You are on page 1of 10

Journal of Materials Processing Technology 209 (2009) 54945503

Contents lists available at ScienceDirect

Journal of Materials Processing Technology

journal homepage:

Fabrication of 3D biocompatible/biodegradable micro-scaffolds using dynamic

mask projection microstereolithography
Jae-Won Choi a , Ryan Wicker a , Seok-Hee Lee b, , Kyung-Hyun Choi c , Chang-Sik Ha d , Ildoo Chung d

W.M.Keck Center for 3D Innovation, The University of Texas at El Paso, El Paso, TX 79968, USA
School of Mechanical Engineering, Pusan National University, San 30 Jangjeon-dong, Geumjeong-gu, Busan 609-735, South Korea
Department of Mechatronics Engineering, Cheju National University, Jejudaehakno, Jeju-Si, Jeju-do 692-756, South Korea
Department of Polymer Science and Engineering, Pusan National University, San 30 Jangjeon-dong, Geumjeong-gu, Busan 609-735, South Korea

a r t i c l e

i n f o

Article history:
Received 19 December 2008
Received in revised form 28 April 2009
Accepted 1 May 2009
Poly(propylene fumarate) (PPF)
Microstereolithography (SL)

a b s t r a c t
Microstereolithography (SL) technology can fabricate three-dimensional (3D) tissue engineered scaffolds with controlled biochemical and mechanical micro-architectures. A SL system for tissue
engineering was developed using a Digital Micromirror Device (DMDTM ) for dynamic pattern generation and an ultraviolet (UV) lamp ltered at 365 nm for crosslinking the photoreactive polymer solution.
The SL system was designed with xy resolution of 2 m and a vertical (z) resolution of 1 m. To
demonstrate the use of SL in tissue engineering, poly(propylene fumarate) (PPF) was synthesized with
a molecular weight of 1200 Da. The viscosity of the PPF was reduced to 150 cP (at 50 C) by mixing
with diethyl fumarate (DEF) in the ratio of 7:3 (w/w). Finally, 2% (w/w) of bis(2,4,6-trimethylbenzoyl)
phenylphosphine oxide (BAPO) was added to the solution to serve as a photoinitiator. Cure depth experiments were performed to determine the curing characteristics of the synthesized PPF, and the resulting
system and prepolymer were used to construct a 3D porous scaffold with interconnected pores of
100 m. Scanning electron microscopy (SEM), and micro-computed tomography (CT) images of the
micro-architecture illustrate that the developed SL system is a promising technology for producing
biodegradable and biocompatible 3D micro-scaffolds with fully interconnected pores.
2009 Elsevier B.V. All rights reserved.

1. Introduction
Microstereolithography (SL) technology, which evolved from
conventional stereolithography, was suggested for producing
micro-scale complex structures. The rst types of SL systems
were based on the vector-based scanning SL method, referred to
as line-scan (Ikuta and Kirowatari, 1993; Takagi and Nakajima,
1993). These laser-based scanning methods also led to the development of nanostereolithography methods using two photon initiated
photopolymerization where several precisely detailed, threedimensional microstructures have been demonstrated (Maruo and
Kawata, 1998; Park et al., 2005; Yang et al., 2007). A number of SL
researchers continue using laser-based scanning systems today,
although much focus of current efforts is in mask-based SL systems, where entire layers are projected at a single time using a
physical mask instead of scanning the surface with a laser beam.
The mask-based SL method was suggested by Nakamoto and
Yamaguchi (1996). Further developments led to a dynamic mask
projection SL system in 1997 by Bertsch et al. where a Liquid Crystal Display (LCD) was used as a dynamic mask instead of the physical

Corresponding author. Tel.: +82 51 510 2327; fax: +82 51 514 0685.
E-mail address: (S.-H. Lee).
0924-0136/$ see front matter 2009 Elsevier B.V. All rights reserved.

mask as used by Nakamotoa and Yamaguchi. In 2001, the resolution

limitation of the LCD system was enhanced by replacing the LCD
with a Digital Micromirror Device (DMDTM ) (Bertsch et al., 2001).
In SL, the LCDs and DMDs are used as dynamic masks or
dynamic pattern generators that create any desired pattern on
the resin surface. Since the rst SL system was suggested, many
researchers have developed similar SL systems because of the
superior capability for producing complex 3D microstructures
(Farsari et al., 2000; Bertsch et al., 2004; Sun et al., 2005; Choi et al.,
2005, 2006, 2009; Limaye and Rosen, 2007; Ha et al., 2008; Han et
al., 2008).
These SL systems can produce implantable scaffolds for tissue
engineering applications using biomaterials that are biocompatible and biodegradable in the human body. A biodegradable scaffold
suggests that the scaffold is gradually degraded chemically in vivo
leaving the desired shape of the regenerated tissue without causing
cytotoxicity (Yang et al., 2001). Biocompatibility means that the surface of the scaffold remains chemically compatible such that cells
attach to and grow on the scaffold without inducing any undesired
reactions (or immune responses) with neighboring tissues (Yang
et al., 2001). Fortunately, biodegradable and biocompatible materials can be synthesized for use in SL. Cooke et al. (2002) rst
produced a poly(propylene fumarate) (PPF) scaffold that was fabricated in a conventional SL system. The micro-architecture of the

J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503
Table 1
The pore sizes used in the literature.


