University of Southern California Dissertations and Theses

A wireless implantable MEMS micropump system for site-specific anti-cancer drug delivery

(USC Thesis Other)

A wireless implantable MEMS micropump system for site-specific anti-cancer drug delivery

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

A Wireless Implantable MEMS
Micropump System for Site-specific
Anti-cancer Drug Delivery

by

Heidi Marie Gensler

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

December 2013

Copyright 2013 Heidi Marie Gensler
ii

Acknowledgements

A German proverb says "to begin is easy, to persevere an art." I believe the key to
persevering is surrounding yourself with good people to guide and support you. I am grateful to
many people who have helped me through all the ups and downs of my academic career through
encouragement, advice, and support.
First and foremost, I must thank my mother Diane Gensler for always believing in me
and making sure I had all the best opportunities to help me succeed. She made sure that I went to
the best possible schools to develop my academic skills and encouraged me to get involved in
extracurricular activities to develop leadership and teamwork skills. My mother not only has
given me great advice over the years but also has exemplified what she teaches. She works hard
at everything she does and is one of the kindest people anyone could ever meet. For these
reasons and many more, she is my ultimate role model.
Thank you to my family and friends who have shared encouraging words and have
cheered me on. To my brothers Joe and Jeff, thank you for providing lots of entertaining STEM-
related “demonstrations” during my childhood that sparked my interest in engineering,
sometimes literally. I am grateful for them but even more grateful that I still have all my fingers,
toes, and eyebrows. To my sisters-in-law Jane and Amy, I am lucky to have several successful
and intelligent individuals who have already been down the path and happily share the lessons
they have learned. Special thanks to Amy Wilson, Courtney Williams, and Renee Carlson for
your friendship and for being there to listen when I needed it most. To Calvin Chow, thank you
for letting me take over your dining table in my final days of dissertation writing. To Dr. Rania
Daily, thank you for helping me keep the big picture in mind and giving me perspective on
navigating the PhD journey. Your mentorship has helped me so much.
There are several people who inspired me and made it possible for me to go to graduate
school. Thank you, Chris Green, for introducing me to Dr. Lisa Brannon-Peppas at UT Austin. It
was in her laboratory that I really learned what graduate school was about and where I met Dr.
Kimberly Homan, who inspired me to pursue a PhD. Kim, you are an amazing role model as
both a researcher and a person. To Dr. Brannon-Peppas, thank you for the opportunity to work in
your laboratory and for your advice and guidance during my graduate school application process.
To the Terry Foundation, you have heard it many times and you will hear it many more, but my
iii

college and graduate experiences were made possible by your financial support. I will do my best
to carry on the Terrys’ legacy by encouraging others to pursue higher education and helping
them achieve their goals. To the National Science Foundation, thank you for your support and I
hope the program continues to grow and inspire the next generation of researchers. Thank you
Dr. Kirsten Belgum, for truly making me feel that the sky is the limit and that I can achieve
anything I set my mind to.
When I applied to graduate school, I searched for a place where the research would have
a notable impact in improving medical technology. After only a brief tour of Dr. Ellis Meng’s
laboratory at USC, and without having even met Dr. Meng in person, it was immediately
apparent that this particular laboratory was special. To Dr. Meng, thank you for mentoring me
both academically and personally. You encouraged me to grow by pushing me beyond my
comfort zone and helped me redefine the boundaries of my abilities. You always want students
to achieve their best, and it is evident in the amount of time and care you take with guiding each
and every student. I know that your work will have a great impact on healthcare, and that this
will be magnified even more through those who you have inspired.
I would also like to acknowledge my collaborators at the USC Keck School of Medicine,
in particular Dr. Uttam Sinha, Dr. Rizwan Masood, and Sutao Zhu. You all taught me how to
communicate with a broader audience and provided excellent opportunities for me to see and
better understand the nuances of translating benchtop work to in vivo studies. Thanks to Dr.
Donghai Zhu for his expertise and supervision of the Keck Photonics Laboratory where some of
the system components were fabricated.
Special thanks to Dr. David D’Argenio, Dr. Jesse Yen, Dr. Qifa Zhou, and Dr. Geoffrey
Shiflett for serving on my qualifying and dissertation defense committees. I appreciated your
helpful feedback and insightful questions. To Mischal Diasanta, Sandra Johns, Daisy Ruesli, and
Karen Johnson in the biomedical engineering department, thank you for all the ways you have
supported me over the years. Many thanks to Jennifer Gerson, who patiently helped me through
the administrative side of finishing my PhD. To Meredith Drake Reitan and Kate Tegmeyer in
the Graduate School, thank you for your help and for organizing great networking events. To my
friends in GSBME, thank you for the fun times on hiking trails, at football tailgates, and playing
ultimate frisbee on the Natural History Museum lawn. I hope the organization continues to grow
and help improve the sense of community that made my graduate school years more enjoyable.
iv

To the Meng Lab, you have been my family and I can only hope to have as great a group
of colleagues in the future. To my colleague and one of my closest friends Roya Sheybani, much
of my work would not have been possible without you. I have been fortunate to work with
someone so intelligent, dependable, and easy to work with. You entertained my thoughts and
ideas, both good and bad, and were always supportive both professionally and personally. Your
successful career has already begun and more great things are to come as a result of your
achievements. To Dr. Jonathan Kuo, thanks for putting up with me as your lab neighbor for five
years. Your ability to get serious work done while keeping a relaxed and cheerful attitude is
admirable, and I look forward to hearing about your future endeavors as an entrepreneur. To Seth
Hara and Curtis Lee, thank you for prompting interesting philosophical discussions during lunch
and coffee time. To Lawrence Yu, Brian Kim, and Angelica Wood, thanks for the fun games of
racquetball that helped me destress. To Dr. Christian Gutierrez, your wisdom and experience has
helped all of us, and it has been a privilege working with you. To all of my labmates, I know you
all will be very successful in whatever you do. To former lab members Dr. Brian Li, Dr. Ronalee
Lo Mann, and Dr. Gabriela Mallen-Ornelas, thank you for welcoming me to the group and for
readily answering all my questions and sharing your knowledge. Thanks to Dr. Tuan Hoang for
your support and advice on graduate school and career planning. I am also grateful to Heather
Chen, Maneesh Gujrati, Diya Dwarakanath, Jason Hoffman, David Johnson, and Nethika
Ariyasinghe, all of whom helped develop and polish the experiments and protocols that aided in
my research.
And finally, to my fiancé Geo Tu, I will never be able to thank you enough. You have
supported me through the tough days and celebrated with me on the successful ones. When
things seemed to be getting out of control, you grounded me and gave me hope. Thank you for
having the patience and trust to wait out several years of a long-distance relationship and for
being my shoulder to lean on. I would not have been able to do it without you, and I am so happy
to be starting our life together, finally in the same city!

v

Table of Contents
Acknowledgements ............................................................................................................. ii
List of Tables ....................................................................................................................... x
List of Figures ..................................................................................................................... xi
Abstract .............................................................................................................................. xx
1 Drug Administration Technology for Rodents .......................................................... 1
1.1 Introduction ............................................................................................................. 1
1.1.1 Motivation ....................................................................................................... 1
1.1.2 Challenges and limitations of current technology .......................................... 1
1.1.3 Approach ......................................................................................................... 2
1.2 Current Drug Administration Methods and Technologies ................................. 2
1.2.1 Acute Methods ................................................................................................ 2
1.2.2 Chronic Methods ............................................................................................. 4
1.3 Development of a Wireless Implantable MEMS Micropump System ............. 10
1.3.1 System overview ........................................................................................... 10
1.3.2 Electrolysis actuation .................................................................................... 13
1.3.3 Previous work ............................................................................................... 14
1.4 Example applications for chronic periodic dosing ............................................. 17
2 Application of System for Anti-Cancer Drug Delivery ........................................... 18
2.1 Motivation .............................................................................................................. 18
2.2 Background ........................................................................................................... 19
2.2.1 Role of sphingosine-kinase-1 in head and neck squamous cell carcinoma .. 19
2.2.2 siRNA and the ribonucleic acid interference pathway ................................. 19
2.2.3 Blocking production of SphK1 using siRNA ............................................... 20
2.2.4 Limitations of GNR-SphK1siRNA nanoplex delivery ................................. 24
vi

2.3 Preliminary In Vivo Testing of Prototype Wired System.................................. 25
2.4 1
st
Generation MEMS Micropump System ........................................................ 27
2.4.1 Design ........................................................................................................... 27
2.4.2 Fabrication .................................................................................................... 28
2.4.3 Methods......................................................................................................... 29
2.4.4 Benchtop testing of 1
st
generation micropump system ................................. 30
2.4.5 In vivo evaluation ......................................................................................... 31
2.4.6 Benchtop testing of valves ............................................................................ 31
2.4.7 Benchtop testing of valved system ............................................................... 33
2.5 Challenges for Next Generation Wireless System Integration ......................... 34
2.5.1 Packaging ...................................................................................................... 34
2.5.2 One-way valve .............................................................................................. 34
2.5.3 Bellows electrochemical actuator ................................................................. 35
2.5.4 Wireless power.............................................................................................. 35
3 2
nd
Generation Micropump System ........................................................................... 36
3.1 2
nd
Generation Goals ............................................................................................ 36
3.2 Packaging ............................................................................................................... 36
3.2.1 Polymers for low-permeability packaging .................................................... 36
3.2.2 Design modifications .................................................................................... 38
3.2.3 Fabrication .................................................................................................... 38
3.3 Experimental Methods ......................................................................................... 39
3.3.1 Filling protocol.............................................................................................. 39
3.3.2 Benchtop testing of dosing regimen ............................................................. 39
3.3.3 Viscosity testing ............................................................................................ 40
3.3.4 Back pressure testing .................................................................................... 41
vii

3.3.5 Valve testing setup and integration with micropump ................................... 41
3.3.6 Benchtop testing of 2
nd
generation valved micropump system .................... 42
3.4 Results .................................................................................................................... 44
3.4.1 Valve characteristics ..................................................................................... 44
3.4.2 Benchtop dosing regimen results .................................................................. 44
3.4.3 Viscosity results ............................................................................................ 46
3.4.4 Back pressure results.................................................................................... 47
3.4.5 Valved system operation results ................................................................... 47
3.5 Summary ................................................................................................................ 49
3.6 Remaining Challenges .......................................................................................... 49
4 Rapid Fabrication and Characterization of Parylene C Bellows for Large Deflection
Applications ....................................................................................................................... 50
4.1 Motivation for Bellows Design and Fabrication Study ..................................... 50
4.2 Background ........................................................................................................... 50
4.2.1 Bellows fabrication methods and applications ............................................. 50
4.2.2 Bellows deflection theory and modeling ...................................................... 54
4.3 Approach ............................................................................................................... 54
4.4 Design and Fabrication......................................................................................... 55
4.5 Experimental Methods ......................................................................................... 59
4.5.1 Finite element models and simulations ......................................................... 59
4.5.2 Mechanical characterization setup ................................................................ 59
4.5.3 Demonstration in an electrochemical actuator .............................................. 61
4.5.4 Statistical analysis ......................................................................................... 61
4.6 Results .................................................................................................................... 62
4.6.1 Finite element model(FEM) and simulations ............................................... 62
viii

4.6.2 Mechanical characterization ......................................................................... 63
4.7 Discussion............................................................................................................... 71
4.8 Conclusion ............................................................................................................. 74
4.9 Bellows Integration with 3
rd
Generation Wireless Drug Delivery System ...... 75
5 Redesign, Integration, and Implementation of the 3
rd
Generation Wireless Micropump
System ................................................................................................................................ 76
5.1 Goals ....................................................................................................................... 76
5.1.1 Improvements upon 2
nd
generation ............................................................... 76
5.1.2 Application design specifications ................................................................. 76
5.2 Design ..................................................................................................................... 76
5.2.1 Bellows electrochemical actuator ................................................................. 77
5.2.2 Reservoir ....................................................................................................... 77
5.2.3 Valve port and connections ........................................................................... 79
5.2.4 Catheter assembly ......................................................................................... 80
5.2.5 Wireless power components ......................................................................... 81
5.3 Fabrication............................................................................................................. 82
5.3.1 Bellows electrochemical actuators ................................................................ 82
5.3.2 Reservoirs ..................................................................................................... 82
5.3.3 Septa .............................................................................................................. 83
5.4 Assembly ................................................................................................................ 84
5.5 Experimental Methods ......................................................................................... 86
5.5.1 Valve screening ............................................................................................. 86
5.5.2 Refill procedure ............................................................................................ 86
5.5.3 Septa robustness to multiple punctures ......................................................... 87
5.5.4 Benchtop wireless testing ............................................................................. 88
ix

5.5.5 Physiological environment simulation (soak test) ........................................ 89
5.5.6 Room versus body temperature operation .................................................... 89
5.5.7 Viscosity ....................................................................................................... 89
5.5.8 Back pressure ................................................................................................ 90
5.5.9 Dosing regimen ............................................................................................. 91
5.6 Results and Discussion .......................................................................................... 91
5.6.1 Valve screening ............................................................................................. 91
5.6.2 Septa robustness to multiple punctures ......................................................... 92
5.6.3 Benchtop wireless testing ............................................................................. 93
5.6.4 Physiological environment simulation (soak test) ........................................ 93
5.6.5 Room versus body temperature operation .................................................... 94
5.6.6 Viscosity ....................................................................................................... 95
5.6.7 Back pressure ................................................................................................ 96
5.6.8 Dosing regimen ............................................................................................. 98
5.7 Summary .............................................................................................................. 102
6 Conclusion ................................................................................................................. 105
References ........................................................................................................................ 107
APPENDIX A: Electrode Fabrication Process Flow ................................................... 114
APPENDIX B: 2
nd
Generation Wireless Micropump System Specifications ............ 115
APPENDIX C: Bellows Fabrication Protocol .............................................................. 116
APPENDIX D: 3
rd
Generation Wireless Micropump System Specifications ............ 121
APPENDIX E: 3
rd
Generation Wireless Micropump System Drawings ................... 122
APPENDIX F: 3
rd
Generation Micropump System Assembly Chart and Bill of Materials
123

x

List of Tables
Table 1. Commercial implantable pumps for infusion in rodents .................................................. 6
Table 2. Comparison of valve opening pressures ......................................................................... 32
Table 3. Summary of water vapor transmission rates for four polymers ...................................... 37
Table 4. Injection molding parameters for PP and PETG ............................................................ 39
Table 5. D-glucose solutions for modeling various viscosities .................................................... 40
Table 6. MEMS fabrication techniques for polymer bellows ([100] © 2012 IOP) ...................... 53
Table 7. Summary of the fabricated bellows designs ([100] © 2012 IOP) .................................. 56
Table 8. Finite element model material properties and bellows dimensions ([100] © 2012 IOP) 59
Table 9. Summary of burst pressures for the various bellows designs ([100] © 2012 IOP) ........ 64
Table 10. Summary of effects of individual bellows design parameters ...................................... 70
Table 11. Application specifications for the 3
rd
generation micropump system .......................... 76
Table 12. Comparison of reservoir dimensions for several generations of micropumps ............. 78
Table 13. Glucose solutions prepared for viscosity testing of the 3
rd
generation micropump...... 90
Table 14. Reverse leakage due to inadequate valve sealing between dosing ............................. 101
Table 15. Application specifications for the 3
rd
generation micropump system ........................ 105

xi

List of Figures
Figure 1. Gastric gavage (feeding tube through the esophagus and into the stomach) is used for
precise and accurate dosing, but the animal must be restrained and the drug must overcome
metabolism and absorption barriers. (Images courtesy of American Association for
Laboratory Animal Science, from Working with the Laboratory Mouse, AALAS Learning
Library) ................................................................................................................................... 3
Figure 2. Injection is common for acute bolus drug administration, but requires restraint and is
not suitable for chronic applications. Images © 2013 Newcastle University. ........................ 4
Figure 3. Commercial tethered infusion system for rodents. Image courtesy of SAI Infusion
Technologies. .......................................................................................................................... 5
Figure 4. Examples of active MEMS pumps [5] include the a) piezoelectric-based active
microport system worn externally by rats [23] (reprinted with kind permission from
Springer Science and Business Media) and b) dual drug delivery system intended for
humans [24] © 2011 IEEE. ..................................................................................................... 9
Figure 5. Examples of passive MEMS devices include a) microreservoir arrays demonstrated in
rats (reprinted from [29] with permission from Elsevier) and a b) microbolus pump
demonstrated in rats and mice [35] © 2009 IEEE. ............................................................... 10
Figure 6. Wireless implantable MEMS micropump system. a) Bellows electrochemical actuator
(BEA), packaging, wireless power, and check valve components, b) 3-D model of system
and c) in vivo testing setup with wireless power source. ...................................................... 11
Figure 7. Electrolysis-based operation of the MEMS micropump system. a) With the system off
(no applied current), a one-way valve prevents biological fluids from mixing with drug
contained in the reservoir. b) With applied current, water electrolysis produces hydrogen
and oxygen gases that extend the bellows and drive surrounding drug out of the reservoir
and catheter. c) After refill of the drug reservoir, dosing cycles continue as described in a)
and b). ................................................................................................................................... 12
xii

Figure 8. a) The electrochemical (EC) actuator consists of Nafion
®
-coated interdigitated
platinum (Pt) electrodes with 100 m finger width and 100 m spacing. b) Electrical
current applied to the electrodes dissociates the water into hydrogen and oxygen gases,
which then expels fluid from the chamber. ........................................................................... 15
Figure 9. Intraocular drug delivery device with refillable silicone rubber reservoir. Reprinted
from [37] with permission from Elsevier.............................................................................. 16
Figure 10. Side profile of the bellows structure (left) and a bellows mounted on an
electrochemical actuator (right). The bellows isolates the electrolysis reaction from the drug
reservoir. Adapted from [36] © 2009 IEEE. ......................................................................... 16
Figure 11. In vitro results for two HNSCC cell lines after treatment with radiation only,
SphK1siRNA only, and combined SphK1siRNA plus radiation. a) Cells were exposed to
radiation levels ranging from 0 to 50 Gray (Gy). b) Cells were transfected with 0 to 200 nM
of SphK1siRNA using a lipid-based transfection agent. c) Comparison of cells treated with
low-dose radiation (2.5 Gy) only, the lipid-based transfection agent only (as a control), an
intermediate amount of SphK1siRNA (100 nM), and SphK1siRNA (100 nM) plus
subsequent low-dose (2.5 Gy) radiation [64]. Copyright © 2010 Wiley Periodicals, Inc. ... 20
Figure 12. In vivo results for pre-treated tumor cells injected into nude mice. Tumor volumes
were tracked over a period of 18 days for GFPsiRNA control and SphK1siRNA, with
subsets of each group (n=6) irradiated with 1.0 Gy every 3 days [64]. Copyright © 2010
Wiley Periodicals, Inc. .......................................................................................................... 21
Figure 13. The positively-charged gold nanorods (GNR) bind to the cell membrane, facilitating
uptake via endocytosis. An innate endonuclease called Dicer separates SphK1siRNA into
single sense and antisense strands. The antisense strand is incorporated by RNA-induced
silencing complex (RISC; also innate), and then binds to the complementary sequence of
the messenger RNA (mRNA) sense strand. The mRNA is then cleaved, which blocks
production of the protein SphK1 which contributes to radiation resistance. ........................ 23
xiii

Figure 14. a) HNSCC cells were subjected in vitro to no treatment (NT), control GNR, control
GNR-GFPsiRNA (non-specific siRNA), and GNR-SphK1siRNA. A subset of each group
was irradiated with 1.0 Gy four hours after treatment. Tumor cell viability was significantly
(p <0.05) lower for GNR-SphK1siRNA nanoplex-treated groups compared to controls. b)
HNSCC tumor xenografts were grown in nude mice, then treated with PBS, GNR-
GFPsiRNA, and GNR-SphK1siRNA (n=6 per group, experiment repeated twice).
Treatments were administered via local injection to three sites proximal to the tumor three
times per week. A subset of each group was irradiated with 1.0 Gy every 3 days. GNR-
SphK1siRNA plus irradiation resulted in the greatest reduction in tumor volume at 25 days.
Adapted from [48] with permission of The Royal Society of Chemistry. ............................ 24
Figure 15. a) Conceptual depiction of placement of implanted micropump to deliver anti-cancer
drug (nanoplexes) to tumors. b) In vivo delivery with wired prototype micropump powered
with a battery pack. ............................................................................................................... 26
Figure 16. Photographs comparing size of the tumors after delivery from active pumps and
passive pumps (diffusion-based, no actuator). Adapted from [83] © 2010 IEEE. ............... 27
Figure 17. Assembly of 1
st
generation wired micropump system. a) A CNC mill was used to
make acrylic molds for casting silicone reservoirs. b) The bellows electrochemical actuator
(BEA) is inserted into the cast silicone rubber reservoir, then additional uncured silicone
rubber is used to seal the slot and adhere the lid to the reservoir. c) An active micropump
(top) included the wired BEA, while a passive micropump (bottom) consisted of the
reservoir shell only. 17b Reprinted from [82] with kind permission from Springer Science
and Business Media. ............................................................................................................. 28
Figure 18. Flow rate testing setup for 1
st
generation wired micropump system. .......................... 30
Figure 19. Flow rate results for the 1
st
generation micropump system. Constant current was
applied for 15 minutes, with a flow rate in the linear region of 4.72 ± 0.35 L/min, then
turned off for 45 minutes to allow the gases to recombine. Reprinted from [82] with kind
permission from Springer Science and Business Media. ...................................................... 30
xiv

Figure 20. Commercial one-way disc valve (left), commercial in-line check valve (middle), and
MEMS in-line check valve (right). Reprinted from [82] with kind permission from Springer
Science and Business Media. ................................................................................................ 31
Figure 21. Stand-alone valve testing setup to evaluate opening pressure of a MEMS valve and
two commercial valves. ........................................................................................................ 32
Figure 22. Flow rate measurement setup for 1
st
generation wired system in a silicone reservoir
with a MEMS valve integrated into the cannula. Commercial valves were tested similarly,
but with the inlet connected to an identical cannula without the integrated MEMS valve. . 33
Figure 23. Performance of non-valved and valved silicone reservoir systems. Reprinted from
[82] with kind permission from Springer Science and Business Media. .............................. 34
Figure 24. Water vapor transmission rates (WVTR; mean  SE, n = 4) for four polymers at 20 °C
and 20 %RH. ......................................................................................................................... 37
Figure 25. Injection molded reservoir design. a) Photo and c) conceptual drawing of base with
bellows electrochemical actuator (BEA) seated in slot. ....................................................... 38
Figure 26. Setup for applying physiologically relevant back pressures during system operation.41
Figure 27. Pressure testing setup for evaluation of a commercial one-way valve and photograph
(bottom left) of the commercial duckbill valve. ................................................................... 42
Figure 28. Constant current of 0.5 mA was applied to three wired and valved micropumps on the
benchtop to demonstrate performance with a periodic dosing regimen. .............................. 43
Figure 29. a) Diagram of wireless operation with inductive power system. b) Photographs of the
wireless components with prototype micropump system. c) No more than 20 cm should
separate the external transmitter and subcutaneously implanted receiver to ensure adequate
power transfer. ...................................................................................................................... 44
Figure 30. 2 mA constant current was applied to six wired rigid reservoir systems (photo) for
11.5 minutes. Consistent flow rates 17.40 ± 0.55 L/min (mean ± SE, n=6) and repeatability
xv

of total dosing volume 183.11 ± 5.58 L (mean ± SE, n=6) were demonstrated across six
micropump systems. ............................................................................................................. 45
Figure 31. 2 mA constant current was applied for ~3 minutes per day for three consecutive days
to show repeatability in periodic dosing of 60 L. ............................................................... 46
Figure 32. Constant currents of 0.5, 1, 2, and 5 mA were applied to a bellows electrochemical
actuator (BEA) in a rigid reservoir. With glucose solutions ranging in viscosity from 1 to
6.21 cP, standard error of less than 6% across the flow rates was observed. ....................... 46
Figure 33. A bellows electrochemical actuator (BEA) in a rigid reservoir was operated against
physiologically relevant back pressures. 5 mA constant current was applied for 1 minutes at
each value (n=4). Adapted from [42] with kind permission from Springer Science and
Business Media. .................................................................................................................... 47
Figure 34. Periodic dosing in wired valved micropumps (2 of 3 micropumps shown). ............... 48
Figure 35. Wireless power transfer was achieved on the 2
nd
generation micropump system when
it was within 20 cm of the transmitter coil, resulting in a forward flow rate on the order of
µL/min. When the transmitter was powered off, minimal reverse leakage was observed. .. 48
Figure 36. a) Standard profile of a bellows with design parameters labeled. b) Application of
load to a bellows results in axial extension, axial bending, or a combination thereof. c)
Cross-sections of flat and corrugated diaphragms (not to scale). Bellows can achieve higher
deflection than flat or corrugated diaphragms of similar dimensions. Image from [100] ©
2012 IOP. .............................................................................................................................. 51
Figure 37. Two part molding process for fabrication of bellows. a) Three modules of PEG-filled
PDMS molding sheets with punched holes were used in various combinations to rapidly
form any desired number of convolutions, and then PEG forms acted as b) a sacrificial
template for Parylene C coating. Image from [100] © 2012 IOP. ........................................ 58
Figure 38. Photographs of the sidewall of a) PDMS molding sheet and b) Parylene C-coated
bellows template prior to sacrificial PEG removal. Image from [100] © 2012 IOP. ........... 58
xvi

Figure 39. Load-deflection testing of the bellows. Nitrogen supply was regulated to obtain
discrete pressures and a compound microscope (100x objective, 1 m vertical resolution)
was used to measure deflection. Image from [100] © 2012 IOP. ......................................... 60
Figure 40. Bellows integrated with MEMS electrochemical actuators were mounted in a
reservoir for flow rate testing. Image from [100] © 2012 IOP. ............................................ 61
Figure 41. Finite element model simulation of (a) deflection and (b) von Mises stress for bellows
with 1, 2, and 3 convolutions, but all other parameters constant (6 mm ID, 9 mm OD, 0.4
mm H, 13.5 m wall thickness). Image from [100] © 2012 IOP. ........................................ 62
Figure 42. (a) Hysteresis of the bellows (6 mm ID, 9 mm OD, 0.4 mm H, 3 convolutions, 13.5
m wall thickness) upon unloading was observed after load cycling the bellows up to 4.19
kPa (0.6 psi). (b) Above 4.19 kPa, dimpling (arrows) occurred at the outer edges of the
convolutions and plastic deformation was observed. Image from [100] © 2012 IOP.......... 63
Figure 43. Photographs of bellows (6 mm ID, 9 mm OD, 0.4 mm H, 13.5 m wall thickness)
with 1, 2, and 3 convolutions subjected to pressure just below burst pressure and exhibiting
plastic deformation. Image from [100] © 2012 IOP. ............................................................ 64
Figure 44. Mechanical testing of three identical bellows (6 mm ID, 9 mm OD, 0.4 mm H, 2
convolutions, 13.5 m wall thickness) demonstrating uniform performance during load-
deflection testing. Image from [100] © 2012 IOP. ............................................................... 65
Figure 45. Mechanical testing of two bellows designs, with inner diameters of 6 and 5 mm, but
all other parameters constant (9 mm OD, 0.4 mm H, 2 convolutions, 13.5 m wall
thickness). Image from [100] © 2012 IOP. .......................................................................... 66
Figure 46. Mechanical testing of two bellows designs, with inner diameters of 5 and 7 mm, but
all other parameters constant (10 mm OD, 0.4 mm H, 2 convolutions, 13.5 m wall
thickness). Image from [100] © 2012 IOP. .......................................................................... 66
xvii

Figure 47. Mechanical testing of two bellows designs, with outer diameters of 9 and 10 mm, but
all other parameters constant (5 mm ID, 0.4 mm H, 2 convolutions, 13.5 m wall
thickness). Image from [100] © 2012 IOP. .......................................................................... 67
Figure 48. Mechanical testing of two bellows designs, with wall thicknesses of 13.5 and 15.5
m, but all other parameters constant (6 mm ID, 9 mm OD, 0.4 mm H, 2 convolutions).
Image from [100] © 2012 IOP. ............................................................................................. 68
Figure 49. Mechanical testing of two bellows designs, with layer heights of 0.3 and 0.4 mm, but
all other parameters constant (6 mm ID, 9 mm OD, 2 convolutions, 13.5 m wall
thickness). Image from [100] © 2012 IOP. .......................................................................... 68
Figure 50. Mechanical testing of three bellows designs with 1, 2, and 3 convolutions, but all
other parameters constant (6 mm ID, 9 mm OD, 0.4 mm H, 13.5 m wall thickness). Image
from [100] © 2012 IOP. ........................................................................................................ 69
Figure 51. Flow rates (Mean  SE, n = 3) generated by electrochemical actuators with bellows of
two different designs (5 mm ID, 10 mm OD, 0.4 mm H, 2 convolutions, 13.5 m wall
thickness; 6 mm ID, 9 mm OD, 0.4 mm H, 3 convolutions, 13.5 m wall thickness). Image
from [100] © 2012 IOP. ........................................................................................................ 71
Figure 52. The redesigned packaging and configuration of the 3
rd
generation micropump system
in a) an exploded assembly view and b) a cross-sectional view. .......................................... 77
Figure 53. Three valve seating options were considered in the reservoir redesign. a) A simple
tunnel with port sized for the valve inlet. b) an external flange that would be mounted to the
back of the valve, and then inserted into a wall recess. c) a built-in flange over which the
valve would be seated. .......................................................................................................... 80
Figure 54. a) Schematic of inductive wireless power source, b) photograph of wireless benchtop
testing setup, and c) conceptual drawing of planned in vivo study setup. ............................ 82
Figure 55. The 3
rd
generation reservoir domes and bases were fabricated using high-resolution
stereolithography................................................................................................................... 83
xviii

