PAPER
www.rsc.org/loc | Lab on a Chip
Direct patterning of composite biocompatible microstructures using
microfluidics{
Yuk Kee Cheung, Brian M. Gillette, Ming Zhong, Sharmilee Ramcharan and Samuel K. Sia*
Received 19th January 2007, Accepted 20th March 2007
First published as an Advance Article on the web 10th April 2007
DOI: 10.1039/b700869d
This study demonstrates a versatile and fast method for patterning three-dimensional (3D)
monolithic microstructures made of multiple (up to 24 demonstrated) types of materials, all
spatially aligned, inside a microchannel. This technique uses confocal scanning or conventional
fluorescence microscopy to polymerize selected regions of a photocurable material, and
microfluidics to automate the delivery of a series of washes and photocurable reagents. Upon
completion of lithographic cycles, the aligned 3D microstructures are suitable for microfluidic
manipulation and analysis. We demonstrated the fabrication of composite 3D microstructures
with various geometries, size scales (up to 1 mm2), spatial resolution (down to 3 mm), and
materials. For a typical multi-cycle process, the total fabrication time was tens of minutes,
compared to tens of hours for conventional methods. In the case of 3D hydrogels, a potential use
is the direct patterning of inhomogeneous 3D microenvironments for studying cell behavior.
Introduction
Monolithic microstructures made of a single type of material
(such as silicon, photoresist, or PDMS) have proven widely
useful in the development of microfabricated systems for
materials science and biology.1,2 Mask-based or direct-write
photolithography are well suited for fabricating such microstructures.3,4 Composite 3D microstructures, made of multiple
types of materials, are useful for a variety of applications,
including nanoparticle composite materials,5 photonic
devices,6,7 microelectromechanical systems,1,8 and cell biology
(to mimic the native extracellular environment).9–17 Spatial
control over composite microstructures of regular patterns can
be achieved by templating18 or self-assembly.19 For arbitrary
patterns, conventional top-down microfabrication typically
requires manual and time-consuming steps (over 1 h for
alignment, exposure, and baking steps) between lithographic
cycles to produce correctly aligned composite structures.
Microfluidics offers a simple and powerful method to deliver
different photocurable materials for making composite 3D
microstructures. It also allows for efficient washing, and
requires small amounts of reagents. During each lithographic
cycle, we aimed to fabricate a microstructure that would
adhere to the surface of the channel, and wash away the
unpolymerized pre-polymer; since the sample does not need to
be removed from the stage in between lithographic cycles,
time-consuming steps of alignment and re-registration are
circumvented. Previous work on direct-write lithography
(without microfluidics) demonstrated the patterning of two
different molecules on a surface,20–22 and studies that
combined microfluidics and lithography demonstrated the
Columbia University, Department of Biomedical Engineering, 351
Engineering Terrace, 1210 Amsterdam Ave., New York, NY, 10027.
E-mail: [email protected]
{ Electronic supplementary information (ESI) available: Additional
data. See DOI: 10.1039/b700869d
574 | Lab Chip, 2007, 7, 574–579
synthesis of structures made of one or two types of
material.23–25 We hypothesized the use of microfluidics in
multi-step lithography could potentially achieve the fabrication and patterning of 3D microstructures of many materials
in a timely and automated fashion, and eliminate the need of
using expensive specialized equipment that would otherwise be
required in an industrial setting.
Results and discussion
Our microfluidic device consisted of a chamber made of
poly(dimethylsiloxane) (PDMS) with cover glass at the top
and bottom (Fig. 1a); this configuration limited an influx of
oxygen which would inhibit radical-induced photopolymerization. For acrylate-based hydrogels, we pre-treated the cover
glass with 3-(trimethoxysilyl) propyl methacrylate to promote
adhesion of the polymerized structures.26 We then placed the
microfluidic device on the stage of an inverted fluorescence
microscope. Initially, we demonstrated the patterning of
monolithic 3D structures made of poly(ethylene glycol)diacrylate (PEG-DA); PEG-DA is a well studied photopolymerizable substrate for encapsulating cells,20,23 and exhibits a
sufficiently low viscosity (,35 cP) to allow for reasonable flow
rates (1 mL min21) using a syringe.23 After scanning the
desired region, we washed away unpolymerized material and
injected a different photocurable material for the subsequent
cycle. We fabricated microstructures of different thicknesses
by varying the height of the microchannel (data not shown).