Pore size (m)

Boyan et al. (1996)



Whang et al. (1999)

Adult mammalian skin


Hadlock et al. (2000)

Tienen et al. (2002)
Leong et al. (2003)
Cho et al. (2005)

Peripheral nerve
Fibrocartilage formation
Liver tissues
Human chondrocytes


scaffold, however, was limited because of the resolution of the system. Lee et al. (2007, 2008) fabricated porous PPF scaffolds with
micro-architecture using a scanning SL system and demonstrated
cell attachment on the fabricated scaffolds. However, the geometries demonstrated by Lee et al. were limited to simple shapes
and true three-dimensionality, particularly with 3D interconnecting channels that are difcult to be fabricated in SL. As a result, the
capability of SL to fabricate complex 3D micro-architectures has
not yet been widely explored, and optimization of these scaffold
geometries requires examination with regard to cell attachment,
proliferation, and differentiation.
The biodegradable and biocompatible scaffold for tissue engineering requires a desired tissue shape, and a porous and
interconnected micro-architecture, which provides a pathway for
transport of nutrients and metabolic waste to and from a cell and/or
tissue. In addition, porous scaffolds should have high surface area
for the initial cell attachment, and a specic micro-scale pore size
depending on cell/tissue types (Hutmacher, 2001; Yang et al., 2001).
Table 1 shows the reported pore sizes according to the cell/tissue
types used in various research. As shown in Table 1, the pore size
depends on the desired cell/tissue type ranging from 5 m to hundreds of microns. This shows the optimal size of the pore requires
investigation using scaffolds with controlled micro-architectures.
These scaffolds can further be used for examination of optimum
pore geometry with respect to cell/tissue type.
In this work, to produce biodegradable and biocompatible scaffolds having a controlled micro-architecture, the development of
a DMDTM (Digital Micromirror Device)-based microstereolithography (SL) system, PPF synthesis and characterization, and
fabrication of 3D PPF micro-scaffolds using medical imaging data
were performed. PPF was synthesized using DEF (diethyl fumarate)
and PG (propylene glycol) along with the hydroquinone and zinc
chloride. The DMD-based SL system was developed with the
lateral resolution of 2 m, and vertical resolution of 1 m.
The synthesized PPF was mixed with the DEF to reduce the PPF
viscosity for use in the SL system. In order to characterize
PPF/DEF, curing tests for penetration depth and critical energy were
performed. Finally, the fabricated microstructures using medical
imaging data were examined by optical microscopy, Scanning electron microscopy (SEM), and micro-computed tomography (CT) to
verify the capability of the developed SL system for producing
micro-architectures with interconnecting micro-pores. The following sections describe these experiments and their results in more

layer by layer in the system (Choi, 2007). The achievable resolution

in SL is within 10 m, whereas it is at best several tens of m
in conventional SL. There are mainly two types of SL: one is the
scanning-based SL using a focused laser spot and the other is the
projection-based SL using a projected light pattern. In scanning
SL, the laser spot is focused and scanned onto the resin surface
using an XY-stage to move the spot instead of a galvanometer mirror (as is used in macro-SL), which induces a defocusing problem
(Ikuta and Kirowatari, 1993). In projection SL, an illuminated light
is patterned using a high-resolution pattern generator, and projected onto the resin surface. Projecting the entire image onto the
surface results in solidifying the entire layer at a time without having to scan the surface (Bertsch et al., 1997, 2004; Sun et al., 2005).
In this work, DMD-based dynamic mask projection SL was used,
which was developed and described previously (Choi, 2007; Choi
et al., 2006, 2009; Ha et al., 2008), and the details of the system are
contained in the following.
DMD-based dynamic mask projection SL system consists of
the light emission subsystem (a lamp, an optical ber, a lter, and a
collimating lens set), the light delivery subsystem (a LightGateTM , a
tube lens, and a reecting mirror), the dynamic pattern generation
subsystem (the DMDTM ), the image focusing subsystem (a modular
focusing unit and an objective lens), and the build subsystem (a
Z-stage, a platform, a resin vat, and a hot plate) as shown in Fig. 1
(Choi, 2007; Choi et al., 2009).
In order to make the light delivery subsystem compact, the light
path was reduced. To accomplish this, the LightGateTM (Unaxis Co.,
USA) was coated for UV light and used as a prism. The LightGateTM
reects the light to the DMDTM and transmits the returned pattern
to the tube lens. To deliver the patterned light toward the objective
lens, the tube lens (Achromat doublet lens, MellesGriot Co., USA)
with the focal length of 120 mm and reecting mirror were used.
The tube lens helped to collimate the patterned light, and was positioned at its focal length from the DMDTM . The reected light from
the DMDTM was assumed as the source; therefore the light between
the tube lens and the objective lens could be collimated (Choi, 2007;
Choi et al., 2009).
In the pattern generation subsystem, the DMDTM (DMD Starter
Kit, Texas Instruments, USA) was chosen for a dynamic pattern generator. The DMDTM consists of 786,000 micromirrors
(1024 768), in which each mirror is 13.68 m along each side, and
independently tilted at 12 by an electrostatic force. The principle
of generating the light pattern using DMDTM is that the incident
light is reected in two directions depending on the mirror tilt
angles, and one of the reected light bundles is the pattern. The

2. Dynamic mask projection microstereolithography

2.1. Principle and components
In SL, the fabrication method works in the same fashion as
in SL. That is, 3D microstructures can be produced by slicing a 3D
model with a computer program, solidifying, and stacking images


Fig. 1. Developed dynamic mask projection SL system.