Figure 56. Discs of medical grade silicone rubber act as septa in the refill port. A ledge built in to
the port prevents septa from occluding the lumen to the reservoir. ...................................... 84
Figure 57. Overview of micropump assembly. Biocompatible epoxy secures a) the valve to the
built-in flange, b) the valve sleeve, silicone connection, and catheter together, and c) the
bellows to the bottom of the dome. d) The EC actuator, base, and dome with attached
bellows e) come together with a marine epoxy seal. f) The fully assembled 3
rd
generation
wireless micropump system is then coated with Parylene C and Class VI silicone rubber. . 85
Figure 58. Valves were clamped temporarily in a custom acrylic fixture for screening prior to
integration with the system. The duckbill portion of the valve extends to the right into a hole
drilled according to the size recommendations for the valve sleeve. ................................... 86
Figure 59. The micropump reservoir is refilled using two ports, one for fluid introduction, and
the other for fluid extraction. ................................................................................................ 87
Figure 60. A reservoir dome with septa was mounted in a) a test fixture and subjected to several
stages of multiple punctures using a b) needle guide, followed by pressure testing until leaks
or up to 775 mmHg (15 psi). ................................................................................................. 88
Figure 61. The wireless micropump was subjected to back pressure values up to 20 mmHg. ..... 91
Figure 62. Variation in appearance of valve slits was not an indicator of future performance. ... 92
Figure 63. The micropump system generated flow rates for 21 days in simulated in vivo
conditions, 50% longer than the intended duration of in vivo studies. ................................. 94
Figure 64. Wireless micropump flow rate performance was not significantly affected by
increasing the environmental temperature from room (21 °C) to body temperature (37 °C).
............................................................................................................................................... 94
Figure 65. Flow rate deviation from baseline was less than a) 10% up to 6 cP in one micropump
and b) less than 10% up to 3 cP and less than 20% up to 6 cP in another micropump. ....... 95
xix

Figure 66. Flow rate performance between micropumps when the mean flow rates for each pump
were normalized to 1 was within 10% of baseline up to 3 cP. ............................................. 96
Figure 67. Mean flow rate for two separate micropumps varied less than 10% for most cases with
up to 20 mmHg back pressure applied against the catheter. ................................................. 97
Figure 68. Flow rate performance between micropumps was similar when the mean flow rates
are normalized to 1................................................................................................................ 98
Figure 69. Week 1 of benchtop testing of the wireless dosing regimen that will be followed for in
vivo studies. Mean flow rate for the seven days is shown for each micropump, except for
Pump 2, which was a mix of wired and wireless testing. ..................................................... 99
Figure 70. Week 2 of benchtop testing of the dosing regimen that will be followed for in vivo
studies. .................................................................................................................................. 99
Figure 71. Pump 15 was operated wirelessly and delivered daily doses of 30 µL for three
consecutive weeks on the benchtop. ................................................................................... 100
Figure 72. Increasing time to start of dosing was observed throughout the consecutive dosing
cycle. Micropump was operated wirelessly and delivered daily doses of 30 µL for three
consecutive weeks on the benchtop, with reservoir refill occurring at the beginning of each
week. ................................................................................................................................... 102

xx

Abstract

The manner in which a drug is delivered to the body plays a major role in its efficacy. There
are several implantable pump technologies for chronic drug administration in humans and large
animals. However, development of new drug therapies generally involves initial evaluation via
human disease models in small animals such as mice. The technology for chronic drug
administration in mice is currently limited to constant flow rates determined at the time of pump
manufacture. With the emerging demand for personalized medicine, there is a need for drug
administration technology at the early stages of drug development that offers flexibility in dosing
schedules. Such a technology would create new opportunities in drug research that is not
possible with currently available tools.
This dissertation describes the development of a wireless implantable micropump system for
mice that allows flexible dosing to be performed post-implantation. The first chapter begins with
a review of the current state-of-the-art drug administration technology for rodents and its
limitations. It then introduces the wireless micropump system, its advantages over current
technology, and an application in anti-cancer drug delivery that drove its design. The second
chapter elaborates the details of the specific needs of the anti-cancer drug and demonstrates a
wired micropump prototype that utilizes electrochemical actuation. The preliminary benchtop
tests and in vivo studies with the first generation system elucidate the need for better packaging,
valve control, and wireless power to enable chronic drug delivery. The third chapter presents the
second generation system, which addresses packaging concerns and valve selection and
evaluation. It concludes with a demonstration of the micropump with a wireless power source.
The fourth chapter presents the mechanical characterization of the bellows component of the
actuator and its effect on the overall system configuration and design. The fifth chapter brings
together the advancements from the first two generations into a fully integrated and wirelessly
powered implantable system. Characterization of this third generation system shows its ability to
perform the intended drug dosing schedule reliably in the environmental conditions that would
occur in vivo.
1

1 Drug Administration Technology for Rodents

1.1 Introduction
1.1.1 Motivation
Laboratory animal research plays a key role in the study and understanding of human
diseases, and is a critical step in the developmental pathway for novel drug therapies for humans.
Many transgenic disease models have been developed in rodents due to their short gestation
period, high fertility, ease of handling, and relatively low required maintenance [1]. Rats and
mice represent an estimated 95% of laboratory animals used in research studies [2], and offer a
relatively inexpensive way to screen for and investigate the most appropriate delivery of novel
therapeutics. Methods for administering novel drugs in rodent models have a critical role in the
success of drug studies, and thus are an important area of research. It is necessary to have drug
administration technologies suitable for both rodent models, as disease models developed in one
do not necessarily transfer to another.

1.1.2 Challenges and limitations of current technology
The efficacy of a given drug therapy is greatly impacted by location, timing, and dosage.
Ideally, clinicians and researchers want to have control over all of these variables. Drugs can be
administered to the body locally (to a specific site or tissue) or systemically. Local delivery can
reduce the total amount of drug required and potentially minimize side effects to tissues that are
not the target of the treatment, but can be challenging for some areas of the body. Depending on
the drug and intended application, dosing can be acute (defined in this work as < 3 days) or
chronic (> 3 days). Drug regimens can vary from constant levels of drug to periodic boluses to
more complex programmable doses.
The optimal combination of these factors depends on the type of drug and the application for
which the drug is intended. Although there are numerous ways to administer drugs in humans
and large animals, there remains a lack of suitable technologies for local, chronic, periodic
dosing to rodents, and in particular mice. Administration of the drug should occur with minimal
handling and restraint to minimize stress and without sedation or anesthesia that could interfere
2

with the drug. In addition, the administration technology should not be susceptible to damage or
removal by the animal.
Current drug delivery methods include oral administration, injection, or infusion via tethered
or implanted pumps. These methods have limitations, including inducing stress on the animals
and lack of control over location, timing, and dosage. Of these methods, implanted pumps would
offer freedom of movement (and thus minimize stress), but the only commercial pump available
for animals as small as mice provides continuous infusion at a flow rate set at the time of
manufacture. The magnitude of mouse-based researched is indicated by The Jackson Laboratory,
a nonprofit organization and major distributor of research mice, who reported 3 million mice
shipped in fiscal year 2012 to approximately 20,000 investigators or laboratories representing
over 900 institutions [3]. In order to advance and promote new understanding and treatments of
disease, new technologies for drug administration in mice are needed.

1.1.3 Approach
My approach is to design, fabricate, and demonstrate a wireless implantable drug delivery
system featuring a micro electro mechanical systems (MEMS) bellows electrochemical actuator
for site-specific, chronic controlled drug delivery in mice. The current design is based on the
requirements for subcutaneous delivery of anti-cancer drug to tumors in mice. The micropump is
intended to be implanted adjacent to the tumor site, circumventing the need for intravenous
administration that would present challenges for drug stability and specificity, as well as the risk
of systemic side effects. The refillable reservoir contains multiple doses of the anti-cancer drug
enabling chronic studies with less frequent handling of the animal. The wirelessly powered
micropump allows control over the dosage timing and amounts, even after implantation. This
platform will enable researchers to conduct customizable and more complex drug studies in mice
that are not possible with current technology.

1.2 Current Drug Administration Methods and Technologies
1.2.1 Acute Methods
1.2.1.1 Enteral administration
Enteral (digestive tract) administration routes are common because they are economical and
convenient (oral) or offer precise and accurate dosing (gastric gavage). But these routes are slow,
3

the drug must undergo metabolism by the liver, and enteral absorption is unpredictable. These
routes can be difficult in animals, as it is dependent on palatability for voluntary consumption or
requires restraint when using gastric gavage (tubing directly inserted into the stomach) (Figure 1)
[4]. Gavage can provide more precise and accurate dosing, but requires a skilled technician to
avoid physical trauma to the esophagus and stomach. Required restraint can induce significant
stress in the animals, which is undesirable both for ethical reasons and to avoid confounding
factors in drug studies. For drugs with a specific target location in the body, it is undesirable to
expose the entire system to the drug and its potential side effects.

Figure 1. Gastric gavage (feeding tube through the esophagus and into the stomach) is used for
precise and accurate dosing, but the animal must be restrained and the drug must overcome
metabolism and absorption barriers. (Images courtesy of American Association for Laboratory
Animal Science, from Working with the Laboratory Mouse, AALAS Learning Library)

1.2.1.2 Injection
In order to bypass absorption delays or for drugs that are not formulated for the oral route,
researchers often use injection. Injection is a convenient method for site-specific delivery but is
labor-intensive, only allows for bolus delivery, and requires frequent handling of the animal
subjects (Figure 2). Two examples of common injection sites are the tail vein and intraperitoneal
space, both of which require skilled technicians in order to avoid injury to the animal.
Intravenous injection (often through the tail vein as in Figure 2a) can have serious complications,
including injection trauma to the anatomy and potential fluid overload. The intraperitoneal route
(Figure 2b) avoids the problem of fluid overload because excess fluid in the subcutaneous space
is quickly cleared by the kidneys, but care must be taken to not accidentally puncture organs or
vasculature. It is well documented that frequent handling of animals induces stress that can
4

confound study results [6,7]. It is recommended for use only in acute applications, but some
applications require chronic infusion.

Figure 2. Injection is common for acute bolus drug administration, but requires restraint and is
not suitable for chronic applications. Images © 2013 Newcastle University.

1.2.2 Chronic Methods
There is evidence that administration of therapy at a particular time can increase
effectiveness, such as in cancer treatment during a specific phase of a tumor cell cycle or
administration of chronic pain medication at a certain time of day [8,9]. Due to the rat’s
relatively large size (~300 grams), there are several chronic drug infusion options available.
However, for the mouse, which can be as small as 25 to 30 grams, chronic drug administration
technologies are still quite limited.

1.2.2.1 Tethered infusion systems
For chronic dosing in animals both large and small, researchers use tethered infusion
systems, in which the animal is directly connected to an infusion pump with a tether (Figure 3).
The external syringe pump provides control over flow rates and volumes. The tether connects to
an external port, integrated into a jacket or harness worn by the animal, or a subcutaneously
implanted injection port. The port provides routing through the skin, often to a blood vessel.
Although use of an external pump can offer precise and accurate dosing, tethering is
impractical for extended chronic studies as the animals can become entangled or break the
tethers by scratching and chewing on them. Rodent jackets and harnesses can be difficult to
5

manage as rodents are often adept at removing them. These methods also require frequent
handling of laboratory animals and constant care to prevent infection and other complications. A
serious concern with tethering is the stress placed on the animal. Continuously-tethered systems
for large animals have even been banned in some European countries because untethered
systems were demonstrated to result in significantly less stress on the animals [5].

Figure 3. Commercial tethered infusion system for rodents. Image courtesy of SAI Infusion
Technologies.

1.2.2.2 Commercial implantable pumps
Implantable pumps offer an alternative to injections or tethered systems, but there are no
commercially available implantable pumps suitable for drug infusion in mice with control over
dosage and timing in chronic experiments. The requirements for these types of applications
include: refillability, small size, on/off operation, and low power consumption. For subcutaneous
implantation to be a minor procedure, the surgery should involve less than 10% of the animal’s
surface area [6]. A summary of specifications and features of commercial implantable pumps for
rodents is shown in Table 1.
6

Table 1. Commercial implantable pumps for infusion in rodents
ALZET
®
* [10]

Med-e-Cell
Infu-Disk™ [11]

Primetech iPrecio
®
[12]

Driving
mechanism
Osmotic Battery Peristaltic (with battery)
Sized for
implantation
in mice
Yes No No
Post-
implantation
flow rate
control
No No No
Battery-less
operation
Yes No No
One-way flow
regulation
No No Yes
Refillable or
reusable
No No Refillable
Reservoir
capacity
100 µL to 200 µL 4.9 mL to 5.1 mL 900 µL
Duration 1 to 6 weeks < 1 week 6 months (1.0 μL/h)
1 week (30.0 μL/h)
Flow rate 0.11 to 8.0 µL/h 0.03 to 1.0 mL/h 1.0 to 30 L/h
Weight 0.4 to 1.1 g Under 12 g (empty) 7.9 g
Size 0.6 cm to 0.7 cm diam.
1.5 to 3.0 cm length
1.6 in (4.1 cm) diam.
0.5 in (1.3 cm) height
38.7 mm width
19.2 mm length
9.7 mm height
*Image courtesy of Durect Corporation

The single-use osmotically-driven ALZET
®
pumps [10] are of a form factor suitable for
implantation in mice but provide only continuous infusion; the flow rate is predetermined by the
water permeability of the pump’s semipermeable membrane and is fixed at the time of
manufacture. Drug payload lasts from 1 to 6 weeks depending on the flow rate. If higher flow
rates are required, the pump duration may not be adequate for extended chronic studies. The non-
refillable Med-e-Cell Infu-Disk™ [11] is too large for implantation into mice (typically weigh
~25 to 30 g), as the smallest available reservoir is 5 mL and weighs ~12 g when empty. The
7

actuation is achieved with an electrochemical cell module that provides only continuous drug
delivery at a factory set flow rate. Even at the lowest flow rate, 0.03 mL/h, the maximum
delivery duration for the 5 mL reservoir pump is less than a week, making this pump unsuitable
for chronic studies lasting longer than 7 days. The Primetech iPrecio
®
pumps [12] are
programmable prior to implantation, but the regimen cannot be modified afterwards. In addition,
this battery-powered pump is too large for use in mice, is single-use only, and the flow rate is
limited to a maximum of 30 L/h.
A key feature that is not available in any commercial pump is the ability to control flow rate
after the pump has been implanted. In addition, only one of these pumps is small enough for
implantation in mice. In order to expand the capabilities of small animal research technology,
these features need to be realized in conjunction with advantages of current technology.

1.2.2.3 Micro electro mechanical systems (MEMS) for drug delivery
The issues of controllable delivery in a small platform can be addressed by employing micro
electro mechanical systems (MEMS) technology, which utilizes fabrication techniques borrowed
from the semiconductor industry to produce miniaturized structures, sensors, actuators, and
systems. Many MEMS pumps have been reported and are reviewed elsewhere [13–18], but few
of these devices feature an integrated reservoir and show feasibility for implantation in rodents.
MEMS-based micropumps are generally divided into dynamic (non-mechanical) and
displacement (mechanical) types. Non-mechanical micropumps typically are limited in flow rate
(maximum of 10 µL/min) [17], have relatively slow response compared to mechanical
micropumps, and often require interaction with a working solution with particular electrical
properties, such as conductivity (e.g. electrokinetic pumps) [14,16,17]. Examples of dynamic
actuation include electroosmosis, electrowetting, electrophoresis, electrohydrodynamics, and
magnetohydrodynamics.

1.2.2.3.1 Active MEMS pumps
Displacement actuation schemes offer distinct advantages for drug delivery applications in
rodents, including operation independent of the fluid properties (such as conductivity), higher
flow rates, and faster response times [17]. There are numerous displacement-type actuation
methods including electrostatic, piezoelectric, thermopneumatic, shape-memory alloy (SMA),
8

bimetallic, and electromagnetic, and ionic conductive polymer films (ICPF). Of these,
piezoelectric pumps are the most common.
Smits [19] described a three piezoelectric valve peristaltic pump designed for insulin
delivery. Maillefer et al. presented a piezoelectric ceramic disk-actuated micropump [20,21] that
includes a reservoir and battery intended for transcutaneous delivery in humans and weighs over
20 grams [22]. Geipel et al. developed a piezoelectrically-actuated microport system, but it has
only been worn externally by rats [23] (Figure 4a). Evans et al. [24] reported an intrathecal drug
delivery system with lead zirconate titanate (PZT) valves that consumes low power, but is
intended for human use (Figure 4b).Other piezoelectric micropumps have been reported and
comparisons between them can be found in [13–17]. Piezoelectric pumps offer high forces and
fast response times, but disadvantages include high voltage requirements and mounting
procedures of PZT disks. Electrostatic pumps were introduced and demonstrated by Judy et al.
[25] and Zengerle et al. [26] starting in the early 1990’s. Although electrostatic actuation allows
fast response times and low power consumption, only small stroke volumes and forces are
achieved [17]. In thermopneumatic and phase-change type actuators, large pressures and
deflection are achieved, but at the cost of high power consumption, slow response times, and the
added fabrication difficulty of a fluid-filled chamber that must be sealed. SMA micropumps
feature high deflection, but require high power and special materials. Bimetallic micropumps
feature relatively simple fabrication and lower driving voltage, but deflection is small and
operation at high frequency is limited [16]. Electromagnetic actuation offers faster mechanical
response than thermopneumatic [15] and large forces, but suffers from high power consumption,
heat generation, and requirement of an external magnet [15,17]. In ICPF pumps, applied voltage
to metal electrodes deposited on either side of a perfluorosulfonic acid polymer membrane
induces bi-directional bending used in pumping. Although there is concern of repeatability issues
related to batch fabrication, this scheme boasts low driving voltage, fast response,
biocompatibility, and operation in aqueous environments [16,17].
9

Figure 4. Examples of active MEMS pumps [5] include the a) piezoelectric-based active
microport system worn externally by rats [23] (reprinted with kind permission from Springer
Science and Business Media) and b) dual drug delivery system intended for humans [24] © 2011
IEEE.

1.2.2.3.2 Passive MEMS drug delivery systems
In addition to micropumps, there are several passive MEMS drug delivery systems consisting
of membrane-sealed microreservoir arrays for controlled release [27–31]. Although
microreservoir technology has been demonstrated in acute studies in rats (~150 to 200 g) [29]
(Figure 5a), there are still some limitations in terms of chronic drug delivery in rodents.
Microreservoirs allow for storage and release of one or multiple drugs, but the reservoirs are not
refillable. Thus, there must be a compromise between duration of the implant and size of the
implant to minimize surgeries for removal and/or replacement of the device. In the case of [29],
membrane dissolution is sensitive to and must be modified for in vivo environmental conditions,
as protein adsorption affects the gold surface of the membrane and impedes corrosion [29]. The
microreservoir seals required a cleaning cycle and at least 10 minutes of a square wave cycle for
the membranes to open consistently. Lastly, delivery from microreservoirs occurs via diffusion,
so delivery rates to the tissue are not adjustable and could vary based on the drug formulation
and in vivo environment.
An implantable microbolus pump was developed at the University of Southern California for
use in neuroimaging studies in rats and mice [32–35]. First, the animal was intravenously
injected with a radiotracer. When the animal engaged in the behavior of interest the pump’s
electrothermal valve was externally triggered to release a euthanasia agent, thus capturing the
brain’s state for functional brain imaging analysis. Operation was demonstrated in vivo in mice
[35] (Figure 5b), but dosing occurred via diffusion with the aid of hydraulic pressure provided by
10

the elastomeric reservoir [33]. Due to the nature of the study, the device was designed for single
use only.

Figure 5. Examples of passive MEMS devices include a) microreservoir arrays demonstrated in
rats (reprinted from [29] with permission from Elsevier) and a b) microbolus pump demonstrated
in rats and mice [35] © 2009 IEEE.

1.3 Development of a Wireless Implantable MEMS Micropump System
Despite abundant work in micropumps and drug delivery, there is still an unmet need for site-
specific, controllable infusion for chronic studies in small animals, with a particular lack of
delivery technologies for animals as small as mice. My approach is to design, fabricate, and
demonstrate a wireless implantable drug delivery system featuring a MEMS bellows
electrochemical actuator (BEA) for controlled, site-specific chronic drug delivery in mice.

1.3.1 System overview
The system is composed of a bellows electrochemical actuator (BEA), packaging, wireless
power, and valve components. The block diagram in Figure 6a shows an overview of the drug
delivery system components, Figure 6b a 3-D model of the system, and Figure 6c the in vivo
wireless drug delivery setup.
This work is part of an ongoing effort in the University of Southern California Biomedical
Microsystems Laboratory to develop new drug administration technologies. Prior efforts
contributed to the design of the electrochemical actuator and motivation for inclusion of a
bellows, as well as a one-way valve candidate. Wireless power components were also recently
developed in the group and were integrated into the system to eliminate the need for a battery.
This work focuses on bellows and packaging design, selection and evaluation of valves, and
integration of all components into a wirelessly-powered refillable implantable system (Figure 6).
11

Figure 6. Wireless implantable MEMS micropump system. a) Bellows electrochemical actuator
(BEA), packaging, wireless power, and check valve components, b) 3-D model of system and c)
in vivo testing setup with wireless power source.

The BEA serves as the actuation component of the system. It consists of a pair of
interdigitated electrodes for electrolysis-based gas generation enclosed within a flexible bellows
structure, which separates the electrolysis reaction from the drug surrounding the bellows. The
packaging includes the drug reservoir and its encapsulation materials, the reservoir refill ports,
and the catheter. Wireless power requires an implanted receiver coil and circuit, and an external
(not implanted) transmitter coil, circuit, and power supply for the transmitter. The valve, which
prevents mixing of biological fluids and the drug contents of the reservoir, and its packaging are
the final components of the system.

12

1.3.1.1 System operation
The BEA consists of a pair of interdigitated platinum (Pt) electrodes contained within a
bellows filled with electrolyte (water). Electrical current applied to the electrodes splits the water
molecules into hydrogen and oxygen gases in a process known as electrolysis. The bellows
expands under the imposed pneumatic driving force of the gases generated by the electrolysis
reaction, which then pushes the fluid (drug) surrounding the bellows out of the reservoir and
catheter to the delivery site (Figure 7b). When the current is turned off, the Pt acts as a catalyst
for the gases to recombine as water [36] (Figure 7a). The BEA can be set to on/off states and
provide adjustable flow rate (through applied current magnitude) to achieve a desired delivery
regimen. This electrolysis-based actuation method allows multiple dosing cycles, which
combined with a refillable reservoir, makes the micropump system appropriate for chronic use.

Figure 7. Electrolysis-based operation of the MEMS micropump system. a) With the system off
(no applied current), a one-way valve prevents biological fluids from mixing with drug contained
in the reservoir. b) With applied current, water electrolysis produces hydrogen and oxygen gases
that extend the bellows and drive surrounding drug out of the reservoir and catheter. c) After
refill of the drug reservoir, dosing cycles continue as described in a) and b).

1.3.1.2 System advantages
The system’s advantages include refillability, small size, on/off operation, and low power
consumption. The system is intended for subcutaneous implantation adjacent to the target site
13

and delivers drug directly to the tissue (Figure 6c). The system’s built-in refillable drug reservoir
can be sized to deliver adequate dosages of the drug and minimize refills without exceeding the
appropriate size for implantation in mice. Electrochemical (EC) actuation is desirable for
implantable drug delivery systems because it consumes low power (~ W to mW), generates low
heat, creates a large driving force, and allows flow rate control from nL/min to L/min through
the adjustment of applied electrical current [37,38]. The BEA is powered wirelessly to avoid use
of a battery, which would add significant mass to the implanted system and for chronic studies
limit implant lifetime.

1.3.2 Electrolysis actuation
The system’s actuation mechanism is based on water electrolysis; applying an electric
potential across a pair of electrodes submerged in electrolyte (water) results in the dissociation of
water into hydrogen and oxygen gas. Other common actuation methods employed in MEMS,
such as piezoelectric, electrostatic, and thermopneumatic, require large amounts of power or
generate heat that is undesirable for implantable devices. Additionally, in contrast to
piezoelectric and electrostatic actuation, electrolysis does not subject the fluid to high frequency
mechanical pumping which could damage the drug.
Oxygen gas, cations, and electrons are produced at the anode (1), after which cations
(protons) pass through the electrolyte (water) to the cathode and combine with electrons in the
external circuit to form hydrogen gas (2).

E° (25°C) = -1.23V (1)

E° (25°C) = 0.00V (2)

The net reaction converts a 3:2 stoichiometric ratio of gas to liquid and produces hydrogen at a
rate twice that of oxygen (3), the limiting step of the electrolysis reaction [39].

E° (25°C) = -1.23V (3)

A significant increase in volume induced by the liquid to gas phase change provides a large
mechanical force that can be utilized for actuation. When constant current (as opposed to
14

voltage) is applied, gas generation rates are also constant [40]. When the current is turned off the
gases recombine into water. There are no side reactions in water electrolysis, and the reaction is
reversible and repeatable [41,42].

1.3.3 Previous work
1.3.3.1 Development of electrochemical actuators and devices
Platinum (Pt) was chosen as the electrode material due to its biocompatibility, role as a
catalyst in recombination, and its resistance to oxidation and corrosion [43]. Pt and a titanium
adhesion layer were patterned on a glass substrate using a dual-layer photolithography and liftoff
process as described in [36,41]. The soda lime wafer was treated with hexamethyldisilazane
(HMDS) for improved adhesion between the wafer and photoresist. The dual-layer photoresist
led to formation of an undercut on the photoresist sidewall that assisted with metal liftoff and
better defined electrode features (see APPENDIX A: Electrode Fabrication for process flow).
After liftoff the electrodes were coated with Nafion
®
,
a biocompatible solid polymer electrolyte,

which provided increased and more efficient gas generation during electrolysis [41]. The
Nafion
®

coating also reduced high current-induced delamination of and electrodeposition on the
electrodes [41]. The interdigitated layout of the electrode reduced the resistive path through the
electrolyte, which translated to improved efficiency and lower heat generation [44]. Based on
electrode geometry studies for maximizing efficiency of the EC actuator [36] and improved
robustness with increased finger width [41], the interdigitated fingers were patterned 100 m
wide and spaced 100 m apart. The final EC actuator has a diameter of ~8 mm plus traces for
electrical connections (Figure 8), and achieved flow rates of L/min that are linearly related to
applied constant currents between 0.25 and 13 mA. When the electrolysis reaction is confined in
a chamber, the phase-change induced pressure increase acts as a driving force to expel fluid from
the reservoir (Figure 8b).

15

Figure 8. a) The electrochemical (EC) actuator consists of Nafion
®
-coated interdigitated
platinum (Pt) electrodes with 100 m finger width and 100 m spacing. b) Electrical current
applied to the electrodes dissociates the water into hydrogen and oxygen gases, which then
expels fluid from the chamber.

Additional tests on the current EC actuator design (coated with Nafion
®
) have shown that
neither orientation (right side up, 90°, or 180°) nor operating temperature (20 °C versus 37 °C)
result in significantly different generated flow rates [41]. Thus, performance of the EC actuator
when it is implanted in mice with the integrated system should be affected by neither mouse
movement nor body temperature.
A microfabricated polymer reservoir system for intraocular drug delivery applications [45]
was previously demonstrated. Multiple punctures with a 30-gauge beveled-tip (non-coring)
needle were performed in vitro to investigate refillability of the device reservoir through a
silicone rubber membrane, which led to the use of medical grade silicone for reservoir prototypes
and refill port septa in current devices. These studies also led to rounded reservoir designs to
minimize dead volume [46] and prevent skin irritation or erosion when implanted.
The next milestone in the prototype system development was to provide electronic control
over dosing. Electrolysis was demonstrated in [37] (Figure 9) for active control over dosing flow
rates from pL/min to L/min, which were relevant for ocular therapy. Testing pump performance
against physiologically relevant backpressures up to 9.33 kPa (70 mmHg) showed that sufficient
flow rates could be achieved even under abnormal ocular pressure conditions. Continuous and
bolus operation modes were carried out and target volumes of 250 nL were delivered. The
refillable reservoir was packaged for acute surgical studies and preliminary experiments were
16

conducted ex vivo in porcine eyes. Although electrolysis was successfully demonstrated in a
packaged system, an important limitation also became apparent. The drug solution was
intermixed with the electrolyte (water) and was oxidized [47] during electrolysis. This concern
led to the next important milestone for the early generations of electrolysis-based drug delivery
devices: a means for separating the electrolysis reaction from the drug solution.

Figure 9. Intraocular drug delivery device with refillable silicone rubber reservoir. Reprinted
from [37] with permission from Elsevier.

1.3.3.2 Introduction of bellows electrochemical actuator
Li, et al. introduced the first bellows electrochemical actuator (BEA) [36] (Figure 10). The
bellows served to isolate the electrolysis reaction in a separate chamber from the drug solution,
preventing undesirable pH changes or oxidation that could affect the drug. A fabrication process
utilizing a sacrificial lost-wax like process was developed for generating bellows made of
biocompatible Parylene C. Mechanical characterization and simulations showed that the bellows
achieved large deflections relative to flat and corrugated diaphragms. The integrated EC actuator
was operated under low power and feasibility of using a wireless power source was discussed.

