University of Southern California Dissertations and Theses

Modular bio microelectromechanical systems (bioMEMS): intraocular drug delivery device and microfluidic interconnects

(USC Thesis Other)

Modular bio microelectromechanical systems (bioMEMS): intraocular drug delivery device and microfluidic interconnects

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

MODULAR BIO MICROELECTROMECHANICAL SYSTEMS
(bioMEMS): INTRAOCULAR DRUG DELIVERY DEVICE
AND
MICROFLUIDIC INTERCONNECTS

by

Ronalee Lo

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BIOMEDICAL ENGINEERING)

December 2009
Copyright 2009 Ronalee Lo
ii

Dedication

This dissertation is dedicated to my family
for their unwavering love and support throughout this journey.
iii
Acknowledgements

This dissertation is a culmination of several years of research; however, I could not
have done it alone. There are many people to whom I am indebted for making this
work possible. To everyone who has helped me along this path, I am truly grateful.

Thank you to Dr. Ellis Meng, my research advisor. She provided invaluable
discussion and suggestions throughout my graduate career. I thank my qualification
and dissertation committee members, Dr. Rajat Agrawal, Dr. David D’Argenio, Dr.
Eun Sok Kim, Dr. Ellis Meng and Dr. Jim Weiland for their helpful advice and
insightful questions. Additionally, I would like to thank President Dr. Steven B.
Sample for always making time for me, Dr. Geoffrey Spedding for his mentorship
and advice, and Dr. Michael Khoo for welcoming me to USC.

I also thank our collaborators Dr. Mark S. Humayun, Dr. Rajat Agrawal, and Dr.
Saloomeh Saati for their help in providing medical insights and surgical support.
Thank you to Dr. Donghai Zhu and Mr. Merrill Roragen for maintaining the
cleanroom facilities.

Thank you to the Biomedical Microsystem Lab members for their willingless to give
of their time to discuss ideas, listen to practice presentations, and brainstorm
solutions to tough problems. I especially thank Dr. Brian Po-Ying Li, who partook
iv
in this journey with me. He is a great friend and labmate; providing both
encouragement and camaraderie.

To my friends (you know who you are), who made life these past several years so
enjoyable; I thank you for your time, the adventures that crossed our paths, and new
memories forged during my time at USC. Special thanks go to Dr. Jessica Lee for
all of the laughs and creating gastronomical experiences, be it turkey legs on Tom
Sawyer’s Island, or your gourmet meals at family dinner night. Dr. Hilton Kaplan,
Dr. Nicolas Sachs, Daren Kwok, Jeremy Selan, David Kleinman and Jacques
Kaplan-Abrahms, thank you for all of the lively conversations at our weekly family
dinners. Albert Lai, thank you for all of the shared experiences at USC, time spent in
GPSS, URSC, and 101 Terrace, and late night food runs to Jerry’s or Diddy Riese.

Dr. James C. Eckert, thank you for your friendship, wise counsel, and willingness to
just drive around when I needed a break. I truly appreciate all of your precious time,
reading my dissertation, looking at slides, listening to frustrations, and celebrating
triumphs.

And finally, to my family, Elsie Yu Lo, Katherine Dintenfass, and Dr. David Mann,
for whom this dissertation is dedicated, mere words cannot describe nor thank you
for all of the love, encouragement, and support you have given me, not only during
my pursuit of a doctorate, but throughout my entire life. All of you have been
v
instrumental in providing me with the strength and desire to achieve my goals. I
express my deepest gratitude to you.

vi
Table of Contents
Dedication ............................................................................................................ii
Acknowledgements.....................................................................................................iii
List of Tables ............................................................................................................x
List of Figures ..........................................................................................................xii
List of Equations ......................................................................................................xxx
Abstract .......................................................................................................xxxi
Chapter 1 Introduction ........................................................................................1
Chapter 2 MEMS Based Drug Delivery Devices................................................6
2.1 Introduction..........................................................................................6
2.1.1 Ocular Drug Delivery Methods............................................................7
2.1.1.1 Topical and Oral Medications..........................................................7
2.1.1.2 Intraocular Injections.......................................................................8
2.1.1.3 Implants............................................................................................9
2.2 A MEMS Approach to Drug Delivery: Manually-Actuated Device .10
2.2.1 Device Design....................................................................................14
2.2.2 The Components................................................................................15
2.2.2.1 Refillable Reservoir and Refill Guides ..........................................15
2.2.2.2 Cannula and Check Valve..............................................................17
2.2.2.3 Support Posts..................................................................................19
2.2.2.4 Suture Tabs....................................................................................19
2.2.3 Device Fabrication.............................................................................20
2.2.3.1 The Silicon and Acrylic Masters....................................................20
2.2.3.2 Layer Fabrication...........................................................................24
2.2.3.3 Device Assembly...........................................................................32
2.2.3.4 Dimensions of Assembled Device .................................................36
2.2.4 Benchtop Experiments- Methods and Results ...................................37
2.2.4.1 PDMS Bonding..............................................................................37
2.2.4.2 Check Valve...................................................................................42
2.2.4.3 Refillability....................................................................................52
2.2.5 In Vivo and In Vitro Experiments- Methods and Results...................61
2.2.5.1 Device Placement...........................................................................61
2.2.5.2 Device Functionality......................................................................63
2.2.6 Summary............................................................................................66
2.3 Electrically-Actuated Device with Dual Check Valve.......................68
2.3.1 Device Design....................................................................................69
2.3.1.1 The Components............................................................................69
vii
2.3.1.2 Device Fabrication.........................................................................84
2.3.1.3 Benchtop Experiments- Methods and Results .............................101
2.3.2 Summary..........................................................................................119
2.4 Surgical Shams.................................................................................121
2.4.1 Solid Surgical Shams .......................................................................121
2.4.1.1 Design..........................................................................................122
2.4.1.2 In Vivo Experiments ....................................................................126
2.4.2 Hollow Surgical Shams....................................................................130
2.4.2.1 Design..........................................................................................131
2.4.2.2 Acute and Chronic In Vivo Experiments- Methods and Results .143
2.4.3 Summary..........................................................................................152
2.4.4 Additional Applications...................................................................153
2.4.4.1 Rat Retinitis Pigmentosa Drug Delivery Device .........................153
2.4.4.2 Cancer Treatment Device.............................................................159
2.5 Future Work.....................................................................................166
Chapter 3 Microfluidic Interconnects .............................................................168
3.1 Introduction......................................................................................168
3.2 Single Interconnect..........................................................................173
3.2.1 Interconnect Design.........................................................................173
3.2.1.1 Proof-of-Concept.........................................................................174
3.2.1.2 SU-8 Anchors...............................................................................175
3.2.1.3 Septum.........................................................................................177
3.2.1.4 Interconnect Integration...............................................................177
3.2.2 System Fabrication...........................................................................178
3.2.2.1 Test Interconnect..........................................................................178
3.2.2.2 Integrated System.........................................................................180
3.2.2.3 Septum Formation........................................................................183
3.2.3 Benchtop Experiments- Methods and Results .................................184
3.2.3.1 Coring vs. Non-coring Needle Tip Type......................................185
3.2.3.2 FEM Analysis of Stress Distribution ...........................................186
3.2.3.3 Pull-out Force and Reusability.....................................................187
3.2.3.4 Maximum Operating Pressure......................................................199
3.2.3.5 Prolonged Pressure Operation......................................................203
3.2.3.6 Failure Modes..............................................................................204
3.2.4 Summary..........................................................................................206
3.3 Multiple Interconnects.....................................................................207
3.3.1 Design..............................................................................................207
3.3.1.1 Septa Design................................................................................208
3.3.1.2 Needle Guides..............................................................................213
3.3.1.3 Side Ports.....................................................................................214
3.3.1.4 Microchannels..............................................................................217
3.3.1.5 Metal Structures...........................................................................220
3.3.1.6 Arrayed Interconnect Permeations...............................................220
3.3.2 Fabrication.......................................................................................224
viii
3.3.2.1 SU-8 Microchannel......................................................................225
3.3.2.2 Parylene C Microchannel.............................................................236
3.3.2.3 Needle Array................................................................................249
3.3.3 Experimental Methods and Results..................................................253
3.3.3.1 FEM Analysis of Stress Distribution ...........................................253
3.3.3.2 Photoelastic Stress........................................................................257
3.3.3.3 Insertion Test................................................................................260
3.3.3.4 Pressure Test................................................................................275
3.3.3.5 Electrolysis Pressure Generation..................................................282
3.3.3.6 Sideport Functionality..................................................................285
3.3.3.7 Parylene C Microchannel Functionality.......................................287
3.3.4 Summary..........................................................................................291
3.4 Future Work.....................................................................................294
Chapter 4 Conclusion......................................................................................295
References ........................................................................................................297
Appendices ........................................................................................................303
Appendix A- Fabrication Process for Silicon Masters ........................................303
Appendix B- Steps to Mount Silicon Master to Glass Substrate.........................304
Appendix C- Fabrication Steps for Creating Acrylic Master..............................305
Appendix D- Mask Used to Fabricate Bottom Layer Silicon Master .................306
Appendix E- Fabrication Steps for Creating Bottom and Middle Layers ...........307
Appendix F- Mask Used to Fabricate Middle Layer Silicon Master ..................308
Appendix G- Fabrication Process for Creating Top Layer from Acrylic Master 309
Appendix H- Pattern to Cut PDMS Reservoirs...................................................310
Appendix I- Cleaning Process for Device Layers Prior to Oxygen Plasma
Treatment .........................................................................................311
Appendix J- Oxygen Plasma Treatment Process for Bonding Bottom and Middle
Layers...............................................................................................312
Appendix K- Process for Making PDMS Members of a Certain Thickness.......313
Appendix L- Mask Used to Make Metal Alignment Marks for All of the Modular
Valve Processes................................................................................314
Appendix M- Mask Used to Pattern First Layer of SU-8 Valve Plate/ Pressure
Limiter..............................................................................................315
Appendix N- Mask Used to Pattern Second Layer of SU-8 Valve Plate/ Pressure
Limiter..............................................................................................316
Appendix O- Fabrication Process for SU-8 Valve Seat and Pressure Limiter....317
Appendix P- Mask Used to Pattern SU-8 Spacer Plate.......................................320
Appendix Q- Fabrication Process for SU-8 Spacer Plate....................................321
Appendix R- Mask Used to Pattern SU-8 Mold for Silicone Valve Plate ..........323
Appendix S- Fabrication Process for SU-8 Mold to Create PDMS Valve Plate.324
Appendix T- Mask Used to Pattern SU-8 Mold for Silicone Valve Plate with
Optional Bossed Feature ..................................................................326
ix
Appendix U- Fabrication Process for SU-8 Mold to Create Silicone Valve
Plate with Bossed Feature ................................................................327
Appendix V- Mask Used to Pattern SU-8 Mold for Silicone Valve Plate with
Optional Bossed and Overhang Features .........................................330
Appendix W- Fabrication Process for SU-8 Mold to Create Silicone Valve
Plate with Bossed and Overhang Features.......................................333
Appendix X- Assembling Heat Shrink Packaged Valve SOP.............................336
Appendix Y- Procedure to Test Heat-Shrink Packaged Valve............................341
Appendix Z- File Used To Make Custom-Designed Cut Puncture Jig ...............343
Appendix AA- File for Making the Puncture Force Jigs and Illustration of
the Jig Assembly ..............................................................................344
Appendix BB- Laser File for Making the Molds for Possible Layouts for
Version 1 of the Solid Surgical Shams. ...........................................345
Appendix CC- Laser File User to Create Solid Surgical Sham v2_large and
v2_small ...........................................................................................346
Appendix DD- Laser File Used to Create Solid Surgical Sham Mold v3_1.......347
Appendix EE- Laser File Used to Create Hollow Surgical Sham Molds v3_2,
v3_2, and v4_1.................................................................................348
Appendix FF- Laser File Used to Create Hollow Surgical Sham Molds v5_1
and v6_1...........................................................................................349
Appendix GG- Laser File Used to Create Hollow Surgical Sham Mold v7 .......350
Appendix HH- Fabrication Process for Making Hollow Shams .........................351
Appendix II- Surgical Protocol for In Vivo Implantation of Hollow Surgical
Shams ...............................................................................................352
Appendix JJ- Fabrication Process to Create the Interconnect Test Structure......353
Appendix KK- File for Making the Puncture Force Jigs and Illustration of
the Jig Assembly ..............................................................................355
Appendix LL- File for Creating a Layer of the Parylene C Deposition Holder..357
Appendix MM- Masks Used to Fabricate the Single Interconnect Designs .......358
Appendix NN- Wafer Level Pictures of Arrayed Interconnect...........................362
Appendix OO- Mask Used To Fabricate Arrayed Interconnect SU-8 Wafer 1 ..364
Appendix PP- Masks Used to Fabricate Arrayed interconnect SU-8 Wafer 2....365
Appendix QQ- Masks Used to Fabricate Arrayed interconnect Parylene C
Wafer 1.............................................................................................368
Appendix RR- Masks Used to Fabricate Arrayed interconnect Parylene C
Wafer 2.............................................................................................371
Appendix SS- Fabrication Process for Arrayed Interconnects with SU-8
Microchannels..................................................................................376
Appendix TT- Fabrication Process for Arrayed interconnects with SU-8
Microchannels and Metal Components............................................377
Appendix UU- Fabrication Process for Arrayed Interconnects with
Parylene C Microchannels ...............................................................379
Appendix VV- Fabrication Process for Arrayed Interconnects with Parylene C
Microchannels with Metal Components ..........................................381
Appendix WW- 4 and 8 Needle Insertion Force Jigs and Assembly ..................383
x
List of Tables

Table 2-1 Dimensions of Drug Delivery Device .......................................................37
Table 2-2 Summary of Results from In Vivo Delivery using the Manually-
Actuated Drug Delivery Device ..................................................................66
Table 2-3 Summary of values used in theoretical calculations of large
deformations in uniform thin plates.............................................................79
Table 2-4 Dimensions of valve components, including the three valve
designs (hole, straight arm, s-shape arm). All components are 900
μm in diameter.............................................................................................80
Table 2-5 Summary of heat-shrink tube characterization results for two tube
gauge sizes (22 AWG and 18 AWG) ........................................................107
Table 2-6 Summary of the FEM results for displacement and stress on an
assembled valve.........................................................................................109
Table 2-7 Summary of packaged valve operating characteristics for hole,
straight arm, and s-shaped arm valve. .......................................................113
Table 2-8 Summary of closing time constants for the packaged valve....................115
Table 2-9 Dimensions of Fabricated Version 1 Surgical Shams. ............................124
Table 2-10 Solid sham timeline and description of solid sham
characteristics. ...........................................................................................126
Table 2-11 Timeline for hollow surgical sham, including major device
characteristics. ...........................................................................................137
Table 2-12 Summary of Results from In Vivo Delivery using Hollow
Surgical Sham............................................................................................146
Table 2-13 Summary of endothelial cell density at the conclusion of the 6
month study for eyes that were implanted and refilled 6 times
during the course of the study and an implant were the refills were
terminated 2 months prior to the end of the study at month 4...................151
Table 2-14 Summary of the component diamensions for the electrolysis
driven device..............................................................................................160
Table 3-1 Comparison of Connector Design Options..............................................170
xi
Table 3-2 Values used for theoretical pull-out force calculation.............................192
Table 3-3 Summary of Connector Parameters for Published Connectors
(Chiou and Lee 2004, Li and Chen 2003, Yao, et al. 2000)......................196
Table 3-4 Summary of Leakage Pressure Results...................................................202
Table 3-5 Summary of fabricated arrayed interconnect combinations. ...................222
Table 3-6 Summary of dimensions and design specifications for the SU-8
microchannel and Parylene C microchannel arrayed interconnect............223
Table 3-7 Summary of the SU-8 microchannel arrayed interconnects which
were fabricated. These interconnects do not have any metal
structures....................................................................................................228
Table 3-8 Summary the SU-8 microchannel arrayed interconnects with
metal, which were fabricated.....................................................................234
Table 3-9 Summary of the Parylene C microchannel arrayed interconnects
which were fabricated. These interconnects do not have any metal
structures....................................................................................................239
Table 3-10 Summary of the Parylene C microchannel arrayed interconnects
with metal structures, which were fabricated. ...........................................246
Table 3-11 Summary of relationship between insertion and removal forces
and the needle type (coring vs. non-coring), needle gauge (27G or
33G), number of needles (1, 4, or 8), and rate of insertion (0.5 or 1
mm/sec) (mean ± SE, n = 4). .....................................................................274
Table 3-12 Summary of failure pressure and failure locations for all septa
designs. Arrows indicate failure points. ...................................................278
xii
List of Figures

Figure 2-1 Illustration of device functionality A) Device is comprised if 3
molded silicone layers. B) Layers are assembled and bonded to
form device which contains a refillable reservoir, flexible cannula,
suture tabs, support posts, and check valve. C) Device is sutured
to the sclera. D) Flexible cannula is inserted into the anterior or
posterior segments of the eye via a scleral tunnel; device is covered
by the conjunctiva (not shown). E) The patient manually-actuates
the device by pressing on the reservoir with their finger. F) The
change in device volume causes an internal pressure to build up
until the check valve opens and fluid is expelled from the device
into the eye interior. G) After several dispensing events, the
device is depleted. H) A surgeon can, in a minimally invasive
manner, refill the device using a 30G (O.D. 305 μm) non-coring
needle. Figure adapted from images courtesy of Tun Min Soe..................13
Figure 2-2 Exploded view of the drug delivery device. The device is
comprised of three layers (bottom, middle, and top) which define
the components of the device. .....................................................................15
Figure 2-3 Placement of the drug delivery device. Note, conjunctiva is not
shown. Figure adapted from image courtesy of Tun Min Soe. ..................16
Figure 2-4 Illustration of the refill ring used to prevent the needle from
penetrating through the base of the device. .................................................17
Figure 2-5 Image of several ring guides placed along a 30 gauge needle.
For application, only one ring guide per need is necessary.........................17
Figure 2-6 Check valve operations for forward and reverse pressure. ......................18
Figure 2-7 Image of the assembled check valve. Dyed liquid is used to
provide contrast. ..........................................................................................19
Figure 2-8 Cross-section of fabrication process to create silicon masters.................21
Figure 2-9 Image of a silicon mold used to create the (a) bottom and (b)
middle layers for the drug delivery device. The silicon masters
were coated in Parylene C to facilitate mold release of the PDMS
layer from the master...................................................................................22
xiii
Figure 2-10 SolidWorks image of the three layers that comprise the drug
delivery device. The entire device is 17 mm in length. Note,
suture tabs are not shown.............................................................................24
Figure 2-11 SolidWorks image of the bottom layer. Dimensions of the
bottom layer are given [mm]. Note, suture tabs are not shown..................25
Figure 2-12 Cross-sectional image of the fabrication process for the bottom
layer silicon master and individual silicone layer. The cross-
section is taken through the line of symmetry (line indicated on
Figure 2-13). ................................................................................................27
Figure 2-13 Red line indicates location of cross-section image for Figure
2-12..............................................................................................................27
Figure 2-14 SolidWorks image of the middle layer. Dimensions of the
bottom layer are given [mm]. ......................................................................28
Figure 2-15 Cross-sectional image of the fabrication process for the middle
layer silicon master and individual silicone layer. The cross-
section is taken through the line of symmetry of the middle layer
(line is indicated in Figure 2-16). ................................................................29
Figure 2-16 Red line indicates location of cross-section image for Figure
2-15..............................................................................................................29
Figure 2-17 SolidWorks image of top layer. The interior cavity of the top
layer defines the reservoir volume. Dimensions are indicated
[mm]. ...........................................................................................................30
Figure 2-18 Process for making drug delivery device reservoirs. A) Use
epoxy to affix acrylic squares onto a glass slide, B) Pour PDMS
prepolymer onto acrylic mold and half-cure PDMS, C) Cut
reservoirs from molded PDMS piece. D) Remove reservoirs from
mold, E) Reservoirs are ready for assembly................................................32
Figure 2-19 Illustration of how the three layers are fabricated and assembled
to form the manually-actuated drug delivery device. ..................................33
Figure 2-20 Placement of pairs of pieces on glass slide to facilitate
placement and alignment of layers after oxygen plasma treatment.............34

xiv

Figure 2-21 Adding reinforcing layer to drug delivery device. A) Device is
placed on an inclined glass slide, B) PDMS prepolymer is poured
around the device, covering the edge of the device but not
occluding the check valve opening, C) The device is removed from
the slide and excess PDMS is cut from the device. .....................................36
Figure 2-22 Procedure used to qualitatively determine bond strength after
oxygen plasma or wet chemical treatment...................................................38
Figure 2-23 Testing setup used to quantitatively measure bond strength after
oxygen plasma or wet chemical treatment...................................................39
Figure 2-24 A) Implanted in vivo device (plasma bonded) with bond failure
location due to surgical handling identified. B) Reinforcing layer
added to bonded device in order to provide more mechanical
robustness to the device. C) Excess silicone was removed from the
device to create the desired device outline, ruler divisions measure
1 mm. ...........................................................................................................42
Figure 2-25 Time-lapsed photographs of dyed DI water being dispensed
from the check valve under manual actuation. ............................................43
Figure 2-26 Typical Diode Current versus Voltage Curve ........................................44
Figure 2-27 Typical Check Valve Pressure versus Flow Rate Curve........................45
Figure 2-28- A) Exploded SolidWorks view of the custom-made laser-
machined jig to characterize the check valve operation. B)
Pressure setup used to open the valve with pressurized water. ...................46
Figure 2-29 Flow Rate vs. Pressure curve for check valve (mean ± SE, n =
4)..................................................................................................................48
Figure 2-30 Check valve control of dosing under 250 mmHg and 500
mmHg (33.3 kPa and 66.7 kPa) of applied pressure. Duration of
applied pressure was varied using a solenoid valve controlled using
a 50% duty cycle square waves. ..................................................................50
Figure 2-31 A representative graph depicting the volume dispensed after the
applied pressure (250 mmHg and 500 mmHg, 33.3 kPa and 66.7
kPa) is removed from the valve. The dashed lines indicate when
the accumulated volume reached 63.2% of the final value, the time
at which this point occurred was defined as the closing time
constant for the valve...................................................................................52
xv
Figure 2-32 Refill needle determination, A) Coring versus non-coring 30
gauge (305 μm OD) needle illustration and SEM images, B) Top
view of needle track through punctured PDMS slab using each
needle, C) Side view of needle track through PDMS slab. .........................54
Figure 2-33 Exploded SolidWorks image of the custom-made laser
machined jig used ensure multiple puncture events pierce the
membrane in the same location for worst-case scenario testing..................56
Figure 2-34 A) Exploded SolidWorks view of the custom-made laser-
machined jig to apply pressure to punctured membranes. B) Setup
used to provide measure leakage pressure of punctured
membranes...................................................................................................57
Figure 2-35 Leakage pressure for 250 μm and 673 μm thick membranes
punctured 8, 12 and 24 times through the same location with a 30G
non-coring needle (n = 4). ...........................................................................59
Figure 2-36 Surgical verification of liquid delivery in vitro using the drug
delivery device.............................................................................................63
Figure 2-37 Surgical verification of drug device refill was completed in
vitro using a commercially-available, standard 30 gauge non-
coring needle................................................................................................65
Figure 2-38 A MEMS ocular drug delivery device, which is sutured to the
eye, contains a refillable drug reservoir, contoured morphology,
cannula, and modular valve. The valve comprises four stacked
disks (valve seat, valve plate, spacer plate, and pressure limiter.
The cannula is inserted into the anterior or posterior segments of
the eye for targeted delivery of drugs..........................................................72
Figure 2-39 Heat-shrink packaged valve integrated into a silicone surgical
sham device. Drug reservoir with metal ring indication refill port
location, heat-shrink tubing, and valve are indicated. Ruler
divisions are 1 mm.......................................................................................73
Figure 2-40 Valve operation (from left to right): initially normally-closed,
valve opens under forward pressure that exceeds cracking pressure,
excessive pressures close the valve, and valve remains closed
under reverse pressure. ................................................................................74
Figure 2-41 Photo of the valve components (valve seat, valve plate, spacer
plate, and pressure limiter), pre-shrink heat–shrink tube, and fully
assembled valve...........................................................................................75
xvi
Figure 2-42 a) Side view and b) top view of the packaged valve in a FEP
heat-shrink tube. The valve was placed inside the tube with a
custom jig. The entire fixture was heated to 215 ºC at 1.5 ºC/min
and cooled at the same rate to room temperature. .......................................76
Figure 2-43 a) Parylene C cannula integrated with a drug delivery pump, b)
clogging of Parylene C cannula after ex vivo testing...................................77
Figure 2-44 Three different valve plate designs a) hole, b) straight arm, and
c) s-shape arm; and the corresponding fabricated valve plates d)
hole (through holes are indicated by the arrows), e) straight arm,
and f) s-shaped arm......................................................................................80
Figure 2-45 Fabrication process for the valve seat and pressure limiter
plates. Fabrication steps are cross-section views at the A-A’ line. ............86
Figure 2-46 Fabrication process for the SU-8 spacer plate. Fabrication steps
are cross-section views at the A-A’ line. .....................................................87
Figure 2-47 Fabrication process for the valve plate using an SU-8 master
mold. Fabrication steps are cross-section views at the A-A’ line.
Straight arm valve is shown; hole and s-shaped arm valves are
fabricated in an identical manner.................................................................89
Figure 2-48 Illustration of the additional features (bossed and/or overhang)
which can be added to the simple valve plate designs.................................90
Figure 2-49 Top and side views of valve assembly. a) valve seat, b) valve
plate added to valve seat, c) spacer plate placed on valve plate, d)
pressure limiter added to assembled valve. .................................................91
Figure 2-50 a) Heat-shrink jig setup. Teflon base and top each contains a
centering pin. The top and base are aligned with machine screws.
Nuts set the top and base distance. b) Close up view of an
assembled valve with pre-shrink heat-shrink tube surrounding the
valve.............................................................................................................92
Figure 2-51 Process steps to package assembled valve. a) FEP heat-shrink
tube is placed around bottom centering pin, assembled valve is
placed on centering pin, b) jig top is added, valve is clamped
between top and bottom centering pins, FEP tube is lifted around
valve, c) jig and valve assembly is placed in vacuum oven, and d)
packaged valve is removed from jig............................................................93
Figure 2-52 Top and side views of the molds used for fabrication the second
generation drug delivery device reservoir. ..................................................94
xvii
Figure 2-53 Assembled surgical sham using molds shown in Figure 2-52. ..............95
Figure 2-54 Components needed to create a fully integrated electrically-
actuated device with dual regulation check valve. ......................................99
Figure 2-55 Fully integrated electronically-actuated drug delivery device.
Device includes a refillable reservoir, electrolysis pump, separate
refill area, refill ring, PEEK baseplate, silicone cannula, heat-
shrink packaged dual check valve, and suture tabs. ..................................101
Figure 2-56 Valve plate deflection setup. ................................................................102
Figure 2-57 Comparison of measured valve plate deflection to the
theoretical values for a flat plate................................................................104
Figure 2-58 Pre and post heat-shrink tubing. Solid disk packaged in heat-
shrink tubing to test robustness of the adhesiveless packaging
method. ......................................................................................................105
Figure 2-59 Visualization of flow rate through a packaged valve using
Rhodamine B. ............................................................................................110
Figure 2-60 Test setup to determine valve operating characteristics.......................111
Figure 2-61 Flow profiles from 4 runs on same valve, valve was kept
hydrated in double distilled water between runs to prevent valve
from drying out. The hole valve plate was used in this packaged
valve...........................................................................................................112
Figure 2-62 Accumulated volume measurements to determine closing time
constant. Closing time constants were calculated by determining
the amount of time for 63.2% of the total accumulated volume to
exit the valve..............................................................................................115
Figure 2-63 Accumulated volume expelled from the packaged valve using
an electrolysis pump and custom-made jig (mean ± S.E., n = 4). .............118
Figure 2-64 Images of the assembled device electrolysis structure. The
electrolysis structure delaminated during current application...................119
Figure 2-65 Definition of major and minor axes on surgical sham devices. ...........123
Figure 2-66 First version of the implanted solid surgical shams and the
dimensions. ................................................................................................125
Figure 2-67 Implanted surgical shams, A) 2 mm x 12 mm x 15.9 mm and B)
1 mm x 13 mm x 29.4 mm.........................................................................127
xviii
Figure 2-68 Mold used to fabricate version 2 of the surgical shams.
Dimensions are the same as version 1 with additional sutures on
the 1mm thick sham (v2_large) and the sutures removed from the
silicone cannula from both shams..............................................................128
Figure 2-69 Implanted solid surgical sham with stainless steel ring visible
through the conjunctiva. The surgeon was able to simulate refill
by targeting the center of the stainless steel ring using a
commercially available 30 gauge needle. The stainless steel ring is
outlined in this image to help indicate its location. ...................................130
Figure 2-70 Illustration of the hollow surgical sham. A refill needle access
the sham interior by piercing the refill port location (designated by
a refill ring). A PEEK baseplate prevents the needle from piecing
through the entire device. ..........................................................................131
Figure 2-71 Illustration of the application of a refill ring on the refill needle.........133
Figure 2-72 Image of the PEEK baseplate to limit refill needle insertion
depth. .........................................................................................................134
Figure 2-73 Top and side view of the initial reservoir design for the
integrated drug delivery device. The reservoir body is separated
from the refill port to prevent the pump chamber and Parylene C
bellows from accidentally being punctured by the refill needle................136
Figure 2-74 Illustration of acrylic molds used to fabricate the hollow
surgical sham. ............................................................................................138
Figure 2-75 Fabrication steps for making the hollow surgical sham.......................141
Figure 2-76- Hollow surgical sham being filled on benchtop..................................142
Figure 2-77 Benchtop demonstration of manual dispensation of dyed liquid
from within a hollow sham device.............................................................142
Figure 2-78 Illustration of the hollow sham placement in the eye...........................144
Figure 2-79 Images of hollow sham device implantation for acute and
chronic in vivo studies ...............................................................................145
Figure 2-80 Transilluminated eye with implanted hollow sham device. The
device and stainless steel ring outlines are clearly visible through
the eye tissue (i.e. conjunctiva) covering the device. ................................147

xix
Figure 2-81 Still images of surgical video taken during device refill in a
chronic in vivo study. A) Transillumination of the eye helps locate
and identify the target refill area. B) A 30G needle is inserted
through the center of the refill area. Needle insertion stops when
the needle tip encounters the rigid baseplate embedded in the
device base. C) Trypan blue dye is injected into the device; the
dye can be observed spreading through the device. ..................................148
Figure 2-82 Images of in vivo device refilling and dispensing. First, the
refill site is checked for any damage, infection, or scarring from
previous refills. Next, the eye is transilluminated to help identify
the refill ring location (the refill ring appears as a darker shadow).
The refill needle (30G non-coring) is inserted through the center of
the refill ring until the needle progressing is stopped by the device
baseplate. Trypan blue dye is injected into the device; dye can be
seen spreading through the device as a dark plume. The dye exits
the cannula and into the anterior chamber. Finally, the puncture
site is inspected for damage or leakage. ....................................................149
Figure 2-83 Assembled device with bone screws used to secure device to rat
skull............................................................................................................154
Figure 2-84 Image of proposed device superimposed on a image of a rat
skull............................................................................................................155
Figure 2-85 Laser file to create molds for the rat retinitis pigmentosa drug
delivery device...........................................................................................156
Figure 2-86 Image of a) curing stand to prevent suture tabs from becoming
sealed during assembly, b) device assembly on curing stand....................157
Figure 2-87 Laser file to create molds for mouse cancer drug delivery
device.........................................................................................................162
Figure 2-88 Image of the components to fabricate the rat cancer drug
delivery device (device body and electrolysis actuator)............................163
Figure 2-89 Fully assembled rat cancer drug delivery device with
electrolysis actuator. Heat-shrink wrapping on wires not shown. ............164
Figure 3-1 Microfluidic system with integrated circular interconnects. 33
gauge non-coring needles were inserted into the input and output
septa. Rhodamine was introduced into the system to demonstrate
system functionality. PDMS septum is outlined to indicate its
location. .....................................................................................................172
xx
Figure 3-2 Exploded view of the setup used to demonstrate horizontal
interconnect proof-of-concept. ..................................................................174
Figure 3-3 Images showing fluid progression in proof-of-concept setup................175
Figure 3-4 A) Image of an assembled circle septum interconnect. B) Top
view of three different septum connector shapes (circle, barbed,
and square) that were designed and integrated into the test
microfluidic system. Needle, PDMS septum, SU-8 housing, and
microchannel in the designs are indicated. C) Side view of the
needle piercing the PDMS septum. Image is not drawn to scale..............176
Figure 3-5 Integrated interconnect with microfluidic system. System
contains SU-8 layer which defines the septum housing,
microchambers, and microchannel. Electrolysis structure and flow
sensors are fabricated with a Ti/Pt metal layer. PDMS septum and
glass cover plate are not present. ...............................................................178
Figure 3-6 Cross-sectional fabrication steps for the test interconnect. Cross-
section is taken through the microchamber. ..............................................180
Figure 3-7 Simplified fabrication process for the microfluidic chip with
integrated interconnect. Cross section views are through the
PDMS septum and microchamber with interdigitated electrodes for
an electrolysis pump. .................................................................................182
Figure 3-8 Edge view of the needle insertion location. The 33 gauge non-
coring needle pierces the PDMS septum through the edge of the
system, creating an in-plane connection....................................................184
Figure 3-9 Stress induced on the septum during needle insertion at a) needle
pre-puncture, and b) after the needle has fully pierced the septum. ..........187
Figure 3-10 Pull-out test setup. Connector is held perpendicular to the
ground by placing the microfluidic device in the Plexiglas test
fixture. Weights are added to a container attached to the luer lock
portion of the needle. Pull-out force is determined by multiplying
gravity by the combined mass of the weights, needle, and
container. Image is not drawn to scale......................................................191
Figure 3-11 Pull-out force of the interconnects are compared to the
calculated theoretical values......................................................................192
Figure 3-12 Comparison of the interconnect (circular, square, and barbed)
and that of other published connectors of the first pull-out force
with respect to contact length. ...................................................................194
xxi
Figure 3-13 Comparison of the interconnect (circular, square, and barbed)
and that of other published connectors of the first pull-out force
with respect to contact area........................................................................194
Figure 3-14 Comparison of the pull-out force for our interconnects (circle,
square, barbed) compared to other published connectors. Pull-out
force varies over subsequent pull-outs and is dependent on contact
area.............................................................................................................197
Figure 3-15 Comparison of normalized pull-out force with respect to
contact area. ...............................................................................................198
Figure 3-16 Test setup for leakage pressure test and prolonged pressure test
using pressurized water. Output needle is blocked using an
Upchurch plug. ..........................................................................................200
Figure 3-17 Test setup for leakage pressure test using pressurized N2.
Leakage is visualized by N2 bubbles escaping from the submerged
the microfluidic chip..................................................................................201
Figure 3-18 Interconnect failure at the PDMS septum and stainless steel
needle interface. A) Water surrounds the needle shaft as PDMS is
debonded from the needle and B) seeps from the needle insertion
point. ..........................................................................................................203
Figure 3-19 Interconnect failure due to Parylene C delamination. Dyed
water can be seen spreading between the Parylene C and substrate
layers..........................................................................................................205
Figure 3-20 Illustration of the a) side view of the arrayed SU-8 and
Parylene C microchannel, b) top view of the needle array and
septa, and c) image of a connected arrayed interconnect. .........................208
Figure 3-21 Schematic indicating key features of our interconnect
technology. Here, interconnects with surface micromachined
Parylene C channels are shown. Needle guides to help align the
needle arrays to the septa. Additional features which can be added
to the arrayed interconnect include sideports and interdigitated
electrodes for electrolysis or electrochemical sensing...............................210
Figure 3-22 Septa configurations used in the arrayed interconnect designs............211
Figure 3-23 Illustration of the needle guides designed to align the trajectory
of the needle for the arrayed interconnect design......................................213
xxii
Figure 3-24 Parallel sideport structures integrated into arrayed interconnect
designs that have A) individual septum, and B) combined septa..............215
Figure 3-25 Illustration of the perpendicular sideports in the arrayed
interconnect design. ...................................................................................216
Figure 3-26 Converging microchannel designs of A) 4 rectangular, B) 8
rectangular, and C) 4 oval overlapping septum designs............................219
Figure 3-27 Fabricated arrayed interconnect with Parylene C microchannels
and sideports. Salient features of the arrayed microfluidic system
with integrated interconnects are highlighted. External access via
needles is not shown in these photographs................................................221
Figure 3-28 Fabrication steps for the SU-8 wafer that contains interconnect
designs that do not require metal structures. The cross-section is
taken horizontally along the microchannel. The SU-8 is lighter in
color at step D because after SU-8 patterning, no SU-8 exists along
the cross-section line. However, the lighter SU-8 represents the
SU-8 material remaining surrounding the needle guide and
microchannel in order to better visualize the process flow after step
D. This process flow is used for designs shown in Figure 4-37A. ...........227
Figure 3-29 Fabrication steps for the SU-8 wafer that contains interconnect
designs that do not require metal structures. The cross-section is
taken horizontally along the microchannel. The SU-8 is lighter in
color at step K because after SU-8 patterning, no SU-8 exists along
the cross-section line. However, the lighter SU-8 represents the
SU-8 material remaining surrounding the needle guide and
microchannel in order to better visualize the process flow after step
D. This process flow is used for designs shown in Figure 4-37B. ...........233
Figure 3-30 Process flow for Parylene C microchannel arrayed interconnect
designs. The cross-section line is taken along the microchannel.
Translucent SU-8 represents SU-8 which surrounds a component
but not within the cross-sectional line. The translucent SU-8 is
included to aid in illustrating the fabrication process. This
fabrication process is used for designs shown in Figure 4-38A. ...............238
Figure 3-31 Time-lapsed images of the working electrolysis structures prior
to packaging. 0.3 mA of current was applied to the electrodes.
Bubble formation was visually confirmed.................................................242

xxiii
Figure 3-32 Process flow for Parylene C microchannel arrayed interconnect
designs. The cross-section line is taken along the microchannel.
Translucent SU-8 represents SU-8 which surrounds a component
but not within the cross-sectional line. The translucent SU-8 is
included to aid in illustrating the fabrication process. This
fabrication process is used for designs shown in Figure 4 38B.................244
Figure 3-34 Needle arrays which provide shared or separate input
capabilities to needles and thus microchannels. a) 4 shared 4
needles, 1 mm spacing, b) 4 shared needles, 2.54 mm spacing, c) 8
shared needles, 1 mm spacing, d) 4 separate needles, 1 mm
spacing, and e) 8 separate needles, 2.54 mm spacing. The scale bar
represents 10 mm.......................................................................................249
Figure 3-35 Custom-made laser machined molds for creating an array of
needles (4 or 8). All layers are made of acrylic and are color coded
to illustrate assembly. ................................................................................252
Figure 3-36 Centerline of FEM analysis of needle insertion induced stress in
arrayed interconnect...................................................................................255
Figure 3-37 FEM images of stress distribution within septa during needle
insertion, a) pre-puncture, b) partial puncture, and c) complete
insertion. ....................................................................................................256
Figure 3-38 Experimental setup to visualize photoelastic stress. ............................258
Figure 3-39 Photoelasic stress in PDMS from needle insertion for a single
(18G) and needle array (four 27G). Yellow arrows indicate areas
of stress. .....................................................................................................259
Figure 3-40 Illustration of membrane behavior during pre-puncture,
puncture, and post-puncture stages of needle insertion.............................261
Figure 3-41 Custom-made laser-machined jigs used to measure insertion
force through PDMS using a A) 4 or B) 8 needle assembly. ....................264
Figure 3-42 Illustration of needle tip displacement relative to the PDMS
membrane for frictional force measurements............................................267

xxiv
Figure 3-43 Generic results from insertion force tests. a) needle touches
surface of the PDMS sample, b) material being pierced by needle
(combination of stiffness and puncture forces), c) needle moving
through PDMS material (friction force), d) needle stops moving
and material relaxes, e) needle in process of being removed from
material (max removal force), f) needle fully removed from
material. Inset: needle displacement over time.........................................270
Figure 3-44 Relationship of post-puncture insertion forces (friction and
cutting forces) and removal forces with respect to the number of
insertion needles. .......................................................................................271
Figure 3-45 Illustration of “tenting” effect. .............................................................277
Figure 3-46 Common failure modes of the arrayed interconnect. a) Leakage
through needle insertion path between needle and septa, b) leakage
at output septa through needle track, c) delamination between the
septa and Parylene C coated substrate, d) delamination between the
Parylene C coating and the substrate. Examples of needle
misalignment are shown in c) and d).........................................................281
Figure 3-47 Internal pressure change due to electrolysis. Pressure increases
when current (0.3 mA) is applied to the interdigitated electrodes,
pressure decreases when current is turned off and the oxygen and
hydrogen gas recombine into water...........................................................284
Figure 3-48 Experimental setup for sideport testing................................................285
Figure 3-49 Time lapse images of sideport function: a) dyed water
introduced in the main septum and un-dyed water through the
sideport, b) close up image of the laminar flow within the
microchannel..............................................................................................286
Figure 3-50 Time-lapsed images of IPA evaporating from within the
arrayed interconnect Parylene C microchannel. ........................................289
Figure 3-51 Time-lapsed images of Rhodamine B moving through the
arrayed interconnect Parylene C microchannel via capillary action.
Scale bar is 1 mm.......................................................................................290
Figure 3-52 Time-lapsed images of Rhodamine B moving through the
arrayed interconnect Parylene C microchannel in a partially
packaged device. Scale bar is 1 mm. ........................................................291

xxv
Figure 3-53 Illustration of the modularity of scale for the arrayed
interconnect. The septa portion of the interconnect can be
elongated or repeated any amount of times to fit the needs of the
system. .......................................................................................................292
Figure 3-54 Example of "plug and play" modularity ..............................................293
Figure 4-1 6 mm x 6 mm acrylic squares to form the acrylic mold of the
device reservoir. Note, image is not shown true to scale..........................305
Figure 4-2 Mask used to create silicon master for the bottom layer of the
drug delivery device. The white sections are etched 100 μm into
the silicon substrate to create a negative of the desired structure..............306
Figure 4-3 Mask used to create silicon master for the middle layer of the
drug delivery device. The white sections are etched 250 µm into
the silicon substrate to create a negative of the desired structure..............308
Figure 4-4 Pattern used to cut uniform reservoirs from molded PDMS sheet.........310
Figure 4-5 Mask used to create the metal alignment marks. This mask
patterns a photoresist layer. Metal is then deposited on the
photoresist; the photoresist is removed using acetone, isopropyl
alcohol, and water......................................................................................314
Figure 4-6 Mask used to pattern the bottom layer of the SU-8 valve seat and
pressure limiter. SU-8 is a negative resist, therefore, the white
areas indicate locations where SU-8 structures will remain, while
the black areas will be removed.................................................................315
Figure 4-7 Mask used to pattern the top layer of the SU-8 valve seat and
pressure limiter. SU-8 is a negative resist, therefore, the white
areas indicate locations where SU-8 structures will remain, while
the black areas will be removed.................................................................316
Figure 4-8 Mask used to pattern the SU-8 spacer plate. SU-8 is a negative
resist, therefore, the white areas indicate locations where SU-8
structures will remain, while the black areas will be removed..................320
Figure 4-9 Mask used to pattern the SU-8 mold used to cast the silicone
valve plate. SU-8 is a negative resist, therefore, the white areas
indicate locations where SU-8 structures will remain, while the
black areas will be removed. .....................................................................323

xxvi
Figure 4-10 Mask used to pattern the optional second layer for an SU-8
mold used to cast the silicone valve plate with bossed feature. SU-
8 is a negative resist, therefore, the white areas indicate locations
where SU-8 structures will remain, while the black areas will be
removed. ....................................................................................................326
Figure 4-11 Mask used to pattern the first layer for an SU-8 mold used to
cast the silicone valve plate with bossed and overhang feature. The
thickness of this layer determines how far the overhang will extend
beyond the valve plate. SU-8 is a negative resist, therefore, the
white areas indicate locations where SU-8 structures will remain,
while the black areas will be removed.......................................................330
Figure 4-12 Mask used to pattern the second layer for an SU-8 mold used to
cast the silicone valve plate with bossed and overhang feature.
This layer defines the bossed structure. SU-8 is a negative resist,
therefore, the white areas indicate locations where SU-8 structures
will remain, while the black areas will be removed. .................................331
Figure 4-13 Mask used to pattern the third layer for an SU-8 mold used to
cast the silicone valve plate with bossed and overhang feature.
This layer is used to define the thickness of the valve plate and the
shape of the valve plate arms (i.e. through-holes). SU-8 is a
negative resist, therefore, the white areas indicate locations where
SU-8 structures will remain, while the black areas will be removed. .......332
Figure 4-14 Three valve plate types: a) hole, b) straight arm, c) s-shaped
arm .............................................................................................................336
Figure 4-15 Steps to prepare valve plate..................................................................337
Figure 4-16 Items needed to packaging valve..........................................................337
Figure 4-17 Front and side views of SU-8 pieces....................................................338
Figure 4-18 Aligning pieces (top view) : valve plate only, valve plate with
valve seat, valve plate & valve seat & spacer plate, valve plate &
valve seat & spacer plate & pressure limiter .............................................338
Figure 4-19 Aligning pieces (side view) : valve plate only, valve plate with
valve seat, valve plate & valve seat & spacer plate, valve plate &
valve seat & spacer plate & pressure limiter .............................................338
Figure 4-20 Process needed to shrink FEP tubing around valve .............................339
Figure 4-21 Assembled valve in heat-shrink prior to placing in the oven...............339
xxvii
Figure 4-22 Side and top view of a packaged valve ................................................340
Figure 4-23 Testing setup to apply pressure to packaged valves or solid
disks. ..........................................................................................................341
Figure 4-24 The three custom-designed laser-machined layers that are
stacked to form the puncture jig. ...............................................................343
Figure 4-25 Corel Draw files to create puncture force jigs. Jig assembly is
also shown. The colors are just used to indicate corresponding
layers and are not present in the laser file..................................................344
Figure 4-26 Corel draw files used to create custom-made, laser-machined
molds of various shapes and sizes for the first version of the solid
surgical shams. The 0.75 mm to 2 mm labels indicate the
thickness of the sham.................................................................................345
Figure 4-27 File used to fabricate the redesigned solid surgical sham molds
(v2_large and v1_small). The dimensions are the same as in
version 1 with additional sutures on the 1 mm thick sham and the
sutures removed from the silicone cannula from both shams....................346
Figure 4-28 Drawing used to create solid sham mold v3_1. The mold is a
convex dome which was filled with silicone, leveled, and cured to
create a solid sham.....................................................................................347
Figure 4-29 Drawing used to create hollow sham molds v3_2, v3_3, and
v4_1. Shaded portions are etched to create domes or flat surfaces. .........348
Figure 4-30 Drawing used to create hollow sham molds v5_1 and v6_1.
Shaded portions are etched to create domes or flat surfaces. ....................349
Figure 4-31 Drawing used to create hollow sham mold v7. Shaded portions
are etched to create domes or flat surfaces. ...............................................350
Figure 4-32 Corel Draw file to create Parylene C deposition holder. 6
layers using this pattern were cut. The assembled holder is
modular and can be placed in the Parylene C deposition chamber
with up to all six layers in place. The Parylene C deposition holder
layers are separated using plastic standoffs. ..............................................357
Figure 4-33 Mask used to pattern photoresist to create a metal liftoff layer
for the single interconnect design..............................................................358
Figure 4-34 Mask used to pattern photoresist to create an etch mask for the
Parylene C and expose the metal electrodes and electrolysis
structure on the single interconnect deign. ................................................359
xxviii
Figure 4-35 Mask used to pattern SU-8 layer to create a 300 μm channel.
This produces a structure that is compatible with using a 33 gauge
needle to pierce the septum........................................................................360
Figure 4-36 Mask used to pattern SU-8 layer to create a 500 μm channel.
This produces a structure that is compatible with using a 30 gauge
needle to pierce the septum........................................................................361
Figure 4-37 Wafers with SU-8 microchannels, A) which do not contain
metal structures, and B) microchannels integrated with electrolysis
and conductance structures. Wafer B can also be fabricated
without metal creating additional SU-8 microchannel interconnects
without metal. The blue areas identify the septum locations....................362
Figure 4-38 Wafers with Parylene C microchannels, A) that do not contain
metal structures, and B) microchannels integrated with electrolysis
and conductance structures. Wafer B can also be fabricated
without metal creating additional SU-8 microchannel interconnects
without metal. The blue areas identify the septum locations....................363
Figure 4-39 Mask used to pattern SU-8 layer for the SU-8 microchannel
arrayed interconnects found in Figure 4-37A............................................364
Figure 4-40 Mask used to pattern photoresist to create a metal liftoff layer
for the SU-8 microchannel arrayed interconnects found in Figure
4-37B. ........................................................................................................365
Figure 4-41 Mask used to pattern photoresist to create an etch mask for the
Parylene C and expose the metal electrodes and electrolysis
structure on the SU-8 microchannel arrayed interconnects found in
Figure 4-37B..............................................................................................366
Figure 4-42 Mask used to pattern SU-8 layer for the SU-8 microchannel
arrayed interconnects found in Figure 4-37B. This mask can be
used without the masks in Figure 4-40 and Figure 4-41 to create
additional verions of the arrayed interconnect with SU-8
microchannels without metal.....................................................................367
Figure 4-43 Mask used to pattern photoresist to create a sacrificial
photoresist structure that defines the microchannel interior for the
Parylene C arrayed interconnect designs found in Figure 4-38A..............368
Figure 4-44 Mask used to pattern photoresist to create an etch mask for the
Parylene C covering the microchannel opening of the Parylene C
microchannel designs found in Figure 4-38A. ..........................................369
xxix
Figure 4-45 Mask used to pattern SU-8 layer for the Parylene C
microchannel arrayed interconnects found in Figure 4-38A. ....................370
Figure 4-46 Mask used to pattern photoresist to create an etch mask which
etches the Parylene C for the Parylene C microchannel arrayed
interconnects found in Figure 4-38B. The etched areas will allow
the electrodes of the metal structure to come into direct contact
with the glass substrate. .............................................................................371
Figure 4-47 Mask used to pattern photoresist to create a metal liftoff layer
for the Parylene C microchannel arrayed interconnects found in
Figure 4-38B..............................................................................................372
Figure 4-48 Mask used to pattern photoresist to create a sacrificial
photoresist structure that defines the microchannel interior for the
Parylene C arrayed interconnect designs found in Figure 4-38B..............373
Figure 4-49 Mask used to pattern photoresist to create an etch mask for the
Parylene C covering the microchannel openings, electrodes, and
electrolysis structures for Parylene C microchannel designs found
in Figure 4-38B..........................................................................................374
Figure 4-50 Mask used to pattern SU-8 layer for the Parylene C
microchannel arrayed interconnects found in Figure 4-38B This
mask, along with Figure 4-48 and Figure 4-49can be used without
the masks in Figure 4-46 and Figure 4-47 to create additional
verions of the arrayed interconnect with Parylene C microchannels
without metal. ............................................................................................375
Figure 4-51 Corel Draw file used to create custom-made, laser-machined,
jigs to measure insertion force of 4 and 8 needles arrays. Jig
assembly is also shown. Single needle insertion tests completed
using the 4 neede jig and aligning a single needle through one hole. .......383

xxx
List of Equations

Equation 2-1 Large deflection in a homogenous, thin film plate...............................78
Equation 2-2 Flexural rigidity equation to determine deflection in thin film
plate..............................................................................................................78
Equation 2-3 Electrochemical reaction during electrolysis of water .........................82
Equation 2-4- Volume of an Ellipsoid .....................................................................122
Equation 2-5- Volume of a Dome............................................................................122
Equation 3-1 Adhesion Force ..................................................................................188
Equation 3-2 Frictional Pull-Out Force ...................................................................188
Equation 3-3 Frictional Pull-Out Force Independent of Pressure............................189
Equation 3-4 Total Pull-Out Force...........................................................................189
Equation 3-5 Critical width to prevent stiction in cantilevers..................................218
Equation 3-6 Phase difference due to stress.............................................................257
Equation 3-7 Insertion Force of a Needle into a Membrane....................................260
Equation 3-8- Stiffness Force of a Membrane Deflecting .......................................261
Equation 3-9- Frictional Force of a Needle Through a Membrane..........................262
Equation 3-10- Cutting Force of a Needle Piercing a Membrane............................262
Equation 3-11 Insertion force equation....................................................................265
Equation 3-12 Removal force equation....................................................................265
Equation 3-13 Volume of gas generation during electrolysis..................................282
xxxi
Abstract

Presented in this work are two devices, an ocular drug delivery device with a dual-
regulation check valve, and an arrayed, horizontal microfluidic interconnect. Both
devices were designed to contain modular components; modularity will increase the
flexibility, versatility, and functionality of these devices.