J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503

new solution surface with the desired layer thickness by moving

downward deeply, moving upward to the predetermined position,
and then waiting for a certain time for the solution in order to
be evenly distributed. This settling time depends on the solution
viscosity and can be experimentally determined. Fig. 3 shows the
process of refreshing the solution surface. Because the recoating
process involves resin ow, the viscosity is a dominant factor in
SL, and a low viscosity solution is needed to ensure precise layer
thicknesses and fast fabrication by reducing the settling time (Choi,
2007; Choi et al., 2009).
Fig. 2. The distances among DMDTM , tube lens, objective lens, and resin surface.

bundle of light reected at +12 makes the desired pattern, which

is projected on the resin surface through the tube lens and objective
lens, whereas the other bundle is projected to a dummy direction.
The DMDTM makes a certain pattern by tilting each mirror at 12
according to the binary information of each pixel (one pixel is the
same as one micromirror) from a binary image that is transferred
to the DMDTM board (Choi, 2007; Choi et al., 2009).
In the image focusing subsystem, two kinds of objective lenses
with numerical aperture (N.A.) of 0.13 and 0.3 (CFI Plan Flour, Nikon,
Japan) were selectively used as a projection lens. The focal lengths
(fo ) of the objective lenses with the N.A. of 0.13 and 0.3 were 50 mm
and 20 mm, respectively. The objective lenses were selectively used
according to the reduction ratio, which determined the achievable
resolution and working area. Using optical system design software,
the distances between the DMDTM surface and the tube lens (ldt )
and the tube lens and the objective lens (lto ), as shown in Fig. 2, were
optimized. Table 2 shows the original and optimized distances. The
magnication can be calculated by the ratio of the distances ldt and
fo by the basic lens equation, because the light between the tube
lens and the objective lens was collimated. Therefore, the magnications using the two objective lenses were 0.434, and 0.174,
so that 1 pixel (13.68 m) on the DMDTM would represent 5.9 m
and 2.4 m on the image plane (resin surface), respectively (Choi,
2007; Choi et al., 2009).
The build subsystem consists of a Z-stage with the resolution
of 100 nm for stacking, a platform for attaching a substrate, a vat
for containing the photopolymer liquid solution, and a hot plate for
controlling the temperature of the solution. The Z-stage makes a

2.2. Fabricated microstructures using commercial photopolymers

Commercial photocurable resins are candidates for use in SL,
because many resins are sufciently reactive with UV light wavelengths used in SL. Alternatively, commercial monomers along
with a photoinitiator can also be designed and used for specic
purposes. Using the developed SL system, several microstructures with complex features were fabricated as shown in Fig. 4.
Each microstructure was fabricated using different fabrication conditions and materials as shown in Table 3. Fig. 4 shows SEM
images of a micro-wineglass, micro-cup, micro-bishop, and microsprings, respectively. In particular, the XY-stage under the optical
system was used for mass production (Choi, 2007; Ha et al.,
2008), and Fig. 4(c) shows nine microstructures in the same
build. The achievable feature size is 30 m, and the volumes
for each are 1 mm 1 mm 1.5 mm, 1.95 mm 1.95 mm 2.4 mm,
1.7 mm 1.2 mm 2.7 mm, and 0.5 mm 0.5 mm 1.2 mm (for
one micro-spring), respectively.
3. PPF/DEF prepolymer
3.1. PPF synthesis
Diethyl fumarate (DEF), propylene glycol (PG), zinc chloride, and
hydroquinone were purchased from Fisher Scientic Korea (Seoul,
South Korea), and bis(2,4,6-trimethylbenzoyl) phenylphosphine
oxide (BAPO) was purchased from Ciba Specialty Korea (Seoul,
South Korea). Poly(propylene fumarate) (PPF) was synthesized with
a two-step reaction as previously reported (Peter et al., 1999; Fisher
et al., 2001, 2002a,b; Shung et al., 2002). Briey, DEF and PG were

Table 2
The original and optimized distances between the optics.
Objective lens

N.A. 0.13
N.A. 0.3

ldt (mm)

lto (mm)

Magnication (fo /ldt )

Original distance

Optimized distance

Original distance

Optimized distance





50/115.2 (0.434)
20/115.2 (0.174)

Fig. 3. Building process: (a) irradiation and curing; (b) deep dip traversing so the neighboring solution can ow on top of the part; (c) traversing upward to the desired
position (1 layer thickness); (d) waiting until the solution surface becomes uniform.

J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503


Fig. 4. Fabricated microstructures using the developed SL system: (a) micro-wineglass (Choi, 2007); (b) micro-cup (Choi, 2007); (c) micro-bishops (Ha et al., 2008); (d)
micro-springs (Choi et al., 2009).

prepared as main components with the molar ratio of 1:3 along

with the zinc chloride (0.01 mol) as a catalyst, and hydroquinone
(0.002 mol) as a crosslinking inhibitor. In the rst reaction, the prepared solution was mixed by an overhead mechanical stirrer, and
heated from 100 C to 150 C with an increase of 10 C every 20 min
under nitrogen. It was maintained for 7 h (the time is equivalent
to 90% of reaction) once the temperature reached 150 C. Diester
intermediate as a main product, and ethanol as a byproduct were
produced through transestrication. In the second reaction, the
solution was maintained at 130 C under vacuum for 1 h (the time is
equivalent to the predetermined molecular weight of the PPF). PPF
as a main product and PG as a byproduct were produced. The synthesized PPF was not puried because this initial work was focused
solely on manufacturability using the SL system, and cell studies of PPF scaffolds manufactured using the SL system were left
for future work (as mentioned previously, however, PPF has been
a widely investigated biopolymer (Peter et al., 1999; Fisher et al.,
2001, 2002a,b; Shung et al., 2002), and manufacturing PPF with
the SL system is not expected to alter the biocompatibility of the

3.2. PPF characterization

To verify the utilization of synthesized PPF in the developed SL
system, the viscosity was measured and adjusted. In general, if the
solution is too viscous (usually the viscosity of the solution in SL
is maintained below 200 cP), the solution surface cannot be easily
refreshed because of the building process (see Fig. 3). The viscosity
of the synthesized PPF was measured using a viscometer (SV-10, A
& D Co., Japan) and gel permeation chromatography (GPC) (Waters
510, USA) was used to measure its molecular weight. The measured
viscosity was 12,000 cP at 47 C as shown in Fig. 5, where the
viscosity higher than 12,000 cP could not be measured because of
the measurement limitations. In addition, the average molecular
weight was determined to be 1200 Da.
3.3. PPF/DEF prepolymer preparation
The viscosity of the PPF depends on the time of the second
reaction, and the viscosity of the solution was modied for SL
as previously mentioned. Therefore, a diluent was needed to lower

Table 3
Fabrication conditions for the microstructures in Fig. 4.