Figure 10. Side profile of the bellows structure (left) and a bellows mounted on an
electrochemical actuator (right). The bellows isolates the electrolysis reaction from the drug
reservoir. Adapted from [36] © 2009 IEEE.

17

1.4 Example applications for chronic periodic dosing
It is well established in the drug delivery literature that both dosage and timing are critical to
the study of drug effects on the body (pharmacodynamic responses) [6,8,9], as drug function is
often tied to biological rhythms. Chronotherapeutics entails drug therapy that is tuned to
biological rhythms which play a role in disease activity and thus impact drug efficacy [9]. A few
examples of diseases or conditions that may be affected by treatment with control over dosage
and timing are cancer, chronic pain, and diabetes. Cancer studies have shown that administration
of chemotherapy at a particular time in a tumor cell cycle can increase effectiveness while
reducing toxicity to normal cells [8,9]. For chronic pain, the time of day when the drug is
administered may be critical for successful treatment [8,9]. In the case of diabetes, control over
both basal insulin levels as well as carefully timed boluses at mealtimes is necessary to mimic
natural levels [8].
Recently, researchers at the USC Keck School of Medicine and SUNY Buffalo have
developed a drug for treating head and neck squamous cell carcinoma (HNSCC). The drug is
based on small interfering ribonucleic acid (siRNA) that silences a gene that confers radiation
resistance in certain types of tumors. In vitro and in vivo studies have demonstrated the potential
to greatly reduce or even eliminate the need for radiation therapy in HNSCC treatment [48], but
a new method for administering the drug is needed to be able to expand the studies with more
precision and control over dosage timing and location in animals, as well as demonstrate future
prospects in clinical settings. This work leverages MEMS technology and past achievements in
BEA development to address the delivery needs of this promising anti-cancer therapeutic.

18

2 Application of System for Anti-Cancer Drug
Delivery

2.1 Motivation
Head and neck cancer is the sixth most common cancer in the world, with an estimated
incidence of approximately 650,000 [49]. In 2012 in the United States, there will be an estimated
52,610 new cases of head and neck cancer and 11,500 deaths [50]. Approximately 90% of head
and neck cancers are diagnosed as squamous cell carcinoma. Early stage detection and treatment
of head and neck squamous cell carcinoma (HNSCC) can lead to cure rates of 70-90% [51].
However, two-thirds of patients present with advanced stage HNSCC, for which recurrence rates
are as high as 55% [52]. The 5-year survival rate across all stages is only 40-50% [48].
Treatment typically involves surgery to remove cancerous tissues, followed by radiation and in
some cases chemotherapy. Radiotherapy is used to initiate intrinsic apoptotic (cell death)
pathways, but radiation is non-specific and affects non-cancerous tissues as well. Mucositis
(mouth sores), dysphagia (difficulty swallowing), and speech disruption are some of the common
side effects contributing to morbidity of high-dose radiotherapy in HNSCC treatment [51,52].
New therapies are needed that can treat advanced-stage and recurrent HNSCC and mitigate
or eliminate associated treatment side effects. A novel drug for treating HNSCC has been
developed by collaborators at the USC Keck School of Medicine and SUNY Buffalo. The drug
was based on small interfering ribonucleic acid (siRNA) that silences a gene that confers
radiation resistance in certain types of tumors, with the potential to greatly reduce or even
eliminate the need for radiation therapy [48]. Human tumor xenografts were implanted into a
nude mouse model and treated with a novel anti-cancer drug injected at the tumor site. Gold
nanorods were developed to help the drug enter the cells once the drug has arrived at the target
site, but new technology is needed to deliver siRNA to the target site.
My approach is to implant the wireless implantable MEMS micropump adjacent to the
tumor. The system provides site-specific delivery of the drug directly to the tumor,
circumventing the need for intravenous administration that would present challenges for drug
stability and specificity, as well as the risk of systemic side effects. The refillable reservoir can
store multiple doses of the anti-cancer drug enabling chronic studies with less frequent handling
of the animal. The wirelessly powered micropump allows control over the dosage timing and
19

amounts, even after the micropump has been implanted. The micropump system provides the
platform for further investigation of this novel siRNA-based therapeutic, which could lead to
improved cancer therapy and quality of life for patients both during and after treatment.

2.2 Background
2.2.1 Role of sphingosine-kinase-1 in head and neck squamous cell carcinoma
The anti-cancer drug is based on siRNA that blocks production of sphingosine-kinase-1
(SphK1), a protein that plays a major role in regulating cell proliferation, apoptosis, and
radiosensitivity. Apoptosis and proliferation in cells are regulated by sphingolipids, including
ceramide, sphingosine, and sphingosine-1-phosphate (S1P). Ceramide and sphingosine promote
apoptosis and inhibit proliferation. S1P promotes cell survival and proliferation [53,54], and has
also been classified as a regulator of angiogenesis [55,56], adhesion, and metastasis [56–58].
According to “sphingosine rheostat” theory, sphingosine kinases (SphK) regulate sphingolipid
metabolism, and thus control whether the cell is in an apoptotic or proliferative state [59–62]. Of
the two SphK enzymes, SphK1 is better understood with regards to carcinogenesis [63]. It is
upregulated in many human cancers [64], and evidence has been shown that SphK1 is related to
radiosensitivity in prostate cancer [65–67]. Recently, HNSCC tumors were also shown to have
upregulated SphK1 expression, particularly in advanced stage and recurrent tumors, and
exhibited radiation resistance [48,64]. If HNSCC tumors were more sensitive to radiation, then
there would be great potential for reducing tumor recurrence and radiation-associated morbidity
[64].

2.2.2 siRNA and the ribonucleic acid interference pathway
The siRNA-based anti-cancer drug utilizes cellular mechanisms of the ribonucleic acid
interference (RNAi) pathway. RNAi is a pathway in eukaryotic cells in which siRNA cleaves
messenger RNA (mRNA) with a complementary sequence to silence genes post-translation [68].
There are numerous applications for gene silencing technology, but siRNA has a short in vivo
half-life (~sec to min) [69] and difficulties entering the target cell. Techniques for introducing
siRNA into the cell via carriers is an intensely active area of research for developing genetic-
based therapies for cancer, viruses, inflammation, and other diseases [70]. A carrier (or vector)
should protect the siRNA from extracellular enzymes, facilitate uptake of siRNA into the cells
20

via endocytosis, and allow siRNA to be released once inside the cell [71]. Ideally, the carrier
itself does not elicit an immune response and helps avoid uptake by non-target cells [72].
Carriers include viral vectors [73], lipids and peptides [68,72,74], cationic polymers
[68,70,72,74], and inorganic nanomaterials [48,72]. Carriers may facilitate systemic
(intravenous) delivery by 1) protecting siRNA until it reaches the target site (with surface
modifications such as polyethylene glycol coatings) and 2) by targeting a specific site using
ligands. Examples of carriers designed to act as ligands include glycosylated molecules,
peptides, proteins, antibodies, and engineered antibody fragments [72]. With these chemical
modifications the half-life of siRNA may be extended to minutes or days, but at the cost of
decreased potency [69]. Some success has been found with these targeted systems, even entering
clinical trials, but each drug and carrier molecule combination requires significant laboratory and
clinical testing to ensure safety and efficacy. As an alternative to systemic delivery, injection has
been used in vivo for site-specific administration of siRNA-based compounds.

2.2.3 Blocking production of SphK1 using siRNA
Recently, siRNA targeting of SphK1 (SphK1siRNA) in radiation resistant tumor cells
resulted in promising outcomes in several in vitro and in vivo studies. SphK1siRNA delivered to
HNSCC cell lines via lipid-mediated transfection inhibited SphK1, resulting in reduced tumor
cell proliferation and increased radiosensitivity in vitro (Figure 11) [64]. The highest reduction in
tumor cell viability in vitro occurred with lipid-mediated SphK1siRNA transfection followed by
low-dose radiation (Figure 11c, far right column pair).

Figure 11. In vitro results for two HNSCC cell lines after treatment with radiation only,
SphK1siRNA only, and combined SphK1siRNA plus radiation. a) Cells were exposed to
radiation levels ranging from 0 to 50 Gray (Gy). b) Cells were transfected with 0 to 200 nM of
SphK1siRNA using a lipid-based transfection agent. c) Comparison of cells treated with low-
21

dose radiation (2.5 Gy) only, the lipid-based transfection agent only (as a control), an
intermediate amount of SphK1siRNA (100 nM), and SphK1siRNA (100 nM) plus subsequent
low-dose (2.5 Gy) radiation [64]. Copyright © 2010 Wiley Periodicals, Inc.

In the next part of the study, tumor cells were transfected with SphK1siRNA in vitro and
then the pre-treated cells were injected into Balb/C athymic (nude) mice. Tumor growth in vivo
(Figure 12) was only slightly reduced for the tumor cells treated with control GFPsiRNA (non-
specific siRNA) plus low-dose radiation, but significantly reduced for tumor cells pre-treated
with SphK1siRNA alone. An even greater reduction of tumor volume was achieved with
combined SphK1siRNA and low-dose radiation treatment. Reduction of Ki-67, a marker for
active cellular proliferation, and an increase in Caspase 3, a marker for apoptosis, provided
further evidence of increased apoptosis with SphK1 inhibition.

Figure 12. In vivo results for pre-treated tumor cells injected into nude mice. Tumor volumes
were tracked over a period of 18 days for GFPsiRNA control and SphK1siRNA, with subsets of
each group (n=6) irradiated with 1.0 Gy every 3 days [64]. Copyright © 2010 Wiley Periodicals,
Inc.

However, the preliminary study was small and treatment of the tumor cells prior to
implantation did not represent a true therapeutic situation. The lipid-based transfection method
used in [64] was undesirable for direct treatment in vivo due to pulmonary toxicity concerns [75]
and the nature of the transfection process. In a follow-up study, a new formulation of the
22

SphK1siRNA utilized a gold nanorod (GNR) carrier to form GNR-SphK1siRNA nanoplexes. In
vitro studies were conducted to evaluate the new formulation with GNR. For in vivo studies, the
nanoplexes were not used to pre-treat the tumor cells, but rather administered post-implantation
to the tumors via local injection.
GNRs were favored as carriers as they are resistant to physiological degradation and have
shown biocompatibility [71,76–79]. In addition, GNR distribution post-delivery could be
monitored through inherent longitudinal surface plasmon resonance properties [80,81]. The use
of synthetic nanoparticles avoided potential safety concerns of virus-based carriers. The
nanoparticle surface was made cationic (+) to allow for electrostatic complexation with anionic
(-) SphK1siRNA in a stable but reversible fashion. The positively-charged GNR bound to the
cellular membrane to facilitate endocytosis (uptake into the cell) [48]. The electrostatic bond
between the GNR and SphK1siRNA was weak, such that the nanoplexes were decoupled at the
interior cellular membrane layer or by charged cellular components within the cytoplasm,
resulting in release of SphK1siRNA to the cytoplasm. A complex called Dicer then cleaved the
double-stranded SphK1siRNA into single sense and antisense strands. The antisense strand was
incorporated by the RNA-induced silencing complex (RISC), an innate compound of the RNAi
pathway [68], which then cleaved the mRNA that codes for SphK1. A simplified overview of the
siRNA pathway is shown in Figure 13.
In vitro studies by Masood et al.demonstrated that GNR-SphK1siRNA nanoplexes and GNR
alone, but not free siRNA, were taken up by the cell. Similar to results seen in [64], GNR-
SphK1siRNA treatment plus irradiation showed the greatest reduction of viable cells as
compared to controls. HNSCC cells (untreated) were injected into the flank of nude mice and
grown for four days prior to treatment [48] in order to more closely recapitulate the therapeutic
situation. Treatment (PBS, GNR-GFPsiRNA, or GNR-SphK1siRNA) was then administered
post-implantation at three proximal sites around the tumor three times per week. A subset of each
group was irradiated every three days with 1.0 Gy, and tumor volume was measured (Figure 14).

23

Figure 13. The positively-charged gold nanorods (GNR) bind to the cell membrane, facilitating
uptake via endocytosis. An innate endonuclease called Dicer separates SphK1siRNA into single
sense and antisense strands. The antisense strand is incorporated by RNA-induced silencing
complex (RISC; also innate), and then binds to the complementary sequence of the messenger
RNA (mRNA) sense strand. The mRNA is then cleaved, which blocks production of the protein
SphK1 which contributes to radiation resistance.

In vivo, delivering SphK1siRNA blocked the protein’s production and facilitated treatment of
the tumor with lower doses of ionizing radiation (up to 5x less than clinical levels) in mice [48].
In addition, there was no statistically significant difference in SphK1 expression between the
irradiated and non-irradiated SphK1siRNA groups in vivo (data not shown here), suggesting that
the treatment with the novel drug may even lead to reduced or eliminated radiotherapy.
24

Figure 14. a) HNSCC cells were subjected in vitro to no treatment (NT), control GNR, control
GNR-GFPsiRNA (non-specific siRNA), and GNR-SphK1siRNA. A subset of each group was
irradiated with 1.0 Gy four hours after treatment. Tumor cell viability was significantly (p <0.05)
lower for GNR-SphK1siRNA nanoplex-treated groups compared to controls. b) HNSCC tumor
xenografts were grown in nude mice, then treated with PBS, GNR-GFPsiRNA, and GNR-
SphK1siRNA (n=6 per group, experiment repeated twice). Treatments were administered via
local injection to three sites proximal to the tumor three times per week. A subset of each group
was irradiated with 1.0 Gy every 3 days. GNR-SphK1siRNA plus irradiation resulted in the
greatest reduction in tumor volume at 25 days. Adapted from [48] with permission of The Royal
Society of Chemistry.

2.2.4 Limitations of GNR-SphK1siRNA nanoplex delivery
Chemical modifications that facilitate systemic delivery of the GNR-siRNA nanoplexes do
not yet exist, so preliminary in vivo studies have thus far been limited to manual intratumoral
injection. For studies with large numbers of animals, which is necessary to confirm results with
statistical significance, manual injection is tedious and imprecise. Injection is a simple method
for site-specific delivery but only allows for bolus delivery and requires frequent handling of the
animal subjects. The GNR-siRNA nanoplexes require a finite time for adsorption, thus injection
is not the optimal delivery paradigm for studying the GNR-siRNA nanoplex function. As
discussed earlier, frequent handling of animals induces stress that can confound study results
[6,7].
Significant research hurdles remain for development and adequate testing of a systemic
formulation of GNR-SphK1siRNA, and these targeted systems could potentially have side
effects to normal tissues encountered prior to arrival at the site. For systemic delivery, the carrier
25

and its modifications must overcome serum nuclease degradation, cellular uptake, and rapid
renal clearance [75]. As mentioned previously, the extensive chemical modifications that enable
systemic delivery can decrease potency of the drug [69].
Having a means of delivering GNR-SphK1siRNA directly to the site of action without
delays, systemic side effects, and need for additional chemical modifications could provide
greater efficacy and safety. The nanoplexes have successfully shown suitability for entering the
cell and releasing siRNA within the cell, but a new method is needed to provide efficacious
delivery of siRNA to the tumor site for chronic, periodic dosing studies in the mouse model. The
goal of my work is to provide the technology to satisfy this unmet need in GNR-siRNA nanoplex
delivery and thereby further development of this promising new therapeutic.

2.3 Preliminary In Vivo Testing of Prototype Wired System
Prototype wired systems consisting of a BEA packaged in silicone rubber (no valve) were
implanted into nude mice in a preliminary study to evaluate the subcutaneous delivery of GNR-
siRNA nanoplex delivery [82]. HT29 (human colon cancer) cells were implanted subcutaneously
in both sides of the flank of nude mice to induce tumors, with the left tumor receiving treatment
and the right acting as a control with no treatment. The system was implanted adjacent to the
tumor and delivered GNR-siRNA nanoplexes directly to the tumor site (Figure 15).

26

Figure 15. a) Conceptual depiction of placement of implanted micropump to deliver anti-cancer
drug (nanoplexes) to tumors. b) In vivo delivery with wired prototype micropump powered with
a battery pack.

Wires exiting the pump were routed underneath the skin and exited the skin behind the
mouse’s head. A custom external battery pack was temporarily connected to the pump wires
during delivery periods to supply power to the pump. With a flow rate of ~2.5 L/min a bolus of
~50 L was delivered daily for two weeks, with reservoir refills twice per week. Passive pumps
(silicone reservoir only, no BEA) were also implanted and delivered control treatments including
phosphate buffered saline (PBS), gold nanorods (GNR), and free siRNA. Without the actuator
the drug moved by diffusion from the reservoir, through the catheter, and finally to the tissue.
Injections of PBS, free siRNA, and GNR-siRNA nanoplexes were also conducted at the same
frequency and bolus volume as the active pump. A subset of the treated tumors were exposed to
1.0 Gy (100 rad) of radiation. Active delivery of the GNR-siRNA nanoplexes in combination
27

with radiation treatment showed significant tumor regression (~50%) over both diffusion and
injection, with the greatest effect occurring adjacent to the catheter tip [82] (Figure 16).

Figure 16. Photographs comparing size of the tumors after delivery from active pumps and
passive pumps (diffusion-based, no actuator). Adapted from [83] © 2010 IEEE.

Although this study showed a positive outcome, there were several caveats. Only 1-2 animals
were used per test group, therefore a follow-up study with a larger number of subjects would be
needed to confirm these results. The current supplied by the battery pack was approximately
~0.78 mA, but may have not have been consistent for the entire duration of the study,
particularly at the end when the batteries were partially drained. Upon explantation, the
electrodes, which were fabricated on Parylene substrate instead of glass, showed signs of
delamination. Therefore, the actuator may have been damaged and the exact flow rate of each
delivery period may have varied. The preliminary results of active delivery of GNR-siRNA
nanoplexes were promising, but significant modifications to the system were needed.

2.4 1
st
Generation MEMS Micropump System
2.4.1 Design
BEAs for the 1
st
generation MEMS micropump system were made with a glass substrate
(instead of Parylene) and Nafion
®
coating to improve performance and prevent delamination
[41]. The BEAs were then integrated into reservoirs and powered through wired connections for
benchtop testing. The diameter of the reservoir (~21 mm) was chosen based on the dimensions of
28

the BEA footprint. The height of the reservoir (~10 mm) balanced the expected extension height
of the bellows during electrolysis and the space limitation in vivo. The reservoir included built-in
rounded suture bulbs for anchoring to tissue in vivo.

2.4.2 Fabrication
The prototype reservoirs for the in vivo study were made with laser-cut molds that resulted in
shallow reservoir lids and limited bellows movement and thus displacement volume. These lids
had to be elevated from the pump base with a spacer ring of silicone to allow height within the
reservoir for the bellows to expand. Sealing these multiple reservoir parts complicated the
integration process. For the 1
st
generation system, silicone rubber casting in custom computer
numerical control (CNC) milled acrylic produced molds with more accurate vertical dimension
control and eliminated the need for a spacer, thus simplifying integration (Figure 18).

Figure 17. Assembly of 1
st
generation wired micropump system. a) A CNC mill was used to
make acrylic molds for casting silicone reservoirs. b) The bellows electrochemical actuator
(BEA) is inserted into the cast silicone rubber reservoir, then additional uncured silicone rubber
is used to seal the slot and adhere the lid to the reservoir. c) An active micropump (top) included
the wired BEA, while a passive micropump (bottom) consisted of the reservoir shell only. 17b
Reprinted from [82] with kind permission from Springer Science and Business Media.

Polydimethylsiloxane (PDMS), or silicone rubber, of medical grade (MDX-4 4210 or A-103;
Factor II, Lakeside, AZ) was mixed (Super Mixer AR-250, Thinky Corp., Tokyo, Japan) in a
10:1 base:curing agent ratio and degassed under vacuum (Model V0-914A Lindberg Blue,
Asheville, NC) for 45 minutes. The uncured solution was poured into molds and baked in an
29

oven (EC0A Environmental Chamber, Sun Electronic Systems Inc., Titusville, FL) for 1 hour at
80 °C to cure. USP Class VI silicone tubing (VWR International, Radnor, PA) for the catheter
was set in the mold during curing, as integration post-cure was difficult. Additional MDX-4 4210
was later added near the tip of the catheter to provide a suture flap during in vivo studies.
Kynar™ wire-wrap wires (30 AWG, Jameco Electronics, Belmont, CA) were affixed to the
electrodes using conductive epoxy (EPO-TEK
®
H20E, Epoxy Technology, Billerica, MA) and
further coated with nonconductive marine epoxy (Loctite; Henkel Corp., Rocky Hill, CT) for
insulation [42]. Bellows were filled with electrolyte (water) and assembled onto electrodes using
double-sided pressure sensitive adhesive film (3M™ Double Coated Tape 415, 3M, St. Paul,
MN) and reinforced with marine epoxy to form the BEA. The BEA was inserted through a slit in
the back of the reservoir, then additional PDMS was used to adhere the BEA to the inner base of
the reservoir and to seal the slit and lid. The fully packaged device was cured for approximately
45 minutes at 80 °C. The refillable reservoirs were then filled with ~1 mL of deionized water
using a 30-gauge non-coring needle as previously demonstrated in [84].

2.4.3 Methods
The custom battery pack used for the preliminary in vivo study did not provide precise and
accurate current for the entire duration of the study. For 1
st
generation benchtop testing, a high
quality current source (2400 Sourcemeter, Keithley Instruments Inc., Cleveland, OH) supplied
constant current of 2.0 mA for 15 minutes to the 1
st
generation wired micropump system (Figure
18). The catheter outlet was connected to a 100 µL calibrated micropipette (VWR International,
Radnor, PA) and flow rates were calculated according to the equation below.

Current was turned off for 45 minutes to allow recombination to occur and the bellows to
deflate. This on/off cycle was repeated four times.
30

Figure 18. Flow rate testing setup for 1
st
generation wired micropump system.

2.4.4 Benchtop testing of 1
st
generation micropump system
The system produced a flow rate of 4.72 ± 0.35 L/min (mean ± SE, n=4) and consumed low
power of approximately 3 mW (Figure 19). A delay of approximately 5 minutes occurred prior to
positive fluid flow, which was predominately due to the highly compliant nature of the silicone.
Without a valve and with the end of the catheter (or micropipette) exposed to air or an uncovered
container of liquid, the fluid moved along the catheter in reverse towards the reservoir during
recombination. However, the fluid front did not fully return to the original starting location,
which was also an effect of the compliant silicone packaging.

Figure 19. Flow rate results for the 1
st
generation micropump system. Constant current was
applied for 15 minutes, with a flow rate in the linear region of 4.72 ± 0.35 L/min, then turned
off for 45 minutes to allow the gases to recombine. Reprinted from [82] with kind permission
from Springer Science and Business Media.

31

2.4.5 In vivo evaluation
Several non-valved 1
st
generation micropump systems were implanted into nude mice. These
studies provided insight into realistic dimension restrictions for the implanted system, techniques
for suturing the catheter adjacent to the tumor, quantitative input for refill port size (in order to
feel beneath the skin), and the absolute need for wireless delivery. Wires, even kept short at the
skin surface and behind the mouse’s head to avoid chewing and scratching, were not robust
enough to endure an in vivo study. The soft silicone packaging posed concerns in active mice
that were constantly rubbing up against the cage walls and were capable of extremely flexible
body movements (can bend the body into a u-shape), thus potentially stressing the devices and
posing a risk of accidental dosing. Presence of biological fluids was evident up to ~1 cm in the
catheter, indicating potential mixing of biological fluids with the drug in the reservoir.

2.4.6 Benchtop testing of valves
A check valve was needed in the system to prevent biological fluids from mixing with
reservoir contents during recombination and between dosing periods. The valve would also
facilitate reservoir refill by allowing introduction of the desired drug refill volume exclusively
from the refill port and blocking unintended introduction of fluids from the catheter outlet. The
refill procedure will be discussed in a later chapter. The integrated system was tested on the
benchtop without a valve, with the MEMS in-line check valve (packaged in a fluorinated
ethylene propylene cannula) [85], and with two commercial valves, shown in Figure 20. The
commercial valves were chosen based on biocompatibility and manufacturer specified opening
(cracking) and sealing pressures.

Figure 20. Commercial one-way disc valve (left), commercial in-line check valve (middle), and
MEMS in-line check valve (right). Reprinted from [82] with kind permission from Springer
Science and Business Media.

32

The first commercial valve (240270524 one-way disc valve, Halkey Roberts, St. Petersburg,
FL), constructed of Class VI biocompatible polycarbonate housing and silicone seal disc, was
chosen for its low opening and sealing pressures (<1.72 kPa or 0.25 psi and ≤3.45 kPa or 0.50
psi, respectively, according to the manufacturer). The second, a normally closed in-line check
valve (80031, Qosina, Edgewood, NY), was made of acrylic and ethylene propylene diene
monomer and had an even lower manufacturer specified opening pressure of <0.35 kPa or 0.05
psi. The MEMS valve and two commercial valves opening pressures were evaluated on the
benchtop (Figure 21) and results are shown in Table 2.

Figure 21. Stand-alone valve testing setup to evaluate opening pressure of a MEMS valve and
two commercial valves.

Table 2. Comparison of valve opening pressures
Valve Opening Pressure
(kPa / mmHg / psi)
MEMS check valve 1.31 / 9.82 / 0.19
One-way disc valve 0.69 / 5.17 / < 0.10
In-line check valve 0.35 / 2.59 / < 0.05
Reprinted from [82] with kind permission from Springer Science and Business Media.

33

2.4.7 Benchtop testing of valved system
A cannula identical to that used for the MEMS valve packaging was integrated into the
micropump reservoir. The commercial valve inlets were connected to the end of this cannula and
the outlets to a calibrated micropipette (Accu-Fill 90, Becton, Dickinson, and Co., NJ). Constant
current of 1.0 mA was applied (2400 Sourcemeter, Keithley Instruments Inc., Cleveland, OH) for
15 minutes to the actuators then turned off for 45 minutes to allow recombination to occur and
the bellows to deflate (Figure 22). This on/off cycle was repeated four times in each non-valved
and valved system configuration. The duration of current application was calculated based on
total dose requirements and desired flow rate for the application of GNR-siRNA nanoplex
delivery.

Figure 22. Flow rate measurement setup for 1
st
generation wired system in a silicone reservoir
with a MEMS valve integrated into the cannula. Commercial valves were tested similarly, but
with the inlet connected to an identical cannula without the integrated MEMS valve.

None of the valves fully prevented backflow during the recombination period (Figure 23).
The valves caused slightly longer (~sec) delays to forward flow, but did not prevent backflow
during recombination. Added flow path resistance reduced flow rates to 4.67 L/min (one-way
disc valve), 4.49 L/min (in-line check valve), and 4.28 L/min (MEMS valve). The system was
capable of achieving the opening pressure for all valves tested, but the slow rate of
recombination did not provide enough force to reach the necessary valve sealing pressures.
34

Figure 23. Performance of non-valved and valved silicone reservoir systems. Reprinted from
[82] with kind permission from Springer Science and Business Media.

2.5 Challenges for Next Generation Wireless System Integration
2.5.1 Packaging
For 1
st
generation wired system prototypes, packaging was made of medical grade silicone.
Although biocompatible, the compliant material was causing undesirable effects on performance
of the system in flow rate testing and was not suitable for precise metering. The compliant
reservoir could also lead to accidental dosing. The next generation system for in vivo studies
needed a new material for the reservoir and packaging that was both biocompatible and rigid,
and had low permeability. A robust and quick method for fabrication of packaging was needed,
as the animal studies would require a minimum of four devices for each test group: one GNR-
siRNA nanoplex and three control solution groups. The 2
nd
generation packaging will be
discussed in Chapter 3, while further improvements for the 3
rd
generation will be discussed in
Chapter 5.

2.5.2 One-way valve
The MEMS and commercial valves did not perform in a manner appropriate for the current
micropump system. The commercial valves were also quite large relative to the micropump. Part
of Chapter 3 describes evaluation of a miniature commercial valve found later that provided
significantly better performance as a one-way valve to minimize reverse leakage during
recombination and mixing of biological fluids with the drug. Another important consideration
35

with the addition of valving is refill. With a valve, a two-port reservoir design is needed in the
reservoir for refill and flushing, and was incorporated into later system reservoir designs.

2.5.3 Bellows electrochemical actuator
Initial studies on the EC actuator and BEA have demonstrated the capability to generate
L/min flow rates specified for the application while consuming low power. However, the
dimensions of the 1
st
generation device were already pushing the limit for tolerable volume for
implantation in mice. In addition, the 1
st
generation micropump required multiple refills per
week during in vivo studies due to the limited displacement capacity of the bellows. The bellows
dimensions are the primary factor in determining the final reservoir dimensions, and thus must
be chosen carefully to minimize device footprint while still meeting the dosing regimen
requirements for daily dose volume and total dosing volume between refills. The bellows
dimensions and relative performance will be discussed Chapter 4, and the application of this
knowledge will be addressed with the 3
rd
generation packaging redesign in Chapter 5.

2.5.4 Wireless power
The preliminary in vivo study with the 1
st
generation wired micropump system results
showed that wired operation was not feasible for chronic studies. An inductive wireless power
source was developed in the Biomedical Microsystems Laboratory and will be discussed in later
chapters.
36

3 2
nd
Generation Micropump System

3.1 2
nd
Generation Goals
This chapter will focus primarily on addressing two of the challenges presented at the end of
Chapter 2: evaluating micropump performance with a rigid reservoir, and further investigation to
find a passive mechanical check valve suitable for low flow rate (several µL/min) drug delivery.

3.2 Packaging
The 1
st
generation micropump system was housed in a medical grade silicone rubber
reservoir. Although these reservoirs were simple to cast, sealing of the reservoirs was difficult,
and the soft material’s compliance led to issues in flow rate testing. When implanted, the soft
reservoir could be compressed during normal mouse movement and activity and also cause
unintended dosing. In addition, silicone rubber is highly permeable to moisture and gases, and
thus does not provide an adequate barrier between the reservoir contents and external bodily
fluids for chronic studies. To address these issues, the 2
nd
generation reservoir was made of a
rigid, low-permeability medical grade polymer. The packaging was fabricated by injection
molding, so that highly consistent reservoirs could be made rapidly, and seams connecting the
parts could be more easily sealed.

3.2.1 Polymers for low-permeability packaging
Alternative materials for the rigid reservoir shell were evaluated based on the following
criteria: biocompatibility, low permeability, and performance after injection molding.
Water vapor transmission tests were conducted on four samples each of 1 mm-thick
polypropylene (20-melt PP; Chase Plastics), 1 mm-thick polyethylene terephthalate glycol
(PETG; Chase Plastics), 2 mm-thick silicone rubber (MDX-4 4210, Factor II, Lakeside, AZ), and
14 µm-thick Parylene C (Specialty Coating Systems, Indianapolis, IN). These materials are all
available as biocompatible USP Class VI materials. Each sample was sealed with marine epoxy
over the mouth of a 50 mL flask filled with 50 mL of double distilled deionized water and
allowed to cure for 24 hours. The mass of each sealed beaker was measured with a high precision
balance (0.1 mg readability) at the beginning and end of the 7-day testing period. In between
measurements the samples were stored in a closed container maintained at 20  0.4 C and 20 
37

3%RH, monitored with a humidity and temperature sensor. Indicating dessicant was added each
day to keep the relative humidity as constant as possible.
It was demonstrated that permeability of the packaging was reduced 17-fold or 27-fold using
injection molded PETG or PP, respectively. Permeability could be further reduced by coating the
injection molded plastic with a thin layer of Parylene C, which shows 1100 times less water
vapor transmission than silicone rubber. These results, shown in Figure 24, were in accordance
with published values found in the literature [86] (Table 3).

Figure 24. Water vapor transmission rates (WVTR; mean  SE, n = 4) for four polymers at 20
°C and 20 %RH.

Table 3. Summary of water vapor transmission rates for four polymers
Polymer Mean WVTR, n=4
(g mm/m
2
day)
Standard
Error
Massey Values*
(g mm/m
2
day)
Parylene C 0.036 0.0009 0.083,0.1
PP 1.45 0.14 0.12-0.59,100+ (<0.5 mm films)
PETG 2.35 0.11 1.5, 1.6
MDX-4 39.95 2.11 20.83 (silicone elastomer)
*Varies with testing conditions

PETG was preferred over PP because it is transparent, which simplified assembly and
debugging, and it was readily available in biocompatible formulation from commercial sources
38

in quantities appropriate for our application. Injection molded PETG could withstand drilling for
modifications such as adding a valve port, refill port, or cannula port. PETG samples were
soaked over two weeks in 1X phosphate buffered saline (PBS), the GNR-siRNA nanoplex
solution base, and no effects on clarity or mechanical integrity were observed.

3.2.2 Design modifications
The 2
nd
generation packaging consisted of a circular base and reservoir dome. The base had a
recessed slot for the actuator for more efficient use of vertical space and rounded edges to cover
the corners of the actuator’s glass substrate (Figure 25a). The base thickness is 1.1 mm to cover
the glass substrate of the EC actuator while providing mechanical robustness for handling. The
base included a small ledge to aid with adhesion of the dome (Figure 25b). The dome consisted
of a cylindrical reservoir with filleted edges to maintain a smooth profile for subsequent
encapsulation and potential implantation. Dimension details can be found in APPENDIX B: 2
nd

Generation Wireless Micropump System Specifications.

Figure 25. Injection molded reservoir design. a) Photo and c) conceptual drawing of base with
bellows electrochemical actuator (BEA) seated in slot.

3.2.3 Fabrication
Test molds for the 2
nd
generation reservoir dome and base parts were drawn in SolidWorks
(Dassault Systèmes SolidWorks Corp., Concord, MA) and aluminum molds were fabricated in
house with a computer numerical control (CNC) mill (MicroMill 2000 HD/LE; Microproto
Systems, Chandler, AZ). The appropriate parameters for injection molding of PP and PETG
(Table 4) were determined and several prototype reservoirs with 1
st
generation dimensions were
made with a benchtop plastic injector (AB-100; AB Machinery, Montreal, Canada). The PETG
pellets were dried thoroughly before injection molding to prevent cloudiness and brittleness,
39

however this was not necessary for PP. Refill ports of approximately 3 mm in diameter were
drilled into the injection molded parts and filled with medical grade silicone rubber (A-103,
Factor II, Lakeside, AX) and cured at 80 °C for 1 hour to form septa. A band of marine epoxy
was added at the base of the septa to reinforce the seal.

Table 4. Injection molding parameters for PP and PETG
Parameter Setting
Polypropylene (PP) Polyethylene terephthalate glycol
(PETG)
Mold temperature 70 °C (158 °F) 23 °C (73 °F)
Injection
temperature
204 °C (400°F) 254 °C (490 °F)
Injection pressure 482 kPa (70 psi) 620 kPa (90 psi)
Mold release None None
Pellet drying time Not necessary 4+ hours, 65 °C (149 °F)

3.3 Experimental Methods
3.3.1 Filling protocol
The device is (re)filled by inserting a 30-gauge beveled-tip (non-coring) needle into the refill
port septum. The device is held with the catheter pointing upwards so that air is pushed out
preferentially until the reservoir is completely filled with the drug solution. Alternatively, the
catheter can be submerged in the drug solution, and air extracted from the refill port. Generally
the prior method was more effective in removing air bubbles because the catheter was directly
adjacent to the base of the reservoir, while the refill port was offset from the wall.

3.3.2 Benchtop testing of dosing regimen
Prototype (wired) systems with rigid polymer packaging were assembled to test actuation
behavior for a 180 L total dose volume, the experimentally determined maximum volume that
the bellows design could safely displace. The same BEA was used with these rigid reservoir
systems as the 1
st
generation silicone reservoir systems discussed in Chapter 2. An integrated
0.020” (0.51 mm) inner diameter silicone catheter was connected to a calibrated micropipette
(100 L; VWR International, Radnor, PA) to quantify flow rate and volumes. 2 mA constant
current was applied to each system. 2 mA was chosen for convenience, but the flow rate is
40

linearly dependent with the applied current and can be adjusted according to the desired dosing
regimen [42].
The same systems were also used to test repeatability of dosing across several days of
consecutive dosing. 2 mA constant current was applied to each system for ~3 minutes to deliver
a 60 L dose once daily for three consecutive days, corresponding to a total dosed volume of 180
L. In between delivery periods the current was turned off to allow the gases in the bellows to
completely recombine.

3.3.3 Viscosity testing
Drug solutions could vary in viscosity depending on the type of base solution (PBS in the
case of GNR-siRNA nanoplexes) and concentration of the drug. A protocol for preparation of
model glucose solutions with varying viscosity and a testing setup were developed to evaluate
the actuation performance under this condition. Anhydrous D-glucose (VWR International,
Radnor, PA) was dissolved into deionized water in four different concentrations as shown in
Table 8 and viscosities were measured with a Cannon-Fenske Routine Viscometer. Viscosities
were chosen and verified based on work by Migliori, et al. [87].

Table 5. D-glucose solutions for modeling various viscosities
D-glucose concentration
in water (%, w/w)
Viscosity at 20 C
(cP, 1 cP = 1 mPa s)
0 1.00
5 1.36
10 1.58
20 2.13
40 6.21

The BEA was integrated into a rigid reservoir. The outlet cannula of the reservoir was
connected to a calibrated micropipette and flow rate calculated as discussed previously. The flow
rate was measured at four (0.5, 1, 2, and 5 mA) different applied constant current levels for each
of the five test solutions.

41

3.3.4 Back pressure testing
The BEA in a rigid reservoir was subjected to back pressure values up to ~1.03 kPa (~7.76
mmHg or 0.15 psi) [42], the approximate upper limit of human central venous pressure [88],
which is much higher than back pressures expected in subcutaneous delivery. 5 mA constant
current was applied for 1 minute four times at each back pressure value. Flow rates were
determined by measuring the distance traveled in a calibrated micropipette as described
previously. Back pressure was applied with a custom pressure setup controlled by a LabVIEW
interface as shown in Figure 26.

Figure 26. Setup for applying physiologically relevant back pressures during system operation.

3.3.5 Valve testing setup and integration with micropump
None of the valves evaluated for the 1
st
generation micropump system prevented reverse
leakage during recombination. The search for a valve with lower sealing pressures led to a
commercial miniature duckbill valve made of medical grade silicone (2.0 mm; Minivalve,
Cleveland, OH). The valve base was attached with epoxy around the end of a 0.030” (0.76 mm)
inner diameter rigid polyetheretherketone (PEEK; VWR International, Radnor, PA) cannula,
which was later substituted for the silicone catheter for integration with the system. A rigid
tubing segment (FEP; Zeus International, Orangeburg, SC) was used as a sleeve and for making
a connection to the calibrated micropipette. The PEEK cannula end opposite the valve was
42

connected to a pressure setup for determining opening (cracking) pressure. The cannula and
valve were reversed for back pressure (sealing) testing (Figure 27). Four trials were conducted in
each position. It should be noted that the lower limit of the pressure testing setup was 0.34 kPa
(2.6 mmHg).

Figure 27. Pressure testing setup for evaluation of a commercial one-way valve and photograph
(bottom left) of the commercial duckbill valve.

3.3.6 Benchtop testing of 2
nd
generation valved micropump system
3.3.6.1 Wired testing
After stand-alone valve testing, the silicone catheter was removed from the reservoir and
replaced with the PEEK cannula with attached duckbill for benchtop testing. Three wired
micropump systems were assembled with a valve and tested on the benchtop to demonstrate
performance with a periodic dosing regimen (Figure 28). Constant current of 0.5 mA was applied
(Keithley SourceMeter) to achieve a 30 L dose each day for three days. The daily dose volume
was previously 60 L, but drug concentration was altered so that only 30 L per day was
required. Flow rates were measured using a calibrated micropipette attached to the cannula
outlet. Then a valved micropump was integrated with the wireless power source.
43

Figure 28. Constant current of 0.5 mA was applied to three wired and valved micropumps on the
benchtop to demonstrate performance with a periodic dosing regimen.

3.3.6.2 Wireless testing
Wireless operation was demonstrated on the benchtop for a prototype micropump through a
Class E inductive power system. The wireless system was split into the transmitter (external)
components and the receiver (subcutaneously implanted) components. A 12V power supply unit
(PSU) powered the transmitter coil/PCB (printed circuit board), which would be seated
underneath the animal cage for in vivo studies. Energy was wirelessly transferred to the receiver
coil/PCB (Figure 29). Prototypes of the wireless components were assembled and tested with the
valved rigid reservoir system. For future in vivo studies, a Parylene C coating will insulate the
electronics and reduce permeability through the reservoir wall. A further coating of medical
grade silicone rubber will soften the packaging edges for reduced tissue irritation. The
transmitter coil required placement within 20 cm of the receiver coil to attain adequate power
transfer, thus the mouse’s range of movement within the standard laboratory cage would need to
be limited. This could easily be achieved by placing a plastic container as a “trap” over the
mouse.

44

Figure 29. a) Diagram of wireless operation with inductive power system. b) Photographs of the
wireless components with prototype micropump system. c) No more than 20 cm should separate
the external transmitter and subcutaneously implanted receiver to ensure adequate power
transfer.

3.4 Results
3.4.1 Valve characteristics
2.0 mm duckbill valves were tested independently of the micropump and were found to have
a opening pressure of less than 0.69 kPa (5.17 mmHg) and no reverse leakage was observed with
pressures applied to a reversed valve up to 1.72 kPa (12.93 mmHg).

3.4.2 Benchtop dosing regimen results
Current application for 11.5 minutes produced an average flow rate of 17.40 ± 0.55 L/min
(mean ± SE, n=6 pumps) and an average delivered volume of 183.11 ± 5.58 L (mean ± SE,
n=6). Delay to the start of visible dosing was < 1 second, confirming our hypothesis that the
delays to start of dosing seen previously in the 1
st
generation were related to the compliant
45

silicone reservoir material. The flow rate was highly consistent for all six systems as shown in
Figure 30.

Figure 30. 2 mA constant current was applied to six wired rigid reservoir systems (photo) for
11.5 minutes. Consistent flow rates 17.40 ± 0.55 L/min (mean ± SE, n=6) and repeatability of
total dosing volume 183.11 ± 5.58 L (mean ± SE, n=6) were demonstrated across six
micropump systems.

The systems also showed consistent performance across three days of daily dosing (Figure
31). However, because of recombination, consecutive boluses require compensation for the
previous dose(s) in order to deliver 60 L each time. Thus, Day 2 would effectively require a
120 L bolus to compensate for the first 60 L dose and deliver 60 L, and Day 3 would require
a 180 L to compensate for the first two 60 L doses and deliver 60 L. Without a valve, the
fluid front returns during recombination to the original starting point, which for benchtop testing
was within the micropipette. A valve would be required in vivo to prevent the fluid front from
reaching and potentially introducing biological fluids to the reservoir.
46

Figure 31. 2 mA constant current was applied for ~3 minutes per day for three consecutive days
to show repeatability in periodic dosing of 60 L.

3.4.3 Viscosity results
As shown in Figure 32, viscosities up to 6.21 cP of the pumped fluid had minimal effects on
flow rate performance. The standard error at a given current was less than 6% of the flow rate
across all viscosities, and less than 2% in the case of 0.5 mA and 5 mA.

Figure 32. Constant currents of 0.5, 1, 2, and 5 mA were applied to a bellows electrochemical
actuator (BEA) in a rigid reservoir. With glucose solutions ranging in viscosity from 1 to 6.21
cP, standard error of less than 6% across the flow rates was observed.

47

3.4.4 Back pressure results
The results (Figure 33) showed that the flow rates under the various back pressures were
not significantly different from the flow rate with no applied back pressure.

Figure 33. A bellows electrochemical actuator (BEA) in a rigid reservoir was operated against
physiologically relevant back pressures. 5 mA constant current was applied for 1 minutes at each
value (n=4). Adapted from [42] with kind permission from Springer Science and Business
Media.

3.4.5 Valved system operation results
3.4.5.1 Wired
The measured valve opening pressure was less than 5.17 mmHg (0.69 kPa). Valved and
wired micropump systems demonstrated consistent flow rates of 3 to 4 L/min across all doses.
With each subsequent dose, an increasing dosing duration corresponded to pressure build up
compensating for reduction in reservoir volume [89]. Once the reservoir was refilled, the curve
shifted back to the “Dose 1” position. Representative graphs are shown in Figure 34. Accurate
dosing was indicated across all devices with standard error between 3 and 10% of the flow rate
magnitude. Reverse leakage of the valve was ~1-2 L, or 3-9% of the dose volume, validating
the one-way capability of the valve in this periodic dosing application.
48

Figure 34. Periodic dosing in wired valved micropumps (2 of 3 micropumps shown).

3.4.5.2 Wireless
The valved wireless micropump system was operated on the benchtop. Power transfer
occurred when the receiver coil was within 20 cm of the transmitter coil. The valve opened
immediately to allow forward flow and flow rates on the order of L/min were observed (Figure
35). When current was removed to turn the system off, < 0.33 L reverse leakage through the
valve was observed over a 24-hour period.

Figure 35. Wireless power transfer was achieved on the 2
nd
generation micropump system when
it was within 20 cm of the transmitter coil, resulting in a forward flow rate on the order of
µL/min. When the transmitter was powered off, minimal reverse leakage was observed.

49

The 2
nd
generation micropump’s wireless system had one set flow rate, which was chosen at
the time of assembly. Dose volumes were determined by the duration of applied current. The
wireless circuit could be set to a lower current output to lower flow rate to values desired for
future in vivo studies.

3.5 Summary
The goals of the 2
nd
generation micropump system were to evaluate performance with a rigid
reservoir and find and evaluate a suitable one-way valve for the intended application. Benchtop
studies showed that use of a rigid reservoir eliminated the majority of the delay to the start of
drug dosing and produced more reliable flow rate behavior. In addition, back pressure and drug
viscosity showed no significant effects on micropump performance, further demonstrating the
robustness of flow rate performance. A miniature one-way valve was selected and evaluated with
the micropump. Minimal added flow resistance and minimal reverse leakage indicated that the
valve was suitable for the micropump’s intended operating range and application. In addition,
wireless on/off control with a generated flow rate on the order of µL/min was demonstrated for
the valved micropump system.

3.6 Remaining Challenges
Although the rigid reservoir improved dosing performance, design changes need to be
incorporated into the 3
rd
generation wireless micropump system packaging. The reservoir was
much larger than the bellows, and resulted in a large inaccessible (dead) volume. The design
discussed in Chapter 5 optimizes the system configuration, such that the overall implant size and
mass (with wireless power components) are no larger than that of the wired 1
st
generation
system. The valve connection was sufficient for benchtop studies, but would not have been
robust enough for in vivo use, thus integration of the valve with the reservoir needed to be
addressed. In order to find the most efficient reservoir dimensions, an understanding of the
bellows design parameters and their effects on deflection was necessary.

50

4 Rapid Fabrication and Characterization of
Parylene C Bellows for Large Deflection
Applications

4.1 Motivation for Bellows Design and Fabrication Study
Polymer bellows have a lower Young’s modulus, thus achieve higher deflection under lower
loads than more traditional metal and ceramic bellows. This characteristic gives rise to
interesting applications, but limited information exists on design parameters of polymer bellows
and how these parameters affect load-deflection (axial extension) performance. In addition,
fabrication methods for polymer bellows shown thus far pose difficulties for high throughput
fabrication. A rapid high-yield fabrication process and characterization tests were developed to
analyze how the different features of the bellows structure contribute to load-deflection
performance. Effects of varying five bellows design parameters were studied using mechanical
load-deflection testing and simulations with finite element models. This work presented the first
study of polymer bellows design parameters and their effects on axial extension. Utility of these
polymer bellows were then demonstrated in electrochemical actuators. The knowledge gained
from this study provided the understanding needed to balance requirements of the MEMS
micropump system’s dosing volumes and size constraints for the application of GNR-siRNA
nanoplex delivery in mice.

4.2 Background
4.2.1 Bellows fabrication methods and applications
A bellows is a thin-walled corrugated tube [90] typically used as a pressure-responsive
device (switch, gauge), flexible shaft coupling, or as a hermetic housing [91]. The bellows shape
was chosen over corrugated or flat membranes because it can achieve higher deflection with less
applied pressure ([36]; Figure 36). Miniature metal and ceramic bellows have been employed
extensively in many applications, but often require large driving pressures (>MPa) to produce
only modest deflections (10’s of m) due to the relatively high Young’s modulus of the materials
[92]. A polymer bellows typically has a Young’s modulus orders of magnitude less than those of
metal and ceramic bellows, thus they require lower pressure (and thus less power) to achieve
51

larger deflections than metal or ceramic bellows. Applications of polymer bellows include
electrostatic actuators [93], endoscopic pressure sensors [94], microfluidic channel connectors
[95], fuel cell reservoirs [96], piston actuators [97], pneumatic artificial rubber muscles [98],
bending pneumatic actuators [99], and electrochemical actuators [36,82].
A standard profile of a bellows fixed at one end is shown in Figure 36, where t is the wall
thickness, ID the inner diameter of the bellows, OD the outer diameter of the bellows, and H the
height of one layer. The number of convolutions will be referred to as N. As shown in Figure 36,
under applied loads, and assuming no twisting, bellows undergo axial extension (the focus of this
work), axial bending, or a combination thereof.

Figure 36. a) Standard profile of a bellows with design parameters labeled. b) Application of
load to a bellows results in axial extension, axial bending, or a combination thereof. c) Cross-
sections of flat and corrugated diaphragms (not to scale). Bellows can achieve higher deflection
than flat or corrugated diaphragms of similar dimensions. Image from [100] © 2012 IOP.

Microelectromechanical systems (MEMS) bellows made of non-polymer materials [92] have
been produced using surface micromachining in which alternating layers of structural and
sacrificial materials are deposited and patterned until the final stacked multi-convolution
structure is achieved. However, the layer-by-layer nature of the process is time consuming and
impractical for producing polymer bellows having more than a few convolutions. MEMS
polymer bellows generally have been fabricated using an alternate process in which a sacrificial
bellows template is first produced using a microfabricated mold. The template is then coated
with the desired polymer and the sacrificial material removed to release the polymer bellows
structure. Mold features from m to mm can be achieved by either subtractive or additive MEMS
52

processes. A brief survey of recent work in MEMS polymer bellows fabrication is given in Table
6 and is grouped according to the manner in which the template molds were produced.
Minami, et al. [93] used subtractive laser ablation for mold generation paired with metal
oblique evaporation and Parylene deposition to fabricate a composite metal-polymer bellows for
application in an electrostatic actuator. Although resolution of this subtractive process was
relatively high (23 m), the process required multiple intermediate sacrificial materials that were
not reusable and was time consuming due to the serial nature of producing the laser ablated
molds. In additive processes [36,82,95,96,101], molds are constructed in a bottom-up approach.
Luharuka, et al. [96] fabricated bellows for storage reservoirs in a fuel delivery system.
Stereolithography (SLA) negative molds achieved 50 m minimum features but subsequent
processing steps limited the final part’s minimum feature size to 0.5 mm. Wax impressions
formed with the SLA molds were coated with Parylene. The wax was then removed with boiling
borax and a brush-coat of PDMS was applied to the Parylene bellows for structural strength.
Boiling, if using a high melting temperature wax, could induce thermal stress on the Parylene
and contamination with borax is undesirable for in vivo applications. Feng, et al. [95] presented a
room temperature sacrificial wax molding technique for Parylene bellows, but it also required a
non-biocompatible solvent to remove the wax. Although sub- m resolution is possible, current
additive methods are limited by the use of expensive equipment, to select materials, or by low
throughput. Finally, the production of bellows-like structures directly by focused-ion-beam
chemical vapor deposition (FIBCVD) was reported with sub-µm feature sizes [101], but at the
cost of long exposure time and was limited to phenanthrene. Extrapolating to a 1 mm tall
structure, fabrication would require over 13 hours.
Of these methods, a sacrificial wax process is attractive as it does not require specialized
equipment and can be relatively low-cost. Our group previously developed a sacrificial molding
technique utilizing polyethylene glycol (PEG) for MEMS electrochemically-driven bellows
actuators for drug delivery. The actuator consisted of a Parylene bellows filled with electrolyte
(water) and attached to platinum interdigitated electrodes [36,83]. In simulations, the bellows
achieved 1.5 mm, while a corrugated diaphragm of similar dimensions achieved only 0.8 mm for
the same pressure [36]. When activated, the actuator pumped drug out of an adjacent reservoir.
PEG was selected as the template for its biocompatibility and ability to dissolve in water.
53

However, only a 1.5 convolution bellows was achieved, and the bellows fabrication process was
time intensive and had low yield.

Table 6. MEMS fabrication techniques for polymer bellows ([100] © 2012 IOP)

Fabrication Sacrificial
Material
Bellows
Material
Application Dimensions Operation Range Ref.
Mold Patterning via Subtractive Process
Excimer laser
ablation (PI
mold), oblique
evaporation (Au),
vapor deposition
(Parylene)
Polyimide,
brass
Parylene C-
gold-Parylene
C-gold
Electrostatic
actuator
1.2 mm diameter,
~900 m L, 23 m
thread pitch
NA [93]
Mold Patterning via Additive Process
Sacrificial wax
molding, vapor
deposition
Wax (not
specified)
Parylene C Microfluidic
channel
connectors
NA NA [95]
Stereolithography,
sacrificial wax
molding, vapor
deposition
Wax (not
specified)
Parylene C-
PDMS-Parylene
C
Fuel cell
delivery
systems
reservoir
~100 m total wall
thickness; 10 mm
OD x 12.7 L; 15.7
mm L
44.8 kPa (linear
design), 13.8 kPa
(rotary design)
[96]
Sacrificial PDMS
and wax molding;
vapor deposition
PDMS,
PEG
Parylene C Electrochem-
ical actuator
6 mm ID x 9 mm
OD; 10 m thick
up to ~12 kPa [36]
Reusable PDMS
and sacrificial
wax molding;
vapor deposition
PEG Parylene C Drug delivery
device actuator
6 mm ID x 9 mm
OD; 13.5 m thick
up to 3.45 kPa [82]
Reusable PDMS
and sacrificial
wax molding;
vapor deposition
PEG Parylene C Electrochem-
ical actuator
Varies with design,
see Table 3
up to 3.45 kPa;
higher for burst
testing
This
work
Direct Fabrication of Bellows via Additive Process
Focused-Ion-
Beam Chemical
Vapor Deposition
(FIBCVD)
None Phenanthrene Demonstrate
FIBCVD
fabrication
method
0.8 m pitch, 0.1
m thickness, 2.75
m OD, 6.1 m H
NA [101]
NA = information not available
54

The process required mold reinforcement and multiple curing stages for the molds, which
were single-use only and had to be remade for each new group of bellows. To foster new
applications with MEMS bellows, fabrication needs to progress from serial to parallel or batch
production, molds should be reusable or inexpensive, the mold release step should not induce
thermal stress in the polymer, and the intermediate sacrificial materials should not pose a
contamination concern when the bellows are integrated with microfluidic systems. For
biomedical applications, the materials must also be biocompatible.

4.2.2 Bellows deflection theory and modeling
Metal and ceramic bellows are typically treated as a series of stacked diaphragms (or plates)
[91], each of which behaves according to classical plate theory as they undergo small deflection
relative to the diaphragm thickness. The total bellows deflection is the sum of the individual
diaphragm deflections and can be approximated using linear analytical models. Thin polymer
bellows, which exhibit highly nonlinear behavior, violate the assumption of small or moderately
large deflection and are not adequately described by diaphragm or even thin diaphragm
(membrane) theory. Due to the complex geometry of the bellows and highly nonlinear behavior
of thin polymers, an analytical closed-form solution is impractical [102] and instead finite
element modeling (FEM) can be employed to model deflection and stress under various loads.
FEM has previously been used for characterization and modeling of complex polymer structures,
such as in [103,104].

4.3 Approach
My approach uses a low Young’s modulus, biocompatible, chemically inert polymer to
enable novel applications with MEMS actuators and systems. Using a polymer, rather than stiff
metal or ceramic materials, allows large deflection under low applied pressures. A biocompatible
material allows integration with implantable systems, and chemical inertness imparts
compatibility with a wide variety of drugs and solutions. Previous efforts in my group led to use
of a water-soluble polymer as a sacrificial molding material. I developed an improved sacrificial
lost wax-like process for more economical and high-throughput fabrication. The new process
reduced fabrication time (1 week down to 1 day), featured reusable molds, did not require
55

expensive equipment, achieved wax removal at room temperature, and achieved higher yields in
contrast to [36].
Finite element models and simulations paired with mechanical characterization of various
designs provided insight into the bellows design parameters and their relationship to load-
deflection performance. An example of the utility of bellows in isolating the electrolysis reaction
from pumped fluid was demonstrated in electrochemical (EC) actuators. Integration of
electrochemical actuators into microfluidic system brings electrolysis reaction in contact with
reservoir fluid, thus limiting fluids that can be pumped and applications in which EC actuators
can be used. The bellows overcome this limitation, opening up possibilities for EC actuators that
have not been possible until now, including biomedical applications such as siRNA delivery in
vivo.

4.4 Design and Fabrication
Poly(para-xylylene), commonly known as Parylene, was discovered in the 1940’s and
commercialized in the early 1970’s. Parylene is favored for many applications due to its low
permeability to moisture and gases and high chemical resistance (no known solvents at room
temperature). Parylene C (the chlorinated type) is preferred for biomedical applications due to its
additional characteristics of high purity, resistance to fungus, non-thrombogenicity, and
neutrality to bacteria (meets ISO 10993 and USP Class VI standards). Implantation studies up to
26 weeks have demonstrated biocompatibility, and just a few examples of long-term commercial
applications include cardiac assist devices, stents, cochlear and intraocular implants, and
neurostimulators [105]. It is also used frequently to insulate electronics or as coating for medical
surgical tools. Parylene C has a low Young’s modulus of 2.76 GPa (manufacturer) to 4.75 GPa
[106], which imparts significantly more flexibility than traditional metal and ceramic materials
used for miniature bellows.
Bellows were made with varying dimensions in order to evaluate the effect of design
parameters (wall thickness, inner diameter, outer diameter, layer height, number of convolutions)
on overall bellows performance. The inner and outer diameters were chosen to correspond with
the active area of the electrochemical actuator, in which bellows utility was demonstrated.
Bellows overall height was intentionally kept low to minimize profile. The bellows wall
thickness was varied in order to determine the appropriate thickness for robustness and to
56

evaluate the effect of thickness on deflection. Each bellows design is shown in Table 7 along
with its inner to outer diameter ratio. The naming convention used for designs was as follows:
ID-OD-H,N,t. Thus, 6-9-0.4, 2, 13.5 describes a bellows with inner diameter of 6 mm, outer
diameter of 9 mm, layer height of 0.4 mm, 2 convolutions, and a wall thickness of 13.5 m.

Table 7. Summary of the fabricated bellows designs ([100] © 2012 IOP)
Dimensions
a
(ID-OD-H)
Number of
Convolutions
ID/OD

Parylene-coated
PEG template
5-10-0.4 2 0.50

5-9-0.4 2 0.56

6-9-0.4 3 0.67

6-9-0.4 2 0.67

6-9-0.4 1 0.67

6-9-0.4 (15.5 m) 2 0.67

6-9-0.3 2 0.67

7-10-0.4 2 0.70

a
All wall thicknesses 13.5 m unless otherwise specified

Fabrication of the bellows consisted of a two part molding process (Figure 37, Appendix C:
Bellows Fabrication Protocol). First, a set of reusable PDMS (Sylgard 184; Dow Corning Corp.,
Midland, MI) sheets were made of a specified thickness (0.3 and 0.4 mm) using a custom frame
57

of brass shims (Precision Brand, Downers Grove, IL) mounted to a flat glass plate (Nanofilm,
Westlake Village, CA). Uncured PDMS was poured into the frame and excess removed with a
squeegee. After curing in an oven at 80 C for one hour, the PDMS sheet was cut into 15 x 15
mm squares. Perforations equal to the dimensions of the inner or outer diameters of the bellows
were made in the center of the PDMS squares using metal arch punches (C.S. Osborne & Co.,
Harrison, NJ). The punches were polished with ultrafine sandpaper to improve smoothness of the
cut in the PDMS molding sheets (Figure 38).
The sheets were visually aligned (horizontal alignment within ~50 m with use of a
microscope) and stacked as shown in Figure 37 to form three different modules. Alignment of
PDMS molding sheets for a dozen bellows required less than 1 hour. Glass slides served as the
base substrate for stacking. Polyester tape (8403; 3M, St. Paul, MN) placed below the PDMS
stack facilitated removal of bellows after Parylene coating. Middle and top modules included an
additional flat solid sheet of PDMS to facilitate transfer of the modules during stacking. The
number of convolutions was increased by adding additional middle modules.
The reusable modules were filled with molten (50 C) low molecular weight (M
n
1,000)
polyethylene glycol (PEG; Alfa Aesar, Ward Hill, MA). With lower molecular weight PEG
(1,000 instead of 14,000) smoother and less brittle replicas were obtained upon cooling than in
[36]. In addition, the lower melting temperature of the PEG 1,000 eliminated the need for
vacuuming and mold reinforcement as described in [36], which was used to prevent the molten
PEG from leaking between the stacked PDMS molding sheets. PEG 1,000 did not leak between
our PDMS sheets as long as the temperature was kept below 60 C. Removing the reinforcement
step saved time during the mold preparation steps and allowed us to reuse the molding sheets,
such that total fabrication time was reduced from 1 week to 1 day. Solidified PEG modules,
consisting of one or two layers each as shown in Figure 37, were stacked and fused by
moistening the opposing faces of the modules to create bellows templates in increments of 1
convolution.

58

Figure 37. Two part molding process for fabrication of bellows. a) Three modules of PEG-filled
PDMS molding sheets with punched holes were used in various combinations to rapidly form
any desired number of convolutions, and then PEG forms acted as b) a sacrificial template for
Parylene C coating. Image from [100] © 2012 IOP.

Bellows templates (1, 2, or 3 convolutions) were coated with 13.5 or 15.5 m of Parylene C
(PDS 2010; Specialty Coating Systems, Indianapolis, IN). After coating, a large perimeter
around the bellows was cut with a razorblade and the slides were soaked in room temperature
deionized water to remove the sacrificial PEG and release the Parylene C bellows.

Figure 38. Photographs of the sidewall of a) PDMS molding sheet and b) Parylene C-coated
bellows template prior to sacrificial PEG removal. Image from [100] © 2012 IOP.

The vapor deposition process involves neither solvents nor plasticizers and produces a thin,
transparent, and conformal coating that penetrates all exposed surfaces [105]. The deposition
process occurs at room temperature, so substrates are not subjected to thermal stress. Normal
coating rate is 3-5 m/hour and the coating thickness depends on the amount of dimer and
59

loading of the chamber. This new improved bellows fabrication process features reusable molds,
does not induce thermal stress in the bellows material, uses a sacrificial material that is available
in biocompatible formulations, achieves yields of up to 90%, and takes only one day.

4.5 Experimental Methods
4.5.1 Finite element models and simulations
Three-dimensional finite element models (FEM) were developed for nonlinear static
simulations (Solidworks Simulation 2010, Dassault Systèmes SolidWorks Corp., Concord, MA)
of three bellows designs: 1, 2, and 3 convolution bellows of dimensions 6-9-0.4, 13.5 m.
Quarter models were used given the geometric symmetry and to minimize processing time. The
large displacement formulation and direct sparse solver were used to account for the highly
nonlinear nature of the polymer bellows. Loads from 0.00 to 3.45 kPa (25.86 mmHg, 0.50 psi)
were applied and the resulting deflection and von Mises stress values were recorded. Material
properties and dimensions used for the FEM and analysis are shown in Table 8.

Table 8. Finite element model material properties and bellows dimensions ([100] © 2012 IOP)
Dimension Value
Wall thickness ( m) 13.5
Inner Diameter (mm) 6
Outer Diameter (mm) 9
Layer Height (mm) 0.4
Number of convolutions 1, 2, and 3
Material Properties Value
Young’s modulus (GPa) 2.76

Tensile strength (MPa) 68.9
Yield strength (MPa) 55.2
Poisson’s ratio 0.40
Density (g/cm
3
) 1.289

4.5.2 Mechanical characterization setup
Bellows were clamped in a custom acrylic test fixture connected to a custom pressure setup
(Figure 39). An electronic pressure regulator (900X; ControlAir Inc., Amherst, NH) controlled
60

using a LabVIEW (National Instruments, Austin, TX) interface regulated nitrogen supplied from
a pressurized nitrogen gas cylinder. Loads were applied at room temperature to the bellows
mounted in the fixture and deflection of the center of the top of the bellows was recorded using a
compound microscope (PSM-1000; Motic China Group Co., Xiamen, China) with a 100x
objective lens. The fine focus knob has a calibrated resolution of 1 m/division. The center of
the bellows top surface was brought into focus under no load, then refocused under loading to
determine the deflection in m based on the number of divisions. To minimize backlash, the
knob was continually adjusted in one direction only. Verification with a pressure calibrator
mounted at the test fixture outlet ensured that delays were minimal (<5 seconds) between the
pressure measured at the regulator and at the test fixture.

Figure 39. Load-deflection testing of the bellows. Nitrogen supply was regulated to obtain
discrete pressures and a compound microscope (100x objective, 1 m vertical resolution) was
used to measure deflection. Image from [100] © 2012 IOP.

The onset of plastic deformation was determined by load cycling to successively higher
pressures until hysteresis of the deflection curve was observed. Between each load cycle, the
bellows was left unloaded for approximately 15 minutes to observe relaxation. Load cycling was
previously used to characterize flat Parylene C membranes [106]. Burst pressure (ultimate tensile
strength) was evaluated by increasing pressure until the bellows burst or leaked. For mechanical
characterization in the elastic range, loads from 0.00 to 3.45 kPa (25.86 mmHg, 0.50 psi) were
applied in discrete steps of 0.69 kPa every two minutes. The upper limit of 3.45 kPa ensured that
the bellows were below the elastic limit and hysteresis was not observed. For integration with the
MEMS actuators, the bellows should operate within the elastic range for predictable and
repeatable performance.

61

4.5.3 Demonstration in an electrochemical actuator
Bellows were integrated with electrochemical actuators as described in [36,41] Platinum
electrodes with a titanium adhesion layer were fabricated on a glass substrate with a dual-layer
photolithography and liftoff process, after which they were coated with Nafion
®
[41]. Bellows
were filled with electrolyte (water) and assembled onto electrodes using double-sided pressure
sensitive adhesive film (3M™ Double Coated Tape 415, 3M, St. Paul, MN) and reinforced with
marine epoxy (Loctite; Henkel Corp., Rocky Hill, CT) to form bellows electrochemical actuators
(BEA). Kynar™ wire-wrap wires (30 AWG; Jameco Electronics, Belmont, CA) were attached to
the electrodes with silver epoxy (EPO-TEK

H20E, Epoxy Technology, Inc., Billerica, MA) and
further insulated with marine epoxy.
Acrylic reservoirs were custom machined and the BEA placed within for flow rate testing at
room temperature (Figure 40). Constant current of 2.0 or 5.0 mA was applied (2400
Sourcemeter; Keithley Instruments Inc., Cleveland, OH) to the wires of the BEA to induce
electrolysis within the bellows. The phase changed-induced pressure increase extended the
bellows and applied pressure to reservoir fluid surrounding the bellows to force fluid out of the
reservoir. Flow rates and volumes were measured by a calibrated micropipette (100 L; VWR
International, Radnor, PA) attached at the outlet. Note that in contrast to the pressure setup, in
which pressure was applied to a closed system and held at static pressure values, the integrated
reservoir system is open (via the cannula outlet) and the pressure values were dynamic.

Figure 40. Bellows integrated with MEMS electrochemical actuators were mounted in a
reservoir for flow rate testing. Image from [100] © 2012 IOP.

4.5.4 Statistical analysis
Each of the bellows design parameters (outer diameter, inner diameter, wall thickness, layer
height, number of convolutions) were subjected to a two-tailed t-test for two independent
62

samples with unequal variances. For load-deflection testing, each sample consisted of three
bellows of the same design tested three times. For flow rate testing, three measurements were
made for each bellows design at each applied current value.

4.6 Results
4.6.1 Finite element model(FEM) and simulations
The resulting curves for 3-D nonlinear static FEM deflection simulations (Figure 41) showed
that the yield stress of Parylene (55.2 MPa according to manufacturer and 59 MPa according to
[106]) was not exceeded at 3.45 kPa, the maximum pressure applied. The highest stresses were
observed at the 90-degree corners at the base of the bellows (Figure 41b). The 1 and 2
convolution bellows FEM simulations of deflection were slightly larger but on the same order of
magnitude as mechanical testing, but the 3 convolution was noticeably underestimated in the
simulation as compared to the mechanical results.

Figure 41. Finite element model simulation of (a) deflection and (b) von Mises stress for bellows
with 1, 2, and 3 convolutions, but all other parameters constant (6 mm ID, 9 mm OD, 0.4 mm H,
13.5 m wall thickness). Image from [100] © 2012 IOP.

63

4.6.2 Mechanical characterization
4.6.2.1 Elastic range and burst pressure
The upper limit of the elastic range was determined by load cycling one of the bellows (6-9-
0.4, 3, 13.5 m) until hysteresis was observed in the load-deflection curve (Figure 42). Cycling
from 0.00 to 1.38 kPa and 0.00 to 2.76 kPa (0.00 to 0.20 psi and 0.00 to 0.40 psi) showed
negligible hysteresis between loading and unloading. Noticeable hysteresis was observed upon
unloading during testing over the span of 0.00 to 4.19 kPa (0.00 to 0.60 psi). A waiting period
between cycles (on the order of ten minutes) allowed the bellows to return to the starting
unloaded position. A loading cycle up to 5.52 kPa (0.80 psi) resulted in significant dimpling at
the outer edge of the convolutions (Figure 42b), and plastic deformation was observed.

Figure 42. (a) Hysteresis of the bellows (6 mm ID, 9 mm OD, 0.4 mm H, 3 convolutions, 13.5
m wall thickness) upon unloading was observed after load cycling the bellows up to 4.19 kPa
(0.6 psi). (b) Above 4.19 kPa, dimpling (arrows) occurred at the outer edges of the convolutions
and plastic deformation was observed. Image from [100] © 2012 IOP.

Bellows were loaded (1 bellows per design) until bursting, which occurred at ~60 kPa for some
designs and over 100 kPa as shown in Table 9. The bellows designs stretched significantly and
were permanently deformed. Photographs of deformed bellows are shown in Figure 43.
64

Table 9. Summary of burst pressures for the various bellows designs ([100] © 2012 IOP)
Dimensions
a

ID-OD-H (mm)
Number of convolutions ID/OD Burst Pressure
(kPa)
5-10-0.4 2 0.50 83
5-9-0.4 2 0.56 96
6-10-0.4 2 0.60 61
6-9-0.4 3 0.67 >100
6-9-0.4 (15.5 m)
2 0.67 >100
6-9-0.4 2 0.67 >100
6-9-0.4 1 0.67 >100
6-9-0.3 2 0.67 >100
7-10-0.4 2 0.70 65
a
All wall thicknesses 13.5 m unless otherwise specified

Figure 43. Photographs of bellows (6 mm ID, 9 mm OD, 0.4 mm H, 13.5 m wall thickness)
with 1, 2, and 3 convolutions subjected to pressure just below burst pressure and exhibiting
plastic deformation. Image from [100] © 2012 IOP.

4.6.2.2 Repeated loading and uniformity between bellows of same design
Three bellows of the exact same design were each subjected to the same load-deflection test
three times. Individual bellows showed no signs of plastic deformation up to 3.45 kPa for the
three cycles. The load-deflection curves of three bellows of the same design were extremely
similar. Typically, the standard error for a group of bellows of the same design was 5-10% of the
deflection value. As shown in Figure 44, the standard error can be as low as 3% (at 3.45 kPa).
65

Figure 44. Mechanical testing of three identical bellows (6 mm ID, 9 mm OD, 0.4 mm H, 2
convolutions, 13.5 m wall thickness) demonstrating uniform performance during load-
deflection testing. Image from [100] © 2012 IOP.

4.6.2.3 Inner diameter
The effect of inner diameter was characterized by fabricating a pair of bellows designs in
which the inner diameter was decreased by 1 mm and a second pair that differed by 2 mm. All
other parameters for a given pair were kept constant. Decreasing the inner diameter from 6 mm
to 5 mm while keeping all other design parameters constant did not significantly alter the overall
load-deflection curve (Figure 45). The results were not significantly different (p > 0.05).

66

Figure 45. Mechanical testing of two bellows designs, with inner diameters of 6 and 5 mm, but
all other parameters constant (9 mm OD, 0.4 mm H, 2 convolutions, 13.5 m wall thickness).
Image from [100] © 2012 IOP.

Decreasing the inner diameter of the bellows from 7 mm to 5 mm, keeping all other design
parameters constant, resulted in increased deflection for all pressures tested (a difference of 581
m at 3.45 kPa) as shown in Figure 46. However, the difference between the two data sets was
significant only for 3.45 kPa (p < 0.05).

Figure 46. Mechanical testing of two bellows designs, with inner diameters of 5 and 7 mm, but
all other parameters constant (10 mm OD, 0.4 mm H, 2 convolutions, 13.5 m wall thickness).
Image from [100] © 2012 IOP.

67

4.6.2.4 Outer diameter
Increasing the outer diameter from 9 mm to 10 mm, keeping all other parameters constant,
resulted in a significant upward shift in the overall load-deflection curve (Figure 47; p < 0.05).
At 3.45 kPa, the largest applied pressure for testing with the elastic range, the difference in
deflection was 776 m.

Figure 47. Mechanical testing of two bellows designs, with outer diameters of 9 and 10 mm, but
all other parameters constant (5 mm ID, 0.4 mm H, 2 convolutions, 13.5 m wall thickness).
Image from [100] © 2012 IOP.

4.6.2.5 Wall thickness
A 2 m reduction in wall thickness resulted in a significant increase in the overall deflection
of a given bellows design (Figure 48; p < 0.05). At 3.45 kPa the deflections were 1218 and 1595
m (a difference of 377 m) for 15.5 and13.5 m wall thicknesses, respectively.

68

Figure 48. Mechanical testing of two bellows designs, with wall thicknesses of 13.5 and 15.5
m, but all other parameters constant (6 mm ID, 9 mm OD, 0.4 mm H, 2 convolutions). Image
from [100] © 2012 IOP.

4.6.2.6 Layer height
Increasing layer height from 0.3 mm to 0.4 mm, keeping all other parameters constant,
shifted the overall load-deflection curve at 3.45 kPa upwards from 1350 to 1595 m (245 m) as
shown in Figure 49 (p < 0.05).

Figure 49. Mechanical testing of two bellows designs, with layer heights of 0.3 and 0.4 mm, but
all other parameters constant (6 mm ID, 9 mm OD, 2 convolutions, 13.5 m wall thickness).
Image from [100] © 2012 IOP.

69

4.6.2.7 Number of convolutions
Increasing the number of convolutions from 1 to 2 and 2 to 3 resulted in a statistically
significant upward shift of the load-deflection curve (Figure 50; p < 0.05). The shift from 2 to 3
convolutions was noticeably greater (1595 to 2941 m at 3.45 kPa) than the shift from 1 to 2
convolutions (1080 to 1595 m at 3.45 kPa).

Figure 50. Mechanical testing of three bellows designs with 1, 2, and 3 convolutions, but all
other parameters constant (6 mm ID, 9 mm OD, 0.4 mm H, 13.5 m wall thickness). Image from
[100] © 2012 IOP.

4.6.2.8 Summary of load-deflection testing
The effects of varying the bellows design parameters are summarized in Table 10. Entries
with a * indicate statistical significance, as obtained from a t-test. Decreasing the inner-to-outer-
diameter (ID/OD) ratio resulted in an increase in overall deflection, but was more effective with
an increase in OD rather than a decrease in ID. Wall thickness, layer height, and convolution
number were varied (individually) and resulted in statistically significant changes in deflection.

70

Table 10. Summary of effects of individual bellows design parameters
on load-deflection performance ([100] © 2012 IOP)
Parameter Variation Effect on Deflection
at 3.45 kPa
Outer Diameter
5-9-0.4 vs 5-10-0.4 (2 convo, 13.5 m)
Increase
1 mm (11%)
*Increase
776 m (50%)
Inner Diameter
7-10-0.4 vs 5-10-0.4 (2 convo, 13.5 m)
Decrease
2 mm (29%)
Increase
581 m (33%)
Wall thickness
15.5 m vs 13.5 m (6-9-0.4, 2 convo)
Decrease
2 m (13%)
*Increase
377 m (31%)
Layer Height
6-9-0.3 vs 6-9-0.4 (2 convo, 13.5 m)
Increase
0.1 mm (33%)
*Increase
245 m (18%)
Convolution Number
1 vs 2 convo (6-9-0.4, 13.5 m)
Increase
1 to 2 (100%)
*Increase
515 m (48%)
Convolution Number
2 vs 3 convo (6-9-0.4, 13.5 m)
Increase
2 to 3 (50%)
*Increase
1345 m (84%)
*
indicates statistical significance (p < 0.05)

4.6.2.9 Integration with electrochemical actuators
It was shown previously that increasing the convolution number from 1 to 2 did not have a
significant effect on flow rate when bellows were integrated with electrochemical actuators [21].
Here, two different bellows designs with varying inner and outer diameters were integrated with
electrochemical actuators and tested in an acrylic reservoir. Flow rates generated by the two
bellows electrochemical actuators at 2 and 5 mA applied constant current were not significantly
different (Figure 51; p > 0.05).

71

Figure 51. Flow rates (Mean  SE, n = 3) generated by electrochemical actuators with bellows
of two different designs (5 mm ID, 10 mm OD, 0.4 mm H, 2 convolutions, 13.5 m wall
thickness; 6 mm ID, 9 mm OD, 0.4 mm H, 3 convolutions, 13.5 m wall thickness). Image from
[100] © 2012 IOP.

4.7 Discussion
Linear bellows equations and thin membrane approximations are not applicable for our
Parylene C bellows, but with finite element models and simulations, deflection was
approximated for several bellows designs. Load-deflection curves from the nonlinear FEM
simulations were comparable to the experimental results for 1 and 2 convolution bellows, but
underestimated the deflection of 3 convolution bellows. The yield strength was not exceeded in
the simulations of all three bellows designs, demonstrating operation within the elastic range for
applied loads up to 3.45 kPa (0.50 psi). The bellows top surface was more dome-like in benchtop
testing than suggested by simulation results, indicating that FEM boundary conditions were too
strict and underestimated the deflection at the edges of the topmost bellows convolution.
Simulations achieved deflection approximations within 10% of the experimental load-deflection
curve for 1 and 2 convolution designs above 2.76 kPa (0.30 psi) when the Young’s modulus for
Parylene was increased from 2.76 GPa to 4.75 GPa, but greatly underestimated center deflection
for 3 convolution bellows.
Yang, et al. [92] argued that reduced rigidity provides significant improvement in deflection
for a single convolution bellows over a flat diaphragm of the same material and dimensions. In
[106], a Parylene C flat membrane of dimensions 2.8 x 1.6 mm
2
and thickness 2.8 m deflected
72

only up to 65 m under 4 psi applied pressure. Flat and corrugated Parylene C diaphragms of
dimensions 4.3 x 4.3 mm
2
achieved deflections of ~200 m and <50 m, respectively, at 1.5 kPa
[107]. Li et al. simulated deflection of a Parylene bellows and corrugated diaphragm of
approximately the same dimensions under the same load conditions (3.45 kPa applied pressure)
and observed maximum deflections of 1.5 and 0.8 mm, respectively [36].
FEM simulations predicted stress exceeding the Young’s modulus for Parylene C at ~5.52
kPa (0.80 psi), which was similar to the value seen for plastic deformation in mechanical testing.
The simulations also showed greater stress along the right angle convolution edges, particularly
in the lowest convolution near the clamping site. Incorporation of a convolution profile with
rounded edges may reduce stress concentration at these locations. Sites of high stress
concentration identified by FEM simulation were highly correlated to sites of burst failure of
fabricated bellows.
The fabrication process was relatively quick and used minimal resources. A large number of
molds can be made for the initial fabrication steps, but the overall throughput of the process is
limited by the manual stacking step. Automation or an improved method for stacking the PEG
modules could further increase the fabrication process throughput. The sidewall profiles of this
simple hole punching method are restricted to the smoothness obtained from the polished metal
punches, and could be a factor in material performance. Improved mold forming technologies
need to be explored for further improvement in mold surface quality.
Smaller inner-to-outer-diameter (ID/OD) ratios and larger convolution depths (difference
between outer and inner radius) may allow for greater achievable deflection of bellows on this
size scale, particularly at higher pressures. Although a power analysis with the t-test parameters
indicated that the studies were relatively low power, significance was still detectable in all but
one design parameter, ID. Small changes in ID from 1 to 2 mm did not have a noticeable effect
except at the highest pressure tested for the 2 mm case (3.45 kPa). In the case of 5-9-0.4 vs 6-9-
0.4, the ID to OD ratio increases from 0.56 to 0.67 and in 5-10-0.4 vs 7-10-0.4, the increase is
from 0.50 to 0.70. This suggests that larger ID or ID to OD ratio changes or possibly sample
sizes are required to elucidate the effect on achievable deflection. It was expected that increasing
OD while keeping all other parameters constant would enhance mobility and allow for greater
deflection of the bellows. This was confirmed in the load-deflection testing for the magnitude of
dimension change tested and the difference between the two bellows designs (varying only OD)
73

was statistically significant. Reducing layer height (H), and thus overall bellows height,
produced a statistically significant difference in deflection for the dimension change tested, but
the deflection change was relatively small compared to the reduction in H and would need to be
evaluated in the context of a given application. The number of convolutions (N) were varied to
evaluate the potential benefit of additional convolutions to increase potential deflection range,
but was kept low to minimize the overall bellows height in consideration of minimizing
dimensions for integration in low profile MEMS devices.
For the designs parameters evaluated in this work, the most effective way to increase bellows
deflection was to increase outer diameter, increase the number of convolutions, or decrease wall
thickness (Table 10). Modifying these parameters provided the greatest change in deflection
while requiring relatively small changes in initial bellows volume under no load. Thus, there are
several options for addressing the balance between mechanical strength and anticipated
displacement volume requirements with the overall volume occupied by the bellows once
integrated with a MEMS actuator or system.
The load-deflection curves showed hysteresis at lower applied pressure loads than with
Parylene flat membranes tested by Shih, et al., where ~28 kPa (~4 psi) was the onset of
hysteresis. Time for mechanical relaxation between cycles (after dropping from 0.69 kPa to 0.00)
was on the order of ten to twenty minutes, which is in agreement with stress relaxation time
constants reported in [108]. Dimpling was observed at approximately the onset of hysteresis in
the loading-unloading cycle. This was likely due to compressive stresses that developed near the
edge of the convolutions, as is seen in simply supported diaphragms [102].
Burst pressure testing indicated that the designs with larger convolution depths (table 5)
tended to burst at lower loads, corresponding with the relatively larger deflection and higher
associated stress. The pressure transducer’s range was limited to (~100 kPa) which in turn
limited investigation of some parameters (such as wall thickness) on mechanical strength. In
addition, only one bellows was used for each destructive test so further studies with larger
sample sizes and a transducer with greater range would be needed to evaluate the effects of
design parameter changes with statistical significance.
Two different bellows designs integrated with electrochemical actuators showed a
statistically insignificant difference in generated flow rate. It is important to note that the
pressure testing and the flow rate testing setups were in different fluidic environments (air vs.
74

liquid), which may have contributed to the insignificant effect of the tested design parameter
variations (number of convolutions, diameter) on flow rate. For pressure testing, bellows were
clamped in an open test fixture and deflected in ambient air. Static pressure was applied at the
previously described values. During flow rate testing pressure was continuously increasing via
electrolysis within the bellows, resulting in dynamic pressure values. The bellows deflected
against fluid (water) in the rigid reservoir and the coupled fluid column in the rigid cannula,
which was open at one end. Further investigation of bellows design parameters and their effect
on flow rate is warranted and will be examined in future work. An individual BEA was tested
over 100 times (much greater than required by our intended applications) without noticeable
changes in performance when operated in the elastic range (up to 3.45 kPa). Large deflection
was achieved under relatively low applied pressures (~kPa) compared with a bellows integrated
with a thermopneumatic actuator [92] while consuming significantly less power (~mm with ~3
mW versus ~ m with ~720 mW).

4.8 Conclusion
A rapid, high-yield fabrication process for MEMS Parylene C bellows for large deflection
applications was demonstrated. The fabricated thin film polymer bellows exhibited repeatable
behavior under loading within the determined elastic range and bellows of same design
demonstrated uniform load-deflection performance. FEM simulations provided approximations
of load-deflection curves for several bellows designs. The onset of hysteresis for the Parylene C
bellows structure (4.19 kPa or 0.60 psi) was determined in mechanical testing. Large deflection
(~mm) was achieved under relatively low applied pressure (~kPa). Bellows design parameters
were evaluated and it was found that convolution number, wall thickness, and outer diameter had
the greatest effect on load-deflection (axial extension) performance for the bellows designs
fabricated. Bellows were combined with interdigitated electrodes to form electrochemical
actuators and fluid pumping under low power (~3 mW) was demonstrated. Several design
parameters can be adjusted to achieve a desired magnitude of deflection while maintaining
dimensions appropriate for incorporation with MEMS actuators and devices. The bellows offer
complete separation of the electrolyte from the reservoir fluid when integrated with
electrochemical actuators and could have applications in other actuators where a separation of
the actuation fluids and mechanisms from other system chambers is necessary. When constructed
75

with Parylene C, which boasts biocompatibility, inertness to a broad range of chemicals, and low
permeability, bellows are poised to become an enabling technology for novel applications in
MEMS actuators and microfluidic systems.

4.9 Bellows Integration with 3
rd
Generation Wireless Drug Delivery System
Through mechanical characterization and simulations the dimensions and parameters
favorable for the GNR-SphK1siRNA nanoplex delivery application were determined. The
structure is robust and mechanical failure of material occurs far outside the normal operating
range. The bellows can withstand repeated use within the elastic range (up to 3.45 kPa or 0.5 psi
for designs tested). Increases in outer diameter of the bellows and decrease in inner-to-outer
diameter ratio provide greater extension, which provide greater displacement volumes. It was
determined that dimension changes do not interfere with electrolysis, and that flow rates
generated by the EC actuator are not significantly affected by bellows design. With slight
dimension changes relative to the bellows footprint, the maximum displacement volume can be
greatly improved without significantly affecting overall system size. Several bellows EC
actuators were continuously operated until failure (leak in the bellows) to determine the
maximum displacement volume. A 6-9-0.4, 2,13.5 bellows (as used for the 1
st
and 2
nd
generation
devices) could deliver at most ~225 L. A 5-10-0.4, 2,13.5 bellows could deliver over 400 L
before leaking or bursting. For the prior, the total “safe” displacement volume was set at 180 L,
and the latter 320 L. With this knowledge of the behavior of different bellows designs under
various load conditions and integrated with the EC actuator, the next generation micropump
system could be explored.

76

5 Redesign, Integration, and Implementation of the
3
rd
Generation Wireless Micropump System

5.1 Goals
5.1.1 Improvements upon 2
nd
generation
The goals of the 3
rd
generation micropump system redesign were to incorporate wireless and
valving components, but without increasing mass and volume of the implant. The EC actuator
dimensions were fixed and determined the footprint of the implant. However, with the results of
the bellows characterization, modifications to the bellows design provided an opportunity to
more efficiently use the surrounding reservoir space.

5.1.2 Application design specifications
The following design specifications were determined based on discussions with the
collaborators for the anti-cancer drug application. The micropump should be able to deliver a
daily dose of 30 µL at a rate of several µL/min. The study duration is anticipated to last at least
two weeks. The reservoir dimensions must balance the need to keep the reservoir size to a
minimum with the need to minimize refills and animal handling. The target for the mass of the
pump is <10% of the mouse’s weight, or 3 grams. The height and diameter should also be at least
as small as the 1
st
generation silicone micropumps, which were 10 mm and 22 mm, respectively.

Table 11. Application specifications for the 3
rd
generation micropump system
Parameter Specification
Flow rate ~µL/min
Dosing volume (daily) 30 µL
Study duration > 2 weeks
Refill 1 per week
Mass < 3 g
Height < 10 mm
Diameter < 22 mm

5.2 Design
An overview of the complete implantable wireless micropump system is shown in Figure 52.
Extensive details of component dimensions can be viewed in APPENDIX D: 3
rd
Generation
Wireless Micropump System Specifications, and several drawings in APPENDIX E: 3
rd

77

Generation Wireless Micropump System Drawings. Class VI materials were used as much as
possible and the entire system was encapsulated with a layer of biocompatible Parylene C and
medical grade silicone rubber. As will be discussed shortly, a high-resolution rapid prototyping
technique called stereolithography (SL) was used to fabricate the reservoir dome and base.

Figure 52. The redesigned packaging and configuration of the 3
rd
generation micropump system
in a) an exploded assembly view and b) a cross-sectional view.

5.2.1 Bellows electrochemical actuator
The previous system’s bellows (6 mm ID, 9 mm OD, 0.4 mm layer height, and 2
convolutions) could only achieve 180 µL displacement volume. In order to reduce refill
frequency, the bellows needs to displace more volume. But this increase in displacement volume
should not occur at the expense of added height to the pump profile. Based on the knowledge
from chapter 4, a slight increase in outer diameter would accomplish this outcome, so a 6 mm
ID, 9.5 mm OD bellows design was chosen. The EC actuator is exactly the same as the one used
in the 2
nd
generation micropump.

5.2.2 Reservoir
The 3
rd
generation reservoir base was rectangular rather than circular to more closely mimic
the EC actuator dimensions. Though the perimeter of the shape changed, the redesigned base
78

took advantage of the “slot” concept used in the 2
nd
generation to reduce height and still provide
additional structural reinforcement to the actuator substrate and cover its sharp corners.
The reservoir dome dimensions were selected according to the bellows outer diameter and
expected deflection height and are shown in Table 12. Although increasing the outer diameter of
the bellows results in a slight increase in dead volume between the convolutions, it represents
only a small percentage of the fill volume. With the reduction of the reservoir inner diameter
close to that of the bellows outer diameter, the dead volume between the wall and bellows was
drastically reduced from nearly 1100 L to less than 100 L. Compared to the 2
nd
generation
reservoir, the 3
rd
generation design reduces the reservoir inner diameter by 46%, the fill volume
by 65%, the footprint by 18%, and the dead volume (inaccessible drug) by nearly 90%. These
changes made it possible to incorporate the wireless powering components and valve without
exceeding the dimensions of the 1
st
generation system.

Table 12. Comparison of reservoir dimensions for several generations of micropumps
Parameter
1
st
/ 2
nd
Generation

3
rd
Generation

Height

8.3 mm 8.1 mm
Footprint

21.6 mm, OD 20 x 15 mm, L x W
Reservoir

19.6 mm ID 10.5 mm ID
Reservoir fill
volume

1300 L 450 L
Dead volume*
Between bellows
convolutions
28 L
34 L
Between wall and
bellows OD
1087 L 96 L
*1
st
and 2
nd
generations used a 6 (ID)-9 (OD)-0.4(H) bellows design;
3
rd
generation uses a 6 (ID)-9.5(OD)-0.4(H)design

The new reservoir dome also featured two refill ports instead of one to facilitate refill in the
valved configuration of the system, which will be discussed later in the chapter. The ports were
79

each 4 mm in diameter, which provided a reasonably large target area for aiming the refill
needle, but did not extend the reservoir dome dimensions past the footprint of the implant. A
small ledge located 2.8 mm deep in the introduction port and 1.2 mm in the extraction port
prevented septa from occluding the lumen to the reservoir. Below these ledges, the channel
tapered to minimize dead volume and trapped air bubbles. The extraction side septum was
thinner due to the placement of the reservoir outlet to this port at the ceiling of the dome to
improve extraction of air bubbles. The refill ports were offset from the reservoir and had
connecting channels with small diameters relative to the refill port diameter to avoid adding dead
volume to the drug reservoir. The reservoir dome had a footplate for increased surface area for
adhesion and to facilitate alignment during assembly. A tunnel protruding from the reservoir
dome housed the valve and catheter assembly. The wireless components were integrated around
the circumference of the reservoir to minimize added volume.

5.2.3 Valve port and connections
The one-way valve used in the 2
nd
generation system was also used in the 3
rd
generation. The
valve manufacturer (Minivalve) recommends an approximately 2.0 mm diameter sleeve to fit
over the duckbill portion of the valve and an access hole no larger than 1.4 mm on the back side
of the valve. From handling during earlier benchtop testing, it was clear that the silicone valve
material is sensitive to mechanical forces and pressures, which is advantageous for low opening
pressures, but necessitates proper housing to ensure the valve only opens when desired. Thus, a
tunnel was incorporated into the reservoir dome that housed the valve and connections to the
catheter. The valve sleeve consisted of a segment of Class VI extruded PTFE (Thin wall 12
AWG, Zeus, Inc.), which had an outer diameter less than that of the tunnel and an inner diameter
matching the specifications of the valve manufacturer. Several design concepts for seating the
valve in the reservoir dome were considered: a simple housing where the valve base aligns over
the access hole, an external flange that would be fit inside a recess on the interior of the
reservoir, and a built-in flange in the valve housing (Figure 53).
The valve seating option shown in Figure 53a was the simplest design and required no
additional SL parts. However, the alignment of the valve over the hole (1.4 mm diameter) in the
tunnel would be difficult, clogging of the hole would be likely during epoxy application or cure,
80

and the valve could be deformed during attachment. This could unintentionally open the slit
feature in the duckbill portion of the valve and affect one-way operation.

Figure 53. Three valve seating options were considered in the reservoir redesign. a) A simple
tunnel with port sized for the valve inlet. b) an external flange that would be mounted to the back
of the valve, and then inserted into a wall recess. c) a built-in flange over which the valve would
be seated.

The second option (Figure 53b), an externally assembled flange, had the advantage of
extra reinforcement to keep the valve in its best conformation, the valve and flange orifices could
be aligned externally to the reservoir, and the likelihood of epoxy clogging is greatly reduced.
But this option required an extra SL part which would be pushing the resolution limits of the SL
process and material, and required two alignment steps: 1) the valve and the flange, and 2) the
flange and the reservoir wall recess. The third option (Figure 53c) was selected because it was
less likely to result in valve clogging during epoxy application and cure, provided extra
reinforcement for the valve, aided in alignment, and did not require an extra SL part. In order to
fit this built-in flange inside the valve base and have flange wall thicknesses robust enough for
the SL process and handling during assembly, the access hole must be rather small at 0.4 mm
diameter. This was approximately the same size as the catheter diameter and the length of the
port was only 1 mm, so minimal added flow resistance would occur. More views of the valve
port design can be found in APPENDIX E: 3
rd
Generation Wireless Micropump System
Drawings.

5.2.4 Catheter assembly
81

A Class VI polyurethane catheter with an inner diameter of 0.64 mm (0.025” ID; SAI
Infusion Technologies) was chosen for the 3
rd
generation system because it is more rigid and less
permeable than the previously used silicone catheter, yet more pliable than PEEK, PTFE, and
polyethylene. It is important for implantation surgeries that the material is easy for the surgical
personnel to handle and place near the tumor. The diameter is approximately the same as the
silicone catheter diameter used in earlier generations. The catheter lumen selected is large
enough to avoid clogging with biological tissue and fluids, but small enough that a minimal
volume of drug is stored after the valve and potentially mixing with biological fluids. A short
segment of Class VI silicone rubber tubing with an inner diameter of 1.02 mm (0.040” ID, VWR
International) provides the necessary step from the smaller catheter outer diameter to the larger
valve sleeve tubing with an inner diameter of 2.16 mm (0.085” ID; Zeus, Inc., Orangeburg, SC)
(Figure 52).

5.2.5 Wireless power components
A Class E inductive wireless power source developed in the Biomedical Microsystems
Laboratory was used. It has a transmission frequency of 2 MHz and requires 9 volts (modified
after the 2
nd
generation wireless demonstration) to power the external transmitter circuit and coil,
the latter of which will be situated under the animal cage during in vivo studies. The receiver
circuit and coil are mounted on the implanted micropump. A current regulator on the implanted
receiver controls the current applied to the EC actuator, and can be adjusted to the desired flow
rate prior to implantation. The current wireless system operates at a set flow rate (related to on
current output from the receiver), but can be turned on and off after implantation to provide
various dosing volumes.

82

Figure 54. a) Schematic of inductive wireless power source, b) photograph of wireless benchtop
testing setup, and c) conceptual drawing of planned in vivo study setup.

5.3 Fabrication
5.3.1 Bellows electrochemical actuators
EC actuators will be made in the same manner as those for the 2
nd
generation micropump as
described in Chapter 3. The bellows will be made of Class VI Parylene C and fabricated
according to the process described in chapter 4.

5.3.2 Reservoirs
The new reservoir has very small features, in particular the refill ports and connection
channels, that are not compatible with basic injection molding but can be addressed by using
another type of rapid prototyping technology called stereolithography (SL). The process begins
with a vat of liquid thermoset resin. An ultraviolet laser scans the XY cross-section of the part,
solidifying the resin at the focal point. The part is built up layer by layer to form a three-
83

dimensional object. Due to the fine features of the reservoir, the requirement for
biocompatibility, and the relatively small quantity of parts needed, the reservoirs were
outsourced to a commercial prototyping company (FineLine Prototyping, Inc., Raleigh, NC). The
company specializes in high-resolution SL and has a Class VI SL material called WaterShed
®

XC 11122. WaterShed
®
, which has properties similar to the common thermoplastic acrylonitrile
butadiene styrene (ABS), is water resistant and durable. The material is also optically clear,
which is beneficial during assembly and diagnosing issues in benchtop testing (Figure 55). One
limitation of this process and the WaterShed
®
material is that the minimum feature size is 0.5
mm. But beyond this size, the resolution is 50 µm.

Figure 55. The 3
rd
generation reservoir domes and bases were fabricated using high-resolution
stereolithography.

5.3.3 Septa
Holes ranging in diameter from 3.99 mm (0.157” #22 drill bit) to 4.39 mm (0.173” #17 drill
bit) were drilled into ½” thick acrylic molds. These holes were then filled with a 10:1 ratio of
base to curing agent uncured and degassed silicone rubber (Class VI MDX-4 4210; Factor II,
Lakeside, AZ). The mold was cured at 80 °C for 1.5 hours, or until the silicone plugs were fully
cured. The plugs were then sliced with a razorblade into 1.2 and 2.8 mm thick discs. Any
sections of plugs that had air pockets were discarded to avoid weakened structural integrity that
could compromise the sealing of the septa. These discs were then inserted into the refill ports of
the SL dome and the perimeters reinforced with marine epoxy (Figure 56). Biocompatible
epoxies were evaluated, but were not robust enough to withstand potential punctures by the refill
needle and the slight movement that occurs during insertion and removal of the needle.
84

Figure 56. Discs of medical grade silicone rubber act as septa in the refill port. A ledge built in
to the port prevents septa from occluding the lumen to the reservoir.

The 4.21 mm (0.166” #19 drill bit) diameter holes provided the best combination of ease of
insertion into the port and compression that prevents dislodging the septa during needle removal.
With a port diameter of 4 mm, this represents 5% compression of the silicone septa.

5.4 Assembly
A thorough description of the assembly order and details is provided in APPENDIX F: 3
rd

Generation Micropump System Assembly Chart and Bill of Materials. Briefly, several
components, such as the bellows, septa, and EC actuators, were made individually. Then these
parts were built into subassemblies, which were then combined into the final micropump system
as shown in Figure 57. The combined mass of the components was approximately 2.2 grams.
With the mass of the fluid in the drug reservoir, this value approached 2.7 grams. Potting and
encapsulation steps added nearly 1 gram, therefore this assembly step is a target for
improvement.
85

Figure 57. Overview of micropump assembly. Biocompatible epoxy secures a) the valve to the
built-in flange, b) the valve sleeve, silicone connection, and catheter together, and c) the bellows
to the bottom of the dome. d) The EC actuator, base, and dome with attached bellows e) come
together with a marine epoxy seal. f) The fully assembled 3
rd
generation wireless micropump
system is then coated with Parylene C and Class VI silicone rubber.

Several considerations were taken into account during fabrication and assembly. The SL
material used has a heat deflection temperature (HDT) of 46 to 54 °C (115 to 130 °F) at 0.46
MPa, so care was taken to ensure that none of the steps involving the SL parts exceeded this
temperature. 301, 730 unfilled, and OE145-3 (all EPO-TEK®) biocompatible epoxies were
evaluated for use in assembly based on handling and adhesion to the materials used for the
various system components. Of these, the 730 unfilled was chosen because it adhered well to all
the necessary parts and was an appropriate viscosity for ease of handling. The drawback of this
epoxy is that it requires a 24-hour cure time at room temperature. Because of the HDT of the
reservoir material, cure times were rather long. For some assembly steps, the next phase could
begin once the epoxy was set (after about 6 to 8 hours), in order to speed up the assembly
timeline.
The WaterShed
®
material is water resistant, but a coating of Parylene C was deposited on all
interior and exterior surfaces of the reservoir base and dome to further reduce permeability. This
coating was done after septa insertion and epoxy reinforcement to also reduce permeability
86

through the septa. After assembly and wireless receiver mounting, the entire system was
encapsulated with Parylene C to insulate the wireless components and further coated with a thin
layer of medical grade silicone to provide a softer interface with the tissue when implanted.

5.5 Experimental Methods
5.5.1 Valve screening
The commercial valves showed varying performance on the benchtop, thus the valves were
pre-screened using a custom test fixture (Figure 58). A wired BEA inside a rigid reservoir
received 0.5 mA current for approximately 13 minutes to deliver a 60 µL bolus through the
valve, which was temporarily clamped without epoxy in a custom fixture at the reservoir outlet.

Figure 58. Valves were clamped temporarily in a custom acrylic fixture for screening prior to
integration with the system. The duckbill portion of the valve extends to the right into a hole
drilled according to the size recommendations for the valve sleeve.

After testing, the valve could be later integrated into a system. The outlet of the valve clamp
was connected to a calibrated micropipette to track delivery volume and reverse leakage
volumes. Valves that resulted in significant reverse leakage during recombination (> 20%) were
deemed underperforming and discarded. Valves were examined under a microscope before and
after testing to determine if the shape of the slit at the tip of the valve could predict performance,
and if testing affected the opening.

5.5.2 Refill procedure
To refill and flush the pump, a syringe with 30-gauge beveled non-coring needle is inserted
into the extraction port and another 30-gauge needle into the introduction port (Figure 59). A
Luer adapter was used to connect a long piece of Class VI silicone tubing to the introduction
needle, and could be filled with the drug or test solution. Pulling on the extraction syringe
plunger pulls in fluid from the introduction port. To flush the catheter, the introduction needle
87

and tubing are left in the port and the extraction needle removed. Depressing the plunger of a
syringe connected to the introduction tubing forces drug out of the catheter.

Figure 59. The micropump reservoir is refilled using two ports, one for fluid introduction, and
the other for fluid extraction.

5.5.3 Septa robustness to multiple punctures
Previous work in the lab demonstrated the self-sealing capabilities of silicone rubber [84].
The septa, made of the same silicone rubber (MDX-4 4210, Factor II, Lakeside, AZ), were
subjected to stages of multiple punctures followed by pressure testing. A 30 gauge beveled tip
(non-coring) needle (0.012” outer diameter) was chosen for puncture tests (and for refill) because
it minimizes damage to the silicone. In [84], repeated punctures in the same location resulted in
failure at lower pressures than repeated punctures in random locations, so the prior was chosen
for failure testing here.
An acrylic test fixture was built with a pressure inlet and two reservoir ports (Figure 60a).
Four micropump domes with epoxied septa, coated with Parylene C, were each mounted over the
pressure inlet and secured with 5-minute epoxy. The 5-minute epoxy adhered the dome to the
smooth acrylic without leaks beyond the upper limit of testing, 775 mmHg (15 psi). The test
fixture had interchangeable covers, one of which formed a reservoir over the outside of the
dome, and the other a needle guide (Figure 60b) for making repeated punctures to the same
location on the septa. The guide has two small (0.015” diameter) holes to provide a narrow guide
for the refill needles.
88

Figure 60. A reservoir dome with septa was mounted in a) a test fixture and subjected to several
stages of multiple punctures using a b) needle guide, followed by pressure testing until leaks or
up to 775 mmHg (15 psi).

The mounted dome was punctured 5 times using the guide. Then this piece was replaced by
the reservoir cover and the reservoir was filled with deionized water. The reservoir inlet was
plugged and the outlet connected to a calibrated micropipette to observe any fluid displacement
due to leaks through the septa. Pressurized nitrogen gas was applied in increments of 50 mmHg
up to 775 mmHg (15 psi), or until nitrogen bubbles were seen leaking through the septa or the
fluid front in the catheter moved. The reservoir was then drained and replaced with the alignment
piece for an additional 7 punctures (for a total of 12). The pressure testing was repeated on the
domes, then 12 additional punctures (for a total of 24) were made and the domes again pressure
tested.

5.5.4 Benchtop wireless testing
The micropump with mounted receiver circuit (with a current output of 0.3 mA) and coil
were placed on an acrylic shelf that sits on top of the coil at approximately the same height as an
animal cage. All micropumps were tested in the same location within the coil perimeter to
eliminate any slight variations in flow rate that could occur due to slight differences in the
magnetic field across the coil area. The micropump’s catheter was connected to a calibrated
micropipette for flow rate measurements as described in previous chapters.

89

5.5.5 Physiological environment simulation (soak test)
The Parylene C and silicone rubber encapsulation layers are expected to provide a temporary
moisture barrier and thereby protect the electronics. A fully assembled and encapsulated
micropump was soaked in 1X phosphate buffered saline (PBS) at body temperature (37 °C) to
simulate in vivo environmental conditions and verify that the encapsulation layers are adequate
for the duration of the anticipated in vivo study. Each day the micropump was removed from the
bath, refilled with deionized water, and subjected to a flow rate test of approximately ten
minutes. Valve opening times were recorded, and once the flow rate stabilized, flow rate
measurements were taken each minute for 5 minutes.

5.5.6 Room versus body temperature operation
A previous experiment comparing EC actuator flow rate performance at room and body
temperature showed no statistically significant difference [41], so it is expected that the
integrated system will also have no significant changes in flow rate with an increase from room
to body temperature. To verify, a fully assembled and encapsulated micropump was soaked in a
room temperature (21 °C) or body temperature (37 °C) deionized water bath for 30 minutes. The
reservoir and catheter were refilled and flushed of any air bubbles. While still submerged in the
bath, the micropump was subjected to a flow rate test. After valve opening and movement of the
fluid front in the catheter was stable, current was applied for 5 minutes and turned off for 2
minutes. Four cycles of on/off current application were performed for each temperature. The
average flow rate was measured over the 5 minute interval. The temperature of the baths was
monitored throughout testing with a temperature probe and less than 1 °C fluctuation was
observed.

5.5.7 Viscosity
Prior viscosity tests were all conducted at a higher applied current and higher flow rates than
the flow rate intended for in vivo studies, therefore viscosity testing was repeated with the new
wireless system at the lower flow rate to ensure that there would still be no significant effect.
Similar to preparations for the 2
nd
generation viscosity testing, five binary D-glucose in water
solutions were made to simulate viscosities ranging from approximately 2 to 6 cP. Glucose was
90

added to 50 °C MilliQ (Millipore) water according to Table 13. The containers were kept
covered throughout mixing and cooling.
Table 13. Glucose solutions prepared for viscosity testing of the 3
rd
generation micropump
D-glucose in water
(%, w/w)
Grams of glucose
in 50 mL water
Viscosity (cP)
0.00 0.000 1.0
21.00 13.291 2.0
29.80 21.225 3.0
35.25 27.220 4.0
39.18 32.203 5.0
42.19 36.490 6.0

Solutions were chosen based on a calibration curve determined in the lab using a Cannon-
Fenske Routine Viscometer. The same micropump was used for all solutions, including water
only. In between tests the drug reservoir was flushed three times with deionized water followed
by an air flush to minimize contamination between the different concentrations. Current was
turned on for approximately 7 minutes, with the first 2 minutes allowing time for the valve to
open and the last 5 minutes for taking flow rate measurements.

5.5.8 Back pressure
Micropumps were also subjected to back pressure values in increments of 5 mmHg up to 20
mmHg. The wireless micropump was connected to one end of a calibrated micropipette, and the
opposite end of this micropipette to the outlet of a custom pressure setup similar to Figure 26
(Figure 61). The previous generation micropump was tested with back pressures up to 15 mmHg,
but at much higher flow rates of nearly 50 µL/min. The in vivo flow rate will be much lower at 2
to 3 µL/min, so it is important to confirm that at lower flow rates the performance is still
unaffected by back pressure. 20 mmHg was selected as upper limit for testing because it exceeds
the central venous pressure for mice (and humans), a value much higher than what is anticipated
for subcutaneous drug delivery.
Micropumps were refilled prior to each set of runs. Each set consisted of a baseline flow rate
measurement over 5 minutes after valve opening, followed by 2 minutes with current off. Then
current was turned on again for another 5 minutes simultaneously with one of the back pressures,
followed by a 2 minute period with neither current nor back pressure. This was repeated until
91

flow rate measurements were taken at each of the back pressure values. The order in which the
back pressures were tested was purposely staggered for each of the sets. Four sets of runs were
conducted and the mean flow rate over the 5-minute intervals were calculated at each back
pressure value.

Figure 61. The wireless micropump was subjected to back pressure values up to 20 mmHg.

5.5.9 Dosing regimen
Six micropumps were operated wirelessly to simulate the in vivo dosing study. Pumps were
powered on and delivered 30 µL daily over approximately 10 to 15 minutes depending on the
individual micropump’s flow rate. Volume delivered as well as reverse leakage in between doses
was tracked in a calibrated micropipette. Just prior to the eighth day’s dose the reservoirs were
refilled, as would be done in vivo, and consecutive daily dosing continued.

5.6 Results and Discussion
5.6.1 Valve screening
Approximately 50% of the valves exhibited reverse leakage greater than 20% and therefore
did not pass the valve screening and were not integrated into the micropump reservoirs. The
valves were examined before and after screening. No changes were observed in the shape of the
slit opening that would indicate the screening test affected structural integrity. It was observed,
92

however, that the slits varied significantly from one valve to the next. Some of the slits were off
center (Figure 62).

Figure 62. Variation in appearance of valve slits was not an indicator of future performance.

Unfortunately, these variations in appearance did not provide a pattern for predicting one-
way function. However, the valve screening itself was quite successful and the performance
results of micropumps that used the pre-screened valves will be shown later.

5.6.2 Septa robustness to multiple punctures
All four tested domes showed no septa leaks for up to one dozen punctures and a maximum
applied pressure within the dome of 775 mHg (15 psi). After two dozen punctures, the pressure
at which the septa leaked varied. One dome did not leak up to the testing setup maximum of 775
mmHg, and another dome leaked at 700 mmHg (13.5 psi). The other two domes leaked at 100
mmHg (1.9 psi). In all cases, the leak occurred solely at the thinner of the two septa. This is in
agreement with [84], which also found that the thinner silicone membrane leaks at lower
pressures.
For benchtop testing prior to in vivo testing, the septa must be punctured once each for both
initial filling and refilling between weeks 1 and 2. Adding two additional punctures per fill/refill
to account for potential “misses” or inadequate punctures brings the total number of punctures
per septum to six. Doubling this amount for refill prior to and during the in vivo study would
bring the total number of punctures per septum to twelve. The punctures during robustness tests
were purposely made in the same location using a needle guide. In a similar setup in [84],
repeated punctures in the same location resulted in worse performance than punctures made in
different locations. Without the needle guide, it is unlikely that the punctures will occur as in the
“worst case scenario” of same location punctures. For all four pumps, no leakage occurred up to
twelve punctures.
93

In addition, the reservoir pressure is not expected to reach a value greater than the opening
(cracking) pressure of the valve, which was experimentally determined to be < 0.69 kPa (5.17
mmHg or 0.10 psi). Thus, the pressure would have to increase by nearly 20 times the value
expected in normal operating conditions for the septa to leak.

5.6.3 Benchtop wireless testing
A single micropump was tested on the benchtop with five different wireless receivers. An
average flow rate of 2.66 ± 0.16 µL/min (mean ± standard error, n=5) was measured and the
individual flow rates ranged from 2.20 to 3.20 µL/min. Because the range of flow rates is
relatively large compared to the magnitude of the flow rate, each micropump would need to be
calibrated prior to in vivo study.

5.6.4 Physiological environment simulation (soak test)
After 21 days of soaking in 1X PBS at 37 °C, the micropump was still functioning
consistently and as expected (Figure 63). On days 7 and 9, lower flow rates were attributed to an
air bubble that was observed during refill on Day 10. This air bubble was likely introduced just
prior to the day 7 dose. The filling protocol was modified to ensure adequate flushing to remove
any potentially trapped air bubbles for further testing days. Excluding these outliers, the flow
rates vary within the expected range for a given wireless micropump under normal benchtop
testing conditions (dry, at room temperature). The mean (excluding outliers on days 7 and 9)
after 21 days of soaking was 2.80 µL/min. After 21 days, the micropump no longer generated
forward flow. During debugging it was determined that the electrical connection between the
wireless receiver and the BEA was no longer functional. When tested individually the wireless
receiver and BEA were still operating as expected.
Although the micropump exceeded the intended in vivo duration by 50%, it would be
preferable to have a greater safety margin between the study duration and implant duration. In
addition, only 1 micropump was tested under these conditions. Thus, an improved electrical
connection between the receiver and BEA is needed, and more micropump systems should be
evaluated in simulated physiological conditions.
94

Figure 63. The micropump system generated flow rates for 21 days in simulated in vivo
conditions, 50% longer than the intended duration of in vivo studies.

5.6.5 Room versus body temperature operation
As expected based on EC actuator testing, body temperature did not have a significant effect
on flow rate performance of the micropump (Figure 64). Pump 4 generated a flow rate of 2.83 ±
0.03 µL/min (mean ± standard error, n=4) at room temperature (21 °C) and 2.77 ± 0.25 µL/min
(mean ± standard error, n=4) at body temperature (37 °C). Pump 21 generated a flow rate of 1.75
± 0.03 µL/min (mean ± standard error, n=4) at room temperature (21 °C) and 1.69 ± 0.11 µL/min
(mean ± standard error, n=4) at body temperature (37 °C).

Figure 64. Wireless micropump flow rate performance was not significantly affected by
increasing the environmental temperature from room (21 °C) to body temperature (37 °C).

95

Similar to results seen with the stand-alone EC actuator [41], micropump flow rate
performance is not affected by increasing the micropump environment to body temperature, thus
as far as temperature is concerned, benchtop performance is expected to adequately represent in
vivo performance.

5.6.6 Viscosity
The same micropumps were also subjected to solutions of varying viscosity, and minimal
effects on flow rate performance were observed up to the viscosity of blood, which is 3 to 4 cP at
body temperature. Delivery of water, with a viscosity of 1 cP at 21 °C, was used to determine the
baseline flow rate measurement. The mean flow rate at viscosities up to 6 cP were within Pump
4’s baseline flow rate of 3.67 µL/min (Figure 65a). Pump 21 generated a baseline flow rate of
1.83 µL/min (Figure 65b). Up to 3 cP, the mean flow rate fell within 10% of the baseline, and up
to 6 cP the mean flow rate fell within 20% of the baseline.

Figure 65. Flow rate deviation from baseline was less than a) 10% up to 6 cP in one micropump
and b) less than 10% up to 3 cP and less than 20% up to 6 cP in another micropump.
96

The means across the two micropumps were different, but consistent with the variation seen in
an individual micropump tested with different receivers (section 5.6.3). When the micropump
mean flow rates were normalized to 1, flow rates were still within 10% of the baseline for both
pumps up to 3 cP (Figure 66). Above 3 cP, flow rates were within 10% for Pump 4 and within
20% for Pump 21. Error was higher for one of the micropumps above 3 cP, indicating even more
potential variation.

Figure 66. Flow rate performance between micropumps when the mean flow rates for each
pump were normalized to 1 was within 10% of baseline up to 3 cP.

The siRNA-based anti-cancer drug is delivered in a phosphate buffered saline (PBS) base
and has a viscosity much less than 3 cP, thus the expected variation in flow rate would be less
than 10% due to any slight variations in drug concentration.

5.6.7 Back pressure
Minimal effects on flow rate performance were generally observed with up to 20 mmHg
applied against the catheter. The flow rate at 0 mmHg of applied back pressure was used as the
baseline flow rate measurement. It was observed with all tested values of back pressure that the
mean flow rates were lower than baseline, but for the most part were within 10%. Pump 4
generated a baseline flow rate of 3.44 µL/min and variation was less than 10% from this value up
to 20 mmHg (Figure 67a). Pump 21 generated a baseline flow rate of 2.60 µL/min (Figure 67b).
97

The mean flow rates at each back pressure were within 10% of the baseline except for the mean
flow rate at 10 mmHg, which was within 18% of baseline.

Figure 67. Mean flow rate for two separate micropumps varied less than 10% for most cases
with up to 20 mmHg back pressure applied against the catheter.

The means across the two micropumps were different, but as discussed in the viscosity
testing, the difference was consistent with the variation seen in an individual micropump tested
with different receivers (section 5.6.3). When the micropump mean flow rates were normalized
to 1, flow rate performance between the micropumps was comparable up to 15 mmHg (Figure
68). Pump 21 showed variable flow rate performance at 20 mmHg, but the lower end of the
standard error overlapped with the baseline flow rate.
98

Figure 68. Flow rate performance between micropumps was similar when the mean flow rates
are normalized to 1.

For the intended in vivo studies, the subcutaneous delivery location is expected to subject the
micropump to back pressure much less than the central venous pressure (8 mmHg, [88]). Based
on the back pressure testing, even at nearly double this value, variation of less than 10% from the
baseline flow rate is expected to occur.

5.6.8 Dosing regimen
5.6.8.1 Flow rate performance
The first week of the consecutive daily dosing regimen for each of the six micropumps is
shown in Figure 69. Micropumps successfully delivered the 30 µL dose each day. The mean
flow rate is shown on each graph, excluding the first pump shown (Pump 2), which was wired
for the first three days for comparison to performance with the wireless receiver the last four
days. For the fifth pump, days 6 and 7 are not shown because the catheter was accidentally
damaged during wireless system debugging. After repair the micropump was functioning
normally, but was effectively “reset” to day 1 because the reservoir had been exposed to ambient
pressure with the catheter damage. Micropumps were refilled just prior to the eighth dose, and
performance during the eighth dose was (as expected) comparable to the first dose of week 1.

99

Figure 69. Week 1 of benchtop testing of the wireless dosing regimen that will be followed for
in vivo studies. Mean flow rate for the seven days is shown for each micropump, except for
Pump 2, which was a mix of wired and wireless testing.

Three micropumps (2, 4, and 15) were tested for an additional week (Figure 70) and showed
similar flow rate performance to week 1 of dosing. An electrode to wireless receiver connection
failed prior to Pump 2’s week 2 dose 6. A more robust connection was made during assembly for
future micropumps to prevent recurrence of the issue. The other two micropumps delivered all
seven doses for week 2.

Figure 70. Week 2 of benchtop testing of the dosing regimen that will be followed for in vivo
studies.
100

Pump 15 was tested daily for three consecutive weeks (Figure 71). An electrical connection
was repaired prior to week 3, and resulted in a higher flow rate. It is possible that the connection
had affected the first two weeks of dosing, however the 30 µL dose was still achieved. On the
benchtop this can be observed and accounted for, but in vivo this would present an issue with
dosing accuracy.

Figure 71. Pump 15 was operated wirelessly and delivered daily doses of 30 µL for three
consecutive weeks on the benchtop.

5.6.8.2 Valve performance
Valve performance varied across the micropumps, but for the most part significantly reduced
back flow due to recombination in between dosing (Table 14). It was observed in one
micropump that the valve was more effective in preventing back flow during week 2 than in
week 1, and slightly better in week 3 than in week 2. This suggests the possibility that over
multiple dosing cycles, the valve behavior may improve. Further study with additional valved
micropumps over longer study durations is needed to confirm this phenomenon.

101

Table 14. Reverse leakage due to inadequate valve sealing between dosing
Pump
Reverse Leakage (µL)
Mean ± SE, n = 7 unless noted otherwise
Week 1 Week 2 Week 3
2 NA 0.38 ± 0.25, n=5 NA
4 6.05 ± 0.64 7.57 ± 1.25 NA
15 9.76 ± 3.04 3.26 ± 0.31 3.05 ± 0.32
17 6.42 ± 1.19 NA NA
19 8.40 ± 0.54, n=5 NA NA
21 3.24 ± 1.97 NA NA

5.6.8.3 Compensation for sequential dosing
A trend of increasing current duration (current on to current off) to achieve the daily volume
was observed across the micropumps throughout the consecutive daily dosing study. This
phenomenon has been noted previously in a valved system and is attributed to pressure build up
compensating for the reduced volume in the reservoir [42,89]. For micropumps with similar flow
rates above 2 µL/min and weeks with data on compensation time for all seven doses, the
compensation times (mean ± standard error, n=4) were plotted against the day in the consecutive
dosing cycle. As shown in Figure 72, the compensation time had an approximately linear
relationship with the day of consecutive dosing. After refill, the curve effectively “reset” as the
volume in the reservoir was replenished. Although the errors were relatively large compared to
the mean value on a given day, the curve would help to more accurately predict the start of
dosing and thus improve dosing accuracy.
102

Figure 72. Increasing time to start of dosing was observed throughout the consecutive dosing
cycle. Micropump was operated wirelessly and delivered daily doses of 30 µL for three
consecutive weeks on the benchtop, with reservoir refill occurring at the beginning of each week.

5.7 Summary
The redesigned reservoir and configuration of the 3
rd
generation micropump system reduced
the reservoir mass and volume, which led to successful wireless receiver and valve integration
without increasing the overall implant form factor. Including all encapsulation, the assembled
micropump system height was under 10 mm. With the length less than 22 mm and width less
than 16 mm, the footprint is smaller than the 1
st
generation. The mass of the components was
only 2.7 grams (with filled reservoir), but the addition of potting and encapsulation materials
brought the mass above the 3 gram limit. However, the total filled encapsulated system mass
including the wireless power components is still approximately the same as the wired
generations. The amount of potting material could be reduced in future designs by instead using
a low density polymer cap that fits over the electronics and reservoir. A minimal amount of
epoxy would then be needed to seal the cap to the current packaging, and could be followed with
very thin layers of encapsulation for reduced permeability. Based on the volume requirements for
the anti-cancer drug application, the reservoir size has been reduced as much as possible. If the
requirements were to be modified such that less delivered volume is required per bolus and
between refills, further reductions in system size could be achieved by using an even smaller
reservoir, bellows, and electrode design.
103

Two refill ports, rather than just one, were incorporated into the 3
rd
generation reservoir
design. This facilitated filling and flushing of the reservoir, without delivering fluid through the
catheter. When extraction occurred to (re)fill the reservoir, the duckbill feature in the valve
pinched shut and the amount of fluid in the catheter that entered the reservoir was negligible.
This ensures that flushing and filling will cause neither accidental dosing nor contamination of
the reservoir fluid with fluid from the catheter, which could contain trace amounts of biological
fluids.
While visual inspection of valves did not provide adequate information to predict valve
performance once integrated with the micropump, screening with a flow rate test was quite
successful in doing so. All of the valves that passed screening (~50%) and were integrated into
pumps performed reasonably well in reducing reverse leakage to a small fracture of dosed
volume in the periodic dosing regimen. Future work on biocompatible normally-closed valves is
needed for more adequate sealing during recombination and between delivery periods. These
valves should also be passive and have low opening pressures to minimize flow delays.
The fully integrated wireless systems were tested on the benchtop and achieved flow rates
on the order of µL/min that were relevant for the application of anti-cancer drug delivery.
Although the mean flow rate for an individual pump demonstrated minimal variation, the mean
flow rates across micropumps were sometimes significantly different. Future work with the
wireless components is needed to avoid needing calibration for each micropump system, but the
duration of dosing periods and overall study would still be on the same order of magnitude in
spite of the flow rate differences in the current micropump system.
Characterization tests of flow rate performance under various environmental conditions
verified that the micropumps were robust to changes in drug or test solution viscosity,
temperature, and back pressure variations. With regards to the anti-cancer drug delivery
application, the performance exceeded the thresholds for each of these conditions. Although a
small number of micropumps were tested, these results were in agreement with earlier generation
results and stand-alone actuator (prior to integration) results. In addition, the micropump
demonstrated consistent operation in a simulated in vivo environment for a duration significantly
longer than the intended in vivo study duration of two weeks. The micropump system
demonstrated the capability to carry out a dosing regimen consistent with the schedule which
will be conducted in later in vivo studies. The variation coefficient (standard deviation divided by
104

the mean) for flow rates of individual pumps ranged from 5% to 34%, which overlaps with the
10% variation coefficient of commercially available ALZET
®
osmotic pumps [10].

105

6 Conclusion

The goals of this work were to design, fabricate, and demonstrate a wireless implantable drug
delivery system featuring a MEMS bellows electrochemical actuator for site-specific, chronic
controlled drug delivery in mice. Several iterations of micropumps were designed, fabricated,
and implemented in both benchtop and in vivo studies.
The application of subcutaneous delivery of anti-cancer drug to tumors in mice drove the
design of the system. All but one of the specifications for the application were met (Table 15).
Even after adding wireless power components and increasing the bellows dimensions, the
footprint of the fully assembled and encapsulated 3
rd
generation micropump system was less than
that of the 1
st
and 2
nd
generation systems. Although the mass of the 3
rd
generation system
exceeded the desired specification, the mass is still on par with the 1
st
generation system. This
system was implanted in preliminary in vivo studies, but it was determined that rigid packaging
and wireless power were needed before additional in vivo work. The 3
rd
generation system could
be implanted and the mass likely tolerated by the mouse, but it would be preferable to reduce the
mass to a smaller percentage of the animal’s body weight. Dosing accuracy and repeatability is
highly dependent on the mechanical characteristics of all system components along the fluid
pathway, and thus rigid materials were implemented as much as possible to minimize effects of
compliance.

Table 15. Application specifications for the 3
rd
generation micropump system
Parameter Specification Met?
Flow rate ~µL/min Yes, 1 to 3 µL/min
Dosing volume (daily) 30 µL Yes
Study duration > 2 weeks Yes, 3+ weeks
Refill 1 per week Yes
Mass < 3 g 3.8 g, but can reduce potting amount
Height < 10 mm Yes, 9 mm
Diameter < 22 mm Yes, 21 x 16 mm

The actuation mechanism makes repeated dosing possible, but requires a one-way valve to
prevent potential mixing of biological fluids with drug in the micropump reservoir. Investigation
of commercial and MEMS valve options led to a suitable choice that was integrated and
evaluated with the micropump system. The dosing regimen planned for in vivo studies was
106

successfully demonstrated on the benchtop and pumping behavior was characterized for several
micropumps. The micropumps performed reliably even with variation in environmental
conditions, which included temperature, back pressure, and solution (drug) viscosity. A
micropump was also subjected to simulated in vivo conditions by submersion into a saline bath at
37 °C. The variation coefficient of the test systems was as low as 5%. Ideally, with further
improvements and fine tuning of the micropump system, the variation coefficient across all
micropumps would be 10% or less, on par with commercial implantable pumps [10]. These
characterization studies have shown that the wireless micropump system is suitable for in vivo
study.
The technology discussed here represents a portion of a much broader effort for improved
drug delivery. Future research can build upon the knowledge from this work to add “smart”
features to enhance performance and improve interfacing with the micropump system. The
current system operates at a set flow rate which can be turned on and off to achieve a desired
dosage amount. Having adjustable flow rate control in addition to on/off control would take this
technology to the next level of sophistication. This work focused on a micropump system
suitable for implantation in mice. However, the key actuation components are made using
MEMS processes, which makes it possible to scale the system both smaller and larger to provide
research opportunities in many different animal models and in novel applications that are limited
by current technology.
Because the components in contact with the reservoir fluid are inert and biocompatible and
the mechanism is independent of the drug, the micropump system could be modified for
applications other than the siRNA-based drug. As mentioned in the introduction, there is
increasing evidence that efficacy of drug therapies may be closely tied with biological rhythms.
Drug administration technology that has flexible programmability post-implantation would be an
invaluable tool in elucidating these more complex drug and biological rhythm interactions,
which could lead to improved treatment and quality of life for patients.

107

References
1. George J. Krinke The Laboratory Rat (Handbook of Experimental Animals); 1st ed.;
Academic Press: San Diego, Calif.; London, 2000.
2. Trull, F. L.; Rich, B. A. More Regulation of Rodents. Science 1999, 284, 1463–1463.
3. The Jackson Laboratory Fast facts - The Jackson Laboratory
http://www.jax.org/fastfacts/index.html (accessed Oct 12, 2012).
4. Turner, P. V.; Brabb, T.; Pekow, C.; Vasbinder, M. A. Administration of substances to
laboratory animals: routes of administration and factors to consider. Journal of the
American Association for Laboratory Animal Science 2011, 50, 600–613.
5. Infusion Products for Laboratory Research - SAI-Infusion http://sai-infusion.com/
(accessed Oct 16, 2012).
6. Urquhart, J.; Fara, J. W.; Willis, K. L. Rate-controlled delivery systems in drug and
hormone research. Annu. Rev. Pharmacol. Toxicol. 1984, 24, 199–236.
7. Kramer, K.; Kinter, L.; Brockway, B. P.; Voss, H. P.; Remie, R.; Van Zutphen, B. L. The
use of radiotelemetry in small laboratory animals: recent advances. Contemp Top Lab Anim
Sci 2001, 40, 8–16.
8. Youan, B. B. Chronopharmaceutics: gimmick or clinically relevant approach to drug
delivery? Journal of Controlled Release 2004, 98, 337–353.
9. Smolensky, M. H.; Peppas, N. A. Chronobiology, drug delivery, and chronotherapeutics.
Advanced Drug Delivery Reviews 2007, 59, 828–851.
10. Alzet ALZET® Osmotic Pumps - Implantable pumps for research http://www.alzet.com/
(accessed Oct 19, 2011).
11. Med-E-Cell Welcome to Med-e-Cell: infusion pumps, fluid delivery devices, oxygen
generation/control and fuel cell technology for licensing and development partnerships
http://medecell.com/ (accessed Oct 19, 2011).
12. iPrecio iPRECIO®, Innovative Drug Infusion Technology for Laboratory Animals
http://www.iprecio.com/ (accessed Oct 19, 2011).
13. Nguyen, N. T.; Huang, X.; Chuan, T. K. MEMS-micropumps: a review. J. Fluids Eng
2002, 124, 384–392.
14. Laser, D. J.; Santiago, J. G. A review of micropumps. Journal of Micromechanics and
Microengineering 2004, 14, R35–R64.
15. Woias, P. Micropumps--past, progress and future prospects. Sensors and Actuators B:
Chemical 2005, 105, 28–38.
16. Tsai, N. C.; Sue, C. Y. Review of MEMS-based drug delivery and dosing systems. Sensors
and Actuators A: Physical 2007, 134, 555–564.
17. Amirouche, F.; Zhou, Y.; Johnson, T. Current micropump technologies and their
biomedical applications. Microsystem Technologies 2009, 15, 647–666.
18. Yun, K.-S.; Yoon, E. Micropumps for MEMS/NEMS and Microfluidic Systems. In
MEMS/NEMS; Leondes, C. T., Ed.; Springer US: Boston, MA, 2006; pp. 1112–1144.
19. Smits, J. G. Piezoelectric micropump with microvalves. In Proc.
University/Government/Industry Microelectronics Symposium; IEEE, 1989; pp. 92–94.
20. Maillefer, D.; Gamper, S.; Frehner, B.; Balmer, P.; van Lintel, H.; Renaud, P. A high-
performance silicon micropump for disposable drug delivery systems. In Proc. IEEE Int.
Conf. MEMS; IEEE: Interlaken , Switzerland, 2001; pp. 413–417.
108

21. Maillefer, D.; van Lintel, H.; Rey-Mermet, G.; Hirschi, R. A high-performance silicon
micropump for an implantable drug delivery system. In Proc. IEEE Int. Conf. MEMS;
IEEE: Orlando, FL , USA, 1999; pp. 541–546.
22. Debiotech Debiotech - Medical Devices - Switzerland http://www.debiotech.com/
(accessed Nov 18, 2011).
23. Geipel, A.; Goldschmidtboeing, F.; Jantscheff, P.; Esser, N.; Massing, U.; Woias, P.
Design of an implantable active microport system for patient specific drug release.
Biomedical Microdevices 2008, 10, 469–478.
24. Evans, A. T.; Chiravuri, S.; Gianchandani, Y. B. A Multidrug Delivery System Using a
Piezoelectrically Actuated Silicon Valve Manifold With Embedded Sensors. Journal of
Microelectromechanical Systems 2011, 20, 231–238.
25. Judy, J. W.; Tamagawa, T.; Polla, D. L. Surface-machined micromechanical membrane
pump. In Proc. IEEE Int. Conf. MEMS; IEEE: Nara, Japan, 1991; pp. 182–186.
26. Zengerle, R.; Richter, A.; Sandmaier, H. A micro membrane pump with electrostatic
actuation. In Proc. IEEE Int. Conf. MEMS; IEEE: Travemunde , Germany, 1992; pp. 19–
24.
27. Santini Jr, J. T.; Richards, A. C.; Scheidt, R.; Cima, M. J.; Langer, R. Microchips as
controlled drug-delivery devices. Angew. Chem. Int. Ed 2000, 39, 2396–2407.
28. Grayson, A. C. R.; Choi, I. S.; Tyler, B. M.; Wang, P. P.; Brem, H.; Cima, M. J.; Langer,
R. Multi-pulse drug delivery from a resorbable polymeric microchip device. Nature
Materials 2003, 2, 767–772.
29. Li, Y.; Shawgo, R. S.; Tyler, B.; Henderson, P. T.; Vogel, J. S.; Rosenberg, A.; Storm, P.
B.; Langer, R.; Brem, H.; Cima, M. J. In vivo release from a drug delivery MEMS device.
Journal of Controlled Release 2004, 100, 211–219.
30. Elman, N.; Patta, Y.; Scott, A.; Masi, B.; Ho Duc, H.; Cima, M. The Next Generation of
Drug-Delivery Microdevices. Clin Pharmacol Ther 2009, 85, 544–547.
31. Staples, M. Microchips and controlled-release drug reservoirs. Wiley Interdisciplinary
Reviews: Nanomedicine and Nanobiotechnology 2010, 2, 400–417.
32. Moore, W. H.; Holschneider, D. P.; Givrad, T. K.; Maarek, J.-M. I. Transcutaneous RF-
Powered Implantable Minipump Driven by a Class-E Transmitter. Biomedical Engineering,
IEEE Transactions on 2006, 53, 1705–1708.
33. Givrad, T. K.; Maarek, J.-M. I.; Moore, W. H.; Holschneider, D. P. Powering an
Implantable Minipump with a Multi-layered Printed Circuit Coil for Drug Infusion
Applications in Rodents. Annals of Biomedical Engineering 2009, 38, 707–713.
34. Givrad, T. K. Induction Power Microbolus Infusion Pump Used for Functional
Neuroimaging Applications in Rodents. Ph.D. Dissertation, University of Southern
California, 2007.
35. Li, P.-Y.; Givrad, T. K.; Holschneider, D. P.; Maarek, J.-M. I.; Meng, E. A Parylene
MEMS Electrothermal Valve. Journal of Microelectromechanical Systems 2009, 18, 1184–
1197.
36. Li, P.-Y.; Sheybani, R.; Gutierrez, C. C.; Kuo, J. T. W.; Meng, E. A Parylene Bellows
Electrochemical Actuator. Journal of Microelectromechanical Systems 2010, 19, 215–228.
37. Li, P. Y.; Shih, J.; Lo, R.; Saati, S.; Agrawal, R.; Humayun, M. S.; Tai, Y. C.; Meng, E. An
electrochemical intraocular drug delivery device. Sensors and Actuators A: Physical 2008,
143, 41–48.
109

38. Neagu, C. R.; Gardeniers, J. G. E.; Elwenspoek, M.; Kelly, J. J. An electrochemical
microactuator: principle and first results. Journal of Microelectromechanical Systems 1996,
5, 2–9.
39. Böhm, S.; Timmer, B.; Olthuis, W.; Bergveld, P. A closed-loop controlled
electrochemically actuated micro-dosing system. Journal of Micromechanics and
Microengineering 2000, 10, 498–504.
40. Sillen, C.; Barendrecht, E.; Janssen, L.; Vanstralen, S. Gas bubble behaviour during water
electrolysis. International Journal of Hydrogen Energy 1982, 7, 577–587.
41. Sheybani, R.; Meng, E. High-Efficiency MEMS Electrochemical Actuators and
Electrochemical Impedance Spectroscopy Characterization. Journal of
Microelectromechanical Systems 2012, 21, 1197–1208.
42. Sheybani, R.; Gensler, H.; Meng, E. A MEMS electrochemical bellows actuator for fluid
metering applications. Biomedical Microdevices 2012.
43. Meng, E.; Li, P.-Y.; Tai, Y.-C. A biocompatible Parylene thermal flow sensing array.
Sensors and Actuators A: Physical 2008, 144, 18–28.
44. Belmont, C.; Girault, H. H. Coplanar interdigitated band electrodes for synthesis Part I:
Ohmic loss evaluation. Journal of Applied Electrochemistry 1994, 24, 475–480.
45. Lo, R.; Li, P.-Y.; Saati, S.; Agrawal, R.; Humayun, M. S.; Meng, E. A refillable
microfabricated drug delivery device for treatment of ocular diseases. Lab Chip 2008, 8,
1027.
46. Lo, R.; Kuwahara, K.; Li, P.-Y.; Agrawal, R.; Humayun, M. S.; Meng, E. A Passive
Refillable Intraocular MEMS Drug Delivery Device. In; IEEE, 2006; pp. 74–77.
47. Li, P.-Y.; Sheybani, R.; Kuo, J. T. W.; Meng, E. A parylene bellows electrochemical
actuator for intraocular drug delivery. In Proc. IEEE Int. Conf. TRANSDUCERS; Denver,
CO, USA, 2009; pp. 1461–1464.
48. Masood, R.; Roy, I.; Zu, S.; Hochstim, C.; Yong, K.-T.; Law, W.-C.; Ding, H.; Sinha, U.
K.; Prasad, P. N. Gold nanorod–sphingosine kinase siRNA nanocomplexes: a novel
therapeutic tool for potent radiosensitization of head and neck cancer. Integrative Biology
2012, 4, 132.
49. Parkin, D. M.; Bray, F.; Ferlay, J.; Pisani, P. Global Cancer Statistics, 2002. CA: A Cancer
Journal for Clinicians 2005, 55, 74–108.
50. American Cancer Society Cancer Facts & Figures 2012; American Cancer Society:
Atlanta, Georgia, USA, 2012.
51. Argiris, A.; Karamouzis, M. V.; Raben, D.; Ferris, R. L. Head and neck cancer. The Lancet
2008, 371, 1695–1709.
52. Wise-Draper, T. M.; Draper, D. J.; Gutkind, J. S.; Molinolo, A. A.; Wikenheiser-Brokamp,
K. A.; Wells, S. I. Future directions and treatment strategies for head and neck squamous
cell carcinomas. Translational Research 2012, 160, 167–177.
53. Hannun, Y. A.; Obeid, L. M. Principles of bioactive lipid signalling: lessons from
sphingolipids. Nature Reviews Molecular Cell Biology 2008, 9, 139–150.
54. Ogretmen, B.; Hannun, Y. A. Biologically active sphingolipids in cancer pathogenesis and
treatment. Nat Rev Cancer 2004, 4, 604–616.
55. Lee, O.-H.; Kim, Y.-M.; Lee, Y. M.; Moon, E.-J.; Lee, D.-J.; Kim, J.-H.; Kim, K.-W.;
Kwon, Y.-G. Sphingosine 1-Phosphate Induces Angiogenesis: Its Angiogenic Action and
Signaling Mechanism in Human Umbilical Vein Endothelial Cells. Biochemical and
Biophysical Research Communications 1999, 264, 743–750.
110

56. Wang, L.; Dudek, S. M. Regulation of vascular permeability by sphingosine 1-phosphate.
Microvascular Research 2009, 77, 39–45.
57. Wang, D.; Zhao, Z.; Caperell-Grant, A.; Yang, G.; Mok, S. C.; Liu, J.; Bigsby, R. M.; Xu,
Y. S1P differentially regulates migration of human ovarian cancer and human ovarian
surface epithelial cells. Molecular Cancer Therapeutics 2008, 7, 1993–2002.
58. Bergelin, N.; Blom, T.; Heikkila, J.; Lof, C.; Alam, C.; Balthasar, S.; Slotte, J. P.;
Hinkkanen, A.; Tornquist, K. Sphingosine Kinase as an Oncogene: Autocrine Sphingosine
1-Phoshate Modulates ML-1 Thyroid Carcinoma Cell Migration by a Mechanism
Dependent on Protein Kinase C- and ERK1/2. Endocrinology 2008, 150, 2055–2063.
59. Cuvillier, O.; Pirianov, G.; Kleuser, B.; Vanek, P. G.; Coso, O. A.; Gutkind, J. S.; Spiegel,
S. Suppression of ceramide-mediated programmed cell death by sphingosine-1-phosphate.
Nature 1996, 381, 800–803.
60. Kohama, T. Molecular Cloning and Functional Characterization of Murine Sphingosine
Kinase. Journal of Biological Chemistry 1998, 273, 23722–23728.
61. Mandala, S. M.; Thornton, R.; Tu, Z.; Kurtz, M. B.; Nickels, J.; Broach, J.; Menzeleev, R.;
Spiegel, S. Sphingoid base 1-phosphate phosphatase: A key regulator of sphingolipid
metabolism and stress response. Proceedings of the National Academy of Sciences 1998,
95, 150 –155.
62. Olivera, A. Sphingosine Kinase Expression Increases Intracellular Sphingosine-1-
phosphate and Promotes Cell Growth and Survival. The Journal of Cell Biology 1999, 147,
545–558.
63. Spiegel, S.; Milstien, S. Functions of the Multifaceted Family of Sphingosine Kinases and
Some Close Relatives. Journal of Biological Chemistry 2006, 282, 2125–2129.
64. Sinha, U. K.; Schorn, V. J.; Hochstim, C.; Chinn, S. B.; Zhu, S.; Masood, R. Increased
radiation sensitivity of head and neck squamous cell carcinoma with sphingosine kinase 1
inhibition. Head & Neck 2011, 33, 178–188.
65. Nava, V. E.; Cuvillier, O.; Edsall, L. C.; Kimura, K.; Milstien, S.; Gelmann, E. P.; Spiegel,
S. Sphingosine Enhances Apoptosis of Radiation-resistant Prostate Cancer Cells. Cancer
Research 2000, 60, 4468 –4474.
66. Mahdy, A. E.; Cheng, J. C.; Li, J.; Elojeimy, S.; Meacham, W. D.; Turner, L. S.; Bai, A.;
Gault, C. R.; McPherson, A. S.; Garcia, N.; Beckham, T. H.; Saad, A.; Bielawska, A.;
Bielawski, J.; Hannun, Y. A.; Keane, T. E.; Taha, M. I.; Hammouda, H. M.; Norris, J. S.;
Liu, X. Acid Ceramidase Upregulation in Prostate Cancer Cells Confers Resistance to
Radiation: AC Inhibition, a Potential Radiosensitizer. Molecular Therapy 2008, 17, 430–
438.
67. Taha, T. A. Down-regulation of Sphingosine Kinase-1 by DNA Damage: DEPENDENCE
ON PROTEASES AND p53. Journal of Biological Chemistry 2004, 279, 20546–20554.
68. Whitehead, K. A.; Langer, R.; Anderson, D. G. Knocking down barriers: advances in
siRNA delivery. Nature Reviews Drug Discovery 2009, 8, 129–138.
69. Dykxhoorn, D. M.; Palliser, D.; Lieberman, J. The silent treatment: siRNAs as small
molecule drugs. Gene Ther 2006, 13, 541–552.
70. Howard, K. A. Delivery of RNA interference therapeutics using polycation-based
nanoparticles. Advanced Drug Delivery Reviews 2009, 61, 710–720.
71. Ghosh, P.; Han, G.; De, M.; Kim, C.; Rotello, V. Gold nanoparticles in delivery
applications. Advanced Drug Delivery Reviews 2008, 60, 1307–1315.
111