The drug delivery device aims to provide a new method for mitigating the
progression of chronic retinal diseases. Each of the major device components
(refillable reservoir, bandpass regulating valve and cannula, electrically-actuated
pump, device base) are fabricated separately; several devices can therefore be
assembled in parallel. These components are easily replaced with an identical part,
or exchanged for an alternate design. Alternate designs lead to a multitude of
additional drug delivery applications.

The microfluidic interconnect provides a standardized method of connecting the
macro world to a micro one. Current interconnects are custom-made solutions which
are specific to each device. The interconnect presented allows multiple connections
to be made simultaneously while being simple to incorporated with existing devices.
The most unique aspect of a standardized interconnect scheme is opening the
possibility of “plug and play” microfluidics. Individual microfluidic elements (e.g.
microchambers, microchannels, mixers, heaters, etc) can be fabricated. These
xxxii
elements can then be connected on-demand to form a complete system. “Plug and
play” microfluidics makes available any system which can be constructed from the
basic microfluidic building blocks without necessitating access to a cleanroom.
Additionally, systems can be expanded, simplified, or altered simply by making
adjustments to the number and order of the connected components.

1
Chapter 1 Introduction

The field of bioMEMS (bio Microelectromechanical Systems) has seen a dramatic
increase in research and applications in recent years. The merging of traditional
MEMS fabrication techniques with unique medical needs has helped this field grow.
Though bioMEMS devices have proven useful in research and real-world situations,
many bioMEMS devices are designed for a single application, and because of their
complete wafer-level process, are difficult to alter without requiring an entirely new
fabrication process. Additionally, non-modular designs lack the flexibility and
versatility to allow MEMS designs from being easily adapted to new applications.
Modular devices provide several advantages such as creating interchangeable parts,
allowing flexibility in designed operation range and principle, simply “plug and
play” components which are easily replaced, and rapid prototyping.

Microelectromechanical Systems (MEMS) have many advantages over traditional
manufacturing processes. First, MEMS devices can be fabricated with a high degree
of precision (micron scale, 10
-6
m). Additionally, due to their miniature size, many
devices can be batch fabricated on a single wafer during the fabrication process.
However, fabrication processes can be costly, and small modifications to a
component within the device often require an entire redesign of the entire process.

2
MEMS fabrication processes are costly in finances, time, and resources. In a
research setting, a process run may require several masks, each can cost a couple
hundred dollars, whereas, commercial processes may require masks which are
several thousands of dollars apiece. Additional costs may also include device
materials, especially of parts, components, or devices cannot be reused in the new
design, as well as the additional use and time in cleanroom facilities. Furthermore,
redesigning masks and re-characterizing changes to the process run are very time
consuming tasks, which can affect research or delivery deadlines. Finally, additional
man-hours and equipment availability must be considered. Modular designs, which
allow for increased flexibility and versatility of designs, can mitigate some of the
additional time and cost requirements. Modular designs provide opportunities to
accommodate component changes, extend device applications, and aid in device
assembly.

Modularity can be defined in a number of ways. It can refer to replacing an entire
component within a device, similar to a “plug and play” model, or it can mean
replacing a part within a component. Modular devices can also be defined as scaling
the design by repeating a design feature or altering the dimensions of a feature or
part. Each of these definitions is not mutually exclusive and has a common theme of
increasing device functionality and adaptability.

Device designs which allow entire components to be easily replaced allows for quick
repairs without having to fabricate an entirely new device. Additionally, once a
3
working device is established, components can be swapped for to verify specific
component functionality, which can be used to help debug and identify faulty
components.

Replacing a part within a component has several advantages. First, MEMS
fabrication lends itself to creating multiple identical copies of a specific part. These
parts are all interchangeable; interchangeable parts are essential to the assembly-line
fabrication process, which allows for inexpensive mass production of reliable
devices. Changing a single part, which is different in dimensions, shape or materials,
from the replaced part can result in a component with a different operation. This
change results in a new component, which can be tailored to fit a separate
application, without having to redesign the entire device. Rapid prototyping of new
devices can be accomplished by changing small parts within device components.

Finally, modular designs may scale the current design either by changing the
dimensions of a component or part, or repeating a desired element. Scaling by
scaling either by increasing a part/component’s dimensions can shift the operating
range of a device. Additionally, trade-offs based on need and available resources can
be optimized when the dimensions can be scaled. For example, a more robust design
may require more space; however, there is a finite amount of real-estate, available.
Therefore, once the space constraint is determined, the component can be scaled up
to fill all of the available space, resulting in the strongest device possible. The
second version of scaling modulation is to repeat elements. Repetition of specific
4
design elements means that a design can be extended from a single device to an
arrayed device, or even vice versa, where an arrayed device can be reduced to a
single device, as desired functionality dictates.

Presented in this work are two different bioMEMS designs that contain modular
systems. The first is an ocular drug delivery system (Section Chapter 2). Current
drug delivery systems include 1) biocompatible matrices infused with drugs and 2)
membrane coated reservoirs that release drugs upon membrane rupture. Modular
designs, including interchangeable parts, redesigned parts, and replaceable
components allowed for rapid prototyping and testing of designs in order to
accommodate physiological needs. Presented here is a manually-actuated prototype
device containing a refillable reservoir, a microchannel or cannula for targeted drug
delivery, a valve to control fluid flow, and suture tabs to secure the device in place.
Each component was tested on benchtop; the integrated device was tested using
acute in vitro and in vivo studies to demonstrate the proof-of-concept of a refillable
drug delivery device. The results from the manually-actuated device were used to
design and fabricate an electrically-actuated system with a refillable reservoir, refill
port, flexible cannula, modular dual regulation check valve, suture tabs, electrolysis
pump, and separate drug and pump chambers. Surgical shams were created to refine
the surgical protocol. Chronic in vivo tests (6 month studies) were conducted to
determine long-term biocompatibility of the device, and to demonstrate dispensation
and refill in vivo.

5
The second project is a device that can be used to connect the microfluidic world to
the macro or benchtop world (Section 0). Applications of microfluidic devices are
found in the biological and chemical research areas as well as lab-on-a-chip and
micro total analysis systems (µTAS). The connections (or interconnects) used to
access microsystem are generally custom-made solutions which are single-use,
require precision alignment, additional and complex fabrication steps, have increased
dead volume, and utilize adhesives to secure them in place. These challenges limit
interconnects from being batch fabricated or easily integrated into microfluidic
devices, thus lowering device yield and wide-spread implementation. A modular
interconnect, which can be incorporated into a wide variety of existing devices was
created. The design, fabrication, testing, and advantages of a reusable horizontal
interconnects is presented. Single interconnect versions were fabricated to
demonstrate proof-of-concept. The horizontal interconnect design was scaled and
extended to create an arrayed interconnect where multiple connections can be made
simultaneously. Finite element analysis of the insertion stresses as well as
experimental determination of insertion and removal forces were conducted.
Operating ranges of both the single and arrayed interconnect were determined and a
discussion of interconnect design improvements is presented.

6
Chapter 2 MEMS Based Drug Delivery Devices

2.1 Introduction

Targeted and precisely controlled delivery of therapeutic compounds into the body is
an ongoing challenge in many areas of drug delivery. Drug delivery to ocular tissues
is particularly difficult due to several factors including physiological barriers and
trauma to the eye during invasive therapies. Furthermore, drug delivery devices
need to be small in order to fit within the spatial limitations in and surrounding the
eye. Ocular drug delivery has been investigated as a treatment method for chronic
eye diseases such as glaucoma, age-related macular degeneration (AMD), diabetic
retinopathy, and retinitis pigmentosa. These diseases are the leading causes of
irreversible blindness (Geroski and Edelhauser 2000). Lifelong treatment in the
form of therapeutic medications and surgical procedures are necessary to slow or
prevent disease progression. Current methods of treatment include topically and
orally administered medications, intraocular injections, surgical intervention, and
biodegradable implants. However, limitations in current methods suggest a need for
a new solution to ocular drug delivery.

A manually-actuated drug delivery prototype device presented here is a first
generation device which is used to demonstrate the feasibility of a MEMS refillable
ocular drug delivery device. This prototype is capable of targeted delivery, refill,

7
and fluid flow control. It also allows individual device components to be quickly
integrated into a system where device components can be tested in vitro and in vivo.
These results will be used to design a second generation device (electrically-actuated
device with dual check valve) capable of targeted delivery, device refill/reusability,
fluid flow control under operating parameters, and precise dosage control.

2.1.1 Ocular Drug Delivery Methods

Current ocular drug delivery methods are categorized as, 1) topical or oral, 2)
intraocular injections, and 3) implants (Lee, et al. 2004). Concerns such as patient
compliance, invasive therapies, resulting side-effects, targeted delivery, and dosage
control must be considered when choosing a treatment method.

2.1.1.1 Topical and Oral Medications

Drug delivery to the anterior and posterior segments of the eye is especially difficult
due to physiological barriers. Non-invasive methods of delivery, including eye drops
and oral medications, must permeate through the modified mucosal membrane of the
cornea, or the blood-retina barrier, respectively. For eye drops, it has been reported
that only 5% of the dispensed drug may reach the anterior intraocular tissues through
the cornea (Geroski and Edelhauser 2000). Furthermore, drug dilution due to
lacrimation, tear drainage, and turnover limit the drug contact time with the cornea.
Oral medications can be used to treat diseases that affect the posterior ocular tissues;

8
however the blood-retina barrier significantly impedes the penetration of drugs.
Larger doses need to be administered in order to obtain therapeutic levels, however,
this may result in serious systemic side effects (Fraunfelder 1979, Fraunfelder 1980,
Fraunfelder 1977, Fraunfelder 1990, Fraunfelder and Meyer 1984, Fraunfelder 2004,
Fraunfelder and Fraunfelder 2004). While topical and oral medications are the
simplest and least invasive method of treatment, they rely on patient compliance and
self-dosing.

2.1.1.2 Intraocular Injections

Intraocular injections directly bypass physiological barriers and are an effective
method for delivering precise dosages into the ocular space. However, the limited
half-life of drugs in the vitreous cavity necessitates frequent injections (1-3 per
week) for disease management (Lee, et al. 2004). This method of treatment is a
viable solution for bacterial infections, such as endophthalmitis, but is not suitable
for chronic diseases such AMD or cytomegalovirus (Mamalis, et al. 2002). In
addition, patient acceptance of this method is poor. Frequent injections also may
result in trauma to the eye tissues that can lead to cataracts, vitreous hemorrhage, and
retinal detachment (Ambati, et al. 2000).

9
2.1.1.3 Implants

Due to the limitations of currently available drug delivery methods, there is a need
for advanced drug delivery systems that can provide both accurate and targeted
dosing. Ideal drug delivery devices are biocompatible, minimize trauma and
inflammation, provide localized delivery with minimal exposure to other tissues,
provide sustained therapy (i.e. can be refilled), and are placed to facilitate medical
inspection without impeding the vision of the patient (Metrikin and Anand 1994).
The system should have broad drug compatibility and have a small, minimally-
invasive form factor; limited real-estate exists for comfortable placement of the
device within the ocular orbit. The device must be able to reliably dose medication
under normal intraocular pressure (IOP) conditions (15.5 ± 2.6 mmHg (mean ± SD),
2.06 ± 0.35 kPa) (Ritch, et al. 1989), elevated conditions caused by diseases such as
glaucoma (IOP > 22 mmHg, 2.93 kPa) (Wilensky 1999), and transient conditions of
fluctuating pressures due to outside influences such as patient sneezing, eye rubbing,
or changes in altitude (e.g. flying).

Existing drug delivery devices can be categorized as (1) biodegradable or non-
biodegradable (2) implantable pump systems, and (3) atypical implantable systems.
Biodegradable and non-biodegradable systems typically consist of a polymer matrix
infused with drug or a reservoir containing drug, respectively. The drug is released
as the matrix dissolves or as the drug diffuses from the non-biodegradable reservoir.
This form of drug delivery is dependent on drug load within the polymer and in vivo
polymer degradation; this method has limited volume and cannot be refilled.

10
Implantable pump systems provide control over delivery rate and volume by using
an active pumping mechanism to dispense drug from an internal reservoir. Five
types of implantable pump systems have been investigated: infusion pumps,
peristaltic pumps, osmotic pumps, positive displacement pumps, and controlled
release micropumps. Atypical systems provide targeted delivery at a constant rate
and reduce the amount of drug necessary for treatment (Dash and Cudworth 1998).
For example, hydrogel systems consisting of a polymer matrix infused with drug
swell due to the absorption of biological fluids. Drug release occurs proportionally
to the rate of swelling. Existing ocular drug delivery devices fall into the first or last
category, while implantable pump systems have been successfully executed for other
drug delivery needs (e.g. insulin delivery, neurological compounds) (Grayson, et al.
2004, Razzacki, et al. 2004, Ziaie, et al. 2004). Two commercially available devices,
Vitrasert® and Retisert™, distributed by Bausch and Lomb utilize this method of
delivery. The Vitrasert® and Retisert™ have a reported lifetime of 5-8 and 30
months, respectively. However, commercial sustained ocular drug delivery implants
require repeated surgical interventions to implant and replace the device resulting in
similar side effects to those found in injection therapies.

2.2 A MEMS Approach to Drug Delivery: Manually-
Actuated Device

A drug delivery device capable of providing targeted and precisely controlled
delivery to intraocular tissues can be fabricated using microelectromechanical

11
systems (MEMS) techniques. This device will be an improvement over current
sustained-delivery devices because a MEMS device (1) utilizes techniques that can
miniaturize the system to fit within the space constraints, (2) may contain valving
and electronics which are able to control the dosage and provide custom drug
regiments, and (3) can be refilled. The polymer MEMS delivery device consists of a
refillable drug reservoir, transscleral cannula, check valve, support posts, and suture
tabs. This is the first MEMS drug delivery device to feature a refillable reservoir
(Lo, et al. 2006, Lo, et al. 2009). A refillable reservoir provides several advantages
over existing ocular drug delivery systems: (1) it extends the usable lifetime of the
device without increasing the device size; (2) it can be replenished without
significant surgery as opposed to sustained implant devices that need to be removed
and re-implanted; and (3) it can take advantage of newly available pharmaceutical
solutions simply by changing the drug solution within the reservoir. Additionally,
the device is assembled using interchangeable parts. Interchangeable parts allow all
of the parts to be fabricated in parallel and then assembled.

The device is surgically implanted with the reservoir placed underneath the
conjunctiva, a membrane surrounding the eye (Figure 2-1C). The cannula is inserted
through the eye wall with the drug dispensing tip terminating in either the anterior or
posterior segment depending on the site of treatment (Figure 2-1D). A specific dose
of medication is dispensed from the device when the reservoir is mechanically
actuated by the patient’s finger (Figure 2-1E). The increase in pressure within the
reservoir forces the drug to travel down the enclosed microchannel within the

12
transscleral cannula. A flow-regulating check valve (one-way valve) was
incorporated near the tip of the cannula. The check valve responds to the pressure
increase and opens, allowing drug to flow out of the device but prevents back flow of
bodily fluids into the device (Figure 2-1F). Support posts are contained within the
microchannel and the reservoir to prevent stiction following the collapse of the
device walls when the drug is depleted. When the reservoir has been depleted,
medical personnel can, in a minimally invasive manner, refill the device using a
standard needle and syringe (Figure 2-1G,H).

Figure 2-1 Illustration of device functionality A) Device is comprised if 3 molded silicone layers. B) Layers are assembled and bonded to form device which
contains a refillable reservoir, flexible cannula, suture tabs, support posts, and check valve. C) Device is sutured to the sclera. D) Flexible cannula is inserted
into the anterior or posterior segments of the eye via a scleral tunnel; device is covered by the conjunctiva (not shown). E) The patient manually-actuates the
device by pressing on the reservoir with their finger. F) The change in device volume causes an internal pressure to build up until the check valve opens and
fluid is expelled from the device into the eye interior. G) After several dispensing events, the device is depleted. H) A surgeon can, in a minimally invasive
manner, refill the device using a 30G (O.D. 305 μm) non-coring needle. Figure adapted from images courtesy of Tun Min Soe.

13

14
2.2.1 Device Design

The drug delivery device is formed from three layers of molded
polydimethylsiloxane, PDMS, (Sylgard 184, Dow Corning, Midland, MI) (Figure
2-1A, Figure 2-2). The bottom layer defines the base of the device and includes
suture tabs through which sutures are threaded to secure the device to the eye. This
layer also contains support posts to prevent stiction of device walls when the drug
has been depleted. The final support post at the tip of the cannula also serves as the
check valve seat. The middle layer forms a portion of the cannula and the check
valve orifice. The topmost layer completes the refillable reservoir and defines the
maximum drug volume that can be housed. Multiple identical copies of each layer
can be fabricated simultaneously, providing interchangeable components.
Additionally, alternate top layer pieces can be fabricated to increase or decrease the
interior volume without requiring new bottom or middle pieces.

Figure 2-2 Exploded view of the drug delivery device. The device is comprised of three layers
(bottom, middle, and top) which define the components of the device.

2.2.2 The Components

2.2.2.1 Refillable Reservoir and Refill Guides

The total device size and the reservoir volume, is limited by the available
subconjunctival area and the space in between the rectus muscles (Figure 2-3).
Patient comfort must also be considered. The internal volume is defined by the
desired volume per dose and frequency of the dosage. Ophthalmic surgeons estimate
that the entire device should measure no more than 2-3 mm in thickness to maximize
patient comfort.

15

Figure 2-3 Placement of the drug delivery device. Note, conjunctiva is not shown. Figure adapted
from image courtesy of Tun Min Soe.

When the drug contained within the reservoir is depleted, the reservoir can be
refilled with a non-coring needle. Device refill is aided by the addition of a needle
refill guide. The refill guide limits the needle insertion depth to prevent the needle
from penetrating the entire device and entering the eye and to prevent the needle tip
from becoming embedded into the bottom wall of the reservoir (Figure 2-4). If the
tip becomes embedded in the reservoir wall, the needle lumen may be occluded and
prevent the drug from being dispensed into the reservoir. The refill guide is a PDMS
ring affixed to the needle shaft (Figure 2-5). The ring is set at the maximum depth
the needle can penetrate into the device. The maximum depth is determined by the
combined values of the reservoir wall thickness and the height of the interior
reservoir volume.

16

Figure 2-4 Illustration of the refill ring used to prevent the needle from penetrating through the base
of the device.

Figure 2-5 Image of several ring guides placed along a 30 gauge needle. For application, only one
ring guide per need is necessary.

2.2.2.2 Cannula and Check Valve

The cannula and integrated check valve cross the eye wall and enable targeted
delivery to intraocular tissues either in the anterior or posterior segment. The
cannula dimensions are determined by both ocular anatomy and surgical
considerations. The cannula tip must not obscure the visual pathway (i.e. pupil) and
the cannula should not come into contact with the cornea. These constraints limit the
length of the cannula. The width and height of the cannula should be no more than 1
mm (0.04 inches) which corresponds to the maximum incision size the eye is able to
self-seal. The cannula measures 10 mm x 1 mm x 1 mm (0.4’’ x 0.04’’ x 0.04’’).
17

The internal dimensions of the cannula were chosen so as to minimize flow
resistance (cross-section of 0.5 mm x 0.1 mm or 0.02’’ x 0.004’’).

A check valve is a microfluidic component that serves as a flow restrictor within the
system. The integrated check valve is a normally-closed one-way valve. A
normally-closed valve remains closed until enough pressure is generated to open the
valve. This opening pressure is known as the cracking pressure for the valve; the
valve opens when the applied pressure exceeds the cracking pressure. If the external
pressure on the valve is higher than the internal pressure, the valve remains closed
(Figure 2-6). This prevents bodily fluids from entering the device and contaminating
the drug. A check valve directs one-way flow, preventing backflow in a system.

Figure 2-6 Check valve operations for forward and reverse pressure.

A post located at the tip of the cannula serves as the valve seat. The check valve is
formed by aligning a 203 μm diameter orifice over the valve seat at the end of the
cannula (Figure 2-7).

18

Figure 2-7 Image of the assembled check valve. Dyed liquid is used to provide contrast.

2.2.2.3 Support Posts

Support posts resembling extruded square pillars (0.4 mm x 0.4 mm x 0.1 mm), are
located along the interior length of the cannula and in the reservoir. The support
posts prevent PDMS stiction within the cannula and reservoir.

2.2.2.4 Suture Tabs

Four suture tabs provide surgeons a means for anchoring the device to the eye.
Suture tabs are circular in shape (O.D. 2 mm) and placed at each corner of the
reservoir so that the anchoring sutures will not accidentally occlude or damage the
cannula (Figure 2-2).

19

20
2.2.3 Device Fabrication

The device was fabricated by combining three separate layers of molded PDMS.
The PDMS was molded using either a silicon or acrylic master. Once the pieces
were fabricated and separated, the individual layers were assembled and bonded to
create the final device.

2.2.3.1 The Silicon and Acrylic Masters

2.2.3.1.1 Silicon Masters

The master molds for the bottom and middle layers were constructed from silicon
wafers. Silicon was used as the substrate because etching fine structures into the
silicon is well understood and controllable using available etching techniques.

First, 4” silicon wafers were dehydration baked at 100 °C for at least 30 minutes
(Figure 2-8A). The wafers were vapor coated with hexamethyldisilazane (HMDS)
adhesion promoter. Photoresist (AZ 4620, AZ Electronic Materials, Branchburg,
NJ) was spin coated at 2 krpm for 40 seconds to obtain a 10 μm layer (Figure 2-8B).
After the photoresist is exposed and patterned, native oxide (SiO
2
) was removed with
a 10% hydrofluoric acid (HF) dip (Figure 2-8C). 100 and 250 μm etch depths for the
bottom and middle molds, respectively, were achieved using deep reactive ion
etching (DRIE) (PlasmaTherm SLR-770B, Unaxis Corporation, St. Petersburg, FL)
(Figure 2-8D). Photoresist was removed using acetone, isopropyl alcohol (IPA), and

deionized (DI) water (Figure 2-8E). Then the wafers were cleaned using oxygen
plasma (400 mTorr, 400 W, 4 minutes) (PEIIA, Technics Plasma, Kirchheim,
Germany).

Figure 2-8 Cross-section of fabrication process to create silicon masters.

The silicon masters were cleaned using the RCA standard-clean-1 (SC-1) process
(5:1:1 DI H
2
O:H
2
O
2
:NH
4
OH) to remove any organic compounds. The masters were
then coated with approximately 5 μm of vapor deposited Parylene C (Specialty
Coating Systems, Inc., Indianapolis, IN) (Figure 2-8F, Figure 2-9). This release
layer facilitates the removal of the PDMS replica from the mold. A list of the
process steps to fabricate silicon masters can be found in Appendix A.
21

Figure 2-9 Image of a silicon mold used to create the (a) bottom and (b) middle layers for the drug
delivery device. The silicon masters were coated in Parylene C to facilitate mold release of the PDMS
layer from the master.

Once the silicon master was processed, it was mounted on a glass substrate in order
to prepare for replicating a PDMS bottom layer. First, a 5’’ x 5’’ (127 mm x 127
mm) glass plate was cleaned using isopropyl alcohol. The plate was placed on a
hotplate (90 °C). The value of the hotplate cannot exceed 120 °C or it may cause
thermal degradation of the Parylene C on the wafer. A few grams of paraffin wax is
placed on the center of the glass plate and allowed to melt. When the wax is fully
melted, the silicon wafer is placed (patterned side up) on the wax. Firm pressure is
applied on unpatterned locations on the wafer in order to ensure the wafer is flush
against the glass. Any excess wax that seeps out from under the wafer is removed.
The hotplate was turned off and the entire setup was allowed to slowly return to
room temperature. The steps to mount the master can be found in Appendix B.

Middle and bottom layers of the device were created using soft-lithography
techniques. Soft-lithography is a technique used to mold ductile materials with a
micromachined mold. One of the most common uses of soft-lithography is to mold
PDMS prepolymer using a silicon master. PDMS (Sylgard 184, Dow Corning,
Midland, MI) was mixed (10:1 base to curing agent ratio; AR-250 Hybrid Mixer,
22

23
Thinky Corp., Tokyo, Japan) then poured and spread over the silicon master. The
PDMS was degassed in a vacuum oven (Model VO914A, Lindberg/ Blue, Asheville,
NC) and cured (90 °C for 1 hour). The molded sheet of PDMS was gently separated
from the silicon master. Individual device layers were dissected from the PDMS
sheet using a fine-tipped blade. A more detailed fabrication process for the bottom
and middle layers can be found in Sections 2.2.3.2.1- The Bottom Layer and
2.2.3.2.2- The Middle Layer, respectively.

2.2.3.1.2 Acrylic Master

The top layer was molded using a conventionally machined acrylic master. An
acrylic master was used because the dimensions for the top layer did not require the
precision of a silicon master. Furthermore, the size of the top layer changed
frequently, resulting in the need for a rapid and inexpensive method for creating the
mold.
The acrylic master is created by cutting a square piece with rounded corners (6 mm x
6 mm, 0.236’’ x 0.236’’) from a 0.8 mm (0.031’’) thick sheet of acrylic. The corners
were rounded to distribute the stresses which are concentrated in the corners of the
reservoir. The pieces were cut using a laser cutter (35 Watt Helix, Epilog Laser,
Golden, CO). The laser pattern is defined by a software package (CorelDraw);
vector cuts are represented as lines where etching locations are designated with
shaded shapes. The squares of acrylic and several glass slides (25.4 mm x 50.8 mm,
1’’ x 2’’) were cleaned using isopropyl alcohol. Five minute epoxy was used to affix

the squares on the glass slide 15 mm apart. A glass slide was used as the base of the
master in order to ensure a flat bonding surface for the PDMS replicas. The epoxy
was allowed to cure at room temperature for 24 hours. The file for the squares and
acrylic master fabrication process can be found Appendix C.

2.2.3.2 Layer Fabrication

The manually-actuated drug delivery device is comprised of three individually
structured layers of PDMS.

Figure 2-10 SolidWorks image of the three layers that comprise the drug delivery device. The entire
device is 17 mm in length. Note, suture tabs are not shown.

24

2.2.3.2.1 The Bottom Layer

The bottom layer forms the base of the device outlining the shape of the reservoir,
cannula and suture tabs. It also contains mechanical support posts and check valve
seat structures. This layer also defines the size of the microchannel within the
cannula. The support posts prevent flow restriction if the device were to collapse on
itself.

Figure 2-11 SolidWorks image of the bottom layer. Dimensions of the bottom layer are given [mm].
Note, suture tabs are not shown.

The bottom layer is fabricated using a silicon master (Figure 2-12C). The master
was created using traditional lithography and etching techniques to fabricate a
negative relief of the desired shape (Figure 2-12A, B, C). The fabrication process for
the silicon master is discussed in Section 2.2.3.1.1- Silicon Masters. The mask used
to create the bottom layer silicon master can be found in Appendix D.

25

26
Once the master was fabricated, the bottom layer for the device was replicated. First,
PDMS was prepared using a 10:1 base to curing agent ratio (Sylgard 184, Dow
Corning, Midland, MI). Approximately 15 grams of PDMS was poured onto the
center of the master. The master was tipped to manually spread the PDMS over the
entire surface of the wafer, ensuring PDMS coverage on each component (Figure
2-12D). Once all of the components have been covered, the wafer is tipped to allow
the excess PDMS to flow off of the wafer. The master is placed in a vacuum oven
(<30 mmHg (4 kPa), Model VO914A, Lindberg/ Blue, Asheville, NC) to degas the
PDMS. The degassed PDMS can then be cured at room temperature (24 hours) or
rapidly cured at 70 ºC for 30 minutes. Once the PDMS is cured, a razor blade is used
to cut along the edge of the master. A circular PDMS sheet containing 20 replicas of
the bottom later is carefully peeled from the master mold (Figure 2-12E). The sheet
is then transferred to a glass substrate where the individual replicas are cut from the
sheet (Figure 2-12F, G, H). The fabrication process steps can be found in Appendix
E.

Figure 2-12 Cross-sectional image of the fabrication process for the bottom layer silicon master and
individual silicone layer. The cross-section is taken through the line of symmetry (line indicated on
Figure 2-13).

Figure 2-13 Red line indicates location of cross-section image for Figure 2-12.

27

2.2.3.2.2 The Middle Layer

The middle layer defines the top cover to the cannula and check valve orifice.

Figure 2-14 SolidWorks image of the middle layer. Dimensions of the bottom layer are given [mm].

Fabricating the silicon master for the middle layer is identical to the bottom layer
except the silicon master is etched 250 μm deep using DRIE. The process is
described in Section 2.2.3.1.1- Silicon Masters. The mask for fabricating the silicon
master is shown in Appendix F.

Fabricating the middle layer component is similar to the bottom layer process
described in Section 2.2.3.2.1- The Bottom Layer. However, the check valve orifice
must be made in the component prior to cutting each replica from the PDMS sheet
(Figure 2-15G). The check valve orifice is made by puncturing the PDMS using a
33-gauge coring needle (O.D. 203 μm). Fabrication steps for the middle layer can be
found in Appendix E.
28

Figure 2-15 Cross-sectional image of the fabrication process for the middle layer silicon master and
individual silicone layer. The cross-section is taken through the line of symmetry of the middle layer
(line is indicated in Figure 2-16).

Figure 2-16 Red line indicates location of cross-section image for Figure 2-15.

29

2.2.3.2.3 The Top Layer

The top layer forms the reservoir. The reservoir interior volume is defined by the
shape and size of the acrylic mold. The top layer is prepared after the bottom and
middle layers have been assembled. The top and bottom layer assembly is discussed
in Section 2.2.3.3- Device Assembly.

Figure 2-17 SolidWorks image of top layer. The interior cavity of the top layer defines the reservoir
volume. Dimensions are indicated [mm].

To fabricate the top layer, PDMS is poured over the top layer acrylic mold (Figure
2-18B). The mold and PDMS prepolymer is placed into a vacuum oven to degas the
PDMS. The mold is then placed into an oven for 15 minutes at 70 °C. After 15
minutes, check to see if the PDMS has reached a “half-cured” state. To check if the
PDMS is “half-cured,” gently touch the surface of the PDMS with a mixing rod. If
the PDMS sticks to the rod and deforms slightly when the rod is removed, the PDMS
is “half-cured.” The “half-cured” state helps to attach the top layer to the middle
30

31
layer and creates a stronger bond between the layers. The procedure to create a
“half-cured” top layer can be found in Appendix G.

The mold is then aligned over the reservoir cutting pattern (Figure 4-4). A fine-
tipped blade is used to carefully cut a 1 mm border around the edge of the acrylic
square (Figure 2-18C). Nest, obtain a pre-assembled bottom/middle layer pair. The
bonding and assembly process for the bottom and middle pairs are described in
Sections 2.2.3.3.1- Cleaning, 2.2.3.3.2- Oxygen Plasma Bonding, and 2.2.3.3.3-
Bonding Top Layer to Middle and Bottom Layers. Using a pair of tweezers,
carefully lift the top layer from the mold (Figure 2-18D). Align the top layer over
the middle/bottom layer stack such that the edge of the reservoir interior matches the
interior edge of the middle piece. Place the assembled device into a Petri dish and
allow the reservoir to finish curing at room temperature for 24 hours. The
fabrication process and reservoir cutting pattern can be found in Appendix H.

Figure 2-18 Process for making drug delivery device reservoirs. A) Use epoxy to affix acrylic
squares onto a glass slide, B) Pour PDMS prepolymer onto acrylic mold and half-cure PDMS, C) Cut
reservoirs from molded PDMS piece. D) Remove reservoirs from mold, E) Reservoirs are ready for
assembly.

2.2.3.3 Device Assembly

After each layer is molded, the three layers must be assembled to form the complete
device (Figure 2-19). The bottom and middle layers are first cleaned using a
hydrochloric acid (HCl) solution. Next, the surface properties of the PDMS bottom
and middle layers are altered using oxygen plasma to promote bonding between the
pieces without the need for adhesives.

32

Figure 2-19 Illustration of how the three layers are fabricated and assembled to form the manually-
actuated drug delivery device.

2.2.3.3.1 Cleaning

Each PDMS bottom and middle layer piece is submerged in a 1:10 DI H
2
O:HCl
solution. This cleaning process removes particles or dust that may have accumulated
on the pieces after mold release. The piece is rinsed off using DI water and blown
dry. The cleaning process can be found in Appendix I.
33

2.2.3.3.2 Oxygen Plasma Bonding

Oxygen plasma was used to modify the surface of the PDMS pieces in order to assist
the bonding of the individual layers together. Pairs of bottom and middle layers
were placed on a glass slide (one pair per slide) with the desired bonding surface
facing up. The exposed side is modified using oxygen plasma (Figure 2-20).
Pressure, power, and time parameters for oxygen plasma were varied to find the
optimal settings which yield the strongest bonds. The most durable bonds were
found to form after a pressure of 100 mTorr and power of 100 W were applied for 45
seconds. A more detailed discussion of oxygen plasma bonding can be found in
Section 2.2.4.1.1- Oxygen Plasma Bonding Method.

Figure 2-20 Placement of pairs of pieces on glass slide to facilitate placement and alignment of layers
after oxygen plasma treatment.

Each pair needed to be assembled quickly because oxygen plasma modified PDMS
will revert back to the initial hydrophobic state. A polar liquid, such as ethanol, can
be used to slow the reversion (Duffy, et al. 1998). Ethanol also lubricates the pieces
and allows for easier alignment. The oxygen plasma treatment process is described
in Appendix J. The layers are assembled under a microscope to ensure the edges of
the layers and the check valve opening and valve seat are aligned.
34

35

2.2.3.3.3 Bonding Top Layer to Middle and Bottom Layers

The top layer of the device is added after the bottom and middle layers have been
bonded. Section 2.2.3.2.3- The Top Layer describes the process of bonding the top
layer to the bottom/middle layer assembly.

2.2.3.3.4 Reinforcing Layer

The reinforcing layer is a thin layer of PDMS that is cured around the assembled
device. This layer became necessary because oxygen plasma bonding is not robust
enough to withstand the stresses placed on the device during surgical handling. The
reinforcing layer is fabricated by placing an assembled device into a crystallization
dish with an inclined glass slide (Figure 2-21A). PDMS is poured around the device
until the edge of the device is fully covered, ensuring PDMS does not cover the
check valve (Figure 2-21B). The crystallization dish is then placed into the oven (70
°C, 1 hour) to cure the PDMS. Excess PDMS is removed from the device to obtain
the final shape of the reinforced device (Figure 2-21C).

Figure 2-21 Adding reinforcing layer to drug delivery device. A) Device is placed on an inclined
glass slide, B) PDMS prepolymer is poured around the device, covering the edge of the device but not
occluding the check valve opening, C) The device is removed from the slide and excess PDMS is cut
from the device.

2.2.3.4 Dimensions of Assembled Device

A summary of the device dimensions are found in
Table 2-1. Given a potential dosage range of between 50-250 nL per dose, the
device can hold between 1143-228 doses if the entire volume is depleted.

36

37
Table 2-1 Dimensions of Drug Delivery Device
Specification Value
Fabrication Material PDMS (Silicone Rubber)
External Wall of Reservoir Body (W x L) 7 mm x 7 mm
Internal Wall of Reservoir Body (W x L x H) 6 mm x 6 mm x 1.58 mm
Reservoir Internal Volume 57.15 µL
Reservoir
Support Pillars in Reservoir (W x L x H) 0.5 mm x 0.5 mm x 0.1 mm
Fabrication Material PDMS (Silicone Rubber)
External Wall of Tube (W x L) 1 mm x 10 mm
Internal Wall of Tube (W x L x H) 0.5 mm x 9.5 mm x 0.1 mm
Tube Volume 0.475 µL
Cannula
Support Pillars in Tube (W x L x H) 0.4 mm x 0.4 mm x 0.1 mm
Valve Orifice Diameter 203 µm
Valve
Valve Seat (W x L x H) 0.4 mm x 0.4 mm x 0.1 mm
17B
2.2.4 Benchtop Experiments- Methods and Results

2.2.4.1 PDMS Bonding

A method for bonding the individual layers together without using adhesives was
investigated. The bonding method needed to produce irreversible bonding and be
compatible with the PDMS material. Adhesives were not considered as the primary
method for bonding because adhesives can seep into the device and clog the
microchannel or the check valve.

Oxygen plasma and chemical treatments were evaluated as methods to bond PDMS
in both benchtop prototypes and surgical models. Bonding strength was qualitatively
measured by observing the fracture resistance of the bond to an applied shear force.
PDMS sample coupons were prepared (Figure 2-22A). Prior to oxygen plasma or
wet chemical treatment, PDMS coupons (20 mm x 10 mm) were cleaned using in a

0.01% HCl solution. Pairs of coupons were then exposed to oxygen plasma or wet
chemical treatment and assembled such that half of each coupon overlapped the
other coupon (Figure 2-22B,C). Shear force was applied by pulling on the non-
overlapping sections (tabs) of the coupons (Figure 2-22D).

Figure 2-22 Procedure used to qualitatively determine bond strength after oxygen plasma or wet
chemical treatment.

Quantitative measurements of bonding strength were evaluated by bonding two
pieces to form an enclosed reservoir (interior volume, 6 mm x 6 mm x 1.59 mm).
The enclosed reservoir was fabricated by bonding the top piece of the device (Figure
2-17) to a PDMS coupon. Again, samples were cleaned prior to treatment (Figure
2-23A, B, C). Pressurized water was introduced into the cavity until the bond failed
(Figure 2-23D).

38

Figure 2-23 Testing setup used to quantitatively measure bond strength after oxygen plasma or wet
chemical treatment.

2.2.4.1.1 Oxygen Plasma Bonding Method

Oxygen plasma treatment of PDMS results in a hydrophilic surface due to the
formation of silanol groups (Si-OH). When two treated surfaces are joined,
irreversible covalent bonds (Si-O-Si) form. The number of silanol groups formed
depends on conditions of the plasma treatment (duration, power, and pressure).
Polar solutions such as water and ethanol can extend the reactive time of the silanol
groups and slows the return of the surface to its original hydrophobic state
(McDonald, et al. 2000). Baking the bonded sample can also increase the strength of
39

40
the bond (McDonald and Whitesides 2002). Oxygen plasma power, pressure, and
duration were examined to determine the optimal conditions for maximum bonding
strength. Water and ethanol were used to facilitate alignment prior to bonding.
Variations in baking time and temperature were also examined.

2.2.4.1.2 Wet Chemical Bonding Method

Chemical treatment of PDMS to promote bonding was also evaluated. Samples of
PDMS were immersed in 0.012 M HCl. Samples were removed and rinsed with
ethanol, blown dry, and assembled. Baking temperature, baking time, and bond
compression during baking were examined.

2.2.4.1.3 Results

Bonds created following oxygen plasma were found to be more reliable than those
created following wet chemical treatment. Optimized oxygen plasma conditions
were empirically determined to be 100 W, 100 mTorr, and 45 seconds. Once the
surfaces of the layers were treated, the layers were aligned with the aid of a polar
liquid (e.g. ethanol or water). The entire assembly was then baked at 100 °C for 45
minutes in order to complete the bonding process (Duffy, et al. 1998).

The device was originally assembled using oxygen plasma treatment to bond all
three layers. However, the bond strength was insufficient to survive handling during
surgical implantation of the device. For example, the surgical procedure required

41
bending the cannula back on itself at 180° to insert it into the eye. The bonds
connecting the two layers that formed the cannula were observed to fail during this
procedure. Additionally, leakage from the reservoir was observed. Failure points of
the device may have resulted from the hand assembly method to align the layers
during bonding. Each layer has a 1 mm border which is used as the bonding surface.
However, a slight misalignment will result in a weaker bond if the full 1 mm surface
is not utilized. Weaker bonds may also be in part due to the time elapsed between
treatment and bonding. The strongest bonds are formed within 1 minute of
treatment, however, due to the limitations of the oxygen plasma machine, the layers
may have been assembled outside of this 1 minute window.

It was observed that half-cured PDMS retained some of the malleable properties of
the PDMS prepolymer. It maintained the basic shape but was tackier than the fully
cured PDMS. Half-cured PDMS also created a strong irreversible bond when placed
on a fully cured piece of PDMS. The half-cured reservoirs bonded more strongly to
the bottom and middle layer stack than oxygen plasma treated bonds. Half-cured
bottom and middle layers could not be used because the individual replicas could not
be separated from the thin PDMS sheets. The half-cured thin PDMS replicas could
not be aligned because they did not have sufficient structural integrity to maintain
their shape.

The device assembly process was altered to take advantage of half-cured material
and PDMS prepolymer to strengthen the bonds between the layers in order to survive

surgical manipulation. The bonding between the bottom and middle layers
continued to use oxygen plasma treatment, a half-cured top layer was added. The
assembled device was unable to survive the stresses from surgical handling (Figure
2-24A), therefore, the assembled device was strengthened using a reinforcing layer
of PDMS prepolymer, as described in Section 2.2.3.3.4-Reinforcing Layer (Figure
2-24B). The reinforcing layer was trimmed to remove excess material and to create
suture tabs (Figure 2-24C). The reinforced device was able to survive extensive
surgical handling and did not fail during benchtop testing or implantation.

Figure 2-24 A) Implanted in vivo device (plasma bonded) with bond failure location due to surgical
handling identified. B) Reinforcing layer added to bonded device in order to provide more
mechanical robustness to the device. C) Excess silicone was removed from the device to create the
desired device outline, ruler divisions measure 1 mm.

2.2.4.2 Check Valve

2.2.4.2.1 Benchtop Operation

The check valve for the device was tested on benchtop to verify its functionality.
The device was filled with dyed DI water. The dyed liquid was observed moving
through the cannula and out of the check valve. When pressure was removed from
42

the reservoir the droplet of dispensed liquid remained outside of the check valve and
did not backflow into the device. Time-elapsed photographs were taken of the check
valve operation (Figure 2-25).

Figure 2-25 Time-lapsed photographs of dyed DI water being dispensed from the check valve under
manual actuation.

2.2.4.2.2 Characterization

2.2.4.2.2.1 Check Valve Opening and Closing Pressures

2.2.4.2.2.1.1 Methods

The check valve can be characterized by determining the cracking pressure and
generating a pressure versus flow rate curve over a range of pressures. The check
valve is analogous to a diode in the electrical world. A current versus voltage curve
can be generated for a diode where current is measured as voltage is applied to the
diode. The ideal diode has no current output when zero voltage is applied. When
43

voltage is positive, the current varies linearly with respect to voltage; when the
voltage is negative there is no reverse current. Similarly, an ideal check valve can be
characterized with a pressure versus flow rate curve, where pressure is applied and
the resulting flow rate is measured. Again, in the ideal case, the valve remains
closed when no pressure is applied, and opens and allows linear fluid flow with
respect to pressure. Under reserve pressure the valve remains closed.

Figure 2-26 Typical Diode Current versus Voltage Curve

44

Figure 2-27 Typical Check Valve Pressure versus Flow Rate Curve.

However, both the diode and check valve do not operate under ideal conditions. The
diode conducts voltage after a bias voltage threshold is achieved. Furthermore, a
negative voltage will result in leakage current; a large negative voltage (or
breakdown voltage) will cause the diode to fail (Figure 2-26). In the case of the non-
ideal check valve, the valve remains closed until the pressure exceeds the cracking
pressure of the valve. For reverse pressures, a small amount of leakage may occur
prior to the valve sealing closed; a large reverse pressure will cause irreversible
damage to and failure of the check valve (Figure 2-27).

Operating characteristics of the check valve, such as pressure versus flow rate curve
and cracking pressure, were obtained by aligning the bottom and middle layers and
clamping the structure in a custom laser-machined testing apparatus and attaching
the fixture to a water chamber pressurized by a nitrogen cylinder (Figure 2-28A).
Pressurized water is forced through the delivery tube and out of the check valve. A
45

calibrated pipette (50 μL, Clay Adams, Parsippany, NJ) was attached to the output of
the check valve (Figure 2-28B). Flow rate is measured by timing fluid passage
through a precision calibrated pipette. The pipette was primed by pre-filling it with
water. A small air bubble was introduced at the input of the pipette. Small
increments of pressurized water were applied to the check valve and the bubble
position was monitored. After each pressure increase, the system was held for five
minutes at a constant pressure to allow the system to equilibrate. The cracking
pressure of the check valve is identified as the lowest pressure required that resulted
in visual confirmation of bubble movement over a five minute period. The cracking
pressure was recorded. The flow rate was determined by applying pressures larger
than the cracking pressure and measuring the elapsed time to move an air bubble
along the entire length of the 50 μL pipette.

Figure 2-28- A) Exploded SolidWorks view of the custom-made laser-machined jig to characterize
the check valve operation. B) Pressure setup used to open the valve with pressurized water.

46

47
2.2.4.2.2.1.2 Results

The integrated check valve was characterized to determine the cracking pressure and
forward flow rates. Cracking pressure was determined to be 62 kPa (465 mmHg)
(Lo, et al. 2006). The cracking pressure is much higher than normal pressures found
within the eye. A higher opening pressure than normal intraocular pressures is
necessary to prevent the device from accidental dispensing due to normal activities
that may put pressure on the device, such as sneezing or rubbing the eyes. In this
case, the high cracking pressure may be partially attributed to the flow resistance of
microchannel. The microchannel is 500 μm wide; however, the microchannel is
narrower at the support posts. The support post is 400 μm x 400μm, leaving only 50
μm on either side of the support post for fluid flow. Another component that adds
resistance to the system is the check valve. The check valve is formed by
positioning a 203 μm diameter hole above a 400 μm x 400 μm square. The
overlapping area between the two layers is approximately 0.127 mm
2
. This overlap
area may have been weakly bonded together during the oxygen plasma treatment
step. The bond may have artificially created an elevated cracking pressure because a
higher pressure is needed to break the bond in order to get the check valve to
function. Furthermore, variations in the cracking pressure may also be inherent to
the PDMS material.

However, these device or material properties can be exploited to modify the valve
design to fit a target operation within a specified pressure range if necessary. PDMS

stiffness can be changed due to differences in base: curing agent ratio and curing
temperature/time. Additionally, the check valve orifice diameter can be increase or
decrease; which can be accomplished by using a smaller or larger gauge coring
needle to create the orifice, respectively. A smaller orifice will have a greater
overlap area with the valve seat (shifting the cracking pressure higher), and greater
flow resistance. A larger orifice will have the opposite effect.

Flow rate increased when the pressure on the system was increased. The range of
actuation pressures was determined by the range of interest for ocular dispensation as
well as the limits of the pressure testing system. A maximum flow rate of 321.29
μL/min was obtained at a pressure of 290.61 kPa (2179.78 mmHg) (Figure 2-29).

Figure 2-29 Flow Rate vs. Pressure curve for check valve (mean ± SE, n = 4).
48

49
2.2.4.2.2.2 Check Valve Closing Time Constant

2.2.4.2.2.2.1 Methods

Check valve dosing while varying the applied pressure and the duration of the
pressure was also measured. An electronically-controlled solenoid valve
(LHDA0533115H, The Lee Company, Westbrook, CT) was used to vary the
duration of the applied pressure to simulate a patient’s finger. A square wave control
signal with a 50% duty cycle and frequencies ranging from 18.5 mHz to 500 mHz
was used to control a pressurized water source. In the “on” state the solenoid
permits pressurized water to be passed onto the jig. The volume of fluid exiting the
valve is measured by noting the distance an air bubble traveled in a calibrated pipette
(100 μL) over the “on” and “off” portion of the wave. The closing time constant for
the valve was determined by measuring the elapsed time between removing the
pressure from the valve and when the accumulated volume exiting the valve reached
63.2% of the final volume.

2.2.4.2.2.2.2 Results

Dosed volume was measured for two different applied pressures 250 mmHg (33.3
kPa) and 500 mmHg (66.7 kPa) controlled using a 50% duty cycle square wave
control signal in the frequency range 18.5 mHz to 500 mHz (53 to 2 sec periods).
Dosed volume and pressure duration were found to be linearly proportional for both
applied pressures, resulting in a consistent flow rate independent of the dosing period

(Figure 2-30). The steady state flow rates were 1.57 ± 0.2 µL/sec and 0.61 ± 0.2
µL/sec (mean ± SE, n = 4) at 500 mmHg and 250 mmHg (66.7 kPa and 33.3 kPa),
respectively.

Figure 2-30 Check valve control of dosing under 250 mmHg and 500 mmHg (33.3 kPa and 66.7 kPa)
of applied pressure. Duration of applied pressure was varied using a solenoid valve controlled using a
50% duty cycle square waves.

Due to the finite closing time of the valve, flow was observed after removal of the
pressure source. The time constant associated with valve closing was found to be
10.2 sec for 500 mmHg (66.7 kPa) and 14.2 sec for 250 mmHg (33.3 kPa) (Figure
2-31). Dispensed volume and closing time calculations could not be measured in
real-time, therefore these data were extracted from video footage of the air bubble
moving through the calibrated pipette. Dosed volume was calculated using initial
50

51
and final bubble positions in the pipette and the applied pressure duration was
calculated from the time stamps in the video. The closing time constant was
calculated by noting the bubble location starting from when pressure was removed
and at specific time intervals until the bubble movement stopped. The closing time
constant was calculated by determining the time at which the accumulated volume
reached 63.2% of the final value. The volume dispensed over the duration of closing
time is 3.5 μL and 6.3 μL for 250 mmHg and 500 mmHg (33.3 kPa and 66.7 kPa),
respectively. The long closing time constant can significantly increase the dosage
amount, especially for very small dosages. Additionally, the valve does not prevent
accidental dosing due to transient fluctuations in intraocular pressures (e.g. sneezing,
flying). Therefore, improvements in response time and overpressure protection will
be incorporated in future drug delivery device prototypes.