Total layer

Layer thickness







SI40a :IBXAb = 1:1 (w/w)

IBXA:HDDAc :BEDd = 8:1:1
(by wt.%); DMPAe 5%
(w/w) as a photoinitiator

SI40 was purchased from 3D systems (USA), and suitable for conventional SL
IBXA represents isobornyl acrylate, and purchased from Woorim Chem Tech Co.,
South Korea.
HDDA represents 1,6-hexanediol diacrylate, and purchased from Miwon Commercial Co., South Korea.
BED represents bisphenol-A-ethoxylated (4) diacrylate, and purchased from
Hannong Chemicals Co., South Korea.
DMPA represents dimethoxy-2-phenylacetophenone, and purchased from
Fisher Scientic Korea, Seoul, South Korea.

Fig. 5. PPF viscosity variation with temperature.


J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503

Fig. 6. PPF/DEF viscosity variation with temperature.

Fig. 7. 3D model used for curing depth experiments.

the viscosity of the solution without changing the desired biomaterial properties. DEF, which is a main material for PPF synthesis,
was used as the diluent as it has been reported that DEF does not
signicantly alter the biomaterial properties of PPF. In addition, DEF
has a carbon double bond, so it participates in crosslinking (Fisher
et al., 2002b). In this work, DEF was added to PPF with the ratio of
3:7 (w/w), and the ratio was obtained from the work of Fisher et al.
(2002b). Fisher et al. (2002b) showed that the mixture of DEF/PPF
(3:7 by wt.%), where the molecular weight of the synthesized PPF
was 1260 Da, had the highest elastic modulus and fracture strength.
The viscosity of the PPF/DEF as a function of temperature is shown
in Fig. 6 and was measured using the same viscometer as above. The
PPF/DEF prepolymer was prepared by stirring for 12 h along with
the 2% (w/w) of BAPO as a photoinitiator. In Fig. 6, the viscosity of the
PPF/DEF represents 700 cP at room temperature. This viscosity is
relatively high compared to the desired value of 200 cP. However,
the viscosity of the solution was logarithmically reduced by elevating the temperature, and the viscosity was 150 cP at 50 C. This
value is suitable for SL (Choi, 2007), so a hot plate was installed
under the vat to maintain the temperature of the solution at 50 C.
3.4. Cure depth experiment
A photocurable solution is crosslinked by connecting a
monomer, oligomer or polymer chain, where they have carbon double bonds that can be broken down by a radical. The photons from
the projected light break down the photoinitiator into a radical
along the penetration direction of the light. The radicals then bond
the neighboring monomer, oligomer, or polymer by breaking carbon double bonds. Therefore, in terms of 3D micro-fabrication, the
penetration depth of the light and critical energy at photoinitiation
are important and need to be controlled.
To examine the penetration depth and critical energy, curing
depth experiments were conducted using the PPF/DEF prepolymer.
The energy delivered on the solution surface (Emax ) penetrates into
the solution. The energy inside the solution at the depth z (E(z)) is
dened by BeerLambert law as described in Eq. (1), where Dp is
the penetration depth of the solution (Jacobs, 1993). By introducing
the critical energy (Ec ) into Eq. (1), the curing depth (Cd ) can be
dened as in Eq. (2), where Ec is the energy at the gel point. The
gel point is the point at which solidication begins. Therefore, two
important characteristics of the photocurable solution are Ec and
Dp . These can be experimentally determined through measuring
of the curing depth according to the exposure energy. From the
determined values of Ec and Dp , the exposure energy and stacking
thickness can be chosen (Choi, 2007; Choi et al., 2009). In addition,

it is important to note that the smaller the curing depth, the ability
to fabricate down-facing and complex microstructures is improved:
E(z) = Emax e(z/Dp )
Cd = z(Emax ) = Dp ln





To conduct the curing depth experiment, the curing model as

shown in Fig. 7 was used. The curing model consists of posts and
crossbeams. The posts were stacked with 10 layers with a layer
thickness of 100 m using 30 s exposure time, giving a total length
for the posts of 1 mm. The crossbeams were fabricated in the
last layer with the given exposure energy. The irradiance in the
developed SL was 33.8 mW/cm2 , and the exposure energy was
controlled by opening the shutter for 1, 2, 3, 4, and 5 s. The thicknesses of the four crossbeams were measured at the center using a
microscope (DFC 280, LEICA, Germany).
The specimens for curing characteristics were fabricated
as shown in Fig. 8. The crossbeams for 1 s exposure time
(=33.8 mJ/cm2 ) were not obtained because the thickness of each
individual crossbeam was so small that it was broken during rinsing.
Fig. 9 represents the cure depth graph by measuring the thickness
of the center of the crossbeams. In accordance with Fig. 9 and Eq.
(2), the critical energy and penetration depth were determined to
be 32.4 mJ/cm2 and 78 m, respectively.
4. Fabrication results and discussion
4.1. 3D model reconstruction
To demonstrate the capability for producing 3D micro-scaffolds
using the synthesized PPF solution in the developed SL system,
a model was desired that contained three-dimensionality, downfacing surfaces and micro-architectures. A human kidney model
was selected for demonstration as it has complex 3D architecture.
Fig. 10 shows the reverse engineering process using medical CT
imaging data. The process consists of using previously segmented
anatomical information from CT imaging data (provided by Materialise Inc.), extracting and saving the data for the kidney using
MimicsTM (Materialise Inc.), and then importing the kidney data
into Imageware 10.0 (SDRC Co.), which treated the data as point
clouds. To provide a nal kidney model, unnecessary point clouds
were removed, cross-sectional slices were created to make B-Spline
curves, lofting was used to create B-Spline surfaces, and the surfaces were capped to make a water-tight solid model. To make a
porous scaffold model, rectangular pores (with scaled size equiv-

J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503


Fig. 8. Fabricated crossbeam specimens with exposure times of: (a) 2 s; (b) 3 s; (c) 4 s; (d) 5 s exposure time.

alent to 120 m 120 m or 50 pixels 50 pixels) were generated

along the horizontal and vertical directions and Boolean operations
in a CAD system were used to incorporate these channels in the scaffold model. Finally, the scaffold model was scaled down by a factor
of 0.013 to t within the build size of the SL system, sliced with
the thickness of 4 m, and converted to binary images for building
in the SL system as described in the previous section.
4.2. Fabrication
To produce the kidney scaffold using the synthesized PPF/DEF in
the developed SL system, the exposure energy was chosen to be
68 mJ/cm2 , which is equivalent to 2 s exposure in the previously
described cure depth experiment. The selected energy resulted in a
curing depth of 60 m for down-facing surfaces according to the
previous curing experiments. To maintain the viscosity of PPF/DEF
solution to be 150 cP, the temperature was maintained at 50 C
using a hot plate placed under the vat. A settling time of 30 s was

used after the deep dip process, and the SL system was equipped
with the objective lens with the N.A. of 0.3.
Fig. 11 shows the SEM image of the fabricated PPF/DEF kidney
scaffold with a volume of 1400 m 820 m 700 m and pore
size of 100 m. The scaffold was built with 169 layers and the
layer thickness of 4 m. The exterior of the scaffold appears to have
reproduced the 3D geometry quite well, including the freeform
outer shape of the kidney as well as the rectangular vertical pores.
However, the fabricated pore size (100 m) is smaller than the
designed size (120 m) due to material shrinkage. Furthermore,
micro-CT imaging (CT 80, SCANCO Medical AG Co., Switzerland)
was used to show cross-sectional images, internal pore architectures, and interconnectivity. The reconstructed image from the
micro-CT data, shown in Fig. 12, illustrates that the rectangular
pores have good interconnectivity in the vertical direction. However, the horizontal channels could not be found in any of the
micro-CT imaging planes. This could be due to overcure on the
down-facing surfaces, crosslinking of uncured PPF/DEF prepolymer
in the channels due to successive scanning (i.e., similar to overcure,
UV penetration into the uncured trapped material in the channels),
and/or the inability to remove uncured PPF/DEF from the channels
when rinsing and subsequently curing this uncured material in the
post-curing process.
4.3. Discussion

Fig. 9. Cure depth variation with exposure energy.

To better understand the manufacturing capabilities of PPF/DEF

in SL, the PPF/DEF kidney scaffold was compared with a similar
model that was fabricated using a commercial resin (Choi, 2007).
Fig. 13 shows the fabricated kidney scaffold made from the same
commercial resin that was used in the models shown in Fig. 4(a)
and (b). For this fabrication, the system was equipped with an
objective lens of 0.13 N.A. The scaffolds shown in Fig. 13 for the
commercial resin and Fig. 11 for PPF/DEF show comparable feature
detail with clear rectangular pores and similar freeform surfaces.
However, upon closer examination, the commercial resin offers
some improved feature detail over PPF/DEF. This can be seen, for
example, in Fig. 13 with clear evidence of successful fabrication
of the horizontal channels and interconnectivity of the channels


J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503

Fig. 10. Reverse engineering process to reconstruct human kidney model from CT data: (a) CT imaging data; (b) extracted point cloud; (c) cleaned point cloud; (d) crosssectional point series; (e) cross-sectional B-Spline curves; (f) lofted surface; (g) solid geometry model; (h) scaffold model; (i) binary image series used to fabricate the scaffold
in SL.

Fig. 11. SEM image of fabricated PPF/DEF kidney scaffold.

with the vertical rectangular channels. As mentioned previously,

the horizontal channels were not successfully reproduced in the
PPF/DEF scaffold (see Fig. 12). There may be several factors that
can enhance the manufacturability of PPF/DEF to achieve improved
three-dimensional architectures, but it should be noted that this
initial fabrication of PPF/DEF in SL is quite successful as much
research and development has been invested into manufacturability of commercial resins.
One factor that could account for the difculty in fabricating
the horizontal channels is the inability to remove uncured resin
from these channels during the rinsing process. As shown in Fig. 6,
the viscosity of the uncured PPF/DEF surrounding the fabricated
scaffold increased to more than 700 cP at room temperature after
removing the scaffold from the heated vat (which is required to
clean the part). During rinsing, it was noted that the uncured
PPF/DEF was strongly adhered to the structure and difcult to
remove. The scaffold was rinsed by dipping it in a beaker with iso-

Fig. 12. Reconstructed micro-CT image of PPF/DEF scaffold: (a) reconstructed image; (b) horizontal cross-sectional image (the scale bar is 100 m).

J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503

Fig. 13. SEM image of fabricated kidney scaffold using the commercial resin.

propyl alcohol (IPA) for several minutes. Because of its orientation

on the glass slide (i.e., the scaffold was rst attached to a substrate,
or glass slide, during fabrication), it was easier to clean the vertical channels using simple agitation in the beaker. Although this
does not fully explain the inability to fabricate horizontal channels, the authors believe that there was uncured resin that was not
removed from the scaffold and the cleaning process is an important
area of research for fabricating micro-architectures with biomaterials such as PPF/DEF. Furthermore, the relatively long rinsing
time before post-curing also contributed to distortion of the scaffold. The drying process also distorted the scaffold due to capillary
forces (Sun et al., 2005). Excessive shrinkage and distortion of the
PPF/DEF scaffolds are also important characteristics of the manufacturing process and experimental biopolymer that should be
accounted for and minimized in the process. Methods for minimizing shrinkage and distortion are currently under investigation.
One immediate solution under investigation that could improve the
removal of uncured polymer from micro-channels and also improve
the effects of distortion could be to elevate the temperature of the
rinsing solution.
To illustrate that the SL system with this biopolymer could
successfully fabricate both horizontal and vertical channels with
interconnectivity, a simple PPF/DEF cubic scaffold was designed and
manufactured. This scaffold, shown in Fig. 14, demonstrates that
the manufacturing process can successfully fabricate these channels in both orientations. In this scaffold, the interconnectivity of
the channels was increased so that the uncured resin could be more
easily removed. The cubic scaffold shown in Fig. 14 also further
illustrates the excessive shrinkage experienced by this polymer.
Although shrinkage remains an issue, the successful fabrication and
cleaning of the cubic scaffold conrms that the system can be used
to fabricate scaffolds with complex micro-architectures, and like


commercial resins, the polymer and manufacturing process require

optimization for manufacturability.
For photocuring systems, a few factors work simultaneously and
competitively on the properties and manufacturability of the fabricated structures. The most important factor in photocuring systems
is the degree of crosslinking (or crosslinking density). Crosslinking
density is strongly dependent on the exposure time, the wavelength
of the light, the type of photoinitiator, as well as the reactivity of
monomers. In general, increasing the crosslinking density of the
polymers increases the strength of cured polymers (Ha et al., 1996;
Jung et al., 1998). In terms of the crosslinking density, the PPF used
in the present research was not puried, and so some unreacting
materials could have remained in solution, and these materials did
not participate in crosslinking, and thus, reduced the crosslinking
density (Varadan et al., 2001).
Furthermore, the molecular weight of the synthesized PPF was
1200 Da, which resulted in the high viscosity of the solution. As
described previously, SL requires a viscosity of 200 cP for the
recoating process. This was achieved using DEF as a diluent, a heated
vat, and a settling time of 30 s, and although micro-fabrication was
successful, some dimensional inaccuracies undoubtedly resulted
from temperature variations within the vat (i.e., from the free surface during settling). In addition to the molecular weight affecting
the viscosity, the overall manufacturability of polymers, in general,
are inversely proportional to the molecular weights of polymers
(Aiello and Mantia, 2001). That is, on increasing the molecular
weight of polymers, the polymers are more difcult to use in manufacturing due to an increased melt viscosity (Menon et al., 2002),
despite the strength of the polymer increasing with viscosity. The
manufacturability of polymers is further impacted by the molecular weight distribution within the polymer (Ha et al., 1996; Jung
et al., 1998). The DEF content used in this work was xed at 30%,
similar to that used in the work of Fisher et al. (2002b), because
Fisher et al. showed that a solution of PPF with 30% (w/w) of DEF
had the highest elastic modulus and fracture strength. In terms of
manufacturability, different concentrations of DEF will affect the
curing characteristics, and the effects of these concentrations on
micro-fabrication need to be explored. Finally, 2% (w/w) of BAPO
was added to the PPF/DEF solution to serve as a photoinitiator.
The curing depth and width vary according to the photoinitiator
concentration, and the effects of photoinitiator concentration on
dimensional accuracy, manufacturability, etc. also require further
investigation (Varadan et al., 2001). Thus, future studies will involve
investigations on the effects of PPF molecular weight, DEF concentration, and photoinitiator concentration.
In addition to the polymer solution, there are other controllable
parameters in the fabrication process. To obtain ne microstructures with overhanging features, the light irradiance and exposure
pattern can be controlled that impact the dimensional accuracy
(Choi et al., 2009). The same energy with a different irradiance
produces different curing width and depth (Varadan et al., 2001);

Fig. 14. Fabricated PPF/DEF cubic scaffold with more pores: (a) SEM image; (b) horizontal cross-sectional CT image.


J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503

multiple exposures can also be used to enhance the dimensional

accuracy by dividing the given exposure energy into multiple exposures (Wu, 2005). These controllable factors in the manufacturing
process that can impact dimensional accuracy and the ability to
fabricate micro-features will also be investigated in future work.
5. Conclusions
In this work, a 3D micro-scaffold was fabricated using a previously developed dynamic mask projection SL system and the
biodegradable/biocompatible polymer, poly(propylene fumarate)
(PPF), which was synthesized for this research. This research
showed that SL has the potential to fabricate complex 3D microscale structures with interconnecting pores out of a biopolymer. The
viscosity of the synthesized PPF was reduced by adding DEF to the
solution for fabrication in the SL system, and the viscosity was further reduced during fabrication by heating the material in the SL
vat. Using the PPF/DEF prepolymer and the developed SL system, a
curing experiment was conducted to determine the critical energy
and penetration depth for the solution. The ability of the system to
fabricate a complex 3D micro-scaffold was demonstrated by fabricating a miniature 3D human kidney scaffold with interconnecting
pores. The fabricated micro-scaffold was observed using SEM and
optical microscopy, and the cross-sections were investigated using
a micro-CT imaging system. Although the results illustrated some
manufacturing limitations of the experimental material when compared to commercial resins, it is concluded that the developed SL
system and the use of PPF show promise for fabricating complex
micro-scaffolds with prescribed micro-architectures. In the future,
PPF/DEF scaffolds with controlled micro-architectures will be fabricated and used in in vitro and in vivo studies, and the authors believe
this system can be used to fabricate complex micro-architectures
out of a variety of additional biomaterials that will ultimately have
an impact on many tissue engineering and regenerative medicine
This work was supported by the Korea Research Foundation
Grant (KRF-2007-357-D00023) funded by Korean Government
(MOEHRD), the Basic Program of the Korea Science & Engineering
Foundation (Grant No. R01-2004-000-10507-0), the Mr. and Mrs.
MacIntosh Murchison Chair I in Engineering endowment at the University of Texas at El Paso (UTEP), and a research contract from the
U.S. Army Space and Missile Defense Command and the Homeland
Protection Institute to the Center for Defense Systems Research at
UTEP. We would like to thank Mr. Nathan Castro for taking micro-CT
images. The ndings and opinions presented in this paper are those
of the authors and do not necessarily reect those of the sponsors
of this research.
Aiello, R., Mantia, F.P.L., 2001. On the improvement of the processability of UHMWHDPE by adding a liquid crystalline polymer and a uoroelastomer. Macromol.
Mater. Eng. 286 (3), 176178.
Bertsch, A., Zissi, S., Jezequel, J.Y., Corbel, S., Andre, J.C., 1997. Microstereophotolithography using a liquid crystal display as dynamic mask-generator. Microsyst.
Technol. 3 (2), 4247.
Bertsch, A., Zissi, S., Jezequel, J.Y., Corbel, S., Andre, J.C., 2001. Microstereolithography:
concepts and applications. In: Proceedings of 8th IEEE International Conference
on Emerging Technologies and Factory Automation, pp. 289298.
Bertsch, A., Jiguet, S., Renaud, P., 2004. Microfabrication of ceramic components by
microstereolithography. J. Micromech. Microeng. 14, 197203.
Boyan, B.D., Hummert, T.W., David, D.D., Schwartz, Z., 1996. Role of material surfaces
in regulating bone and cartilage cell response. Biomaterials 17, 137146.
Cho, S.H., Oh, S.H., Lee, J.H., 2005. Fabrication and characterization of porous alginate/polyvinyl alcohol hybrid scaffolds for 3D cell culture. J. Biomater. Sci. Polym.
Ed. 16 (8), 933947.

Choi, J.W., Ha, Y.M., Won, M.H., Choi, K.H., Lee, S.H., 2005. Fabrication of 3dimensional microstructures using dynamic image projection. In: Proceedings of
International Conference on Precision Engineering and Micro/Nano Technology
in Asia (ASPEN 2005), Shenzhen, China, pp. 472476.
Choi, J.W., Ha, Y.M., Lee, S.H., Choi, K.H., 2006. Design of microstereolithography
system based on dynamic image projection for fabrication of three-dimensional
microstructures. J. Mech. Sci. Technol. 20 (12), 20942104.
Choi, J.W., 2007. Development of projection-based microstereolithography apparatus adapted to large surface and microstructure fabrication for human
body application. Ph.D. Dissertation. Pusan National University, Busan, South
Choi, J.W., Wicker, R.B., Cho, S.H., Ha, C.S., Lee, S.H., 2009. Cure depth control for
complex 3D microstructure fabrication in dynamic mask projection microstereolithography. Rapid Prototyping J. 15 (1), 5970.
Cooke, M.N., Fisher, J.P., Dean, D., Rimnac, C., Mikos, A.G., 2002. Use of stereolithography to manufacture critical-sized 3D biodegradable scaffolds for bone ingrowth.
J. Biomed. Mater. Res. B 64B, 6569.
Farsari, M., Claret-Tournier, F., Huang, S., Chatwin, C.R., Budgett, D.M., Birch, P.M.,
Young, R.C.D., Richardson, J.D., 2000. A novel high-accuracy microstereolithography method employing an adaptive electro-optic mask. J. Mater. Process.
Technol. 107, 167172.
Fisher, J.P., Holland, T.A., Dean, D., Engel, P.S., Mikos, A.G., 2001. Synthesis and properties of photocross-linked poly(propylene fumarate) scaffolds. J. Biomater. Sci.
Polym. Ed. 12 (6), 673687.
Fisher, J.P., Vehof, J.W.M., Dean, D., Waerden, J.P., Holland, T.A., Mikos, A.G., Jansen,
J.A., 2002a. Soft and hard tissue response to photocrosslinked poly(propylene
fumarate) scaffolds in a rabbit model. J. Biomed. Mater. Res. 59, 547
Fisher, J.P., Dean, D., Mikos, A.G., 2002b. Photocrosslinking characteristics and
mechanical properties of diethyl fumarate/poly(propylene fumarate) biomaterials. Biomaterials 23, 43334343.
Ha, C.S., Jung, S.J., Kim, E.S., Kim, W.S., Lee, S.J., Lee, S.J., 1996. Properties of UV-curable
polyurethane acrylates using nonyellowing polyisocyanate for oor coating. J.
Appl. Polym. Sci. 62, 10111021.
Ha, Y.M., Choi, J.W., Lee, S.H., 2008. Mass production of 3-D microstructures using
projection microstereolithography. J. Mech. Sci. Technol. 22 (3), 514521.
Hadlock, T., Sundback, C., Hunter, D., Cheney, M., Vacanti, J.P., 2000. A polymer foam
conduit seeded with Schwann cells promotes guided peripheral nerve regeneration. Tissue Eng. 6 (2), 119127.
Han, L.H., Mapili, G., Chen, S., Roy, K., 2008. Projection microfabrication of threedimensional scaffolds for tissue engineering. J. Manuf. Sci. Eng. Trans. ASME 130
(2), 021005-1-4.
Hutmacher, D.W., 2001. Scaffold design and fabrication technologies for engineering
tissuesstate of the art and future perspectives. J. Biomater. Sci. Polym. Ed. 12
(1), 107124.
Ikuta, K., Kirowatari, K., 1993. Real three dimensional micro fabrication using
stereo lithography and metal molding. In: Proceedings of 6th IEEE Workshop
on Micro Electro Mechanical Systems (MEMS93), New York, NY, USA, pp.
Jacobs, P.F., 1993. Rapid prototyping and manufacturing: fundamentals of stereolithography. Soc. Manuf. Eng..
Jung, S.J., Lee, S.J., Cho, W.J., Ha, C.S., 1998. Synthesis and properties of UV-curable
waterborne unsaturated polyester for wood coating. J. Appl. Polym. Sci. 69,
Lee, J.W., Lan, P.X., Kim, B., Lim, G., Cho, D.W., 2007. 3D scaffold fabrication with
PPFDEF using micro-stereolithography. Microelectron. Eng. 84, 17021705.
Lee, J.W., Lan, P.X., Kim, B., Lim, G., Cho, D.W., 2008. Fabrication and characteristic
analysis of a poly(propylene fumarate) scaffold using micro-stereolithography
technology. J. Biomed. Mater. Res. B 87B (1), 19.
Leong, K.F., Cheah, C.M., Chua, C.K., 2003. Solid freeform fabrication of threedimensional scaffolds for engineering replacement tissues and organs.
Biomaterials 24, 23632378.
Limaye, A.S., Rosen, D.W., 2007. Process planning method for mask projection microstereolithography. Rapid Prototyping J. 13 (2), 7684.
Maruo, S., Kawata, S., 1998. Two-photon-absorbed near-infrared photopolymerization for three-dimensional microfabrication. J. Microelectromech. Syst. 7 (4),
Menon, A.R.R., Aigbodion, A.I., Pillai, C.K.S., Mathew, N.M., Bhagawan, S.S., 2002. Processability characteristics and physicmechanical properties of natural rubber
modied with cashewnut shell liquid and cashewnut shell liquid-formaldehyde
resin. Eur. Polym. J. 38, 163168.
Nakamoto, T., Yamaguchi, K., 1996. Consideration on the producing of high
aspect ratio micro parts using UV sensitive photopolymer. In: Proceedings
of 7th International Symposium on Micro Machine and Human Science, pp.
Park, S.H., Lee, S.H., Yang, D.Y., 2005. Subregional slicing method to increase threedimensional nanofabrication efciency in two-photon polymerization. Appl.
Phys. Lett. 87, 154108.
Peter, S.J., Kim, P., Yasko, A.W., Yaszemski, M.J., Mikos, A.G., 1999. Crosslinking characteristics of an injectable poly(propylene fumarate)/-tricalcium phosphate paste
and mechanical properties of the crosslinked composite for use as a biodegradable bone cement. J. Biomed. Mater. Res. 44, 314321.
Shung, A.K., Timmer, M.D., Jo, S., Engel, P.S., Mikos, A.G., 2002. Kinetics of
poly(propylene fumarate) synthesis by step polymerization of diethyl fumarate
and propylene glycol using zinc chloride as a catalyst. J. Biomater. Sci. Polym. Ed.
13 (1), 95108.

J.-W. Choi et al. / Journal of Materials Processing Technology 209 (2009) 54945503
Sun, C., Fang, N., Wu, D.M., Zhang, X., 2005. Projection micro-stereolithography
using digital micro-mirror dynamic mask. Sens. Actuator A: Phys. 121 (1),
Takagi, T., Nakajima, N., 1993. Photoforming applied to ne machining. In: Proceedings of 4th International Symposium on Micro Machine and Human Science
(MHS93), pp. 173178.
Tienen, T.G., Heijkants, R.G.J.C., Buma, P., Groot, J.H., Pennings, A.J., Veth, R.P.H., 2002.
Tissue ingrowth and degradation of two biodegradable porous polymers with
different porosities and pore sizes. Biomaterials 23, 17311738.
Varadan, V.K., Jiang, X., Varadan, V.V., 2001. Microstereolithography and Other Fabrication Techniques for 3D MEMS. John Wiley & Sons Ltd.


Whang, K., Healy, K.E., Elenz, D.R., Nam, E.K., Tsai, D.C., Thomas, C.H., Nuber, G.W., Glorieux, F.H., Travers, R., Sprague, S.M., 1999. Engineering bone regeneration with
bioabsorable scaffolds with novel microarchitecture. Tissue Eng. 5 (1), 3551.
Wu, D.M., 2005. Micro fabrication of 3D structures and characterization of molecular
machine. Ph.D. Dissertation. UCLA, Los Angeles.
Yang, D.Y., Park, S.H., Lim, T.W., Kong, H.J., Yi, S.W., Yang, H.K., Lee, K.S., 2007. Ultraprecise microreproduction of a three-dimensional artistic sculpture by multipath
scanning method in two-photon photopolymerization. Appl. Phys. Lett. 90,
Yang, S., Leong, K.F., Du, Z., Chua, C.K., 2001. The design of scaffolds for use in tissue
engineering. Part I. Traditional factors. Tissue Eng. 7 (6), 679689.