72. Jeong, J. H.; Mok, H.; Oh, Y.-K.; Park, T. G. siRNA Conjugate Delivery Systems.
Bioconjugate Chemistry 2009, 20, 5–14.
73. Couto, L. B.; High, K. A. Viral vector-mediated RNA interference. Current Opinion in
Pharmacology 2010, 10, 534–542.
74. Lam, J. K.-W.; Liang, W.; Chan, H.-K. Pulmonary delivery of therapeutic siRNA.
Advanced Drug Delivery Reviews 2012, 64, 1–15.
75. Ozpolat, B.; Sood, A. K.; Lopez-Berestein, G. Nanomedicine based approaches for the
delivery of siRNA in cancer. Journal of Internal Medicine 2010, 267, 44–53.
76. Connor, E. E.; Mwamuka, J.; Gole, A.; Murphy, C. J.; Wyatt, M. D. Gold Nanoparticles
Are Taken Up by Human Cells but Do Not Cause Acute Cytotoxicity. Small 2005, 1, 325–
327.
77. Huang, X.; Jain, P. K.; El-Sayed, I. H.; El-Sayed, M. A. Gold nanoparticles: interesting
optical properties and recent applications in cancer diagnostics and therapy. Nanomedicine
2007, 2, 681–693.
78. Stone, J. W.; Sisco, P. N.; Goldsmith, E. C.; Baxter, S. C.; Murphy, C. J. Using Gold
Nanorods to Probe Cell-Induced Collagen Deformation. Nano Letters 2007, 7, 116–119.
79. Sonavane, G.; Tomoda, K.; Makino, K. Biodistribution of colloidal gold nanoparticles after
intravenous administration: Effect of particle size. Colloids and Surfaces B: Biointerfaces
2008, 66, 274–280.
80. Ding, H.; Yong, K.-T.; Roy, I.; Pudavar, H. E.; Law, W. C.; Bergey, E. J.; Prasad, P. N.
Gold Nanorods Coated with Multilayer Polyelectrolyte as Contrast Agents for Multimodal
Imaging. Journal of Physical Chemistry C 2007, 111, 12552–12557.
81. Sokolov, K.; Follen, M.; Aaron, J.; Pavlova, I.; Malpica, A.; Lotan, R.; Richards-Kortum,
R. Real-time vital optical imaging of precancer using anti-epidermal growth factor receptor
antibodies conjugated to gold nanoparticles. Cancer Res. 2003, 63, 1999–2004.
82. Gensler, H.; Sheybani, R.; Li, P.-Y.; Mann, R. L.; Meng, E. An implantable MEMS
micropump system for drug delivery in small animals. Biomedical Microdevices 2012, 14,
483–496.
83. Gensler, H.; Sheybani, R.; Li, P.-Y.; Lo, R.; Zhu, S.; Yong, K.-T.; Roy, I.; Prasad, P. N.;
Masood, R.; Sinha, U. K.; Meng, E. Implantable MEMS drug delivery device for cancer
radiation reduction. In Proc. IEEE Int. Conf. MEMS; Hong Kong, SAR, China, 2010; pp.
23–26.
84. Lo, R.; Li, P.-Y.; Saati, S.; Agrawal, R. N.; Humayun, M. S.; Meng, E. A passive MEMS
drug delivery pump for treatment of ocular diseases. Biomedical Microdevices 2009, 11,
959–970.
85. Lo, R.; Meng, E. A Modular Heat-Shrink-Packaged Check Valve With High Pressure
Shutoff. Journal of Microelectromechanical Systems 2011, 20, 1163–1173.
86. Massey, L. Permeability properties of plastics and elastomers : a guide to packaging and
barrier materials; 2nd ed.; Plastics Design Library/William Andrew Pub.: Norwich NY
USA, 2003.
87. Migliori, M.; Gabriele, D.; Di Sanzo, R.; de Cindio, B.; Correra, S. Viscosity of
Multicomponent Solutions of Simple and Complex Sugars in Water. Journal of Chemical
& Engineering Data 2007, 52, 1347–1353.
88. Klingensmith, M. E.; Washington University (Saint Louis, M. ). D. of S. The Washington
manual of surgery; Wolters Kluwer Health/Lippincott Williams & Wilkins: Philadelphia,
2008.
112

89. Gensler, H. M.; Sheybani, R.; Meng, E. A MEMS micropump system with one-way valve
for chronic drug delivery. In; Transducer Research Foundation, Inc.: Marina Del Rey, CA,
2013.
90. Wilson, J. F. Mechanics of bellows: A critical survey. International Journal of Mechanical
Sciences 1984, 26, 593–605.
91. Di Giovanni, M. Flat and corrugated diaphragm design handbook; M. Dekker: New York,
1982.
92. Yang, X.; Tai, Y.-C.; Ho, C.-M. Micro bellow actuators. In Proc.; IEEE: Chicago, USA,
1997; Vol. 1, pp. 45–48.
93. Minami, K.; Morishita, H.; Esashi, M. A bellows-shape electrostatic microactuator. Sensors
and Actuators A: Physical 1999, 72, 269–276.
94. Bonanomi, G.; Rebello, K.; Lebouitz, K.; Riviere, C.; Di Martino, E.; Vorp, D.; Zenati, M.
A. Microelectromechanical systems for endoscopic cardiac surgery. J Thorac Cardiovasc
Surg 2003, 126, 851–852.
95. Guo-Hua Feng; Eun Sok Kim Universal concept for fabricating micron to millimeter sized
3-D parylene structures on rigid and flexible substrates. In Proc.; Kyoto, Japan, 2003; pp.
594–597.
96. Luharuka, R.; Wu, C.-F.; Hesketh, P. J. Design, fabrication, and testing of a near constant
pressure fuel delivery system for miniature fuel cells. Sensors and Actuators A: Physical
2004, 112, 187–195.
97. Liu, H.; Pi, X.; Zhou, C.; Zheng, X.; Whensheng Hou; Zhiyu Wen Design of site specific
delivery capsule based on MEMS. In Proc.; 2008; pp. 498–501.
98. Takashima, K.; Noritsugu, T.; Rossiter, J.; Guo, S.; Mukai, T. Development of curved type
pneumatic artificial rubber muscle using shape-memory polymer. In Proc.; Tokyo, Japan,
2011; pp. 1691–1695.
99. Shapiro, Y.; Wolf, A.; Gabor, K. Bi-bellows: Pneumatic bending actuator. Sensors and
Actuators A: Physical 2011, 167, 484–494.
100. Gensler, H. M.; Meng, E. Rapid fabrication and characterization of MEMS Parylene C
bellows for large deflection applications. Journal of Micromechanics and
Microengineering 2012, 22, 115031.
101. Matsui, S. Three-dimensional nanostructure fabrication by focused-Ion-beam chemical
vapor deposition. In Proc. IEEE Int. Conf. TRANSDUCERS; Boston, USA, 2003; Vol. 1,
pp. 179–181 vol.1.
102. Ugural, A. C. Stresses in beams, plates, and shells; CRC Press: Boca Raton, 2009.
103. Metz, P.; Alici, G.; Spinks, G. M. A finite element model for bending behaviour of
conducting polymer electromechanical actuators. Sensors and Actuators A: Physical 2006,
130-131, 1–11.
104. Kim, K. H.; Yoon, H. J.; Jeong, O. C.; Yang, S. S. Fabrication and test of a micro
electromagnetic actuator. Sensors and Actuators A: Physical 2005, 117, 8–16.
105. Wolgemuth, L. Innovative Conformal Coating Technologies for Advanced Medical
Devices 2012.
106. Shih, C. Y.; Harder, T. A.; Tai, Y. C. Yield strength of thin-film parylene-C. Microsystem
Technologies 2004, 10, 407–411.
107. Sim, W.; Kim, B.; Choi, B.; Park, J.-O. Theoretical and experimental studies on the
parylene diaphragms for microdevices. Microsystem Technologies 2005, 11, 11–15.
113

108. Lin, J. C.-H.; Lam, G.; Tai, Y.-C. Viscoplasticity of parylene-C film at body temperature.
In Proc.; IEEE: Paris, France, 2012; pp. 476–479.

114

APPENDIX A: Electrode Fabrication Process Flow

115

APPENDIX B: 2
nd
Generation Wireless Micropump System Specifications
Component Dimensions/Volume/Mass Material(s)
Reservoir base
21.6 mm Diameter, lower portion
Injection molded 20-melt polypropylene (not Class VI, sample from
Chase Plastics) or PETG (Class VI, also from Chase Plastics)
19.6 mm Diameter, raised portion
1.1 mm Height, lower portion
1.6 mm Height, raised portion
12 x 17 mm Milled slot
0.37 / 0.47 g Mass (with / without slot)
Electrode
8 mm Outer diameter of metal
Pt/Ti on glass, Nafion
®
-coated
11 mm x 16 mm Substrate length x width
0.26 g Mass before Nafion®-coating
0.28 g Mass after Nafion®-coating
0.45 g Mass, Nafion®-coated and with soldered wires
Wires 0.17 Mass 13 cm wires + solder 30G wire-wrap wire, solder, marine epoxy (Loctite) seal
Bellows
6 mm - 9 mm ID-OD
Parylene C, Class VI 730 unfilled (EPO-TEK
®
) biocompatible epoxy,
marine epoxy (Loctite) seal over biocompatible epoxy
0.4 mm Layer height
13.5 µm Wall thickness
2 Convolutions
0.005 / 0.105 g Mass (empty / filled with ~100 µL water)
Adhesive rings
8 mm Inner diameter
2 adhesive rings (laser-cut Tape 415, 3M) 10 mm Outer diameter
0.007 g Mass
Reservoir dome
21.6 mm Outer diameter
Injection molded 20-melt polypropylene (not Class VI, Chase Plastics) or
PETG (Class VI, Chase Plastics)
19.6 mm Inner diameter
5.8 mm / 4.8 mm Height (outside/ inside)
1.0 mm Wall thickness
0.84 / 0.69 g Mass (with / without septum and catheter)
Refill port
3.6 mm Hole diameter Filled with pre-cured silicone rubber (Class VI MDX-4 4210, Factor II),
marine epoxy (Loctite) seal 2.5 - 3.0 mm Height of septum
Catheter
25 mm Length (volume of ~5.1 µL)
Silicone tubing (Class VI, VWR), marine epoxy (Loctite) seal to reservoir
wall
0.020” Inner diameter (0.508 mm)
0.037” Outer Diameter (0.940 mm)
0.016 g Mass (25 mm long)
Assembled system
(epoxied, no
wireless)
8.3 / 7.4 mm Height with / without epoxy
Marine epoxy (Loctite)
21.6 mm Diameter
1.3 mL Fill volume (DI water)
2.3 / 3.6 g Mass (empty/ filled with DI water)

116

APPENDIX C: Bellows Fabrication Protocol
The bellows fabrication is a two-part lost wax-like molding procedure. First, make molds out of
silicone rubber sheets with punched holes (representing inner and outer diameter of the bellows).
Second, use these sheets to make polyethylene glycol (PEG) templates. The PEG templates are
then coated with Parylene C. After coating, the PEG sacrificial material is released with room
temperature deionized water.

Materials/Supplies
Flat glass mask plates (Nanofilm)
Polyester (green) tape (3M Tape 8403)
Polydimethylsiloxane (PDMS) (Sylgard 184, Fisher Scientific)
Squeegee
Polyethylene glycol (PEG Mn 1000, Alfa Aesar via VWR International)
Glass beaker for molten PEG
3 mL syringe
Coring needle (~23G)
Large plastic petri dishes, 150 mm diameter
Glass slides, 1”x3”
Parylene C dimer (Specialty Coating Systems)
Glass beaker for dissolving PEG out of Parylene C bellows
DI water

Equipment
Digital balance
Thinky Mixer
Desktop vacuum
Oven
Cutting board
Exacto knife
Caliper (to measure PDMS thickness)
Hole punches (C.S. Osborne Co. via McMaster)
PDS 2010 deposition system (SCS)

117

Methods

I. Fabrication of reusable silicone rubber molding sheets

Each set of PDMS molding sheets will make either a “bottom”, a “middle”, or a “top” module. A
bottom+top forms a bellows with 1 convolution. A bottom+middle+top forms a bellows with 2
convolutions. Additional middle sets would allow even more convolutions, and omission of the
top module will allow for increments of 0.5 convolutions.

1. Clean or obtain an extremely flat glass plate (photolithography mask plates work well).
Measure the thickness of the glass plate along several locations on each side.
2. Carefully line the edges of the plate with thin (~1/4” wide) strips of green polyester tape
to make a frame. Make sure there are no large gaps through which the PDMS could leak
out. Approximately 6-7 smooth layers will yield a PDMS thickness of 0.4 mm. Measure
with calipers to verify the thickness (subtracting the glass plate only thickness to obtain
tape thickness.)
3. Weigh out Sylgard 184 in a 10:1 base:curing agent ratio and mix using the Thinky
Mixer.
4. Pour the uncured Sylgard 184 into the frame. There are several ways to spread the
PDMS:
a. wait for gravity to spread it out evenly (less bubbles)
b. spread carefully with a spatula (usually introduces some bubbles)
c. pour large amount on one end and use squeegee to spread to the other side (best
option if you have a squeegee or extremely flat edge)
5. Place the filled frame into the desktop vacuum and keep under vacuum for
approximately 30-45 minutes, or until the air bubbles are mostly removed.
6. Place the degassed and filled frame into the oven and cure at 80 C for approximately 1
hour or until the silicone rubber is no longer tacky. Beware of unlevel shelves - uncured
PDMS has a low viscosity and over the course of several minutes can run and
accumulate to one side, resulting in non-uniform thickness.
7. Remove cured PDMS frame from the oven.
8. Cut around the perimeter of the silicone rubber and remove from the frame.
9. The tape frame may be reusable if cleaned.
10. Cut the large silicone rubber sheets into smaller squares as shown in Figure C-1. Add 10
mm to the outer diameter of the bellows to determine the width of the sheet (5 mm on
either side for easier handling and stacking.)

118

Figure C-1. After verifying that the silicone rubber sheet thickness is appropriate, squares
will be cut out and holes punched according to the bellows dimensions.

11. Sandwich the silicone rubber sheet between two glass slides. Verify the silicone rubber
sheet thickness in the middle and towards each edge with a caliper. If the thickness
varies more than 10% (e.g. one end is 0.37 mm and the other is 0.43), set aside or discard
the sheet.
12. Store the cut and measured sheets in a covered petri dish to keep clean.
13. Label/track the thickness and variation (e.g. 0.40 mm ± 0.02 mm) of each sheet.

II. Cutting and punching holes into silicone rubber molding sheets
1. Use the metal punches to make holes in the sheets representing the inner and outer
diameter of the bellows.
2. Punch holes by laying the sheet on a clean cutting board. Punching on the desk or
another hard surface could damage the edge of the punch and result in jagged cuts and
tears in the PDMS.
3. Label/track the punched hole diameter with the sheet’s label.

III. Stacking silicone rubber modules
1. Stack the silicone rubber sheets according to the diagram in Figure C-2. (green=polyester
tape, light blue=PDMS, gray=PEG)
2. Align the holes to be as concentric as possible.

Figure C-2. Stacked PDMS sheets without PEG for bottom and middle molds

IV. Making the PEG template

1. Line the oven floor with a TexWipe in case of PEG drips.
Bottom Module Middle Module Top Module
Polyester
tape
PDMS
Glass slide
119

2. Preheat the oven to exactly 50 °C. If the temperature deviates more than 5 °C, you will
have issues with the PEG leaking between PDMS sheets or not filling the molds
completely.
3. Add PEG Mn 1000 to a glass beaker or jar. Put it in the oven to start melting. Keep
covered as much as possible because PEG is extremely hygroscopic (absorbs moisture
from the air).
4. Also place the silicone modules and syringe+coring needle for PEG filling in the oven to
warm up to 50 °C.
5. Put on gloves. Do not handle the slides or the PDMS without wearing gloves because
they need to be kept as clean as possible. PDMS is very sticky and gets dirty easily.
6. Begin by adhering the green polyester tape to glass slides, smoothing out any bubbles.
There should be space for three 9 mm outer diameter bellows on a single 1”x3” glass
slide. Let the green polyester tape hang over the edge and fold to make a tab. This makes
it easier to move the bellows slide after Parylene coating.
7. Fill the molds with PEG Mn 1000 until it is slightly convex (the liquid surface just
barely bulges up). Let the PEG fill underneath the convolution and wick around to avoid
trapping air bubbles.
8. Make sure no bubbles get into the convolutions. Use a small pipette tip to suck out any
bubbles that get stuck in the convolutions or redo that mold.
9. Cover each stack with a solid piece of PDMS to ensure a flat top surface on each
module.
10. Remove the filled molds from the oven and let cool. Note: clean the silicone rubber
molding sheets and store in a covered container for future use.
11. Carefully peel away the PDMS sheets from the hardened PEG.
12. Use a clean, slightly damp mini paintbrush to moisten the top surface of the different
modules. Then stack the PEG modules to generate a PEG bellows template.

Figure C-3. Stack PEG modules to from the template of the bellows for Parylene C
coating (shown left to right: 1, 2, and 3 convolution templates of varying dimensions).

13. Store the stacked PEG in covered petri dishes in the fridge to prevent them from
absorbing too much water (PEG is very hygroscopic). Add to the petri dish a KimWipe
filled with 7-10 dessicant rocks if necessary.
14. If PEG looks “moist” immediately (within a half hour), the PEG has likely expired or
absorbed too much moisture.
15. Coat the templates immediately.

V. Coating the PEG template
1. Coat with Parylene C.
2. Use a razorblade to cut a large square perimeter around each bellows.
120

3. Soak PEG-filled Parylene-coated bellows slides into a beaker of MilliQ water for 2 hours
to dissolve out PEG. Do not use DI water from the general lab sink or from the fume
hoods, as it contains trace contaminants that could ruin the actuator
4. Rinse the bellows three times with MilliQ water and store in a covered container.

VI. Cleanup
1. Rinse PDMS molding sheets with MilliQ water, dry with KimWipes (TexWipes leave
lint), and store in a covered container.
2. Also rinse and dry the slides used for stacking the middle and top modules. Store in a
covered container until the next use.
3. Discard the slides used for making the bottom modules (the slides coated in Parylene.)
4. Discard used PEG. Make sure stock PEG bottle is parafilmed and stored in a cool, dry
place.

121

APPENDIX D: 3
rd
Generation Wireless Micropump System Specifications
Component Dimensions/Volume/Mass Material(s)
Reservoir base 20.00 mm Length (parallel to electrode legs) High-resolution stereolithography-made DSM Somos WaterShed XC
11122 (Class VI, FineLine Prototyping); Coated with 5 µm Parylene C
(Class VI, SCS Coatings) to reduce moisture absorption
15.00 mm Width
1.50 mm Corners radius of curvature
1.00 mm Height, base
0.55 mm Depth, slot
17.50 x 12.00 Slot L x W
0.196 / 0.207 g Mass (without / with 5 µm Parylene)
Electrode 11 mm x 16 mm Substrate length x width *Refer to 2nd generation specifications for more details
0.26 / 0.28 g Mass without / with Nafion® coating
Bellows 6 - 9.5 mm, 2 ID-OD, number of convolutions Parylene C, biocompatible epoxy (Class VI, 730 unfilled, EPO-TEK
®
),
marine epoxy (Loctite) seal 0.4 mm Layer height
13.5 µm Wall thickness
0.005 / 0.105 g Mass (empty / filled with 100 µL water)
Reservoir dome 11.50 mm Outer diameter, 1.5 mm fillet High-resolution stereolithography-made DSM Somos WaterShed XC
11122 (Class VI, FineLine Prototyping); Coated with 5 µm Parylene C
(Class VI, SCS Coatings) to reduce moisture absorption
10.50 mm Inner diameter, 1.0 mm fillet
7.10 , 6.10 mm Outside, inside height
0.50 mm Wall thickness (except reservoir ceiling)
1.00 mm Reservoir ceiling thickness
0.400 g Mass (without septum and catheter)
Refill ports (built-in) 5.0 mm Outside diameter, overlaps edge of dome Filled with silicone rubber (Class VI MDX-4 4210, Factor II, Lakeside,
AZ), marine epoxy (Loctite) seal
4.0, 1.2 mm Diameter, depth of extraction port septum
4.0, 2.8 mm Diameter, depth of introduction port septum
0.08 g Septa pair with marine epoxy seal
Catheter 25 mm Length (volume of ~5 µL) Polyurethane (Class VI, SAI Infusion Technologies)
0.025”, 0.040" Inner, outer diameter
0.015 g Mass (25 mm long)
Valve 0.01 g Mass 2.0 mm duckbill valve (Class VI silicone, Minivalve)
Wireless 0.22 g Receiver circuit without epoxy potting *Refer to assembly chart materials/parts list for wireless component
details 0.47 g Receiver coil
Assembled system
(epoxied, wireless,
before external
reservoir coatings)
8.1 mm Height (7.1+1.0) Marine epoxy (Loctite); biocompatible epoxy (Class VI, 730 unfilled,
EPO-TEK
®
) 11.5 / 20 x 15 mm Diameter of dome / L x W of base
~450 µL Reservoir fill volume (DI water)
2.2 / 2.7 g Mass (empty/ filled with DI water)
Assembled wireless
system coatings
0.4 g
0.1 g
0.4 to 0.5 g
Mass epoxy seal
Mass Parylene C
Marine epoxy (Loctite)
Parylene C (Class VI, SCS Coatings)
MDX-4 4210 (Class VI, Factor II)
Mass silicone encapsulation
122

APPENDIX E: 3
rd
Generation Wireless Micropump System Drawings

123

APPENDIX F: 3
rd
Generation Micropump System Assembly Chart and Bill
of Materials

124

125

126

Materials
M01 Soda lime glass wafer (Mark Optics)
M02 Titanium (99.999% Ti, International Advanced Materials)
M03 Platinum (99.999% Pt, International Advanced Materials)
M04 Nafion® (Ion Power); Note: do not exceed 40 degrees C after curing
M05 Conductive epoxy (H20, EPO-TEK®)
M06 Biocompatible epoxy (730 unfilled, EPO-TEK®)
M07 Silicone rubber, not Class VI (Sylgard 184, Fisher Scientific)
M08 Double-sided adhesive tape (Tape 415, 3M)
M09 Polyethylene glycol (PEG Mn 1000, VWR)
M10 Parylene C dimer (SCS)
M11 MilliQ water (Millipore)
M12 Marine epoxy (Loctite)
M13 Silicone rubber, Class VI (MDX4-4210 or A-103, Factor II, Lakeside, AZ)
M14 Polyurethane tubing, Class VI (0.025" ID x 0.040" OD, SAI Infusion)
M15 Silicone tubing, Class VI (0.040" ID x 0.085" OD, VWR)
M16
Polytetrafluoroethylene (PTFE) tubing,
Class VI (12 AWG ID thin-walled extruded tubing, Zeus, Inc.)
M17 Nickel plated copper wire (30 AWG Kynar Wire-Wrap Wire, Digikey)
M18 Flexible printed circuit board (Flex PCB, Gold Phoenix)
M19 Litz Wire 54/50, for receiver (Wiretron)
M20 Solder (Lead or paste)
M21 Silver epoxy (H20E, EPO-TEK®)
M22 Copper wire (20G, for transmitter, Digikey or McMaster)
M23 Acrylic (CG coil) - not reusable
M24 Biocompatible superglue (MG 30, Adhesive Systems, Frankfort, IL)

Disposables

D01 DI water
D02 Double sided Scotch tape (665, Office Depot)
D03 Photoresist (AZ1518IN, AZP4400; Capitol Scientific, Austin ,TX)
D04 Acetone (VWR)
D05 Isopropyl alcohol (VWR)
D06 Polyester tape (Green, 3M 8403)
D07 Glass slides (VWR)
D08 Flux (Flux Pen, Amazon)
D09 1,1,1,3,3,3-hexamethyldisilazane (HMDS; MP Biomedicals, LLC, Solon, OH)
D10 Scotch Tape (Office Depot)

Parts
P01 Photomasks for electrode
127

P02 Electrode
P03 Bellows
P04 Adhesive rings (2 per actuator)
P05 Reservoir dome (WaterShed, FineLine Prototyping)
P06 Septum
P07 Reservoir base (WaterShed, FineLine Prototyping)
P08 Catheter
P09 Valve sleeve tubing
P10 Catheter sleeve tubing
P11 Valve (2.0 mm duckbill, Minivalve)
P12 Wireless transmitter coil
P13 Wireless receiver coil
P14 Receiver circuit printed circuit board
P15 Current regulator (1, LM334SM, Digikey)
P16 Schottky diode (1, BAT54A, Digikey)
P17 Schottky diode (1, BAT54C, Digikey)
P18 Tuning capacitors (1-4, depends on coil, Digikey)
P19 Resistors (2, depends on current, Digikey)

*Note: does not include transmitter circuitry

Subassemblies
S01 Wired & coated electrode
S02 Actuator on base (no bellows)
S03 Reservoir dome with septa
S04 Reservoir dome with septa, valve & catheter
S05 Wireless receiver PCB
S06 Wireless transmitter PCB
S07 Receiver
S08 Transmitter
S09 Dome with integrated valve, catheter, bellows
S10 Assembled reservoir

Fixtures
F01 Septa mold
F02 Assembly fixture (TG coil) - reusable
F03 Centrifuge tube
F04 Silicone rubber sheets
F05 Bellows testing jig

Manufacturing Standard Operating Procedure
MSOP-01 Electrode Fabrication
128

MSOP-02 Wire Attachment
MSOP-03 Nafion® Coating
MSOP-04 Adhesive Rings Fabrication
MSOP-05 Bellows Fabrication
MSOP-06 Septa Fabrication
MSOP-07 Bellows Attachment to Dome
MSOP-08 Septa Insertion
MSOP-09 Reservoir Dome Coating
MSOP-10 Reservoir Assembly
MSOP-11 Valve & Catheter Assembly into Dome
MSOP-12 Coil Fabrication (Receiver)
MSOP-13 PCB Population
MSOP-14 Coil Tuning
MSOP-15 Electronics Insulation
MSOP-16 Receiver Attachment to Device
MSOP-17 Tubing Preparation
MSOP-18 Coil Fabrication (Transmitter)
MSOP-19 Current Setting (Receiver)
MSOP-20 Implant Coating and Encapsulation

Tests / Inspections

T-01 Bellows Inspection
T-02 Electrode Inspection (incl. scotch tape)
T-03 Electrode Functional Test
T-04 Coated Electrode Functional Test
T-05 Valve Functional Test
T-06 Wireless Functional Test (receiver)
T-07 Wireless Functional Test (transmitter)
T-08 Wireless Functional Test (together)
T-09 Bellows Assembled Functional Test
T-10 Wired Fully Assembled Functional Test
T-11 Wireless Fully Assembled Functional Test

Abstract (if available)

Abstract The manner in which a drug is delivered to the body plays a major role in its efficacy. There are several implantable pump technologies for chronic drug administration in humans and large animals. However, development of new drug therapies generally involves initial evaluation via human disease models in small animals such as mice. The technology for chronic drug administration in mice is currently limited to constant flow rates determined at the time of pump manufacture. With the emerging demand for personalized medicine, there is a need for drug administration technology at the early stages of drug development that offers flexibility in dosing schedules. Such a technology would create new opportunities in drug research that is not possible with currently available tools. ❧ This dissertation describes the development of a wireless implantable micropump system for mice that allows flexible dosing to be performed post-implantation. The first chapter begins with a review of the current state-of-the-art drug administration technology for rodents and its limitations. It then introduces the wireless micropump system, its advantages over current technology, and an application in anti-cancer drug delivery that drove its design. The second chapter elaborates the details of the specific needs of the anticancer drug and demonstrates a wired micropump prototype that utilizes electrochemical actuation. The preliminary benchtop tests and in vivo studies with the first generation system elucidate the need for better packaging, valve control, and wireless power to enable chronic drug delivery. The third chapter presents the second generation system, which addresses packaging concerns and valve selection and evaluation. It concludes with a demonstration of the micropump with a wireless power source. The fourth chapter presents the mechanical characterization of the bellows component of the actuator and its effect on the overall system configuration and design. The fifth chapter brings together the advancements from the first two generations into a fully integrated and wirelessly powered implantable system. Characterization of this third generation system shows its ability to perform the intended drug dosing schedule reliably in the environmental conditions that would occur in vivo.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Parylene-based implantable interfaces for biomedical applications

PDF

Implantable bioMEMS drug delivery systems

PDF

Wireless electrochemical drug delivery micropump with fully integrated electrochemical dose tracking feedback system

PDF

Development of implantable Parylene-based MEMS technologies for cortical applications

PDF

Parylene C bioMEMS for implantable devices with electrochemical interfaces

PDF

Integrated wireless piezoelectric ultrasonic transducer system for biomedical applications

PDF

Strategies for improving mechanical and biochemical interfaces between medical implants and tissue

PDF

Chitosan nanoparticle mediated oral drug delivery of kidney targeting micelles for polycystic kidney disease

PDF

Parylene-based biomems sensors for multiple physiological systems

PDF

Development of fabrication technologies for robust Parylene medical implants

PDF

Next generation neural interfaces for neural modulation and prosthesis

PDF

High-resolution data acquisition with neural and dermal interfaces

PDF

The electrochemical evaluation of Parylene-based electrodes for neural applications

PDF

A percutaneously implantable wireless neurostimulator for treatment of stress urinary incontinence

PDF

Towards a high resolution retinal implant

PDF

Piezoelectric ultrasonic and acoustic microelectromechanical systems (MEMS) for biomedical, manipulation, and actuation applications

PDF

Thin-film impedimetric sensors for chronic in vivo use: design and application to hydrocephalus treatment

PDF

Electronics design and in vivo evaluation of a wirelessly rechargeable fetal micropacemaker

PDF

Understanding anti-angiogenic signaling and treatment for cancer through mechanistic modeling

PDF

Liquid flow rate sensing via solution resistance for in vivo applications

Asset Metadata

Creator Gensler, Heidi Marie (author)

Core Title A wireless implantable MEMS micropump system for site-specific anti-cancer drug delivery

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 05/21/2014

Defense Date 10/15/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag bellows electrochemical actuator,drug delivery,implants,MEMS,micropump,OAI-PMH Harvest,Parylene C

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Meng, Ellis (committee chair), D'Argenio, David Z. (committee member), Shiflett, Geoffrey R. (committee member), Yen, Jesse T. (committee member), Zhou, Qifa (committee member)

Creator Email heidigensler@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-349546

Unique identifier UC11295437

Identifier etd-GenslerHei-2174.pdf (filename),usctheses-c3-349546 (legacy record id)

Legacy Identifier etd-GenslerHei-2174.pdf

Dmrecord 349546

Document Type Dissertation

Format application/pdf (imt)

Rights Gensler, Heidi Marie

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA