Proceedings of the 8th International Conference on MicroManufacturing

University of Victoria, Victoria, BC, Canada, March 25-28, 2013

Micro Injection Molded Microtopographic Polymer Plates used to ICOMM

2013
Mechanically Direct Stem Cell Activity No. 9
1 2 3 4
John W. Rodgers , Meghan E. Casey , Sabrina S. Jedlicka , and John P. Coulter
1
Mechanical Engineering and Mechanics; Lehigh University; Bethlehem, PA; [email protected]
2
Bioengineering; Lehigh University; Bethlehem, PA; [email protected]
3
Materials Science and Engineering, Bioengineering, Center for Advanced Materials and Nanotechnology; Lehigh University;
Bethlehem, PA; [email protected]
4
Mechanical Engineering and Mechanics; Lehigh University; Bethlehem, PA; e-mail: [email protected]

ABSTRACT Silicon has been used in multiple studies as a microfeatured

tooling material for injection molding. It is often considered
A novel micro injection molding assembly which allows for to be used more for prototyping than for industrial mold
firm yet compliant housing of silicon tooling was designed assemblies due to its brittle nature, with metal stampers
and manufactured. Microchannels were etched into a silicon being a more durable alternative [4]. However, the
wafer through the use of ultraviolet lithography (UVL) robustness of silicon tooling over a period of a relatively
combined with deep reactive ion etching (DRIE). Injection large number of cycles has been proven as well [5].
molded polystyrene plates containing microtopography were Molded macroscale polymer parts containing micro and
fabricated in which multiple molding parameter studies were nanotopography have previously been used for the purpose
executed to further understand the effect of mold
of inducing specific stem cell responses in a cell culture
temperature and injection velocity on replication. Micro and
laboratory environment. Microfeatured polymer plates have
nano featured polymeric substrates have tremendous
also been used to differentiate stem cells into particular cell
potential for use in stem cell culture, as cells are exposed to
types and as a means to prevent stem cells from
and controlled by microtopography in their natural
differentiating [6,7]. However, a very limited number of
environment.
studies have used the high throughput manufacturing process
of injection molding, to fabricate such substrates [8,9]. In
INTRODUCTION addition to topography, it has been shown that stem cell
differentiation can be controlled by substrate stiffness, in
Micro injection molding can be used to manufacture which stiffness values equivalent to a particular biological
macroscale parts containing microtopography. During tissue type will cause stem cells to begin differentiating into
conventional injection molding, a molten polymer liquid the same cell type [10].
enters and fills a pre-machined cavity and freezes upon Polystyrene (PS), a relatively stiff thermoplastic, is the
contact with the cooler mold walls. The immediate frozen polymer used to make the standard cell culture dish used in
layer would be a fatal flaw if attempting to fill microfeatures laboratories worldwide. Although biocompatible with proper
present at the cavity surfaces. Consequently, mold plasma treatment, the stiffness of PS is multiple orders of
temperatures are typically maintained around glass transition
magnitude greater than that of biological tissue. To avoid
temperature to provide adequate mobility of the polymer
using a different material, it is hypothesized that a locally
chains to fill channels prior to freezing [1]. Increases in
more compliant surface stiffness can be created through the
injection velocity and melt temperatures also have been
use of molded microfeatures capable of elastically
shown to enhance replication [2,3]. At certain size scales,
deforming.
traditional machining of the mold cavity becomes impossible
and alternate methods must be used. Techniques derived Classical mechanics can be used to determine the optimal
from the electronics industry can be implemented to create cross-sectional shape of a beam to be used that would result
micro and nanofeatured substrates. One such technique is in the most significant reduction in apparent stiffness. The
UVL, in which a 2 dimensional geometric pattern is area moment of inertia is a measurement of the resistance to
projected onto a photosensitive substrate via the combination bending of a beam. To obtain a low apparent stiffness, the
of a selectively light permeable mask and ultraviolet rays. inertia must therefore be minimized, in which the value is
The material most commonly used for lithography is silicon. determined through the integral:
UVL is followed by etching (wet or dry) to permanently
transfer the geometry to the silicon wafer. To fabricate I =
ʃʃ y2 dA (1)
anisotropic features, DRIE, which is a dry etching process,
can be used. By using UVL with DRIE, a large area of
The value “y” is the distance from the axis of rotation and
microchannels can be fabricated over a relatively large
the cross-sectional area. Consequently, a circular cross-
surface area.
section satisfies this requirement, and cylindrical beams are
43
8th ICOMM, March 25-28, 2013

chosen as the stiffness reduction geometry. According to the Table 1: DRIE Conditions used to create microchannels.
Euler-Bernoulli beam bending approximation, the deflection Plasma flow time Source
Cycles
Processing
Type (SCCM) /cycle (s) Power (W) time
is given by:
SF6 125 3 1200 34 2 m 50 s
C4 F8 100 2 1200
υ = 64FL3 / 3πED4 (2) C4 F8 150 20 1500 1 20 s

F is a force applied at the tip of the beam, L is the length of square window present in the aluminum insert. A 0.02’’
the beam, E is the bulk modulus of the material, and D is the thick layer of PTFE was positioned between the wafer and
diameter of the pillar. It is hypothesized that if the cell wafer backing to dampen the impact load of the polymer
comes into contact with the deflecting micropillars (instead during molding. A 0.01’’ thick PTFE gasket was fixed in
of bulk PS) that the mechanical perception of the substrate front of the wafer to provide a mechanical buffer between
from a cellular processing perspective will be a more aluminum and silicon, and to seal off the cavity. The main
compliant substrate. mold cavity includes a sprue, cold slug well, short runner,
The current study focuses on the manufacturability of such and fan gate. The type of gate was chosen to provide a
stiffness reducing microfeatured substrates, and specifically uniform flow front across the part and presumably impart an
the filling behavior of a micro injection molded network of equal filling of microcavities. The resultant macroscale
polystyrene pillars. molded part was a plate with a raised platform containing the
microtopography (Fig. 3). Two cartridge heaters on either
side of the mold which were used to heat the assembly,
EXPERIMENTAL combined with wire thermocouples, provided a closed-loop
feedback control to maintain specific mold temperatures
A. UV PHOTOLITHOGRAPHY throughout the molding process.
A 1564 Å thick layer of silicon dioxide was deposited onto C. MICRO INJECTION MOLDING
the surface of a P-type Crysteco 3 inch silicon wafer with a
Micro injection molding trials were conducted with a 3 ton 2
thickness of 509 ± 26 μm. OCG 825 positive photoresist was
stage injection unit micro injection molding machine (Nissei
spin coated at 5,000 rpms for 40 s onto the wafer, yielding a
AU3E). The polymer molded was general purpose
layer of 9000 Å. A photolithography mask containing
polystyrene (666D, Americas StyrenicsLLC).
transparent holes orthogonally arrayed with a diameter of 2
μm and spacing of 3 and 4 μm was placed over top of the The polymer has a melt flow rate of 8 g/10 min (200 oC / 5
coated wafer, and UV exposure was applied using a Karl kg; D-1238), elastic modulus of 3,000 ± 34 MPa, and Vicat
Suss MJB3 mask aligner at a power of 25 W for 1.4 s. The Softening Temperature of 99 oC. Polystyrene was chosen
wafer was immersed in an OCG 809 solution for 20 s due to its widespread use as a cell culture platform in the
(photoresist developer), and rinsed with water, leaving laboratory. A study on the effect of lower mold
regions of exposed silicon dioxide (SiO2). To remove the temperatures (<Tg) was executed to analyze the progression
SiO2 and expose silicon in preparation for etching, the wafer of the melt flow into microchannels (trial 1).
was submerged in buffered HF for a period of 7 min.
The Bosch process was used, in which activation and A
passivation gases were plasma polymerized for a prescribed
number of cycles. The etching conditions are shown in Table
1. Each cycle consisted of a 3 s period of sulfur hexafluoride
(SF6), which etches isotropically, and a 2 s period of
octafluorocyclobutane (C4F8), which provides a protective
coating to the sidewalls of the microchannel (allowing for a
net anisotropic feature). The wafer was then immersed in
photoresist stripper to reveal bare silicon. A final deposition
of C4F8 was applied as an anti-stiction agent. Fig. 1b shows a B
schematic of the etched profile that is created from the
process.

B. MOLD ASSEMBLY

The silicon wafer was diced into 10 x 10 mm2 squares and

inserted into the mold assembly (Fig. 1), in which the wafer
sits inside of an aluminum backing that is screwed into the
aluminum insert. When screwed in, the wafer occupies a Fig. 1: Micromold Aluminum insert assembly exploded (A)
and assembled (B). TC = thermocouple.

44
8th ICOMM, March 25-28, 2013

Table 3: Molding conditions for Trial 2

Specimen # Flow Rate, Q (cm3/s)
2a 2.01
2b 4.02
2c 6.03

A B C

Fig. 2: Full mold assembly schematic. The red rods are wire
thermocouples.

In addition, a study analyzing the effect of injection velocity

on replication was conducted at mold temperatures above the Fig. 3: Schematic of molded part (A), locations of measured
Tg of polystyrene (which is assumed to be 91 oC, trial 2). The regions (B) and actual molded part (C).
molding conditions for trial 1 are shown in Table 2. The
injection time (filling time + holding time) was maintained RESULTS AND DISCUSSION
at 8 s, cooling time was held constant at 20 s, holding
pressure was 19.3 MPa, and back pressure was 4.1 MPa for UVL with DRIE yielded microchannels on the surface of the
both trials. The temperature zones were set to 232.2 oC / silicon wafer. (Fig. 4). There is a widening near the top of the
226.7 oC / 226.7 oC / 221.1 oC / 201.7 oC from the nozzle to channel. Presumably, photoresist protected the top edge of
the rear of the plastication screw, respectively. For trial 2, the holes for a majority of the etch time and eventually was
injection flow was the test parameter (Table 3). To further etched away, allowing for an increase in diameter near the
facilitate microfeature filling, the barrel temperatures were channel entrance.
increased. Injection time was 8 s, cooling time was dictated The microchannels had a diameter of 1.64 μm and depth of
by the mold temperature reaching 82.2 oC, and holding and 4.75 μm, which corresponds to an etch rate of 1.68 μm/min
back pressures were the same as trial 1. Temperature zones and aspect ratio of 2.9 (Table 1). The channels appear to be
were 248.9 oC / 243.3 oC / 232.2 oC / 221.1 oC / 201.7 oC for straight, with sidewall scallops (an inherent result of the
trial 2. Melt temperature, or the temperature of the polymer Bosch process). The etch rate is dependent on many different
as it leaves the injection unit, was measured using the “30- parameters, such as plasma flow rate, wafer temperature, and
30-30 rule” with a wire thermocouple (30 cycles, 30oC activation to passivation ratio.
above predicted temperature, 30 s wait prior to One common concern associated with injection molding is
measurement). A LEO 1550 Variable Pressure Scanning the presence of undercuts in the tooling cavity. Undercuts
Electron Microscope was used to image the surface of the lead to damage of the molded part, causing final part
wafer and polymer samples in distinct regions of the part. deformation as well as residual polymer inside the cavity. In
For trial 1, images were taken near the gate. For trial 2, 9 fact, it is recommended that some draft be on all positive
distinct locations on the surface of the part were recorded part features to ensure proper part ejection. A draft angle was
(shown in Fig. 3). When measuring pillar height as a intentionally avoided due to the specific geometrical design
function of the distance from the gate, the middle regions (2, requirements of a cylindrical beam with constant cross-
5, and 8) were used. section.
The scallops imparted to the sidewalls during the etching
Table 2: Molding conditions for Trial 1
process can be considered nano-undercuts that could
Specimen # Tmold (oC) Flow Rate, Q (cm3/s) potentially hinder the ejection of the micropillars.
Nevertheless, a taper near the top of the of cavity provides
1a 65.6 7.54
facilitation of ejection and a slightly increased mechanical
1b 76.7 2.51 stability for the molded pillars.When molding using mold
temperatures below Tg, replication values are dramatically
1c 76.7 5.03 reduced, as premature solidification occurs prior to filling
1d 76.7 7.54 the microcavities.

45
8th ICOMM, March 25-28, 2013

A B

Fig. 5: Trial 1 near gate of Tmold = 65.6 oC (A) vs. Tmold = 76.7 oC
at identical flow rate of 7.54 cm3/s.

Fig. 4: Microfeatured silicon mold insert micrograph and

microchannel dimensions.

Table 4: Microchannel dimensions

Dimensions 3 μm C2C
Depth 4.75 ± .08 μm
Diameterchannel 1.64 ± .03 μm
Diametertop 1.84 ± .02 μm Fig. 6: Cross-WLF plot of Viscosity vs. Shear Rate for different
melt temperatures of 666D.

Near the gate, an increase in mold temperature from 65.6 oC

to 76.7 oC allows for minimal flow into the cavity, as
evidenced by Fig. 5B. A common behavior associated with
injection molding is hesitation flow, in which the polymer
fills larger regions of a cavity prior to the smaller domains.
The delayed filling of the microcavities allows the polymer
near the wall to cool for an additional period of time, thereby
increasing its viscosity.Viscosity, defined as the resistance of
fluid to flow, is a function of shear rate, temperature, and
pressure. It is often modeled through the use of the cross-
WLF equation, in which viscosity is shown to decrease with Fig. 7: Viscosity vs. Melt temperature at different shear rates.
increasing shear rate and increasing melt temperature since it
is a non-Newtonian fluid (Fig. 6, constants for polystyrene
obtained from Autodesk® Moldflow database). Injection velocity was also varied at a constant mold
temperature of 76.7 oC. Replication remained relatively
constant even when comparing the lowest and highest flow
ηbulk = ηo / ((1+(ηo/τ))^(1-n)) (3) rates (which differ by a factor of 3) (Fig. 8). It is quite
apparent that although some degree of microchannel filling
ηbulk is the bulk viscoity of the polymer, η0 is the viscosity can occur with a mold temperature below Tg of the polymer
at a zero shear rate, and is determined through a calculation being molded, the achievable replication is fairly low, and is
performed with experimentally determined constants. very difficult to enhance even when altering parameters
Viscosity tends to follow such trends in a linear manner (in a known to affect replication (such as flow rate).
logarithmic scale) above a certain shear rate threshold, and Increasing the mold temperature, although a positive
appears to do so independent of melt temperature (see Fig. influence for replication, often leads to additional processing
-1 o
7). At at a shear rate of 100 s , a 176 C reduction in time, and may impart deleterious defects such as lower
temperature only increases the viscosity by 22 Pa*s, molecular weight and increased brittle behavior. The
compared to a 2928 Pa*s increase in viscosity with the additional process time is accredited to the mold being above
identical temperature difference at a shear rate of 1 s-1. ejection temperature of the polymer, in which the molded
Hesitation flow allows for a decrease in shear rate, therefore part is too hot to allow for ejection. A “variotherm” system
making the microchannel flow susceptible to polymer is required to increase and reduce mold temperature
viscosity which is highly sensitive to the melt temperature. synchronously with the molding cycle. An effective
variotherm process does not sacrifice cycle time.

46
8th ICOMM, March 25-28, 2013

phases of the molding cycle, as shown in Fig. 9 and Fig. 10.

A For the part molded with a flow rate of 2.01 cm3/s, pillars
are barely visible for both the end of fill and middle region.
In contrast, pillars are beginning to form in the same
locations for the part molded at 4.02 cm3/s. The highest level
of replication was achieved through using the highest flow
rate. However, the consistency of pillar height across the part
was greatest for the middle flow rate.
The top surface of the pillars appear to be concave for those
with lesser height, in which the middle of the surface is
B slightly collapsed. The uneven surface could be due to the
presence of trapped air inside of the cavity. Traditionally
injection molding involves venting of air from the tooling
cavity through either shallow channels which run along the
tooling surface or through the clearance between the ejector
pins and ejector pin holes. However, designs which include
microcavities with relatively high aspect ratios create a
situation where the air has no avenue to escape and thus
becomes trapped inside the microfeatures. When attempting
C to fill the cavity, the air pushes against the polymer melt and
prevents further filling from occurring. Moreover, the air
may push in a radial direction, causing the concave feature
on the pillar surface. Vacuum venting has been suggested as
a potential solution to the trapped air dilemma. However,
designing and implement such systems is rather costly and
has proven only marginally successful.
Fig. 8: Trial 1 near gate of Qinj = 2.51 cm3/s (A), 5.03 Another potential cause for the pillar tip shape is the
cm3/s (B), and 7.54 cm3/s (C). presence of a micro sink mark, in which the core of the pillar
is insulated from the relatively cooler walls of the
microcavity. When the hotter internal polymer cools, it
However, it does result in increased wear of the tooling as a
contracts and can pull the solidified polymer down into the
result of rapid heating and cooling [11].The second trial
center of the feature.
imparts such a system, in which the tooling is actively heated
and passively cooled. The uneven pillar tips could also be due to a locally lower
viscosity at the microcavity tooling surface. The nanoridges
In trial 2, the mold temperature was maintained well above
present on microcavity walls may enhance the shear thinning
the Tg of 666D, with polymer flow rate during the filling
effect, as the melt is subjected to a certain nanoroughness
phase being the varied parameter. The set barrel and screw
along the wall. It would be interesting to note whether the
conditions led to a melt temperature of 196.1 oC. SEM
same surface geometry occurred at a smaller scale. It has
micrographs were collected in 9 different regions to provide
been previously suggested by Yao et al. that as feature
a comprehensive view of the filling activity across the part
dimensions are reduced, the lower viscosity regions of an
(Fig. 9, Fig. 10, Fig. 11). With a few exceptions, measured
advancing flow front transition from the outer region (where
pillar heights were greater than that of samples molded at
higher shear rates are present) to the center of the flow [12].
mold temperatures below Tg of polystyrene. The few areas
where replication is poor is most likely due to the presence The consideration of microscale viscosity can be made
of micro sink marks, in which the warmer polymer below the within the context of the Cross-WLF model, in which such
surface has pulled the pillar surface towards the core of the value is given as:
part.
The highest replication occurred near the gate for all three ηmicro = (γdot, T, P) = (1+ ξ(g2/z2))* ηbulk (4)
flow rates. Replication is defined here as the ratio of pillar
height to microcavity depth. For the lowest flow rate (Fig.
ξ is a non-dimensional constant specific to the polymer being
9), the highest achievable replication was 0.257 (pillar height
molded, g is the radius of gyration of the polymer molecular
= 1.22 ± 0.05 μm). The middle flow rate yielded a height
chain, and z is the thickness of the polymer fill region. The
replication value of 0.267 and the high flow rate resulted in a
locally higher viscosity is attributed to molecular effects near
maximum value of 0.609.
the wall, in which the forces associated with molecular
Increasing the flow rate by a factor of even 2 significantly bonds are no longer negligible, but begin to contribute to the
enhanced replication when the mold temperature was resistance to movement of the polymer.
maintained above Tg during the filling and compensation

47
8th ICOMM, March 25-28, 2013




 



 
  
μm)



 




 





 




 




 

 
 


 



Fig. 12: Pillar heights along the part for different flow rates
at a mold temperature of 104.4 oC.
Fig. 9: Micrographs of 9 regions of PS molded with flow rate of
2.01 cm3/s.
case that the flow path narrows as the polymer is
Molecular effects have also been suggested as a cause for filling.Consequently, a local microscale flow regime can
increased elastic modulus with decreasing scale [13]. occur, in which the solidified polymer essentially functions
For the pillars with the highest replication, the tip concavity as a part of the mold cavity. Resultantly, the microscale flow
is relatively reduced. The presence of taller pillars indicate phenomenon of lower viscosity towards the middle of the
that the flow length inside of the microchannel increased. As channel could provide enhanced filling capacity away from
the polymer flows through the channel, it is most likely the the wall and thus decrease the concave features present on
the lesser filled microchannels.
One consideration previously mentioned is the need for a
taper on negative tooling (i.e., positive featured parts). The
scallops present on the sidewalls of the silicon tooling are
seen as witness marks on the sides of the pillars. There also
appears to be a slight stretching of the pillars occurring in
response the interaction between the pillars and silicon
scallops (Fig. 13).

CONCLUSION

Polystyrene substrates containing polymer microtopography

were fabricated using the high throughput high precision
manufacturing method of micro injection molding. Mold
temperature and injection velocity were both shown to alter
replication of the microfeatures. Moreover, the scallops
Fig. 10: Micrographs of 9 regions of PS molded with flow rate
present on the sides of the microcavities were shown to alter
of 4.02 cm3/s.
the geometry of microfeatures.
Further design and experimentation will be executed within
the context of traditional injection molding principles (slight
draft angles on microfeatures) combined with novel micro
and nanoscale processing considerations (microscale
viscosity, size scale dependence of elastic modulus) to move
closer to achieving a highly uniform network of
microfeatures for use in mechanically directing the activity
of stem cells.

Fig. 11: Micrographs of 9 regions of PS molded with flow rate of

6.03 cm3/s.
Fig. 13: Micrograph of pillars with slightly stretched scallop
marks closer to base of pillar.

48
8th ICOMM, March 25-28, 2013

Acknowledgements: National Science Foundation (Grant

#1014987), Ray Filozof, Bill Mushock, Brian Slocum for
laser cut gaskets.

REFERENCES
A

[1] M. Heckele, "Review on micro molding of

thermoplastic polymers," Journal of Micromechanics
and Microengineering, 2004; 14: R1-R14.
[2] B. Sha, "Investigation of micro-injection moulding:
Factors affecting the replication quality," Journal of
Materials and Processing Technology, 2007; 183: 284-
296.
[3] C. Srirojpinyo, "Processing Parameters Affecting
Nanoinjection Molding," NSTI-Nanotech. 2004: 464-
467.
[4] N. Zhang, "Replication of micro/nano-scale features by
micro injection molding with a bulk metallic glass
insert," Journal of Micromechanics and
Microengineering, 2012; 22: 065019.
[5] S.H. Yoon. "Evaluation of novel tooling for nanoscale
injection molding," SPIE International Symposia,
Smart Structures & Materials / NDE; San Diego, 2005.
107-116.
[6] M. Mrksich, "Stem Cell Differentiation: Multipotency
retained," Nature Materials, 2011; 10:559-560.
[7] J. Fu, "Mechanical regulation of cell function with
geometrically modulated elastomeric substrates,"
Nature Methods, 2010; 7: 733-736.
[8] C. Holzer, "Nanostructured polymer surfaces to
optimize cell behavior. ANTEC. Orlando, FL: Society
of Plastics Engineers. 2012.
[9] N. Gadegaard,"Biomimetic Polymer Nanostructures by
Injection Molding," Macromolecular Materials and
Engineering, 2003; 288: 76-83.
[10] Y Pek, "The effect of matrix stiffness on mesenchymal
stem cell differentiation in a 3D Thixotropic gel,"
Biomaterials, 2010; 31:385-391.
[11] J. Giboz, "Microinjection molding of thermoplastic
polymers: a review," Journal of Micromechanics and
Microengineering, 2007; 17: R96-R109.a
[12] D. Yao, "Simulation of the filling process in micro
channels for polymeric materials," Journal of
Micromechanics and Microengineering, 2002;12: 604-
610.
[13] L. Sun. "Modeling the size-dependent elastic properties
of polymeric nanofibers," Nanotechnology, 2008; 19:
455706-455713.

49