The height limitation of our system agreed with previously
published results,24 which is mainly determined by the
‘thickness’ of the focusing light arriving at the channel (i.e.
Depth-of-field of the objective lens used). To maintain high
precision and fidelity among the microstructures, we adjusted
the height of the microchannel to be comparable to the range
of DOF, so that any polymerization by out-of-focus light
could be minimized. For attaining reliable and fast fabrication,
This journal is ß The Royal Society of Chemistry 2007
Fig. 1 Microfluidic device used in fabrication of 3D composite
microstructures, and an example of a 3D microstructure made from
24-cycle fabrication. (a) A schematic diagram of the experimental
setup. A microfluidic device was placed onto the stage of an inverted
fluorescence microscope (shown is the setup for a confocal microscope;
for an epifluorescence microscope, an automated x-y stage is used
instead of an x-y scanning mirror). Photocurable reagents or washes
(with PBS or ethanol) were delivered into a chamber by injection into
the inlet. Each ‘‘lithographic cycle’’ consisted of exposure (1 to 2 s),
washing of unpolymerized pre-polymer (1 min), injection of prepolymer for subsequent cycle (12 s), and alignment (less than 1 min,
including the time needed for manual fine adjustment for a
conventional fluorescence microscope). During each lithographic
cycle, the same microscope, under brightfield mode, can be used to
visualize the spatial fidelity of the fabrication. (b) Schematic patterns
and corresponding brightfield images of microstructures fabricated
after 6, 16, and 24 cycles; the respective total fabrication times were 12,
30, and 45 min. The polymerized microstructures could be visualized
during each lithographic cycle because they exhibited a change of index
of refraction. The polymer consisted of 40% w/v of PEG-DA, with
2% w/v of Irgacure 2959 as photoinitiator. For clarity of visualization,
PEG-DA in each cycle was doped with a combination of the
following fluorescent dyes: fluorescein (green), rhodamine B (red),
and 7-(diethylamino) coumarin-3-carboxylic acid (blue) (see Table I in
ESI{). Schematic diagrams of designed exposure patterns for all cycles
are provided in ESI Fig. I.{ (c) Merged fluorescence image (red, green,
and blue channels) of the final 3D composite microstructure after
24 cycles of fabrication. All scale bars = 200 mm.
This journal is ß The Royal Society of Chemistry 2007
we identified three important technical parameters: (1) flow
rate of less than 1.5 mL min21 for at least 1 mL of washing
solution, for sufficient washing of unpolymerized pre-polymer
without shearing off the polymerized structures; (2) high
concentration of photoinitiator, in order to minimize polymerization time, decrease diffusion of radicals, and increase
the spatial fidelity and resolution; (3) correct focusing of the
objective lens within the chamber, facilitated by introducing a
fluorescent dye into the chamber before the fabrication.
Factors that would increase the total time of fabrication
included: low percentage of prepolymer (which would increase
the exposure time), large scanning area, and large number
of cycles.
Using this setup, we fabricated a 3D microstructure, with
high spatial control, made of 24 cycles of materials in 45 min
(Fig. 1b,c). For clarity of visualization of the final structure,
we doped the PEG-DA in each cycle with a different combination of fluorescent dyes (see Table I in ESI{). During the
fabrication, we verified the x-y spatial fidelity by brightfield
microscopy (Fig. 1b). We observed high spatial fidelity and
fast fabrication over all cycles, with no deterioration in the
24th cycle compared to the 1st. (For fabrication beyond
24 cycles, the total time of fabrication would scale linearly
with the number of cycles.) As expected for hydrogels, some
diffusion of fluorescent molecules occurred between materials
fabricated from different cycles (Fig. 1c).
