fabrication methods for making polymeric microfluidic

devices1–4 to provide for a less expensive alternative to device

fabrication with a wider range of material properties. Soft

lithography with poly(dimethylsiloxane) (PDMS) provides a

means of fabricating microfluidic devices and has recently been

explored by several research groups for developent of various

unit operations including mixing,1 reaction chambers,1 etc.

With soft lithography, microfluidic components may be sealed

to a flat substrate to create a device with enclosed channels,

or components may be treated, aligned and contacted to

other PDMS components to create multilayer devices with

interconnecting channels.

Most recently, a fabrication method utilizing living radical

photopolymerization (LRPP) has been developed.5 LRPP

involves photopolymerization of a monomer formulation that

includes a photoinitiator and a photoiniferter precursor. The

photoinitiator is responsible for the majority of the bulk

polymerization. The photoiniferter provides a source of

reinitiatable radicals on fully cured polymer surfaces,6–8

allowing for building of subsequent layers or surface mod-

ification with covalent atachment. The benefits of the LRPP

fabrication method are many and include the ability to form

covalently bonded, multilayer devices rapidly with three-

dimensional pathways and features and the ability to graft a

range of functional materials to surfaces.

The next step in enhancing the LRPP fabrication method is

to build functional components within devices. An in situ

formed, macroporous polymer plug within a microfluidic

channel, for example, could be used as a static mixer, to

provide a high surface area with specific functionality, or as a

fast acting hydrogel valve. Several groups have worked with

porous polymeric networks for use in microfluidic devices,

including Beebe9 and Svec and Frechet.10,11 These methods

require multiple hours to make porous polymer networks,

while pore size, pore connectivity and porosity are sensitive to*christopher.bowman@colorado.edu

PAPER www.rsc.org/loc | Lab on a Chip

This journal is � The Royal Society of Chemistry 2005 Lab Chip, 2005, 5, 151–157 | 151

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

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

crosslinking agent concentrations and temperature. In addi-

tion, several time-consuming steps are required to attach these

polymer networks to microfluidic channels. The LRPP

fabrication method, in combination with salt-leaching techni-

ques used in tissue engineering scaffolds,12 affords a rapid,

robust method for fabricating porous polymeric networks in

microfluidic devices. The purpose of this article is to describe

the LRPP-leaching method that we have utilized to create

macroporous polymeric plugs and to demonstrate potential

applications such as a static mixer, a high surface area reaction

site, and rapidly responding hydrogel valves for these plugs in

microfluidic devices.

Experimental

Materials

The standard formulation used for making devices and porous

polymer plugs was: 48.3 wt% Ebecryl 4827, an aromatic

urethane diacrylate (donated by UCB Chemicals), 48.3 wt%

triethylene glycol diacrylate (Polysciences), 1.4 wt% Irgacure

184 (photinitiator, Ciba), 1.0 wt% acrylic acid (Aldrich), and

1.0 wt% tetraethylthiuram disulfide (photoiniferter precursor,

Aldrich). Sodium chloride particles were ground and sieved to

known size ranges (¡45 mm, 45–53 mm, and 75–106 mm). All

other materials were used as received.

Porous polymer surfaces were modified with 2-aminoethyl

methacrylate (Aldrich) dissolved in methanol (Fisher

Scientific) and diluted in diethylene glycol (Sigma). Fluos

(Roche) was used to label the amino group of 2-aminoethyl

methacrylate with a green fluorescent molecule.

The acidic hydrogel formulation (HEMA/AA) was: 68.6 wt%

2-hydroxyethyl methacrylate (HEMA, Acros Organics),

29.4 wt% acrylic acid (AA, Aldrich), plus 1 wt% ethylene

glycol dimethacrylate (EGDMA, Aldrich) and 1 wt% Irgacure

184 (Ciba). The basic hydrogel formulation (HEMA/DEAEA)

was: 86 wt% HEMA, 12 wt% 2-(diethyleneamine) ethyl

acrylate (DEAEA, Aldrich), plus 1 wt% EGDMA and 1 wt%

Irgacure 184. For both hydrogel formulations: there were

70 mol% HEMA and 30 mol% of acidic or basic functional

groups with the same weight fraction of ethylene glycol

dimethacrylate for crosslinking and of Irgacure 184 for

photoinitiation.

General description of LRPP technique

The LRPP technique involves the use of a photopolymer

system (standard monomer formulation described above), a

collimated UV light source (365 nm wavelength), patterned

photomasks and a low melting temperature wax. Each layer of

a device is built by exposing the monomer mixture to UV light

through a patterned photomask. All exposed areas are cured,

leaving liquid monomer in the masked areas. As the monomer

formulation contains a photoiniferter precursor, each cured

layer generates radical sites upon re-exposure to UV light.

When a new layer of monomer is placed over a polymerized

layer of a microfluidic device and exposed to UV light, the

monomer forms covalent bonds with the polymerized layer

through these photoiniferter generated radical sites.

Additional layers may be applied by a transfer method, in

which a single layer is made, and then covalently bonded to a

device by sandwiching a thin layer of monomer solution

between the two and photo-curing through appropriate masks

(Fig. 1c). Entire devices, individual device layers, or separate

components are made with various photopolymerizable

materials to provide the required physical or chemical

properties. LRPP adds the ability to graft to the surface of

the polymer substrate (through photoiniferter radical sites),

creating spatially controlled, specific surface properties within

a device. This LRPP fabrication method has great potential for

Fig. 1 Illustration of microfluidic device and porous polymer plug fabrication process using the transfer method. (a) Salt/monomer mixture is

packed into a channel, covered with a photomask and exposed to collimated UV light. (b) Excess salt/monomer mixture has been removed and the

device is placed in a quiescent water bath for 1 h. (c) Transfer method: The device is covered with a thin layer of monomer mixture (cross-hatch

pattern), a polymerized lid layer (white square) and a photomask with the channel pattern, and then exposed to UV light. (d) Complete device after

channels were cleaned with low pressure air. Inset: Image of a mixer device with three Y-channels. The bottom two channels contain porous plugs

just after the Y-shaped channels merge. A scanning electron micrograph image of the porous plug is outlined in black.

152 | Lab Chip, 2005, 5, 151–157 This journal is � The Royal Society of Chemistry 2005

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

View Article Online

rapidly prototyping and mass producing relatively inexpensive

microfluidic devices.

Porous polymer network fabrication, LRPP-leaching method

The same LRPP monomer formulation used for making

devices was used to make porous polymeric networks, except

that sieved sodium chloride (salt) particles were incorporated

into the monomer mixture as the porogenic component.

Various ratios of salt particles and monomer formulation

were mixed, by stirring and agitating with a vortexer, to form a

paste-like mixture. The salt-monomer mixture was then

packed into a channel, covered with a photomask and cured

with 70 mW cm22, 365 nm, collimated light for 300 s (Fig. 1a).

Polymer was formed around salt particles in the exposed

region, leaving excess salt and unreacted monomer in the

masked regions of the channel. Low-pressure air was blown

into the open channels to remove the uncured salt–monomer

mixture. Finally, the salt was leached out of the polymer

composite by placing the device in a quiescent bath of

deionized water for one hour (Fig. 1b). Materials used and

generated in the leaching process are deionized water and salt,

which are non-toxic and inexpensive. The next layer of the

device was attached by the transfer method, with a photomask

to prevent curing in the porous network and channels of the

device (Fig. 1c).

Salt leaching experiments were conducted with the standard

LRPP monomer formulation and various amount of salt of

known size ranges. To make disks, the salt–monomer mixture

was packed into a 1 mm thick mold with a 4 mm diameter

circle. Glass slides were clamped to both sides of the mold and

the entire setup was exposed to 70 mW cm22, 365 nm,

collimated light for 300 s.