Figure 2-31 A representative graph depicting the volume dispensed after the applied pressure (250
mmHg and 500 mmHg, 33.3 kPa and 66.7 kPa) is removed from the valve. The dashed lines indicate
when the accumulated volume reached 63.2% of the final value, the time at which this point occurred
was defined as the closing time constant for the valve.

2.2.4.3 Refillability

2.2.4.3.1 Needle Determination

2.2.4.3.1.1 Methods

The device operating lifetime is closely linked to the ability to refill the reservoir and
the mechanical integrity of the punctured material. Therefore, a choice of refill
needle (e.g. type, gauge, etc) and maximum number of achievable refills must be
determined.
52

53
2.2.4.3.1.2

A commercially available, 30 gauge (305 μm in outer diameter, O.D.) needle was
used to refill the device. This size was selected as a trade-off between utilizing the
needle with the smallest outer diameter to maximum material lifetime and stiffs
enough to puncture the device without the needle buckling or bending. The needle
must pierce the conjunctiva and reservoir wall in order to replenish the reservoir
contents. A small needle size also allows the punctured conjunctiva to self-seal,
therefore avoiding the need for sutures.

Results

Two types of 30 gauge needles, coring and non-coring, were investigated to
determine the most suitable needle profile. The needle must be able to puncture the
PDMS reservoir but cause minimal damage to the material after removal. Both types
of needles were inserted through PDMS slabs. The needle tracks through the PDMS
and at the insertion sites were examined.

Scanning electron microscope (SEM) images of 30 gauge coring and non-coring
needles were taken to examine the needle tip and needle shaft. The coring needle
has a blunt tip with a circular cutting edge. The non-coring needle has a beveled tip
which tapers to a point (Figure 2-32A).

Both needles were pushed into and withdrawn from a sample piece of PDMS. The
resulting needle puncture entrance and needle track were imaged using an optical
microscope. A coring needle removed material as it was pushed through the sample.

When the needle was removed, a circular hole was formed at the puncture location
and a cylindrical shape was cut through the PDMS slab. The material cut from the
PDMS sample remains inside the needle and prevents liquid from being dispensed.
The non-coring needle displaces material as it moves through the sample and creates
a small tear. The material relaxes when the needle is removed and seals the tear
(Figure 2-32B, C). A non-coring needle was selected for refilling to maximize
device lifetime.

Figure 2-32 Refill needle determination, A) Coring versus non-coring 30 gauge (305 μm OD) needle
illustration and SEM images, B) Top view of needle track through punctured PDMS slab using each
needle, C) Side view of needle track through PDMS slab.

54

55
2.2.4.3.2.1
2.2.4.3.2 Maximum Puncture Events and Leakage After Puncture

Methods

PDMS membranes (250 μm and 670 μm thick) were evaluated for mechanical
integrity by testing for leaks after repeated punctures (or refill events). PDMS sheets
were obtained by spreading a thin layer of PDMS (10:1 base to curing agent ratio) on
a glass substrate and cured at 70 ˚C for 20 minutes. 0.5’’ x 0.5’’ (12.7 mm x 12.7
mm) square membranes were cut from the sheet; the procedure used to create the
membranes can be found in Appendix K. Two membrane thicknesses were tested to
determine leakage pressure as a function of material thickness.

The worst-case scenario for refill was investigated and implemented in the leakage
pressure study. The leakage pressure of a membrane punctured 8 times in different
locations within a circular (5 mm diameter) area was compared to the leakage
pressure of a membrane punctured 8 times in the same location using a 30G non-
coring needle. To ensure that repeated punctures were through the same point
membranes were mounted in a custom laser-machined acrylic jig. The jig contains a
small hole to align repeated needle punctures in the same location (Figure 2-33).
The method which resulted in a lower leakage pressure was used for the leakage
pressure test to show how the number of punctures (8, 12, and 24 punctures) affects
leakage pressure.

Figure 2-33 Exploded SolidWorks image of the custom-made laser machined jig used ensure multiple
puncture events pierce the membrane in the same location for worst-case scenario testing.

After puncturing, the membrane was transferred to the second jig in which the
puncture site was aligned between two reservoirs (Figure 2-34A). The input of the
jig was attached to a pressurized water system and the output was connected to a
calibrated pipette (50 μL, Clay Adams, Parsippany, NJ) (Figure 2-34B). Pressurized
water was applied in increments and water leakage, if any, was monitored by
tracking the position of small bubble in the water-filled calibrated pipette. The
leakage pressure is determined by the lowest pressure which causes the bubble to
move any visible distance over a five minute period. The burst pressure, pressure
required to rupture an unpunctured membrane, was also measured.
56

Figure 2-34 A) Exploded SolidWorks view of the custom-made laser-machined jig to apply pressure
to punctured membranes. B) Setup used to provide measure leakage pressure of punctured
membranes.

2.2.4.3.2.2 Results

Two needle puncture arrangements were examined to determine the spatial
arrangement impact on membrane strength. For the case where all punctures were
created at the same position on a 250 μm thick membrane, the resulting leakage
pressure after 8 punctures was 6.98 ± 0.73 kPa (n = 4) (52.36 ± 5.48 mmHg). The
second case involved puncturing a 250 μm membrane punctured 8 times in random
locations in a 5 mm x 5 mm square area and resulted in a leakage pressure of 8.14 ±
0.07 kPa (mean ± SE, n = 3) (61.02 ± 0.52 mmHg). This test verified that the needle
punctures in exactly the same position lowered the leakage pressure of the
membrane, and therefore is the worst-case scenario. This was likely associated with
an increase in puncture size as a result of repeated needle insertions. For the
randomly located puncture case, each puncture location only has one insertion event
57

58
and thus the needle track size was minimal. The aligned punctures method was used
to compare leakage pressure for 8, 12, and 24 punctures.

The maximum pressure a repeatedly punctured membrane can withstand without
leaking decreases as the number of punctures increases. The burst pressure for the
unpunctured 673 μm membrane (525.75 kPa, 3943.5 mmHg) was 5 times larger than
the leakage pressure for 8 punctures (106.2 kPa, 796.4 mmHg), while the burst
pressure for the unpunctured 250 μm membrane (393.5 kPa, 2951.1 mmHg) was 56
times larger than the 8 puncture leakage pressure (6.98 kPa, 52.4 mmHg). This
suggests that the additional area in thicker PDMS membranes provided more self-
sealing ability than thinner membranes.

For all of the repeated puncture tests, though the leakage pressure for both
membranes decreased with increasing number of punctures the thicker membrane
was able to withstand higher pressures prior. Also, the leakage pressure for both
membranes becomes constant for higher number of punctures (Figure 2-35). This
could be attributed to nominal or lack of additional damage to the needle track for
each subsequent needle puncture.

Figure 2-35 Leakage pressure for 250 μm and 673 μm thick membranes punctured 8, 12 and 24 times
through the same location with a 30G non-coring needle (n = 4).

2.2.4.3.3 Benchtop Refill and Dispensation

2.2.4.3.3.1 Methods

An assembled device was placed on a glass slide and filled using a standard 30 gauge
non-coring needle. Dyed DI water was injected into the device to help visualize
fluid flow. The device was filled until liquid was seen exiting the check valve. The
reservoir was manually-depressed to dispense the contents of the device. The refill
and dispensation of the device was repeated three times.

59

60
2.2.4.3.3.2 Results

Device functionality was verified. Refill and delivery were successfully completed
four times without failure. However, it should be noted that benchtop tests benefited
from the use of a glass slide to stop the progression of the needle. Also, the device
was placed on a solid surface, which provided more support to the device and aided
dispensation. When the device is implanted, the needle progression through the
device will not be stopped and may penetrate through the entire device.
Furthermore, ocular tissue is more compliant than a laboratory benchtop. The eye
will move within the ocular orbit and/or the tissue will deform when pressure is
applied to the device. This motion may absorb some of the pressure placed on the
device. Device modifications may be necessary once the device is tested in vitro and
in vivo.

Pressure was applied to the reservoir until the device stopped dispensing any more
liquid. However, it was noted that a non-negligible amount of liquid remained in the
reservoir. The sidewalls of the reservoir prevent the reservoir from being fully
depressed. Also, a cuboid reservoir results in dead volume at the reservoir corners.
Future versions of the reservoir shape need to be changed to a circular or oval shape
in order to minimize dead volume. A circular or oval reservoir shape would also be
beneficial in distributing the stress placed on the device during dispensation as
corners are locations for stress concentration and would be more prone to failure.

61
2.2.5 In Vivo and In Vitro Experiments- Methods and Results

2.2.5.1 Device Placement

2.2.5.1.1 Methods

The device placement and surgical procedure to affix the device and introduce the
cannula into the eye was determined in vitro using enucleated porcine eyes. Suture
tabs were tested to verify the strength and location of the tabs; the device was sutured
to the sclera. Surgeons recommended limbal incisions or scleral tunnel insertions to
insert the cannula into the anterior chamber of the eye. A limbal incision is a small
cut made at the border between the cornea and sclera to access the anterior chamber.
A scleral tunnel is a tunnel made 3 mm from the limbus and penetrates the eye wall.
The device was first sutured to the sclera, and then the cannula was inserted into the
eye via the scleral tunnel. Once the device was secured with the cannula entering the
anterior chamber, device dispensation and refill tests were conducted.

The anterior chamber of the eye was used for the in vitro and in vivo studies because
the cannula can be visualized through the cornea, as opposed to inserting the cannula
into the posterior segment of the eye. Future applications that require delivery to the
posterior segment can use a similar surgical procedure.

62
2.2.5.1.2 Results

Surgical studies determined that only two sutures, located on the front corners of the
device, were needed to secure the device to the sclera (white portion of the eye).
Furthermore, it was difficult to insert the cannula through the limbal incision. Once
inserted, the insertion route demonstrated two challenges, 1) dispensed liquid seeps
out of the incision, thus reducing the amount of time the liquid is in contact with
ocular tissues, and 2) the use of sutures to close the incision more tightly around the
cannula causes the cornea to wrinkle. A scleral tunnel was more successful in
introducing the cannula. During surgical manipulation, a device that was assembled
and bonded using only oxygen plasma assisted bonding failed due to layer
delamination.

A reinforcing layer was introduced to the fabrication process to increase device
robustness. Reinforced devices did not fail during implantation. Furthermore, it was
determined that reversing the surgical protocol to insert the cannula into the eye via a
scleral tunnel and then suturing the device in place places less stress on the cannula.

2.2.5.2 Device Functionality

2.2.5.2.1 In Vitro Delivery

63
2.2.5.2.1.1
2.2.5.2.1.2
Methods

The device was prefilled with dyed DI water. The cannula was inserted through a
scleral tunnel and the device was sutured to the sclera of an enucleated porcine eye.
Surgical forceps were used to manually depress the reservoir. A surgical microscope
with video capture abilities were used to observe dye dispensation into the eye.

Results

Dyed DI water can be seen entering the eye (Figure 2-36). The bloom of dispensed
liquid can be seen to increase in size as the reservoir is continually depressed.

Figure 2-36 Surgical verification of liquid delivery in vitro using the drug delivery device.

However, as predicted from the benchtop model, depressing the reservoir with a
cotton swab did not generate enough force to dispense the liquid. When the reservoir

64
2.2.5.2.2.1
2.2.5.2.2.2
was depressed, the eye and device shifted, preventing the surgeon from putting
constant force on the device. The liquid was dispensed by squeezing the device with
surgical forceps. While using surgical forceps was acceptable in generating the
pressure required for liquid dispensation, this method cannot be used in a practice.

2.2.5.2.2 Device Refill

Methods

The device was prefilled with dyed DI water, the cannula inserted into the anterior
chamber via a scleral tunnel and the device was sutured to an enucleated porcine eye.
The dye was dispensed into the eye by depressing the chamber with surgical forceps.
Once the internal volume of the reservoir was significantly depleted, a 30 gauge
needle was used to pierce the top of the device reservoir. Dyed DI water was refilled
into the device.

Results

Increased in internal volume of the device is visually verified during refill (Figure
2-37). The refilled device was dispensed and refilled several times to verify
successful device functionality and multiple refills.

Figure 2-37 Surgical verification of drug device refill was completed in vitro using a commercially-
available, standard 30 gauge non-coring needle.

The needle progressing was visually monitored and controlled. However, in a
commercially viable device, refill would need to be successful when visual
inspection of the device is not possible. A means to prevent the needle from piercing
through the entire device is necessary. A stiff backing which cannot be punctured by
the needle will need to be incorporated into future designs.

2.2.5.2.3 In Vivo Delivery

2.2.5.2.3.1 Methods

The drug delivery device was sterilized one week prior to implantation. Pre-surgical
preparations include prefilling the device with a phenylephrine (10% concentration)
and Trypan blue mixture. Phenylephrine is a pupil dilating agent; Trypan blue is a
dye that stains necrotic tissues and cells but is not taken up by live cells. A male
pigmented rabbit was prepared for device implantation. The cannula of the device
was introduced into the anterior chamber of the eye via a scleral tunnel and the
device was secured to the eye using sutures. Baseline pupil diameter measurements
65

66
2.2.5.2.3.2
were taken. The chamber was then manually depressed and approximately 25 μL
(2.5 mg) of phenylephrine was delivered into the eye (t = 0). Pupil diameter
measurements were taken immediately after delivery (t = 1) and 10 minutes after
dispensation (t = 10).

Results

The pupil size of the rabbit increased after phenylephrine was introduced into the
eye. A summary of the results can be found in Table 2-2. Observation of the dyed
phenylephrine provided visual confirmation of delivery into the anterior chamber.
Both vertical and horizontal pupil diameter measurements increased following
phenylephrine delivery.

Table 2-2 Summary of Results from In Vivo Delivery using the Manually-Actuated Drug Delivery
Device
Pupil Dimension Vertical [mm] Horizontal [mm]
Baseline Measurement
t = 0
6.5 6.0
After Dispensation
t = 1 min
8.0 6.25
After Dispensation
t = 10 min
8.0 7.0
Total Change 1.5 mm (23%) 1.0 mm (16%)
19B
2.2.6 Summary

A manually-actuated drug delivery device that can be utilized as a method of
treatment for chronic ocular diseases has been demonstrated. Device components

67
such as the refillable reservoir and check valve were characterized. The maximum
internal pressure the reservoir can withstand, after multiple punctures using a 30
gauge non-coring needle, was determined to initially decrease with additional
punctures, but showed no significant decrease after 12 punctures. The check valve
diode curve was measured to determine cracking pressure and pressure versus flow
rate values. Check valve regulated delivery and closing time constants were also
determined. The device was tested in benchtop and ex vivo experiments including
device dispensing and refill. In vivo experiments demonstrated successful drug
delivery resulting in the corresponding physiological response. This ocular drug
delivery system is broadly compatible with existing ophthalmic drugs.

This device also has several modular components, which can simplify device
assembly and/or allow the device to be customized to a specific patient dosing
requirement or additional applications. The device is made of 3 layers; multiple
identical copies of each layer are formed simultaneously, providing interchangeable
components. Additionally, these individual layers allow several devices to be
assembled in parallel. Furthermore, a specific layer can be altered without requiring
changes in the other two layers to accommodate the change. For example, the
interior cavity of the top layer can be increased, thereby increasing the internal
volume of the device, without requiring a redesign of the bottom and middle layers,
or changes to the assembly process for the three layers. Changing the internal
volume can be used change the number of doses the device can contain and the
frequency of refill events. Finally, the check valve orifice size can be scaled

68
(increase or decrease in diameter), which will affect the diode curve of the check
valve.

2.3 Electrically-Actuated Device with Dual Check Valve

The electrically-actuated drug delivery device with dual check valve has the same
basic components (refillable reservoir, cannula, check valve, and suture tabs) found
in the manually-actuated drug delivery device. However, each component has been
redesigned to address any challenges observed in the manually-actuated device.
Each component is again designed to device constraints, tested to verify
functionality, and integrated into one device. The electrically-actuated device with
dual check valve also incorporates an electrolysis pump as a means for electronic-
actuation and a refill port.

The prototype of the electrically-actuated device with dual check valve has been
constructed for benchtop testing. The prototype has a medical-grade silicone drug
reservoir with integrated refill port, a pumping chamber to house the electrolysis
structure and electrolyte, a cannula, and a dual check valve system. The pumping
chamber ensures the electrolyte and drug remain separate, therefore the drug cannot
be contaminated by the electrolyte, nor oxidize due to electrolysis.

The electrically-actuated device with dual check valve improves upon the manually-
actuated device both in safety and functionality. The manually-actuated device was

69
capable of bolus (e.g. pulse) delivery; the second generation will allow for both bolus
and continuous delivery. The electrically-actuated device with dual check valve will
be electrically-actuated using an electrolysis structure. Electrolysis passes current
through water, which causes water to transform from liquid togas phase. This
actuation method provides a versatile dosing (bolus and continuous delivery),
variable delivery rates, and automatic dosing (Li, et al. 2007, Li, et al. 2008, Meng,
et al. 2006). Furthermore, while the manually-actuated single check valve allowed
one-way flow, it cannot reliably deliver a constant volume for each manually-
actuated dispensation event. Nor can the manually-actuated device prevent
accidental dosing from uncontrolled transient pressure spikes (e.g. flying, sneezing,
etc). The electrically-actuated device with dual check valve delivery rate can be
controlled via electrolysis. Additionally, the check valve contains a pressure limiter
which closes the check valve at high pressures.

2.3.1 Device Design

2.3.1.1 The Components

2.3.1.1.1 Check Valve and Cannula

A drug delivery system consisting using a Parylene C cannula has been previous
demonstrated (Li, et al. 2008). Drug was pumped directly into the eye by
electrolysis actuation. However, this prototype lacked a flow control valve; drugs
and bodily fluids could readily diffuse into and out of the device through the

70
intraocular cannula. Furthermore, the delicate, thin-walled Parylene C cannula was
easily damaged during surgical manipulation and implantation. Its rectangular
profile made it difficult to seal the incision through which it was inserted with
sutures alone; leakage paths resulted at the interface between the cannula and
incision site. Therefore, a more robust, circular cannula with an integrated check
valve was developed.

A cannula having circular cross-section facilitates wound closure with sutures and
prevents leakage around the incision. Ideally, the cannula diameter is < 1 mm; an
incision of this size in the eye wall can self-seal even without the use of sutures.
Therefore, the valve design should accommodate a circular cannula. Additionally,
the valve must be in-plane with respect to the fluid flow. This ensures that moving
mechanical parts of the valve do not come in contact with ocular tissues (Lo, et al.
2009). If the valve touches ocular tissue (e.g. the cornea) it may cause damage to the
tissue. Additionally, the valve opening may be occluded by the tissue, preventing
the valve from opening. This valve orientation also minimizes dead-volume. To
prevent diffusion of drug from the device into the tissue, the valve must be normally-
closed. Finally, the valve must be safe and function reliably under normal
intraocular pressure (IOP) conditions (15.5 ± 2.6 mmHg, 2.1 ± 0.35 kPa, (mean ±
SD)) (Ritch, et al. 1989), elevated conditions (glaucoma IOP > 22 mmHg, 2.9 kPa),
and transient pressures fluctuations (e.g. patient sneezing, eye rubbing, or changes in
ambient pressure) (Ritch, et al. 1989, Wilensky 1999).

71
2.3.1.1.1.1
Many MEMS valves exist and were reviewed recently by Oh and Ahn (Oh and Ahn
2006). To minimize the power requirements for the ocular drug delivery device,
only passive mechanical valve designs were considered. These valves allow flow
under forward pressure and exhibit diode-like regulation of flow. Examples include
valves consisting of flow orifices controlled by pressure-sensitive flaps, membranes,
and spherical balls. Lin et al. presented a glaucoma drainage device with Parylene C
checks valves (normally-closed and normally-open) valve in series to achieve
bandpass regulation. Adhesives were used to secure the valves in a Parylene C tube
(Lin, et al. 2009). Lo et al. fabricated a normally-closed silicone valve within a
rectangular cannula by stacking patterned layers of silicone rubber (Lo, et al. 2009).
However, current valve designs and packaging schemes are not suitable for
integration into an advanced ocular drug delivery device, therefore, a new valve and
package are necessary.

Design

Our valve approach is modular and consists of four plates stacked together to form a
normally-closed valve with a pressure limiter feature to provide bandpass fluid
regulation (Lo and Meng 2009). The four plates are the valve seat, valve plate,
spacer plate, and pressure limiter. Each of these plates can be replaced or exchanged
for a different plate design, creating a modular valve design with multiple
permeations. The plates are packaged within heat-shrink tubing; resulting in a robust

and adhesiveless package. The packaged valve is easily integrated into the drug
delivery system (Figure 2-38, Figure 2-39).

Figure 2-38 A MEMS ocular drug delivery device, which is sutured to the eye, contains a refillable
drug reservoir, contoured morphology, cannula, and modular valve. The valve comprises four stacked
disks (valve seat, valve plate, spacer plate, and pressure limiter. The cannula is inserted into the
anterior or posterior segments of the eye for targeted delivery of drugs.

72

Figure 2-39 Heat-shrink packaged valve integrated into a silicone surgical sham device. Drug
reservoir with metal ring indication refill port location, heat-shrink tubing, and valve are indicated.
Ruler divisions are 1 mm.

When a pressure greater than the cracking pressure is applied, the valve plate
deflects and lifts off of the seat, allowing forward fluid flow. Increased pressure on
the valve plate causes it to deflect further. Eventually, at a pressure exceeding the
closing pressure, the valve plate seals against pressure limiter to stop flow altogether.
Thus, the “bandpass” like behavior cuts off flow to prevent accidental dosing when
excessive forward pressures are experienced (Figure 2-40).

73

Figure 2-40 Valve operation (from left to right): initially normally-closed, valve opens under forward
pressure that exceeds cracking pressure, excessive pressures close the valve, and valve remains closed
under reverse pressure.

To simplify the packaging processes and to increase the yield, no adhesives were
used. Instead, biocompatible heat-shrink tubing was investigated as a packaging
method for the valve. Here, the heat-shrink tube also serves as the drug delivery
cannula. Heat-shrink tubing has been widely used in the world of electronics for
applications such as electrical isolation, environmental protection, repair, and
strengthening of joints. Heat-shrink tubes are available in many thermoplastic
materials, such as polyolefin, fluoropolymer (fluorinated ethylene propylene (FEP),
polytetrafluoroethylene (PTFE), polyvinylidene fluoride (PVDF)), polyvinyl
chloride (PVC), neoprene, and silicone elastomer. The heat-shrink mechanism is
achieved by first forming the material into its final shape. Then, ionizing radiation is
used to cross-link the material. Once cross-linked, the part is heated to a temperature
greater than the melting temperature of the material. The part is then stretched or
blown into an expanded configuration and cooled to maintain the expanded state.
The part will then shrink when heat, near the melting point, is again applied to the
part due to the elastic nature of the cross-linked material (Bradley 1984).
Biocompatible heat-shrink tubing (polyolefin) was investigated as a packaging
scheme for flexible sensors (Naito, et al. 2008) and glaucoma drainage implants
74

(Pan, et al. 2006). Due to their conformal nature, circular shape, and
biocompatibility, heat-shrink tubing was selected for packaging the valve for the
ocular drug delivery application.

The four modular valve components are shown in Figure 2-41 along with the heat-
shrink tubing. The SU-8 valve seat and silicone valve plate form the normally-
closed portion of the valve. The SU-8 spacer plate defines the distance the valve
plate must deflect in order to seal against the SU-8 pressure limiter plate. Therefore,
the thickness of the spacer plate, in part, controls the pressure at which the valve
closes. The valve seat and pressure limiter provide the structural support for the
valve plate.

Figure 2-41 Photo of the valve components (valve seat, valve plate, spacer plate, and pressure
limiter), pre-shrink heat–shrink tube, and fully assembled valve.

The assembled valve is packaged into a 22G FEP heat-shrink tube (Zeus Industrial
Products Inc., Orangeburg, SC) (Figure 2-42). FEP is a well known medical material
and is designated a USP class VI biocompatible polymer. FEP is also transparent,
with a refractive index of 1.338 (Zeus Industrial Products), thereby allowing visual
75

inspection of the valve after packaging. FEP is resistant to most chemicals and
solvents and can withstand temperatures in excess of 260 ºC, making it suitable for a
wide range of applications.

Figure 2-42 a) Side view and b) top view of the packaged valve in a FEP heat-shrink tube. The valve
was placed inside the tube with a custom jig. The entire fixture was heated to 215 ºC at 1.5 ºC/min
and cooled at the same rate to room temperature.

Upon heating, the heat-shrink tube contracts around the valve forming a robust
package that securely holds the valve assembly in place without any adhesives. The
previous Parylene C cannula was prone to clogging or damage during surgical
implantation (Figure 2-43), therefore a more robust cannula design was necessary.
The heat-shrink tube wall thickness (approximately 200 μm) is greater than that of
the Parylene C cannula (thickness 7.5 μm), resulting in a more mechanically robust
structure. As mentioned previously, the circular cannula facilitates sealing of the
incision, thereby minimizing leakage at the cannula and tissue interface.

76

Figure 2-43 a) Parylene C cannula integrated with a drug delivery pump, b) clogging of Parylene C
cannula after ex vivo testing.

2.3.1.1.1.2 Theory

The valve diameter was selected to meet the surgical requirements; a maximum
incision length of 1 mm was permitted. Therefore, valve components were limited to
900 μm in diameter, leaving 100 μm for packaging (i.e. a 50 μm annulus). The
dimensions and geometry of the individual valve components were determined using
theoretical equations and finite element modeling.

The analytical solution for large-deflection of a thin flexible plate of uniform
thickness guided selection of the thicknesses of the spacer and valve plates (Equation
2-1, Equation 2-2). The maximum deflection (w
max
) of a uniform and homogenous
place was calculated from the plate thickness (t), applied pressure (p), plate radius
77

(a), and flexural rigidity (D). Flexural rigidity is a function of Young’s modulus (E),
plate thickness (t), and Poisson’s ratio ( ν) (Ugural 1999).

Equation 2-1 Large deflection in a homogenous, thin film plate
D
pa
t
w
64
)
4
2
max
2
= w 486 . 0 1 (
max
+

Equation 2-2 Flexural rigidity equation to determine deflection in thin film plate
3
2
12(1 )
Et
D
ν
=
−

The outer edge (150 μm wide band) of the valve plate was reserved for clamping by
the valve seat and pressure limiter, leaving a 600 μm diameter area for the active
deflecting area of the plate. In the analysis, applied pressure was varied between 0 -
1000 mmHg (0 - 133.3 kPa) for valve thicknesses between 0-150 μm. The resulting
calculated w
max
was assigned as the maximum thickness of the spacer plate. It
should be noted that w
max
is the deflection at the center of the plate; sealing against
the pressure limiter requires greater deflection. The final values were chosen based
on estimated pressure operating ranges and ease of handling. The values used in the
large-deflection equations are shown in Table 2-3.

78

79
Table 2-3 Summary of values used in theoretical calculations of large deformations in uniform thin
plates.
Variable Value
Thickness (t) 0-150 μm
Applied Pressure (p) 0 - 1000 mmHg, 0 - 133.3 kPa
Plate Radius (a) 300 μm
Young’s Modulus (E) 2 MPa
Poisson’s Ratio ( ν) 0.48

2.3.1.1.1.3 Finite-Element Modeling

SolidWorks models of the valve components were created for FEM analyses. Stress
and deformation FEM analyses determined the stress distribution and valve behavior
for forward pressure values ranging from 0 - 1000 mmHg (0 - 133.3 kPa) and the
reverse pressure value of -500 mmHg (-66.66 kPa). FEM results provided
convenient visualization of valve plate movements and its interaction with the
pressure limiter.

Three different valve plate designs (hole, straight arm, and s-shape arm) were
investigated (Figure 2-44). Each design possessed a different effective fluidic
resistance and thus differing bandpass flow regulating characteristics (e.g. opening
and closing pressure). FEM estimations and theoretical analyses guided the
assignment of valve geometries such that the operational pressure range would be
limited at the lower bound by normal intraocular pressure (IOP), <35 mmHg (4.7
kPa), and an upper bound that was arbitrarily chosen to be at least 2 orders of
magnitude greater than normal IOP values (e.g. 2000 mmHg, 266.6 kPa). However,
the operating pressure ranges of the valve are easily customized by altering the

dimensions of the valve plate and spacer plate. The dimensions of the valve
components are presented in Table 2-4.

Figure 2-44 Three different valve plate designs a) hole, b) straight arm, and c) s-shape arm; and the
corresponding fabricated valve plates d) hole (through holes are indicated by the arrows), e) straight
arm, and f) s-shaped arm.

Table 2-4 Dimensions of valve components, including the three valve designs (hole, straight arm, s-
shape arm). All components are 900 μm in diameter.

80

81
2.3.1.1.2 Refillable Reservoir

Similarly to the manually-actuated a refillable reservoir is used to house the drug.
The reservoir is fabricated using MDX4-4210, medical-grade silicone. The refillable
reservoir is separated using a Parylene C bellows to separate the refillable reservoir
into drug and electrolysis chambers.

In the manually-actuate drug delivery device, the refillable reservoir was fabricated
by molding PDMS around a Plexiglas rectangular cuboid. The reservoir shape was
unsuitable for long term implantation due to several reasons, 1) the external shape
had sharp corners and edges, which can aggravate biological tissue and could lead to
conjunctival thinning, and 2) the walls of the reservoir prevented the reservoir from
being fully depressed, thus the corners of the interior space had the potential of
accumulating dead volume; preventing all of the drug from being dispensed and can
lead to contamination of newly refilled drug, and 3) the corners and edges of the
reservoir were areas of stress concentration during reservoir depressing; these areas
have a greater potential of failure.

A new design which provided a more suitable biologically compatible morphology,
while removing areas of stress concentration and dead volume investigated. A
domed shape reservoir was chosen as it satisfied the above criteria.

82
2.3.1.1.3.1
2.3.1.1.3.2
2.3.1.1.3 Electrolysis Pump and Pump Chamber

Theory

Electrolysis pumping was selected as the driving force for expelling drugs from the
device into the body. Electrolysis has many advantages over other pumping schemes
because of its low heat generation, low power consumption, simple construction, and
ability to generate large displacements. Electrolysis is an electrochemical reaction
which breaks water into hydrogen and oxygen gas when a current is passed through a
metal conductor such as gold or platinum. The electrolysis reaction can be described
in the following manner:

Equation 2-3 Electrochemical reaction during electrolysis of water
cathode : 2H
2
O
(l)
↔ O
2(g)
+ 4H
+
(aq)
+ 4e
−

anode : 4H
+
(aq)
+ 4e
−
↔ 2H
2(g)

net : 2H
2
O
(l)
↔ O
2(g)
+ 2H
2
(g)

The efficiency of the electrolysis structure is dependent on many factors including,
electrode design (e.g. spacing, width, thickness, material, and surface area),
electrolysis solution, external conditions (e.g. temperature and pressure), and the
electrical parameters (e.g. current amperage).

Design

The electrolysis structure is two interdigitated platinum electrodes fabricated on a
flexible Parylene C substrate. The overall footprint of the interdigitated section is
circular in order to fit within a rounded reservoir. The electrolysis structure is

83
patterned onto a Parylene C substrate. This allows the electrolysis structure to be
placed onto a non-planer surface.

The pump chamber is fabricated using Parylene C in a bellows structure. This
structure can expand and contract to accommodate gas generation and
recombination. The Parylene C bellows is flexible, and therefore, does not
significantly impeded pumping efficiency. The bellows structure is attached to the
Parylene C electrolysis structure to create an enclosed pump system.

2.3.1.1.4 Cannula

The cannula for the electrically-actuated device with dual check valve is mainly the
heat-shrink package around the dual function valve. The heat-shrink package
(biocompatible FEP) and circular cross-section makes the package well-suited for
implantation.

2.3.1.1.5 Baseplate

A baseplate to prevent the refill needle from penetrating through the entire device is
added to the reservoir base. The baseplate is made of PEEK (polyetheretherketone),
which is a commercially available, bio-compatible material. A Plexiglas stencil is
used to trace the desired baseplate shape onto a 0.01’’ thick PEEK sheet; the shape is
then cut from the sheet. The PEEK baseplate is added to the silicone prepolymer in

84
2.3.1.2.1.1
the reservoir base mold; upon curing, the PEEK sheet is fully encased by the
silicone.

2.3.1.1.6 Suture Tabs

Two suture tabs are added to the reservoir to provide locations through which sutures
can be threaded to secure the device to the eye. Suture tabs are torus-shaped; a
predefined center hole allows sutures to be passed through the suture tab without the
suture needle tearing the silicone material. A tear may propagate through the
material causing the suture tab to fail, or even cause a leakage path to form through
the reservoir.

2.3.1.2 Device Fabrication

2.3.1.2.1 Valve Fabrication

SU-8 Valve Seat and Pressure Limiter

The SU-8 valve seat and pressure limiter shared identical designs. This choice
simplified fabrication and ensured interchangeability of the two parts. A two-layer
SU-8 process was used (Figure 2-45). First, a soda-lime wafer (Mark Optics, Santa
Ana, CA) was dehydrated for 20 minutes at 120 ºC. Then the wafer was treated with
Omnicoat (MicroChem, Newton, MA) to facilitate release of the SU-8 components.

85
Three layers of Omnicoat were spun onto the wafer (3000 rpm, 30 sec) with a bake
step (1 min at 200 ºC) performed after each coat (Figure 2-45a). Multiple Omnicoat
layers reduced the time and temperature required for the release step. SU-8 2100
(MicroChem, Newton, MA) was prespun onto the wafer (30 sec, 500 rpm) to provide
an even coating. Then, 160 μm SU-8 was applied (30 sec, 1750 rpm) to form the
first layer of the component (Figure 2-45b). This layer was softbaked on a hotplate
at 95 ºC for 2 hours (3 ºC/min) and slowly cooled to room temperature. The layer
was then patterned (390 mJ/cm
2
); the energy dosage was determined by using the
suggested energy level for 160 μm (260 mJ/cm
2
) and adjusting with a 1.5 multiplier
for using a glass substrate instead of silicon (Figure 2-45c). The mask used to
pattern the first layer can be found in Appendix M. Dicing saw tape was placed
behind the wafer prior to exposure to prevent unwanted exposure from the reflected
UV from the aligner chuck. A post-exposure bake (12 min, 95 ºC) was completed,
again ramping from room temperature to 95 ºC at 3 ºC/min and slowly cooled back
to room temperature. 40 μm of SU-8 2050 (MicroChem, Newton, MA) was spin
coated (30 sec, 4000 rpm) to form the features in the valve seat and pressure limiter
(Figure 2-45d). The wafer was then baked for 3 hours at 95 ºC (with ramp up and
cool down). The 40 μm was patterned (192 mJ/cm
2
; 160 mJ/cm
2
times a 1.2
multiplier for a SU-8 substrate), and post exposure baked for 14 minutes at 95 ºC
Figure 2-45e). The mask used to pattern the second layer of SU-8 can be found in
Appendix N. The components were developed using SU-8 developer (MicroChem,
Newton, MA) (Figure 2-45f).

To remove the valve seats and pressure limiters from the wafer, the wafer was
immersed in Remover PG (MicroChem, Newton, MA) (Figure 2-45g). The
components were rinsed in isopropyl alcohol (IPA) and DI H
2
O and then hardbaked
at 215 ºC for 1 hour. This final step annealed the SU-8 components to improve
thermal resistance for the subsequent heat-shrink packaging process. This step was
performed under vacuum to prevent oxidation of the SU-8 and reduce the residual
stress in the thick film structure (Daniel, et al. 2001). The hardbake step was later
removed to simplify the fabrication process; the entire assembled valve was
hardbaked during the heat-shrink tubing process (heat-shrink packaging occurred
under vacuum at 215 ºC). The complete fabrication process can be found in
Appendix O.

Figure 2-45 Fabrication process for the valve seat and pressure limiter plates. Fabrication steps are
cross-section views at the A-A’ line.

86

87
2.3.1.2.1.2 SU-8 Spacer Plate

The spacer plate was fabricated on a dehydrated wafer (20 min, 120 ºC) coated with
3 layers of Omnicoat (as described in the fabrication of the valve seat and pressure
limiter) (Figure 2-46a). The 40 μm thick spacer plate was spin coated (SU-8 2050,
30 sec, 4000 rpm) (Figure 2-46b). The layer was softbaked at 95 ºC for 1 hour,
ramping at 3 ºC/min from room temperature to 95 ºC. Dicing saw tape was applied
to the backside of the wafer prior to exposure (240 mJ/cm
2
; 160 mJ/cm
2
times a 1.5
multiplier for a glass substrate) (Figure 2-46c). The mask used to pattern the spacer
plate can be found in Appendix P. The wafer is post-exposure baked for 6 minutes at
95 ºC (ramping from 3 ºC/min from room temperature to 95 ºC, and slowly cooled
back to room temperature). The wafer was immersed in SU-8 developer (Figure
2-46d). The spacer plates were then released from the substrate using Remover PG
and rinsed using IPA and DI H
2
O Figure 2-46e). A complete fabrication recipe for
the spacer plate can be found in Appendix Q.

Figure 2-46 Fabrication process for the SU-8 spacer plate. Fabrication steps are cross-section views
at the A-A’ line.

88
2.3.1.2.1.3 Silicone Valve Plate

The valve plate was fabricated by casting medical-grade silicone pre-polymer
(MDX4-4210, Dow Corning, Midland, MI) on a SU-8 master mold. The simplest
design for the valve plate mold is a single layer of SU-8 which defines the valve
plate thickness, valve plate diameter, and the through holes of the valve plate.

2.3.1.2.1.3.1 Simple Valve Plate: No Bossed or Overhang Features

The SU-8 master was created on a soda lime wafer using SU-8 2050. First the wafer
was treated with A-174 (a silane adhesion promoter) to enhance Parylene C adhesion
to the soda lime wafer. A 4 μm layer of Parylene C (Specialty Coating Systems,
Inc., Indianapolis, IN) was vapor deposited onto the wafer to prevent the SU-8 from
delaminating from the wafer due to mismatch of the thermal coefficients of
expansion between soda lime and SU-8 (Li, et al. 2008). A 75 μm layer of SU-8
2050 was spin coated (30 sec, 2000 rpm) and softbaked for 90 minutes at 95 ºC
(Figure 2-47a). This layer of SU-8 defined the valve plate thickness. The SU-8
layer was patterned (308 mJ/cm
2
; 208 mJ/cm
2
dose times a 1.5 multiplier for the
glass substrate), post-exposure baked at 95 ºC for 7 minutes, and developed using
SU-8 developer (Figure 2-47b,c). The mask used to pattern the valve plate can be
found in Appendix R. MDX4-4210 (10:1 base to curing agent ratio), was poured
onto the mold and degassed under vacuum. Excess silicone was removed by
scraping the mold with a metal squeegee (Figure 2-47d) (Kee Suk, et al. 2004). The

silicone is cured at room temperature for over 24 hours or, for accelerated curing, at
90 ºC for 1 hour. Silicone is less prone to shrinkage if cured slowly at room
temperature. Valve plates where released from the mold; any excess silicone was
manually removed using a fine-tipped blade (Figure 2-47e). The fabrication process
for creating the SU-8 mold and casting the silicone is found in Appendix S.

Figure 2-47 Fabrication process for the valve plate using an SU-8 master mold. Fabrication steps are
cross-section views at the A-A’ line. Straight arm valve is shown; hole and s-shaped arm valves are
fabricated in an identical manner.

2.3.1.2.1.3.2 Valve Plate with Bossed Feature and Valve Plate with Bossed and
Overhang Features

The valve plate can designed to include additional features, which affect valve
function and performance. Two optional features include a bossed structure or an
overhang (Figure 2-48). The bossed structure causes the valve plate to press more
tightly against the valve seat, therefore increasing the cracking pressure of the valve.
The overhang structure is a proposed mechanism for locking the valve plate in place
89

if, during experimental testing, it was determined that the valve plate shifts during
pressure application. Complete process to fabricate the SU-8 mold for a valve plate
with just the bossed feature or both the bossed and overhand features can be found in
Appendix U and Appendix W, respectively.

Figure 2-48 Illustration of the additional features (bossed and/or overhang) which can be added to the
simple valve plate designs.

2.3.1.2.2 Valve Heat Shrink Packaging

Valves were assembled by stacking individual components. First, the components
were gathered onto a silicone sheet under a stereo microscope. Silicone provided a
tacky surface to hold the stack steady while making alignment adjustments.

First, the valve seat was placed face up (sealing rings up) on the silicone working
space (Figure 2-49a). Next, the valve plate was stacked on top (Figure 2-49b). The
valve plate must be aligned such that none of the through holes are placed over the
valve seat opening. Misalignment may compromise the normally-closed function of
the valve, or prevent the valve from closing at elevated pressures. The spacer plate
90

was added on top of the valve plate, again, ensuring that the spacer plate was aligned
and not covering the valve plate through holes (Figure 2-49c). Finally, the bottom
side of the pressure limiter, the side with the sealing rings, was identified and placed
face down on top of the spacer plate (Figure 2-49d).

Figure 2-49 Top and side views of valve assembly. a) valve seat, b) valve plate added to valve seat,
c) spacer plate placed on valve plate, d) pressure limiter added to assembled valve.

The assembled valve was then packaged with 1.3:1 shrink ratio FEP heat-shrink
tubing with the aid of a custom Teflon jig containing stainless steel centering pins
(813 μm diameter) (Figure 2-50a,b). A 22G (inner diameter prior to shrinkage: 914
μm, maximum wall thickness: 254 μm) heat-shrink tube was placed around the
centering pin on the jig base. The heat-shrink tube must be shorter than the centering
pin. Next, the assembled valve (valve seat, valve plate, spacer plate, pressure
limiter) was carefully placed on the centering pin (Figure 2-51a). The jig top, which
has a matching and adjustable stainless steel centering pin, was aligned and secured
to the jig base using four 10-32 machine screws. The distance between the jig base
91

and top was set using hex nuts positioned along the screws. The two centering pins
were aligned and the top pin was slowly lowered until the valve stack was securely
clamped. The FEP tubing was carefully slipped around the valve (Figure 2-50b,
Figure 2-51b). The entire jig was then placed in an vacuum oven and heated to 215
ºC at a rate of 1.5 ºC/min; held at 215 ºC for at least 30 minutes, and then cooled to
room temperature at the same rate (Figure 2-51c). The baking and cooling steps
were ramped to limit the thermally induced stress on SU-8 which may lead to
cracking. The jig was removed from the oven and disassembled. Then the packaged
valve was slipped off the centering pins (Figure 2-51d). A complete SOP for
assembling and packaging the valve can be found in Appendix X.

Figure 2-50 a) Heat-shrink jig setup. Teflon base and top each contains a centering pin. The top and
base are aligned with machine screws. Nuts set the top and base distance. b) Close up view of an
assembled valve with pre-shrink heat-shrink tube surrounding the valve.

92

Figure 2-51 Process steps to package assembled valve. a) FEP heat-shrink tube is placed around
bottom centering pin, assembled valve is placed on centering pin, b) jig top is added, valve is clamped
between top and bottom centering pins, FEP tube is lifted around valve, c) jig and valve assembly is
placed in vacuum oven, and d) packaged valve is removed from jig.

2.3.1.2.3 Refillable Reservoir

The silicone reservoir was fabrication technique was designed and optimized during
the construction of the hollow surgical sham (Section 2.4.2.1.3.1- Reservoir). The
same technique was utilized for fabricating the reservoir for the electrically-actuated
device with dual check valve using reservoir molds designed for this device (Figure
2-52).
93

Figure 2-52 Top and side views of the molds used for fabrication the second generation drug delivery
device reservoir.

In summary, a silicone reservoir can be fabricated using a clam-shell custom-made
Plexiglas mold. The bottom mold, a convex double-dome shape which defines the
reservoir interior, was fabricated using laser ablation of a gradated double-oval
pattern. The internal volume of the reservoir can be calculated by determining the
volume of the convex portion of the mold. The top mold, which is a concave
double-dome, creates the exterior surface of the reservoir.

94

The reservoir was fabricated by filling the bottom mold with MDX4-4210, a
medical-grade silicone. The mold was placed under vacuum to remove any residual
air in the silicone prepolymer. Two metal shafts (1.27 mm O.D.) were placed on the
edge of the mold to create a separation between the top and bottom molds. This
separation dictated reservoir wall thickness. Finally, the top mold was carefully
placed and aligned to the bottom mold. The silicone was cured and the reservoir was
released from the mold. The excess silicone was then removed from the reservoir.

The base of the reservoir is also fabricated using a Plexiglas mold. The base mold
was filled with MDX4-4210 and placed under vacuum. A silicone tube and PEEK
baseplate was carefully placed in the mold and cured. A fully assembled surgical
sham using the molds is shown in Figure 2-53.

Figure 2-53 Assembled surgical sham using molds shown in Figure 2-52.

The silicone reservoir was proposed as a rapid method of fabricating and prototyping
reservoirs. Molds for silicone reservoirs could be easily made and altered to fit any
95

96
design change. However, due to the compliant nature of silicone, which can affect
the dispensation characteristics, the reservoir must be eventually fabricated using a
stiffer material (e.g. injection molded medical-grade polypropylene).

2.3.1.2.4 Cannula

To connect the packaged valve to the reservoir, a silicone tube is cured into the
reservoir base. The silicone tube is positioned using an indentation in the reservoir
base mold. The same silicone tubing (O.D. 0.085’’, 2.159 mm) (REF 60-411-44,
HelixMark, The Netherlands) was used in the testing of both the packaged solid disk
and packaged valve. The interface between the silicone tubing and heat-shrink tube
was leak-tight up to 17 psi without the aid of any adhesives or mechanical cinching.
This seal is adequate for normal valve operation, but can be further strengthened by
applying silicone prepolymer at the heat-shrink tube and silicone tube junction prior
to implantation.

The use of a silicone tube as a connection between the reservoir and packaged valve
provides several valuable features. First, not applying silicone prepolymer to the
heat-shrink tube/ silicone tube junction adds increased modularity of the electrically-
actuated device with dual check valve as packaged valves can be easily exchanged
with no damage to either the assembled reservoir or packaged valve. Additionally,
during implantation, the cannula length can be adjusted to fit patient needs removing

97
2.3.1.2.5.1
the packaged valve, cutting excess silicone tubing, and then replacing the valve.
This would not be possible if the tube was a single piece.

The heat-shrink tube is very stiff. If a more flexible cannula is desired, the length of
the heat-shrink can be shortened such that heat-shrink tubing is only slightly longer
than the height of the stacked valve. In this mode, the cannula is made entirely of the
flexible silicone tubing and the packaged valve can be inserted into the end of the
silicone tube, whereby the packaged valve is entirely contained within the silicone
tube.

2.3.1.2.5 Refill Port

Port Material

The material used to fabricate the refillable portion of the reservoir is being
investigated. The current material, MDX4-4210, is being used to model the device
because it is easily obtained and can be used to rapidly create prototypes.
Additionally, MDX4-4210 is a USP Class IV biocompatible material; however,
silicone has been reported to absorb liquids and solvents.

The requirements of the material include: self-sealing behavior after needle puncture,
be able to withstand the operating pressures of the device, no leakage through
puncture site under normal operating pressures, USP Class IV biocompatibility,

98
2.3.1.2.5.2
moldable, and compatible with the fabrication steps being used to create the other
components of the device.

Refill Port Placement

The refill port is placed in the smaller of the two circular domes, which make up the
reservoir body. The electrolysis bellows occupies the larger of the two domes,
therefore, this placement separates the refill area from the main reservoir body and
prevents the refill needle from compromising the integrity of the pump chamber
during refill.

A stainless steel ring is embedded into the reservoir body to help surgeons locate the
refill portion of the device. The stainless steel ring is clearly visible as a dark
shadow when the eye is transilluminated (Figure 2-80). The refill needle pierces the
center of the ring in order to access the device interior.

2.3.1.2.6 Device Assembly

The device is assembled on benchtop. All of the components are pre-fabricated to
expedite device assembly. The following components are necessary for device
assembly: silicone reservoir base layer (with embedded PEEK baseplate and silicone
cannula), silicone spacer layer, silicone reservoir dome (with embedded stainless
steel refill ring), and electrolysis pump with Parylene C bellows (Figure 2-54).

Figure 2-54 Components needed to create a fully integrated electrically-actuated device with dual
regulation check valve.

First, the suture tabs in the reservoir base are cored using a 19G needle (O.D. 1.067
mm). 20G (O.D. 0.902 mm) needle shafts are then threaded through the suture holes
to prevent the sutures holes from clogging during assembly. The spacer layer is then
affixed to the reservoir base using MDX4-4210 prepolymer; both pieces are placed
in an oven for 5 minutes at 80 ºC to cure.

Next, the electrolysis pump and bellows component is attached to the reservoir base
using MDX4-4210 prepolymer. Special care is needed to ensure the MDX4-4210
does not reflow over the electrodes prior to curing. Again, the assembly is cured at
80 ºC for 5 minutes. DI H
2
O soaked Techwipes are placed around the device during
99

100
curing to increase the local humidity. This mitigates the evaporation of electrolyte
from the pump chamber.

Two different types of spacer layers can be added to the device. The spacer layers
increase the reservoir height by raising the reservoir dome. The increased height
may be necessary to accommodate the bellows expansion during pumping. The
spacer layers are 1 mm thick by 1 mm tall silicone pieces that are shaped like the
device footprint. The first spacer layer has a section missing to facilitate fabrication
by allowing the wires from the electrolysis pump to be fed through the spacer layer.
Prior to the addition of this layer, the device assembly was challenging because the
wires tended to prop up the reservoir dome on one side; sealing the gap between the
reservoir base and dome was very difficult. This layer, in addition to adding
reservoir height, makes it easier to secure the reservoir dome to the rest of the device,
because the wires are now recessed. The second spacer layer’s sole purpose is to add
reservoir height and is optional.

Next, the reservoir dome is secured using MDX4-4210 and cured in place. The fully
assembled device is filled with water to identify any leaks in the reservoir body. The
dual check valve is then plugged into the end of the silicone cannula and the 20G
needle shafts are removed from the suture tabs (Figure 2-55).

Figure 2-55 Fully integrated electronically-actuated drug delivery device. Device includes a
refillable reservoir, electrolysis pump, separate refill area, refill ring, PEEK baseplate, silicone
cannula, heat-shrink packaged dual check valve, and suture tabs.

2.3.1.3 Benchtop Experiments- Methods and Results

2.3.1.3.1 Valve Plate Deflection

2.3.1.3.1.1 Methods

The deflection for each valve plate design (hole, straight arm, s-shaped arm) under
forward pressure was measured and compared to theoretical values for large
deflection of a uniform plate (Equation 2-1, Equation 2-2). Each plate was clamped
at the periphery in a custom-made jig which allowed pressurized air to be applied to
101

the backside (Figure 2-56). Plate deflection was measured using a compound
microscope with 1 μm resolution. The microscope was focused on the center of the
valve plate under zero applied pressure. Pressurized nitrogen gas (0 - 500 mmHg, 0 -
66.7 kPa) was applied to deflect valve plate. The microscope was refocused on the
center of the deflected plate and deflection was calculated from the change in the
microscope fine focus knob position.