We fabricated a variety of composite 3D microstructures of
various interlinking patterns and spatial resolution using a
confocal microscope. Confocal microscopy offered highly
calibrated spatial positioning by region-of-interest scanning.20
First, we fabricated (via two cycles) an interlinking set of gears,
and verified the intended three-dimensionality of the microstructures by confocal microscopy followed by 3D reconstruction (Fig. 2a). We also fabricated (via three cycles) an array of
closely interlinking parallelogram structures (Fig. 2b). For this
fabrication, in which the surface area (1 mm2) was large by the
standard of microfabrication, the total fabrication time was
24 min. In a third pattern, we fabricated (via three cycles) a
chain-like pattern (Fig. 2c). Overall, using the confocal
microscope, we could achieve a spatial resolution of less than
3 mm (ESI Fig. III{), which is slightly larger than the predicted
resolution of 2 mm from the focal volume. We believe this
difference is due to light scattering, and possibly the slight
residual flow of the prepolymer during the exposure (which is
on the order of seconds).
We also fabricated composite microstructures made of
materials other than PEG-DA doped with fluorescent molecules. First, using a conventional epifluorescence microscope,
we fabricated microstructures made from three other acrylatebased polymers: trimethylolpropane triacrylate (TMP-TA),
trimethylolpropane ethoxylate triacrylate (TMpeg-TA), and
1,6-hexanediol diacrylate (HEX-DA). This range of materials
allows for the fabrication, within the same microstructure, of
islands with different chemical compositions, porousities, and
stiffnesses. For example, we measured the Young’s moduli of
these materials (under the experimental conditions of this
study) to range from 10 kPa (comparable to that of adipose
or muscle tissue27) to 200 kPa (comparable to that of
cartilage28). We fabricated (via 3 cycles) aligned 3D monolithic
Lab Chip, 2007, 7, 574–579 | 575
Fig. 2 Fabrication of 3D composite microstructures of different
geometries, sizes, and spatial resolution. All microstructures comprised
40% w/v PEG-DA, with 200 mM Eosin Y as photoinitiator,
polymerized using an argon laser on a confocal laser scanning
microscope. (a) Three dimension-reconstructed images from confocal
microscopy of a 3D gear design fabricated from PEG-DA prepolymer
doped with fluorescein (green) and rhodamine B (red). Total
fabrication time was 9 min. (b) Merged fluorescence image (red, blue,
and green channels) of a 12 by 8 array of parallelograms, fabricated
via 3 lithographic cycles, over a large surface area (1 mm2). Total
fabrication time was 24 min. (c) Merged fluorescence image (pseudo
colored in blue, purple, and magenta for PEG-DA doped with
rhodamine B, 7-(diethylamino) coumarin-3-carboxylic acid, and
fluorescein, respectively) of a chain-like pattern of microstructures,
fabricated via 3 cycles. Total fabrication time was 15 min. Schematic
diagrams of designed exposure patterns for all cycles are provided in
ESI Fig. I.{ All scale bars = 50 mm.
microstructures made of TMP-TA, TMpeg-TA, and HEX-DA
(Fig. 3a). We also fabricated a composite microstructure made
of two sets of overlapping PEG-DA disks in an alternating
geometry, with one set doped with 5 mm-diameter latex beads;
these microbeads, which would be expected to be visible along
the sides of the 3D structures, allowed us to investigate
further the spatial fidelity of the fabrication by scanning
electron microscopy (SEM). SEM images collected at different
magnifications showed clean edges with vertical sidewalls
(Fig. 3b,c), with the expected configuration of disks with and
without the microbeads (Fig. 3c).
The ability to spatially pattern islands of different material
properties may be useful for replicating the anisotropy and
inhomogeneity of biological 3D microenvironments, which
play an important role in the behavior of cells.9,29,30 We first
confirmed the viability of 3T3 fibroblasts in PEG-DA
polymerized by long wavelength-UV light; PEG-DA has been
widely used to encapsulate cells in 3D microenvironments,31,32
and it can be chemically coupled with adhesion and other
peptides.20 Using a LIVE/DEAD assay, we observed a
viability of over 95% of encapsulated cells after photopolymerization (ESI Fig. II{). We fabricated a 3D composite
microstructure of fibroblast-encapsulated PEG-DA within a
network of acellular hydogel (Fig. 4), in a geometry and size
scale that mimics those of a simple tissue. The final patterned
3D composite microstructures can be accessed easily by
removing the PDMS (e.g. for microscale tissue engineering29),
or left inside the microchannel for further microfluidic
interrogation and analysis. The feasibility of studying cell–cell
576 | Lab Chip, 2007, 7, 574–579
Fig. 3 Fabrication and characterization of microstructures made
of materials with different chemical and bead compositions. The
microstructures in this figure were fabricated with a conventional
epifluorescence microscope with a motorized stage. To match the
emission spectrum of the mercury lamp, we used Irgacure 2959 (a UVsensitive photoinitiator) to carry out photopolymerization at 365 nm.
The objective lens (406, 1.25 N.A.) projected the UV light through a
small aperture diaphragm (20 mm in diameter) onto the photocurable
material, inducing rapid radical-induced photopolymerization
(y2 seconds). (a) Merged fluorescence image (red and blue) of
microstructures of three different materials: TMP-TA (red disks),
TMPeg-TA (blue disks), and HEX-DA (red rectangles). The different
shapes of the microstructures were achieved with different aperture
diaphragms. Total fabrication time was 8 min. (b) Scanning electron
micrograph (SEM) at 2106 magnification of a disk-based microstructure that alternates with disks made of PEG-DA and PEG-DA
doped with 5 mm-diameter microbeads. Total fabrication time was
5 min. Insets: (left upper) schematic pattern of the intended design;
(lower right) fluorescence image of 0.5 mm-diameter fluorescent
microbeads inside a PEG-DA microstructure. (c) SEM at 3506
magnification of dotted region in panel (b). Because the microstructures are sandwiched flush between two glass plates, microbeads are
expected to be observed only on the sides of the microstructures, as
observed. Schematic diagrams of designed exposure patterns for all
cycles are provided in ESI Fig. I.{ All scale bars = 50 mm.
communication in the 3D environments2,19,33 of this system is
supported by the observation of diffusion of small molecules
(fluorophores) between materials made in different cycles.
Disadvantages of the technique in this paper include: (1)
difficulty of fabricating thick structures (greater than hundreds
of microns) with high spatial fidelity due to light scattering; (2)
potentially long total time of fabrication for large structures
(several mm) of many different types of materials (hundreds),
since this method is serial rather than parallel; (3) limitations in
choice of materials, which ideally share compatible sets of
solvents (to avoid swelling and collapsing of polymerized
structures), and exhibit sufficiently low viscosity (less than tens
of cP) for reliable microfluidic flow.
Conclusion
Overall, we demonstrated a versatile method that uses
standard microfluidics and fluorescence microscopy to fabricate aligned, composite 3D microstructures made of a large
number of materials more quickly than conventional fabrication methods (e.g. tens of minutes rather than tens of hours for
This journal is ß The Royal Society of Chemistry 2007
4.5 g L21 glucose, supplemented with 10% bovine calf serum
and 1% penicillin/streptomycin, and used at 80% confluence.
To fluorescently label the cells, 2 mL of pre-warmed 5 mM
CellTracker CMRA dye (Orange) was added to a culture flask
and incubated for 30 min. The dye solution was replaced with
fresh, pre-warmed medium and the cells were incubated for
another 30 min. The medium was discarded and the cells were
rinsed a few times with PBS. Fluorescently labeled cells were
trypsinized and re-suspended in fresh medium.
Assembly of microfluidic device
Fig. 4 Fluorescence image of mammalian cells inside a 3D composite
microstructure. The patterned microstructure consists of PEG-DA
that encapsulated 3T3 fibroblasts (labeled with CellTracker orange)
interlinked with a vessel-like acellular hydrogel network (PEG-DA
labeled with rhodamine B). For cell patterning in 3D hydrogels, the
fibroblasts were embedded in a PEG-DA pre-polymer with Eosin Y as
photoinitiator, and photopolymerized at 514 nm. The less-orderly
design of this cellular hydrogel microstructure was intended to mimic
in vivo-like structures. Scale bar = 50 mm.
24 materials). In one important application, biocompatibility
of the acrylate-based hydrogels and the spatial resolution of
the method (below 3 mm) render this technique potentially
useful for studying cell behavior in inhomogeneous and
anisotropic 3D environments that mimic the complexity of
biological microenvironments.
Experimental
Materials
Poly(ethylene glycol)-diacrylate (PEGDA, Mn 400), 1,6Hexanediol
diacrylate
(HEX-DA),
and
Polybead1
Microspheres (0.50 mm) were purchased from Polysciences
(Warrington, PA). Trimethylolpropane Triacrylate (TMPTA),
Trimethylolpropane Ethoxylate Triacrylate (TMpegTA),
Triethanolamine, 3-(trimethoxysilyl) propyl methacrylate,
1-vinyl-2-pyrrolidone (NVP), Rhodamine B, Fluorescein,
Hoechst 33342, Eosin Y, and TRIZMA1 acetate were purchased from Sigma (Milwaukee, WI). Swiss 3T3-L1 fibroblasts
(CL-173) and Dulbecco’s Modified Eagle’s Medium (DMEM)
were purchased from American Type Culture Collection
(Manassas, VA). Other cell culture and cell staining reagents
were purchased from Invitrogen (Carlsbad, CA), unless stated
otherwise. Polydimethylsiloxane (PDMS) is purchased from
Dow Corning (Midland, MI). High intensity UV lamp (Model
B-100A) was purchased from UVP, Inc. (Upland, CA). All
reagents were used without further purification.
Cell culture
Fibroblasts were passaged under 5% CO2 at 37 uC in DMEM
adjusted to 4 mM L-glutamine, 1.5 g L21 sodium bicarbonate,
This journal is ß The Royal Society of Chemistry 2007
A PDMS-based microfluidic device will allow us to fabricate,
wash, and examine the photopolymerizable material under the
microscope without removing the sample from the stage, and
hence eliminating the need for repositioning and re-focusing
the sample between steps of infusion of different types of
photocurable materials. The design was a square shape well
(5 mm 6 5 mm 6 165 mm) with inlet and outlet to allow
passage of materials (Fig. 1a). The assembly steps of the device
are as follows: first, we designed a transparency mask containing the channel features in CleWin (WieWeb Software, the
Netherlands) and printed it at CADArt Services, Inc. (Bandon,
Oregon). Then, using traditional photolithography, a silicon
master with square shape protrusions (5 mm 6 5 mm 6
165 mm), inlet, and outlet channels was fabricated. PDMS
base and curing agent were mixed in the ratio of 10 : 1,
degassed, and cured on the silicon master at 60 uC for 2 h. The
cured PDMS channel mold was cut out and a piece of number
1 or 0 cover glass (4 mm 6 4 mm) was bonded to the top of the
square chamber. The whole device was then plasma treated
and permanently bonded to another piece of larger number 1
coverglass (24 6 40 cm2). Two pieces of 0.38 mm (I.D.)
polyethylene tubing (Intramedic) were inserted at the inlet and
outlet. A needle syringe was connected to the inlet to deliver
solution into the microchannel. In order to promote adhesion
of the fabricated structures on the substrate, we incubated the
microchannel with 3-(trimethoxysilyl) propyl methacrylate for
15 minutes, as described in previous studies.16 After several
washes with 100% ethanol, the entire device was heated on a
hot pate at 110 uC for another 15 min. This leaves free
methacrylate groups on the glass surface, which link covalently
to acrylate-based polymer upon polymerization. In the case
where removal of the PDMS chamber after fabrication is
necessary, we omitted the plasma treatment step of PDMS to
prevent permanent bonding to the underlying glass slide. Also,
only the glass at the bottom of the chamber was treated with
3-(trimethoxysilyl) propyl methacrylate, so that the fabricated
structures would preferentially stick to the bottom substrate
but not to the top of the chamber. These extra steps ensure the
fabricated structures were undisrupted.
Preparation of prepolymer
To demonstrate the compositeness of our system, we used
multiple acrylate-based polymers along with different combinations of fluorescent dyes/beads to fabricate the microstructures. ESI Fig. I and Table I summarize the patterns and
compositions for each cycle.{ All pre-polymers were prepared
and shielded from light until use.
Lab Chip, 2007, 7, 574–579 | 577
Fabrication by confocal laser scanning microscope
The confocal laser scanning microscope we used to pattern the
microstructures in Fig. 2 and 4 is a Zeiss LSM 510 META.
Details on using the microscope for patterning is described
elsewhere.21 Briefly, a pre-defined pattern was drawn as a
region-of-interest in the LSM 5 software. Upon introducing
pre-polymer solution into the microfluidic chamber, a 514 nm
argon laser beam (6.25 mW average power output) was
focused within the chamber and scanned through the regionof-interest at 3.2–12.8 ms per pixel and repeated scanning for
10–15 times. Using a 10 6 0.5 NA objective, the scanning
resolution we used was 512 6 512 pixels, which spanned an
area close to 1 6 1 mm2. Polymerized structures could be
observed through the eye-piece under transmitted bright field.
The chamber was flushed with washes of PBS and then a new
prepolymer was introduced for the next cycle of fabrication.
The final composite microstrucure was imaged directly from
the confocal microscope or conventional fluorescent microscope through the bottom glass of the microfluidic chamber.
Fabrication by conventional fluorescent microscope
A conventional fluorescent microscope can also be used to
fabricate microstructures. In this study, we used a Leica CTR
6000 inverted fluorescent microscope (Leica Microsystems, IL)
equipped with an automated motorized-stage and InVitro
acquisition software (MediaCybernetics, MD) to fabricate the
microstructures in Fig. 1 and 3. Under this system, the shape
of the microstructure in each cycle is defined by the shape of
the aperture diaphragm. The resolution becomes the dimension of the aperture projected onto the focal plane. To define a
pattern, exposure positions were pre-programmed in the
acquisition software using the ‘‘User defined points’’ function.
Exposures were carried out with a Mercury arc lamp at 365 nm
for 1 s. After flushing out the unpolymerized materials with
Trizma buffer or ethanol, we moved to the next exposure point
and repeated the exposure step until all the exposure points
were exposed. The final structures were examined under the
same microscope.
Scanning electron microscopy (SEM) analysis
PEG-DA microstructures (Fig. 3b) were fabricated as
described above, removed from the microfluidic device,
sputter-coated with gold-palladium to a thickness of 3–5 nm,
and then imaged using a Hitachi 4700 SEM at 2–5 kV emission
accelerating voltage.
Mechanical testing
To measure the range of stiffnesses of the acrylate-based
materials under the experimental conditions of the study, we
performed an unconfined compression test using an Instron
5848 Microtester equipped with a 10N load cell and ceramic
platens. Hydrogel discs (Diameter: 9 mm, thickness: y1 mm)
made of 10%, 25%, and 50% w/v polymer were cast in a PDMS
mold sandwiched between glass slides and photopolymerized
under UV lamp for 1 min. The samples were stored in ethanol
for two days prior to testing. Using the integrated Instron
software, the platens were brought into contact with the
578 | Lab Chip, 2007, 7, 574–579
sample until a 10 gf (0.098 N) preload was reached and then
held for 5 min to obtain the equilibrium preload value. For
more compliant gels (i.e. 10% w/v), a 5 gf preload was used.
The height under preload was determined for each sample
from the relative position of the platens. For stiffer gels (25%
and 50% w/v), a 2% strain was applied at a rate of 0.001 mm s21
and held for 2 h to obtain the equilibrium load value.
For 10% w/v gels a 15% strain was used. The Young’s modulus
was calculated by dividing the difference between the
preload and equilibrium load by the sample area times the
percent strain.
Young0 s Modulus~
ðEq: Load{ Pr eload Þ=Area
% Strain
(1)
Cell viability assay
Fibroblasts were suspended in 10% w/v PEG-DA, 0.1% w/v
Irgacure 2959, 0.3% v/v 1-vinyl-2-pyrrolidone at a cell
density of 1 6 106 cells mL21. Cell-prepolymer mixture was
polymerized under UV light at 365 nm and then covered with a
thin layer of medium. The survival rate of fibroblasts was
quantified by LIVE/DEAD cytotoxicity kit immediately after
polymerization. Fluorescent images were taken at five different
locations across the surface of the gel and the number of viable
cells was counted (ESI Fig. II{). Percentage of viable cells was
determined with respect to the number of dead cells counted
from the same sample.
Acknowledgements
We thank the Columbia Optical Microscope facility for
providing help on the confocal microscope, Nancy Diao for
helping with the microfluidics work, Curtis Chin and Richard
Harniman for helping with the SEM analysis, Lux Lu and
Van Mow for helping with the stiffness measurements, and
Vincent Linder, Lance Kam, and members of the Sia lab for
critical reading of the manuscript. This work was supported by
a Scientist Development Grant (to S.K.S) from the American
Heart Association, and a Croucher Foundation Scholarship
(to Y.K.C)
References
1 M. J. Madou, Fundamentals of microfabrication, CRC, Boca
Raton, Florida, 2002.
2 S. K. Sia and G. M. Whitesides, Electrophoresis, 2003, 24,
3563–3576.
3 Y. N. Xia and G. M. Whitesides, Angew. Chem., Int. Ed., 1998, 37,
551–575.
4 B. D. Gates, Q. B. Xu, M. Stewart, D. Ryan, C. G. Willson and
G. M. Whitesides, Chem. Rev., 2005, 105, 1171–1196.
5 A. C. Balazs, T. Emrick and T. P. Russell, Science, 2006, 314,
1107–1110.
6 F. Di Benedetto, A. Biasco, R. Bizzarri, D. Arosio, F. Ricci,
F. Beltram, R. Cingolani and D. Pisignano, Langmuir, 2006, 22,
29–31.
7 M. Lei, Y. D. Gu, A. Baldi, R. A. Siegel and B. Ziaie, Langmuir,
2004, 20, 8947–8951.
8 A. C. R. Grayson, R. S. Shawgo, A. M. Johnson, N. T. Flynn,
Y. Li, M. J. Cima and R. Langer, Proc. IEEE, 2004, 92, 6–21.
9 L. G. Griffith and M. A. Swartz, Nat. Rev. Mol. Cell Biol., 2006, 7,
211–224.
This journal is ß The Royal Society of Chemistry 2007
10 M. D. Tang, A. P. Golden and J. Tien, J. Am. Chem. Soc., 2003,
125, 12988–12989.
11 M. D. Tang, A. P. Golden and J. Tien, Adv. Mater., 2004, 16,
1345–1348.
12 J. Fukuda, A. Khademhosseini, J. Yeh, G. Eng, J. J. Cheng,
O. C. Farokhzad and R. Langer, Biomaterials, 2006, 27,
1479–1486.
13 M. M. Stevens, M. Mayer, D. G. Anderson, D. B. Weibel,
G. M. Whitesides and R. Langer, Biomaterials, 2005, 26,
7636–7641.
14 A. Khademhosseini, S. Jon, K. Y. Suh, T. N. T. Tran, G. Eng,
J. Yeh, J. Seong and R. Langer, Adv. Mater., 2003, 15, 1995–2000.
15 A. Khademhosseini, K. Y. Suh, S. Jon, G. Eng, J. Yeh, G. J. Chen
and R. Langer, Anal. Chem., 2004, 76, 3675–3681.
16 D. R. Albrecht, V. L. Tsang, R. L. Sah and S. N. Bhatia, Lab Chip,
2005, 5, 111–118.
17 V. L. Tsang and S. N. Bhatia, Adv. Drug Delivery Rev., 2004, 56,
1635–1647.
18 A. N. Stachowiak, A. Bershteyn, E. Tzatzalos and D. J. Irvine,
Adv. Mater., 2005, 17, 399–403.
19 P. Lee, R. Lin, J. Moon and L. P. Lee, Biomed. Microdevices, 2006,
8, 35–41.
20 M. S. Hahn, J. S. Miller and J. L. West, Adv. Mater., 2005, 17,
2939–2942.
21 J. S. Miller, M. I. Bethencourt, M. Hahn, T. R. Lee and J. L. West,
Biotechnol. Bioeng., 2006, 93, 1060–1068.
This journal is ß The Royal Society of Chemistry 2007
22 M. S. Hahn, L. J. Taite, J. J. Moon, M. C. Rowland, K. A. Ruffino
and J. L. West, Biomaterials, 2006, 27, 2519–2524.
23 D. J. Beebe, J. S. Moore, J. M. Bauer, Q. Yu, R. H. Liu,
C. Devadoss and B. H. Jo, Nature, 2000, 404, 588–590.
24 D. Dendukuri, D. C. Pregibon, J. Collins, T. A. Hatton and
P. S. Doyle, Nat. Mater., 2006, 5, 365–369.
25 T. Leong, Z. Gu, T. Koh and D. H. Gracias, J. Am. Chem. Soc.,
2006, 128, 11336–11337.
26 V. L. Tsang and S. N. Bhatia, Adv. Drug Delivery Rev., 2004, 56,
1635–1647.
27 D. E. Discher, P. Janmey and Y. L. Wang, Science, 2005, 310,
1139–1143.
28 V. C. Mow and A. Ratcliffe, in Basic orthopaedic biomeahcnis, ed.
V. C. Mow and W. C. Hayes, Lippincott-Raven, Philadelphia, PA,
1997, pp. 113–177.
29 A. Khademhosseini, R. Langer, J. Borenstein and J. P. Vacanti,
Proceedings of the National Academy of Sciences of the United
States of America, 2006, 103, 2480–2487.
30 C. M. Nelson, M. M. Vanduijn, J. L. Inman, D. A. Fletcher and
M. J. Bissell, Science, 2006, 314, 298–300.
31 D. R. Albrecht, G. H. Underhill, T. B. Wassermann, R. L. Sah and
S. N. Bhatia, Nature Met., 2006, 3, 369–375.
32 J. Yeh, Y. Ling, J. M. Karp, J. Gantz, A. Chandawarkar, G. Eng
and J. Blumling, Biomaterials, 2006, 27, 5391–5398.
33 G. M. Walker, H. C. Zeringue and D. J. Beebe, Lab Chip, 2004, 4,
91–97.
Lab Chip, 2007, 7, 574–579 | 579