To make porous plugs in microfluidic channels, the same

salt–monomer mixture was packed into 400 mm 6 400 mm

channels at the appropriate channel segment. The salt mixture

and channels were covered with a patterned photomask,

masking everything except the porous plug area, and exposed

to 70 mW cm22, 365 nm, collimated light for 300 s.

Backpressure measurement

Backpressure was measured with a piezoelectric transducer

(Honeywell 142PC15) that was connected to an inlet stream to

the device. As pressure increased, deflection of the silicone

membrane in the transducer caused a change in voltage output

from the transducer. Change in voltage is linearly proportional

to pressure change and was calibrated with various column

heights of water.

Mixing efficiency

The porous plugs used as mixers were made with the standard

LRPP monomer formulation, with a salt size range of 75–

106 mm. The plug dimensions were 400 mm wide, 400 mm tall,

and 1 mm in length. The test channels consisted of two inlet

(400 mm 6 400 mm) channels that converged into a single

outlet (400 mm 6 400 mm) channel in a ‘‘Y’’ configuration. The

porous polymer plug was located at the convergence point.

One inlet stream consisted of a 0.3 mM solution of Fluorescein

(Sigma) in phosphate buffered saline (pH 7.5). The second

inlet stream consisted of only phosphate buffered saline

(pH 7.5). Phosphate buffered saline was used to minimize

any pH drift over the course of the experiment. The extent of

mixing was determined by measuring the distribution of the

fluorescent dye at the inlet and outlet of the porous polymer

plug using a photomultiplier tube system (PTI 814) and an

inverted microscope (Nikon TE300). The concentration of the

fluorescent dye one-third of the distance across the channel

width is subtracted from the concentration two-thirds of the

distance across the channel width. The differences at the inlet

and outlet are referred to as DI and DO respectively. Finally,

the extent of mixing (EOM) is determined using:

EOM~1{DO

DI(1)

For a perfectly mixed stream, DO would have a value of 0

and the extent of mixing would have a value of 1. For an

unmixed stream, the change across the inlet and the outlet

would be the same and therefore the EOM would be 0.

Results and discussion

Macroporous plug characterization

Three photopolymerizable monomer solutions were used to

build macroporous networks: the standard LRPP formulation

and two hydrogel formulations, as described in the Materials

section. Each polymer readily forms covalent bonds with the

channel walls during the LRPP process, assuring adhesion to

the wall. Salt particle sizes ranged from 45–53 mm to 75–

106 mm. After fully curing the monomer–salt mixture, removal

of salt via dissolution led to formation of pore stucture.

Because the polymer reacts as a separate phase around the salt

particles, pore size, shape, and volume are similar to that of the

salt particles incorporated into the initial monomer–salt

mixture (see Fig. 5 later). There were no solvency issues or

temperature dependencies on pore size or porosity, as salt is

not soluble to any significant extent in the monomer solutions.

With the proposed process, other porogens may be used to

create a specific pore shape or size, or to meet solubility

requirements of various leaching solvents.

To measure weight loss due to salt leaching, calculate

porosity, and test for pore interconnectivity, porous polymer

disks were made of the standard LRPP monomer formulation

as described in the Materials section and left in room

temperature water, on a shaker plate for 24 h. As verified by

visual inspection with confocal microscopy, continuous pores

were created with monomer–salt mixtures containing 60 to

80 wt% salt. Samples made with 90 wt% salt were too porous

and did not remain intact following leaching of the salt. As

seen in Table 1, all of the salt was not removed from networks

made with 60 wt% salt, indicating there are also dead end

pores filled with salt within the network. All of the salt was

leached from samples made with monomer–salt mixtures

containing 70 and 80 wt% salt, demonstrating good pore

interconnectivity. By visual inspection, none of the porous

polymer network was observed to be lost during the mass-loss

study; however, it is possible that a small amount was removed

in the solvent fraction after leaching. This possibility would

This journal is � The Royal Society of Chemistry 2005 Lab Chip, 2005, 5, 151–157 | 153

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

View Article Online

explain the slightly greater than 100% salt mass loss as

reported for several samples in Table 1.

When making porous plugs in channels, salt is leached out

before enclosing the channel with a top layer, leaving the top

side and two ends of the porous plug exposed to water for

faster leaching of salt. An additional weight loss study (data

not shown) indicated that complete leaching of salt takes

approximately one hour in a quiescent water bath for a

400 mm 6 400 mm 6 3 mm plug made with the standard

LRPP monomer formulation and 70 to 80 wt% salt.

Flow characterization

To test flow properties through the porous polymer plugs,

microfluidic devices were constructed that incorporated four

separate channels, each with one porous plug mid-length in the

channel. Two ranges of salt/pore diameters (45–53 mm and 75–

106 mm) and two plug lengths (3 mm and 4 mm) were

incorporated into each device. The channel cross-section

remained constant at 400 mm 6 400 mm. All plugs were made

with the standard LRPP monomer formulation using 80 wt%

salt in the initial monomer–salt mixture. Volumetric flow rates

were varied from 100 ml min21 to 1000 ml min21 with

corresponding Reynolds numbers from 2.2 to 22, based on

channel cross-section, indicating laminar flow for all tests.

Darcy Flow is characteristic of and often used to describe

low Reynolds number flow through porous media. The basic

equation describing Darcy flow relates pore size and flow rate

to a pressure change:13

DP~4QmL

kpd2(2)

where DP is the pressure change between inlet and exit sides of

the porous plug; Q is the volumetric flow rate; m is the fluid

viscosity; L is the length of the porous plug; k is the specific

permeability; and d is the characteristic cross-sectional

diameter through which fluid flows.

Pressure change was measured as a function of porous plug

length and pore diameter. On a plot of DP versus flow rate, the

slope of the line should be inversely proportional to the square

of the characteristic diameter, which is roughly estimated as

the middle range of the particle size distribution. As seen in

Fig. 2, the ratio of the slopes of trend lines for the smaller

pores to the larger pores is about 3.5, which is very close to the

calculated ratio, based on the midpoint of each size range, of

3.4. The effect of plug length (for same size pores) appears to

deviate from Darcy’s law, as a 33% increase in plug length gave

only an approximately 20% increase in pressure for the same

pore sizes at the same flow rates. This difference may be due to

end effects, as the ends of the porous plugs are irregular after

the salt particles are leached out. For each plug length, flow

rate and pore size, the pressure drop was measured and was

used to calculate the specific permeability of the plug from eqn.

(2). The average of the permeabilities calculated this way is 9 ¡

3 6 103 cm2 cps atm21. The average pore size was used to

calculate permeabilities, assuming a symmetric distribution of

particle sizes about the mean. It is also important to note that

the backpressure did not vary after cycling three times through

each range of flow rates (100, 200, 500 and 1000 ml min21),

allowing two or more minutes of flow through each plug at

each flow rate. This result indicates that the porous plugs

maintain their structure with relatively high flow rates. Flow

rates above 1000 ml min21 were not evaluated, based on the

assumption that most microfluidic applications will not

require such high flow rates.

Applications for porous polymer networks

Passive mixer

One area for improvement in microfluidic devices is mixing, as

mixing occurs by Fickian diffusion for the laminar flow regime

and, thus, requires increased channel length or decreased flow

rate to achieve increased mixing in a channel of fixed cross-

section.14 Several channel geometries have been explored to

enhance mixing within microfluidic channels.15–19 To reduce

mixing time and length, a microfluidic device was fabricated,

incorporating a porous polymer plug that acts as a static mixer

(Fig. 1 inset). The porous plugs are meant to increase the

effective mixing rate by splitting incoming streams multiple

times, to increase contact area between fluids and reduce the

necessary diffusion distance between fluids. The mixing device

was fabricated from the standard monomer formulation

described in the Materials section. Porous plugs were made

with the standard LRPP monomer formulation and 80 wt%

salt in the initial salt–monomer mixture. To visualize mixing in

the channels, two fluid streams were introduced into the

Y-shaped mixer channels; one stream contained fluorescein

Table 1 Salt weight loss (mean ¡ one standard deviation for 3–5repeats) and calculated porosity of porous plugs made with 80 wt%salta

% Salt removed (by wt)/calculated porosity (by wt)

Salt size 60 wt% salt 70 wt% salt 80 wt% salt

45–53 mm 79 ¡ 1/63 101 ¡ 2/77 102 ¡ 2/8575–106 mm 88 ¡ 2/66 102 ¡ 2/78 102 ¡ 2/84a 1 mm thick, 4 mm diameter disks were fabricated by packing thestandard LRPP monomer formulation and salt into a mold andexposing to 70 mW cm22, 365 nm light for 300 s.

Fig. 2 Measured backpressure as a function of flow rate for channels

filled with porous plugs of varying lengths and pore sizes. Squares

indicate 45–53 mm pores; circles indicate 75–106 mm pores; open shapes

indicate 3 mm plug length; filled shapes indicate 4 mm plug length.

Smaller pore size and longer plug length result in higher pressures for

the same flow rates.

154 | Lab Chip, 2005, 5, 151–157 This journal is � The Royal Society of Chemistry 2005

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

View Article Online

and the other deionized water. The inlet flow rates were

80 ml min21. Qualitative results (Fig. 3) indicate that porous

plugs do increase mixing, as shown by a more homogenous

distribution of fluorescein at the porous plug exit, relative to

the channel with no porous plug, but with the same

dimensions, fluid streams and flow rates.

Fig. 4 presents quantitative data for the extent of mixing at

two different flow rates, 20 ml min21 and 40 ml min21. The

results show that the porous polymer plugs significantly

enhance microfluidic mixing, with the extent of mixing

increasing more than five-fold over that in the control channels

without porous plugs. However, there is not a statistically

significant difference in mixing between the two flow rates

tested. Apparently, the reduced contact time at higher flow

rates is offset by more vigorous contact between the two

streams as they are forced through the porous medium. Future

experiments will study the effect of pore size distribution and

plug length on the extent of mixing. While the porous plug

causes a greater pressure drop than alternative microfluidic

mixers with less channel constrictions,15–19 the pressure drop is

only about 0.l psi for typical microfluidic flow rates (see Fig. 2).

Increased surface area and functionalized surface for enhanced

reactions

Porous polymer networks not only provide a means to mix

reagents quickly in a microfluidic channel but also increase the

surface-to-volume ratio for heterogeneous reactions.

Functional sites for reactions may be incorporated into the

bulk of the polymer network in several ways, one of which is

through copolymerization of a vinyl terminated functional

group within the standard formulation. Care should be taken

with this approach, as porous network properties such as

stiffness and strength will change as copolymer composition

changes. Alternatively, LRPP provides a facile means for

modifying porous polymer surfaces by grafting vinyl termi-

nated monomers with pendant functional groups via the

surface photoiniferter radical sites.20 Consumption of expen-

sive functional materials is reduced, and the number of

available reactive sites is dramatically increased by modifying

pore surfaces, instead of copolymerizing reactive materials

within the bulk of the polymer network. This technique allows

for completely independent control of the bulk and surface

composition properties.

To demonstrate this concept, 2-aminoethyl methacrylate

was grafted to the surfaces of a porous disk made from the

standard LRPP monomer formulation described in the

Materials section. Porous disks were coated with a solution

of 2-aminoethyl methacrylate, diethylene glycol and methanol,

and the solution was pulled through the porous disks by

vacuum. The porous disks were then exposed to UV light for 0

to 900 s. Samples were flushed with methanol to remove

diethylene glycol and unreacted 2-aminoethyl methacrylate,

then placed in a vacuum oven overnight to remove the

remaining methanol. To verify that 2-aminoethyl methacrylate

was grafted to the pore surfaces, Fluos (Roche) solution was

placed onto the porous disks and pulled through with vacuum.

Fluos is a fluoroprobe that contains a succinimidyl ester group

attached to fluorescein. The succinimidyl ester groups of Fluos

react with amino groups to form carboxamide linkages

between amino and fluorescein groups. Fluos was contacted

with the porous plugs for two hours to react with the grafted

amines, after which all samples were soaked in ethanol and

water for two days to remove unreacted Fluos. Porous disks

were then imaged with a laser scanning confocal microscope

(Zeiss, LSM5 Pascal, Axioplan 2 imaging, 488 nm excitation

wavelength, 505–530 nm emission wavelength). The control

sample was treated as described above, but was not exposed to

UV light to induce grafting. No fluorescence was detected for

the control sample at any depth of the porous disk.

Fluorescence was detected at all depths of the grafted porous

disks, which are 180 micron thick, and clearly showed pore

structures on samples that were exposed to UV light for more

than 400 s (Fig. 5). This result demonstrates that vinyl

Fig. 3 Fluorescein and water streams flowing through a porous plug mixing device. Channels are all 400 mm wide by 400 mm deep with 4 mm plug

length. The image on the left (a) shows inlet streams of water on top and fluorescein on bottom. Flow direction is from left to right. Each stream is

introduced into the Y at 80 ml min21. Images (b), (c), and (d) are of sections of the channels immediately downstream of the porous plug region. (b)

No porous plug. (c) Plug with 45–150 mm pores. (d) Plug with ¡45 mm pores. The plug with the smallest pores provided the best mixing.

Fig. 4 Extent of mixing without (striped bars) and with (solid bars)

porous plugs within microfluidic channels as a function of flow rate.

The porous plugs used for this experiment were 3 mm long, made with

80 wt% salt particles, 75–106 mm. There is a significant increase in

extent of mixing for channels with a porous polymer plug versus the

channels which did not have porous plugs. There was no significant

difference in extent of mixing between the two flow rates evaluated.

This journal is � The Royal Society of Chemistry 2005 Lab Chip, 2005, 5, 151–157 | 155

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

View Article Online

functionalized materials are readily grafted to pore surfaces,

and these reactions take place throughout the porous

structure.

Fast acting pH sensitive hydrogel valve

Hydrogels have been investigated for use in many microfluidic

applications, including valves, pumps, and drug release

mechanisms with various triggers such as pH,21,22 tempera-

ture23 and specific antigens.24 One of the main obstacles to

using hydrogels as valves is the slow response time associated

with diffusional limitations.21,22 If pores were to be introduced

within hydrogels, convective flow through the pores would

allow for increased initial contact between the hydrogel and

the solution, leading to a faster response. By combining the

LRPP-leaching technique with hydrogel substrates, we pro-

duced pH sensitive hydrogel valves that are covalently

attached to microfluidic channels and have pores ¡45 mm.

The acidic (HEMA/AA) hydrogel swells when in contact with

solutions of pH greater than 7, and the DEAEA hydrogel

swells when in contact with solutions of pH less than 10.

To test response times of the macroporous hydrogel valves,

a device was fabricated with similar channel geometries as used

in previous studies (Fig. 6).21 For one test valve, there are three

inlet channels; one is for deionized water, which directs flow

through the valve, and the other two are for various pH

solutions, which direct flow along one side of the valve. For

the macroporous HEMA/AA valve, as a pH 12 solution passes

alongside the valve, swelling occurs and the pores are

reversibly reduced in size to block flow of water through the

valve. A pressure transducer is connected to the water stream

and backpressure is recorded as the valve closes. The valve is

considered to be closed when the pressure vs. time plot

becomes linear. When a pH 2 solution is passed alongside the

closed valve, swelling is reduced, the pores reopen, and water

flows through the valve.

Previous work with similar hydrogel chemistries produced a

valve that closed within 12 s and withheld pressures up to

43.5 psi.21 As presented in Fig. 6, the macroporous HEMA/

AA hydrogel valve, made with 80 wt% salt, ¡45 mm, closed

within 5 s with 20 ml min21 of pH 12 solution and 20 ml min21

of deionized water and withstood backpressures of up to 2 psig

for up to 20 min over eight or more cycles. After each cycle, the

valve effectively blocked fluid flow; however, there were signs

of structural damage as the slope of the line, indicating

pressure increase with time, decreased with each run. This

apparatus tests the integrity of the valve by measuring

backpressure of a continuous steady flow rate of water,

resulting in steadily increasing pressure when the valve is

closed. Depending on the use of this type of macroporous

hydrogel valve, the irreversible damage may not be a problem.

For instance, if the valve is used to divert fluid to another

channel, there should not be a large pressure buildup to

damage the valve. The macroporous DEAEA hydrogel valve

swelled as rapidly as the HEMA/AA valve; however, the

DEAEA valve did not maintain its physical structure with a

flow rate of 20 ml min21 of deionized water, and no significant

backpressure was generated. Further polymer formulation

work is required to produce a physically stable acid swelling

hydrogel valve.

By adjusting swelling rates through chemistry and/or

porosity to provide the appropriate swelling response times,

the acid and base swelling hydrogels could also be used as a

pH switch. This switch could be used downstream of a

reaction to control where product is directed based upon pH,

or to restrict flow of harmful streams from channels containing

cells.

Fig. 5 High surface area-to-volume reactor. Laser scanning confocal

microscopy image of a porous polymer disk (z-stack of 5.5 mm thick

optical sections of a 140 mm thick porous plug), with 2-aminoethyl

methacrylate grafted onto pore surfaces. Fluos reacted with the

primary amines of 2-aminoethyl methacrylate to attach green

fluorescent molecules to the pore surfaces. Scale bar represents 100 mm.

Fig. 6 Porous hydrogel valve device image and hydrogel valve closing

data. (a) The hydrogel valve device was used to measure water

backpressure as a function of time with varying pH solutions. Porous

hydrogels were rectangular shaped (400 mm by 200 mm by 200 mm),

with covalent attachment to three of the channel walls. All channels

were 200 mm wide and deep. Various pH buffer solutions entered the

device through feed channels 1 or 2, and permitted fluid flow along

side the hydrogel valve. As the valve closed, water could no longer pass

through the valve and backpressure increased as measured by a

pressure transducer connected to the water inlet stream. (b)

Backpressure generated by a HEMA/AA valve was measured over

time while feeding water at 20 ml min21 and pH 12 buffer solution at

20 ml min21. The valve closed rapidly as indicated by the initial

increase in backpressure, followed by a linear increase in pressure with

time.

156 | Lab Chip, 2005, 5, 151–157 This journal is � The Royal Society of Chemistry 2005

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

View Article Online

Conclusions

Porous polymer networks were fabricated in microfluidic

channels using LRPP and salt leaching. Pore size and porosity

are readily controlled by varying the size and amount of salt

mixed with the monomer formulation. LRPP provides a rapid

means to fabricate covalently attached porous polymer

networks into polymeric microfluidic devices, to graft vinyl-

functionalized materials to pore surfaces, and to incorporate a

myriad of photopolymerizable materials as porous networks

or other components within a device. These porous polymer

networks are integrated as components in microfluidic devices

that act as static mixers, as high surface-to-volume reactors,

and as rapidly responding hydrogel valves.

Acknowledgements

This work was supported by the SIMBIOSYS Program of the

Defense Advanced Research Projects Agency through the Air

Force office of scientific research, and by graduate fellowships

from the Department of Education, Sandia National

Laboratories, and the Teets family. Special thanks are

extended to Bobby Sebra for his advice relating to grafting

experiments and Melissa Mahoney for lessons on use of the

confocal laser scanning microscope.

Helen M. Simms,a Christopher M. Brotherton,a Brian T. Good,a

Robert H. Davis,a Kristi S. Ansethab and Christopher N. Bowman*ac

aDepartment of Chemical and Biological Engineering, ECCH 111,CB424, University of Colorado, Boulder, CO 80309-0424, USA.E-mail: christopher.bowman@colorado.edubHoward Hughes Medical Institute, University of Colorado, Boulder,CO 80309-0424, USAcUniversity of Colorado Health Sciences Center, Department ofRestorative Dentistry, Biomaterials Research Center, 4200 E. 9th Ave.,Denver, CO 80262, USA

References

1 J. M. K. Ng, I. Gitlin, A. D. Stroock and G. M. Whitesides,Electrophoresis, 2002, 23, 3461.

2 D. Qin, Y. Xia, J. A. Rogers, R. J. Jackman, X. Zhao andG. M. Whitesides, Top. Curr. Chemistry, 1998, 194, 2.

3 J. Lee, C. Park and G. M. Whitesides, Anal. Chem., 2003, 23, 6544.4 H. Becker and L. E. Locascio, Talanta, 2002, 56, 267.5 J. B. Hutchison, K. T. Haraldsson, B. T. Good, R. P. Sebra,

N. Luo, K. S. Anseth and C. N. Bowman, Lab Chip, 2004, 4, 658.6 T. Otsu, J. Polym. Sci., Part A: Polym. Chem., 2000, 38, 2121.7 Y. Nakayama and T. Matsuda, Langmuir, 1999, 15, 5560.8 Y. Nakayama and T. Matsuda, Macromolecules, 1999, 32, 5405.9 J. Moorthy and D. J. Beebe, Lab Chip, 2003, 3, 62.

10 T. Rohr, C. Yu, M. H. Davey, F. Svec and J. M. J. Frechet,Electrophoresis, 2001, 22, 3959.

11 C. Yu, F. Svec and J. Frechet, Electrophoresis, 2000, 21, 120.12 J. A. Burdick, D. Frankel, W. S. Dernell and K. S. Anseth,

Biomaterials, 2003, 24, 1613.13 C. J. Geankoplis, Transport Processes and Unit Operations,

Prentice Hall PTR, Englewood Cliffs, New Jersey 07632, 3rdedn., 1993.

14 R. A. Vijayendran, K. M. Motesgood, D. J. Beebe andD. E. Leckband, Langmuir, 2003, 19, 1824.

15 C. Neils, Z. Tyree, B. Finlayon and A. Folch, Lab Chip, 2004, 4,342.

16 A. D. Strook and G. M. Whitesides, Am. Chem. Soc., 2003, 36,597.

17 D. Gobby, P. Angeli and A. Gavriilidis, J. Micromech. Microeng.,2001, 11, 126.

18 J. Melin, G. Gimenez, N. Roxhead, W. van der Wijngaart andG. Stemme, Lab Chip, 2004, 4, 214.

19 M. A. Holden, S. Kumar, E. T. Castellana, A. Besock andP. S. Cremer, Sens. Actuators B, 2003, 92, 199.

20 N. Luo, J. B. Hutchison, K. S. Anseth and C. N. Bowman,J. Polym. Sci., Part A: Polym. Chem., 2002, 40, 1885.

21 R. H. Liu, Q. Yu and D. J. Beebe, J. Microelectromech. Syst., 2002,11, 45.

22 S. K. De, N. R. Aluru, B. Johnson, W. C. Crone and D. J. Beebe,J. Microelectromech. Syst., 2002, 11, 544.

23 M. M. Ozmen and O. Okay, Eur. Polym. J., 2003, 39, 877.24 T. Miyata, N. Asami and T. Uragami, Nature, 1999, 399, 766.

This journal is � The Royal Society of Chemistry 2005 Lab Chip, 2005, 5, 151–157 | 157

Publ

ishe

d on

08

Dec

embe

r 20

04. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

28/1

0/20

14 1

8:37

:02.

View Article Online