Figure 2-56 Valve plate deflection setup.

2.3.1.3.1.2 Results

The experimentally obtained valve plate deflection for each valve plate design was
compared to theoretical values (Figure 2-57) and found to be in agreement. While
the theoretical model did not completely predict the behavior, it was still a useful
design tool. Deviations from the measured data could be attributed to the
geometrical differences from the theoretical thin plate geometry. The plate thickness
used in the equation was 75 μm, whereas the fabricated valve plates were slightly
102

103
thicker at 84 ± 3.7 μm, 87.8 ± 5.3 μm, and 82.2 ± 5.4 μm (mean ± SE, n = 4) for the
hole, straight arm, and s-shaped arm valve plates, respectively. A thicker plate
generally leads to less deflection. However, changing the geometry from a flat plate
to a selectively perforated plate increased achievable vertical deflection due to
increased flexibility. Thus, the straight arm valve plate deflected more than the hole
design even though the straight arm plates were slightly thicker. The clearest
demonstration of the impact of tether compliance on achievable deflection was the s-
shaped arm valves. The s-shaped tethers bend allowing the plate to twist upward and
away from the valve seat as pressure is applied, providing additional deflection
(Wang, et al. 1999). Therefore, the s-shaped arm valve plate achieved the greatest
deflection for a given applied pressure.

Figure 2-57 Comparison of measured valve plate deflection to the theoretical values for a flat plate.

2.3.1.3.2 Heat-Shrink Packaging Characterization

2.3.1.3.2.1 Methods

The heat-shrink packaging method was evaluated to determine the dimensional
changes during the thermal shrinking process, the robustness of the package, and the
fluidic integrity (Figure 2-58). Two different gauges (22 AWG and 18 AWG, pre-
shrunk O.D. 1.29 mm and 1.88 mm, respectively) of FEP heat-shrink were
characterized (Zeus Inc., Orangeburg, SC). The outer diameters of pre-shrink and
post-shrunk tubing were compared and the percent change in outer diameter was
calculated. The fluidic integrity of each tube was quantified by packaging a solid
104

105
≤ ≤
200 μm thick SU-8 disk. A 900 μm diameter disk was used in the 22 AWG tube; the
solid disk possesses the same diameter as the individual valve components. The 18
AWG tube was packaged with a 1.5 mm diameter disk for comparison. The solid
disks were packaged under the same conditions as the valve (room temperature to
215 ºC at 1.5 ºC/ min and cooled from 215 ºC to room temperature at 1.5
ºC/min).

Figure 2-58 Pre and post heat-shrink tubing. Solid disk packaged in heat-shrink
tubing to test robustness of the adhesiveless packaging method.

Both pressurized water and nitrogen gas (0 - 2000 mmHg, 0 - 266.6 kPa) were
applied through the heat-shrink tube against one side of the solid disk. A 100 μL
calibrated pipette (Clay Adams, Parsippany, NJ, USA) was placed at the outlet to
measure leakage of water between the disk and heat-shrink tubing. For pressurized
N
2
, the tubing outlet was immersed in water to visualize any bubbles due to leakage.

106
2.3.1.3.2.2 Results

The final post-shrink diameter for the 22 AWG cannula used to package the valve
was approximately 1 mm. The outer diameter of the heat-shrink tube decreased
19.4-25% post-shrinkage. In the valve region, the diameter was slightly greater than
1 mm. This slight increase in diameter does not compromise the incision site which
must be sealed around the cannula alone and not the valved portion.

A segment of heat-shrink tube was packaged with a solid SU-8 disk to determine the
robustness of the packaging method and the quality of the seal around object. For
both the 18 AWG and 22 AWG tubes, the solid disks remained in position under
applied pressure and the entire system was leak-tight up to 2000 mmHg (266.6 kPa)
of pressurized water. This value is the pressure limit of our testing apparatus and
almost 2 orders of magnitude greater than normal IOP values. The packaged system
was also able to withstand up to 2000 mmHg (266.6 kPa) of pressurized nitrogen gas
as verified by the absence of bubbles while submerged under water. The maximum
pressure for microfluidic interconnects using heat-shrink tubing also reported leak-
free connections up to 200 kPa (Pan, et al. 2006). These results are comparable to
devices which are secured using adhesives (Lee, et al. 2004, Puntambekar and Ahn
2002, Tsai and Lin 2001). A summary of the results is presented in Table 2-5. This
characterization demonstrates that this packaging technique is extremely robust and
thus suitable for a wide variety of applications. A diverse selection of materials and

107
gauges of heat-shrink tube are commercially available and can be selected to match
specific packaging needs.

Table 2-5 Summary of heat-shrink tube characterization results for two tube gauge sizes (22 AWG
and 18 AWG)
FEP Heat-Shrink
Characterization
22 AWG 18 AWG
Valve O.D. [mm] 0.9 1.5
Initial tube O.D. [mm]
(n = 5, mean ± SE)
1.29 ± 0.014 1.88 ± 0.006
Post-shrink O.D.
Tube + Valve [mm]
(n = 5, mean ± SE)
1.23 ± 0.002 1.84 ± 0.004
Post-shrink O.D.
Tube Only [mm]
(n = 21, mean ± SE)
1.04 ± 0.006 1.41 ± 0.005
% change in O.D. 19.40% 25%
Leakage Pressure
[mmHg, kPa]
>2000, 266.6 >2000, 266.6

2.3.1.3.3 Packaged Valve Characterization

2.3.1.3.3.1 Finite-Element Analyses Results

Finite-element analyses of displacement and stress were conducted on an assembled
hole valve (Table 2-6). A linear FEM model with small-displacement conditions
was used to provide a qualitative understanding of valve behavior. As the applied
pressure increased, displacement and stress values also increased. The valve plate
deflected until it was constrained by the pressure limiter and eventually sealed.
However, the center of the valve plate continued to deflect with increasing pressure,

108
albeit in smaller increments. Under reverse pressure application, the valve plate
deflected <7 μm and maintained an effective seal against the valve seat.

The stress analysis provided guidance on the selection of suitable materials on the
basis of mechanical robustness. At the maximum forward applied pressure (1000
mmHg, 133.3 kPa), stress accumulated in the bottom of the valve seat and valve
plate. The maximum stress was experienced by the valve plate (0.99 MPa) and was
concentrated along the edge in contact with the valve seat. The observed stress was
<20% of MDX4-4210 tensile strength (5 MPa) and significantly less than the tensile
strength of SU-8 (60 MPa). Reverse pressure (500 mmHg, 66.7 kPa) analysis
verified the induced stresses (0.46 MPa) were at least an order of magnitude less than
the tensile stresses of MDX4-4210 or SU-8.

Table 2-6 Summary of the FEM results for displacement and stress on an assembled valve.

2.3.1.3.3.2 Benchtop Operation

The valve operating range was determined by visually observing valve operation and
measuring flow rate at specific pressure set points. Pressurized Rhodamine B was
applied to the cannula inlet and enhanced visualization of fluid flow through the
valve. Rhodamine B was observed to exit only at the pressure limiter through hole.
No leakage at the interface between the assembled valve and heat-shrink tube was
observed (Figure 2-59). This demonstrates robustness of this packaging method for
stacked components fabricated from different polymers.

109

Figure 2-59 Visualization of flow rate through a packaged valve using Rhodamine B.

2.3.1.3.4 Packaged Valve Opening and Closing Pressures

2.3.1.3.4.1 Methods

The bandpass flow regulation behavior of a packaged valve (hole valve plate) was
determined. Pressurized water (0 - 2000 mmHg, 0 - 266.6 kPa) was applied in
incremental steps to the valve inlet. The flow rate from the packaged valve was
measured using a 100 μL calibrated pipette connected to the outlet (Figure 2-60).
The pipette was prefilled with double distilled water and a bubble was introduced
between the valve outlet and pipette inlet. The system was held at each test pressure
set point for 3 minutes to allow the system to equilibrate. The cracking pressure and
flow rates for different pressures were measured. Reverse pressure (0 - 500 mmHg,
110

0 - 66.7 kPa) was also investigated. Several flow profiles (flow rate versus pressure)
for a hydrated valve (valve that was kept in contact with water at all times) were
obtained to verify repeatability of valve operation. The hydrated valve data was also
compared to a dry valve (valve where water was allowed to evaporate between
experiments). The testing process procedure can be found in Appendix Y.

Figure 2-60 Test setup to determine valve operating characteristics.

2.3.1.3.4.2 Results

Consistent flow rate profiles (pressure versus flow rate) were repeatedly obtained in
sequential experiments when the valve remained hydrated between runs. All
experimental trials were performed within a 24 hour period. Preliminary results for
the hole valve is presented in Figure 2-61. The valve cracking pressure was
approximately 25 - 50 mmHg (3.3 - 6.7 kPa) and the maximum flow rate (3.18 ±
0.18 µL/sec, mean ± SE, n = 4) occurred near 500 mmHg (66.7 kPa) (Figure 2-61).
The valve closed between 1750 and 2000 mmHg (233.3 and 266.6 kPa) and
remained leak-free under 500 mmHg (66.7 kPa) of reverse pressure, which is at least
an order of magnitude greater than normal IOP values.

111

The averaged flow rate profile (mean ± SE, n = 4) was compared to that of a dried
valve. For dried valves, the cracking pressure was much higher (200 - 300 mmHg,
26.7 - 40 kPa). The increase might be attributed to stiction between the valve plate
and seat. Additionally, the dried valve had a smaller closing pressure (1000 - 1250
mmHg, 133.3 - 166.7 kPa). Hydration causes the silicone to swell, increasing the
valve plate thickness and altering the deflection behavior. An implanted valve will
remain hydrated from contact with the aqueous or vitreous humors, therefore, the
hydrated flow profile is more representative of long-term valve behavior. However,
this also suggests that valves may need to be preconditioned prior to implantation.

Figure 2-61 Flow profiles from 4 runs on same valve, valve was kept hydrated in double distilled
water between runs to prevent valve from drying out. The hole valve plate was used in this packaged
valve.

112

113
A table summarizing the results of the hole, straight arm, and s-shaped arm packaged
valves can be found in Table 2-7.

Table 2-7 Summary of packaged valve operating characteristics for hole, straight arm, and s-shaped
arm valve.
Parameter Hole
Straight
Arm
S-Shaped
Arm
Cracking Pressure [mmHg, kPa] 50, 6.67 300, 40.0 100, 13.3
a
Closing Pressure [mmHg, kPa] 1750, 233.3 900, 200.0 700, 93.3
b
Leak-tight Reverse Pressure [mmHg, kPa] 500, 66.7 500, 66.7 500, 66.7
Pressure at Max Flow [mmHg, kPa] 500, 66.7 500, 66.7 150, 20.0

As Table 2-7 shows, the s-shaped arm closed at a lower pressure than the straight
arm valve. Additionally, the straight arm closed at a lower pressure than the hole
valve. This result is consistent with the theoretical equation and FEM models where
the s-shaped arm is more compliant, and therefore, has a greater deflection than the
other two valves given a specific applied pressure. All three valves were leak-tight
under 500 mmHg (66.7 kPa) of reverse pressure; no leakage was measured or
observed. The pressure at which maximum flow was achieved was identical for the
hole and straight arm valves. This is partially due to the choice of pressure
increments, which were not small enough to resolve the slight differences in the hole
and straight arm valve responses. Additionally, as shown in Figure 2-57, the arm
and straight arm valve plates have similar deflection values for applied pressures
between 0 - 500 mmHg (0 – 66.7 kPa).

114
2.3.1.3.5.1
2.3.1.3.5.2
2.3.1.3.5 Packaged Check Valve Closing Time Constant

Methods

The closing time constant for the valve was determined by applying pressurized
water (250, 500, and 750 mmHg, 33.3, 66.6, and 100 kPa respectively) to the valve,
shutting off the pressure with a pneumatic solenoid valve, and measuring the
accumulated volume of water exiting the valve after pressure shut-off. A calibrated
100 μL pipette (Clay Adams, Parsippany, NJ) was used to measure the accumulated
volume. A circuit controlled the pneumatic solenoid valve and a light-emitting diode
was used to indicate the valve state (open or closed). Closing time was defined as
the duration between the elapsed time from pressure shut-off to when 63.2% of the
total accumulated volume had exited the valve.

Results

The valve has a finite response time when the applied pressure is removed. The
closing time constants were determined for three different applied pressures (250,
500 and 750 mmHg, 33.3, 66.7, and 100 kPa, respectively) (Figure 2-62). Recall
that the flow rate is maximal near 500 mmHg (66.7 kPa). An electronically
controlled pneumatic valve, with a response time of 3 ms, was used to provide a near
instantaneous application and shut-off of pressure. Dispensed volume data was
extracted from video footage of the flow tracked by a bubble moving in a 100 μL
pipette after pressure removal. The bubble was observed for 5 minutes after the

pressure was removed from the valve to ensure the valve was completely closed.
Accumulated volume experiments for each pressure were repeated four times. The
data are summarized in Table 2-8.

Figure 2-62 Accumulated volume measurements to determine closing time constant. Closing time
constants were calculated by determining the amount of time for 63.2% of the total accumulated
volume to exit the valve.

Table 2-8 Summary of closing time constants for the packaged valve.
Pressure
[mmHg] ([kPa])
Accumulated Volume
[μL] (mean ± SE, n = 4)
63.2%
Volume [μL]
Closing Time
Constant [sec]
250 (33.3) 9.60 ± 0.079 5.98 7.6
500 (66.7) 19.43 ± 0.378 12.11 7.85
750 (100) 30.13 ± 0.401 18.77 8.05

115

116
Closing time constants were very consistent with only a slight increase in response
time at higher pressures. As pressure increases, the valve plate deflects further from
the valve seat and has a greater distance to travel when returning to the initial
position.

One would expect the highest accumulated volume to correspond to the highest flow
rate (500 mmHg, 66.7 kPa). However, 750 mmHg (100 kPa) resulted in the largest
accumulated volume (Figure 2-62) and is attributed to the difference between
transient and steady state flow rates in the closing time and flow rate profile
experiments, respectively.

The total accumulated dosage after the pressure is removed from the valve depends
on the applied pressure and the amount of time the valve was pressurized (i.e. if the
flow rate has reached steady state). However, the accumulated volume can affect the
final dosage, especially for small targeted values. Therefore, shorter closing times
are desirable; designs to shorten the closing time can be investigated for high
precision applications.

The addition of an electrolysis pump to serve as the driving force should further
decrease the error. The electrolysis structure is made of platinum, which catalyzes
the recombination of oxygen and hydrogen gases back into water. Once the current
to drive the electrolysis pump has been terminated, the internal volume of the pump

117
2.3.1.3.6.1
2.3.1.3.6.2
chamber will decrease. The decreased volume will pull a vacuum in the drug
chamber, thus forcing the packaged valve to close faster.

2.3.1.3.6 Valve Pressurized using Electrolysis Structure

Methods

A packaged valve was attached to a jig which housed a pre-made electrolysis
structure. The valve output was attached to a 100 μL calibrated pipette. The
electrolysis structure was fabricated using platinum on glass. The interdigitated
electrodes had a pattern where the metal fingers were 20 μm wide with a 100 μm
space between the fingers. A 1 mA current was applied to the electrodes.

Results

Preliminary results demonstrated that the valve could be opened with the pressure
generated from the electrolysis structure (Figure 2-63). The experiment was
conducted 4 times with a wide range of results. However, the valve opened within
the first 15 seconds of applied current. Additionally, the flow rate plateaued when
the electrolysis structure became obscured with bubbles.

Figure 2-63 Accumulated volume expelled from the packaged valve using an electrolysis pump and
custom-made jig (mean ± S.E., n = 4).

2.3.1.3.7 Assembled Device Electrolysis and Valve Flow Rate

2.3.1.3.7.1 Method

An assembled device was filled with water and the electrolysis leads were attached
to a power supply and given 1 mA of current. The output of the packaged valve was
attached to a 100 μL calibrated pipette.

118

119
2.3.1.3.7.2 Results

The results from the first test on the assembled device were sporadic. A close
inspection of the device showed electrolysis delamination as well as the degradation
of the conductive epoxy (Figure 2-64). The delamination caused the circuit to break,
thus halting the electrolysis process. The degradation of conductive epoxy resulted
in the epoxy residue resettling along the leads and shorting the circuit halting the
formation of bubbles. Additional tests and optimization are planned to optimize the
electrolysis structure and epoxy to prevent pump failure.

Figure 2-64 Images of the assembled device electrolysis structure. The electrolysis structure
delaminated during current application.

2.3.2 Summary

An electrically-actuated drug delivery device which uses electrolysis actuation and
has a dual-regulation check valve has been fabricated and demonstrated. A modular
dual-regulation check valve, which provides bandpass regulation of flow between

120
two pressure ranges, was verified on benchtop and within an integrated system.
Three variations of the valve plate were investigated to determine the best design for
the ocular drug delivery device application. Theoretical equations and FEM analysis
guided valve design. Check valve cracking pressure, closing pressure and closing
time constants were empirically determined.

This device contains multiple modular components, which can be used to customize
this device to each patient, or used to create a drug delivery mechanism for
applications other than ocular drug delivery.

The dual regulation valve consists of 4 plates, all of which can be fabricated in
parallel with multiple identical copies made during each process run. Additionally,
the valve seat and pressure limiter are the same part, thereby simplifying the process
fabrication process as a process for the pressure limiter is not necessary.
Interchangeable parts, like the manually-actuated device, allow multiple valves to be
assembled in parallel.

Furthermore, adjustments to the valve plate (using the hole, straight arm, or s-shaped
arm design) and the spacer plate can affect the valve performance. The each valve
plate has varying flow resistance and deflection characteristics, thereby changing the
band width of the bandpass flow regulations, cracking pressure, and closing pressure.
Scaling the spacer plate thickness, making it thinner or thicker, will also affect the

121
bandwidth. A thinner spacer plate will decrease the bandwidth, while a thicker
spacer plate will increase the bandwidth.

Finally, the entire packaged valve is a modular component which and be easily
removed or replaced in the assembled device. Valve replacement may be necessary
to change a faulty valve, or to replace with a valve which is better tuned to a specific
application. Additionally, hanging the cannula length may be necessary depending
on the reservoir placement and delivery location. However, the preferred valve
placement is at the cannula tip. Valve placement at the cannula tip ensures the dosed
volume exits the cannula. If the valve were placed at the cannula/ reservoir junction,
the delivery mechanism of the dosed liquid would be significantly different than the
current mechanism as the volume would need to diffuse out of the cannula interior.
The silicone cannula can be shortened by removing the valve, cutting the cannula,
and then re-inserting the valve.

2.4 Surgical Shams

2.4.1 Solid Surgical Shams

A solid surgical sham was designed to provide surgeons a model in which to
optimize the surgical procedure. The solid shams can also be used in chronic
implantation studies to determine the biological response to a shape which is similar
to the final design morphology.

2.4.1.1 Design

Surgical models, or surgical shams, were designed to model the final device shape
and size. Design parameters dictated several design requirements: 1) an oval
reservoir shape, 2) a maximum thickness of 2mm and, 3) an interior volume of at
least 200 μL. The cannula could be placed along the major or minor axis centerline
of the oval shape (Figure 2-67). The targeted sham internal volume dictated the
overall sham dimensions. To calculate the dimensions for the sham, the equation to
calculate the volume of an ellipsoid (Equation 2-4) was modified to calculate all
possible values for the major and minor axes, as well as device thickness (Equation
2-5).

Equation 2-4- Volume of an Ellipsoid
4
3222
ab c
ipsoid
llipsoid
llipsoid
π
Ellipsoid
Volume
a height of ell
b legnthof e
c width of e
=
=
=
=

Equation 2-5- Volume of a Dome
1
6
dome
of dome
of dome
πα
dome
Volume
height of
major axis
minor axis
βχ
α
β
χ
=
=
=
=

Here, the minor axis is defined as the axis through which the silicone tube intersects
where the major axis the perpendicular to the silicone tube (Figure 2-65).

122

Figure 2-65 Definition of major and minor axes on surgical sham devices.

Table 2-9 lists the shams which were fabricated using custom-made laser-ablated
acrylic molds. Laser files for the molds can be found Appendix BB.
123

124

Table 2-9 Dimensions of Fabricated Version 1 Surgical Shams.
Depth [mm] Width [mm] Length [mm] Volume [μL]
0.75 16 31.9 200.43
0.75 31.9 16 200.43
0.875 16 27.3 200.12
0.875 15 29.1 199.98
0.875 14 31.2 200.12
0.875 13.6 32.1 200.01
0.875 27.3 16 200.12
0.875 29.1 15 199.98
0.875 31.2 14 200.12
0.875 32.1 13.6 200.01
1 16 23.9 200.22
1 15 25.5 200.28
1 14 27.3 200.12
1 13 29.4 200.12
1 12 31.9 200.43
1 23.9 16 200.22
1 25.5 15 200.28
1 27.3 14 200.12
1 29.4 13 200.12
1 31.9 12 200.43
2 14 13.7 200.85
2 12 15.9 199.81
2 10 19.1 200.01
2 8 23.9 200.22
2 13.7 14 200.85
2 15.9 12 199.81
2 19.1 10 200.01
2 23.9 8 200.22

The molds created solid surgical shams with suture tabs, a stainless steel ring, and a
silicone cannula. The embedded stainless steel ring is used to model the designated
refill location. The dimensions of the silicone body, cannula, and suture tabs (one set
located on the reservoir of the sham, and two sets along the silicone tube) are given
(Figure 2-66). Two size options for the stainless steel rings, 1) 2.51 mm outer
diameter (O.D.), 1.6 mm inner diameter (I.D.), 0.46 mm thick and 2) 5.31 mm O.D.,

125
2.92 mm ID, 0.46mm thick were presented to the surgical staff. The 5.31 mm O.D.
ring was preferable to provide the maximum target area. Surgeons chose two final
sizes to implant into Dutch Belted male pigmented rabbits for chronic tests. After
initial testing, the sutures surrounding the cannula were removed because they
impeded the insertion of the cannula into the anterior chamber of the eye. The shams
were continually modified to accommodate surgical considerations. The full solid
sham design progression can be found in Section 2.4.1.1.1-Solid Sham Timeline.

Figure 2-66 First version of the implanted solid surgical shams and the dimensions.

2.4.1.1.1 Solid Sham Timeline

A list of the progression of the solid shams, as well as a summary of the design, can
be found in Table 2-10.

126
Table 2-10 Solid sham timeline and description of solid sham characteristics.
Solid Sham Timeline
Sham v1_large v1_small v2_large v2_small v3_1
Mold File
Location
Figure
2-66A
Figure
2-66B
Appendix
CC
Appendix
CC
Appendix
DD
Solid/ Hollow Solid Solid Solid Solid Solid
Material PDMS PDMS
PDMS or
MDX4-
4210
PDMS or
MDX4-
4210
PDMS or
MDX4-
4210
Footprint
Shape
Oval Oval Oval Oval Oval
Major Axis
Length [mm]
29.4 15.9 29.4 15.9 16.9
Minor Axis
Length [mm]
13 12 13 12 13
Thickness
[mm]
1.00 2.00 1.00 2.00 2.00
Volume [µL] 200.12 199.81 200.12 199.81 230.07
Internal
Volume [µL]
N/A N/A N/A N/A N/A
Suture Tab
location
Cannula
and
Reservoir
Cannula
and
Reservoir
Reservoir Reservoir Reservoir
Refill Ring ID
[mm]
2.92 2.92 2.92 2.92 2.92
Refill Ring
OD [mm]
5.31 5.31 5.31 5.31 5.31
Baseplate
Size
N/A N/A N/A N/A N/A

2.4.1.2 In Vivo Experiments

2.4.1.2.1 Implantation

2.4.1.2.1.1 Methods

Two surgical shams, v1_large (1 mm x 13mm x 29.4mm) and v1_small (2mm x
12mm x 15.9mm) (Figure 2-67), were tested in vivo. The larger stainless steel ring

(5.31 mm OD, 2.92 mm ID, 0.46mm thick) was used in these shams. The solid
reservoir was sutured under the conjunctiva and the silicone cannula was introduced
into the anterior chamber of the eye via a scleral tunnel.

Figure 2-67 Implanted surgical shams, A) 2 mm x 12 mm x 15.9 mm and B) 1 mm x 13 mm x 29.4
mm

It was determined that the suture tabs surrounding the silicone tube were unnecessary
however, the larger sham required more suture tabs along the reservoir to anchor the
reservoir in place. A second set of surgical shams (v2_large and v2_small) were
fabricated to facilitate surgical implantation (Figure 2-68). The file used to fabricate
the molds using a laser-cutter can be found in Appendix CC.

127

Figure 2-68 Mold used to fabricate version 2 of the surgical shams. Dimensions are the same as
version 1 with additional sutures on the 1mm thick sham (v2_large) and the sutures removed from the
silicone cannula from both shams.

2.4.1.2.1.2 Results

Chronic in vivo data of v2_large demonstrated that the footprint size was not well
tolerated in the limited space available within the eye wall. The device was difficult
to insert, and was found to cause conjuctival thinning over the device. v2_small had
better results and was tolerated by the ocular tissue in acute in vivo studies.

A slightly larger sham, v3_1 was created to test a device that would have the same
overall shape as a sham with a 200 μL interior. The major and minor axes were
increased by 1 mm to mimic a hollow sham whose interior is a similar shape to
v2_small with a 0.5 mm thick wall. This sham was also tested in acute in vivo
studies and demonstrated minimal adverse affects.

128

129
2.4.1.2.2.1
2.4.1.2.2.2
2.4.1.2.2 Refill Ring

Methods

A stainless steel was embedded into the reservoir. The stainless steel ring is visible
through the conjunctiva. A non-coring needle can be inserted through the center of
the stainless steel ring to refill the device.

Results

The stainless steel ring embedded within the solid surgical sham was visible through
the conjunctiva (Figure 2-69). The surgeon was able to insert a commercially-
available 30 gauge non-coring needle (O.D. 305 μm) through the conjunctiva and
through the center of the stainless steel ring (I.D. 2.92 mm) indicating that targeted
refill is possible. However, a needle stop is still needed to indicate when the needle
has reached the base of the device. Also, a means for visualizing the stainless steel
ring may be necessary if fibrous tissue formation or complications from surgery
prevent the ring from being visible.

Figure 2-69 Implanted solid surgical sham with stainless steel ring visible through the conjunctiva.
The surgeon was able to simulate refill by targeting the center of the stainless steel ring using a
commercially available 30 gauge needle. The stainless steel ring is outlined in this image to help
indicate its location.

2.4.2 Hollow Surgical Shams

The hollow surgical sham contained suture tabs, a stainless steel ring, a
polyetheretherketone (PEEK) (0.254 mm thick) baseplate, and a silicone tube. The
stainless steel ring, as with the solid shams, helps surgeons identify the refill
location. The refill location is where the surgeon can puncture the device and
replenish the device interior. This location will be designed using materials that can
withstand multiple punctures without leakage. The PEEK baseplate, which serves as
a needle stop, prevents the needle from puncturing through the device. The silicone
tube is the cannula that allows the liquid held in the device interior to enter the
anterior chamber. Hollow surgical shams were made so that surgeons could practice
130

refilling a device with a designated refill port and baseplate to prevent the needle
from piercing through the entire device (Figure 2-70).

Figure 2-70 Illustration of the hollow surgical sham. A refill needle access the sham interior by
piercing the refill port location (designated by a refill ring). A PEEK baseplate prevents the needle
from piecing through the entire device.

2.4.2.1 Design

2.4.2.1.1 Needle Stop

In vivo testing of the solid sham demonstrated a need for a mechanism to determine
how far the needle has penetrated into the device. Two main failure modes, both
which fail to refill the device, can occur if needle depth is not controlled. If the
needle insertion is too shallow, part of the needle opening may not be fully inserted
into the device. When the refill syringe is depressed, the refill drug will flow out of
the needle tip and onto the eye surface instead of into the device. However, if the
needle is inserted too far during refill, the needle may puncture through the entire
device, where the needle tip will enter the eye interior. Upon refilling, the drug will
be injected directly into the eye. Additionally, this failure mode will cause further
131

132
2.4.2.1.1.1
damage to eye tissues, similar to the side effects of direct ocular injection treatments.
Therefore, a method to guide needle penetration depth is necessary.

Needle Ring Guide

The needle ring guide idea was previous mentioned in Section 2.2.2.1-Refillable
Reservoir and Refill Guides. A needle ring guide is an indication on the needle shaft
which shows the surgeon how far to insert the needle. A silicone ring, placed on the
needle shaft, can be used to mark the desired needle depth. When the device is
implanted, but prior to covering the device with the conjunctiva, the surgeon can
insert a needle into the device and adjust the location of the silicone ring to mark
how far the needle must be inserted to safely reach the device interior (Figure 2-71).
During refill, the surgeon only inserts the needle until the silicone ring touches the
eye surface.

Figure 2-71 Illustration of the application of a refill ring on the refill needle.

However, it was determined that this method of regulating needle depth does not
account for conjunctival thickness, or changes that may occur to the ocular tissue
during the lifetime of the device (e.g. fibrous tissue formation). Therefore, a second
method for controlling needle depth must be developed.

2.4.2.1.1.2 Rigid Device Baseplate

A rigid baseplate was proposed as a method for preventing the needle from
puncturing through the entire device. The device thickness measures 2 mm in
thickness and the opening of a beveled needle was measured to be 1.2 mm long from
133

needle tip to the opening termination. Therefore, if the needle is inserted until the
needle tip touches the baseplate, then the entire needle opening is contained with the
device.

PEEK (polyetheretherketone) was chosen as the baseplate material because it is a
USP Class VI biocompatible material. Furthermore, thin PEEK sheets (0.01’’) are
commercially available; therefore the PEEK baseplate can be added to the device
base without increasing the overall device thickness (Figure 2-72).

Figure 2-72 Image of the PEEK baseplate to limit refill needle insertion depth.

2.4.2.1.2 Hollow Sham Timeline

134
Several versions of hollow surgical shams were designed and fabricated to
accommodate surgical needs and requests. Each sham was created using custom-

135
made laser-machined acrylic molds. A complete timeline of the hollow sham
devices can be found in Table 2-11.

v3_2 is a hollow counterpart to the v3_1 solid sham. However, this hollow sham did
not contain a rigid baseplate because it was a benchtop demonstration prototype.
v3_3 was fabricated using medical-grade silicone (MDX4-4201) and included a
baseplate. The baseplate was made using a hole-punch, and was placed directly
beneath the refill ring. However, it was determined that if the refill needle was
inserted at an angle (as opposed to perfectly perpendicular) to the device surface, it
was possible for the needle tip to miss the PEEK baseplate. Therefore, v4_1 was
fabricated with a larger PEEK baseplate that would cover the majority of the device
footprint.

v5_1 and v6_1 were created to be better suited for chronic in vivo testing with rabbit
eyes. They are approximately 40% smaller than the v3 devices. The major
difference between v5_1 and v6_1 was the introduction of a PEEK baseplate that is
the same shape and size as the device footprint. Additionally, a smaller refill ring
was necessary in order to fit the smaller device.

v7_1 is the initial design for the shape of the modular device which integrates all of
the major device components: refill reservoir, refill ring, PEEK baseplate, suture
tabs, electrolysis pump, electrolysis bellows, and dual regulation check valve. The
reservoir body is different from the previous shams as it is two overlapping dome

136
shapes as opposed to a single dome (Figure 2-73). This layout separates the main
reservoir from the refill port; preventing the refill needle from accidentally
puncturing the pump bellows during refill.

Figure 2-73 Top and side view of the initial reservoir design for the integrated drug delivery device.
The reservoir body is separated from the refill port to prevent the pump chamber and Parylene C
bellows from accidentally being punctured by the refill needle.

In the final integrated device, the cannula diameter was increased to accommodate
the dual check valve. Additionally, the cannula was moved away from the refill port
to prevent refill liquid from exiting the cannula as opposed to refilling the reservoir.
The final integrated device is discussed in Section 2.3.1.2.6- Device Assembly.
.

Table 2-11 Timeline for hollow surgical sham, including major device characteristics.
Hollow Sham Timeline
Sham v3_2 v3_3 v4_1 v5_1 v6_1 v7
Mold Diagram
Appendix
EE

Appendix
EE
Appendix
EE
Appendix
FF
Appendix FF Appendix GG
Solid/ Hollow Hollow Hollow Hollow Hollow Hollow Hollow
Material
PDMS or
MDX4-
4210
PDMS or
MDX4-
4210
PDMS or
MDX4-
4210
MDX4-
4210
MDX4-4210 MDX4-4210
Footprint Shape Oval Oval Oval Oval Oval
Overlapping
Offset Circles
Major Axis
Length [mm]
16.9 16.9 16.9 9.845 9.845 14
Minor Axis
Length [mm]
13 13 13 7.7 7.7
10.285 and
6.655
Thickness [mm] 1.89 1.89 1.89 1.65 1.65 2
Volume [µL] 217.88 217.88 217.88 65.49 65.49 413.43
Internal Volume
[µL]
126.08 126.08 126.08 42.94 42.94 213.12
Suture Tab
location
Reservoir Reservoir Reservoir Reservoir Reservoir Reservoir
Refill Ring ID
[mm]
2.92 2.92 2.92 1.98 1.98 Bump
Refill Ring OD
[mm]
5.31 5.31 5.31 4.76 4.76 Bump
Baseplate Size
[mm]
N/A
Circle
D: 6.53
Circle
D:11.66
Circle
D: 6.53
Oval
9.845 x 7.7
Overlapping
Offset Circles
D: 8.5 & 5.5

137

2.4.2.1.3 Fabrication

The steps to fabricate all of the hollow sham designs were similar. The reservoir is
made by molding PDMS between a convex and concave dome molds, while the base
is fabricated by molding the PDMS into the desired footprint shape and
incorporating a silicone tube (Figure 2-74).

Figure 2-74 Illustration of acrylic molds used to fabricate the hollow surgical sham.

The reservoir dome and the device base were made separately and assembled. The
pieces can be made in parallel; however, for clarity each fabrication process is
described separately. A list of the fabrication process steps for oval shaped hollow
shams can be found in Appendix HH.

138

139
2.4.2.1.3.1
2.4.2.1.3.2
Reservoir

To make the dome portion, PDMS prepolymer was poured into the concave half of
the dome mold (Figure 2-75B). A stainless steel washer (ID 2.92mm, OD 5.31mm)
was placed into the concave dome mold (Figure 2-75C). The mold was then
transferred to a vacuum oven to degas the PDMS prepolymer. The convex half of
the dome mold was then aligned and pressed onto the concave mold (Figure 2-75D).
Two glass microscope cover slips were used to separate the two halves of the mold.
The mold was then placed into an oven for 30 minutes at 70 °C to rapidly cure the
PDMS. Once cured, the dome piece was removed from the mold and excess PDMS
was cut from the dome (Figure 2-75E).

Base

The device base was made by pouring PDMS prepolymer into the base mold (Figure
2-75B). The mold was placed into a vacuum oven to degas the PDMS. Once
degassed, the PEEK baseplate was carefully placed into the base mold, ensuring
bubbles were not introduced into the PDMS (Figure 2-75C). The baseplate was
made by cutting the desired baseplate size from a polyetheretherketone (PEEK) sheet
(0.254 mm thick). A one inch length of silicone tubing (0.305 mm ID, 0.61 mm
OD), threaded with a stripped piece of 30 gauge wire (0.254 mm diameter), is
inserted into the indentation on the base mold which indicates the tube location. The
wire prevents the tube from being clogged by the PDMS prepolymer. The mold was

140
2.4.2.1.3.3
placed in an oven at 70 °C for 30 minutes for rapid curing. The cured base piece was
removed lifted from the base mold; extra PDMS was carefully cut from the base
piece and the wire is removed from the tube (Figure 2-75E).

Device Assembly

Once both pieces of the hollow sham were ready the two pieces were carefully
aligned (Figure 2-75F). A thin line of PDMS prepolymer was applied along the edge
of the two pieces, bonding them together (Figure 2-75F). The bonded device was
then placed in an oven (70 °C) for 10 minutes to completely cure the PDMS.

Figure 2-75 Fabrication steps for making the hollow surgical sham.

2.4.2.1.4 Benchtop Verification

The hollow sham was filled on the benchtop by piercing the device through the
middle of the stainless steel ring using a 30 gauge non-coring needle (Figure 2-76).
The needle was pushed into the device until the needle tip progression through the
device was prevented by the PEEK baseplate. Dye liquid is injected into the device
until excess liquid exits the cannula.

141

Figure 2-76- Hollow surgical sham being filled on benchtop.

The contents within the sham were dispensed by manually depressing the reservoir
(Figure 2-77). The sham was refilled and dispensed multiple times.

Figure 2-77 Benchtop demonstration of manual dispensation of dyed liquid from within a hollow
sham device.

142

143
2.4.2.2.1.1
2.4.2.2 Acute and Chronic In Vivo Experiments- Methods and
Results

2.4.2.2.1 Acute In Vivo Dispensation

Methods

The hollow sham will be implanted into rabbits to demonstrate device functionality
in an acute in vivo study. The device is placed in the superior temporal quadrant of
the eye, with the cannula entering the anterior chamber of the eye. The device fully
enclosed with the eye wall by covering the device with the conjunctiva. This
minimizes the possibility of infection entering the eye interior (Figure 2-78).

Figure 2-78 Illustration of the hollow sham placement in the eye.
Prior to implantation of the device, hollow shams were sterilized. The sterilized
sham was filled on benchtop with saline solution and manually-depressed to verify
device functionality before implantation. The device was then refilled with a
mixture containing phenylephrine (10% concentration) and Trypan blue dye. The
silicone tube was pinched closed by tying a suture around the base of the tube; the
tube was closed off to prevent accidental dispensation of phenylephrine during
implantation.

The device was implanted beneath the conjunctiva and secured with suture tabs. A 3
mm wide scleral tunnel was created 2 mm posterior to the limbus. The cannula of
the device was cut so that the tip of the cannula extended approximately 2 mm into
144

the anterior chamber. The tip was cut at an angle; this created a beveled tip on the
cannula which aided insertion of the cannula through the scleral tunnel. The closing
suture around the cannula was removed and device functionality was verified prior to
suturing the conjunctiva over the device (Figure 2-79).

Figure 2-79 Images of hollow sham device implantation for acute and chronic in vivo studies

The baseline pupil diameters were measured prior to releasing the suture closing the
cannula. The suture around the cannula was cut and the device was manually
depressed using blunt forceps. The pupil dilation was measured after dispensation
and again after the conjunctiva was replaced over the device and sutured in place.
The surgical protocol is listed in Appendix II.
145

146
2.4.2.2.1.2 Results

Two Parylene C coated devices were implanted in the right eye of Dutch-belted
pigmented rabbits, the left eye acted as the control. The baseline horizontal and
vertical pupillary diameters were measured immediately prior to the release of the
suture closing the cannula. The phenylephrine was delivered by supporting the base
of the device with blunt forceps and depressing the reservoir using a Q-tip; this event
was marked as t= 0. At t=1 min 29 sec and t=9 min 58 sec, pupil diameter
measurement was taken. A summary of the pupil diameter values can be found in
Table 2-12.

Table 2-12 Summary of Results from In Vivo Delivery using Hollow Surgical Sham
Pupil Diameter Vertical Horizontal
Baseline
t = 0 minutes
4.5 mm 4.5 mm
After Dispensation
t = 1 min 29 sec
7 mm 7 mm
After Dispensation
t = 9 min 58 sec
7.5 7.5
Total Change 3 mm (66% change) 3 mm (66% change)

It was also shown that the stainless steel ring of an implanted device can be made
more visible by transilluminating the device. A light is placed along the exterior of
the eye. The stainless steel ring is seen as a darker shape compared to the
illuminated eye tissue (Figure 2-80). This minimally invasive and real-time method
of identifying the stainless steel ring will aid medical personnel when refilling the
device.

Figure 2-80 Transilluminated eye with implanted hollow sham device. The device and stainless steel
ring outlines are clearly visible through the eye tissue (i.e. conjunctiva) covering the device.

2.4.2.2.2 Chronic In Vivo Dispensation and Refill

2.4.2.2.2.1 Methods

The chronic study of the device was conducted to verify device functionality over a 6
month period. The device was implanted using the same protocol listed in the acute
study. After the initial verification of dispensation of the implanted device, the
device was refilled once a month with Trypan blue (Figure 2-81).
147

Figure 2-81 Still images of surgical video taken during device refill in a chronic in vivo study. A)
Transillumination of the eye helps locate and identify the target refill area. B) A 30G needle is
inserted through the center of the refill area. Needle insertion stops when the needle tip encounters
the rigid baseplate embedded in the device base. C) Trypan blue dye is injected into the device; the
dye can be observed spreading through the device.

Each month, the surgeons determined if any biofouling prevented them from
identifying the refill location, as well as if the tube was occluded and prevented
successful delivery of the dye into the eye. Biocompatibility was also monitored
using color photography and fluorescein angiography. At the end of the study, the
endothelial cell density was counted, and any damage to ocular tissue was noted.

2.4.2.2.2.2 Results

Surgical procedures were deemed by the surgeons to be minimally invasive and well
tolerated. The time to implant a surgical sham from the first incision in the
conjunctiva to suturing the conjunctiva over the device was approximately 60
minutes. All of the implanted devices were fabricated using USP Class VI,
biocompatible materials and did not show any biofouling to affect device
148

functionality during 6-month follow-up period. Additionally, no leakage around the
insertion site or filtering bleb formation over any of the implants was observed.

Devices were refilled up to 6 times at 4- to 6-week intervals during a period of 4 to 6
months. Transconjunctival refilling was performed in less than 1 minute by a single
surgeon without any complications (Figure 2-82).

Figure 2-82 Images of in vivo device refilling and dispensing. First, the refill site is checked for any
damage, infection, or scarring from previous refills. Next, the eye is transilluminated to help identify
the refill ring location (the refill ring appears as a darker shadow). The refill needle (30G non-coring)
is inserted through the center of the refill ring until the needle progressing is stopped by the device
baseplate. Trypan blue dye is injected into the device; dye can be seen spreading through the device
as a dark plume. The dye exits the cannula and into the anterior chamber. Finally, the puncture site is
inspected for damage or leakage.

149

150
Anterior chamber depth and intraocular pressure were measured after surgical
implantation and monthly refilling; values were normal in all implanted eyes
compared to the contralateral control eyes. Surgeons reported no retinal or optical
disc damage on indirect ophthalmoscopy examination or fluorescein angiogram.
Examinations slit-lamps, anterior color photography, or fluorescein angiograms did
not show any cornea, iris or lens damage. Finally, the surgeon reported observing no
infection of the ocular tissue surrounding the device, all of the devices remained in
place (i.e. did not extrude), and the cannula was not occluded throughout the entire
duration of the study.

At the conclusion of the chronic study, ocular tissue was prepared to determine any
tissue damage from the implant or refill events. Light microscopic examination
showed endothelial cell loss close to the cannula insertion site. The difference in the
final corneal thickness mean between the implanted (0.42 ± 0.02 mm) and control
(0.41 ± 0.02 mm) eyes was not statistically significant; the mean peripheral corneal
thickness in implanted eyes was 0.49 ± 0.03 mm vs. 0.51 ± 0.02 mm in control.
Additionally, no evidence of corneal edema was seen and epithelial integrity was
uniform. However, scanning electron microscope images of the endothelial side of
the cornea showed changes in density, size, structure and morphology of endothelial
cells in all quadrants of the cornea after 6 months.

A summary of the mean endothelial cell density in the implanted and control eyes
are found in Table 2-13. However, it was found that the endothelial cell changes

151
improved significantly in one rabbit in which refill and dispensation of Trypan blue
solution was ceased at 4 months after surgery.

Table 2-13 Summary of endothelial cell density at the conclusion of the 6 month study for eyes that
were implanted and refilled 6 times during the course of the study and an implant were the refills were
terminated 2 months prior to the end of the study at month 4.
Implanted Eye Control Eye
Superior Temporal
Quadrant Cornea
(6 months, 6 refills)
1348 ± 231 cells/mm
2
4511 ± 177 cells/mm
2

Central Cornea
(6 months, 6 refills)
2603 ± 214 cells/mm
2
4305 ± 202 cells/mm
2

Superior Temporal
Quadrant Cornea
(6 months, 4 refills, refill
ceased at 4 months)
4161 ± 278 cells/mm
2
4806 ± 159 cells/mm
2

Central Cornea
(6 months, 4 refills, refill
discontinued at 4 months)
4326 ± 244 cells/ mm
2
4551± 126 cells/mm
2

A device was implanted for 3 weeks and removed prior to any refill or dispensation
to determine if surgical implantation caused any damage. SEM images of the
endothelial cells appear to be similar to the control and endothelial cells of the device
with fewer refills.

Corneal endothelial cell damage in the implants is similar to those found in
commercially available glaucoma drainage device implantation such as the Ahmed
glaucoma valve

(Kim, et al. 2008, Lim 2003). However, endothelial cell loss can be
minimized by restricting the movement of the cannula relative to the cornea. The
most common causes of damage due to macroscopic contact of cannula with the
cornea is non-ideal positioning of a long tube or anterior chamber shallowing (Lim

152
2003). These results suggest a careful placement of a shorter bevel-tipped cannula
with a rounded edge may decrease endothelial damage. The difference in endothelial
cellular damage between the implants that were refilled and dispensed Trypan blue
for the full period of the 6 month study versus the implant that was not refilled for
the last 2 months of the study may be attributed to toxicity of Trypan blue (van
Dooren, et al. 2004).

2.4.3 Summary

Surgical shams to optimize the device shape, component functionality, and
component placement were created. Acute ex vivo studies using the shams allowed
surgeons to create and refine a surgical protocol for implanting and refilling the
electronically-actuated device. Chronic, 6 month, in vivo studies demonstrated
device biocompatibility and tolerance to monthly refills. No biofouling of the device
or cannula was observed. Some endothelial cell damage was found, but may be
attributed to the toxicity of Trypan blue, the dying agent used to verify device
delivery.

Similar to the manually-actuated drug delivery device, the shams are fabricated using
interchangeable layers. Replacement of layers is also possible, where the dome layer
can be change to increase or decrease the interior volume. As demonstrated in the
sham timeline, the dimensions of the refill ring and PEEK baseplate were easily

153
changed and incorporated into the dome or base layers. Additionally, the location of
the refill port can be moved without affecting the base layer or cannula location.

2.4.4 Additional Applications

The applications of MEMS based, refillable drug delivery system can be extended to
any application where precise delivery in a difficult to access, localized area is
necessary. Two additional drug delivery device applications, which are in the initial
phase of development and are currently being investigated, are presented.

2.4.4.1 Rat Retinitis Pigmentosa Drug Delivery Device

A prototype device which is capable to delivering medications which treat retinitis
pigmentosa (RI) into a rat animal model has been proposed. Retinitis pigmentosa is
a degenerative disease that destroys photoreceptors, leading to irreversible damage of
ocular tissues. The final device will contain a refillable reservoir, electrolysis pump,
cannula, flow control valves, and suture tabs.

2.4.4.1.1 Design and Fabrication

The initial device is based on the hollow surgical models described in Section 2.4.2-
Hollow Surgical Shams. Similar to the hollow shams, the Rat RI device is fabricated

using medical-grade silicone (MDX4-4210), and constructed be assembling a
molded dome on a base with an embedded cannula.

The cannula from the hollow shams (silicone tube, O.D. 0.024’’, 0.609 mm) is much
too large for a rat eye; which is typically 4 mm in diameter. Therefore, a smaller
cannula is incorporated into the device. Polytetrafluoroethylene (PTFE) cannula, 38
or 34 AWG in size (Zeus Inc., Orangeburg, SC, I.D.: 0.102 mm or 0.152 mm, wall
thickness: 0.051 mm or 0.076 mm, respectively), were used.

Additionally, the device cannot be sutured into the eye wall (as was proposed for the
large mammal ocular drug delivery device). Instead, the device is secured to the rat
skull using suture tabs and 0-80 thread 1/8’’ hex screw bone screws (92196A052,
McMasterCarr, Santa Fe Springs, CA) (Figure 2-83). The flexible cannula is then
threaded down to the eye for insertion.

Figure 2-83 Assembled device with bone screws used to secure device to rat skull.
154

Initial molds for the rat RI device were fabricated based on the dimensions of a rat
skull (Figure 2-84). The skull is approximately 2 cm in width and 4 cm in length.

Figure 2-84 Image of proposed device superimposed on a image of a rat skull.

However, surgeons asked for a 25% reduction in size in order ensure the device
could be easily implanted. The molds for the smaller device are shown in Figure
2-85. The device made from these molds has an internal volume of 56.2 μL

155

Figure 2-85 Laser file to create molds for the rat retinitis pigmentosa drug delivery device.

The purpose of the reservoir convex and concave molds and the device base mold
are identical to the hollow sham molds. The curing stand piece was created to
prevent the suture tabs from becoming occluded during device assembly. The curing
piece was marked with dots located at the centers of each suture tab. Holes 1.27 mm
in diameter will drilled partially through the Plexiglas mold at each of the dots. A 18
G needle shaft (O.D. 1.27 mm) was placed into each hole to create the stand (Figure
2-86a). Once the device base was fabricated, it was placed onto the stand such that
156

each needle shaft penetrated the corresponding suture tab. The reservoir dome was
then aligned and affixed onto the base (Figure 2-86b). Once cured, the entire device
was removed form the stand.

Figure 2-86 Image of a) curing stand to prevent suture tabs from becoming sealed during assembly,
b) device assembly on curing stand.

The PEEK cutout guide was used to trace a PEEK baseplate that was the same size
and shape as the oval portion of the device baseplate.

2.4.4.1.2 In Vivo Testing

2.4.4.1.2.1 Methods

The device was tested in vivo on mature adult rats. The rat skull was exposed by
making a small incision in the scalp. The device was placed on the skull to
determine the location for 3 securing bone screws. Three holes were drilled into the
skull using a #56 drill bit and hand-held drill pen. The device was then secured to
157

158
2.4.4.1.2.2
the skull by placing the bone screws through the suture holes and pre-drilled holes in
the skull.

A small scleral tunnel was made in the rat eye between the cornea and retina using a
28G needle. The lens in a rat eye is particularly large compared to the rat eye;
additionally, the space between the cornea and retina is very small. Therefore, the
surgeon must take particular care when making the scleral tunnel to not touch the
lens, nor cause retinal detachment.

The cannula was then inserted into the scleral tunnel. Small sutures were used to
close the tunnel around the cannula.

Results

The initial surgery determined the device was both simple to 1) secured to the skull
and, 2) insert the cannula into the eye. However, the bending radius of the cannula
limited cannula motion. The bending radius prevented the cannula from following
the natural curve of the animal head from the skull to the eye. Kinks in the cannula
significantly altered flow resistance. Therefore, it was proposed that the cannula be
made of silicone and terminates with a small length of the PTFE tube. Only the
PTFE portion of the cannula will enter the eye. Additionally, suture tabs will be
placed at the along the silicone tube and silicone/ PTFE junction to help secure the
cannula to the animal as well as to the eye, respectively.

159
2.4.4.2.1.1
2.4.4.2 Cancer Treatment Device

A device which is capable of delivery pharmacological solutions to cancerous
growths in a mouse animal model was proposed. Two different devices and delivery
modes were investigated to determine if continuous delivery to a tumor can
significantly slow the growth of the tumor. Both devices have a refillable reservoir,
flexible silicone cannula, and PEEK baseplate. However, one set of devices relies on
diffusion for drug dosing, which the other will contain an electrolysis pump.

Two cancerous growths were cultivated in on the hindquarters of a mouse animal
model (one per side). An incision was made in the skin covering the back of the
animal. The device was secured to the subcutaneous muscle. The cannula tip was
directed to one of the tumors. The size of the treated tumor was compared to the
untreated tumor to determine the effectiveness of the rat cancer drug delivery device.

2.4.4.2.1 Design and Fabrication

Diffusion Device

The hollow sham molds from the ocular drug delivery device was used as the initial
prototype for the diffusion driven mouse cancer delivery device. Both v4_1 and
v6_1 were used to determine if one size was better suited for the animal model. The
diffusion device was not optimized for this study; however, shams v4_1 and v6_1

160
2.4.4.2.1.2
were determined to be suitable for this application. Fabrication steps for the v4_1
and v6_1 sham devices can be found in Section 2.4.2.1.3- Fabrication.

Electrolysis Delivery Device

The electrolysis delivery device dimensions were driven by the needs of the
electrolysis pump and Parylene C bellows. The device body diameter must be larger
than the pump footprint and the height of the device needed to accommodate a fully
expanded bellows. Diamensions of the device components can be found in Table
2-14.

Table 2-14 Summary of the component diamensions for the electrolysis driven device.
Component Dimensions Volume [µL], Est
dia = 18 mm
Dome
h = 3.5 mm
674.68
length = 63.5 mm, 2.5''
I.D. = 0.305 mm, 0.012''
Cannula
O.D. = 0.610 mm, 0.024''
4.63
dia
1
= 3.5 mm
dia
2
= 4.5 mm
Unactuated Bellows
h = 1.6 mm
81.68
dia
1
=13 mm
dia
2
= 6 mm
Pump Substrate
(PEEK)
t = 0.26 mm
36.35
Total Drug Reservoir Space 561.29

A refill ring was not included in this device because it would not be visible through
the skin of the animal. Therefore, a raised bump was placed on the reservoir. This
bump can be felt through the skin to help indentify the refill location. A refill bump

161
also increases the amount of material at the refill location. As shown in Figure 2-35,
a thicker refill port increases the leakage pressure value and increases the number of
times the device can be refilled. Additionally, the bump was located off-center to
prevent the refill needle from puncturing the Parylene C bellows and compromising
the pump chamber. The pump chamber was also placed off-center so that the edge
of the bellows did not overlap any portion of the refill bump.

The device body was fabricated in a manner similar to the hollow surgical shams; the
molds used to fabricate the pieces of the device body can be found in Figure 2-87. In
summary, the reservoir top and base molds were filled with medical-grade silicone,
placed under vacuum, and cured. However, an additional spacer piece was also
made for this device to add extra height to the device interior in order to
accommodate the Parylene C bellows. The spacer piece also had an opening, which
served as a pass-through for the electrical connections to the electrolysis electrodes.

Figure 2-87 Laser file to create molds for mouse cancer drug delivery device.

162

Device assembly begins by fabricating all of the individual pieces for the device
body: (reservoir dome, reservoir base, spacer layer), and the packaged actuator
(Figure 2-88).

Figure 2-88 Image of the components to fabricate the rat cancer drug delivery device (device body
and electrolysis actuator).

The device was assembled by first securing the electrolysis actuator to the reservoir
base using a small drop of silicone prepolymer. The silicone was cured in an oven
for 5 minutes at 70 ºC. DI H
2
O soaked paper towels were placed near the device in
the oven to increase the humidity around the device in order to mitigate any diffusion
of the water contained within the actuator.

Next, the spacer piece was aligned to the reservoir base. The spacer plate follows the
outline of the base. The wires from the electrolysis actuator were aligned to the
opening in the spacer piece. The spacer piece was secured with silicone prepolymer
and cured in the oven.
163

The holes in the suture tabs were created using a 16G coring syringe needle (O.D.
1.651 mm). 16G needle shafts where then inserted into the holes to prevent the holes
from becoming clogged during the final assembly stages.

The reservoir dome was then affixed to the spacer piece using silicone prepolymer.
The orientation of the dome was adjusted to place the refill bump over a portion of
the reservoir interior not occupied by the electrolysis actuator.

The device was then filled by piercing the refill bump with a 30G needle (O.D. 305
μm) non-coring needle to verify the absence of leaks in the reservoir assembly. The
metal shafts were removed from the suture holes and the device was refilled again to
ensure no leaks were present at the suture locations. The wires were covered using
electrical heat-shrink tubing to prevent the animal from chewing through to the metal
core of the wires (Figure 2-89).

Figure 2-89 Fully assembled rat cancer drug delivery device with electrolysis actuator. Heat-shrink
wrapping on wires not shown.

164

165
2.4.4.2.2.1
2.4.4.2.2.2
2.4.4.2.2 In Vivo Testing

Methods

The device was tested in a mouse animal model. Cancerous tissue was grown on the
hindquarters of the animal (one per side) starting at Day 0 until Day 3. One growth
will be treated with drug from the device, while the contralateral growth will serve as
the control.

An incision was made through the skin on the back of the animal. The device was
sutured to the back of the animal with the cannula tip placed in close proximity to
one of the cancerous growths. The devices were then filled with either siRNA drug,
gold particles, siRNA and gold nanoparticles, or phosphate buffer solution. The
cancer growths were monitored for 19 days following implantation.

The electrolysis pump device was refilled once and was pumped every day for
approximately 30 minutes (0.78 mA) to deliver a total of 50 μL per dosing period.
The diffusion pumps were not refilled.

Results

After 19 days, preliminary data shows only the active pumping device demonstrated
any significant difference between the treated cancer growth and the control growth.

166
The treated growth was approximately 30% smaller in volume than the untreated
growth.

One potential reason the diffusion pumps were not as successful may be attributed to
the slow administration of drug. However, one possible failure mode is an air gap
anywhere in the fluid path between the liquid filled reservoir and the cannula tip.
This air gap would prevent diffusion of the drug into the body.

Further tests are planned for an optimized drug delivery device with an electrolysis
actuator. The optimized device will be smaller to have a better fit with the animal
model and have wireless power actuation. Additionally, the device will be fabricated
with two cannulae in order to allow treatment of either two cancer growths, or to
provide two points of treatment on the same growth.

2.5 Future Work

Proof-of-concept, manually-actuated, electrically-actuated and surgical model drug
delivery devices have been fabricated and tested. However, the final device design
can still be further refined. All of the major components within the fully integrated
device presented in this work: a refillable reservoir and refill port, cannula, dual
regulation check valve can be optimized. A rigid reservoir with a resealable refill
port should be investigated. The cannula diameter can be further reduced in size to
minimize the invasiveness of the device. The check valve can also be reduced in

167
size to fit within the smaller cannula. Achieving repeatability of check valve results
should be investigated but may be realized during commercial fabrication of the
check valve. Additionally, a fully integrated device, along with the electrolysis
pump, and separate pump/ drug chambers needs to demonstrated in a chronic in vivo
study.

168
Chapter 3 Microfluidic Interconnects

3.1 Introduction

Microfluidics, and in particular micro total analysis systems (μTAS) and lab-on-a-
chip (LOC), have many features advantageous for advanced chemical and biological
analyses. The improved performance achieved in part through system
miniaturization, laminar flow, high throughput, reduced sample consumption, and
shorter analysis time has enabled new diagnostic tools and resulted in the application
of these technologies to areas such as chemical synthesis, genetic analysis, drug
screening, and even single cell/molecule analysis (Ho and Tai 1998, Whitesides
2006). To realize the full potential of microfluidics, a suitable packaging technology
to reliably couple micromachined microchannels to the macro-world is required.

The micro-to-macro fluidic interfaces, or interconnects, are usually custom designed
solutions and manually assembled one at a time. The assembly process is time
consuming and can require precision alignment or complicated fabrication steps.
The lack of a batch fabricated approach to interconnects greatly complicates the
packaging of the microfluidic system and is an obstacle to the widespread
commercialization of microfluidic systems. A modular design which can be easily
fabricated, incorporated and customized for each system would help address the
issues facing MEMS packaging.

169
Many interconnects have out-of-plane interfaces in which the tubing and
interconnect are connected perpendicularly with respect to the substrate surface
(Christensen, et al. 2005, Galambos, et al. 2001, Lee, et al. 2004, Li and Chen 2003,
Matsumoto, et al. 2003, Meng, et al. 2001, Pattekar and Kothare 2003, Puntambekar
and Ahn 2002, Yang and Maeda 2003). Out-of-plane interconnects can obstruct
optical viewing of the microfluidic system and interfere with operation of
microscope objectives. To accommodate microscope operation, interconnects are
placed further apart which increases dead volume and overall system size. Robust
mechanical connections in an out-of-plane approach are difficult to achieve due to
limited contact area between the device and tubing. Adhesives can reinforce the
connection (Li and Chen 2003, Meng, et al. 2001, Pattekar and Kothare 2003), but
are difficult to control; device yield is reduced when the adhesive bleeds beyond the
application area and clogs the microfluidic system. When a removable connection or
modular approach is desired as opposed to a permanent connection, adhesives cannot
be used. Most existing interconnects cannot be removed from the system without
leaving a permanent fluidic breach in the system (Baldock, et al. 2004, Chiou and
Lee 2004, Christensen, et al. 2005, Lee, et al. 2004, Li and Chen 2003, Meng, et al.
2001, Pattekar and Kothare 2003, Puntambekar and Ahn 2002). A summary of the
advantages and limitations of connector orientation and attachment methods is
presented in Table 3-1. An alternative approach to forming connections that are
robust but do not require additional complex fabrication steps, precision alignment,
or adhesives is presented.

Table 3-1 Comparison of Connector Design Options
Category References Pros Cons
Orientation
Vertical (Anderson, et al. 2000, Bhagat, et al. 2007,
Christensen, et al. 2005, Li and Chen 2003,
Meng, et al. 2001, Murphy, et al. 2007,
Pattekar and Kothare 2003, Yang and
Maeda 2003, Yao, et al. 2000), Upchurch
Scientific
• Possibility of large surface
area available for fluidic
connections

• Mechanically unstable
• Increased dead volume
• Can obstruct microscope
observations
• Increased lateral dimensions
Horizontal (Baldock, et al. 2004, Chiou and Lee 2004,
Dahlin, et al. 2005, Lo and Meng 2008)
• Mechanically robust
• Less dead volume

• Increased device thickness to
accommodate interconnect/
tubing
Attachment Method
Adhesives (Lee, et al. 2004, Li and Chen 2003, Meng,
et al. 2001, Murphy, et al. 2007, Pattekar
and Kothare 2003)
• Simple
• Commercially-available
materials
• Increased mechanical
strength and maximum
pressure range
• May clog device
• Crack formation within adhesive,
leads to leaking and decreased
operating pressure range

Thermal (Meng, et al. 2001, Murphy, et al. 2007,
Pattekar and Kothare 2003, Puntambekar
and Ahn 2002)
• Simple
• Water-tight seal

• May not be thermally compatible
with other device materials

170

171
Table 3-1 Comparison of Connector Design Options (cont)
Category References Pros Cons
Compression (Anderson, et al. 2000, Christensen, et al.
2005, Dahlin, et al. 2005, Lee, et al. 2004,
Liu, et al. 2003, Yao, et al. 2000)
• Rapid connection
• Tube can be replaced

• System operating pressure a
function of surface area contact
between tube and connector
• Removing tube may cause
leakage
Other (Baldock, et al. 2004, Chiou and Lee 2004,
Yang and Maeda 2003)
N/A N/A
Reusable (Anderson, et al. 2000, Christensen, et al.
2005, Dahlin, et al. 2005, Lee, et al. 2004,
Liu, et al. 2003, Lo and Meng 2008, Yao, et
al. 2000)
• Can replace damaged or
contaminated connections

• Removing tubing may expose
microfluidic interior

A novel interconnect is investigated to address the current challenges in
microfluidics packaging. The interconnect design follows the plug-in format (pin-
and-socket) found in microelectronics. In this case, the “pin” is a commercially-
available, small diameter needle while the “socket” corresponds to an integrated
PDMS septum contained within a SU-8 housing located at the inlet and/or outlet of a
microfluidic system (Figure 3-1). The SU-8 housing also serves to designate the
needle insertion area and forms the microchannel between the inlet and outlet. The
microfluidic system can be accessed by piercing a needle through the septum. By
using a non-coring syringe needle, fluidic connections may be repeatedly established
via septa; once the needle is removed, the PDMS septum reseals preserving the
integrity of the microfluidic system and any contained fluids (Lo, et al. 2006).

Figure 3-1 Microfluidic system with integrated circular interconnects. 33 gauge non-coring needles
were inserted into the input and output septa. Rhodamine was introduced into the system to
demonstrate system functionality. PDMS septum is outlined to indicate its location.

The in-plane layout of the interconnect, in which the interconnect is oriented in the
same plane as the substrate, enables robust microfluidic connections by increasing
the contact area between the macro-world needle and the microchannels in the chip.
Connections are established on-demand and without the need for adhesives,
precision alignment, thru-wafer drilling/etching, or other complex post-processing
172

173
and assembly steps. Furthermore, the interconnect does not impede microscope
observation of the system. This microfluidic interconnect approach incorporates
packaging in the design and layout of the microfluidic system and can be batch
fabricated. These interconnects are easily adapted to other microfluidic system
designs by simply adding a single mask to an existing fabrication process to create
SU-8 anchors for the septa. Additionally, the interconnect design is applicable to
both bulk and surface micromachined microfluidic devices. The single interconnect
design can be extended into an arrayed interconnect design where multiple needles
simultaneously pierce an array of septa. Both the single and arrayed interconnects
are presented.

3.2 Single Interconnect

3.2.1 Interconnect Design

The pin-and-socket interconnect consists of a PDMS septum housed in a SU-8
anchor and a non-coring needle that punctures the septum to establish the macro-
micro interface. The PDMS septum size and shape is defined by the layout of the
SU-8 anchor. Anchor thickness is determined by the outer diameter of the needle
and practical limits of fabricating thick film SU-8 structures (including film
uniformity and process time). In this work, the SU-8 structure height was selected to
be 100-200 μm greater than the outer diameter of the needle. For a 33 gauge needle

(OD 203 μm) the SU-8 layer was 300 μm thick, for a 30 gauge needle (OD 305 μm)
the SU-8 layer was 500 μm thick.

3.2.1.1 Proof-of-Concept

The idea of a horizontal, or in-plane, interconnect was rapidly prototyped using a
simple setup to demonstrate proof-of-concept. Two chambers connected by a
channel were cut into the center of a 500 μm PDMS membrane using a fine tipped
blade. The membrane was then sandwiched between two glass slides to create an
enclosed system (Figure 3-2).

Figure 3-2 Exploded view of the setup used to demonstrate horizontal interconnect proof-of-concept.

Binder clips were used to apply constant pressure on the membrane. Needles were
inserted from the edge of the membrane to access the chambers. Dyed DI water was
injected into one chamber using a syringe. The liquid was observed to flow from
174

one chamber to the other via the channel exit the system through the second needle
(Figure 3-3).

Figure 3-3 Images showing fluid progression in proof-of-concept setup.

3.2.1.2 SU-8 Anchors

The SU-8 anchors mechanically interlock with the PDMS septum preventing septum
damage or shifting during needle insertion and removal. Three anchor shapes
(square, circular, and barbed) were fabricated and examined (Figure 3-4). The length
of the septum also determines the contact length between the needle and PDMS.
Thus, there exists a design trade-off between interconnect strength and footprint.

175

Figure 3-4 A) Image of an assembled circle septum interconnect. B) Top view of three different
septum connector shapes (circle, barbed, and square) that were designed and integrated into the test
microfluidic system. Needle, PDMS septum, SU-8 housing, and microchannel in the designs are
indicated. C) Side view of the needle piercing the PDMS septum. Image is not drawn to scale.

The SU-8 anchor was fabricated on a glass slide substrate. For devices in which
electrical traces were required, Parylene C was deposited between the glass substrate
and the SU-8 to provide electrical isolation, but was also found to alleviate the
thermal stress-induced delamination of the SU-8 (Despont, et al. 1997). SU-8
delamination was observed in 100% of the setups without the Parylene C layer. The
difference in the coefficient of thermal expansion of the glass slide (8.9x10
-6
/°C)
compared to SU-8 (5.2 x 10
-5
/°C) can result in delamination of SU-8 structures
during processing. Parylene C has a high percent elongation (200%), which allows
176

177
the Parylene C to absorb the stresses due to the thermal mismatch between the glass
and SU-8 (Lo and Meng, 2008)..

3.2.1.3 Septum

The SU-8 layer includes the microchannel which is terminated on either end by a
region reserved for the PDMS septum (Sylgard 184, Dow Corning, Midland, MI). A
PDMS septum is located at both the input and output and allows rapid access to the
sealed microchannel (Figure 3-1, Figure 3-4). Mechanical interlocking structures
patterned in the SU-8 layer secure the PDMS septa to the substrate during both
needle insertion and removal.

3.2.1.4 Interconnect Integration

Interconnects were fabricated at the inlet and outlet of a single SU-8 microchannel.
The interconnect was also integrated into a simple microfluidic system that contained
a microchannel, microchambers, electrolysis pump, and resistive sensors without the
need for any additional fabrication masks (Figure 3-5).

Figure 3-5 Integrated interconnect with microfluidic system. System contains SU-8 layer which
defines the septum housing, microchambers, and microchannel. Electrolysis structure and flow
sensors are fabricated with a Ti/Pt metal layer. PDMS septum and glass cover plate are not present.

3.2.2 System Fabrication

3.2.2.1 Test Interconnect

Microfluidic systems each containing one channel and integrated interconnects at
both the inlet and outlet were batch-fabricated using conventional micromachining
techniques. Test chips with only the SU-8 anchor, SU-8 microchannels, and PDMS
septum were fabricated without metal components for the interconnect test devices.
First, the substrate, either a 76 mm (3 inch) soda lime wafer (Silicon Quest
International, Santa Clara, CA) or soda lime slide (75 mm x 50 mm, Corning Glass
Works, Corning, NY), cleaned and coated with Parylene C (Specialty Coating
Systems, Inc., Indianapolis, IN). The Parylene C was vapor deposited (2 μm thick)
to help prevent the SU-8 from delaminating from the glass surface (Figure 3-6B).

Next, a 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) was obtained
by using a two-step spin coating process to provide better thickness uniformity
across the substrate. The first coat was spun at 1.5 krpm (approximately 200 μm
178

179
thick) followed by a second planarization coating spun at 3 krpm (for an additional
100 μm) (Figure 3-6C). After each spin coat, the applied SU-8 layer is left at room
temperature for 3 hours to improve planarization. Then, the layers were each
softbaked at 90 °C; the first layer softbaked for 90 minutes and the second for 3
hours. Baking steps were all performed on a programmable hotplate (Dataplate
Series 730, Barnstead International, Debuque, IA) set to ramp at 3 °C/min. The
lower softbake temperature was selected to avoid thermal degradation of the
underlying Parylene C. Substrates were allowed to slowly cool to room temperature
after each bake step to avoid thermal stress cracks in the SU-8. The SU-8 was
patterned (600 mJ/cm
2
), post-exposure baked for 30 minutes at 90 °C, and then
developed using SU-8 developer (MicroChem Corp., Newton, MA) (Figure 3-6D).
A final hardbake step was performed at 90 °C for 30 minutes. A list of the
fabrication steps for the interconnect test structure can be found Appendix JJ.

The dehydration step (90 °C for 30 minutes) prior to SU-8 coating is lower than the
normal 120 °C in order to prevent the thermal degradation of the Parylene C.
Parylene C is rated to survive for over 10 years at 100 °C in an air environment.
However, Parylene C will thermally degrade in air at 125 °C (Systems). It has been
observed that some hotplates overshoot the temperature while ramping to the desired
set temperature. Therefore, a lower temperature is chosen to ensure that the Parylene
C is not exposed to excessive heat.

Figure 3-6 Cross-sectional fabrication steps for the test interconnect. Cross-section is taken through
the microchamber.

3.2.2.2 Integrated System

A similar process as the one described in Section 3.2.2.1-Test Interconnect was used
to fabricate an interconnect system that included electrolysis and resistive thermal
sensors. After substrate was cleaned prior to Parylene C deposition, the substrate
was spin coated with AZ 4400 photoresist (AZ Electronic Materials, Branchburg,
NJ) (4 krpm, 40 s, 4 μm) (Figure 3-7A). After exposure, a liftoff mask was
produced. Ti/Pt (200 Å/2000 Å) (International Advanced Materials, Spring Valley,
NY) was e-beam evaporated and defined using standard liftoff processes by
removing the photoresist layer in acetone, isopropyl alcohol, and deionized water
180

181
(Figure 3-7B). The resistive thermal sensors and electrolysis structures were formed
during these steps. Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
was vapor deposited (2 μm) to electrically isolate the metal traces and to prevent the
SU-8 from delaminating from the glass (Figure 3-7C). Photoresist, AZ 4400, was
spin coated (4 μm, 4 krpm, 40 s), patterned, and used as an etch mask for oxygen
plasma removal of Parylene C by reactive ion etching to reveal the contact pads
(Figure 3-7D). The process to spin coat and develop the SU-8 is the same as for the
test interconnect structure (Figure 3-7E). The entire processes to fabricate the
integrated system can be found in Appendix KK. Finally, the septum is filled with
PDMS and the microsystem is capped; the steps are described in Section 3.2.2.3.

Figure 3-7 Simplified fabrication process for the microfluidic chip with integrated interconnect.
Cross section views are through the PDMS septum and microchamber with interdigitated electrodes
for an electrolysis pump.

The masks used to fabricate both the 300 μm and 500 μm channel interconnect can
be found in Appendix MM. Only the SU-8 mask was needed to fabricate the test
interconnect system discussed in Section 3.2.2.1-Test Interconnect (Figure 4-35).

182

183
3.2.2.3 Septum Formation

The septa on either end of the microchannel were formed after completing the SU-8
fabrication process the SU-8 anchors were filled with PDMS (Figure 3-7G). First,
the regions adjacent to the septum (microchambers and microchannel) were filled
with deionized water. Water serves as a convenient mask that prevents PDMS
prepolymer from flowing into water-masked areas. PDMS is oil-based and is thus
immiscible in water, allowing water to serve as a mold for casting PDMS (Chao, et
al. 2007). PDMS also has a low surface energy and its hydrophobic property further
enhances the water masking technique. A similar water masking technique has been
previously demonstrated (Li, et al. 2007, Li, et al. 2008). PDMS was mixed in a
10:1 elastomer-to-curing agent ratio (AR-250 Hybrid Mixer, Thinky Corp., Tokyo,
Japan) and then precisely introduced into the septum area through a 20 gauge needle
and syringe. PDMS prepolymer was degassed in a vacuum chamber to remove
trapped bubbles and then partially-cured (65 °C for 20 minutes) in order to facilitate
the final adhesion of the capping layer.

To complete the microfluidic system, the top of the microchannel was formed by
capping the SU-8 microchannel and PDMS septum with a soda lime glass slide (75
mm x 50 mm x 1 mm, Corning Glass Works, Corning, NY) (Figure 3-7H). First, the
glass slide was diced to obtain 10 mm x 50 mm pieces to match the die size. To join
the substrate and cover, PDMS was used as an adhesive sealant. A thin, partially-
cured PDMS membrane (~300-400 μm thick, 65 °C for 20 minutes) was placed on

top of the SU-8 layer to prevent excess PDMS from entering the microchannel
below. PDMS prepolymer was spread on top of the partially-cured PDMS
membrane as an adhesive layer and the glass cover was placed on top of the
microchannel assembly. This prepolymer layer further planarizes any remaining
thickness non-uniformity in the SU-8 layer. The entire structure was left at room
temperature for 24 hours to allow the PDMS to cure completely or rapidly cured at
70 °C for 1 hour. Finally, a 33 gauge (O.D. 203 μm) stainless steel non-coring
syringe needle (Hamilton Company, Reno, NV) was manually inserted into the
interconnect (Figure 3-8).

Figure 3-8 Edge view of the needle insertion location. The 33 gauge non-coring needle pierces the
PDMS septum through the edge of the system, creating an in-plane connection.

3.2.3 Benchtop Experiments- Methods and Results

To optimize the performance of this new microfluidic interconnect method, needle
tip types were assessed and compared. The interconnect performance was evaluated
184

185
using a needle with the most suitable tip type. The robustness and reusability of the
needle-septum connection was evaluated by measuring the pull-out forces associated
with needle removal. Leakage was examined in both short and long term studies
under pressurized operation.

3.2.3.1 Coring vs. Non-coring Needle Tip Type

3.2.3.1.1 Methods

Access to a microchannel was gained by piercing a needle through the PDMS
septum such that the lumen of the needle established a continuous fluidic path to the
microchannel. Stainless steel coring and non-coring needles (Hamilton Company,
Reno, NV) (30 gauge) were examined to determine the needle tip type that could
maximize the lifetime of the interconnect (Lo, et al. 2006). Needle puncture marks
of both coring and non-coring needles through PDMS membranes were compared.

3.2.3.1.2 Results

The results of this study were similar to the conclusions presented in Section
2.2.4.3.1.2- Results. The non-coring needle was determined to be best suited to
access the microchamber via the septum.

186
3.2.3.2 FEM Analysis of Stress Distribution

3.2.3.2.1 Methods

Finite element analysis and modeling estimated the stress on the septum as a non-
coring needle was inserted. The circle septum was chosen as the model for its
simple design, and therefore, the most likely design to be implemented. The model
was used to determine stress dissipation within the septum and model if any stress
reaches the SU-8 housing.

The stress distribution within the septum was modeled as needles touched the surface
of the septum and after the needle fully penetrated the septum. In the model, the
surface area of the septa which would normally be in contact with the septa housing
(e.g. SU-8 anchor housing, device substrate, and device cap) was constrained to
represent the packaging. Though the PDMS septum may be able to debond from the
SU-8 or Parylene C coated surfaces, this dynamic interaction is difficult to model;
therefore, the contacting surfaces are fixed. The septum faces not in contact with the
SU-8 housing, substrate, or packing (e.g. faces through which the needle can enter or
exit) were not constrained.

3.2.3.2.2 Results

Finite element modeling shows the stress concentration remained localized around
the needle. A radial stress pattern was observed at the needle tip pre-puncture. Post
insertion, the stress was uniformly distributed along the needle shaft and did not
appear to extend to the SU-8 housing (Figure 3-9).

Figure 3-9 Stress induced on the septum during needle insertion at a) needle pre-
puncture, and b) after the needle has fully pierced the septum.

3.2.3.3 Pull-out Force and Reusability

3.2.3.3.1 Theory

A needle inserted through a PDMS septum can be modeled as a stiff fiber embedded
in a soft matrix. Total pull-out force, the force required to completely remove the
needle from the PDMS septum, can be used as a measure of the interface strength
between the needle and the septum. Debonding in this model system can be
achieved by the application of tensile force, torque, or a compressive force, however,
187

only the tension-induced debonding is considered here. Pull-out forces for
fiber/matrix joints subjected to tension have been explained by Gent and Liu using a
modified theory based on the Griffith fracture energy criterion for debonding that
also accounts for the work associated with frictional sliding at the debonded interface
(Gent and Liu 1991). The strength of adhesion is associated with the energy
criterion for fracture and can be expressed as:

Equation 3-1 Adhesion Force
188
2
0
4 F =
2
f
ma
ArEG π

where F
0
is the adhesion force, A is the cross-sectional area of the matrix, r is the
fiber radius, E
m
is the Young’s modulus of the matrix material, and G
a
is the
adhesive fracture energy. The frictional pull-out force, F
f
, is expressed as:

Equation 3-2 Frictional Pull-Out Force
FrpX π μ =
kp

where r is the radius of the fiber, µ is the coefficient of friction, p is the compressive
stress, and X is the contact length between the fiber and the matrix. This result
assumes that the coefficient of friction is constant. If there is a 2% or greater
pressure at the fiber tip compared to the Young’s modulus of the matrix, then the
product of the coefficient of friction and compressive stress can be represented by a
constant value, μ = . Under this condition, the coefficient of friction is

independent of pressure and the theoretical pull-out force due to friction can be
simplified (Equation 3-3).

189
2
f
=
f
FF
Equation 3-3 Frictional Pull-Out Force Independent of Pressure
FrkX π

The total pull-out force, F, is a combination of the frictional and adhesion forces
(Equation 3-4).

Equation 3-4 Total Pull-Out Force
00
2 F rkX F π =+ = +

The equation shows a linear relationship between the pull-out force and the
fiber/matrix contact length. The strength of adhesion G
a
is found from Equation 3-1
by extrapolating the pull-out force F to the case where X = 0 to obtain the intercept
F
0
.

3.2.3.3.2 Methods

The reusability of the interconnects was assessed by determining the pull-out forces
required to remove the needle after multiple insertions into the same septum. Pull-
out force is proportional to the contact area between needle and septa, therefore, can
be used as an indicated for how well the septa seals around the needle for each
subsequent use. For these experiments, a 33 gauge non-coring needle was used and

190
the three septum shapes were each evaluated. The microfluidic system was mounted
such that the microchannel was oriented perpendicular to the ground by using a
custom laser machined test setup (Mini/Helix 800, Epilog, Golden, CO) (Figure
3-10). The needle was inserted through a hole in the base of the test fixture and then
through the PDMS septum to gain access to the microchannel. Pull-out forces were
applied to the needle by gradually increasing the brass weights contained in a bag
attached to the needle. The bag was preloaded with a 50 g brass weight. Weights
were then added incrementally with a 10 minute pause between weight additions to
allow the system to equilibrate. The smallest weight used was 1 g which
corresponds to a force resolution of 9.8 mN. The total weight attached to the needle
was recorded after the needle was pulled out of the septum. The resulting pull-out
force was calculated from the final pull-out weight and recorded. The same
interconnect was used repeatedly in up to 8 trials following the same procedure
described above.

Figure 3-10 Pull-out test setup. Connector is held perpendicular to the ground by placing the
microfluidic device in the Plexiglas test fixture. Weights are added to a container attached to the luer
lock portion of the needle. Pull-out force is determined by multiplying gravity by the combined mass
of the weights, needle, and container. Image is not drawn to scale.

3.2.3.3.3 Results

The pull-out forces for each septum shape were obtained and compared to theoretical
values (Figure 3-11). The theoretical maximum (1.19 N) and minimum (0.86 N)
pull-out force for the interconnects were obtained from Equation 3-4 using the values
in Table 3-2. The theory can only be applied to the first pull-out event and
corresponding pull-out force; the equation does not account for damage or effects of
191

multiple pull-outs and therefore is not used for comparison in subsequent pull-out
events.

Table 3-2 Values used for theoretical pull-out force calculation.
Variable Value
E
m
360 kPa - 870 kPa
G
a
180 J/m
2

- 240 J/m
2
r 101.5 μm
μp 0.1 kPa
A 2.1 mm
2

The wide range in theoretical values is attributed to the variation of published values
for the Young’s modulus of PDMS (Armani, et al. 1999, Gent and Liu 1991);
differences in material properties may arise due to varying processing conditions
(e.g. temperature and time for PDMS curing).

Figure 3-11 Pull-out force of the interconnects are compared to the calculated theoretical values.

192

193
As predicted in the simple fiber/matrix model, pull-out force increased with
interconnect contact length and contact area. It is important to note that Gent and
Liu modeled pull-out of stainless steel fibers from PDMS that were embedded into
the material before curing (Gent and Liu 1991). The model may not adequately
account for the increase in compressive force experienced by the needle due to the
displaced material or the irregular path the needle may take. The first pull-out force
value for the square, circular, and barbed interconnects may vary from the model
because of the inability to precisely determine the contact length and area. The
flexible, small diameter needle was manually inserted and the angle and exact
location of needle insertion varied between interconnects. It is possible that some
needles may have been inserted near or at the glass or SU-8 interfaces. Thus, an
incomplete PDMS-to-needle seal would result in which the PDMS would compress
the needle against the hard surface instead of completely surrounding it.
Furthermore, the needle tears the PDMS as it is pushed through the PDMS matrix,
again resulting in uneven compression over the contact area between the needle and
PDMS.

The pull-out force for the first removal with respect to contact length (Figure 3-12)
and contact area (Figure 3-13) for the circle, square, and barbed interconnect was
compared to other pull-out values for published connectors (Table 3-3) (Chiou and
Lee 2004, Li and Chen 2003, Yao, et al. 2000). Deviations from the linear behavior
predicted by Equation 3-4 may be attributed to the differences in operating principle
between each interconnect system and the fiber/matrix materials used.

Figure 3-12 Comparison of the interconnect (circular, square, and barbed) and that of other published
connectors of the first pull-out force with respect to contact length.

Figure 3-13 Comparison of the interconnect (circular, square, and barbed) and that of other published
connectors of the first pull-out force with respect to contact area.

The pull-out force for the different connectors does show a linear relationship
between force and length (Figure 3-12), as predicted in the theoretical equation
194

195
(Equation 3-4). Also, force increases as contact area increases, as expected.
Variations in pull-out force may also be attributed to different setups and fabrication
processes. The interconnect designed by Chiou and Lee used a fused silica capillary
that was placed in PDMS prepolymer and cured in place (Chiou and Lee 2004). This
connector benefited from a perfect match between the PDMS and the capillary,
which may have resulted in elevated pull-out force values. The Li and Chen
connector as well as the Yao et al. connector have a predefined cylinder, which was
smaller than their fiber, in the matrix material (Li and Chen 2003). The fiber was
inserted into the cylindrical hole, where the matrix material compressed along the
entire contact area between the fiber and the matrix. Furthermore, the Yao et al.
matrix was encompassed in a silicon structure, providing further compression force
on the fiber as the matrix was in a limited area (Yao, et al. 2000).

196
Table 3-3 Summary of Connector Parameters for Published Connectors (Chiou and Lee 2004, Li and
Chen 2003, Yao, et al. 2000)
Connector
Diameter
[mm]
Contact
Length
[mm]
Contact
Area
[mm
2
]
Fiber
Material
Matrix
Material
Circle 1 0.203 7.010 4.471
Stainless
Steel
Sylgard 184
Circle 2 0.203 9.400 5.995
Stainless
Steel
Sylgard 184
Square 1 0.203 7.240 4.617
Stainless
Steel
Sylgard 184
Square 2 0.203 10.540 6.722
Stainless
Steel
Sylgard 184
Barbed 1 0.203 8.030 5.121
Stainless
Steel
Sylgard 184
Barbed 2 0.203 8.960 5.714
Stainless
Steel
Sylgard 184
Chiou 1 0.361 14.990 17.000
Fused
Silica
Sylgard 184
Chiou 2 0.361 9.960 11.296
Fused
Silica
Sylgard 184
Chiou 3 0.361 6.260 7.100
Fused
Silica
Sylgard 184
Yao 1 0.500 0.250 0.393 Glass
MRTV1
American
Safety
Technologies,
Inc.
Yao 2 0.400 0.250 0.314 Glass
MRTV1
American
Safety
Technologies,
Inc.
Li 1 1.020 3.000 9.613
Teflon
(PTFE)
Sylgard 184
Li 2 0.840 3.000 7.917 Glass Sylgard 184

The pull-out force data for multiple trials (up to 8 pull-outs) were compared to the
performance of other published microfluidic interconnects (Figure 3-14) (Chiou and
Lee 2004, Li and Chen 2003, Yao, et al. 2000). The measured pull-out force was
consistent over 8 pull-outs though, in some cases, exhibited a slight decrease after
the first pull-out. Again, pull-out force was dependent on the contact area between
the interconnect fiber and matrix. The pull-out force was normalized with respect to
contact area in order to better compare the pull-out force of the existing interconnects
(Figure 3-15). Most of the normalized pull-out forces fell within the range of 0.1 to
0.5 N/mm
2
.

Figure 3-14 Comparison of the pull-out force for our interconnects (circle, square, barbed) compared
to other published connectors. Pull-out force varies over subsequent pull-outs and is dependent on
contact area.

197

Figure 3-15 Comparison of normalized pull-out force with respect to contact area.

Each of the mechanical anchoring schemes for the interconnects (square, circular,
barbed) prevented the septum from dislodging from the SU-8 housing. Though all of
the interconnect shapes survived multiple uses, the circular shaped anchor may be
the most desirable structure for two reasons. First, the circular shape does not
contain as many corners that may allow for stress concentration in the SU-8 layer.
Secondly, the smooth, circular septa did not contain sharp corners and were easier to
fill with PDMS. In the square and barbed septum cavities, the PDMS prepolymer
required manual spreading to ensure complete filling of the cavity corners.

Device robustness can be increased by increasing the contact area between the
septum and needle. Therefore, a trade-off between available space and device
198

199
integrity can be made to tailor the single interconnect to each microfluidic
application.

3.2.3.4 Maximum Operating Pressure

3.2.3.4.1 Methods

The operating range of each of the three interconnect designs was determined for
both pressurized dyed water and nitrogen gas. Dyed water was used to help visualize
fluid flow through the device. Non-coring needles were inserted through the input
and output septa. Dyed water was introduced from the input, through the
microchannel, and to the output needle to ensure a continuous fluidic pathway
between the input and output. The output needle was blocked (P-656 luer lock
assembly and P-770 plug, Upchurch Scientific, Oak Harbor, WA) and the input
needle was attached to a custom pressure testing system (Figure 3-16).

Figure 3-16 Test setup for leakage pressure test and prolonged pressure test using pressurized water.
Output needle is blocked using an Upchurch plug.

For both pressurized water and pressurized nitrogen gas tests, the applied pressure
was slowly increased with a 10 minute pause between pressure increments to allow
the system to equilibrate. In the case of pressurized water, leakage was determined
when using visual confirmation of device failure. When using pressurized nitrogen
gas, the assembly was submerged in deionized water and the escape of nitrogen
bubbles was used to determine the point of leakage failure (Figure 3-17). The
pressure at which the interconnect started to leak and the location of the leakage
were recorded.

200

Figure 3-17 Test setup for leakage pressure test using pressurized N2. Leakage is visualized by N2
bubbles escaping from the submerged the microfluidic chip.

3.2.3.4.2 Results

The leakage test identified the maximum operating pressure as well as the failure
mode in each case (Table 3-4). The maximum operating pressure was 51 kPa for
water operation (barbed interconnect) and 60.4 kPa for N
2
operation (circle
interconnect).

201

202
Table 3-4 Summary of Leakage Pressure Results
Port
Type
Test
Medium
Leakage Pressure
[kPa]
Leakage Location
Circle Water 36.27
Between glass and
Parylene C interface
Square Water 20.82
Between glass and
Parylene C interface
Barbed Water 51.37 Needle insertion at output
Circle N
2
60.4
Between glass and
Parylene C interface
Square N
2
25.72 Needle insertion at output
Barbed N
2
15.72
Between glass and
Parylene C interface

Two failure modes were observed: (1) the needle insertion point and (2) at the glass-
Parylene C interface. The first failure mode suggests the presence of a leakage path
along the needle insertion path or a compromise of the seal between the needle and
the PDMS septum (Figure 3-18). The second failure mode is the result of poor
adhesion between the glass and Parylene C. The failure modes are further discussed
in Section 3.2.3.6- Failure Modes.

Figure 3-18 Interconnect failure at the PDMS septum and stainless steel needle interface. A) Water
surrounds the needle shaft as PDMS is debonded from the needle and B) seeps from the needle
insertion point.

3.2.3.5 Prolonged Pressure Operation

3.2.3.5.1 Methods

Long term operation of the interconnects under pressurized water was also evaluated.
The same custom pressure testing system was used (Figure 3-16). An interconnect
was pressurized at 36 kPa and monitored over a 24 hour period. This pressure value
is the average of the maximum operating pressures using pressurized water.

3.2.3.5.2 Results

The prolonged pressure test demonstrated that the connector can maintain
pressurized water at a pressure of 36.2 kPa for over 24 hours. No measurable
203

204
leakage was observed over this period of time. These interconnects can therefore be
used in devices that have applications with long operation times.

3.2.3.6 Failure Modes

3.2.3.6.1 Delamination

Needle insertion angle through the input and output septa was observed to be a
critical factor in interconnect failure. When the needle was inserted at a downward
angle relative to the glass substrate, the tip of the needle would penetrate into the
Parylene C layer and could potentially become lodged at the glass/Parylene C
interface. If the needle tip did not become embedded under the Parylene C, the
Parylene C layer is still damaged when the tip penetrated the layer, resulting in a
possible leakage exit path. When a test medium, such as dyed deionized water, was
introduced through the needle, failure due to delamination was observed when the
fluids seeped under the interconnect structure (Figure 3-19). Subsequent fabrication
runs were modified to include an adhesion promotion step with silane-based A-174
prior to Parylene C deposition to prevent Parylene C delamination from the glass
substrate. In these interconnects, if the needle was inserted at an angle, the needle tip
would scratch the Parylene C but the film would not readily delaminate from the
glass surface.

Figure 3-19 Interconnect failure due to Parylene C delamination. Dyed water can be seen spreading
between the Parylene C and substrate layers.

3.2.3.6.2 Needle Misalignment

Misalignment of the needle may occur as the needle penetrates the PDMS septum at
an angle parallel to the plane of the substrate. In this case, the needle trajectory
would cause the needle tip to collide with the surrounding SU-8 housing and thus
block the lumen of the needle. The needle could be removed and reinserted;
however, it was demonstrated that repeated insertions were associated with
decreasing needle pull-out forces. Thus, careful alignment of the needle to the
microchamber is necessary to avoid degradation of interconnect strength related to
multiple misaligned needle insertions.

Another failure mode occurred when the needle was inserted at an upward angle
relative to the glass substrate. The needle could become lodged in the PDMS/glass
interface where its circumference was not completely sealed by PDMS. When the
needle was inserted immediately adjacent to the glass cap, the needle was not
205

206
surrounded by PDMS. The small gaps between the needle shaft and the displaced
PDMS form low resistance leakage paths (Figure 3-18).

A future refinement to this interconnect would be to include a needle insertion guide,
either external or integrated, to prevent failure modes resulting from the needle
penetrating through the PDMS septum boundaries and into interfaces. The needle
guide would direct the needle to the center of the PDMS septum and prevent the
needle from being inserted at an angle. For high density microfluidic connections,
multiple interconnects can be positioned in a linear array at the edges of a
microfluidic chip. This interconnect format would be analogous to ribbon cables and
sockets on jumper pins commonly found in the microelectronics industry.

3.2.4 Summary

Practical in-plane integrated interconnects that can be batch fabricated with
microfluidic systems have been demonstrated. Needle pull-out force was modeled as
debonding in which both frictional and adhesion forces were accounted. In this
model, pull-out force scales with needle/septum contact length. This relationship
was confirmed experimentally and by comparing interconnect results against other
published interconnects. The reusability of the interconnects was demonstrated in
multiple pull-out trials in which the pull-out force remained constant over 8 trials. In
experiments where the interconnects were subjected to incrementally increasing gas
or water pressure, the Parylene C-to-glass interface adhesion was found to be critical

207
in preventing leakage. Also needle insertion angle and alignment affected the
robustness of this interconnect scheme. No leakage was observed under prolonged
exposure to pressurized water.

As demonstrated with theoretical equations, empirical data, and comparison with
existing connection schemes, device robustness is proportional to contact area
between the septum and needle. The single interconnect design can be scaled to
increase or decrease the contact area as needed to fit specific operating pressure
criteria.

3.3 Multiple Interconnects

The first version, or single interconnect version, of the horizontal interconnect, like
many published interconnects, allows only one connection to be made at a time. The
second version of the interconnect design allows many connections to be made
simultaneously (arrayed interconnect version); decreasing the footprint of each
connector will result in a higher density of connectors within the same space.

3.3.1 Design

The arrayed interconnect uses the pin-in-socket approach, similar to that of the single
interconnect (Lo and Meng 2008). The high-density design allows multiple needles

to simultaneously access corresponding microfluidic channels via their integrated
septa. The septa were fabricated by filling an SU-8 septa housing with PDMS
prepolymer than allowing it to cure. An array of commercially available non-coring
needles was inserted into the device by piercing the septa to establish a fluidic
connection (Figure 3-20).

Figure 3-20 Illustration of the a) side view of the arrayed SU-8 and Parylene C microchannel, b) top
view of the needle array and septa, and c) image of a connected arrayed interconnect.

3.3.1.1 Septa Design

As in the single interconnect design, PDMS was chosen as the septa material due to
its compliant and resealing properties. In particular, PDMS septa will seal around
the non-coring syringe needles following insertion and removal to avoid formation of
any leakage paths. Several versions of the arrayed interconnect were fabricated to
demonstrate the versatility of the technique. Differing septa shapes (oval, oval
208

209
overlap, rectangular) were fabricated on soda lime substrates; the septa shape
determined the septa spacing (2.54 mm or 1 mm center-to-center spacing).
Additional features such as 1) merged septa for higher-density interconnects, 2)
sideports to incorporate a second input stream in a single channel, 3) needle guides to
align the needle to the septum centers, and 4) SU-8 or Parylene C microchannels
(Figure 3-21). Electrolysis pumps (interdigitated electrodes) were also integrated
into microchannels. Modular interconnects are possible (many different combination
of the arrayed interconnect) simply by choosing between different septa shapes,
septa spacing, merged septa, sideports, needle guides, microchannel material, and
metal structures. Additionally, the arrayed interconnect can turn existing
microfluidic systems into modular components by provide simply and standard
connections which allow entire systems to be connected in series or in parallel.

Figure 3-21 Schematic indicating key features of our interconnect technology. Here, interconnects
with surface micromachined Parylene C channels are shown. Needle guides to help align the needle
arrays to the septa. Additional features which can be added to the arrayed interconnect include
sideports and interdigitated electrodes for electrolysis or electrochemical sensing.

3.3.1.1.1 Septa Shape

Oval, overlapped, and rectangular septa housing shapes were chosen for their ability
to retain PDMS without additional anchoring features (Figure 3-22). We previously
presented the relationship between septa design and pull-out force (Lo and Meng
2008). The theoretical equations predicted and experimental results verified that
pull-out force varies linearly with the contact surface area between the septa and
needle. Septa shape did not affect the pull-out force, however, design choices which
included a septa locking feature (e.g. a shape which prevented the PDMS septa from
being dislodged from the septa housing during needle insertion/ removal) were
preferred. Additionally, shapes which minimized stress concentration locations (e.g.
210

contoured edges instead of corners) and were less time intensive to package were
considered. Therefore, designs which maximized contact length while minimizing
septa width (e.g. footprint) were selected.

Figure 3-22 Septa configurations used in the arrayed interconnect designs.

The square septum of the single interconnect have sharp corners. These locations
were found to increase stress concentration in the SU-8 anchor during insertion and
was a potential failure locations. Removing corners results in a more robust SU-8
layer. Oval septa also are easier to fabricate compared with structures that contain
sharp corners. When PDMS was dropped into the circular structure, the PDMS
completely filled the circular structure without any additional manual manipulation.
PDMS filling of the square and barbed septa required the use of a sharp tool to help
encourage the PDMS to completely fill the septum corners. This manual
manipulation could introduce bubbles into PDMS prepolymer and was a time
consuming process. The corners in the rectangular setpa were rounded to minimize
stress concentration. Furthermore, the overlapping septa allowed multiple septa to
be filled simultaneously, simplifying the packaging process.
211

212
3.3.1.1.2 Septa Spacing

Septa spacing, and consequently, the density of the connections were determined by
the dimensions of the microfluidic microchannels and integrated components, as well
as the septum shape. Two septa spacing arrangements were considered. A 2.54 mm
center-to-center spacing design utilized individual oval septa. This spacing was
chosen to emulate the standard spacing found in electrical pin packages. Denser
interconnects with 1 mm center-to-center spacing were fabricated using overlapping
oval septa and connected rectangular septa. Sideports, as well as converging
microchannel designs, that allow multiple fluids to be introduced into a single input
or channel, respectively, were also investigated.

The oval septa design has 2.54 mm center-to-center spacing between adjacent septa.
Each septum is a distinct unit and therefore must be filled individually. Overlapped
and rectangular septa feature denser interconnect packing (1 mm spacing) and
simultaneous filling of multiple septa as they are fluidically connected. The
thickness of the SU-8 septa housing was determined by the practical fabrication
limits for thick planar SU-8 layers (time and process complexity) and the needle
outer diameter. For 33G needles (203 μm OD), a 300 μm layer of SU-8 was used.

3.3.1.2 Needle Guides

The angle of needle insertion is important for proper alignment and to prevent
interconnect failure. Needle misalignment may cause the needle to veer off-center
during insertion and become lodged against the SU-8 anchor walls. Misalignment in
our previous interconnects resulted in blockage of the needle lumen and fluid path
(Lo and Meng 2008). Thus, this improved arrayed interconnect incorporates needle
alignment structures in the SU-8 housing. The needle guide, which guides the
trajectory of the needle, is included in the arrayed interconnect design. The needle
guide is incorporated into the SU-8 anchor and starts as a large opening that tapers to
a smaller opening (Figure 3-23). A needle inserted through the needle guide will be
aligned to pierce the center of the septum.

Figure 3-23 Illustration of the needle guides designed to align the trajectory of the needle for the
arrayed interconnect design.

213

214
3.3.1.3 Side Ports

Side access ports to the microchamber are placed in the arrayed interconnect to
increase the number of possible applications. A second input can be fed through the
side port to mix two inputs. The side port can be placed in-line with the existing
interconnect inputs so that the connections can be made in parallel for both the non-
overlapping and overlapping septa (Figure 3-24A, B). Parallel sideports have several
advantages. Needles used to access the main interconnect ports as well as the
sideports can be inserted simultaneously. Furthermore, because the sideports are
adjacent to the main interconnect septa the SU-8 piece can be an single rectangular
structure that contains all of the ports and septa.

Figure 3-24 Parallel sideport structures integrated into arrayed interconnect designs that have A)
individual septum, and B) combined septa.

This work demonstrates the functionality of parallel sideports, however, a second
design that places the side-port perpendicular could be fabricated to extend the
sideport functionality. Perpendicular sideport applications include interrogation of
the liquid using sensing elements and introduction/ removal of samples along
different points in the flow path (Figure 3-25). To provide spatial control of sample
introduction, several perpendicular sideports can be placed along a microchannel.
Different fluids can be fed through, or removed from these locations.

215

To interrogate the fluid within the microsystem, a needle can be used to pierce the
septum into the microchamber, a sensor (e.g. fiber optic cable treated with a
fluorescent agent to detect the presence of a molecule) can be fed through the needle
shaft. The needle can then be removed leaving the sensor behind; the PDMS will
seal around the sensor to prevent leakage. While this design may be less robust and
require additional time to connect the sideports compared to its parallel design
counterpart, perpendicular access provides additional sensing application
possibilities.

Figure 3-25 Illustration of the perpendicular sideports in the arrayed interconnect design.

216

217
3.3.1.4 Microchannels

The microchannels in the arrayed interconnect design were fabricated using SU-8 or
Parylene C.

3.3.1.4.1 SU-8 Microchannels

Arrayed interconnects with SU-8 microchannels were fabricated because the SU-8
microchannel was previously tested and optimized in the single interconnect design.
The fabrication process for the SU-8 microchannel is simpler and requires fewer
steps than microchannels using another material for the microchannel (e.g. Parylene
C) because the microchannel, septum, and microchambers can be formed
simultaneously. Because the SU-8 components are fabricated together, the channel
height is by septa requirements on the thickness of the SU-8 layer. Furthermore, the
width of the microchannel was also limited based on the manufacturer’s
recommended aspect ratio for SU-8 structures (10:1 height to width ratio
(MicroChem)). Therefore, for a 300 μm thick SU-8 layer, a minimum channel width
of 30 μm was used.

3.3.1.4.2 Parylene C Microchannels

For applications requiring smaller channel dimensions, Parylene C microchannels
can be fabricated. The channel dimensions and Parylene C thickness must be

carefully selected to prevent the channel from collapsing when the sacrificial
photoresist within the channel is removed (Yao, et al. 2002). The critical width for
microchannels is obtained from the critical length (l
crit
) for preventing the collapse
and stiction of cantilevers. It is governed by the Young’s modulus of the channel
material (E), the material thickness (t), gravity (g), the surface tension between an
air-liquid interface (γ
la
), and the contact angle of the material (θ
c
) (Tas, et al. 1996):

Equation 3-5 Critical width to prevent stiction in cantilevers
32
3
16 cos
la c
Et g
crit
l
γ θ
⎛⎞
⎜⎟
⎝⎠
=

Yao et al. determined a maximum distance of 150 μm between supporting structures
in a 4.5 μm thick Parylene C beam. In a channel, the channel walls serve as the
supporting structure, therefore, a channel width of 100 μm was chosen, eliminating
the need for support posts within the channel (Yao, et al. 2002).

The fabrication process for the Parylene C microchannel interconnects is described
in Section 3.3.2.2- Parylene C Microchannel. Briefly, to fabricate the microchannel,
a 2μm thick layer of Parylene C is vapor deposited. A 4μm layer of photoresist is
patterned to the shape of the microchannel interior. Support posts at the
microchannel opening are defined. A 4μm layer of Parylene C is vapor deposited.
A 4μm layer negates the need for support posts within the microchannel because the
layer has enough structural integrity to prevent stiction (Yao, et al. 2002). Another
218

layer of photoresist is spun and patterned to define the microchannel opening. The 4
μm of Parylene C covering the microchannel opening is etched using reactive ion
etching (RIE). Finally, the sacrificial photoresist is removed using isopropyl alcohol
as acetone affects the structural integrity of SU-8.

3.3.1.4.3 Converging Microchannels

Converging microchannel designs were included to demonstrate another interconnect
application. Parallel connections can be made and multiple streams of liquid are
joined within the microchannel. Both the rectangular and oval septum designs with
converging microchannels are fabricated (Figure 3-26).

Figure 3-26 Converging microchannel designs of A) 4 rectangular, B) 8 rectangular, and C) 4 oval
overlapping septum designs.

219

220
3.3.1.5 Metal Structures

3.3.1.5.1 Electrolysis

The electrolysis structure designed in the arrayed interconnect is similar to the single
interconnect. Metal electrolysis structures were only integrated into the oval septa
designs in the microchamber and microchannel as a demonstration of arrayed
interconnect integration with metal structures. Metal structures could also have been
added to the rectangular septa design. The electrolysis structure also has
interdigitated leads, but is shaped to fit within the microchamber border. The
electrolysis structure found within the Parylene C microchannel interconnects is
smaller in order to fit both the microchannel opening and the electrolysis structure
within the microchamber. The electrolysis structure cannot function if it is covered
by the Parylene C used to create the microchannel, therefore, additional Parylene C
etch masks as necessary to remove the Parylene C coating the metal structures.

3.3.1.6 Arrayed Interconnect Permeations

Microfluidic devices with integrated arrayed interconnects with different septa
shapes (oval, oval overlap, or rectangular), number of septa (4 or 8), septa spacing
(2.54 mm or 1 mm), microchannel material (SU-8 or Parylene), and optional
sideports were designed and fabricated. Some devices also incorporated converging
channels and metal structures to demonstrate the versatility of this interconnect

technique. A representative arrayed interconnect with Parylene C microchannels,
oval overlap septa with 8 inputs and outputs, sideports, and needle guides is shown
in Figure 3-27.

Figure 3-27 Fabricated arrayed interconnect with Parylene C microchannels and sideports. Salient
features of the arrayed microfluidic system with integrated interconnects are highlighted. External
access via needles is not shown in these photographs.

Table 3-5 summarizes the interconnect combinations which were fabricated. Images
of these interconnects can be found in Table 3-7 (SU-8 microchannel, no metal),
Table 3-8 (SU-8 microchannel with metal), Table 3-9 (Parylene C microchannel, no
metal), and Table 3-10 (Parylene C microchannel with metal).
221

222
Table 3-5 Summary of fabricated arrayed interconnect combinations.
Channel
Type
# Septa Septa Type
Converging
Channels
Sideports Metal
SU-8 4 Oval N N N
SU-8 4 Oval N Y N
SU-8 4 Oval N N Y
SU-8 4 Oval N Y Y
SU-8 4 Oval Overlap N Y N
SU-8 4 Oval Overlap N N N
SU-8 4 Oval Overlap Y N N
SU-8 4 Rectangular Y N N
SU-8 4 Rectangular N N N
SU-8 8 Oval N N N
SU-8 8 Oval N N Y
SU-8 8 Oval Overlap N Y N
SU-8 8 Oval Overlap N N N
SU-8 8 Rectangular Y N N
SU-8 8 Rectangular N N N
Parylene C 4 Oval N N Y
Parylene C 4 Oval N Y Y
Parylene C 4 Oval Overlap N Y N
Parylene C 4 Oval Overlap N N N
Parylene C 8 Oval N N Y
Parylene C 8 Oval Overlap N Y N
Parylene C 8 Oval Overlap N N N

Wafer level illustrations of the fabricated interconnects can be found in Appendix
NN.

A summary of the design aspects for the arrayed interconnect with SU-8
microchannels or Parylene C microchannels can be found in Table 3-6.

223
Table 3-6 Summary of dimensions and design specifications for the SU-8 microchannel and Parylene
C microchannel arrayed interconnect.
Design SU-8 Microchannel Parylene C Microchannel
Needle
# of needles inserted
simultaneously
4,8 4,8
Spacing between needles (non-
overlapping septa)
2.54mm 2.54mm
Spacing between overlapping
septa
1mm 1mm
SU-8 Layer

SU-8 height- 30G Needle
500 µm 500 µm
SU-8 height- 33G Needle
300 µm 300 µm
Microchannel

Microchannel material
SU-8 Parylene C
Opening (ID) of microchannel
N/A 100 µm
Opening (OD) of microchannel
N/A 150 µm
Posts inside funnel
N/A/ 20 µm
Parylene Thickness Top of
Channel
N/A 4 µm
Parylene Thickness
(1st Layer)
2 µm (metal isolation + SU-8
adhesion layer)
2 µm (metal isolation, SU-8
adhesion, bottom of
microchannel)
Parylene Thickness
(2nd layer)
N/A 4 µm (top of microchannel)
Channel Width 300 µm, 500 µm 50 µm, 100 µm
SU-8 Needle Guide
Beginning Width 1 mm 1 mm
Ending Width (30G needle) 500 µm 500 µm
Ending Width (33G needle) 300 µm 300 µm
Length 2 mm 2 mm
Height Depends on SU-8 height Depends on SU-8 height
Septa
Oval Septa Length
(non-overlap)
5 mm 5 mm
Oval Septa Width
(non-overlap)
2 mm 2 mm
Oval Septa Length
(overlap)
5 mm 5 mm
Oval Septa
Overlap Portion
1 mm 1 mm
Rectangular Septa Length 5 mm 5 mm
Rectangular Septa Width 500 µm 500 µm
Metal Layer
Resistors Included Included
Pumping Electrolysis Electrolysis
Temp Sensing Included Included

224
Table 3-6 Summary of dimensions and design specifications for the SU-8 microchannel and Parylene
C microchannel arrayed interconnect. (cont.)
Design SU-8 Microchannel Parylene C Microchannel
Additional Features
Side Ports Included Included
Connecting Microchannels Included Included
Integration with other
materials
N/A SU-8 and Parylene

3.3.2 Fabrication

The fabrication processes for both the SU-8 and Parylene C microchannel
interconnects are discussed below. The SU-8 interconnect has a simpler fabrication
process than the Parylene C interconnects. The mask files for SU-8 microchannel
interconnects, SU-8 microchannel interconnects with metal structures, Parylene C
microchannel interconnects, and Parylene C interconnects with metal structures can
be found in Appendix OO, Appendix PP, Appendix QQ, and Appendix RR,
respectively.

The fabrication of the SU-8 microchannel is similar to the fabrication of the single
interconnect device. The fabrication steps for the arrayed interconnect structures
which contain either the SU-8 microchannel or the Parylene C microchannel are
presented below.

225
3.3.2.1 SU-8 Microchannel

3.3.2.1.1 Arrayed Interconnect, SU-8 Microchannel without Metal

To fabricate the arrayed interconnect designs that have a SU-8 microchannel and no
metal features, first obtain a 3 inch (76 mm) soda lime substrate (Silicon Quest
International, Santa Clara, CA) (Figure 3-28A). Treat the wafer with A-174, a
Parylene C adhesion promoter. Next, vapor deposit 2 μm of Parylene C (Specialty
Coating Systems, Inc., Indianapolis, IN) on one side of the wafer (Figure 3-28B).
The Parylene C layer provides mechanical support to prevent the SU-8 from
delaminating from the glass surface. The SU-8 layer (300 μm) is fabricated using a
two-step spin-coating method using SU-8 2100 (MicroChem Corp., Newton, MA)
(Figure 3-28C). The first coat results in a 200 μm (1.5 krpm, 30 sec). The layer is
then left at room temperature for 3 hours to improve planarization. The first layer is
then softbaked at 90 °C (ramped at 3 °C/ min from room temperature to 90 °C) for
90 minutes. The layer was then allowed to slowly cool to room temperature in order
to prevent the formation of thermal stress cracks in the SU-8. Next, the planarization
layer of SU-8 was applied (100 μm, 3krpm, 30 sec). Again, the layer was left at
room temperature and then softbake for 3 hours using the same softbake and ramping
scheme was used for the first layer. Dicing saw tape was applied to the underside of
the wafer to prevent any undesired UV reflection from the aligner chuck. The SU-8
was then patterned (600 mJ/cm
2
) and post-exposure baked at 90 °C (3°C/ min ramp)
for 30 minutes. The SU-8 layer was developed using SU-8 Developer (MicroChem

226
Corp., Newton, MA); developing requires approximately 30 minutes (Figure
3-28D). The structure is then hardbaked for 30 minutes at 90 °C (3 °C/ min ramp)
and allowed to slowly cooled to room temperature. The wafer is then diced to
separate the various structures. The septa are filled and the setup is capped with a
glass slide, these procedures were previously described in Section 3.2.2.3- Septum
Formation (Figure 3-28E, F). The masks used to fabricate these structures are
presented in Appendix OO. A detailed fabrication process for the arrayed
interconnect structures with SU-8 microchannels can be found in Appendix SS.

Images of the fabricated arrayed interconnects with SU-8 microchannels without
metal structures can be found in Table 3-7.

227

Figure 3-28 Fabrication steps for the SU-8 wafer that contains interconnect designs that do not
require metal structures. The cross-section is taken horizontally along the microchannel. The SU-8 is
lighter in color at step D because after SU-8 patterning, no SU-8 exists along the cross-section line.
However, the lighter SU-8 represents the SU-8 material remaining surrounding the needle guide and
microchannel in order to better visualize the process flow after step D. This process flow is used for
designs shown in Figure 4-37A.

Table 3-7 Summary of the SU-8 microchannel arrayed interconnects which were fabricated. These interconnects do not have any metal structures.
Channel Type,
# Septa,
Septa Type

Converging
Channels
Sideports Metal Image
SU-8
4
Oval
N N N

SU-8
4
Oval
N Y N

SU-8
4
Oval Overlap
N N N

SU-8
4
Oval Overlap
N Y N

10 mm

228

Table 3-7 Summary of the SU-8 microchannel arrayed interconnects which were fabricated. These interconnects do not have any metal structures. (cont)
Channel Type,
# Septa,
Septa Type

Converging
Channels
Sideports Metal Image
SU-8
4
Oval Overlap
Y N N

SU-8
4
Rectangular
N N N

SU-8
4
Rectangular
Y N N

SU-8
8
Oval
N N N

10 mm

229

230
l

Ta of the SU-8 microchannel arrayed interconnects which were fabricated. These interconnects do not have any metal structures. (cont) ble 3-7 Summary
Channel Type,
# Septa,
Septa Type

Converging
Channels
Sideports Meta Image
SU-8
8
Oval Overlap
N N N

SU-8
8
Oval Overlap
N Y N

SU-8
8
Rectangular
N N N

SU-8
8
Rectangular
Y N N

10 mm

231
3.3.2.1.2 Arrayed Interconnect, SU-8 Microchannel with Metal

The fabrication for the arrayed interconnects which have SU-8 microchannels and
metal structures are similar to the process described in Section 3.3.2.1.1. The
additional steps for these structures are as follows. Prior to the Parylene C
deposition, the wafer is spin-coated with AZ 4400 photoresist (AZ Electronic
Materials, Branchburg, NJ) (4 krpm, 40 s, 4 μm) (Figure 3-29B). The photoresist is
patterned and developed to create a lift-off layer for the metal (Figure 3-29C). Ti/Pt
(200 Å/2000 Å) (International Advanced Materials, Spring Valley, NY) is
evaporated onto the wafer surface using an electron-beam metal deposition system
(Figure 3-29D). The underlying photoresist is dissolved using acetone and lifts-off
portions of the metal layer (Figure 3-29E). The wafer is cleaned using oxygen
plasma (100W, 100mT, 5 minutes) to remove any remaining photoresist residue.
The wafer is treated with A-174, a Parylene C adhesion promoter and then coated
with 2 μm of Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN) to
provide mechanical support of the SU-8 and to electrically isolate the metal
structures (Figure 3-29F). The wafer is spin-coated with AZ 4400 photoresist (4 μm,
4 krpm, 30 sec) (Figure 3-29G). The photoresist is patterned and developed to create
a Parylene C etch mask. The Parylene C coating the electrode pads and electrolysis
structures is removed using oxygen plasma (Figure 3-29H). The photoresist is
stripped from the wafer using acetone, isopropyl alcohol, and DI water (Figure
3-29I). The steps to fabricate the SU-8 layer, dice the wafer, form the septum, and

232
cap the microsystem are the same as those in Section 3.3.2.1.1- Arrayed
Interconnect, SU-8 Microchannel without Metal (Figure 3-29J-M). The masks used
for this fabrication process can be found in Appendix PP. A complete fabrication
process for these structures can be found in Appendix TT.

Images of the fabricated arrayed interconnects with SU-8 microchannels and metal
structures can be found in Table 3-8.

233

Figure 3-29 Fabrication steps for the SU-8 wafer that contains interconnect designs that do not
require metal structures. The cross-section is taken horizontally along the microchannel. The SU-8 is
lighter in color at step K because after SU-8 patterning, no SU-8 exists along the cross-section line.
However, the lighter SU-8 represents the SU-8 material remaining surrounding the needle guide and
microchannel in order to better visualize the process flow after step D. This process flow is used for
designs shown in Figure 4-37B.

Table 3-8 Summary the SU-8 microchannel arrayed interconnects with metal, which were fabricated.
Channel Type,
# Septa,
Septa Type

Converging
Channels
SideportsMetal Image
SU-8
4
Oval
N N Y

SU-8
4
Oval
N Y Y

10 mm

234

235

Table 3-8 Summary the SU-8 microchannel arrayed interconnects with metal, which were fabricated. (cont).
Channel Type,
# Septa,
Septa Type

Converging
Channels
SideportsMetal Image
SU-8
8
Oval
N N Y

10 mm

236
3.3.2.2 Parylene C Microchannel

The fabrication of the arrayed interconnect structures that contain a Parylene C
microchannel is more complicated process than its SU-8 microchannel counterpart
because two layers of Parylene C are required to create the microchannel. The
fabrication steps for the Parylene C interconnect designs are described below.

3.3.2.2.1 Arrayed Interconnect, Parylene C Microchannel without
Metal

The arrayed interconnect which has a Parylene C microchannel is fabricated on a 3
inch (76 mm) soda lime wafer (Silicon Quest International, Santa Clara, CA)
(Figure 3-30A). The wafer is treated with a Parylene C adhesion promoter, A-174.
Next, 2 μm of Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN) is
vapor deposited on one side of the wafer (Figure 3-30B). Photoresist, AZ 4400 was
spin-coated (4 μm, 4krpm, 40 sec) and patterned to define the interior shape of the
Parylene C microchannel (Figure 3-30C,D). A second layer of Parylene C (4 μm) is
vapor deposited over the photoresist and initial Parylene C layer (Figure 3-30E).
The two layers of Parylene C are then annealed together under vacuum at 160 °C
(Figure 3-30F). Parylene C will thermally degrade at temperatures above 100 °C in
air, however, it can withstand higher temperatures under vacuum (Systems). This
step is option and can be used if delamination between the two Parylene C layers is
observed.

237

Next, a layer of photoresist, AZ 4400 (AZ Electronic Materials, Branchburg, NJ) is
spin-coated (4 μm, 4 krpm, 40 s) (Figure 3-30G). The photoresist is patterned in
order to create a Parylene C etch mask (Figure 3-30H). The Parylene C coating the
channel opening is etched using oxygen plasma (Figure 3-30I). The photoresist is
then stripped using an acetone, isopropyl alcohol, and DI water rinse (Figure 3-30J).

The SU-8 layer is applied, exposed, and developed using the two-step process
described in Section 3.2.2.1- Test Interconnect (Figure 3-30K,L). The individual
arrayed interconnect structures are diced apart using a dicing saw. The structures are
placed in an isopropyl alcohol bath (at room temperature) for 21 days to remove the
sacrificial photoresist inside the channel (Figure 3-30M). The septa are filled and the
entire structure is capped using the process described in Section 3.2.2.3- Septum
Formation (Figure 3-30N,O). The masks used to fabricate the arrayed interconnect,
Parylene C microchannel setups can be found in Appendix QQ. A detailed
fabrication process is presented in Appendix UU.

Images of the fabricated arrayed interconnects with Parylene C microchannels
without metal structures can be found in Table 3-9.

238

Figure 3-30 Process flow for Parylene C microchannel arrayed interconnect designs. The cross-
section line is taken along the microchannel. Translucent SU-8 represents SU-8 which surrounds a
component but not within the cross-sectional line. The translucent SU-8 is included to aid in
illustrating the fabrication process. This fabrication process is used for designs shown in Figure
4-38A.

Table 3-9 Summary of the Parylene C microchannel arrayed interconnects which were fabricated. These interconnects do not have any metal structures.
Channel Type,
# Septa,
Septa Type

Converging
Channels
SideportsMetal Image
Parylene C
4
Oval Overlap
N Y N

Parylene C
4
Oval Overlap
N N N

Parylene C
8
Oval Overlap
N Y N

Parylene C
8
Oval Overlap
N N N

10 mm

239

240
3.3.2.2.2 Arrayed Interconnect, Parylene C with Metal

The fabrication process for the arrayed interconnect with Parylene C microchannels
and integrated metal components is the most complex of all of the setups. Additional
steps are needed to create the metal components and etch the Parylene C covering
the metal.

The fabrication starts with a soda lime substrate (76 mm or 3 inch wafer, Silicon
Quest International, Santa Clara, CA) (Figure 3-32A). 2 μm of Parylene C
(Specialty Coating Systems, Inc., Indianapolis, IN) is vapor deposited on one side of
the wafer (Figure 3-32B). Photoresist, AZ 4400 (AZ Electronic Materials,
Branchburg, NJ), is spin coated (4 krpm, 40 s, 4 μm) on the Parylene C layer (Figure
3-32C). The photoresist is exposed and patterned to form a Parylene C etch mask
(Figure 3-32D). The Parylene C is etched using oxygen plasma in order to access
the substrate and deposit the metal electrodes (Figure 3-32E). The electrodes must
be formed on the substrate in order to create robust connections that can withstand
electrical connections. The photoresist is then stripped from the surface using an
acetone, isopropyl alcohol, DI water rinse (Figure 3-32F).

Another photoresist layer, AZ 4400, is spin-coated (4 μm, 4 krpm, 40 s), exposed,
and developed to create the metal lift-off layer (Figure 3-32G). Ti/Pt (200 Å/3000
Å) (International Advanced Materials, Spring Valley, NY) is evaporated using an

241
electron-beam deposition system (Figure 3-32H). The photoresist lift-off layer is
then removed using acetone to remove the unnecessary metal (Figure 3-32I). Any
acetone or photoresist residue is cleaned using oxygen plasma. Current (0.3 mA) is
applied to the electrolysis structures to verify functionality (e.g. determine if the
electrodes are shorted, determine if electrodes will delaminate, visually verify bubble
formation).

AZ 4400 is spin coated (4 μm, 4 krpm, 40 s) and pattern to define the interior of
Parylene C microchannel (Figure 3-32J-K). The second layer of Parylene C (4 μm)
is vapor deposited (Figure 3-32L). An optional step to anneal the two layers of
Parylene C together can be completed under vacuum at 160 °C for 48 hours (Figure
3-32M).

AZ 4400 photoresist (AZ Electronic Materials, Branchburg, NJ) is spin-coated (4
μm, 4 krpm, 40 s), exposed and patterned in order to create a Parylene C etch mask
(Figure 3-32M,N). The Parylene C coating the channel opening, electrolysis
structure, and electrode pads is etched using oxygen plasma (Figure 3-32O). The
photoresist is then stripped using an acetone, isopropyl alcohol, and DI water rinse
(Figure 3-32P). Again, electrolysis functionality is verified to determine if any
Parylene C coating remains on the metal structures.

Again, the SU-8 layer is applied, exposed, and developed using the two-step process
previous described in Section 3.2.2.1- Test Interconnect (Figure 3-32Q,R). All of the

electrolysis structures are tested for a final time prior to packaging to verify their
functionality (Figure 3-31).

Figure 3-31 Time-lapsed images of the working electrolysis structures prior to packaging. 0.3 mA of
current was applied to the electrodes. Bubble formation was visually confirmed.

242
The arrayed interconnect structures are diced apart using a dicing saw and then
placed in an isopropyl alcohol bath (at room temperature) for 21 days to remove the
sacrificial photoresist inside the channel (Figure 3-32S). The septa are filled and the
entire structure is capped using the process described in Section 3.2.2.3- Septum
Formation (Figure 3-32T,U). The masks used to fabricate the arrayed interconnect,
Parylene C microchannel setups can be found in Appendix RR. A detailed list of the

243
steps needed to fabricate arrayed interconnects with Parylene C microchannels and
metal components can be found in Appendix VV.

Images of the fabricated arrayed interconnects with Parylene C microchannel and
metal structures can be found in Table 3-10.

Figure 3-32 Process flow for Parylene C microchannel arrayed interconnect designs. The cross-
section line is taken along the microchannel. Translucent SU-8 represents SU-8 which surrounds a
component but not within the cross-sectional line. The translucent SU-8 is included to aid in
illustrating the fabrication process. This fabrication process is used for designs shown in Figure 4
38B.

244

245
Figure 3-32 Process flow for Parylene C microchannel arrayed interconnect designs. The cross-
section line is taken along the microchannel. Translucent SU-8 represents SU-8 which surrounds a
component but not within the cross-sectional line. The translucent SU-8 is included to aid in
illustrating the fabrication process. This fabrication process is used for designs shown in Figure 4
38B. (cont)

Table 3-10 Summary of the Parylene C microchannel arrayed interconnects with metal structures, which were fabricated.
Channel Type,
# Septa,
Septa Type

Converging
Channels
SideportsMetal Image
Parylene C
4
Oval
N N Y

Parylene C
4
Oval
N Y Y

10 mm

246

Table 3-10 Summary of the Parylene C microchannel arrayed interconnects with metal structures, which were fabricated. (cont)
Channel Type,
# Septa,
Septa Type

Converging
Channels
SideportsMetal Image
Parylene C
8
Oval
N N Y

10 mm

247

248
A first layer of Parylene C is deposited prior to the metal deposition layer because it
simplifies the Parylene C etching step. If the metal was deposited prior to the first
Parylene C coating, the Parylene C etching used to expose the microchannel
opening, electrolysis opening, and electrode openings would be through different
thicknesses of Parylene C. The microchannel opening would be covered with 4μm
of Parylene C while the electrolysis and electrode covered with 6μm. Controlling
the etch to expose all of the metal without damaging the microchannel would be
difficult, therefore the deposition of the metal layer is moved in the fabrication
process. However, the electrodes must be in contact with the substrate so that the
electrical probes will not damage the electrodes. If the electrodes were placed on
Parylene C, the probes would be able to scratch or punch through the metal layer
because the underlying Parylene C layer is soft.

Isopropyl alcohol is used to remove the sacrificial photoresist within the
microchannel because acetone has been observed to cause SU-8 delamination of
initial test structures. Additionally, acetone has been previously demonstrated to be
an effective method of removing SU-8 residue but causes delamination of SU-8
when the acetone rinse is not carefully controlled (Agarwal, et al. 2005). Removing
the sacrificial photoresist from within the microchannel requires several days of
soaking in isopropyl alcohol; it is important to ensure that the isopropyl alcohol bath
does not completely evaporate and leave isopropyl alcohol residue within the
microchannel. The residue may block the microchannel and prevent it from
functioning.

3.3.2.3 Needle Array

The needles were housed in channels (0.0135’’, 342.9 μm) spaced 2.54 mm or 1 mm
apart, center-to-center, to match the septa spacing. Two different types of needle
arrays were fabricated: shared input (Figure 3-33a-c) and separated inputs (Figure
3-33d-e).

Figure 3-33 Needle arrays which provide shared or separate input capabilities to needles and thus
microchannels. a) 4 shared 4 needles, 1 mm spacing, b) 4 shared needles, 2.54 mm spacing, c) 8
shared needles, 1 mm spacing, d) 4 separate needles, 1 mm spacing, and e) 8 separate needles, 2.54
mm spacing. The scale bar represents 10 mm.

3.3.2.3.1 Shared Input Needle Arrays

The shared input needle array was fabricated by drilling channels (0.0135’’ O.D.,
80G drill bit) partially through a Plexiglas block. A larger diameter channel (0.04’’
249

250
O.D., 60G drill bit) was drilled through the side of the block that intersected all of
the smaller channels. A 10-32 threaded hole allowed connection of a liquid or gas
source to the needle array with conventional fittings. Commercially available 33G
non-coring needles (21033A Point Style 2, Hamilton Company, Reno, NV) were
carefully placed in each of the 80G holes and affixed using epoxy. In this
configuration, pressurized media was applied to all of the needles simultaneously;
however, needle arrays with separate access to each needle were also fabricated.

3.3.2.3.2 Separate Input Needle Array

A custom-made Plexiglas mold was used to fabricate a needle array with individual
needle access. Individual needles will be encased in liquid plastic casting resin with
the distance between needles (2.54 mm or 0.1’’ center-to-center and 1mm or 0.04’’
center-to-center) the same distance as between the non-overlapping and overlapping
interconnects (center-to-center), respectively. Silicone tubes were attached to each
needle, providing separate input into each needle.

The needle array is made by using a custom-made jig to align the needles and
provide the mold for the surrounding plastic resin (Figure 3-34). The jig is
comprised of four levels. The bottom level serves as the base for the plastic resin
mold and has an etched outline of the holes cut from the middle two layers. The two
middle layers align the needles. The middle layers have a rectangular hole cut from
the jig to allow the plastic resin to coat around the needles. Each middle piece has

251
grooves cut into a surface to designate the needle locations. The needle extends from
one edge of the piece to 18.6 mm (0.732 inches) into the piece. The length of the
groove and placement of the holes results in the beveled end of the needle extending
10 mm (0.394 inches) from the resin block on one side, and a 5 mm (0.197 inches)
long needle base extending from the other side of the resin block. The needles are
aligned by sliding the needle along the groove until its progress is stopped when it
reaches the end of the groove. A thinner rectangular hole is cut from the layer to
prevent the plastic resin from wicking along the needle/groove interface and
clogging the needle tip. The top layer to the jig contains an outline of the holes
found in the middle layers as well as two holes. Once the four jig layers and needles
are assembled, liquid plastic resin is poured into one hole with the other hole serving
as an outlet for the displaced air. The entire jig is held together using four standard
10-32 screws located at the corners of the jig. When the resin is cured, the jig can be
disassembled to remove the molded needle array. The jig can then be reused to
make more needle arrays.

8 s
Figure 3-34 Custom-made laser machined molds for creating an array of needles (4 or 8). All layers
are made of acrylic and are color coded to illustrate assembly.
252

253
3.3.3 Experimental Methods and Results

3.3.3.1 FEM Analysis of Stress Distribution

3.3.3.1.1 Methods

Finite element analysis and modeling estimated the stress on the septa as the needles
were inserted. The oval overlap septa were chosen as the model because this septa
design was the simplest design to fabricate and implement, and therefore, the most
likely candidate for future use. Additionally, the septa were combined in one
complete piece; any stress, induced during insertion, which may affect neighboring
septa/needle pairs, will be visible in this design. The stress distribution in the PDMS
septa set a practical limit to the achievable needle density by dictating the minimum
spacing between the needle path and SU-8 housing. A suggested design rule is a
minimum distance between the needle shaft and the SU-8 housing equaling twice the
length required to dissipate 36.8% of the maximum stress value.

The stress distribution within the septa was modeled at three needle insertion points:
1) as needles touched the surface of the septa, 2) after the needles pierced the septa
and are partially inserted, and 3) after needles were fully inserted. In the model, the
surface area of the septa which would normally be in contact with the septa housing
(e.g. SU-8 anchor, device substrate, and device cap) was fixed to represent the

254
packaging. The bonding strength between the PDMS and SU-8 or Parylene C coated
substrate does not preclude the PDMS from delaminating from either surface,
however, debonding between the PDMS and SU-8 or substrate boundaries is a
dynamic interaction which is difficult to approximate in the model. The septa faces
not in contact with the SU-8 housing, substrate, or packing (e.g. faces through which
the needle can enter or exit) were not constrained. The insertion force at the needle
tip and friction force along the needle shaft were applied using force data obtained
experimentally with a Bose 3100 ElectroForce mechanical fatigue test instrument
(results are presented in Section 3.3.3.3.3- Results). The modeling results were also
compared to real-time photoelastic stress images.

3.3.3.1.2 Results

Finite element modeling of both a single septum and multiple septa designs showed
the stress concentration remained localized around the needle. A radial stress pattern
was observed at the needle tip pre-puncture, and was asymmetric during insertion
due to the beveled tip of the needle. Post insertion, the stress was uniformly
distributed along the needle shaft and did not appear to affect neighboring septa
(Figure 3-36). The maximum stress was observed at the needle/septa interface;
stress decayed exponentially with respect to distance from the needle. The distance
over which the stress decayed to 1.35 x 10
5
N/mm
2
(36.8% of the maximum stress,
3.63 x 10
5
N/mm
2
) was approximately 60 μm. Measurements were taken at the
center of the septa Figure 3-35.

Figure 3-35 Centerline of FEM analysis of needle insertion induced stress in arrayed
interconnect.

The FEM results showed the stress distribution for ideal conditions where: 1) the
needles are all aligned within the septa, 2) the needles do not deviate from a straight
path, 3) the needles maintain a uniform distance from one another, 4) all of the
needles are the same length, 5) all of the needle tips (e.g. the bevels) are orientated in
the same direction, and 6) the needle insertion force is only directed parallel to the
needle shaft. The resultant stress distribution identified the minimum spatial
distance necessary for the interconnect design to prevent unwanted stress
interference on the SU-8 septa housing or adjacent septum/needle pairs.

255

Figure 3-36 FEM images of stress distribution within septa during needle insertion, a) pre-puncture,
b) partial puncture, and c) complete insertion.

The arrayed septa and needle design allows for multiple simultaneous macro-to-
micro connections to be established in a microfluidic device. PDMS is an excellent
choice as a septa material. The compliant nature of PDMS is beneficial for three
reasons, 1) it allows the septa to deform around the needle, providing a reliable and
robust seal for multiple use; 2) the displaced septa material during insertion returns
to the original state after the needle are removed, sealing access to the device
interior; and 3) allows a high density of needles to be inserted simultaneously
without insertion stresses interfering with neighboring needles.

FEM results showed needle insertion induced stress within the septa dissipates
exponentially with respect to distance from the needle shaft. The stress decayed to
36.8% of the maximum value within 60 μm. To prevent the stress from extending to
the SU-8 anchor, a minimum distance between the needle shaft and SU-8 housing of
120 μm is recommended. The rectangular septa design, the one having the smallest
spacing between the needle and SU-8 anchor, had a septum width of 500 μm.
Insertion of a 33G needle with an outer diameter of 203 μm left approximately 150
μm of PDMS between the needle and SU-8 anchor components. Denser rectangular
256

septa are possible; however, additional space may be prudent to account for any
deviations in the needle path (i.e. needle misalignment) from the septa center.

3.3.3.2 Photoelastic Stress

3.3.3.2.1 Methods

PDMS is a photoelastic material; stresses in the PDMS during needle array insertion
and removal can be visualized using polarized light. The principle stresses ( σ
x
, σ
y
)
can be visually observed as a phase difference ( Δ)  that is dependant on the
wavelength ( λ), the material stress-optical coefficient (C), and material thickness (h)
(Timoshenko and Goodier 1970).

Equation 3-6 Phase difference due to stress
257
) (
2
y x
C
h
σ σ
λ
π
− = Δ

PDMS slabs were pierced using a single needle and needle array to visualize stresses
in the material. The sample was placed between two polarizing plates and
illuminated with a broadband light source positioned below the stack. The polarizing
plates were rotated to an orientation which provided the largest contrast in stressed
versus non-stressed areas (Figure 3-37). Low stress areas appear as a white haze
where higher stress areas exhibit a rainbow effect. Photoelastic images were taken

during initial needle puncture and partial needle insertion into the PDMS sample to
compare with FEM results.

Figure 3-37 Experimental setup to visualize photoelastic stress.

3.3.3.2.2 Results

In order to observe the photoelastic stress, larger needles (single needle: 18G, needle
array: 27G) were necessary to enhance the visible stress during needle insertion.
Photoelastic stress using 33G needles was not visible with the current imaging setup.
From the FEM analysis, the visible stress within the PDMS would extend
approximately 100 μm from the needle; the recording camera could not resolve this
distance, therefore, larger diameter needles were necessary.

258

Stress concentrations were visible along the needle shaft and as a plume at the needle
tip during insertion. Photoelastic visualization of insertion of a 2.54 mm spaced
needle array demonstrated that the stress distribution from one needle shaft did not
overlap with that of a neighboring needle (Figure 3-38). This suggests stresses from
each needle/septum pair do not affect the stress of the neighboring pairs. These
results are similar to the FEM prediction for stress distribution (Figure 3-38);
however, a direct comparison cannot be made due to the difference in needle size
and variation in needle orientation in hand made needle arrays.

Figure 3-38 Photoelasic stress in PDMS from needle insertion for a single (18G) and needle array
(four 27G). Yellow arrows indicate areas of stress.

259

3.3.3.3 Insertion Test

3.3.3.3.1 Theory

The force required to pierce a PDMS sample can be modeled using equations which
calculate the force required for a needle to puncture a piece of tissue. The total
insertion force can be seen as a combination of forces that occur before (stiffness
force) and after (frictional force and cutting force) the needle punctures the material
(Equation 3-7) (Simone and Okamura 2002).

260
Axial Insertion Force Stiffness Force Frictional Force Cutting Force =+ +
Equation 3-7 Insertion Force of a Needle into a Membrane

Stiffness force, or the force the material exerts on the material as the material
deforms, is present after the needle touches the material and prior to the needle tip
entering the material. The stiffness force disappears when the needle punctures the
material (Equation 3-8, Figure 3-39). Note, z
2
is less than z
3
because the membrane
relaxes after it has been pierced. The stiffness force can be modeled as a non-linear
spring, however, it has been shown that a second-order polynomial may provide a
better fit to the data and a lower root mean square (rms) value (Simone and Okamura
2002).

Equation 3-8- Stiffness Force of a Membrane Deflecting
261
1
2
3
stiffness
z positio
z posit
z posit
z positi
⎧
⎪
⎨
⎪
⎩
=
=
=
=
1
12 1 2
3
0
0
zz
fazazzzz
zz
nof needletip
ionof membrane prior to needletouch
ionof membraneimmediately before needle pierces membrane
onof membrane after membrane hasbeen
<
=+ ≤≤
>
punctured

Figure 3-39 Illustration of membrane behavior during pre-puncture, puncture, and post-puncture
stages of needle insertion.

The frictional force is the force exerted on the needle as it passes through the
material. Frictional force is a combination of material adhesion and damping
between the material and the needle shaft. It can be modeled using a modified
Karnopp Model (

Equation 3-9) (Karnopp 1985, Richard, et al. 1999).

Equation 3-9- Frictional Force of a Needle Through a Membrane
2
2
2
2
sgn( )
max( , ) 0
min( , ) 0
sgn( )
,&
,&
v
nn
v
na
v
pa
v
pp
Czbz z
DF z
DF z
Czbz z
negative positivevalues of dynamic friction
gative positivevalues of damping coefficient
nega
−Δ
−Δ
Δ
Δ
+ ≤
<≤
<<
+ >
&& &
&
&
&& &
,
friction
np
np
np
f
C C
b b ne
D D
⎧
⎪
⎪
=
⎨
⎪
⎪
⎩
=
=
=
2
v
a
z relativ
va
F sum
Δ
=
±=
&
&
&
tive positivevalues of static friction
evelocity between needle material
luebelow whichvelocity is considered tobe zero
of non frictional forces applied tothe system =−

The final force is the cutting force (Equation 3-10) (Simone and Okamura 2002).
This is the force required for the needle tip to slice through the material. Ideally,
cutting force is constant as the needle progresses through the material and is
independent of needle depth into the material.

Equation 3-10- Cutting Force of a Needle Piercing a Membrane
262

263
23
cutting
tip
2
3
,
0
,
,,
tip p
tip p
stiffness
nzt t
nzt t
R
ositionof needletip
cation of membrane as defined for f
eofpuncture
≤<
>≥
p
f
R constant
n p
z z lo
t time
ttim
⎧
=
⎨
⎩
=
=
=
=
=

3.3.3.3.2 Methods

The force required to insert and remove a single needle or needle array into a PDMS
sample (thickness 2 mm ± 0.1 mm) was measured using a Bose 3100 ElectroForce
mechanical fatigue test instrument and custom-made laser-cut Plexiglas jigs.
Measuring the stiffness, frictional, and cutting forces can be empirically obtained and
compared to the theoretical values described in the above equations. The load-cell in
the Bose machine can provide real-time measurements of the force exerted on a
PDMS sample during the different stages of the needle insertion can isolate the
stiffness and frictional forces. The cutting force is calculated by subtracting the
frictional force from the total force measured after membrane puncture.

The insertion force to insert 1, 4 and 8 needles into the septum will be measured
using a Bose 3100 ELF machine. A custom-made jig is used to measure the
insertion force for multiple needles (Figure 3-40). The jig holds a PDMS membrane

in place as the multiple needles are inserted into the membrane. For simplicity, a
membrane is used instead of 4 or 8 individual PDMS pieces. Individual pieces are a
more true-to-life representation of the arrayed interconnect system, however, the
membrane is supported such that the stress on the membrane from each needle is
independent of its neighbor. The drawing used to create the jigs can be found in
Appendix WW.

Forces from inserting single needles and needle arrays will be measured for multiple
insertions through the same location in the sample. Results using a single needle will
be compared to the needle arrays to determine how the insertion force varies with
respect to the number of needles. The number of needles, needle tip, needle
diameter, and insertion rates will be varied to determine their impact on insertion and
removal force.

Figure 3-40 Custom-made laser-machined jigs used to measure insertion force through PDMS using a
A) 4 or B) 8 needle assembly.

264

The jigs held the sample PDMS piece between two Plexiglas plates. The plates had
2.54 mm spaced through holes (1, 4 or 8 holes) which guided the corresponding
needles arrangements through the PDMS sample piece. The jig was attached to a
load cell at the base of the Bose instrument; needles were lowered at a constant rate
and the force required to puncture the PDMS sample (insertion force) and to remove
the needles from the sample (pull-out force) were recorded.

Insertion force can be described as a combination of pre-puncture and post-puncture
forces. The pre-puncture force, stiffness force (f
stiffness
), is the force due to material
deformation. Post-puncture force is a combination of the force required to push a
needle through the PDMS material (f
cutting
) and the friction force (f
friction
) between the
needle shaft and the PDMS(Abolhassani, et al. 2007, Okamura, et al. 2004). Thus,
the insertion force is expressed as:

265
insertion
f
out pull
f
−
Equation 3-11 Insertion force equation
friction cutting stiffness
f f f + + =

The force required to remove the needle (pull-out force) is a combination of
frictional and debonding forces (Gent and Liu 1991):

Equation 3-12 Removal force equation
friction debonding
f f + =

266
A discussion of the pull-out force was previously presented for a single interconnect
case(Lo and Meng 2008). In summary, Gent and Liu modeled debonding between a
fiber (i.e. the needle) embedded in a matrix (i.e. the septa) using a modified theory
based on the Griffith fracture energy criterion (Gent and Liu 1991). The debonding
force is a function of the cross-sectional area of the matrix, fiber radius, the Young’s
modulus of the matrix, and the adhesive fracture energy between the matrix and the
fiber. The frictional force varies linearly with respect to the coefficient of friction
between the fiber and matrix, the compressive stress, and the contact area between
the fiber and matrix.

The stiffness force can be determined by examining the resulting force versus
displacement graph generated by the Bose machine. It has been reported that the
force versus displacement curve will start at zero and reach a negative peak
immediately before the initial puncture event. The lowest point corresponds to the
maximum stiffness force.

The frictional force can be determined by inserting the needle completely into the
membrane. The needle is displaced in a sinusoidal pattern, however, the needle tip
must remain on one side of the membrane (Figure 3-41) (O'Leary, et al. 2003,
Simone and Okamura 2002). The measured force is purely the frictional force
between the material and the needle shaft. However, frictional force may change
over multiple punctures, therefore, frictional force is determined by the force
measured after the needle has pierced the material when cutting force is zero.

Figure 3-41 Illustration of needle tip displacement relative to the PDMS membrane for frictional
force measurements.

The cutting force is the difference between the force measured by the Bose machine
after puncture and the frictional force determined by the methods described above.
The cutting force may also be verified by puncturing the membrane multiple times
through the same hole; cutting force is expected to be zero for all insertions
following the first insertion.

As described in the Section 2.2.4.3.2.1- Methods, puncturing a location through the
same location models the worst-case scenario. The measured forces from multiple
puncture events in the same location may be correlated to sample strength. If the
needle passes through the same hole, the cutting force would no longer be present for
punctures after the initial puncture. Therefore, the force measured by the Bose
machine for subsequent puncture events after the first puncture can be attributed to
frictional forces. Each jig will be tested with 4 samples. The sample will be
punctured multiple times with the measured puncture force recorded for each of the
punctures. Forces for each puncture event will be averaged across the four
membranes.

267

268
Membrane puncture forces can also be affected by many variables. These variables
need to be controlled or accounted for in order to compare the results. Parameters
which may affect force include: needle tip shape, needle diameter, needle depth (in
the case of frictional force), axial rotation of the needle, clamping forces on the
membrane as the needle is inserted, and needle insertion velocity (Abolhassani, et al.
2007, O'Leary, et al. 2003, Okamura, et al. 2004, Richard, et al. 1999).

The relationship between insertion force and the number of needles (1, 4, or 8
needles), needle type (C: coring, or NC: non-coring), needle gauge (33G, O.D. 203
μm or 27G, 406 μm), and insertion rate (0.5 or 1 mm/sec) was determined. All
needles were obtained from Hamilton Company (Reno, NV). Additionally, needle
insertion was completed multiple times on a single sample to examine magnitude of
the insertion and removal forces with each subsequent use. The two needle types
used in this experiment are classified as coring (C) or non-coring (NC). The name
describes the needle tip. The coring needle has a blunt tip which cores a cylindrical
section from the material; the non-coring needle has a beveled tip and displaces
material as the needle is pushed through the material. The needle gauge describes
the needle outer and inner diameter; a higher gauge corresponds to a smaller
diameter needle.

The insertion rates were selected based on the limitations of the Bose 3100
ElectroForce instrument. The maximum stroke distance of the displacement motor is
4 mm. A 1 mm/sec rate modeled a practical insertion rate while maintaining a high

269
sample rate resolution from the Bose instrument. Additionally, this insertion rate is
comparable to the achievable manual insertion rate. A second rate, which was half
of the base rate value, was selected for comparison.

3.3.3.3.3 Results

Stiffness, insertion (post-puncture insertion force is a combination of friction and
cutting forces), and removal forces were identified for several needle arrangements
(Table 3-11). Figure 3-42 shows a representative graph of the force measurement for
a needle array insertion during needle array insertion and removal. The stiffness
force was measured by identifying a slight dip in force between the needle touching
the PDMS surface (Figure 3-42a) and full needle puncture (Figure 3-42b) (only
present in single needle experiments and not for needle array graphs because the
needle array tips are not perfectly aligned, therefore, stiffness force cannot be
measured). The post-puncture force (cutting and frictional forces) was measured at
Figure 3-42b. Frictional force was measured at Figure 3-42c; cutting force was
calculated by subtracting the frictional force value (Figure 3-42c) from the post-
puncture force (Figure 3-42b). The maximum pull-out force was measured at point
Figure 3-42e. Previous work demonstrated that puncturing the same location
multiple times decreased sealing ability of the material, compared to random
insertion locations(Lo, et al. 2008). Therefore, needle insertion and removal cycles

were performed at least 9 times at the same location in a PDMS slab to determine
how multiple uses affect the insertion and removal forces.

Figure 3-42 Generic results from insertion force tests. a) needle touches surface of the PDMS
sample, b) material being pierced by needle (combination of stiffness and puncture forces), c) needle
moving through PDMS material (friction force), d) needle stops moving and material relaxes, e)
needle in process of being removed from material (max removal force), f) needle fully removed from
material. Inset: needle displacement over time

The measured forces are summarized in Table 3-11. As predicted from insertion
force equations (Gent and Liu 1991), the forces for the 33G single, four needle array
270

and eight needle array, on average, varied linearly with the number of needles
(Figure 3-43). This result is also in agreement with the results presented by
Okamura et al. where the forces of insertion in soft tissue was presented (Okamura,
et al. 2004). Practically speaking, the maximum number of needles that can be
simultaneously inserted is limited by overall device robustness as related to the
magnitude of force necessary to insert the needle array.

Figure 3-43 Relationship of post-puncture insertion forces (friction and cutting forces) and removal
forces with respect to the number of insertion needles.

Stiffness force increased with needle diameter (i.e. decreasing needle gauge). As
expected, coring needles were associated with larger stiffness force due to the blunt
271

272
profile compared to the beveled profile for non-coring needles. Stiffness force could
not be determined for needle arrays because the needle tips are not aligned.
Therefore, each tip punctured the PDMS sample at slightly different times and a
reliable stiffness force could be determined.

The insertion force, the combination of friction and cutting forces, was also expected
to increase with surface area. As surface area increases, frictional force should
likewise increase. Frictional force in 33 C was larger than 33G NC even though both
have same outer diameter due to the area contributed by the lumen of the coring
needle which was also in contact with the PDMS. Coring needles also had higher
cutting force in general due to their blunt profile compared to the beveled tip of the
non-coring needle.

A comparison of forces from differing insertion rates (1 and 0.5 mm/sec) identified a
slight change in the frictional forces. The lower rate (0.5 mm/sec) had a small
increase in insertion and removal forces; this may be caused by a change of dynamic
frictional force between the needle and PDMS. As expected, the cutting force was
not affected.

In all cases, multiple insertions reduced the insertion and removal forces, and
therefore affected the sealing capability of the reusable arrayed interconnect. Data
on the sealing capability of a PDMS slab that was punctured multiple times in the
same location was previously presented (Figure 2-35, Figure 3-15). As the number

273
of punctures increased, the pressure at which induced leakage through the puncture
site (leakage pressure) decreased. A decrease in sealing ability was attributed to
damage of the PDMS from the repeated insertion and removal damage. However,
the relative change in leakage pressure between each additional puncture and
removal event decreased with each insertion/ removal event; this suggested a
saturation of damage. The pull-out force of the first puncture for the 4 needle array
(33G), normalized with respect to contact surface area between the needle and septa,
was 0.52 to 0.48 N/mm
2
for the tenth removal event. This result was similar to
removal forces of comparable designs of other published reusable connectors, which
range from 0.08 to 0.95 N/mm
2
for the first removal event to 0.02 to 0.22 N/mm
2
for
the tenth removal (Chiou and Lee 2004, Li and Chen 2003, Lo and Meng 2008, Yao,
et al. 2000).

Table 3-11 Summary of relationship between insertion and removal forces and the needle type (coring vs. non-coring), needle gauge (27G or 33G), number
of needles (1, 4, or 8), and rate of insertion (0.5 or 1 mm/sec) (mean ± SE, n = 4).
Needle
Gauge
(O.D. [µm])
Needle
Point Type
# of
Needles
Insertion
Rate
[mm/sec]
# of
Insertions
Stiffness
Force [N]
Insertion
Force [N]
Friction
Force [N]
Cutting
Force [N]
Removal
Force [N]
27G (406) Coring 1 1 1
1.87 ± 0.14 1.67 ± 0.16 1.3 ± 0.11 0.37 ± 0.07 1.4 ± 0.2
33G (203) Non-coring 1 1 1
0.18 ± 0.02 0.87 ± 0.02 0.74 ± 0.02 0.14 ± 0.04 0.6 ± 0.03
33G (203) Coring 1 1 1
0.83 ± 0.03 1.2 ± 0.01 0.83 ± 0.02 0.37 ± 0.02 0.89 ± 0.02
27G (406) Non-coring 4 1 1
N/A 5.41 ± 0.34 5.24 ± 0.49 0.18 ± 0.11 2.53 ± 0.17
33G (203) Non-coring 4 1 1
N/A 3.3 ± 0.05 2.81 ± 0.06 0.49 ± 0.04 2.65 ± 0.07
33G (203) Non-coring 4 0.5 1
N/A 3.48 ± 0.08 3 ± 0.05 0.49 ± 0.05 2.7 ± 0.1
33G (203) Non-coring 8 1 1
N/A 5.72 ± 0.15 4.88 ± 0.16 0.84 ± 0.02 4.6 ± 0.15
27G (406) Coring 1 1 9
N/A 1.25 ± 0.09 1.25 ± 0.09 0 ± 0 1.11 ± 0.02
33G (203) Non-coring 1 1 10
N/A 0.61 ± 0.01 0.61 ± 0.01 0 ± 0 0.54 ± 0.03
33G (203) Coring 1 1 9
N/A 0.66 ± 0.01 0.66 ± 0.01 0 ± 0 0.71 ± 0.02
27G (406) Non-coring 4 1 10
N/A 4.67 ± 0.21 4.67 ± 0.21 0 ± 0 2.64 ± 0.2
33G (203) Non-coring 4 1 10
N/A 2.45 ± 0.08 2.42 ± 0.07 0.03 ± 0.03 2.43 ± 0.09
33G (203) Non-coring 4 0.5 10
N/A 2.71 ± 0.07 2.7 ± 0.07 0.01 ± 0 2.49 ± 0.02
33G (203) Non-coring 8 1 10
N/A 4.42 ± 0.15 4.42 ± 0.13 0 ± 0 4.37 ± 0.12

274

275
3.3.3.4.1.1
As shown in the insertion and removal force tests, force scaled linearly with the
number of needles and decreased for needles with 1) smaller diameters and 2) non-
coring tip. The number of needles in an array can be maximized by using the
smallest diameter non-coring needles possible. However, the needles must be rigid
enough to penetrate the PDMS septa without buckling which may lead to needle
misalignment.

3.3.3.4 Pressure Test

3.3.3.4.1 Maximum Leakage Pressure

Methods

Pressurized DI H
2
O was applied to assembled microfluidic systems to determine
failure pressures and modes. Failure pressures were obtained for all three septa
designs (oval, oval overlap, and rectangular) in the 4 microchannel configuration.
The design with the greatest failure pressure was compared to that of the matching
the 8 microchannel design. Needles were inserted into the input and output septa.
Dyed DI H
2
O was first introduced into the system to ensure the microchannel was
open and free from obstructions. The output needles were removed; due to the self-
sealing nature of PDMS, the pressure gradient required to cause leakage from the
needle track is expected to be much higher than the leakage pressure of the insertion

276
3.3.3.4.1.2
site or failure pressure of the device (Lo, et al. 2008, Lo and Meng 2008).
Pressurized DI H
2
O was applied in increments of 0.5 psi (3.45 kPa) with a 5 minute
hold period to allow system equilibration. The failure pressure was recorded when
water leakage was observed.

Results

The leakage pressure associate with each of the three septa designs were determined
(Table 3-12). Several factors were identified to explain the wide range of failure
pressures for each setpa design. The oval overlap may fail at lower pressures since
septa delaminated from the substrate. During needle array insertion, slight
adjustments to align a single needle in the array induced torque force on the other
adjacent needles. This torque resulted in separation of the PDMS from the Parylene-
coated substrate; the adhesion between these materials was poor. Parylene C was
previously used as a release layer for PDMS because of their low adhesion strength
(Lo, et al. 2008).

The oval design has a distinct septum for each needle; therefore, the torque effect is
minimized in this design. However, leakage pressure was determined by the weakest
needle/septum pair. The data for this design in Table 3-12 were obtained for a single
device. Failure occurred at the interface between the needle and septa (Figure
3-45a). Following failure of one pair (at 2 kPa), the septum was resealed using
additional PDMS and repressurized. The same input location failed again at 26 kPa,

whereas the other three input locations did not fail. Again, needle alignment affects
sealing. If the needle was inserted at a large angle relative to the substrate surface,
then the needle may contact the substrate. In this case, the needle is no longer sealed
around its circumference by PDMS and causes the PDMS to partially delaminate
from the substrate near the contact site. This “tenting” effect creates a potential
leakage path along the needle shaft (Figure 3-44). The failure point at 67 kPa was at
the output septum (Figure 3-45b). Needle alignment for the output septa has the
same concerns as the input septa. However, this result indicated the input septa that
did not fail at 2 and 26 kPa points were well aligned.

Figure 3-44 Illustration of “tenting” effect.

The robustness of the rectangular septa interconnect was also highly dependant on
needle alignment. The SU-8 anchoring posts in this design resulted in a narrow
needle path through the septa area. Slight needle misalignment may result in needle
tips lodging in the SU-8 posts or, with sufficient insertion force, dislodging SU-8
posts from the substrate. The 4 septa rectangular design failed at the device edge.
This failure was not related to the interconnect portion of the device; instead it was
an adhesion problem between the SU-8 and Parylene C or the SU-8 and substrate.
The 4 septa rectangular design had a higher failure pressure than the 8 septa
277

rectangular design because it was much simpler to align and insert 4 needles
compared to 8. The additional force necessary to push 8 needles into the septa
caused the needles within the array to buckle. The buckling resulted in the inserting
force being redirected as torque. The delamination between the septa and the
Parylene C, visible when the pressurized dyed water flows under the septa, was
directly caused by this torque. Additionally, several SU-8 posts were dislodged from
the substrate, creating additional areas where the septa and substrate were not in
contact.

Table 3-12 Summary of failure pressure and failure locations for all septa designs. Arrows indicate
failure points.
Channel
Material
Septa
Type
# of
Septa
Failure
Pressure [kPa]
Failure Location
3.79
Insertion site,
delamination at septa/
substrate interface

SU-8
Oval
Overlap
4
15.86
Delamination at setpa/
substrate interface and
Parylene C /substrate
interface

2.054
Edge input port
(delamination between
septa and substrate)
26.2

Edge input port
(delamination between
septa and substrate)

SU-8 Oval 4
67.57 Output port

SU-8 Rectangular 4 36.54
Delamination between
substrate and Parylene
C interface

SU-8 Rectangular 8 8.96
Insertion site,
delamination at
septa/substrate
interface

278

279
3.3.3.4.2.1
3.3.3.4.2 Failure Modes

Needle Misalignment

Needle misalignment, leading to septa/substrate delamination, was the main cause of
failure (Figure 3-45c,d). Needle guides in the SU-8 housing helped ensure the
needles were inserted in the center of the septa, however, the needle guides could not
prevent the needles from veering within the septa, or for the needle to be inserted at
an angle relative to the substrate surface. Additionally, Okamura et al. demonstrated
needle bending is more prevalent for beveled tip needles (Okamura, et al. 2004).
Beveled needles are necessary minimize insertion and removal forces, therefore,
maximizing the reusability of this connector. Commercial bevel-tipped needles are
readily available. Therefore, other methods to decrease instances or causes of needle
misalignment need to be implemented.

Needle misalignment can be mitigated by improving needle stiffness, however,
trade-offs need to be considered. The needles used to establish the microfluidic
connections were 1 inch in length and the septa were 4 – 4.5 mm in length. The
needle length can be shortened to better match the septa length, however, needle
length determines the maximum length of the septa which is linked to the achievable
needle removal force. Removal force is an indicator of the maximum leakage
pressure the septa can withstand. A smaller needle gauge (i.e. larger diameter) may
be used. Larger needles require greater insertion force and greater septa thicknesses

280
3.3.3.4.2.2
to accommodate the increased diameter. Finally, the needle guides can be altered
(e.g. lengthened) to help establish and maintain the needle position and absorb any
buckling effect, but lengthening the needle guides increases the device footprint.
Septa and needle alignment can be more tightly controlled in precision commercial
manufacturing. Therefore, needle alignment improvements may only be necessary
in prototype and research devices.

Delamination

Septa delamination from the Parylene C or the SU-8 was also failure mode (Figure
3-45a,b). Bonding between the septa to the SU-8 anchor and substrate package can
be improved. First, the Parylene C can be removed using oxygen plasma from the
area within the SU-8 housing prior to adding the SU-8 layer. The glass surface can
be treated with oxygen plasma in order to form irreversible bonds between the glass
and PDMS (Bhattacharya, et al. 2005, Duffy, et al. 1998, McDonald, et al. 2000,
McDonald and Whitesides 2002). The SU-8 sidewalls can also be roughened to
increase the contact surface area.

Figure 3-45 Common failure modes of the arrayed interconnect. a) Leakage through needle insertion
path between needle and septa, b) leakage at output septa through needle track, c) delamination
between the septa and Parylene C coated substrate, d) delamination between the Parylene C coating
and the substrate. Examples of needle misalignment are shown in c) and d).

3.3.3.4.3 Prolonged Pressure

3.3.3.4.3.1 Methods

Prolonged pressure will be applied to a device to ensure the arrayed interconnect can
withstand pressure application for an extended period of time. The applied pressure
will be determined by calculating 50% of the average maximum pressures of all
281

tested interconnects. The pressure will be applied for 24 hours and observed to
ensure no leakage during the entire 24 hours.

282
3.3.3.4.3.2 Results

The oval design survived continuous application of pressure (25 kPa) for over 24
hours without any visible leakage. This result demonstrated the interconnect’s
ability to be used in applications which require survival in extended pressurized
conditions.

3.3.3.5 Electrolysis Pressure Generation

3.3.3.5.1 Theory

Electrolysis reaction was previously presented in Section 2.3.1.1.3- Electrolysis
Pump and Pump Chambe. The application of current to the electrolysis electrodes
generates hydrogen and oxygen gas within the microchamber, thus driving liquid
through the microchannel. The theoretical volume (V
theoretical
) can be calculated
using Equation 3-13.

Equation 3-13 Volume of gas generation during electrolysis
3
4
theoretical theoretical m
i
Vq t Vt
F
⎛⎞
==
⎜⎟
⎝⎠

283
Given the theoretical gas generation rate, q
theoretical
[m
3
/s], duration of applied
current, t [sec], current, i [A], Faraday’s constant, F = 96.49x10
3
[C/mol], and the
molar gas volume at 25 °C and atmospheric pressure V
m
= 24.7x10
-3
[m
3
/mol].

3.3.3.5.2 Methods

Internal pressure can be generated using the electrolysis structures in the microfluidic
device. The water within the device was converted to hydrogen and oxygen gas
when current is applied to the electrolysis electrodes. The internal pressure was
measured using a commercial pressure sensor (ASDX 015D44R, Honeywell
International, Morristown, NJ).

3.3.3.5.3 Results

The functionality of including an electrolysis structure for inducing pressure, or
displacing water within the system was demonstrated. The pressure generated
during electrolysis was measured using a pressure sensor. The electrolysis pressure
that can be generated depends on the electrolysis variables (e.g. current, electrode
design, electrolyte), the internal volume of the device, and volume of air already
present in the device or testing setup.

A representative graph of the pressure change due to gas generation and
recombination is shown in Figure 3-46. A baseline reading of atmospheric pressure

was obtained prior to each test. Current was applied to the device until the entire
device interior was voided of any visible water (approximately 100 sec). This is in
good agreement with the theoretical time (97.4 seconds) to generate enough gas to
fill the entire internal volume of the microfluidic device (Equation 3-13, given
V
theoretical
= 5.61 mm
3
and i = 0.3mA). The generated pressure within the device was
well below the failure pressure of the septa (Table 3-12). Additionally, when the
current was removed from the electrodes, recombination of the hydrogen and oxygen
gases was observed. Complete recombination of generated gas occurred within 1
hour. This suggests that none of the generated gas escaped from the testing setup.

Figure 3-46 Internal pressure change due to electrolysis. Pressure increases when current (0.3 mA) is
applied to the interdigitated electrodes, pressure decreases when current is turned off and the oxygen
and hydrogen gas recombine into water.

284

3.3.3.6 Sideport Functionality

3.3.3.6.1 Methods

The sideport feature on the SU-8 interconnects were demonstrated showing the
combination of two input streams. Sideports allow multiple inputs to be combined
into a single microchannel. One sideport was demonstrated in this arrayed
interconnect design, however, the sideport functionality can be extended to include
multiple sideports for additional input lines, although at the expense of chip real
estate.

To demonstrate the functionality and feasibility of the sideport, a 33G non-coring
needles were inserted through the septa of the sideport and main channel. A 30G
non-coring needle was inserted into the output port to minimize any fluidic
resistance at the output. Syringes containing either deionized water (DI H
2
O) or
dyed DI H
2
O were connected to each input needle. A syringe pump was used to
deliver a constant steady flow to both needles (Figure 3-47).

Figure 3-47 Experimental setup for sideport testing
285

3.3.3.6.2 Results

The functionality of the sideports was verified by introducing clear liquid stream
through the sideport into a dyed liquid stream. Both streams were injected at a rate
of 500 μL/min. Due to laminar flow characteristics in microfluidic devices, distinct
and adjacent streams were observed in the channel (Figure 3-48).

Figure 3-48 Time lapse images of sideport function: a) dyed water introduced in the main septum and
un-dyed water through the sideport, b) close up image of the laminar flow within the microchannel.

The sideport allowed the injection of two distinct fluids into a single microchannel.
However, the sideport feature can be extended to allow multiple inputs into the same
microchannel, thus providing a modular design which uses none, some, or all of the
sideports. Sideports can also be placed perpendicular to the microchannel; these
sideports would allow samples to be drawn from any location along the fluidic
stream, or for the introduction of sensors into the microchannel.

286

287
3.3.3.7 Parylene C Microchannel Functionality

3.3.3.7.1 Methods

An arrayed interconnect with a Parylene C microchannel was placed in isopropyl
alcohol (IPA) to remove the sacrificial photoresist in the Parylene C microchannel.
The interconnect was removed from the IPA bath, the liquid front was observed over
time to determine if the channel inlet and outlet orifices were open. Once the
channel was completely dry, a drop of Rhodamine B dye was then placed at the
channel opening. Channel functionality was verified through observing Rhodamine
B movement in the channel via capillary action.

Next, a device was partially packaged where the channel inlet was packaged, as
described in Section 3.2.2.3, while the outlet was not packaged. The outlet was not
packaged in order to decrease the chances of damaging or clogging the microchannel
during packaging; demonstration of channel functionality is still possible with
partially packaged device. A 33G non-coring needle was inserted into the inlet
septum and Rhodamine B dye was pushed into the device. Again, Rhodamine B
flow was observed in the channel.

288
3.3.3.7.2 Results

The liquid/ air interface within the Parylene C microchannel verified the inlet and
outlet ports of the Parylene C microchannel were open. IPA evaporated from the
channel opening and the air/ liquid front was seen advancing through the
microchannel (Figure 3-49). Additionally, capillary action of Rhodamine B through
the channel demonstrated the channel was open and was unobstructed (Figure 3-50).

Functionality of the partially packaged device was verified (Figure 3-51). However,
because pressure was used to inject the Rhodamine B into the channel, time lapsed
images were not fast enough to capture the movement of the Rhodamine B in the
channel. However, Rhodamine B movement through the outlet port funnel was
captured.

It should be noted that the dissolution of the sacrificial photoresist in IPA takes
several weeks before the channels are clear for photoresist. Dissolution time can be
estimated using Fick’s Law of Diffusion. Dissolution can be accelerated by heating
the IPA bath (~ 50 ºC); excessive heat should not be used because SU-8 is very
thermally sensitive. Undiluted photoresist developer can also be used to remove the
sacrificial photoresist and takes roughly 1 day as opposed to 21 days. No adverse
affects on the microfluidic structure were observed when using the developer.

Figure 3-49 Time-lapsed images of IPA evaporating from within the arrayed interconnect Parylene C microchannel.

289

290
Figure 3-50 Time-lapsed images of Rhodamine B moving through the arrayed interconnect Parylene C microchannel via capillary action. Scale bar is 1 mm.

Figure 3-51 Time-lapsed images of Rhodamine B moving through the arrayed interconnect Parylene
C microchannel in a partially packaged device. Scale bar is 1 mm.

3.3.4 Summary

An arrayed interconnect design capable of rapid and multiple simultaneous
connections to a microfluidic device. Additionally, this device is reusable, where
connections can be established, broken, and re-established up to 10 times. Optimal
needle size and style was determined (33G non-coring), and the linear relationship
between insertion and removal forces and the number of needles was theoretically
291

and experimentally verified. While the failure pressure of the arrayed interconnect
was limited by the weakest connection, up to 62 kPa of pressure was supported.
Also, interconnects were able to maintain 25 kPa of pressure for over 24 hours.
Connector spacings of 2.54 and 1 mm were fabricated, however, FEM analysis of
stress distribution shows 33G needles can be spaced as close as 120 μm, center-to-
center. Functionality of additional features such as needle guides, sideports for
combining two fluids, and electrolysis structures were demonstrated.

Modular arrayed interconnects are possible by choosing different combinations of
the septa shape, septa spacing, side ports, needle guides, microchannel material, and
metal structures. Similar to the single interconnect, septa length can be increased or
decreased based on available space and operating pressure requirements.
Additionally, the arrayed interconnect can be scaled to have any number of septa
(not just limited to 1, 4 or 8 septa designs) (Figure 3-52).

Figure 3-52 Illustration of the modularity of scale for the arrayed interconnect. The septa portion of
the interconnect can be elongated or repeated any amount of times to fit the needs of the system.

292

A “plug and play” model for microfluidic designs is now possible using standardized
connections and individual microfluidic components. For example, separate
components such as microchambers, microchannels, mixers, diffusers, heaters, etc
can be fabricated with septa at the input and output of these components. The
components can then be connected in any combination to create an entire
microfluidic system (Figure 3-53). This “plug and play” concept makes microfluidic
systems more readily available. Components, along with the needle arrays, can be
used to create3 on-demand microfluidic systems, which are easily adjusted. Arrayed
interconnects provide a rapid and standard means for connecting to microfluidic
systems to laboratory setups or even to other microfluidic systems.

Figure 3-53 Example of "plug and play" modularity .

293

294
3.4 Future Work

Several design variations of the arrayed interconnect using difference septa shapes,
septa spacing, microchannel material, sideports, and metal structures has been
fabricated and demonstrated. Further work on minimizing failure modes of needle
misalignment and delamination between septa and septa housing can be conducted.
Some possible solutions for preventing one form of needle misalignment (where the
needle path veers off-center) is to make the needles within the needle array stiffer by
shortening the needles, increasing the needle diameter, and investigating the
possibility of pyramidal tipped needles. To prevent the other observed instance of
needle misalignment, where the needle pierces the device at an angle relative to the
device substrate, is to fabricate the SU-8 housing in a three layer process creating a
needle guide which only exists in the middle layer. Additionally, the needle guide
length can be increased. However, needle misalignment may be further reduced
when the needle and septa pair is mass produced in a regulated mass manufacturing
facility.

The delamination of the septa can be mitigated by removing the Parylene C located
between the septa and substrate. The SU-8 housing can be roughened using oxygen
plasma.

295
Chapter 4 Conclusion

The drug delivery device was the first MEMS based ocular drug delivery pump.
Presented here were a proof-of-concept manually-actuated device, electrically-
actuated device and surgical shams. The fully integrated device consists of a
refillable reservoir, cannula, dual regulation check valve, suture tabs, and an
electrolysis pump. The device is made of biocompatible materials. The reservoir is
capable of multiple refills, which extend device lifetime. The check valve controls
the fluid flow and provides bandpass regulation. The valve prevents drugs from
backflowing or diffusing from the device when not in operation, it also protects the
device from accidental dosing due to transient pressure spikes. Furthermore, the
entire device is made of interchangeable parts, allowing parallel assembly and the
ability to easily exchange components.

The microfluidic interconnect provides a means for a standard packaging scheme for
micro-to-macro devices. Both a single and arrayed horizontal interconnect were
presented, which were both capable of multiple uses. These adhesiveless
connections between the macro and micro worlds could be established, broken, and
re-established multiple times. While the interconnect provides a standard
connection, several features (e.g. septa shape, septa spacing, microchannel design,
sideports, metal components, needle arrays) can be chosen to tailor the interconnect
to a specific device or application. Additionally, separate microfluidic components

296
with interconnects at the inlet and outlet can be created, and then assembled in any
order using the needle arrays. This results in a “plug and play” approach to
microfluidic device design.

Devices which can incorporate modular components are very powerful designs
which increase the versatility, flexibility, and potential functionality of the device.
Furthermore, if many of the parts are pre-fabricated, than any number of
combinations are possible and available on-demand without the need of a cleanroom
facility. The two devices presented in this work show how simple changes to a
device can produce an entirely different set of operating characteristics. By
providing the basic building blocks (much like a box of micro-Legos® ) and having
an understanding of how modular components can affect a system, the combinations
of possible devices are endless.

297
References

N. Abolhassani, R. Patel and M. Moallem, Med Eng Phys, 29, 4 (2007)

M. Agarwal, R. A. Gunasekaran, P. Coane and K. Varahramyan, Journal of
Micromechanics and Microengineering, 15, 1 (2005)

J. Ambati, E. S. Gragoudas, J. W. Miller, T. T. You, K. Miyamoto, F. C. Delori and
A. P. Adamis, Invest Ophthalmol Vis Sci, 41, 5 (2000)

J. R. Anderson, D. T. Chiu, R. J. Jackman, O. Cherniavskaya, J. C. McDonald, H. K.
Wu, S. H. Whitesides and G. M. Whitesides, Analytical Chemistry, 72, 14 (2000)

D. Armani, C. Liu and N. Aluru, Proceedings of the IEEE Micro Electro Mechanical
Systems (MEMS'99), Orlando, FL, pp. 222-227 (1999)

S. J. Baldock, P. R. Fielden, N. J. Goddard, H. R. Kretschmer, J. E. Prest and B. J. T.
Brown, J Chromatogr A, 1042, 1-2 (2004)

S. J. Baldock, P. R. Fielden, N. J. Goddard, H. R. Kretschmer, J. E. Prest and B. J. T.
Brown, Journal of Chromatography A, 1042, 1-2 (2004)

A. A. S. Bhagat, P. Jothimuthu, A. Pais and I. Papautsky, Journal of Micromechanics
and Microengineering, 17, 1 (2007)

S. Bhattacharya, A. Datta, J. M. Berg and S. Gangopadhyay, Journal of
Microelectromechanical Systems, 14, 3 (2005)

R. Bradley, Radiation Technology Handbook. (Marcel Dekker, Inc., New York, NY,
1984)

S. H. Chao, R. Carlson and D. R. Meldrum, Lab Chip, 7, 5 (2007)

C. H. Chiou and G. B. Lee, J Micromech Microeng, 14, 11 (2004)

C. H. Chiou and G. B. Lee, Journal of Micromechanics and Microengineering, 14,
11 (2004)

A. M. Christensen, D. A. Chang-Yen and B. K. Gale, J Micromech Microeng, 15, 5
(2005)

298
A. M. Christensen, D. A. Chang-Yen and B. K. Gale, Journal of Micromechanics
and Microengineering, 15, 5 (2005)

A. P. Dahlin, S. K. Bergstrom, P. E. Andren, K. E. Markides and J. Bergquist,
Analytical Chemistry, 77, 16 (2005)

J. H. Daniel, B. Krusor, R. B. Apte, M. Mulato, K. V. Schuylenbergh, R. Lau, T. Do,
R. A. Street, A. Goredema, D. C. Boils-Boissier and P. M. Kazmaier, Journal of
Vacuum Science and Technology A, 19, (2001)

A. K. Dash and G. C. Cudworth, 2nd, J Pharmacol Toxicol Methods, 40, 1 (1998)

M. Despont, H. Lorenz, N. Fahrni, J. Brugger, P. Renaud and P. Vettiger,
Proceedings of the IEEE Micro Electro Mechanical Systems (MEMS'97), Nagoya,
Japan, pp. 518-522 (1997)

D. C. Duffy, J. C. McDonald, O. J. A. Schueller and G. M. Whitesides, Analytical
Chemistry, 70, 23 (1998)

F. T. Fraunfelder, Ophthalmology, 86, 1 (1979)

F. T. Fraunfelder, Ophthalmology, 87, 2 (1980)

F. T. Fraunfelder, Jama, 237, 5 (1977)

F. T. Fraunfelder, Am J Ophthalmol, 110, 4 (1990)

F. T. Fraunfelder and S. M. Meyer, Aust J Ophthalmol, 12, 2 (1984)

F. W. Fraunfelder, Insight, 29, 2 (2004)

F. W. Fraunfelder and F. T. Fraunfelder, Ophthalmology, 111, 7 (2004)

P. Galambos, G. L. Benavides, M. O. M. W. Jenkins and D. Hetherington, Proc.
2001 ASME Conf, New York, New York, pp. 1-8 (2001)

A. N. Gent and G. L. Liu, J Mater Sci, 26, 9 (1991)

A. N. Gent and G. L. Liu, Journal of Materials Science, 26, 9 (1991)

D. H. Geroski and H. F. Edelhauser, Invest Ophthalmol Vis Sci, 41, 5 (2000)

A. C. R. Grayson, R. S. Shawgo, Y. W. Li and M. J. Cima, Advanced Drug Delivery
Reviews, 56, 2 (2004)

299
C. M. Ho and Y. C. Tai, Annu Rev Fluid Mech, 30, (1998)

D. Karnopp, Journal of Dynamic Systems Measurement and Control-Transactions of
the Asme, 107, 1 (1985)

R. Kee Suk, W. Xuefeng, K. Shaikh and L. Chang, Journal of
Microelectromechanical Systems, 13, 4 (2004)

C. S. Kim, J. H. Yim, E. K. Lee and N. H. Lee, Clinical and Experimental
Ophthalmology, 36, 2 (2008)

E. S. Lee, D. Howard, E. Liang, S. D. Collins and R. L. Smith, Journal of
Micromechanics and Microengineering, 14, 4 (2004)

E. S. Lee, D. Howard, E. Liang, S. D. Collins and R. L. Smith, J Micromech
Microeng, 14, 4 (2004)

S. S. Lee, P. Yuan and M. R. Robinson, Encyclopedia of Biomaterials and
Biomedical Engineering, (2004)

P.-Y. Li, J. Shih, R. Lo, B. Adams, R. Agrawal, S. Saati, M. S. Humayun, Y.-C. Tai
and E. Meng, Proceedings of the IEEE Micro Electro Mechanical Systems
(MEMS'07), Kobe, Japan, pp. 15-18 (2007)

P. Y. Li, J. Shih, R. Lo, S. Saati, R. Agrawal, M. S. Humayun, Y. C. Tai and E.
Meng, Sensors and Actuators A-Physical, 143, 1 (2008)

S. F. Li and S. C. Chen, IEEE Trans on Advanced Packaging, 26, 3 (2003)

S. F. Li and S. C. Chen, Ieee Transactions on Advanced Packaging, 26, 3 (2003)

K. S. Lim, Cornea, 22, 4 (2003)

J. C.-H. Lin, P.-J. Chen, B. Yu, M. Humayun and Y.-C. Tai, IEEE 22nd International
Conference on Micro Electro Mechanical Systems, pp. 196-199 (2009)

J. Liu, C. Hansen and S. R. Quake, Analytical Chemistry, 75, 18 (2003)

R. Lo, K. Kuwahara, P. Y. Li, R. Agrawal, M. S. Humayun and E. Meng, Okinawa,
Japan, pp. 4 pp. (2006)

R. Lo, P.-Y. Li, S. Saati, R. Agrawal, M. Humayun and E. Meng, Biomedical
Microdevices, (2008)

300
R. Lo, P.-Y. Li, S. Saati, R. N. Agrawal, M. S. Humayun and E. Meng, Biomedical
Microdevices, (2009)

R. Lo and E. Meng, IEEE 22nd International Conference on Micro Electro
Mechanical Systems, pp. 236-239 (2009)

R. Lo and E. Meng, Sensors and Actuators B: Chemical, 132, 2 (2008)

R. Lo and E. Meng, Sensors and Actuators B-Chemical, 132, 2 (2008)

N. Mamalis, L. Kearsley and E. Brinton, Curr Opin Ophthalmol, 13, 1 (2002)

S. Matsumoto, J. Xie and Y.-C. Tai, 10th International Conference on Miniaturized
Systems for Chemistry and Life Sciences Squaw Valley, California, pp. 551-4
(2003)

J. C. McDonald, D. C. Duffy, J. R. Anderson, D. T. Chiu, H. K. Wu, O. J. A.
Schueller and G. M. Whitesides, Electrophoresis, 21, 1 (2000)

J. C. McDonald and G. M. Whitesides, Accounts of Chemical Research, 35, 7 (2002)

E. Meng, J. Shih, P.-Y. Li, R. Lo, M. S. Humayun and Y.-C. Tai, 11th International
Conference on Miniaturized Systems for Chemistry and Life Sciences (µTAS 2006),
Tokyo, Japan, pp. 633-635 (2006)

E. Meng, S. Wu and Y.-C. Tai, Fresenius Journal of Analytical Chemistry, 371, 2
(2001)

E. Meng, S. Y. Wu and Y. C. Tai, Fresen J Anal Chem, 371, 2 (2001)

D. C. Metrikin and R. Anand, Curr Opin Ophthalmol, 5, 3 (1994)

E. R. Murphy, T. Inoue, H. R. Sahoo, N. Zaborenko and K. F. Jensen, Lab on a Chip,
7, 10 (2007)

J. Naito, M. Shikida, M. Hirota, Z. Y. Tan and K. Sato, IEEE International
Conference on Micro Electro Mechanical Systems, Tucson, AZ, United states, pp.
924-927 (2008)

M. D. O'Leary, C. Simone, T. Washio, K. Yoshinaka and A. M. Okamura, pp. 1774-
1780 vol.2 (2003)

K. W. Oh and C. H. Ahn, Journal of Micromechanics and Microengineering, 16, 5
(2006)

301
A. M. Okamura, C. Simone and M. D. O'Leary, Biomedical Engineering, IEEE
Transactions on, 51, 10 (2004)

T. Pan, A. Baldi and B. Ziaie, Journal of Microelectromechanical Systems, 15, 1
(2006)

T. Pan, J. D. Brown and B. Ziaie, IEEE Engineering in Medicine and Biology
Society, Piscataway, NJ, USA, pp. 4 pp. (2006)

A. V. Pattekar and M. V. Kothare, J Micromech Microeng, 13, 2 (2003)

A. V. Pattekar and M. V. Kothare, Journal of Micromechanics and
Microengineering, 13, 2 (2003)

A. Puntambekar and C. H. Ahn, Journal of Micromechanics and Microengineering,
12, 1 (2002)

A. Puntambekar and C. H. Ahn, J Micromech Microeng, 12, 1 (2002)

S. Z. Razzacki, P. K. Thwar, M. Yang, V. M. Ugaz and M. A. Burns, Adv Drug
Deliv Rev, 56, 2 (2004)

C. Richard, M. R. Cutkosky and K. MacLean, American Society of Mechanical
Engineers, Dynamic Systems and Control Division (Publication) DSC, 67, (1999)

R. Ritch, M. B. Shields and T. Krupin, The Glaucomas. (Mosby, St. Louis, 1989)

C. Simone and A. M. Okamura, pp. 2085-2091 vol.2 (2002)

N. Tas, T. Sonnenberg, H. Jansen, R. Legtenberg and M. Elwenspoek, Journal of
Micromechanics and Microengineering, 6, 4 (1996)

S. Timoshenko and J. N. Goodier, Theory of Elasticity. (McGraw-Hill, New York,
1970)

J. H. Tsai and L. W. Lin, Journal of Micromechanics and Microengineering, 11, 5
(2001)

A. C. Ugural, Stresses in Plates and Shells. (McGraw-Hill, New York, 1999)

B. T. H. van Dooren, W. H. Beekhuis and E. Pels, Archives of Ophthalmology, 122,
5 (2004)

X.-Q. Wang, Q. Lin and Y.-C. Tai, Twelfth IEEE International Conference on Micro
Electro Mechanical Systems, pp. 177-182 (1999)

302

G. M. Whitesides, Nature, 442, 7101 (2006)

J. T. Wilensky, Curr Opin Ophthalmol, 10, 2 (1999)

Z. Yang and R. Maeda, J Chromatogr A, 1013, 1-2 (2003)

Z. Yang and R. Maeda, Journal of Chromatography A, 1013, 1-2 (2003)

T.-J. Yao, S. Lee, W. Fang and Y.-C. Tai, Proceedings of the IEEE Micro Electro
Mechanical Systems (MEMS'00), Miyazaki, Japan, pp. 624-627 (2000)

T.-J. Yao, S. Lee, W. Fang and Y.-C. Tai, Proceedings of the IEEE Micro Electro
Mechanical Systems (MEMS), (2000)

T. J. Yao, X. Yang and Y. C. Tai, Sensors and Actuators a-Physical, 97-8, (2002)

B. Ziaie, A. Baldi, M. Lei, Y. Gu and R. A. Siegel, Adv Drug Deliv Rev, 56, 2
(2004)

303
Appendices

Appendix A- Fabrication Process for Silicon Masters

1. Dehydrate wafer at 100 °C for 30 minutes.
2. Vapor coat wafer with hexamethyldisilazane (HMDS) adhesion promoter.
3. Spin coat photoresist (AZ 4620) 2 krpm, 40 seconds results in 10 μm
layer (Figure 2-12A, Figure 2-15A).
4. Pattern photoresist using bottom (Figure 4-2) or middle layer (Figure 4-3)
masks (Figure 2-12B, Figure 2-15B).
5. Remove native oxide with 10% hydrofluoric acid dip.
6. Etch DRIE
a. 100 μm etch for bottom layer (Figure 2-12C).
b. 250 μm etch for middle layer (Figure 2-15C).
7. Strip photoresist (acetone, isopropyl alcohol, deionized water).
8. Clean with plasma oxygen (400 mTorr, 400 W, 4 minutes).
9. Clean with RCA-1 cleaning process.
10. Vapor deposit 5 μm Parylene C.

304
Appendix B- Steps to Mount Silicon Master to Glass
Substrate
1. Clean a 5’’x 5’’ (127 mm x 127 mm) glass plate (IPA wipe).
2. Place glass plate on hotplate (90 °C), do not exceed 120 °C or Parylene C
will thermally degrade.
3. Place a few scraps of paraffin wax in center of glass plate.
4. When wax is complete melted, place wafer (unpatterned side down) on
melted wax.
5. Apply pressure on wafer (avoid patterned areas) so wafer is flush with
glass surface.
6. Wipe excess wax that seeps out from underneath the wafer.
7. Turn off hotplate and allow setup to cool to room temperature.
8. Remove glass/ wafer setup from hotplate.
9. Create an aluminum foil boat around glass plate.
a. A boat is made by placing the glass plate onto a sheet of aluminum
foil.
b. The foil sheet should extend beyond the glass slide edge by
approximately 2’’ on all sides.
c. The excess foil is turned up and the corners folded. The aluminum
walls contain excess PDMS. The boat is also necessary in order to
degas the PDMS. During the degassing process, PDMS may expand
up to 4 times in volume as trapped air pockets are removed.

Appendix C- Fabrication Steps for Creating Acrylic
Master

1. Laser-cut square pieces with rounded corners (6mm x 6mm, or 0.24’’ x
0.24’’) from a 0.8 mm (0.03 inches) thick piece of acrylic (Figure 4-1)
2. Clean acrylic pieces using IPA.
3. Clean glass microscope slide (1’’ x 2’’) with IPA
4. Mix 5 minute epoxy.
5. Using tweezers, carefully dip one side (6mm x 6mm face) into epoxy.
6. Wipe excess epoxy from square.
7. Place square onto glass slide, ensure square is at least 1 cm from edge of slide
and no closer than 1 cm to another square.
8. Press down on the square, make sure no epoxy seeps out from underneath the
square and that square does not shift (leaving an epoxy trail). Excess epoxy
will result in leaks when the top layer is assembled with the rest of the layers.
9. Allow the epoxy to cure (24 hours).
10. After 24 hours, gently push on the edge of each square to test if the square is
securely fastened to the slide.
11. Create an aluminum foil boat around each glass slide.

Figure 4-1 6 mm x 6 mm acrylic squares to form the acrylic mold of the device reservoir. Note,
image is not shown true to scale.
305

Appendix D- Mask Used to Fabricate Bottom Layer
Silicon Master

Figure 4-2 Mask used to create silicon master for the bottom layer of the drug delivery device. The
white sections are etched 100 μm into the silicon substrate to create a negative of the desired
structure.
306

307
Appendix E- Fabrication Steps for Creating Bottom
and Middle Layers

1. PDMS was prepared for creating the base layer.
2. Pour small puddle (approximately 15 g) of PDMS on center of wafer.
3. Spread PDMS over entire wafer by manually tipping the wafer (Figure
2-12D).
4. Ensure each component is fully covered with PDMS.
5. Tip wafer so excess PDMS is removed from wafer.
6. Place wafer in vacuum oven for 30 minutes at less than -30 inHg, or until
bubbles within PDMS are fully removed.
7. Place wafer into oven (30 minutes, 70 °C). When PDMS is fully cured,
remove wafer from oven. Alternately, PDMS can be cured for 24 hours
at room temperature (make sure glass substrate is placed on flat surface).
8. Using a razor blade, cut the PDMS following the outer diameter of the
wafer.
9. Clean a 5’’ x 5’’ glass plate (IPA wipe).
10. With a clean pair of tweezers, slowly lift the PDMS layer from the wafer
(Figure 2-12E).
11. If fabricating middle layer:
a. Obtain two additional 5’’ x 5’’ glass plates (clean with IPA).
b. Place the plates adjacent to each other with a 1cm gap between the
plate edges.
c. Place the PDMS sheet onto two glass plates with the side that had
been in contact with the wafer facing up. Align the PDMS sheet such
that the pattern for the cannula and check valve is over the 1cm gap.
d. Using a clean 33 gauge coring needle (OD 203 μm), align the needle
over the mark which indicates the location of the check valve.
e. Press down on the needle until it punctures the PDMS sheet. Visually
inspect the puncture location to ensure the needle did not tear the
PDMS sheet.
f. Using the tweezers, lift the PDMS sheet and realign another set of
check valve orifices over the gap. Continue until all of the check
valve orifices are created.
12. Place the PDMS sheet onto the glass plate with the side that had been in
contact with the wafer facing up. (i.e. the patterned side of the PDMS
sheet is packing up)
13. Using a clean and fine-tipped blade, separate out each bottom or middle
layer piece from the PDMS layer (cutting under the microscope leads to
more precise cuts) by cutting along the outline of each piece.
14. Place pieces into a Petri dish and cover for later use.

Appendix F- Mask Used to Fabricate Middle Layer
Silicon Master

Figure 4-3 Mask used to create silicon master for the middle layer of the drug delivery device. The
white sections are etched 250 µm into the silicon substrate to create a negative of the desired
structure.
308

309
Appendix G- Fabrication Process for Creating Top
Layer from Acrylic Master

1. Prepare PDMS.
2. Pour PDMS into the boat. Pour enough PDMS such that the PDMS extends
beyond the top of each acrylic square by 1 mm (0.04 inches).
3. Degas the PDMS.
4. Place mold in oven (70 °C for 20 minutes).
5. Check PDMS to see if the PDMS is “half-cured.” To check if PDMS is half-
cured, gently touch surface of PDMS with a mixing rod. Lift the rod, if the
PDMS sticks to the rod and deforms slightly then the PDMS is half-cured. If
piece is not half-cured, replace mold into oven and check every minute. Be
careful not to fully cure the PDMS.
6. Remove the glass side from the aluminum boat. Remove any excess PDMS
that is on the bottom surface of the slide.
7. Place slide on the cutting pattern (Appendix H).
8. Align the acrylic square to the inner square on the pattern.
9. Using a fine-tipped blade, separate the top layer from the mold by cutting
along the outer square on the pattern.
10. Get a bottom and middle layer that have already been oxygen plasma treated,
aligned and bonded.
11. Lift the piece from the mold using clean tweezers.
12. Carefully align the reservoir on the bottom/middle layer structure. Make sure
the edges of the reservoir match the edge on the middle layer.
13. Place assembled device into a Petri dish and cover. Let reservoir finish
curing in room temperature (24 hours).

Appendix H- Pattern to Cut PDMS Reservoirs

1. Remove the glass slide from the aluminum boat.
2. Remove any excess PDMS that cured to the underside of the glass side.
3. Place slide on the cutting pattern (Figure 4-4).
4. Align the acrylic square to the inner square on the pattern. (Align 6 mm x
6mm acrylic square over the 6mm x 6mm pattern.)
5. Using a fine-tipped blade, separate the top layer from the mold by cutting
along the outer square on the pattern (7 mm x 7 mm square). Tip: it is easier
to lift the square out if an extra piece is removed to create a hollow area next
to the square (i.e. cut a small rectangular piece from one of the sides of the
7mm x 7mm cut).
6. Get a bottom and middle layer that have already been oxygen plasma treated,
aligned and bonded.
7. Lift the piece from the mold using clean tweezers.
8. Carefully align the reservoir on the bottom/middle layer structure. Make sure
the edges of the reservoir match the edge on the middle layer.
9. Place assembled device into a Petri dish and cover. Let reservoir finish
curing in room temperature (24 hours).

Figure 4-4 Pattern used to cut uniform reservoirs from molded PDMS sheet.

310

311
Appendix I- Cleaning Process for Device Layers Prior
to Oxygen Plasma Treatment

1. Wipe two beakers using IPA, and place into fume hood.
2. Prepare diluted HCl solution in one beaker (1:10 DI H
2
O:HCl). Fill other
beaker with DI H
2
O.
3. Using plastic tweezers, gentle place all of the bottom and middle PDMS
layers that are to be cleaned into the dilute HCl solution for 30 minutes.
(Make sure that the number of bottom layers to be cleaned equals the number
of middle layers.)
4. Remove the pieces using the tweezers and place into DI H
2
O beaker to rinse.
5. Clean glass microscope slides (IPA wipe). The number of glass slides
needed equals the number of bottom pieces that are cleaned.
6. Remove one bottom piece from the DI H
2
O, dry using N
2
gas and place onto
glass slide with the patterns facing up (this side will be treated with oxygen
plasma).
7. Remove one middle piece from t he DI H
2
O, dry using N
2
gas and place onto
glass slide so that piece is orientated in a “mirror-image” of the bottom slide
(Figure 2-20). This placement expedites assembly after oxygen treatment.
8. Place glass slide into Petri dish and cover.

312
Appendix J- Oxygen Plasma Treatment Process for
Bonding Bottom and Middle Layers

1. Place one slide into RIE machine.
2. Setup RIE machine and expose slide to 100 mTorr, 100 W, 45 seconds.
3. Remove the slide from the RIE machine.
4. Squirt a small amount of 95% ethanol onto each piece.
5. Gently lift the middle piece and place onto bottom piece.
6. Use microscope to help align check valve orifice over the check valve seat
and edge of both layers. Alignment must be completed within 1 minute of
removing pieces from RIE machine.
7. Bake assembly for 30 minutes at 75°C to increase the strength of the bonds.
8. Place aligned pieces and glass slide into covered Petri dish.

313
Appendix K- Process for Making PDMS Members of
a Certain Thickness

1. Obtain a clean 5’’x5’’ glass plate.
2. Make an aluminum foil boat which encloses the glass plate.
3. Determine how much PDMS is necessary to obtain desired thickness
(estimated value is a guide, but remember that some PDMS is loss inside the
mixing cup and along the edge between the glass plate and Al boat)
4. Make PDMS according to PDMS SOP.
5. Pour PDMS on glass plate. Spread PDMS on plate evenly either by tipping
the plate or using a clean glass slide to help spread PDMS.
6. Place glass plate into vacuum oven to remove the bubbles from the PDMS.
7. Remove plate from vacuum, place plate into oven (70°C for 20 minutes).
8. Remove plate from oven. Remove aluminum foil.
9. Align plate over cutting pattern.
10. Using a fine-tipped blade, cut 0.5’’x0.5’’ PDMS squares from PDMS slab.
11. Measure 2 clean glass cover slips using thickness gauge, record thickness.
12. Place PDMS square between cover slips, measure thickness and record
thickness.
13. Place PDMS square onto a marked glass slide to help catalog square and for
storage.
14. Repeat steps 12 and 13 until all squares are measured.
15. Subtract thickness of glass cover slips from cover slip and PDMS square
stack to obtain thickness of PDMS square.
16. Note if the prepared amount of PDMS resulted in squares of desired
thickness, if not, adjust the amount of PDMS in step 3.

Appendix L- Mask Used to Make Metal Alignment
Marks for All of the Modular Valve Processes

Figure 4-5 Mask used to create the metal alignment marks. This mask patterns a photoresist layer.
Metal is then deposited on the photoresist; the photoresist is removed using acetone, isopropyl
alcohol, and water.
314

Appendix M- Mask Used to Pattern First Layer of
SU-8 Valve Plate/ Pressure Limiter

Figure 4-6 Mask used to pattern the bottom layer of the SU-8 valve seat and pressure limiter. SU-8 is
a negative resist, therefore, the white areas indicate locations where SU-8 structures will remain,
while the black areas will be removed.
315

Appendix N- Mask Used to Pattern Second Layer of
SU-8 Valve Plate/ Pressure Limiter

Figure 4-7 Mask used to pattern the top layer of the SU-8 valve seat and pressure limiter. SU-8 is a
negative resist, therefore, the white areas indicate locations where SU-8 structures will remain, while
the black areas will be removed.
316

317
Appendix O- Fabrication Process for SU-8 Valve Seat
and Pressure Limiter

Metal Alignment Marks
1. Dehydrate glass wafer (120 ºC, 20 minutes)
2. Spin HMDS on wafer
a. Prespin 5sec 1k
b. Spin 30 sec 3k
3. Spin AZ 4400
a. Prespin 5sec 1k
b. Spin 30 sec 4k
4. Softbake at 90 ºC, 3 minutes
5. Expose (387 mJ/cm
2
) using Metal Mark Mask (Appendix L)
6. Develop
a. 1:4 AZ 351: DI H
2
O
b. Rinse in DI H
2
O
7. Check pattern
8. RIE (only descum immediately before metal dep)
a. ASH machine: 100 W, 100mT, 20 minutes
b. Descum 60W, 100mT, 1 minute
9. Metal deposition of Cr
a. 300 Å of Cr (follow Metal Dep SOP)
10. Liftoff PR in acetone, may need to ultrasound the wafer
11. Check pattern

Bottom and Top SU-8 Plate
1. Clean substrate
2. Dehydrate substrate (90 ºC, 20 minutes)
3. Apply OmniCoat
a. Dispense 1-4 mL of OmniCoat
b. Prespin 500 rpm for 5 sec (acceleration of 100rpm/sec)
c. Spin 3000rpm for 30 sec (acceleration of 300 rpm/sec)
d. Bake for 1 min at 200 ºC
4. Repeat step 2 a total of three times for triple coat of OmniCoat
(as per recommendation from MicroChem, Rob)
5. Spin SU-8 2100 to get ~160 µm thickness (thickness here does not have to be
precise, additional thickness makes structure stronger but need to make sure
exposure energy is enough to develop)
a. Coat wafer in SU-8 2100
b. Prespin at 500 rpm for 10 sec (acceleration 100 rpm/sec)

318
c. Spin 1750 rpm for 30 sec (acceleration of 300 rpm/sec)
d. Let sit at room temp for 2 hours to allow the SU-8 to planarize
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
f. Stay at 65 ºC for 5 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 30 minutes (might need to reduce this time because
this layer gets double-baked)
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
6. Expose SU-8 (~260 mJ/cm
2
* 1.5 if on glass, 390 mJ/cm
2
)
a. Apply mask for Bottom SU-8 Disk 1 (Appendix M)
b. Consider apply energy in 50 mJ/cm
2
doses to prevent overheating of
SU-8
7. Post exposure bake for 160 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 5 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 12 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
8. Spin SU-8 2050 to get 40 µm thickness (times are adjusted for a total
thickness of 200 µm)
a. Coat wafer in SU-8 2050
b. Prespin at 500rpm for 10 sec (acceleration 100rpm/sec)
c. Spin 4000rpm for 30 sec (acceleration of 300rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
f. Stay at 65 ºC for 7 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 3 hours
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
9. Expose SU-8 for 40 µm (~160 mJ/cm
2
* 1.2 if on SU-8, 192 mJ/cm
2
)
a. Apply mask for Bottom SU-8 Disk 2 (Appendix N)
10. Post exposure bake for 200 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 5 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 14 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
11. Develop SU-8
a. Pour SU-8 developer into one tray
b. Pour IPA into another tray

319
c. Place substrate into developer, mark time it takes to develop (will take
~17 minutes according to spec sheet)
d. Rinse in IPA (of white substance appears in unexposed portions,
development is not done)
12. Hard bake (optional)
a. Bake for 10 minutes at a temp 10 ºC higher than final device
operation temp. May need to do this if SU-8 reacts unfavorably
during heat-shrink
13. Liftoff SU-8 structures by using Remover PG
a. Place wafer in heated Remover PG (40-60 ºC)
b. As soon as structures lift from wafer, remove from Remover PG
c. Rinse with IPA
d. Rinse with water

Appendix P- Mask Used to Pattern SU-8 Spacer Plate

Figure 4-8 Mask used to pattern the SU-8 spacer plate. SU-8 is a negative resist, therefore, the white
areas indicate locations where SU-8 structures will remain, while the black areas will be removed.
320

321
Appendix Q- Fabrication Process for SU-8 Spacer
Plate

SU-8 Spacer Plate
1. Clean substrate
2. Dehydrate substrate (90 ºC, 20 minutes)
3. Apply OmniCoat
a. Dispense 1-4 mL of OmniCoat
b. Prespin 500 rpm for 5 sec (acceleration of 100rpm/sec)
c. Spin 3000rpm for 30 sec (acceleration of 300 rpm/sec)
d. Bake for 1 min at 200 ºC
4. Repeat step 2 a total of three times for triple coat of OmniCoat
(as per recommendation from MicroChem, Rob)
5. Spin SU-8 2050 to get 40 µm thickness
a. Coat wafer in SU-8 2050
b. Prespin at 500rpm for 10 sec (acceleration 100rpm/sec)
c. Spin 4000rpm for 30 sec (acceleration of 300rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
f. Stay at 65 ºC for 3 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 1 hour
i. Turn off hotplate and allow substrate to slowly cool to room temp
6. Expose SU-8 (~160 mJ/cm
2
* 1.5 if on glass, 240 mJ/cm
2
)
a. Apply mask for Center SU-8 Disk (Appendix P)
7. Post exposure bake for 40 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 1 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 6 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
8. Develop SU-8
a. Pour SU-8 developer into one tray
b. Pour IPA into another tray
c. Place substrate into developer, mark time it takes to develop
(approximately 5 minutes)
d. Rinse in IPA (of white substance appears in unexposed portions,
development is not done)
9. Hard bake (optional)

322
a. Bake for 10 minutes at a temp 10 ºC higher than final device
operation temp. May need to do this if SU-8 reacts unfavorably
during heat-shrink
10. Liftoff SU-8 structures by using Remover PG
a. Place wafer in heated Remover PG (40-60 ºC)
b. As soon as structures lift from wafer, remove from Remover PG
c. Rinse with IPA
d. Rinse with water

323
Appendix R- Mask Used to Pattern SU-8 Mold for
Silicone Valve Plate

Figure 4-9 Mask used to pattern the SU-8 mold used to cast the silicone valve plate. SU-8 is a
negative resist, therefore, the white areas indicate locations where SU-8 structures will remain, while
the black areas will be removed.

324
Appendix S- Fabrication Process for SU-8 Mold to
Create PDMS Valve Plate

Metal Alignment Marks
1. Dehydrate glass wafer (120 ºC, 20 minutes)
2. Spin HMDS on wafer
a. Prespin 5sec 1k
b. Spin 30 sec 3k
3. Spin AZ 4400
c. Prespin 5sec 1k
d. Spin 30 sec 4k
4. Softbake at 90 ºC, 3 minutes
5. Expose (387 mJ/cm
2
) using Metal Mark Mask (Appendix L)
6. Develop
e. 1:4 AZ 351: DI H
2
O
f. Rinse in DI H
2
O
7. Check pattern
8. RIE (only descum immediately before metal dep)
g. ASH machine: 100 W, 100mT, 20 minutes
h. Descum 60W, 100mT, 1 minute
9. Metal deposition of Cr
i. 300 Å of Cr (follow Metal Dep SOP)
10. Liftoff PR in acetone, may need to ultrasound the wafer
Check pattern

SU-8 Mold (flat membrane, no overlap, no bossed structure)
1. Clean substrate
2. Apply A174 adhesion promoter
3. Deposit ~5 µm Parylene C on substrate
4. Dehydrate substrate (90 ºC, 20 minutes)
5. Spin SU-8 2050 to get 75 µm thickness
a. Coat wafer in SU-8 2050
b. Prespin at 500 rpm for 10 sec (acceleration 100 rpm/sec)
c. Spin 2000 rpm for 30 sec (acceleration of 300 rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
f. Stay at 65 ºC for 3 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 90 minutes

325
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
6. Expose SU-8 (205 mJ/cm
2
* 1.5 if on glass, 308 mJ/cm
2
)
a. Apply mask for SU-8 PDMS Mold 3 (Appendix R)
7. Post exposure bake for 75 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 2 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 7 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
8. Develop SU-8
a. Pour SU-8 developer into one tray
b. Pour IPA into another tray
c. Place substrate into developer, mark time it takes to develop (~ 7
minutes according to spec sheet)
d. Rinse in IPA (of white substance appears in unexposed portions,
development is not done)
9. Hard bake
a. Bake for 10 minutes at 150 ºC (ramp from room temp to 150 ºC at 3
ºC/min), turn off heater and allow hot plate and SU-8 to return to
room temp slowly.
b. May help SU-8 survive if need to accelerate PDMS curing in oven
using this mold

MDX4-4210 Layer
1. Pour MDX4-4210 onto SU-8 MDX4-4210 Mold
2. Vacuum MDX4-4210 to remove bubbles
3. Scrap off excess MDX4-4210 slowly (try to prevent meniscus from forming)
4. Cure MDX4-4210
5. Lift off molded MDX4-4210, separate individual pieces, remove lift-off tab

Appendix T- Mask Used to Pattern SU-8 Mold for
Silicone Valve Plate with Optional Bossed Feature

Figure 4-10 Mask used to pattern the optional second layer for an SU-8 mold used to cast the silicone
valve plate with bossed feature. SU-8 is a negative resist, therefore, the white areas indicate locations
where SU-8 structures will remain, while the black areas will be removed.
326

327
Appendix U- Fabrication Process for SU-8 Mold to
Create Silicone Valve Plate with Bossed Feature

Metal Alignment Marks
1. Dehydrate glass wafer (120 ºC, 20 minutes)
2. Spin HMDS on wafer
a. Prespin 5sec 1k
b. Spin 30 sec 3k
3. Spin AZ 4400
a. Prespin 5sec 1k
b. Spin 30 sec 4k
4. Softbake at 90 ºC, 3 minutes
5. Expose (387 mJ/cm
2
) using Metal Mark Mask (Appendix L)
6. Develop
a. 1:4 AZ 351: DI H
2
O
b. Rinse in DI H
2
O
7. Check pattern
8. RIE (only descum immediately before metal dep)
a. ASH machine: 100 W, 100mT, 20 minutes
b. Descum 60W, 100mT, 1 minute
9. Metal deposition of Cr
a. 300 Å of Cr (follow Metal Dep SOP)
10. Liftoff PR in acetone, may need to ultrasound the wafer
Check pattern

MDX4210 Mold (flat membrane, with 40 µm bossed structure, no overlap)
1. Clean substrate
2. Apply A174 adhesion promoter
3. Deposit ~5 µm Parylene C on substrate
4. Dehydrate substrate (90 ºC, 20 minutes)
5. Spin SU-8 2050 to get 40 µm thickness
a. Coat wafer in SU-8 2050
b. Prespin at 500rpm for 10 sec (acceleration 100rpm/sec)
c. Spin 4000 rpm for 30 sec (acceleration of 300rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
f. Stay at 65 ºC for 3 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 1 hour

328
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
6. Expose SU-8 thickness 40 µm (160 mJ/cm
2
* 1.5 if on glass, 240 mJ/cm
2
)
a. Apply mask for SU-8 PDMS Mold 1 (Appendix T) (this is the bossed
structure layer)
7. Post exposure bake for 40 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 1 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 6 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
8. Spin SU-8 2050 to get 75 µm thickness
a. Coat wafer in SU-8 2050
b. Prespin at 500rpm for 10 sec (acceleration 100rpm/sec)
c. Spin 2000 rpm for 30 sec (acceleration of 300rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
(times adjusted for a total of 115 µm thickness)
f. Stay at 65 ºC for 5 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 2 hours
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
9. Expose SU-8 thickness 75 µm (205 mJ/cm
2
* 1.2 if on SU-8, 246 mJ/cm
2
)
a. Apply mask for SU-8 PDMS Mold 2 (Appendix R) (this is the valve
plate layer)
10. Post exposure bake for 115 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 5 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 10 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
11. Develop SU-8
a. Pour SU-8 developer into one tray
b. Pour IPA into another tray
c. Place substrate into developer, mark time it takes to develop (~ 11
minutes according to spec for 115 µm thickness)
d. Rinse in IPA (of white substance appears in unexposed portions,
development is not done)
12. Hard bake
a. Bake for 10 minutes at 150 ºC (ramp from room temp to 150 ºC at 3
ºC/min), turn off heater and allow hot plate and SU-8 to return to
room temp slowly.

329
b. May help SU-8 survive if need to accelerate PDMS curing in oven
using this mold

MDX4-4210 Layer
1. Pour MDX4-4210 onto SU-8 MDX4-4210 Mold
2. Vacuum MDX4-4210 to remove bubbles
3. Scrap off excess MDX4-4210 slowly (try to prevent meniscus from forming)
4. Cure MDX4-4210
5. Lift off molded MDX4-4210, separate individual pieces, remove lift-off tab

Appendix V- Mask Used to Pattern SU-8 Mold for
Silicone Valve Plate with Optional Bossed and
Overhang Features

Figure 4-11 Mask used to pattern the first layer for an SU-8 mold used to cast the silicone valve plate
with bossed and overhang feature. The thickness of this layer determines how far the overhang will
extend beyond the valve plate. SU-8 is a negative resist, therefore, the white areas indicate locations
where SU-8 structures will remain, while the black areas will be removed.
330

Figure 4-12 Mask used to pattern the second layer for an SU-8 mold used to cast the silicone valve
plate with bossed and overhang feature. This layer defines the bossed structure. SU-8 is a negative
resist, therefore, the white areas indicate locations where SU-8 structures will remain, while the black
areas will be removed.
331

Figure 4-13 Mask used to pattern the third layer for an SU-8 mold used to cast the silicone valve plate
with bossed and overhang feature. This layer is used to define the thickness of the valve plate and the
shape of the valve plate arms (i.e. through-holes). SU-8 is a negative resist, therefore, the white areas
indicate locations where SU-8 structures will remain, while the black areas will be removed.
332

333
Appendix W- Fabrication Process for SU-8 Mold to
Create Silicone Valve Plate with Bossed and Overhang
Features

Metal Alignment Marks
1. Dehydrate glass wafer (120 ºC, 20 minutes)
2. Spin HMDS on wafer
a. Prespin 5sec 1k
b. Spin 30 sec 3k
3. Spin AZ 4400
a. Prespin 5sec 1k
b. Spin 30 sec 4k
4. Softbake at 90 ºC, 3 minutes
5. Expose (387 mJ/cm
2
) using Metal Mark Mask (Appendix L)
6. Develop
a. 1:4 AZ 351: DI H
2
O
b. Rinse in DI H
2
O
7. Check pattern
8. RIE (only descum immediately before metal dep)
a. ASH machine: 100 W, 100mT, 20 minutes
b. Descum 60W, 100mT, 1 minute
9. Metal deposition of Cr
a. 300 Å of Cr (follow Metal Dep SOP)
10. Liftoff PR in acetone, may need to ultrasound the wafer
Check pattern

SU-8 Mold (flat membrane, with 40 µm bossed structure, with overlap)
1. Clean substrate
2. Apply A174 adhesion promoter
3. Deposit ~5 µm Parylene C on substrate
4. Dehydrate substrate (90 ºC, 20 minutes)
5. Spin SU-8 2100 to get 260 µm thickness
a. Coat wafer in SU-8 2100
b. Prespin at 500rpm for 10 sec (acceleration 100rpm/sec)
c. Spin 1000 rpm for 30 sec (acceleration of 300rpm/sec)
d. Let sit at room temp for 3 hours
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
f. Stay at 65 ºC for 7 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min

334
h. Stay at 95 ºC for 2.5 hours
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
6. Expose SU-8 thickness 260 µm (365 mJ/cm
2
* 1.5 if on glass, 548 mJ/cm
2
)
a. Apply mask for SU-8 PDMS Mold 1 (Figure 4-11) (this is the
overhang layer)
b. Consider apply energy in 50 mJ/cm
2
doses to prevent overheating of
SU-8
7. Post exposure bake for 260 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 5 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 19 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
8. Spin SU-8 2050 to get 40 µm thickness
a. Coat wafer in SU-8 2050
b. Prespin at 500 rpm for 10 sec (acceleration 100 rpm/sec)
c. Spin 4000 rpm for 30 sec (acceleration of 300 rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
(times adjusted for a total of 300 µm thick)
f. Stay at 65 ºC for 8 minutes
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 3 hours
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
9. Expose SU-8 thickness 40 µm (160 mJ/cm
2
* 1.2 if on glass, 192 mJ/cm
2
)
a. Apply mask for SU-8 PDMS Mold 2 (Figure 4-12) (this is the bossed
layer)
10. Post exposure bake for 300 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 5 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 21 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
11. Spin SU-8 2050 to get 75 µm thickness
a. Coat wafer in SU-8 2050
b. Prespin at 500rpm for 10 sec (acceleration 100 rpm/sec)
c. Spin 2000 rpm for 30 sec (acceleration of 300 rpm/sec)
d. Let sit at room temp for 1 hour
e. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
(adjusted for 375 µm thickness)
f. Stay at 65 ºC for 8 minutes

335
g. Ramp from 65 ºC to 95 ºC at 3 ºC/min
h. Stay at 95 ºC for 3 hours
i. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
12. Expose SU-8 thickness 75 µm (205 mJ/cm
2
* 1.2 if on SU-8, 246 mJ/cm
2
)
a. Apply mask for SU-8 PDMS Mold 3 (Figure 4-13) (this is the valve
plate layer)
13. Post exposure bake for 375 µm thick SU-8
a. Place wafer on hot plate, ramp from room temp to 65 ºC at 3 ºC/min
b. Stay at 65 ºC for 5 minutes
c. Ramp from 65 ºC to 95 ºC at 3 ºC/min
d. Stay at 95 ºC for 23.5 minutes
e. Turn off heater and allow hotplate and wafer to slowly cool to room
temp
14. Develop SU-8
a. Pour SU-8 developer into one tray
b. Pour IPA into another tray
c. Place substrate into developer, mark time it takes to develop (~ 23.5
minutes according to spec sheet for 375 µm)
d. Rinse in IPA (of white substance appears in unexposed portions,
development is not done)
15. Hard bake
a. Bake for 10 minutes at 150 ºC (ramp from room temp to 150 ºC at 3
ºC/min), turn off heater and allow hot plate and SU-8 to return to
room temp slowly.
b. May help SU-8 survive if need to accelerate PDMS curing in oven
using this mold

MDX4-4210 Layer
6. Pour MDX4-4210 onto SU-8 MDX4-4210 Mold
7. Vacuum MDX4-4210 to remove bubbles
8. Scrap off excess MDX4-4210 slowly (try to prevent meniscus from forming)
9. Cure MDX4-4210
10. Lift off molded MDX4-4210, separate individual pieces, remove lift-off tab

Appendix X- Assembling Heat Shrink Packaged Valve
SOP

Preparing the valve plate
1. Use mold with valve plate of 900 μm diameter pieces “squeegee 900 μm
membrane only”
2. Select the type of valve you want (Figure 4-14)

Figure 4-14 Three valve plate types: a) hole, b) straight arm, c) s-shaped arm

3. Gentle cut around valve plate and tab using x-acto knife (Figure 4-15a)
4. Lift valve plate with needle tweezers using tab and place on glass slide with
excess PDMS facing up (Figure 4-15b)
5. Gently open through-holes by removing excess PDMS (Figure 4-15c)
6. Cut excess PDMS from around the valve plate using x-acto knife (Figure
4-15d)
7. Remove tab (Figure 4-15e)

336

Figure 4-15 Steps to prepare valve plate

Assembling Valve

Figure 4-16 Items needed to packaging valve

1. Gather required pieces
a. 2 valve seat/ pressure limiter
b. 1 valve plate
c. 1 spacer plate
d. Needle tweezers
e. Assembly jig
f. 22G FEP heatshrink
g. Glass slide with PDMS square (assembly area)
337

2. Place valve seat, pressure limiter, valve plate, and spacer plate on assembly
area
3. Place assembly area under Lynx microscope
4. Check valve seat/ pressure limiter to make sure raised portion is facing up

Figure 4-17 Front and side views of SU-8 pieces

5. Align valve plate over valve seat. Make sure valve seat has raised portion
facing up and that the through-holes on the valve plate are not over the
through-hole in the valve seat)
6. Align spacer plate on valve plate
7. Align pressure limiter with raised portion facing down

Figure 4-18 Aligning pieces (top view) : valve plate only, valve plate with valve seat, valve plate &
valve seat & spacer plate, valve plate & valve seat & spacer plate & pressure limiter

Figure 4-19 Aligning pieces (side view) : valve plate only, valve plate with valve seat, valve plate &
valve seat & spacer plate, valve plate & valve seat & spacer plate & pressure limiter

8. Cut a section of 22G FEP heat-shrink tubing and place over needle on the jib
base. FEP tubing needs to be shorter than the needle
9. Place jig top onto jig bottom to make sure needles align, bend needles if
necessary, remove jig top
338

10. Place jig bottom on Teflon sheet

Figure 4-20 Process needed to shrink FEP tubing around valve

11. Lift stack and place on the tip of the needle with the FEP tubing, try to center
on needles as much as possible
12. Place jig top on and push top needle down on top of stack
13. Lift FEP tubing so it surrounds valve and is touching the top of the jig

Figure 4-21 Assembled valve in heat-shrink prior to placing in the oven

14. Take sheet to cleanroom, and place in vacuum oven
15. Remove air from vacuum oven
16. Set temp to 215 ºC with 1.5 ºC/min ramp
a. Turn on oven with switch on front of machine
b. Use arrow button to change target temp (215 ºC)
c. Press the enter button to accept new temp
d. Set overtemp control
339

340
i. Hold enter button down until menu appears
ii. Set the valve to 5 ºC more than target
iii. Hit enter to accept
e. Set ramp
i. Hold enter button down until menu appears
ii. Press enter button until LoC appears
iii. Change LoC valve to -1
iv. Press enter button until Upr appears
v. Change value to 1.5
vi. Press enter button until dpr appears
vii. Change value to 1.5
17. When oven reaches 215 ºC, decrease temp to 20 ºC
18. Remove jigs from oven when oven is cool
19. Remove valves from jigs, take pictures of top and side view, label valves
(description of which membrane inside and valve #)

Figure 4-22 Side and top view of a packaged valve

Appendix Y- Procedure to Test Heat-Shrink
Packaged Valve

Pressure System
1. Gather the following materials
a. Blank pressure test form
b. Valve
c. Silicone tubing (Standard Silicone Tubing REF 60-411-44, 0.04’’ ID,
0.085’’ OD)
d. 50 μL pipette (has a green tip)
e. Upchurch parts: luer connector (black and brown), brown tubing,
small brown screw to connect luer and tube
f. Ruler
g. 16G non-coring needle
h. Timer
2. Make sure images have been taken of the valve prior to testing
3. Attach Upchurch setup to pressure system

Figure 4-23 Testing setup to apply pressure to packaged valves or solid disks.

4. Place 16G needle on end of Upchurch luer connector
5. Cut ~ ¾’’ length of silicone tubing and connect one end to the 16G needle
and other end to the packed valve inlet. Push both parts into the tube as far as
possible to reduce the amount of compliant tubing
6. Prefill pipette with some water (so you can see meniscus)
7. Cut ~ ½’’ length of silicone tubing; put output end of packaged valve into
silicone tubing and connect other end to the 50 μL pipette
8. Tape ruler below the pipette
9. Fill out the pressure table with the information about the valve
10. Apply pressure to the valve
a. Make sure pressure system outlet valve is closed
b. Change pressure to target value
c. Mark location of meniscus in pipette
d. Set time to 3 minutes
341

342
e. Open pressure system outlet valve
f. When timer beeps, mark location of meniscus
11. If valve worked, try to test again immediately after first test is complete and
repeat as many times as you can. If the valve does not behave the same way,
try testing the valve after waiting 24 hours between tests
12. If valve seems to be stuck closed, try applying pressure to the outlet side and
try again
13. Examine valve under a microscope and see if you can observe any changes/
damage to the valve after testing is complete (take pictures).

Appendix Z- File Used To Make Custom-Designed
Cut Puncture Jig

Figure 4-24 The three custom-designed laser-machined layers that are stacked to form the puncture
jig.

1. The 30 gauge hole (lower right) is placed at the top of the stack to guide the
needle.
2. The 30 gauge hole with the membrane outline (upper right) is the middle
layer, this piece aligns the membrane such that the holes puncture the center
of the membrane.
3. The 20 gauge hole (lower left) is placed on the bottom of the stack. The hole
allows the needle to keep penetrating through the membrane until the luer
portion of the needle comes into contact with the top plate. This ensures that
each puncture is identical.
343

Appendix AA- File for Making the Puncture Force Jigs and Illustration of the Jig
Assembly

Figure 4-25 Corel Draw files to create puncture force jigs. Jig assembly is also shown. The colors are just used to indicate corresponding layers and are not
present in the laser file.

344

Appendix BB- Laser File for Making the Molds for
Possible Layouts for Version 1 of the Solid Surgical
Shams.

Figure 4-26 Corel draw files used to create custom-made, laser-machined molds of various shapes
and sizes for the first version of the solid surgical shams. The 0.75 mm to 2 mm labels indicate the
thickness of the sham.
345

346
Appendix CC- Laser File User to Create Solid
Surgical Sham v2_large and v2_small

Figure 4-27 File used to fabricate the redesigned solid surgical sham molds (v2_large and v1_small).
The dimensions are the same as in version 1 with additional sutures on the 1 mm thick sham and the
sutures removed from the silicone cannula from both shams.

Appendix DD- Laser File Used to Create Solid
Surgical Sham Mold v3_1

Figure 4-28 Drawing used to create solid sham mold v3_1. The mold is a convex dome which was
filled with silicone, leveled, and cured to create a solid sham.
347

Appendix EE- Laser File Used to Create Hollow
Surgical Sham Molds v3_2, v3_2, and v4_1

Figure 4-29 Drawing used to create hollow sham molds v3_2, v3_3, and v4_1. Shaded portions are
etched to create domes or flat surfaces.
348

Appendix FF- Laser File Used to Create Hollow
Surgical Sham Molds v5_1 and v6_1

Figure 4-30 Drawing used to create hollow sham molds v5_1 and v6_1. Shaded portions are etched
to create domes or flat surfaces.
349

Appendix GG- Laser File Used to Create Hollow
Surgical Sham Mold v7

Figure 4-31 Drawing used to create hollow sham mold v7. Shaded portions are etched to create
domes or flat surfaces.

350

351
Appendix HH- Fabrication Process for Making
Hollow Shams

1. Obtain acrylic molds: reservoir dome concave mold, reservoir dome convex
mold, device base mold. (Figure 2-75A).
2. Make PDMS and pour into reservoir dome concave and device base molds
(Figure 2-75B).
3. Cut a piece of silicone tubing (0.305 mm ID, 0.61 mm OD), thread a stripped
piece of wire-wrap wire (0.254 mm diameter) into the silicone tube. Ensure
that the wire extends beyond both sides of the tube.
4. Place stainless steel washer into bottom of reservoir dome concave mold
(Figure 2-75C).
5. Place molds in the vacuum oven and degas PDMS.
6. Gently place the PEEK baseplate into the device base mold, take care not to
introduce any bubbles into the PDMS. Place the wired silicone tube into tube
location indentation on the reservoir dome mold.
7. Place 2 glass cover slips on the reservoir dome mold, gently place the
reservoir dome convex mold on top of the reservoir dome concave mold
(Figure 2-75D), do not introduce any bubbles.
8. Place both molds into the oven at 70°C for 30 minutes.
9. Remove PDMS from molds (Figure 2-75E).
10. Cut extra PDMS from the pieces (Figure 2-75F). Reservoir dome concave
mold has a guiding groove to help cut the piece to the oval shape.
11. Align PDMS dome over PDMS base piece. Seal with a thin layer of PDMS
prepolymer (Figure 2-75G).
12. Optional- coat entire system with Parylene C.

352
Appendix II- Surgical Protocol for In Vivo
Implantation of Hollow Surgical Shams

1. Pre-surgical preparation:
a. Fill hollow shams with saline solution.
b. Manually-depress the device to verify functionality.
c. Fill the device with phenylephrine.
d. Pinch off the silicone tube by tying a suture around the base of the
tube (as close to the device body as possible).
2. Remove the conjunctiva.
3. Create a scleral tunnel, 3 mm wide and 2 mm posterior to the limbus.
4. Measure the appropriate length for the cannula.
5. Cut the device cannula to the desired length. Cut the cannula at an angle to
create a beveled tip. This tip shape aides insertion through the scleral tunnel.
6. Insert the cannula into the anterior chamber.
7. Suture the device to the sclera using the suture tabs.
8. Measure the baseline pupil size (horizontal and vertical).
9. Release the suture closing the cannula.
10. Manually press the device using blunt forceps. Mark time of dispensation.
11. Measure immediate pupillary response, mark time.
12. Replace conjunctiva.
13. Measure pupillary response at 10 minutes following dispensation.
14. Inject (subconjunctival) steroids and antibiotics were administered inferior
nasally.

353
Appendix JJ- Fabrication Process to Create the
Interconnect Test Structure

1. Obtain 2’’ x 3’’ glass slides or soda lime wafer (Figure 3-6A).
2. Clean the slides using a piranha clean procedure.
3. Treat the slides with A-174 adhesion promoter.
4. Vapor deposit Parylene C onto the slides (Figure 3-6B).
5. Dehydrate the slides at 90°C for 30 minutes.
6. Spin 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) using a
two step process (complete substep “a” for steps10-13 first before repeating
“b” substeps for steps 10-13):
a. First layer: spin at 1.5 krpm (approximately 200 μm thick).
b. Planarization layer: spin at 3 krpm (for an additional 100 μm) (Figure
3-6C).
7. a,b Leave applied SU-8 layer rested at room temperature for 3 hours to
improve planarization.
8. Softbaked layers at 90 °C. Baking steps were all performed on a
programmable hotplate (Dataplate Series 730, Barnstead International,
Debuque, IA) set to ramp at 3 °C/min.:
c. First layer softbaked for 90 minutes.
d. Planarization layer for 3 hours.
e. The lower softbake temperature was selected to avoid thermal
degradation of the underlying Parylene C.
9. a,b Slowly cool SU-8 to room temperature after each bake step to avoid
thermal stress cracks in the SU-8 (repeat steps 10-13 for substeps “a” and
then substeps “b”).
10. Pattern SU-8 (600 mJ/cm
2
) with mask in Figure 4-35.
11. Post-exposure baked for 30 minutes at 90 °C
12. Developed using SU-8 developer (MicroChem Corp., Newton, MA) (Figure
3-6D).
13. Final hardbake step was performed at 90 °C for 30 minutes.

For the Parylene C deposition step (step #4) the slides were placed in a commercial
Parylene C vapor deposition chamber. Parylene C is a pin-hole free, biocompatible,
and conformal material. Normally, the slides would be placed on the base of the
machine, however, to increase the capacity of the deposition chamber, a slide holder
was custom-designed and laser-cut. The slide holder is a modular design containing

354
6 levels, where each level is capable of holding 4 slides or wafers (for a maximum of
24 slides or wafers). Furthermore, both sides of the slide and/or wafer can be coated.
The file to laser-cut the acrylic needed to fabricate each level can be found in
Appendix LL.

355
Appendix KK- Fabrication Process to Create
Integrated Interconnect System

1. The substrate, either a 76 mm (3 inch) soda lime wafer (Silicon Quest
International, Santa Clara, CA) or soda lime slide substrate (75 mm x 50 mm,
Corning Glass Works, Corning, NY), was spin coated with AZ 4400
photoresist (AZ Electronic Materials, Branchburg, NJ) (4 krpm, 40 s, 4 μm)
(Figure 3-7A).
2. Expose PR and develop to create liftoff mask of metal layer.
3. E-beam evaporate Ti/Pt (200 Å/2000 Å) (International Advanced Materials,
Spring Valley, NY).
4. Liftoff metal in an acetone bath. Use a cleanroom swab to gently remove
extra metal. Last bits of metal can be removed by placing the substrate in an
acetone bath and quickly place the acetone bath into an ultrasound sonicator.
standard liftoff processes by removing the photoresist layer in acetone. Rinse
substrate/metal in isopropyl alcohol and deionized water (Figure 3-7B).
5. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(2 μm thick) to electrically isolate the metal traces (Figure 3-7C).
6. Spin coat AZ 4400 was spin coated (4 μm, 4 krpm, 40 s) to pattern Parylene
C.
7. Pattern PR and develop.
8. Use oxygen plasma to remove Parylene C to reveal the contact pads (Figure
3-7D).
9. Remove PR in acetone, isopropyl alcohol and deionized water.
10. Spin 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) using a
two step process (complete substep “a” for steps10-13 first before repeating
“b” substeps for steps 10-13):
a. First layer: spin at 1.5 krpm (approximately 200 μm thick).
b. Planarization layer: spin at 3 krpm (for an additional 100 μm) (Figure
3-7E).
11. a,b Leave applied SU-8 layer rested at room temperature for 3 hours to
improve planarization.
12. Softbaked layers at 90 °C. Baking steps were all performed on a
programmable hotplate (Dataplate Series 730, Barnstead International,
Debuque, IA) set to ramp at 3 °C/min.:
a. First layer softbaked for 90 minutes.
b. Planarization layer for 3 hours.
c. The lower softbake temperature was selected to avoid thermal
degradation of the underlying Parylene C.
13. a,b Slowly cool SU-8 to room temperature after each bake step to avoid
thermal stress cracks in the SU-8 (repeat steps 10-13 for substeps “a” and
then substeps “b”).
14. Pattern SU-8 (600 mJ/cm
2
).
15. Post-exposure baked for 30 minutes at 90 °C

356
16. Developed using SU-8 developer (MicroChem Corp., Newton, MA) (Figure
3-7F).
17. Final hardbake step was performed at 90 °C for 30 minutes.

Appendix LL- File for Creating a Layer of the
Parylene C Deposition Holder

Figure 4-32 Corel Draw file to create Parylene C deposition holder. 6 layers using this pattern were
cut. The assembled holder is modular and can be placed in the Parylene C deposition chamber with
up to all six layers in place. The Parylene C deposition holder layers are separated using plastic
standoffs.

1. The dark areas are etched down to show indentations for where the slides and
wafers are placed.
2. Tabs are cut to allow the wafers and slides to be removed easily.
3. Through holes are cut to encourage Parylene C circulation to other layers.

357

Appendix MM- Masks Used to Fabricate the Single
Interconnect Designs

The masks used to create the single interconnect are presented in the order they are
used. There are three masks for each interconnect design:
1. Metal Liftoff
2. Parylene C Etch
3. SU-8 Patterning
a. 300 μm channel
b. 500 μm channel

Figure 4-33 Mask used to pattern photoresist to create a metal liftoff layer for the single interconnect
design.

358

Figure 4-34 Mask used to pattern photoresist to create an etch mask for the Parylene C and expose
the metal electrodes and electrolysis structure on the single interconnect deign.

359

Figure 4-35 Mask used to pattern SU-8 layer to create a 300 μm channel. This produces a structure
that is compatible with using a 33 gauge needle to pierce the septum.

360

Figure 4-36 Mask used to pattern SU-8 layer to create a 500 μm channel. This produces a structure
that is compatible with using a 30 gauge needle to pierce the septum.

361

Appendix NN- Wafer Level Pictures of Arrayed
Interconnect

Four distinct wafers of designs that combine similar connector designs were
fabricated. The first two wafers contain SU-8 microchannels designs (Figure 4-37).
The remaining two wafers have designs that utilize Parylene C microchannels
(Figure 4-38). Each pair of wafers is further delineated by combining structures that
are integrated with metal structures (electrolysis and conductance) (Figure 4-37B,
Figure 4-38B). Table 3-5summarizes the possible combinations which can be
achieved using the masks to fabricate the four wafers presented in Figure 4-37 and
Figure 4-38.

Figure 4-37 Wafers with SU-8 microchannels, A) which do not contain metal structures, and B)
microchannels integrated with electrolysis and conductance structures. Wafer B can also be
fabricated without metal creating additional SU-8 microchannel interconnects without metal. The
blue areas identify the septum locations.

362

363

Figure 4-38 Wafers with Parylene C microchannels, A) that do not contain metal structures, and B)
microchannels integrated with electrolysis and conductance structures. Wafer B can also be
fabricated without metal creating additional SU-8 microchannel interconnects without metal. The
blue areas identify the septum locations.

Appendix OO- Mask Used To Fabricate Arrayed
Interconnect SU-8 Wafer 1
The mask used to create the SU-8 Wafer 1 of the arrayed interconnect designs is
presented in the order they are used. There is one masks for the SU-8 Wafer 1
layout:
1. SU-8 Patterning

Figure 4-39 Mask used to pattern SU-8 layer for the SU-8 microchannel arrayed interconnects found
in Figure 4-37A.

364

Appendix PP- Masks Used to Fabricate Arrayed
interconnect SU-8 Wafer 2

The masks used to create the SU-8 Wafer 2 of the arrayed interconnect designs are
presented in the order they are used. There are three masks for the SU-8 Wafer 2
layout:
1. Metal Liftoff
2. Parylene C Etching
3. SU-8 Patterning

Figure 4-40 Mask used to pattern photoresist to create a metal liftoff layer for the SU-8 microchannel
arrayed interconnects found in Figure 4-37B.

365

Figure 4-41 Mask used to pattern photoresist to create an etch mask for the Parylene C and expose
the metal electrodes and electrolysis structure on the SU-8 microchannel arrayed interconnects found
in Figure 4-37B.

366

Figure 4-42 Mask used to pattern SU-8 layer for the SU-8 microchannel arrayed interconnects found
in Figure 4-37B. This mask can be used without the masks in Figure 4-40 and Figure 4-41 to create
additional verions of the arrayed interconnect with SU-8 microchannels without metal.
367

Appendix QQ- Masks Used to Fabricate Arrayed
interconnect Parylene C Wafer 1

The masks used to create the Parylene C Wafer 1 of the arrayed interconnect designs
are presented in the order they are used. There are three masks for the Parylene C
Wafer 1 layout:
1. Sacrificial PR for Channel Definition
2. Parylene C Etching
3. SU-8 Patterning

Figure 4-43 Mask used to pattern photoresist to create a sacrificial photoresist structure that defines
the microchannel interior for the Parylene C arrayed interconnect designs found in Figure 4-38A.

368

Figure 4-44 Mask used to pattern photoresist to create an etch mask for the Parylene C covering the
microchannel opening of the Parylene C microchannel designs found in Figure 4-38A.

369

Figure 4-45 Mask used to pattern SU-8 layer for the Parylene C microchannel arrayed interconnects
found in Figure 4-38A.

370

Appendix RR- Masks Used to Fabricate Arrayed
interconnect Parylene C Wafer 2

The masks used to create the Parylene C Wafer 2 of the arrayed interconnect designs
are presented in the order they are used. There are three masks for the Parylene C
Wafer 2 layout:
1. Parylene C Etching 1
2. Metal Liftoff
3. Sacrificial PR for Channel Definition
4. Parylene C Etching 2
5. SU-8 Patterning

Figure 4-46 Mask used to pattern photoresist to create an etch mask which etches the Parylene C for
the Parylene C microchannel arrayed interconnects found in Figure 4-38B. The etched areas will
allow the electrodes of the metal structure to come into direct contact with the glass substrate.

371

Figure 4-47 Mask used to pattern photoresist to create a metal liftoff layer for the Parylene C
microchannel arrayed interconnects found in Figure 4-38B.

372

Figure 4-48 Mask used to pattern photoresist to create a sacrificial photoresist structure that defines
the microchannel interior for the Parylene C arrayed interconnect designs found in Figure 4-38B.

373

Figure 4-49 Mask used to pattern photoresist to create an etch mask for the Parylene C covering the
microchannel openings, electrodes, and electrolysis structures for Parylene C microchannel designs
found in Figure 4-38B.

374

Figure 4-50 Mask used to pattern SU-8 layer for the Parylene C microchannel arrayed interconnects
found in Figure 4-38B This mask, along with Figure 4-48 and Figure 4-49can be used without the
masks in Figure 4-46 and Figure 4-47 to create additional verions of the arrayed interconnect with
Parylene C microchannels without metal.
375

376
Appendix SS- Fabrication Process for Arrayed
Interconnects with SU-8 Microchannels

1. If necessary, clean substrate, 76 mm (3 inch) soda lime wafer (Silicon Quest
International, Santa Clara, CA), using standard piranha clean (1:4-5, H
2
O
2
:
H
2
SO
4
) (Figure 3-28A).
2. Treat substrate with A-174, an adhesion promoter.
3. Cover the back of the wafer with dicing saw tape.
4. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(2 μm thick) to prevent the SU-8 from delaminating (Figure 3-28B).
5. Spin 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) using a
two step process (complete substep “a” for steps12-15 first before repeating
“b” substeps for steps 10-13) (Figure 3-28C):
a. First layer: spin at 1.5 krpm, 30 sec (approximately 200 μm thick).
b. Planarization layer: spin at 3 krpm, 30 sec (for an additional 100 μm).
6. a,b Leave applied SU-8 layer rested at room temperature for 3 hours to
improve planarization.
7. Softbaked layers at 90 °C. Baking steps were all performed on a
programmable hotplate (Dataplate Series 730, Barnstead International,
Debuque, IA) set to ramp at 3 °C/min. A lower softbake temperature than the
manufacturer suggested value was selected to avoid thermal degradation of
the underlying Parylene C.
c. First layer softbaked for 90 minutes.
d. Planarization layer for 3 hours.
8. a,b Slowly cool SU-8 to room temperature after each bake step to avoid
thermal stress cracks in the SU-8 (repeat steps 12-15 for substeps “a” and
then substeps “b”).
9. Pattern SU-8 (600 mJ/cm
2
) ((Figure 3-28D), mask is shown in Appendix OO.
10. Remove dicing saw tape.
11. Post-exposure baked for 30 minutes at 90 °C. Ramp temperature from RT to
90˚C at 3˚C/min to avoid thermal stress. Once the post-exposure bake is
completed, slowly ramp down the temperature as well.
12. Developed using SU-8 developer (MicroChem Corp., Newton, MA). Be sure
to agitate the developer to speed up the development process.
13. Final hardbake step was performed at 90 °C for 30 minutes. Again, ramp
temperature from RT to 90˚C at 3˚C/min to avoid thermal stress. Once the
post-exposure bake is completed, slowly ramp down the temperature as well.
14. Dice designs from wafer.
15. Fill septa with PDMS (Figure 3-28E).
16. Cap device with glass slide (Figure 3-28F).

377
Appendix TT- Fabrication Process for Arrayed
interconnects with SU-8 Microchannels and Metal
Components

1. If necessary, clean substrate, 76 mm (3 inch) soda lime wafer (Silicon Quest
International, Santa Clara, CA), using standard piranha clean (1:4-5, H
2
O
2
:
H
2
SO
4
) (Figure 3-29A).
2. Dehydrate substrate, 120˚C for 20 min.
3. Spin coated with AZ 4400 photoresist (AZ Electronic Materials, Branchburg,
NJ) (4 krpm, 40 s, 4 μm) (Figure 3-29B).
4. Expose PR and develop to create liftoff mask of metal layer (Figure 3-29C),
mask shown in Figure 4-40.
5. Descum surface of substrate with oxygen plasma to remove any residues
prior to metal deposition.
6. E-beam evaporate Ti/Pt (200 Å/2000 Å) (International Advanced Materials,
Spring Valley, NY) (Figure 3-29D).
7. Liftoff metal in an acetone bath. Use a cleanroom swab to gently remove
extra metal. Last bits of metal can be removed by placing the substrate in an
acetone bath and quickly place the acetone bath into an ultrasound sonicator.
standard liftoff processes by removing the photoresist layer in acetone. Rinse
substrate/metal in isopropyl alcohol and deionized water. Dry with N
2
gas
(Figure 3-29E).
8. Treat substrate with A-174, an adhesion promoter.
9. Cover backside of wafer with dicing saw tape.
10. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(2 μm thick) to electrically isolate the metal traces and to prevent the SU-8
from delaminating (Figure 3-29F).
11. Spin coat AZ 4400 (4 μm, 4 krpm, 40 s) to pattern Parylene C (Figure
3-29G).
12. Pattern PR, removing dicing saw tape, and develop photoresist (mask in
Figure 4-41).
13. Etch Parylene C using oxygen plasma (Figure 3-29H).
14. Remove PR in acetone, isopropyl alcohol and deionized water (Figure 3-29I).
15. Spin 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) using a
two step process (complete substep “a” for steps12-15 first before repeating
“b” substeps for steps 10-13) (Figure 3-29J):
e. First layer: spin at 1.5 krpm (approximately 200 μm thick).
f. Planarization layer: spin at 3 krpm (for an additional 100 μm).
16. a,b Leave applied SU-8 layer rested at room temperature for 3 hours to
improve planarization.
17. Softbaked layers at 90 °C. Baking steps were all performed on a
programmable hotplate (Dataplate Series 730, Barnstead International,

378
Debuque, IA) set to ramp at 3 °C/min. A lower softbake temperature than the
manufacturer suggested value was selected to avoid thermal degradation of
the underlying Parylene C.
g. First layer softbaked for 90 minutes.
h. Planarization layer for 3 hours.
18. a,b Slowly cool SU-8 to room temperature after each bake step to avoid
thermal stress cracks in the SU-8 (repeat steps 12-15 for substeps “a” and
then substeps “b”).
19. Pattern SU-8 (600 mJ/cm
2
) (Figure 3-29K) (mask shown in Figure 4-42).
20. Post-exposure baked for 30 minutes at 90 °C. Ramp temperature from RT to
90˚C at 3˚C/min to avoid thermal stress. Once the post-exposure bake is
completed, slowly ramp down the temperature as well.
21. Developed using SU-8 developer (MicroChem Corp., Newton, MA).
22. Final hardbake step was performed at 90 °C for 30 minutes. Again, ramp
temperature from RT to 90˚C at 3˚C/min to avoid thermal stress. Once the
post-exposure bake is completed, slowly ramp down the temperature as well.
23. Dice structures from wafer.
24. Fill septa with PDMS (Figure 3-29L).
25. Cap entire structure with a glass slide (Figure 3-29M).

379
Appendix UU- Fabrication Process for Arrayed
Interconnects with Parylene C Microchannels

1. If necessary, clean substrate, 76 mm (3 inch) soda lime wafer (Silicon Quest
International, Santa Clara, CA), using standard piranha clean (1:4-5, H
2
O
2
:
H
2
SO
4
) (Figure 3-30A).
2. Treat wafer with A-174 adhesion promoter.
3. Cover one side of the wafer with dicing saw tape so that Parylene C will only
coat one side of the wafer. This will help the dicing process easier.
4. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(2 μm thick) to define the bottom layer of the microchannel (Figure 3-30B).
5. Remove dicing saw tape.
6. Spin coat AZ 4400 (4 μm, 4 krpm, 40 s) and pattern to define interior of
Parylene C microchannel (Figure 3-30C-D), mask shown in Figure 4-43.
7. Apply dicing saw tape to backside of the wafer.
8. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(4 μm thick) to define the bottom layer of the microchannel (Figure 3-30E).
9. Remove dicing saw tape.
10. Anneal two Parylene C layers without reflowing PR (optional step if Parylene
C layers start delaminating) (Figure 3-30F).
11. Apply dicing saw tape to the backside of the wafer, this helps with
controlling the exposure dosage for the photoresist and SU-8 (prevents
reflected UV light from overexposing the bottom of the photosensitive layer).
12. Spin coated with AZ 4903 photoresist (AZ Electronic Materials, Branchburg,
NJ) (4 krpm, 40 s, 6 μm) (Figure 3-30G), mask shown in Figure 4-44.
13. Remove dicing saw tape.
14. Pattern PR and develop to create Parylene C etch mask to etch microchannel
opening (Figure 3-30H).
15. Parylene C removed using oxygen plasma (Figure 3-30I).
16. Remove PR in acetone, isopropyl alcohol and deionized water (Figure 3-30J).
17. Spin 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) using a
two step process (complete substep “a” for steps12-15 first before repeating
“b” substeps for steps 10-13) (Figure 3-30K):
18. First layer: spin at 1.5 krpm (approximately 200 μm thick).
19. Planarization layer: spin at 3 krpm (for an additional 100 μm).
20. a,b Leave applied SU-8 layer rested at room temperature for 3 hours to
improve planarization.
21. Softbaked layers at 90 °C. Baking steps were all performed on a
programmable hotplate (Dataplate Series 730, Barnstead International,
Debuque, IA) set to ramp at 3 °C/min. A lower softbake temperature than the
manufacturer suggested value was selected to avoid thermal degradation of
the underlying Parylene C.
22. First layer softbaked for 90 minutes.

380
23. Planarization layer for 3 hours.
24. a,b Slowly cool SU-8 to room temperature after each bake step to avoid
thermal stress cracks in the SU-8 (repeat steps 12-15 for substeps “a” and
then substeps “b”).
25. Apply dicing saw tape to the backside of the wafer, this helps with
controlling the exposure dosage for the photoresist and SU-8 (prevents
reflected UV light from overexposing the bottom of the photosensitive layer).
26. Pattern SU-8 (600 mJ/cm
2
) (Figure 3-30L), mask shown in Figure 4-45.
27. Remove dicing saw tape.
28. Post-exposure baked for 30 minutes at 90 °C. Ramp temperature from RT to
90˚C at 3˚C/min to avoid thermal stress. Once the post-exposure bake is
completed, slowly ramp down the temperature as well.
29. Developed using SU-8 developer (MicroChem Corp., Newton, MA).
30. Final hardbake step was performed at 90 °C for 30 minutes. Again, ramp
temperature from RT to 90˚C at 3˚C/min to avoid thermal stress. Once the
post-exposure bake is completed, slowly ramp down the temperature as well.
31. Dice structures from wafer using dicing saw.
32. Place structure in isopropyl alcohol in room temperature for 3 days to remove
sacrificial photoresist inside the microchannel (Figure 3-30M).
33. Fill septa with PDMS (Figure 3-30N).
34. Cap entire structure with a glass slide (Figure 3-30O).

381
Appendix VV- Fabrication Process for Arrayed
Interconnects with Parylene C Microchannels with
Metal Components

1. If necessary, clean substrate, 76 mm (3 inch) soda lime wafer (Silicon Quest
International, Santa Clara, CA), using standard piranha clean (1:4-5, H
2
O
2
:
H
2
SO
4
) (Figure 3-32A).
2. Place dicing saw tape on one side of the wafer so that Parylene C only coats
one side of the wafer. This will help when dicing the setups apart.
3. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(2 μm thick) to define the bottom layer of the microchannel (Figure 3-32B).
4. Spin coated with AZ 4400 photoresist (AZ Electronic Materials, Branchburg,
NJ) (4 krpm, 40 s, 4 μm) (Figure 3-32C).
5. Expose PR and develop to create Parylene C etch mask, mask shown in
Figure 4-46. The electrodes must be deposited on the substrate in order for it
to be robust enough to withstand electrical connections (Figure 3-32D).
6. Etch Parylene C using oxygen plasma (Figure 3-32E).
7. Remove PR in acetone, isopropyl alcohol and deionized water (Figure
3-32F).
8. Spin coated with AZ 4400 photoresist (AZ Electronic Materials, Branchburg,
NJ) (4 krpm, 40 s, 4 μm).
9. Expose PR and develop to create metal liftoff layer (Figure 3-32G), mask
shown in Figure 4-47.
10. E-beam evaporate Ti/Pt (200 Å/3000 Å) (International Advanced Materials,
Spring Valley, NY) (Figure 3-32H).
11. Liftoff metal in an acetone bath. Use a cleanroom swab to gently remove
extra metal. Last bits of metal can be removed by placing the substrate in an
acetone bath and quickly place the acetone bath into an ultrasound sonicator.
standard liftoff processes by removing the photoresist layer in acetone. Rinse
substrate/metal in isopropyl alcohol and deionized water. Dry with N
2
gas
(Figure 3-32I).
12. Spin coat AZ 4400 (4 μm, 4 krpm, 40 s) and pattern to define the interior of
Parylene C microchannel (Figure 3-32J-K), mask shown in Figure 4-48.
13. Vapor deposit Parylene C (Specialty Coating Systems, Inc., Indianapolis, IN)
(4 μm thick) to define the bottom layer of the microchannel (Figure 3-32L).
14. Spin coated with AZ 4903 photoresist (AZ Electronic Materials, Branchburg,
NJ) (6 μm, 4 krpm, 40 s) (Figure 3-32M).
15. Pattern PR and develop to create Parylene C etch mask to etch microchannel
opening (Figure 3-32N), Figure 4-49.
16. Parylene C removed using oxygen plasma (Figure 3-32O).
17. Remove PR in acetone, isopropyl alcohol and deionized water (Figure
3-32P).

382
18. Spin 300 μm layer of SU-8 2100 (MicroChem Corp., Newton, MA) using a
two step process (complete substep “a” for steps12-15 first before repeating
“b” substeps for steps 10-13): (Figure 3-32Q)
19. First layer: spin at 1.5 krpm (approximately 200 μm thick).
20. Planarization layer: spin at 3 krpm (for an additional 100 μm).
21. a,b Leave applied SU-8 layer rested at room temperature for 3 hours to
improve planarization.
22. Softbaked layers at 90 °C. Baking steps were all performed on a
programmable hotplate (Dataplate Series 730, Barnstead International,
Debuque, IA) set to ramp at 3 °C/min. A lower softbake temperature than the
manufacturer suggested value was selected to avoid thermal degradation of
the underlying Parylene C.
23. First layer softbaked for 90 minutes.
24. Planarization layer for 3 hours.
25. a,b Slowly cool SU-8 to room temperature after each bake step to avoid
thermal stress cracks in the SU-8 (repeat steps 12-15 for substeps “a” and
then substeps “b”).
26. Pattern SU-8 (600 mJ/cm
2
) (Figure 3-32R), mask shown in Figure 4-50.
27. Post-exposure baked for 30 minutes at 90 °C. Ramp temperature from RT to
90˚C at 3˚C/min to avoid thermal stress. Once the post-exposure bake is
completed, slowly ramp down the temperature as well.
28. Developed using SU-8 developer (MicroChem Corp., Newton, MA).
29. Final hardbake step was performed at 90 °C for 30 minutes. Again, ramp
temperature from RT to 90˚C at 3˚C/min to avoid thermal stress. Once the
post-exposure bake is completed, slowly ramp down the temperature as well.
30. Dice structures from wafer.
31. Place structure in isopropyl alcohol in room temperature for 3 days to remove
sacrificial photoresist inside the microchannel (Figure 3-32S).
32. Fill septa with PDMS (Figure 3-32T).
33. Cap entire structure with a glass slide (Figure 3-32U).

Appendix WW- 4 and 8 Needle Insertion Force Jigs and Assembly

Figure 4-51 Corel Draw file used to create custom-made, laser-machined, jigs to measure insertion force of 4 and 8 needles arrays. Jig assembly is also
shown. Single needle insertion tests completed using the 4 neede jig and aligning a single needle through one hole.

383

Abstract (if available)

Abstract Presented in this work are two devices, an ocular drug delivery device with a dual-regulation check valve, and an arrayed, horizontal microfluidic interconnect. Both devices were designed to contain modular components

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Implantable bioMEMS drug delivery systems

PDF

Parylene C bioMEMS for implantable devices with electrochemical interfaces

PDF

Wireless electrochemical drug delivery micropump with fully integrated electrochemical dose tracking feedback system

PDF

Parylene-based biomems sensors for multiple physiological systems

PDF

Piezoelectric MEMS for acoustic sensing, contactless RF switching, and resonant mass sensing

PDF

Integrated wireless piezoelectric ultrasonic transducer system for biomedical applications

PDF

A high frequency array- based photoacoustic microscopy imaging system

PDF

Spatial mapping of real-time quantitative shear stress with vascular oxidative stress

PDF

Parylene-based implantable interfaces for biomedical applications

PDF

Development of implantable Parylene-based MEMS technologies for cortical applications

PDF

Strategies for improving mechanical and biochemical interfaces between medical implants and tissue

PDF

Acoustic ejector employing lens with air-reflectors and piezoelectrically actuated tunable capacitor

PDF

A wireless implantable MEMS micropump system for site-specific anti-cancer drug delivery

PDF

The electrochemical evaluation of Parylene-based electrodes for neural applications

PDF

Development of micromachined technologies for neural interfaces

PDF

Piezoelectric ultrasonic and acoustic microelectromechanical systems (MEMS) for biomedical, manipulation, and actuation applications

PDF

Enzymatically crosslinked scaffold with controllable cell and growth factor delivery for use in tissue engineering

PDF

Development of back-end processing system for high frequency ultrasound b-mode imaging

PDF

Expanded functionality and scalability of modular fluidic and instrumentation components

PDF

Thin-film impedimetric sensors for chronic in vivo use: design and application to hydrocephalus treatment

Asset Metadata

Creator Lo, Ronalee (author)

Core Title Modular bio microelectromechanical systems (bioMEMS): intraocular drug delivery device and microfluidic interconnects

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 10/01/2009

Defense Date 07/28/2009

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag bioMEMS,drug delivery,interconnects,intraocular,microfluidic,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Meng, Ellis (committee chair), Kim, Eun Sok (committee member), Weiland, James D. (committee member)

Creator Email rlo.usc@gmail.com,rlo@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2633

Unique identifier UC1139829

Identifier etd-Lo-3200 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-601081 (legacy record id),usctheses-m2633 (legacy record id)

Legacy Identifier etd-Lo-3200.pdf

Dmrecord 601081

Document Type Dissertation

Rights Lo, Ronalee

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu