University of Southern California Dissertations and Theses

Parylene-based implantable interfaces for biomedical applications

(USC Thesis Other)

Parylene-based implantable interfaces for biomedical applications

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
PARYLENE-BASED IMPLANTABLE INTERFACES FOR
BIOMEDICAL APPLICATIONS

by

Angelica Maria Cobo

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 2017

Copyright 2017 Angelica Maria Cobo
I

To my mother

II

ACKNOWLEDGEMENTS
My life has been profoundly influenced and shaped by my family, friends, and
mentors. I have been very fortunate to have many of them helped me overcome my
challenges and celebrate my success. It is impossible to name all the people that have
helped me achieve my goals and become the person that I am today, nonetheless I will try.
First and foremost, I thank my mother to whom this dissertation is dedicated. Thank
you for making sure I had a loving and caring home. You sacrificed a lot so that my sister
and I could have the opportunities that you never had. Your determination, strength, work
ethic, and perseverance inspire me every day. You taught me that knowledge is a person’s
greatest tool and helped me stay focused on my education. To my sister, who I love and
look up to. Who has helped me grow and lifted my spirit. Thank you for your listening ear
and encouragement. To the rest of my family, thank you for all your love, support, and
faith in me.
To Shayne, my husband, my love, my friend, thank you for being by my side all
this time and supporting me through the best and worst of times. You have always believed
in me even in the moments when I doubted myself. Thank you for your patience, love, and
for celebrating my success.
I would like to acknowledge Dr. Antonello Pileggi, Dr. Cherie Stabler, and Dr.
Christopher Fraker, who I had the fortune to worked with during my undergraduate career.
I thank Dr. Pileggi and Dr. Stabler for accepting me into their laboratories and for allowing
me to prove myself in the challenging field of research. Under the guidance of my then
mentor Dr. Fraker, I discovered my passion for research and solidified my desire to obtain
a doctoral degree. Your dedication to research and passion for teaching are admirable.
Thank you for believing in me and challenging me to become better.
I owe a very special thanks to Dr. Ellis Meng. I still cannot believe I was so
fortunate to have you as my graduate advisor and mentor. Thank you for your guidance,
patience, and support along the Ph.D. path. Your dedication, work ethic, knowledge,
III

creativity, and brilliance make you the perfect role model. You have been an inspirational
mentor and a strong female influence in my life. I will always be grateful to you for
enriching my life with skills and knowledge.
To the members of the Biomedical Microsystems Laboratory (BML), who have
become more like a family to me, thank you for all your support and friendship. I owe a lot
of my success to the collaborative and caring environment of our laboratory. Dr. Heidi Tu,
thank you for welcoming me into lab and guiding me in my first project. You were an
example of great organization skills and hard work. Dr. Roya Sheybani, thank you for being
the best project partner anyone could ask for and a dear friend. I admire your passion for
research and strong work ethic. Dr. Jonathan Kuo, you were an invaluable source of
microfabrication and BML history. Thank you for your honesty and for being a foodie like
me. Dr. Brian Kim, your enthusiasm for research and confidence in our success were great
sources of encouragement. I miss our early morning music sessions and your high-spirited
personality. Dr. Lawrence Yu, thank you for all the technical support, for always reminding
me to do the “good old restart” to fix most of my computer problems, and for being a great
desk neighbor. Dr. Seth Hara, your deep knowledge in electrochemical testing and your
willingness to help me at all times have been invaluable to my research. Dr. Curtis Lee,
thank you for all the insightful conversations and for being our racquetball coach. Dr.
Christian Gutierrez, thank you for sharing your deep knowledge on drug delivery systems
and helping me find answers to many of my questions. Alex Baldwin, thank you for your
patience when sharing your expertise in electrical engineering concepts. Jessica Ortigoza,
thank you for your kindness and hard work on improving Parylene etching processes. I will
miss our Korean BBQ and Colombian lunches. Ahuva Weltman, thank you for listening
to all my frustrations about fabrication and for always looking at the bright side. I admire
your honesty and strength to stand up for what you believe is right. Dr. Kee Scholten, your
knowledge on almost everything is fascinating. Your expertise and guidance have proven
invaluable to my research. I admire your devotion to research and I look forward to your
future accomplishments. Christopher Larson, your dedication and hard work in the LACE
project have allowed a great amount of progress in a short time. I feel confident that you
IV

will take this project even further. James Yoo, thank you for sharing your expertise in
SolidWorks and machining. You have all necessary qualities to become a great professor.
Trevor Hudson, thank you for your eagerness to always help others in lab. Eugene Yoon,
thank you for your contagious energy and enthusiasm for research. I must also thank the
Game of Thrones/ Walking Dead debriefing groups (Dr. Seth Hara, Chris Jones, Alex
Baldwin, James Yoo, Dr. Kee Scholten, and Dr. Lawrence Yu). Your theories and insight
have made these shows very interesting.
I would like to thank Dr. Victor Pikov from GlaxoSmithKline (GSK) for sharing
his expertise in peripheral nerve recording and stimulating interfaces. Dr. Tuan Hoang,
thank you for managing the lab and for providing great career advice. I would also like to
express my gratitude to all the administrative staff in the Department of Biomedical
Engineering who guided me and supported me during my graduate years. To Mischalgrace
Diasanta, Karen Johnson, Sandra Johns, Daisy Rusli, and Walter Lam, thank you. Dr.
Donghai Zhu, thank you for maintaining the cleanroom facilities and helping me
troubleshoot equipment problems.
Finally, I would like to thank the Viterbi School of Engineering, Alfred E. Mann
Institute (AMI), and Women in Science and Engineering (WiSE) for the fellowships and
awards that supported me through my graduate career. My research was also made possible
through grants from the National Institute of Health (NIH), Wallace H. Coulter Foundation,
and Defense Advanced Research Projects Agency (DARPA). Your financial support
helped me realized my academic potential.
V

TABLE OF CONTENTS
ACKNOWLEDGEMENTS .................................................................................................... II
TABLE OF CONTENTS ...................................................................................................... V
LIST OF TABLES ............................................................................................................. VII
LIST OF FIGURES ......................................................................................................... VIII
ABSTRACT ............................................................................................................ XVIII
MICROFABRICATED IMPLANTABLE DEVICES .............................................. 1
IMPLANTABLE MEDICAL DEVICES .................................................................... 1
MICROELECTROMECHANICAL SYSTEMS ........................................................... 3
FUNDAMENTAL MICROFABRICATION TECHNIQUES .......................................... 3
NON-CONVENTIONAL MICROFABRICATION METHODS ..................................... 6
BIOMEMS ....................................................................................................... 8
BIOMEMS IMPLANTABLE DEVICES ................................................................. 9
OBJECTIVES .................................................................................................... 14
REFERENCES ................................................................................................... 15
WIRELESS DRUG DELIVERY PUMP ........................................................... 19
BACKGROUND ................................................................................................ 19
DESIGN ........................................................................................................... 23
FABRICATION ................................................................................................. 27
EXPERIMENTAL METHODS ............................................................................. 34
RESULTS ......................................................................................................... 39
DISCUSSION .................................................................................................... 46
CONCLUSION .................................................................................................. 50
REFERENCES ................................................................................................... 52
LYSE-AND-ATTRACT CUFF ELECTRODE ................................................. 57
BACKGROUND ................................................................................................ 57
DESIGN ........................................................................................................... 63
FABRICATION ................................................................................................. 68
PACKAGING .................................................................................................... 83
EXPERIMENTAL METHODS ............................................................................. 90
RESULTS ......................................................................................................... 95
DISCUSSION .................................................................................................. 102
CONCLUSIONS .............................................................................................. 105
REFERENCES ................................................................................................. 106
VI

RELIABILITY TESTING OF THE LYSE-AND-ATTRACT CUFF ELECTRODE
TOWARDS IN VIVO IMPLEMENTATION ................................................... 114
BACKGROUND .............................................................................................. 114
LONG-TERM SIMULATED IN VIVO CHARACTERIZATION ................................ 116
SIMULATED AND ACUTE IN VIVO INFUSION ................................................... 129
INTERLOCKING MECHANISM ACUTE IN VIVO TESTING ................................. 138
SUMMARY .................................................................................................... 139
REFERENCES ................................................................................................. 141
CONCLUSION ......................................................................................... 143
APPENDIX A PEDIATRIC DRUG DELIVERY PUMP FOR THE TREATMENT OF
LEPTOMENINGEAL METASTASES ....................................................... 146
MEDICAL APPLICATION ............................................................................................... 146
APPROACH ................................................................................................................... 146
FABRICATION ............................................................................................................... 147
EXPERIMENTAL METHODS ........................................................................................... 152
RESULTS AND DISCUSSION .......................................................................................... 154
CONCLUSION ............................................................................................................... 156
REFERENCES ................................................................................................................ 158
APPENDIX B LYSE-AND-ATTRACT CUFF ELECTRODE SHAM DEVICE PROCESS
FLOW ................................................................................................... 159
APPENDIX C LYSE-AND-ATTRACT CUFF ELECTRODE COMPLETE DEVICE
PROCESS FLOW ................................................................................... 162
APPENDIX D LACE FABRICATION BATCH ......................................................... 166
APPENDIX E SACRIFICIAL PHOTORESIST REMOVAL POST-OXYGEN PLASMA
EXPOSURE ............................................................................................ 167
APPENDIX F REMOVAL OF OXYGEN PLASMA EXPOSED PHOTORESIST ............ 169
APPENDIX G ELECTRICAL CROSS-TALK SYSTEM AND TESTING PROTOCOL .... 171
VII

LIST OF TABLES
Table 2-1. Primary differences between the previous generation refillable implantable
micropump system (RIMS) and our current implantable micropump for drug
delivery in small animals (reprinted from [10] with permission from Elsevier). . 23
Table 2-2. Glucose solutions prepared for viscosity testing. ............................................ 38
Table 3-1. Reference values for Pt electrode impedance as a function of electrode surface
area. ....................................................................................................................... 66
Table 3-2. Microfluidic channel sacrificial photoresist removal method and dissolution
times. ..................................................................................................................... 78
Table 3-3. Oxygen plasma exposed photoresist removal methods. .................................. 80
Table 3-4. Medical grade adhesives tested for securing LACE fluidic connection. ......... 88
Table 3-5. Electrochemical impedance spectroscopy in 1x PBS at 1 kHz (mean ± SE) for
a LACE device in flat, curled, and uncurled configuration. ............................... 101
Table 3-6. Electrochemical impedance spectroscopy in 1x PBS at 1 kHz (mean ± SE).
Partially embedded electrodes are located underneath the fluidic outlet port
(reprinted with permission from [74] © 2017 IEEE). ......................................... 102
Table 4-1. Average diameter of outlet port stain calculated using ImageJ (NIH) software
(n = 3) (modified from [5] © 2017 IEEE). ......................................................... 133
Table 4-2. LACE in vivo infusion of methylene blue dye on the sciatic nerve of
anesthetized rats. *No functional channels post implantation. ........................... 135

VIII

LIST OF FIGURES
Figure 1-1. The 1960 Medtronic implantable pacemaker (reprinted from [2] with
permission from Images in Pediatric Cardiology). ................................................. 2
Figure 1-2. Schematic diagram of the cross-sectional views of bulk and surface
micromachining techniques. ................................................................................... 4
Figure 1-3. Schematic representation of photolithography involving the patterning of
photoresist on polymer. ........................................................................................... 5
Figure 1-4. Cyclical nature of deep reactive ion etching process. Steps 2 to 3 are repeated
until the end of etch................................................................................................. 6
Figure 1-5. Process for creating a PDMS mold by casting. ................................................ 7
Figure 1-6. Chemical structure of Parylene C. ................................................................... 9
Figure 1-7. Photographs of the (a) Utah (Reprinted by permission from Macmillan
Publishers Ltd: Nature Neuroscience [22] © 2002) and (b) Michigan arrays
(reprinted with permission from [23] © 2008 IEEE). ........................................... 10
Figure 1-8. CardioMEMS, Inc wireless microelectromechanical systems implantable
pressure sensor part of the EndoSure® AAA Wireless Pressure Measurement
System (reprinted from [7] with permission from The Royal Society of Chemistry).
............................................................................................................................... 11
Figure 1-9. (a) Schematic diagram of the Argus II electrode array with respect to the treated
eye and (b) the Argus II electrode array implanted in a subject with Retinitis
Pigmentosa (reprinted with permission from [26, 28] © 2011 IEEE). ................. 11
Figure 1-10. Photograph of the refillable micropump system (reprinted from [27] with
permission from Springer). .................................................................................. 11
Figure 1-11. Photograph of a Michigan electrode array with integrated microfluidic
channel (reprinted from [32] with permission from the Journal of Neurosurgical
Focus). .................................................................................................................. 13
Figure 1-12. Photograph of drug delivery micropump with integrated sensors and
electronics (reprinted from [34] with permission from Springer). ...................... 14
Figure 2-1. Schematic diagram of in vivo testing setup for small animal research with (a)
commercial tethered infusion system, and (b) wireless implantable drug delivery
pump (reprinted from [10] with permission from Elsevier). ................................ 20
IX

Figure 2-2. Schematic diagram of wireless drug delivery system (modified from [27] ©
2014 IEEE)............................................................................................................ 23
Figure 2-3. (a) Schematic diagram of the custom-made normally closed spring-loaded ball
check valve (reprinted from [28] with permission from Springer) and (b) Shows
fluid path when the valve is open with a forward pressure. (c) Closed valve in the
presence of reverse pressure or no forward pressure being applied. .................... 24
Figure 2-4. Micropump electrolysis actuation operation concept. (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 expands the bellows and drive
surrounding drug out of the reservoir and catheter. .............................................. 25
Figure 2-5. Circuit diagram for the class E wireless inductive powering system. Link
efficiency for concentric transmitting and receiving coils with zero angular
misalignment was calculated to be approximately 0.82 (modified from [10] with
permission from Elsevier). .................................................................................... 26
Figure 2-6. Photographs of first generation valve check valve components. (a) Valve stop,
(b) valve seat, (c) silicone ball, and (d) compression spring. ............................... 27
Figure 2-7. Photograph of assembled spring-loaded ball check valve. ............................ 28
Figure 2-8. Valve seat with fabricated PDMS O-rings. .................................................... 29
Figure 2-9. Photographs of final check valve check valve components. .......................... 29
Figure 2-10. Photograph of final spring-loaded ball check valves (reprinted from [28] with
permission from Springer). .................................................................................. 29
Figure 2-11. Schematic diagram detailing the electrode and bellows fabrication processes
and bellows actuator assembly.............................................................................. 30
Figure 2-12. Photographs of 5:2 PEG-1000 and hydroresin wax bellows mold fabricated
by (a) injection molding and casting (b) top view and (c) side view. ................... 31
Figure 2-13. Wireless implantable micropump individual components. .......................... 33
Figure 2-14. Photograph of a fully assembled wireless micropump (reprinted from [10]
with permission from Elsevier). ............................................................................ 33
Figure 2-15. Schematic overview of pressure testing setup. ............................................ 35
Figure 2-16. Testing setup for characterizing check valve performance as part of a wired
micropump. ........................................................................................................... 35
X

Figure 2-17. (a) schematic diagram (reprinted from [10] with permission from Elsevier)
and (b) photograph of experimental setup for flow rate characterization of wireless
micropumps........................................................................................................... 36
Figure 2-18. Schematic overview of pressure testing setup. ............................................ 37
Figure 2-19. Experimental set up for room temperature versus body temperature testing.
............................................................................................................................... 39
Figure 2-20. Valve characterization with wired micropump: (a) four 30 μL boluses
delivered after valve opening (micropump was refilled after each run to circumvent
pressure buildup during recombination), (b) valve performance averaged for the
four runs and compared to a valve-less system (modified from [28] with permission
from Springer). ..................................................................................................... 40
Figure 2-21. The effects of (a) axial separation distance between concentric transmitter and
receiver coils, and (b) angular misalignment between concentric coils on pump
performance (at 0.33 mA constant current) (reprinted from [10] with permission
from Elsevier). ...................................................................................................... 41
Figure 2-22. Representative data from one stationary micropump wirelessly operated. (a)
Bolus delivery from a micropump operated once a day during week two (at 0.33
mA constant current) and (b) flow rates for the three weeks dosing (reprinted from
[10] with permission from Elsevier). .................................................................... 42
Figure 2-23. The variation in flow rate observed in four micropumps when pumping against
physiological back pressures (up to 20 mmHg). Flow rate was normalized across
pumps at a back pressure of 0 mmHg (reprinted with permission from [27] © 2014
IEEE)..................................................................................................................... 43
Figure 2-24. The variation in flow rate observed in four micropumps when delivering
calibrated glucose solutions of differing viscosity. Blood viscosity is 3 to 4 cP at
body temperature (1cP = 1mPa∙s). Flow rate was normalized across pumps at a
viscosity of 1 cP (reprinted with permission from [27] © 2014 IEEE). ............... 44
Figure 2-25. Representative data from one wireless micropump flow rate performance
when increasing the environmental temperature from room (23 ºC) to body
temperature (37 ºC) (at 0.33 mA constant current) (reprinted from [10] with
permission from Elsevier). .................................................................................... 45
Figure 2-26. Representative flow rate performance results for a micropump under
simulated in vivo conditions for 30 days (at 0.33 mA constant current) (reprinted
from [10] with permission from Elsevier). ........................................................... 46
XI

Figure 2-27. Photograph of micropump outer casing produced using stereolithography.
Casing components shown (a) individually and (b) assembled. ........................... 47
Figure 2-28. Photographs of commercial Duckbill valve prior to incorporation into our
system. Valve slit shows characteristics of a normally opened valve and variability
among valves. ....................................................................................................... 49
Figure 3-1. A compound action potential at the sciatic nerve of an adult cat recorded with
a silicone cuff electrode. The amplitude of the CAP is shown in voltage (modified
from [7] with permission from John Wiley and Sons). ........................................ 58
Figure 3-2. Schematic diagram of the anatomy of a peripheral nerve. ............................. 59
Figure 3-3. Peripheral nerve interfaces necessary for brain-controlled prostheses. ......... 59
Figure 3-4. (a) Extraneural interfaces wrap around the nerve and record/stimulate from the
nerve surface. (b) Intraneural interfaces penetrate the nerve gaining greater access
to individual nerve fibers. ..................................................................................... 60
Figure 3-5. The two types of electrodes applied to interface peripheral nerves classified
regarding invasivity and selectivity. ..................................................................... 63
Figure 3-6. (a) A photograph showing the branching pattern of the rat sciatic nerve with
the location of where the average diameter was measured indicated by the white
arrow. (b) Cross-section through the rat sciatic nerve, showing the four fascicles:
sural (s), tibial (t), peroneal (p), and cutaneous (c) (modified from [53] with
permission from John Wiley and Sons). ............................................................... 64
Figure 3-7. Schematic diagram of the microfluidic channels showing the branching design
and dimensions...................................................................................................... 65
Figure 3-8. Cross sectional view of Parylene-based PN cuff electrode showing electrode’s
placement within a microfluidic channel. Inter-electrode spacing is not to scale. 66
Figure 3-9. Schematic of LACE for targeting individual fascicles within a nerve. Insert
shows the cross-sectional view. ............................................................................ 67
Figure 3-10. SolidWorks design of LACE electrode for sciatic nerve recording and
stimulation. Units in mm....................................................................................... 67
Figure 3-11. Kic software masks. (a) Complete sham device, (b) locking mechanism
serenated teeth, (c) microfluidic channels with ports, and (d) final sham device
mask layout (100 mm wafer). ............................................................................... 68
Figure 3-12. (a) Cross sectional view of sham devices process flow, and (b) fabricated sham
device with photoresist in microfluidic channels. ................................................. 69
XII

Figure 3-13. SEM images of fabricated sham devices showing different side views of the
microfluidic channels. ........................................................................................... 70
Figure 3-14. Kic software masks. (a) Complete LACE device, (b) locking mechanism,
microfluidic channels with ports, and metal traces, and (c) final LACE devices
mask layout (100 mm wafer). ............................................................................... 71
Figure 3-15. LACE fabrication process which utilizes standard surface micromachining
techniques for Parylene. ........................................................................................ 71
Figure 3-16. Fabricated LACE in which microfluidic channels are highlighted by the
presence of photoresist. ......................................................................................... 72
Figure 3-17. Sputtered Pt features obtained from LGA process. Damaged lift-off
photoresist was identified as the wrinkled areas around the device metal features
(right image).......................................................................................................... 73
Figure 3-18. Devices following metal lift-off of sputtered metal. .................................... 73
Figure 3-19. Cracked Pt features following e-beam deposition. ...................................... 74
Figure 3-20. Successful e-beam driven physical vapor deposition................................... 74
Figure 3-21. Devices following metal lift-off of e-beam metal. (a) “UP” logo, (b) electrodes
and traces, (c) locking teeth and etch stop features, and (d) alignment crosses. .. 75
Figure 3-22. Deposited Parylene insulation layer milky in appearance. (a-b) Magnified
photographs reveal the presence of small clusters in polymer areas that were
intensely swabbed during metal lift off. (c) Absence of milky Parylene in areas of
the polymer that were mildly swabbed. ................................................................ 76
Figure 3-23. Parylene bubbles near fluidic structures and low selectivity of Parylene to
photoresist etch rate during DRIE process resulted in damage to devices. .......... 77
Figure 3-24. Out-gassing of sacrificial photoresist solvent during oxygen plasma etching
in the DRIE. Device from a) wafer 1 and (b) wafer 2. ......................................... 78
Figure 3-25. (a) Sacrificial photoresist release fixture involved hanging each device from
a wire attached to a scaffold above the acetone. Before (b) and after (c) photos of a
device soaked via the wire fixture show no significant damage. .......................... 79
Figure 3-26. Oxygen plasma treated AZ 4620 photoresist residue on LACE metal features.
............................................................................................................................... 80
Figure 3-27. Oxygen plasma exposed AZ 4620 photoresist cleaned with homemade
stripper. Photographs show no photoresist residue. .............................................. 81
XIII

Figure 3-28. Side and top view photographs of Teflon® PTFE jig used to anneal LACE
devices................................................................................................................... 82
Figure 3-29. Absorbance measurements from Fourier transform infrared spectroscopy
(FTIR) analysis of Parylene untreated, and acetone soak and thermoformed under
vacuum. ................................................................................................................. 82
Figure 3-30. Cuff electrode with integrated Parylene cable is attached to a zero insertion
force (ZIF) connector. ........................................................................................... 83
Figure 3-31. Electrical packaging for chronic in vivo animal studies. ............................. 84
Figure 3-32. Percutaneous connector inside surgical tunnel. ........................................... 85
Figure 3-33. Fabrication of fluidic connection at the device level. (a) Microwire insertion
into fluidic channel inlet, followed by (b) PEEK insertion and coating with Silicone
MED-4210. ........................................................................................................... 85
Figure 3-34. SAI custom-made fluidic catheter. ............................................................... 86
Figure 3-35. PU catheter inserted into thermoformed LACE device microfluidic channel
inlet. ...................................................................................................................... 87
Figure 3-36. Sylgard 184 silicone wicked into the channel, blocking flow completely.
When dye was flowed into the device, the blockage caused it to leak out of the inlet
opening. ................................................................................................................. 88
Figure 3-37. The fast cure of Loctite 4902 prevents it from blocking the channel. ......... 88
Figure 3-38. The schematic representation of the LACE acute implantation and
experimental setup. ............................................................................................... 89
Figure 3-39 Schematic representation of the LACE chronic implantation and experimental
setup. ..................................................................................................................... 90
Figure 3-40. Proof of concept testing of adjustable interlocking mechanisms. (a) Buckle
serrated mechanism, (b) scaled 180 µm thick paper cuff, and (c) 12 µm thick
Parylene cuff. ........................................................................................................ 91
Figure 3-41. Complete LACE interlocking mechanism. (a) Suture needle passed through
the needle hole, (b-c) tab threaded through the first slit and kept in place by serrated
teeth, and (d) tab secured by threading through second slit to form the buckle locked
structure................................................................................................................. 92
Figure 3-42. Photograph of the LACE interlocking mechanism serrated teeth. ............... 92
XIV

Figure 3-43. Microfluidics evaluation experimental setup. .............................................. 93
Figure 3-44. Standing curled prototype LACE upright into a custom acrylic jig to image
the transverse progression of the dye with a microscope. .................................... 94
Figure 3-45. Sequential photographs of the infusion experiment at 830 nL/min flow rate in
a (a) flat device and (b) curled device. Image analysis using ImageJ software
confirmed that flow rate through channels was uniform within 6% standard error.
............................................................................................................................... 97
Figure 3-46. Microfluidic channel functionality test in thermoformed sham devices in the
curled orientation. (a) Sham device indicating fluidic channel number. (b) Pinching
occurred along the horizontal channel feeder blocking flow to channel 4, and (c)
uniform flow achieved through channels 1 to 3.................................................... 97
Figure 3-47. 1.37% (w/w) agarose nerve phantom after localization testing with fluorescein
dye. ........................................................................................................................ 98
Figure 3-48. Bursting occurred at a maximum flow rate of 1 μL/min. This leak occurred
near the edge of the inlet, signifying it was likely due to delamination of Parylene
layers. .................................................................................................................... 98
Figure 3-49. Cyclic voltammetry in H2SO4 of single recording site yielded standard Pt
electrode characteristics. The integration of the hydrogen desorption peaks to find
QH and the integration of the cathodic current (Qcathodic) for ESA and CSC are
highlighted, respectively. .................................................................................... 100
Figure 3-50. EIS in PBS of single recording site showed a low impedance (< 2 kΩ) at 1
kHz)..................................................................................................................... 101
Figure 3-51. Solidworks diagram of LACE with fluidic channels. ................................ 101
Figure 3-52. Representative cyclic voltammogram of LACE Pt electrodes in sulfuric acid
following priming. (a) surface electrode with no fluidics, (b) partially embedded
electrode (fluidic outlet port), and (c) fully embedded electrode. ...................... 102
Figure 3-53. Representative electrochemical impedance spectroscopy plots of impedance
magnitude and phase in 1×PBS for a (a) surface electrode with no fluidics, (b)
partially embedded electrode (fluidic outlet port), and (c) fully embedded electrode.
EC measurements were taken with a Gamry Reference 600 potentiostat (reprinted
with permission from [74] © 2017 IEEE). ......................................................... 102
Figure 4-1. Optical micrograph of fabricated LACE (reprinted with permission from [5] ©
2017 IEEE).......................................................................................................... 115
XV

Figure 4-2. LACE devices were epoxied into vial caps for long-term soaking experiments.
............................................................................................................................. 117
Figure 4-3. (a) Data output from the automated cross-talk testing system is represented as
a grid. The values shown here represent ideal results from a perfectly-insulated
device. (b) Electrode numbering for the LACE. SE is the surface electrode not
contained within a microfluidic channel. ............................................................ 119
Figure 4-4. Cross-talk testing was performed on several devices that had been soaking for
approximately 3 weeks to reveal differences in cross-talk development in relation
to soak temperature and thermal annealing. For comparison, cross-talk results for
an un-soaked device are also shown (Dry). Values that are negative, greater than
100, or asymmetrical across the diagonal are attributed to noise. ...................... 120
Figure 4-5. Representative cross-talk data for a device (a) pre, and (b) post cyclic
voltammetry cleaning in 0.05M H2SO4. ............................................................. 121
Figure 4-6. Representative long-term cross-talk data for a LACE device after EIS
measurements (except for "Dry"). Percentage cross-talk value is the average across
all measurements (n = 56). .................................................................................. 122
Figure 4-7. Representative EIS (a) magnitude and (b) phase of an annealed LACE device
long-term soak test at 37° C in 1× PBS. Mean ± SE, n = 8 electrodes. .............. 123
Figure 4-8. EIS (a) magnitude and (b) phase of the un-annealed LACE device (control)
undergoing long-term soak test at 37° C in 1× PBS. Mean ± SE, n = 8 electrodes.
............................................................................................................................. 123
Figure 4-9. EIS magnitude and phase for an annealed and un-annealed LACE device
undergoing soak experiments in 1× PBS at 37 °C. Mean ± SE, n = 8 electrodes. (a-
b) annealed device, and (c-d) un-annealed device. ............................................. 124
Figure 4-10. Optical micrograph of an un-annealed LACE device after 1 day of soaking in
1× PBS at 37 °C shows visual delamination at the small metal traces but not other
features. ............................................................................................................... 124
Figure 4-11. Representative EIS magnitude at 1 kHz for an annealed LACE device soaked
in 1× PBS at 37°C. Colored lines and optical micrographs show the beginning of
observable delamination. .................................................................................... 125
Figure 4-12. Electrodes are exposed using a plasma etching process and leaving a thin
border of Parylene defining the electrode boundary. .......................................... 127
Figure 4-13. Representative optical micrographs of a long-term tested device at day 25
showing delamination limited to metal features subjected to current................. 127
XVI

Figure 4-14. Metal delamination (metal wrinkling) after CV in 1× PBS with a scan rate of
50 mV/sec. Only the tested electrodes showed delamination (n=5 electrodes). . 128
Figure 4-15. Schematic of LACE showing three different microfluidic channel
configurations to support channel integrity when the cuff is curled: (a) Design 1:
standard channels (20 µm H x 250 µm W), (b) design 2: incorporation of mid-
channel supporting walls (20 µm H × 20 µm W × 390 µm L), and (c) design 3:
narrower channels (150 µm and 250 µm W sections) (modified from [5] © 2017
IEEE)................................................................................................................... 130
Figure 4-16. Optical micrographs of LACE showing the two new microfluidic channel
configurations: (a) design 2: incorporation of mid-channel supporting walls, and
(b) design 3: narrower channels. ......................................................................... 130
Figure 4-17. Representative optical micrographs of the infusion experiment in curled
devices at 500 nL/min flow rate (transverse view). The channels were primed with
the dye introduced at the microfluidic inlet channel. (a) Design 1, (b) design 2, and
(c) design 3. Scale bar is 1 mm (reprinted with permission from [5] © 2017 IEEE).
............................................................................................................................. 131
Figure 4-18. (a) Transverse view of LACE wrapped around a nerve phantom. Insert shows
side-view. (b) Representative nerve phantom after localized dye delivery (n =3).
Insert scale bar is 2 mm (reprinted with permission from [5] © 2017 IEEE). ... 132
Figure 4-19. Sequence of photographs illustrating the in vivo implantation for the LACE
around the sciatic nerve of an anesthetized rat. (a) Exposed sciatic nerve, (b) device
fed under the nerve with fluidic ports in contact with the nerve, (c-d) tab threaded
through locking slit, (e) tab kept in place by serrated teeth, and (f) LACE securely
wrapped around the sciatic nerve. Scale bar is 1 mm. ........................................ 134
Figure 4-20. Dye diffused longitudinally along the nerve at a 0.3 µL/hr flow rate. ....... 136
Figure 4-21. (a) Localized dye delivery shown as discrete spots of dye on the surface of
the nerve with a flow rate of 1 µL/hr. (b) Histology showed no dye diffusion into
the nerve. ............................................................................................................. 136
Figure 4-22. (a) Sciatic nerve infused with 15 µL/hr flow rate for 5 minutes showed non-
localized dye coverage of the nerve. (b) Histology results showed dye penetration
into the smallest fascicles.................................................................................... 137
Figure 4-23. Optical micrograph of a LACE device indicating the adjustable interlocking
mechanism features. ............................................................................................ 138
Figure 4-24. Photographs of the LACE placement on the sciatic nerve. (a) Unlocked LACE
on the sciatic nerve, (b) locked LACE on the sciatic nerve, and (c-d) locked LACE
XVII

on the dissected sciatic nerve (reprinted with permission from [5] © 2017 IEEE).
............................................................................................................................. 139

XVIII

Implantable medical devices make possible the treatment of chronic conditions by
providing continuous treatment at or near the affected area. BioMEMS technologies allow
for the development of complex, smaller, and versatile implantable devices. However,
reliable chronic performance in vivo has been limited by poor biocompatibility, mechanical
mismatch to the soft biological tissue, inappropriate packaging, and lack of combined
functionalities within a device (e.g. electrodes, drug delivery, sensors). Fortunately, these
challenges can be overcome by utilizing Parylene C as an encapsulation or structural
material. Parylene-based implantable devices hold promise due to high biocompatibility,
flexibility, good water barrier properties, and compatibility with standard MEMS
fabrication techniques. Presented in this work are two bioMEMS devices for clinical
applications designed to achieve reliable chronic performance in vivo.
Human drug therapies are initially evaluated in small animal models such as
rodents; therefore, drug administration technologies for small laboratory animals are
critically important for drug discovery and development. Chapter 2 presents an implantable
micropump intended for evaluation and development of drug therapies in small laboratory
animals for management of chronic conditions. A MEMS approach was implemented to
allow miniature form factor and completely wireless operation of the micropump. The
micropump’s ability to administer a tailored anti-cancer regimen under wireless operation
was investigated under simulated in vivo conditions.
The potential of peripheral nerve (PN) interfaces has been marred by the limited in
vivo lifetime of current technology due to biological and non-biological failure modes. If
PN interfaces functioned reliably under chronic conditions (in the range of multiple years),
the promise of restoring sensory and motor control in amputees can be realized. Chapter 3
presents a novel Parylene-based peripheral nerve interface for restoring sensory and fine
motor control in amputees. In an effort to improve the reliability of the current state-of-the
ABSTRACT

XIX

art, we leveraged Parylene’s thin, flexible substrate and compatibility with batch
microfabrication techniques with a unique drug delivery system to enhance implant-tissue
integration in vivo and prolong chronic performance. The Parylene-based PN interface was
designed, fabricated, and characterized. Finally, chapter 4 focuses on evaluating and
improving the reliability of the Parylene-based PN interface for chronic in vivo
implementation. The electrical, microfluidic, and mechanical functions of the LACE were
tested and characterized under simulated and acute in vivo conditions to evaluate the
interface potential for achieving successful chronic performance.

1

Implantable Medical Devices
Implantable medical devices make possible the treatment of chronic conditions by
providing continuous treatment at or near the affected area. Implantable devices improve
patient compliance, are portable, and can provide tailored therapies. The first implantable
electronic device dates back to 1960, when Wilson Greatbatch and his team successfully
implanted a cardiac pacemaker in a human [1-3]. This device regulated irregular cardiac
rhythm by providing electrical impulses to the heart. The success of the implantable
pacemaker led scientists and engineers in the 1970s and 1980s to consider the treatment of
other chronic conditions using analogous technologies. Emphasis was placed on
implantable drug delivery systems for the treatment of chronic conditions such as diabetes
and cancer. The drug delivery system consisted of a refillable reservoir controlled by
electronics and powered by a battery for localized drug delivery. Among the first
implantable drug delivery systems was a micropump for controlled insulin delivery for the
treatment of diabetes developed by Thomas and Bessman [4, 5]. Another implantable
device introduced around the same time was an implantable defibrillator developed by Dr.
Michel Mirowski and Medrad/Intec Systems [6]. The defibrillator continuously detected
ventricular fibrillation and stopped it with the application of an electrical shock to the heart.

MICROFABRICATED IMPLANTABLE DEVICES

2

Figure 1-1. The 1960 Medtronic implantable pacemaker (reprinted from [2] with permission
from Images in Pediatric Cardiology).

Collaboration between academia, industry, and the medical community has led to
the development of many new implantable devices since then and the emergence of an
implantable device industry [7]. A market research report estimates the United States
implantable medical devices market to reach approximately $73.9 billion by 2018 [8].
Orthopedic implants account for the largest implantable device segment for this forecast.
Other categories include implantable devices for dental care, hearing impairment, drug
delivery, cosmetics, and treatment of cardiovascular diseases and degenerative
musculoskeletal disorders. The market rise is attributed to the increasing elderly population
and prevalence of chronic degenerative diseases.
Implantable electronic medical devices have unique challenges. Lack of
biocompatibility can limit the device functionality and longevity in the body; Implantable
devices must be made of biocompatible materials that will not degrade in the harsh in vivo
environment and, at the same time, must provide mechanical match to the soft biological
tissue to minimize foreign body response and tissue damage [9]. Electrical components
must be hermetically sealed to avoid electrical shorts or corrosion due to extended exposure
to the moist and corrosive saline environment of the body. Devices must also be
compatible with standard sterilization methods to avoid the presence of microorganisms
on the device prior to implantation. Despite advances in battery technologies, batteries are
still one of the largest components of the system and have limited lifetimes reducing the
operational lifetime of the device [7, 10]. Reliability, accuracy, and safety of these devices

3
is critical to obtaining regulatory clearance by the Food and Drug Administration (FDA)
and ultimately translate the device to patient use. Additionally, lack of combined
functionalities (e.g. sensors, drug delivery, electrodes) of these implants limits their ability
to provide optimal disease control therapy and personalized therapies.
Microelectromechanical systems (MEMS) technologies can provide miniaturization,
wireless power operation, and integration of complex functions to implantable devices
providing new opportunities to create novel implants or improve existing ones.
Microelectromechanical Systems
Many of these implantable devices were made possible with the advent of the
transistor in 1947 by John Bardeen and Walter Brattain from Bell Laboratories [11].
Advancements in the semiconductor industry lead to the miniaturization of transistors and
eventually the development of the integrated circuit (IC). Miniaturization in electronic
technology made people wonder what other technologies could also be miniaturized. The
seminal talk by Richard Feynman titled “There’s Plenty of Room at the Bottom: An
Invitation to Enter a New Field of Physics” to the American Physical Society in 1959
marked the birth of the field of microelectromechanical systems (MEMS) [12]. In this talk,
the Nobel Laureate challenged and inspired scientists around the world to consider how to
scale down technologies (e.g. motors, encyclopedia) to micro and nano-sizes. Since then,
the MEMS field has focused on the exploration and development of small devices and
technologies that combine electrical and mechanical modalities.
Fundamental Microfabrication Techniques
MEMS devices are manufactured by leveraging the microfabrication techniques
used in the IC industry. Microfabrication processes can be categorized as bulk and surface
micromachining (Figure 1-2). Bulk micromachining utilizes the substrate film itself,
commonly silicon, to carve out a MEMS device. Various etching and deposition techniques
allow patterning of the substrate. Fabrication of complex or multilayer devices is limited
through the use of bulk micromachining. On the contrary, surface micromachining allows
the fabrication of complex planar structures by deposition and selective removal of

4
structural and sacrificial materials on a carrier substrate. The substrate, usually silicon, only
serves as mechanical support during the fabrication process.

Figure 1-2. Schematic diagram of the cross-sectional views of bulk and surface
micromachining techniques.
MEMS devices are fabricated with a combination of bulk and surface
micromachining steps. The fundamental microfabrication processes are categorized as:
lithography, etching, and deposition. Description of the fundamental microfabrication
processes will be presented with an emphasis on the techniques required to fabricate the
devices described in this work. Detailed information about microfabrication techniques can
be found in [11, 13, 14].
1.3.1 Lithography
Photolithography (Figure 1-3) is a widely used technique to transfer relief patterns
onto a substrate with high resolution and fidelity. A photo sensitive material, photoresist,
is spin coated onto the substrate and patterned by exposure to UV light. Photoresist is
usually an organic polymer infused with a photoactive compound rendering it photo-
patternable. Photoresist is categorized into two types of tones: (1) positive, if light exposed
areas become more soluble in a developing solution, and (2) negative, if the exposed areas
become insoluble in the developer. The photoresist layer can serve as an etching/deposition
mask or to create sacrificial structures of a device.

5

Figure 1-3. Schematic representation of photolithography involving the patterning of
photoresist on polymer.
1.3.2 Etching
Etching is a subtractive process for pattern transfer by selective removal of material.
A masking material (resist or oxide) is patterned on the substrate and the unmasked
material is selectively removed. Etching can be classified by the mechanism used to
remove the material. Wet etching requires substrate immersion in a liquid chemical to etch,
while dry etching is dominated by physical etching in a vacuum environment typically
involving plasma. The side profile resulting from etching can be categorized as isotropic
or anisotropic. In isotropic etching material is removed at equivalent rates in all direction
including horizontal, which results in undercutting of the mask. However, anisotropic etch
has a direction dependence, in most cases the etch rate is significantly faster in the direction
normal to the substrate than the lateral direction. It is important to consider the following
when choosing an etching process: (1) compatibility of the process with the material to be
etched, (2) feature depth, and (3) desired geometry. Deep Reactive Ion Etching (DRIE), a
type of dry etching, was chosen to etch thick polymer films in this work. DRIE combines
a high density plasma with a switch-chemistry process to achieve high aspect ratio
structures with near vertical sidewalls. DRIE uses a Bosch process in which a thin

6
passivating layer is deposited on the sidewalls protecting them during the etch step and
only allowing etching of the bottom feature layer as shown in Figure 1-4.

Figure 1-4. Cyclical nature of deep reactive ion etching process. Steps 2 to 3 are repeated
until the end of etch.
1.3.3 Deposition
Deposition is an additive process in which layers of a material are built-up on a
substrate using techniques such as chemical vapor deposition (CVD), spin coating, physical
vapor deposition (PVD), and thermal oxidation. Quality of the film deposited is dependent
on adhesion to the underlying layer, conformality, film stress, and continuity of the material.
CVD is a chemical process that involves a class of heat and mass transfer deposition
techniques. The technique produces high quality, low stress, and conformal films, is
versatile, and works with a wide range of materials. In PVD the material to be deposited is
first converted into vapor by heating (evaporation) or ions bombardment (sputtering) and
then allowed to condense on the substrate. Compared to evaporation, sputtering can be
used to deposit a large variety of materials and provides a more conformal coverage. In
this work CVD is use for polymer coating, spin coating for planarization of photoresist,
and PVD for metal deposition.
Non-conventional Microfabrication Methods
The microfabrication methods previously described are designed primarily for
traditional MEMS materials such as semiconductors or glasses. There are a group of
fabrication techniques that are applicable for polymers, require minimal processing time,
and in most cases do not require sophisticated equipment.

7
1.4.1 Casting and soft lithography
Casting and soft lithography are techniques commonly used for fast prototyping of
microfluidic systems. A device can be fabricated by relying on molding of a polymer,
usually polydimethylsiloxane (PDMS). Casting can also be performed with other materials
apart from PDMS. A master mold made of etched silicon surfaces or patterned photoresist
is used to define the pattern. The liquid polymer is poured onto the master mold, resulting
in a negative replica of the mold once the polymer cures (Figure 1-5). This process is easily
repeated to create sealed and multilayer structures. The layers are bonded by a reversible
(adhesion material) or irreversible (oxidizing PDMs surface) process. Soft lithography is
limited to only polymers and involves other fabrication methods apart from casting. The
patterned polymer can be used as a stamp (microcontact printing, µ-CP) or mold
(microtransfer molding, µ-TM or micromolding in capillaries, MIMIC) to generate
microstructures or micropatterns.

Figure 1-5. Process for creating a PDMS mold by casting.
1.4.2 Injection molding
Injection molding (IM) is a precision plastic molding technique used to rapidly
replicate polymer microparts. IM requires melting of a thermoplastic material and injecting
it into a heated mold cavity with a controlled injection pressure and hardens when cooled.
Low molecular weight polymers are compatible with IM such as polymethylmethacrylate
(PMMA), polypropylene (PP), and polycarbonate (PC), among others. IM is ideal for
generating microparts with small features and low-aspect ratios. The main challenge of IM

8
for MEMS devices is the fabrication of components with small feature size and high-aspect
ratios, and the high cost of the IM machine.
1.4.3 Stereolithography
Stereolithography (SL) is an additive fabrication process that involves the
polymerization of a liquid photopolymer resin by UV laser into a three-dimensional (3D)
shape. The 3D shape is designed with a CAD software which then feds a two-dimensional
cross-section to the computer-controlled laser to build the desire structure layer by layer.
SL is a rapid fabrication technique that can achieve high feature resolution.
BioMEMS
Interest in MEMS-based devices for biomedical and biological applications has led
to the creation of a specialized field called bioMEMS. BioMEMS devices require aseptic
fabrication techniques, hermetic seals, and biocompatible structural materials for optimal
operation in the in vivo environment [15, 16]. Silicon is an excellent material for electrical
and mechanical applications, but it is not commonly accepted as a biocompatible material
and its rigidity makes it difficult to integrate with the soft biological tissue. Isolation of the
system’s electrical and mechanical components from the harsh and corrosive in vivo
environment remains a major challenge for many bioMEMS devices.
Fortunately, these components can be protected by encapsulation with
biocompatible polymer materials that have good barrier properties. In fact, a few polymer
materials can provide the proper encapsulation and also serve as structural materials for
bioMEMS devices. Such is the case of Parylene, a polymer commercialized in the 1960s
by Union Carbide Corporation [17]. There are many different types of Parylene available,
but the most well-known is Parylene C (poly(monochloro-p-xylylene)), a United States
Pharmacopeia (USP) class VI material (Figure 1-6). This rating is the highest possible for
biocompatibility making it suitable for long-term implantation in vivo. Parylene C is
chemically inert, optically transparent, has excellent water barrier properties, low process
temperatures, and is compatible with standard microfabrication techniques [18]. Parylene

9
C is conformally deposited through CVD, patterned with standard photolithographic
processes, and anisotropically etched with oxygen-based plasma through DRIE. Interest
in free film Parylene C devices is growing due to its compatibility with microfabrication
processes, inertness, and high flexibility.

Figure 1-6. Chemical structure of Parylene C.
BioMEMS Implantable Devices
The advent of bioMEMS 0technology has led to smaller, reliable, and versatile
implantable devices. Several iterations of bioMEMS implantable devices have been
successfully developed. In 1986 the University of Michigan introduced the Michigan probe,
consisting of a micromachined silicon shank with multisite recording electrodes patterned
along the shank and with in-chip signal processing [19]. The Utah electrode array,
introduced in 1991, consists of a silicon-based, microfabricated array of shanks, each with
one recording site at the tip [20]. The Utah array was later commercialized by Blackrock
Microsystems. These intracortical microelectrode arrays can record and stimulate from
individual or small group of neurons making them highly selective (Figure 1-7) [21].
However, their chronic performance is limited by neural tissue and microelectrode damage
during the acute inflammatory reaction resulting from the arrays insertion and by the
presence of the rigid probes in the soft tissue.

10

Figure 1-7. Photographs of the (a) Utah (Reprinted by permission from Macmillan
Publishers Ltd: Nature Neuroscience [22] © 2002) and (b) Michigan arrays (reprinted with
permission from [23] © 2008 IEEE).

CardioMEMS, Inc. developed the first Food and Drug Administration (FDA)
approved MEMS-based implantable device, EndoSure® AAA Wireless Pressure
Measurement System shown in Figure 1-8 [24]. This implantable sensor measures intrasac
pressure during a minimally invasive cardiovascular surgery to repair abnormal thoracic
aortic aneurysm. The Argus II Retinal Prosthesis System was developed to induce visual
perception to blind patients by providing electrical stimulation of the retina [25]. The
implanted portion of the system consists of an antenna, electronics, and an electrode array
[26]. The electrode array is made of a thin film polymer with 60 electrodes (Figure 1-9).
The system is commercially approved to treat advanced retinitis pigmentosa. Gensler, et
al. presented the first implantable MEMS-based drug delivery system for cancer treatment
in mice [27]. The refillable device integrates an electrolysis-based actuation system and a
dosing profile that can be preset to a desired regimen (Figure 1-10). The micropump is now
commercialized by Fluid Synchrony LLC.

11

Figure 1-8. CardioMEMS, Inc wireless microelectromechanical systems implantable
pressure sensor part of the EndoSure® AAA Wireless Pressure Measurement System
(reprinted from [7] with permission from The Royal Society of Chemistry).

Figure 1-9. (a) Schematic diagram of the Argus II electrode array with respect to the treated
eye and (b) the Argus II electrode array implanted in a subject with Retinitis Pigmentosa
(reprinted with permission from [26, 28] © 2011 IEEE).

Figure 1-10. Photograph of the refillable micropump system (reprinted from [27] with
permission from Springer).

12

The functionality requirements of current BioMEMS implantable devices as the
ones previously described have become more complex requiring the development of
multifunctional implants for better disease control and tailored therapies. These
sophisticated implants aim to incorporate a combination of several functions (e.g.
mechanical, electrical, fluidic) to new or existing technologies. Multifunctional neural
implants have been developed to record and modulate neural activity using a combination
of electrical and fluidic functions. Medtronics Inc. integrated drug delivery to their existing
deep brain stimulation electrodes to improve treatment efficacy of neural disorder like
epilepsy and Parkinson’s disease [29]. The e-dura is a soft spinal cord interface for neural
recording and stimulation integrated with a drug delivery channel [30]. The combination
of electrical and chemical stimulations by the e-dura allowed a paralyzed rat to walk. The
Michigan probe previously mentioned is now commercialized by NeuroNexus® [31]. The
company has focused on developing smaller, high density, more complex, and versatile
microelectrodes for neural interfaces. Some of their neural interfaces incorporate
optoelectrodes, drug delivery, or chemical sensing. Rohatgi, et al., incorporated a Michigan
array to a fused silica catheter for simultaneous neural recordings and drug delivery in a
rat cortex (Figure 1-11)[32]. In this study, localized drug delivery aims to prolog the in
vivo electrophysiological performance of their neural interface. Neural interfaces
integrated with microfluidics for drug delivery open new avenues to interface with the
nervous system and improve long-term in vivo performance.

13

Figure 1-11. Photograph of a Michigan electrode array with integrated microfluidic
channel (reprinted from [32] with permission from the Journal of Neurosurgical Focus).

The trend towards personalized drug delivery has led to recent research focus on
the development of complex implantable pumps with closed-loop drug delivery systems
[5, 33]. A feedback system is possible with the incorporation of physical sensors and
integrated electronics. Such a system was presented by Sheybani, et al., where an
electrolysis-based drug delivery micropump integrated with electrochemical dose sensing
allowed tailored and localized drug delivery with real-time device operation feedback
(Figure 1-12)[34]. A dual chamber drug delivery pump with piezoresistive pressure sensors
for active fluid delivery regulation was presented in [35, 36]. The micropump successfully
regulated delivery in various dosing regimens. The incorporation of sensors as closed-loop
feedback mechanisms further improves the safety and accuracy of implantable drug
delivery devices.

14

Figure 1-12. Photograph of drug delivery micropump with integrated sensors and
electronics (reprinted from [34] with permission from Springer).

Multifunctional implantable devices have been successfully achieved; however
most of this work has not yet proven successful in long-term in vivo studies. In part, due to
primary focus on the technical fabrication of the devices as oppose to the in vivo
implementation. The bioMEMS field continues to grow with many devices still being
developed. Areas that will greatly benefit from advancements in the field of bioMEMS
include multifunctional implants for tailored drug delivery systems as well as neural
implants to restore lost sensory and motor functions.
Objectives
BioMEMS technologies have led to the development of several implantable
medical devices. However, very few of these devices can achieve reliable chronic
performance in vivo due to issues with biocompatibility, rigidity of the substrate resulting
in mechanical mismatch between the implant and the surrounding biological tissue,
inappropriate packaging that allows device failure in the harsh in vivo environment, and
lack of multifunctionality. Fortunately, these limitations can be addressed by fabrication of
thin-film implantable devices with polymer materials such as Parylene C, a material that
provides proper device isolation from the wet corrosive biological environment, satisfies
requirements of biocompatibility, flexibility, and amenability to standard MEMS
processing methods. Thin-film Parylene devices can be integrated to existing technologies
(e.g. electrodes, sensors) to provide additional functionality such as flexible microfluidics.

15
The aim of this work is to develop multifunctional Parylene-based implantable
technologies to treat critical medical challenges otherwise unresolvable with current
technologies. The integration of microfluidics to MEMS-based implantable devices will
impart higher levels of functionality by allowing highly localized drug delivery, fluid flow
control in the microscale, improved device-tissue integration, and prolog chronic device
stability. Two specific challenges are examined in the detail: the need for reliable and
localized in vivo infusion of therapeutic pharmaceuticals, and the need for minimally
invasive brain-computer interfaces with high-fidelity signal recording and generation.
These problems are significant obstacles for the fields of drug delivery and prosthetics,
respectively, and share a need for chronically viable medical implants of high complexity
and functionality but minute size. The next chapter will discuss the development of these
implantable devices.
References
[1] W. Greatbatch and C. F. Holmes, "History of implantable devices," IEEE engineering
in medicine and biology magazine: the quarterly magazine of the Engineering in
Medicine & Biology Society, vol. 10, p. 38, 1991.
[2] G. Nelson, "A brief history of cardiac pacing," Texas Heart Institute Journal, vol. 20,
p. 12, 1993.
[3] W. M. Chardack, A. A. Gage, and W. Greatbatch, "A transistorized, self-contained,
implantable pacemaker for the long-term correction of complete heart block," Surgery,
vol. 48, pp. 643-654, 1960.
[4] L. Thomas Jr and S. Bessman, "Prototype For An Implantable Micropump Powered
By Piezoelectric Disk Benders," ASAIO Journal, vol. 21, pp. 516-522, 1975.
[5] A. Cobo, R. Sheybani, and E. Meng, "MEMS: Enabled Drug Delivery Systems,"
Advanced healthcare materials, vol. 4, pp. 969-982, 2015.
[6] M. Mirowski, P. R. Reid, M. M. Mower, L. Watkins, V. L. Gott, J. F. Schauble, et al.,
"Termination of malignant ventricular arrhythmias with an implanted automatic
defibrillator in human beings," New England Journal of Medicine, vol. 303, pp. 322-
324, 1980.
[7] E. Meng and R. Sheybani, "Insight: implantable medical devices," Lab on a Chip, vol.
14, pp. 3233-3240, 2014.

16
[8] Transparency Market Research. (2016, 10/01/2016). U.S Implantable Medical
Devices Market Future Lies in Orthopedic Implants, Expected to Reach US$73.9 bn
by 2018. Available:
http://www.transparencymarketresearch.com/pressrelease/implantable-medical-
devices-market.htm
[9] Y. Onuki, U. Bhardwaj, F. Papadimitrakopoulos, and D. J. Burgess, "A review of the
biocompatibility of implantable devices: current challenges to overcome foreign body
response," Journal of diabetes science and technology, vol. 2, pp. 1003-1015, 2008.
[10] P. Si, A. Hu, J. Hsu, M. Chiang, Y. Wang, S. Malpas, et al., "Wireless power supply
for implantable biomedical device based on primary input voltage regulation," in
2007 2nd IEEE Conference on Industrial Electronics and Applications, 2007, pp.
235-239.
[11] E. Meng, Biomedical microsystems: CRC Press, 2011.
[12] R. P. Feynman, "There's plenty of room at the bottom," Engineering and science, vol.
23, pp. 22-36, 1960.
[13] M. J. Madou, Fundamentals of microfabrication: the science of miniaturization: CRC
press, 2002.
[14] M. J. Madou, "Fundamentals of microfabrication and nanotechnology. Volume two:
Manufacturing techniques for microfabrication and nanotechnology," ed: CRC Press,
2011.
[15] G. Kotzar, M. Freas, P. Abel, A. Fleischman, S. Roy, C. Zorman, et al., "Evaluation
of MEMS materials of construction for implantable medical devices," Biomaterials,
vol. 23, pp. 2737-2750, 2002.
[16] A. R. Grayson, R. S. Shawgo, A. M. Johnson, N. T. Flynn, Y. Li, M. J. Cima, et al.,
"A BioMEMS review: MEMS technology for physiologically integrated devices,"
Proceedings of the IEEE, vol. 92, pp. 6-21, 2004.
[17] Specialty Coating Systems. (2016, 9/30/2016). History of Parylene Available:
http://scscoatings.com/what-is-parylene/parylene-history/
[18] E. Meng, P.-Y. Li, and Y.-C. Tai, "Plasma removal of Parylene C," Journal of
Micromechanics and Microengineering, vol. 18, p. 045004, 2008.
[19] K. Najafi and K. D. Wise, "An implantable multielectrode array with on-chip signal
processing," IEEE Journal of Solid-State Circuits, vol. 21, pp. 1035-1044, 1986.

17
[20] P. K. Campbell, K. E. Jones, R. J. Huber, K. W. Horch, and R. A. Normann, "A
silicon-based, three-dimensional neural interface: manufacturing processes for an
intracortical electrode array," IEEE Transactions on Biomedical Engineering, vol. 38,
pp. 758-768, 1991.
[21] E. Fernández, B. Greger, P. A. House, I. Aranda, C. Botella, J. Albisua, et al., "Acute
human brain responses to intracortical microelectrode arrays: challenges and future
prospects," The chronic challenge-new vistas on long-term multisite contacts to the
central nervous system, p. 10, 2015.
[22] J. P. Donoghue, "Connecting cortex to machines: recent advances in brain interfaces,"
Nature neuroscience, vol. 5, pp. 1085-1088, 2002.
[23] K. D. Wise, A. M. Sodagar, Y. Yao, M. N. Gulari, G. E. Perlin, and K. Najafi,
"Microelectrodes, microelectronics, and implantable neural microsystems,"
Proceedings of the IEEE, vol. 96, pp. 1184-1202, 2008.
[24] Medspace. (2007, 10/01/2016). FDA Approvals: EndoSure AAA, Therapy Cool Path,
AngioJet Spiroflex VG. Available: http://www.medscape.com/viewarticle/554899
[25] Second Sight. (2016, 9/29/2016). Argus® II Retinal Prosthesis System. Available:
http://www.secondsight.com/g-the-argus-ii-prosthesis-system-pf-en.html
[26] D. D. Zhou, J. D. Dorn, and R. J. Greenberg, "The Argus® II retinal prosthesis
system: An overview," in Multimedia and Expo Workshops (ICMEW), 2013 IEEE
International Conference on, 2013, pp. 1-6.
[27] H. Gensler, R. Sheybani, P.-Y. Li, R. L. Mann, and E. Meng, "An implantable MEMS
micropump system for drug delivery in small animals," Biomedical Microdevices, vol.
14, pp. 483-496, 2012.
[28] D. Fitzpatrick, Implantable Electronic Medical Devices: Elsevier, 2014.
[29] D. D. Elsberry, M. T. Rise, and G. W. King, "Method of treating movement disorders
by brain stimulation and drug infusion," ed: Google Patents, 2000.
[30] I. R. Minev, P. Musienko, A. Hirsch, Q. Barraud, N. Wenger, E. M. Moraud, et al.,
"Electronic dura mater for long-term multimodal neural interfaces," Science, vol. 347,
pp. 159-163, 2015.
[31] NeuroNexux. (2016, 9/29/2016). Neural Probes. Available:
http://neuronexus.com/products/neural-probes

18
[32] P. Rohatgi, N. B. Langhals, D. R. Kipke, and P. G. Patil, "In vivo performance of a
microelectrode neural probe with integrated drug delivery: Laboratory investigation,"
Neurosurgical focus, vol. 27, p. E8, 2009.
[33] E. Meng and T. Hoang, "MEMS-enabled implantable drug infusion pumps for
laboratory animal research, preclinical, and clinical applications," Advanced drug
delivery reviews, vol. 64, pp. 1628-1638, 2012.
[34] R. Sheybani, A. Cobo, and E. Meng, "Wireless programmable electrochemical drug
delivery micropump with fully integrated electrochemical dosing sensors,"
Biomedical microdevices, vol. 17, p. 74, 2015.
[35] A. T. Evans, J. M. Park, S. Chiravuri, and Y. B. Gianchandani, "Dual drug delivery
device for chronic pain management using micromachined elastic metal structures
and silicon microvalves," in Micro Electro Mechanical Systems, 2008. MEMS 2008.
IEEE 21st International Conference on, 2008, pp. 252-255.
[36] T. Li, A. T. Evans, S. Chiravuri, R. Y. Gianchandani, and Y. B. Gianchandani,
"Compact, power-efficient architectures using microvalves and microsensors, for
intrathecal, insulin, and other drug delivery systems," Advanced drug delivery reviews,
vol. 64, pp. 1639-1649, 2012.

19

In drug development for disease management, the scientific community relies
heavily on animal studies to provide basis for human clinical trials [1]. The mouse genome
has been completely sequenced allowing accurate comparisons to other mammal genomes
such as humans and its physiologic, metabolic and pathologic processes have been well
documented [2]. As such, human drug therapies are initially evaluated in small animal
models; therefore, drug administration technologies for small laboratory animals are
critically important for drug discovery and development. In this chapter, we present an
implantable micropump with a miniature form factor and completely wireless operation
that enables chronic drug administration intended for evaluation and development of cancer
therapies in freely moving small research animals such as rodents.
Background
Human drug therapies are initially evaluated in small animal models such as
rodents; therefore, drug administration technologies for small laboratory animals are
critically important for drug discovery and development. The most commonly used drug
administration routes for laboratory animals are oral, topical, or intravenous which require
high drug dose concentration, leading to severe side effects, and frequent animal handling.
In chronic studies, frequent animal handling or restraint induces stress, which can confound
study results [3].
The efficacy of chronic drug treatments depend on delivering the appropriate
amount of the therapeutic agent to a specific location at the right time and rate [4]. Delivery
to a specific tissue facilitates the goal of achieving therapeutic drug levels while reducing
harmful side effects, the amount of drug needed, and the frequency of dosing events [5].

WIRELESS DRUG DELIVERY PUMP

20
Localized delivery may be particularly beneficial when administering pharmaceuticals
with short in vivo half-lives such as proteins and peptides. Periodic biological fluctuations
such as circadian rhythms may affect drug kinetics and dynamics [6]. A study of more than
30 anti-cancer drugs showed that the toxicity and efficacy of these drugs varied by more
than 50% depending on the time of dosing [7]. Greater control over delivery rates allows
tailored therapies and drug regimen changes during the course of therapy.
Current technologies for chronic dosing in laboratory animals include tethered
infusion systems and vascular access ports. In tethered systems, the animal is connected to
a precision infusion pump with a tether as shown in Figure 2-1a [8]. The use of the external
pump provides accurate dosing, but is impractical for chronic studies since restriction of
movement by the tether induces stress. Vascular access ports are chronically implanted
subcutaneously minimizing animal movement restrictions [9]. However, vascular ports
require regular cleaning, maintenance, and appropriate protective measures to prevent
infection to or damage by the animal.

Figure 2-1. Schematic diagram of in vivo testing setup for small animal research with (a)
commercial tethered infusion system, and (b) wireless implantable drug delivery pump
(reprinted from [10] with permission from Elsevier).

Implantable infusion pumps further minimize animal handling and permit natural
movement after the initial healing period following implantation. The Med-e-Cell Infu-
Disk is non-refillable, has a preset constant flow rate, and a maximum lifetime of 1 week
[11]. This pump is only suitable for drug delivery in animal models larger than rats due to
its large size. The Primetech iPrecio and the Alzet pumps are the only commercial pumps

21
scaled for mice. The iPrecio peristaltic pump allows programmable infusion profiles, but
is battery powered, only provides up to 47 days of operation at flow rates lower than 0.1
µL/min, cannot be powered off, and is single use [12]. The Alzet osmotic pump provides
a preset continuous flow rate, a drug payload of up to 6 weeks, and is non-refillable [13].
These technical limitations of current technology prohibit flexibility in designing
administration regimens for drug discovery and development research. While
commercially available pumps are not suitable for chronic drug infusion in small animals,
recent advances in microelectromechanical systems (MEMS) technology have led to the
miniaturization and improved efficacy of implantable drug delivery systems suitable for
small laboratory animals [14]. Several ongoing efforts are being carried out in research
laboratories to develop remotely activated drug delivery microsystems. Detailed
descriptions of these efforts are available in [15, 16].
Advanced drug delivery systems incorporate valves for flow regulation. Valves are
crucial for accurate dosing and prevention of reverse flow to avoid device and drug
contamination with interstitial fluid [17]. Several MEMS-based microvalves have been
developed and were reviewed in [18-20]. Microvalves can be categorized as active (with
an actuator) or passive (without an actuator). Active microvalves have been intensely
researched; the published literature describes many active microvalves having novel
designs and actuation mechanisms. However, they are expensive due to the complexity of
their operation mechanism and their fabrication. Passive valves are simple to fabricate,
inexpensive, and most importantly are not controlled by an external actuator. Current
research passive microvalves are limited by their inability to provide leak-free closure, low
cracking pressure (low flow resistance) and consistent performance for low flow regime
devices. Despite many decades of development of microvalves, there is no commercially
available microvalve that is reliable and small enough to be incorporated into our system
[21, 22].
The micropump presented in this work was actuated using electrolysis which offers
low power consumption, large driving force, low heat generation, on-demand activation,

22
and post-implantation flow rate control through the applied current [23]. Wired implantable
MEMS micropumps scaled for use in small laboratory animals (e.g., mice) were previously
demonstrated [23-25]. Preliminary in vivo studies with this wired system demonstrated the
potential of using micropumps to combat radiation resistance tumor in mice by improving
the exposure of tumors to siRNA through sustained localized delivery compared to needle
injections [24-26]. However, the use of transcutaneous wires for powering the pump
required animals to be restrained during dosing, prohibited chronic studies, and placed
restrictions on experimental throughput. Also, the lack of reliable, small form factor check
valves resulted in a slow diffusion of drug from the reservoir into the animal and thereby
limited the accuracy of delivery. The absence of a reliable check valve could also lead to
fluid to flow back through the delivery catheter to the reservoir as a result of the reverse
pressure gradient created during the recombination reaction between electrolysis actuation
cycles [23].
Here, we present a redesigned implantable micropump that incorporates wireless
power telemetry and an integrated miniature check valve that is suitable for chronic drug
delivery in untethered, freely moving small laboratory animals such as mice (Figure 2-1b,
Figure 2-2, Table 2-1). This wireless drug administration method can eliminate the induced
stress from animal handling or tethered and wired drug administration systems that could
potentially confound interpretation of data [3]. Wireless operation was facilitated by
improving electrolysis actuation efficiency which further reduced actuator power
consumption (1 mW from 3 mW). A commercial check valve with small form factor and
improved performance enabled appropriate flow regulation during pumping and resting
states. Concurrently, a custom-made passive check valve was designed, fabricated, and
characterized specifically for our system to achieve improved flow regulation performance.
For this work, the pump performance was characterized for simulated in vivo acute and
chronic conditions, including those relevant to future studies to deliver agents to combat
radiation resistance in a mouse xenograft cancer model. A redesigned drug reservoir
achieved a more compact footprint by reducing dead volume (inaccessible drug) by nearly
91% and check valve dimensions, which were essential to accommodate the addition of

23
wireless circuitry. Integrated refill ports allow long term use of the micropump for chronic
studies.
Table 2-1. Primary differences between the previous generation refillable implantable
micropump system (RIMS) and our current implantable micropump for drug delivery in
small animals (reprinted from [10] with permission from Elsevier).
RIMS pump [24] Current pump
Operation Wired, External constant
current power supply
Wireless power telemetry
One way valve
specifications
Bridge-type membrane valve
0.515 x 1.23 mm, L x OD
1.31 kPa cracking pressure
Duckbill valve
3 x 2 mm, L x OD
0.69 kPa cracking pressure
Height
10 mm 8.1 mm
Footprint
21 mm OD 20 x 15 mm, L x W
Dead volume
1087 µL 96 µL
Power consumption
3 mW 1 mW

Figure 2-2. Schematic diagram of wireless drug delivery system (modified from [27] ©
2014 IEEE).
Design
2.2.1 Check valve
A custom-made normally closed spring-loaded ball check valve (Figure 2-3a) was
included at the pump outlet to prevent backflow of fluids as a result of the reverse pressure
gradient caused by recombination. A passive check valve was selected to minimize the

24
system’s power requirements and due to its lesser fabrication complexity. This type of
valve depends on the pressure of the fluid to operate and have a diode-like flow regulation.
This valve opens with an applied forward pressure greater than its cracking pressure. Once
the cracking pressure has been exceeded, the spring compresses allowing the ball to move
away from the valve seat and consequently allowing forward fluid flow (Figure 2-3b).
Once, the forward pressure is removed, the spring returns to its decompressed state pushing
the ball back into the valve seat (Figure 2-3c). Additional valve design requirements
include: 1) maximum reverse leakage of 10% of dose volume, 2) low cracking pressure
(~0.69 kPa), 3) normally closed, and 4) valve incorporation into micropump catheter outlet.

Figure 2-3. (a) Schematic diagram of the custom-made normally closed spring-loaded ball
check valve (reprinted from [28] with permission from Springer) and (b) Shows fluid path
when the valve is open with a forward pressure. (c) Closed valve in the presence of reverse
pressure or no forward pressure being applied.
2.2.2 Electrolysis actuator
Our micropump system utilizes a low power electrolysis actuation mechanism. The
electrochemical actuator consists of an electrolyte (water) encased by a Parylene bellows
and a pair of interdigitated platinum (Pt) electrodes supported on a rigid glass substrate
Figure 2-4a. An applied electrical current to the electrodes dissociates liquid water into
oxygen and hydrogen gas and induces an increase in pressure. This deflects the bellows,
activates a one-way check valve, and displaces the surrounding fluid out of the rigid
reservoir through the outlet catheter to the delivery site (Figure 2-4b). The gases recombine

25
to water once the current is turned off, allowing the bellows to return to the original rest
position.

Figure 2-4. Micropump electrolysis actuation operation concept. (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 expands the bellows and drive surrounding drug out of the reservoir
and catheter.

The interdigitated electrode design reduces the solution resistive path resulting in
improved electrolysis reaction efficiency [29]. Pt was selected as the electrode material due
to its biocompatibility, good corrosion and oxidation resistance, and ability to catalyze
recombination [30, 31]. Electrolysis efficiency was further improved by coating the
electrodes with Nafion®, a solid polymer electrolyte that provides higher solubility of
oxygen and hydrogen gases than water [23]. Nafion®-coated electrodes also provide faster
and more repeatable recombination, decrease electrode delamination or damage at high
current operation, and increase flow rate [32].
Bellows possess the ability to achieve large deflection with low applied pressure
compared to corrugated or flat diagrams [29]. Parylene C was selected as the bellows
material due to its low Young’s modulus, biocompatibility, inertness, and low permeability
[33, 34]. The bellows consisted of two convolutions each having a 6 mm inner diameter,
9.5 mm outer diameter, 0.4 mm height and 13.5 m thickness. This bellows can safely
displace 185 L of fluid contained within the adjacent drug reservoir without experiencing
plastic deformation [35].

26
2.2.3 Wireless powering system
The pump is normally off until wirelessly activated by a class E inductive powering
system developed in the Biomedical Microsystems Laboratory (Figure 2-5). Inductive
transmission is most suitable for transcutaneous powering compared to other forms of
wireless powering (radiation, conductive, and capacitive) [36]. For this system the power
transmission frequency was chosen to minimize heat generation [37]. It relies on magnetic
field coupling and the transmission frequency can be chosen to minimize power dissipation.
The class E inductive powering transmitter is placed underneath the rodent cage. A current
regulator on the subcutaneously implanted receiver controls the application of current to
the electrolysis actuator; here, a single constant current magnitude was supplied
corresponding to a single flow rate. The volume delivered is controlled by the duration of
pump activation. Heat generation from the wireless system is negligible in part due to
chosen operation frequency [15]. High efficiency off the shelf components were chosen to
minimize power dissipation [38].

Figure 2-5. Circuit diagram for the class E wireless inductive powering system. Link
efficiency for concentric transmitting and receiving coils with zero angular misalignment
was calculated to be approximately 0.82 (modified from [10] with permission from
Elsevier).
2.2.4 System packaging
The micropump design achieves compact footprint by reducing dead volume
between the reservoir wall and bellows. The reservoir inner diameter was reduced by 46%
resulting in a substantial reduction in dead volume from approximately 1100 to 100 µL

27
compared to previous micropump prototypes [39]; this new volume combined with the
capability for refill is adequate for most small animal drug delivery applications. The
reduction in reservoir footprint allowed incorporation of the wireless components while
maintaining a total device mass of approximately 10% of the weight of the animal
(typically adult mice are 30 g) to allow free, unhindered movement [40].
Fabrication
2.3.1 Check valve
Passive spring-loaded ball check valves were fabricated using custom-made parts.
The valve housing consisted of a valve stop and a valve seat produced using
stereolithography (SL) in USP class VI DSM Somos® WaterShed XC 11122 material as
shown in Figure 2-6a-b (FineLine Prototyping, Inc., Raleigh, NC). First generation valve
incorporated balls (Ø 1.524 mm) molded in USP class VI clear silicone and polished to
remove the parting line, (Figure 2-6c, Precision Associates, Inc., Minneapolis, MN).
Compression springs were custom-made from high performance MP35N (nickel-cobalt-
chromium-molybdenum) alloy with a spring rate of 0.035 N/mm (Figure 2-6d, MDC
18725c, Motion Dynamics Corp., Fruitport MI). The last 2-3 coils of one end of the spring
were opened to a larger diameter to allow the ball to connect to the spring without the need
for any adhesives. The small diameter end of the microspring was placed on the valve stop
and then the valve seat and valve stop were compression fit together. The valve was
assembled without the use of adhesives (Figure 2-7).

Figure 2-6. Photographs of first generation valve check valve components. (a) Valve stop,
(b) valve seat, (c) silicone ball, and (d) compression spring.

28

Figure 2-7. Photograph of assembled spring-loaded ball check valve.

Valve leakage was observed during initial characterization of the valve. Leakage
was attributed to inadequate seating of the ball on the valve seat due to the parting line of
the molded silicone ball and to the high surface roughness of the stereolithography parts.
In order to improve sealing on the valve seat, an O-ring was built as follows: a small amount
of USP class VI silicone elastomer (MED-4210, Factor II, Inc., Lakeside, AZ) was poured
directly into the valve seat using a pipette and cured at 40°C. Once cured, the inner diameter
of the membrane was cored using a 20 G needle (Figure 2-8). An average O-ring thickness
of 0.45 mm was achieved with 4% variability among O-rings as confirmed by ImageJ
software analysis (National Institutes of Health). The polymer balls were replaced by
stainless steel balls with no parting line (Ø 1.587 mm, Ultra-Hard Wear-Resistant 440C
Stainless Steel) (McMaster-Carr, Elmhurst, IL) (Figure 2-9). The final valve length was
approximately 5.5 mm with a 2.6 mm outer diameter (Figure 2-10). Following
characterization testing, the check valve was incorporated within the reservoir housing
outlet of a wired micropump, prior to the connection of the silicone delivery catheter.

29

Figure 2-8. Valve seat with fabricated PDMS O-rings.

Figure 2-9. Photographs of final check valve
check valve components.

Figure 2-10. Photograph of final
spring-loaded ball check valves
(reprinted from [28] with permission
from Springer).
2.3.2 Electrolysis actuator
The actuator’s interdigitated electrodes (100 m wide elements separated by 100
m gaps, 8 mm diameter footprint, 300 Å/2000 Å Ti/Pt) were fabricated on a Borofloat 33
glass substrate (University Wafer, Boston, MA) by liftoff following e-beam deposition of
the thin film metal using a previous method described in [24, 32] (Figure 2-11). The
electrodes were potentiostatically cleaned at ± 0.5 V (Gamry Reference 600 Potentiostat,
Warminster, PA) in 1 X phosphate buffered saline (PBS). The electrodes were then dip
coated in Nafion® (Dupont DE521 Solution, Ion Power, INC, New Castle, DE). Kynar™

30
nickel plated copper wires (30 AWG, Jameco Electronics, Belmont, CA) were soldered to
the electrode contact pads. The joint was reinforced and insulated with nonconductive
marine epoxy (Loctite, Westlake, OH) [32].

Figure 2-11. Schematic diagram detailing the electrode and bellows fabrication processes
and bellows actuator assembly.
Parylene bellows were fabricated using a lost-wax two-part molding process
previously described in [35]. Polydimethylsiloxane (PDMS, Sylgard 184; Dow Corning
Corp., Midland, MI) sheets (0.4 mm thick) were perforated with 6 mm and 9.5 mm holes
to mold the inner and outer diameters of the bellows, respectively. The sheets were visually
aligned and stacked together on glass slides to form mold modules. These modules were
filled with molten (50°C) polyethylene glycol (PEG; 1,000 Mn, Sigma Alrich, St. Louis,
MO). Once the PEG solidified, PDMS sheets were peeled. The modules were stacked and
fused together by moistening connecting interfaces with water. Completed PEG molds
were coated with 13.5 µm United States Pharmacopeia (USP) class VI Parylene C
(Specialty Coating Systems, Indianapolis, IN), followed by soaking in deionized water at
room temperature to remove sacrificial PEG.

31
Hollow bellows sacrificial molds can also be fabricated by using injection molding
or casting techniques. These fabrication methods can accelerate the fabrication process
and reduce variability among bellows. A custom-made multipurpose aluminum lost-wax
bellows mold was designed and precision machined with the collaboration of Jeffrey Field
Design, Inc. This mold was designed to inject or cast large bellows (28 mm outer
diameters) molds for a pediatric drug delivery pump application. It is important to find
materials compatibles with these fabrication techniques and at the same time soluble in
water or mild solvents in order to be incorporated into our lost-wax two-part molding
process. Several soluble materials were tested by injecting or gravity casting into the mold.
Materials tested include: freeman optical soluble wax (Freeman Manufacturing & Supply
Company, Avon, OH), riace hydroresin water-soluble injection wax (Rio Grande,
Albuquerque, NM), water-soluble injection wax (Rio Grande, Albuquerque, NM),
parowax household wax (Royal Oak Enterprises, LLC., Roswell, GA), and PEG-1000.
Fabrication methods and materials were optimized to obtained clean wax release from the
mold and complete filling of the bellows mold. A range of temperatures, pressures, and
material composites were tested. Injection and casting methods yielded complete and
reliable bellows wax molds with a material composite of 5 parts PEG-1000 and 2 parts
hydroresin (Figure 2-12). Detailed information about the bellows application, fabrication
methods, and characterization is found in Appendix A.

Figure 2-12. Photographs of 5:2 PEG-1000 and hydroresin wax bellows mold fabricated
by (a) injection molding and casting (b) top view and (c) side view.

32
The Parylene C bellows were filled with double deionized water and attached to the
electrode using double-sided pressure sensitive adhesive film (Tape 415, 3M, St. Paul,
MN). The joint was then reinforced with marine epoxy.
2.3.3 Wireless powering system
A 2 MHz clock oscillator (ECS -2100, ECS international, Olathe, KS) was used to
create the power signal from a 9 V power supply (0.3 W). The generated signal was then
amplified in two stages before being applied to a tuned transmitting coil (8 turns of 20
AWG single strand wire, size: 310 mm x 140 mm). Litz wire (6 turns, 50/54 SPN/SN Litz
Wire, Wiretron, Volcano, CA) was used for the receiving coil (Ø 17 mm). The received
alternating signal was fully rectified using two Schottky diodes (BAT54A and BAT54C,
Fairchild Semiconductor, San Jose, CA). A resistor was used to set the output current on a
3-terminal adjustable current source (LM334, Texas Instruments, Dallas, TX) to 0.33 mA
for driving the pump.
2.3.4 System packaging and assembly
The reservoir domes and bases were produced using SL in USP class VI DSM
Somos® WaterShed XC 11122 material (FineLine Prototyping, Inc., Raleigh, NC). This
material is optically clear, durable, water resistant, and rigid [41, 42].
Medical grade silicone rubber septa (10:1 base-to-curing agent ratio USP class VI
MDX-4 4210; Factor II, Lakeside, AZ) were fabricated based on results of a leak puncture
test allowing for at least one dozen punctures with a maximum applied pressure within the
dome of 775 mmHg (15 psi). The septa were then inserted into the refill ports of the SL
dome and the perimeters reinforced with marine epoxy. This assembly was encapsulated
with 5 m Parylene C to further improve biocompatibility and barrier properties. Then the
electrolysis actuator was situated in the reservoir assembly and secured with biocompatible
epoxy.
Given the long lead time to procure custom parts for the custom-made check valves.
A one-way duckbill valve made of medical grade silicone (2.0 mm duckbill; Minivalve,

33
Cleveland, OH) with a measured cracking pressure of less than 0.69 kPa (5.17 mmHg) was
selected. Valves were prescreened for proper closure and then incorporated into the pump
by attaching the base of the valves to the reservoir housing output path using biocompatible
epoxy (EPO-TEK® 730 unfilled, Epoxy Technologies, Billerica, MA).
The wireless receiver was mounted on the micropump body and encapsulated with
5 m Parylene C, followed by a thin coating of medical grade silicone to provide a softer
interface with tissue. Final micropump components are shown in Figure 2-13. The total
mass of a filled prototype pump was 3.8 g (Figure 2-14).

Figure 2-13. Wireless implantable micropump individual components.

Figure 2-14. Photograph of a fully assembled wireless micropump (reprinted from [10]
with permission from Elsevier).

34
Experimental Methods
Our specific application of interest is a micropump for a siRNA-based therapy
against radiation resistant cancer in which an electrostatically-coupled siRNA-gold
nanorod drug is directly delivered to xenograft tumors in nude mice. This anti-cancer
application requires low flow rate (~ L/min), 30 L daily dose volume, and infusion
duration of at least two weeks with one refill per week. The micropump’s ability to
administer the desired regimen under wireless operation was investigated on benchtop. In
addition, the repeatability of performance across different wireless receivers and impact on
performance of changes in coil orientation were examined. Power transfer can be affected
by coil orientation changes. The wireless micropump is intended to be implanted in a freely
moving animal. Therefore, it is important to investigate the effects of varying the axial
separation distance and angular misalignment between the transmitting and receiving coils
that can result over the course of drug administration in an animal study. Here, we present
micropump performance in simulated in vivo conditions including delivery against a
physiologically relevant back pressure, delivering solutions of varying viscosities, when
delivering at body temperature, and long term pump functionality when subjected to a
simulated in vivo environment (saline soak at body temperature).
2.4.1 Check valve testing
The valve was first characterized with a custom pressure setup in order to obtain an
initial operating range (3 check valves, n = 4 runs per check valve) (Figure 2-15).
Pressurized water was applied to the inlet of the valve in increments of 0.05 psi (0.34 kPa)
from 0 psi until water flow was observed to determine cracking pressure. Reverse pressure
was investigated by applying the pressurized water to the valve outlet. In order to avoid
stiction, valves were flushed with deionized water prior to each run.

35

Figure 2-15. Schematic overview of pressure testing setup.

The check valve was then incorporated into a wired micropump at the reservoir
outlet and characterized (Figure 2-16). Forward flow rate and reverse leakage were
measured by monitoring fluid front movement in a 100 μL calibrated micropipette attached
to the valve outlet. 0.5 mA current was applied to the actuator using a Keithley 2400
sourcemeter (Keithley Instruments Inc., Cleveland, OH) to deliver 30 μL after valve
opening was observed. Current was turned then off and fluid front movement in the
micropipette was monitored for an hour after each run to determine valve reverse leakage
(3 check valves, n = 4 runs per check valves). Flow performance of the same valve-less
pump served as a control.

Figure 2-16. Testing setup for characterizing check valve performance as part of a wired
micropump.
2.4.2 Wireless testing
Flow rate measurements were taken on a single micropump with five different
wireless receivers. Flow rate was calculated by measuring fluid front displacement in a 100

36
L calibrated micropipette (VWR International, Radnor, PA) connected to the catheter
outlet as shown in Figure 2-17. Power transfer between inductive transmission coils is
optimal when concentric transmitter and receiver coils are placed in parallel and as close
to each other as possible [15]. The micropump is intended to be implanted in a moving
subject. Therefore, it is expected that the distance and angle between the transmitting and
receiving coils will vary over the course of administration. The axial separation distance
(parallel coils, 2 – 4 cm, 0.5 cm steps) and angular misalignment (0, 30, 45, and 60°)
between concentric transmitter and receiver coils (2 cm distance between coils) was varied
to study the effects on pump flow rate performance.

Figure 2-17. (a) schematic diagram (reprinted from [10] with permission from Elsevier)
and (b) photograph of experimental setup for flow rate characterization of wireless
micropumps.
2.4.3 Daily dosing
Six stationary (2 cm distance and 0° angular misalignment between concentric
coils) micropumps were operated daily for a period of one week to simulate the dosing
regimen planned for the anti-cancer application. A constant current of 0.33 mA was
supplied to the actuator wirelessly for a duration that allowed 30 L of double deionized
water to be delivered daily. The micropump catheter was connected to a calibrated
micropipette for flow rate measurements as previously described. The fluid front position
was recorded several times during each delivery at discrete time points. Reverse leakage
between dosing events, if any, was recorded. To avoid confounding results from
evaporation, the fluid front being tracked was followed by a small air pocket (10 mm),

37
followed by a column of water resulting in negligible evaporation between dosing events.
Two of the previous micropumps were operated daily for an additional week to simulate
the two week dosing regimen planned for the anti-cancer application. Daily dosing regimen
over a period of three weeks was studied in one of said micropumps. The pumps were only
refilled once per week and the duration of applied current was increased with each
successive dosing event.
2.4.4 Back pressure
To mimic the presence of finite pressure within the various compartments of the
body, micropumps were operated against back pressures varying from 0 to 20 mmHg
applied at the pump outlet. The upper limit of the back pressure was selected because it
exceeds the pressure anticipated for subcutaneous drug delivery in mice [43]. The catheter
outlet was connected to one end of a 100 L calibrated micropipette while the other end
was connected to a custom pressure setup as shown in Figure 2-18. The flow rate was
determined as previously described during trials lasting 5 minutes with 2 minute rest
periods between trials. Four trials were conducted at each back pressure and the flow rate
averaged.

Figure 2-18. Schematic overview of pressure testing setup.

38
2.4.5 Viscosity
Since delivered solutions may possess a range of viscosities, pump performance
using glucose solutions calibrated to different viscosities ranging from 1 to 6 cP was
investigated. Anhydrous D-glucose (VWR International, Radnor, PA) was dissolved into
Millipore water in five different concentrations. Solutions were chosen based on a
calibrated curve determined using a Cannon-Fenske viscometer (Table 2-2). The flow rate
was determined as previously described during trials lasting 5 minutes for each solution.
Four trials were conducted at each viscosity and the flow rate averaged. When changing
solutions, reservoirs were flushed with Millipore water to prevent changes to the viscosity
of the new solution.
Table 2-2. Glucose solutions prepared for viscosity testing.
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

2.4.6 Room temperature versus body temperature
In implantable drug delivery devices, the device will be operated at body
temperature. Three fully assembled and encapsulated micropumps were filled with double
deionized water, flush of any air bubbles and soaked in a room (23 °C) or body temperature
(37 °C) deionized water bath for 30 minutes. Flow rate test was performed while the pump
was submerged in a temperature-controlled water bath as shown in Figure 2-19. The flow
rate was determined as previously described during four cycles each lasting 5 minutes after
valve opening.

39

Figure 2-19. Experimental set up for room temperature versus body temperature testing.

2.4.7 Physiological environment simulation
System performance in simulated in vivo conditions was investigated to verify that
Parylene C and silicone rubber encapsulation layers provide adequate protection for the
electronics for the duration of the anticipated in vivo study. Three fully assembled and
encapsulated micropumps were soaked in 1 X PBS at body temperature (37°C).
Periodically, the micropumps were removed from the bath, refilled with double deionized
water, and subjected to flow rate testing. Valve opening times were recorded and flow rate
measurements were taken for 5 minutes after valve opening.
Results
2.5.1 Check valve
The cracking pressure for the 3 tested check valves using the custom pressure setup
were determined to be 1.46 (10.07), 3.45 (23.79), and 2.0 (13.79) psi (kPa), respectively
(mean value for n = 4 runs per check valve). Differences in cracking pressures can be
attributed to manual fabrication and the tolerance of the stereolithography parts. No reverse
leakage was observed for any of the three valves tested up to 4 psi (27.58 kPa). This
pressure far exceeds values expected to be encountered during recombination and within
various compartments of the body in vivo [44].

40
Representative results of the characterization of one valve are presented in Figure
2-20. As expected, the time required to open each valve varied with the cracking pressure
of the valve tested. However, all three valves opened within one minute of micropump
actuation. Once open, the valves did not present significant flow resistance to flow from
the micrompump. Yet, up to 13% variability was observed between flow rates obtained
from the four runs for each valve. Less than 1% reverse leakage was observed for any of
the three valves during recombination (n = 4 for each valve).

Figure 2-20. Valve characterization with wired micropump: (a) four 30 μL boluses
delivered after valve opening (micropump was refilled after each run to circumvent
pressure buildup during recombination), (b) valve performance averaged for the four runs
and compared to a valve-less system (modified from [28] with permission from Springer).
2.5.2 Wireless testing
Benchtop testing of a single micropump with five different receivers resulted on an
average flow rate of 2.66 ± 0.16 L/min (mean ± SE, n = 5), a 6 % variation between
receivers. The results from changing the axial separation distance between coils and
angular misalignment between transmitter and receiver showed a drop in flow rate by
64.1% and 42.86% for a 3.5 cm separation distance between coils and for 45° angular
misalignment between transmitter and receiver, respectively as shown on Figure 2-21.

41

Figure 2-21. The effects of (a) axial separation distance between concentric transmitter and
receiver coils, and (b) angular misalignment between concentric coils on pump
performance (at 0.33 mA constant current) (reprinted from [10] with permission from
Elsevier).

2.5.3 Daily dosing
Two wirelessly operated micropumps consistently delivered 30 L doses daily over
the course of two weeks as required by the anti-cancer application. These micropumps
showed similar flow rate performance during the two week testing with less than 8%
variation. Figure 2-22a shows representative results from the daily dosing regimen of one
micropump for one week. The data is presented as accumulated volume to study the effects
of increasing pressure build up within the reservoir during recombination periods. Similar
data was obtained for all pumps tested. The shown micropump delivered 30 L doses daily
for three consecutive weeks. Mean flow rates during the first two weeks are not
significantly different as confirmed by statistical analysis (ANOVA, p > 0.05) (Figure
2-22b). Valve performance varied across the micropumps, but it consistently limited back
flow to a maximum of 10 L when the pump was turned off. The slight volume drop seen
before each dosing event correlates to the back flow allowed by the valve. Some valves
showed improved sealing performance in each consecutive week of testing.

42

Figure 2-22. Representative data from one stationary micropump wirelessly operated. (a)
Bolus delivery from a micropump operated once a day during week two (at 0.33 mA
constant current) and (b) flow rates for the three weeks dosing (reprinted from [10] with
permission from Elsevier).
2.5.4 Back pressure
Minimal variation in flow rate compared to the value at 0 mmHg back pressure was
observed up to a back pressure of 20 mmHg. Flow rate decreased slightly as the back
pressure increased, but remained within 10% of the baseline flow rate of each pump
recorded at 0 mmHg. No significant differences were observed between micropumps flow
rates when normalized to 1, as confirmed by statistical analysis (ANOVA, p < 0.05). The
data was plotted using normalized flow rate values (Figure 2-23).

43

Figure 2-23. The variation in flow rate observed in four micropumps when pumping against
physiological back pressures (up to 20 mmHg). Flow rate was normalized across pumps at
a back pressure of 0 mmHg (reprinted with permission from [27] © 2014 IEEE).
2.5.5 Viscosity
The same four micropumps were tested with solutions of varying viscosities and no
significant differences in flow rates were observed. The baseline flow rate is defined by the
value measured by delivering Millipore water with a 1 cP viscosity at room temperature.
Flow rates were within 10% from the baseline flow rate for each pump tested. While the
mean flow rates across the pumps were different as previously explained, the normalized
flow rates between micropumps were not significantly different as confirmed by statistical
analysis (ANOVA, p < 0.05; Figure 2-24).

44

Figure 2-24. The variation in flow rate observed in four micropumps when delivering
calibrated glucose solutions of differing viscosity. Blood viscosity is 3 to 4 cP at body
temperature (1cP = 1mPa∙s). Flow rate was normalized across pumps at a viscosity of 1
cP (reprinted with permission from [27] © 2014 IEEE).
2.5.6 Room temperature versus body temperature
Figure 2-25 shows representative results of the flow rate performance at room
temperature (23 °C) and body temperature (37 °C) for one micropump (mean ± SE, n = 4).
As expected, an increase in flow rate was apparent at body temperature. For two
micropumps, changes in temperature resulted in significant flow rate differences as
confirmed by performing a one-way analysis of variance (ANOVA, p < 0.05).

45

Figure 2-25. Representative data from one wireless micropump flow rate performance
when increasing the environmental temperature from room (23 ºC) to body temperature
(37 ºC) (at 0.33 mA constant current) (reprinted from [10] with permission from Elsevier).
2.5.7 Physiological environment simulation
A 30 day trial with three fully encapsulated and packaged pumps immersed in a 1
X PBS bath at body temperature (37 ºC), showed continued accurate and reliable flow rate
performance with a standard error of less than 4% for each pump. Figure 2-26 shows
representative results of the flow rate performance for one micropump (mean ± SE, n = 4).
This micropump was previously used for the three-week daily dosing study, but has a new
receiver. Mean flow rates between weeks for each pump are not significantly different as
confirmed by statistical analysis (ANOVA, p > 0.05).

46

Figure 2-26. Representative flow rate performance results for a micropump under
simulated in vivo conditions for 30 days (at 0.33 mA constant current) (reprinted from [10]
with permission from Elsevier).

Discussion
The pump components were redesigned to optimize pump performance and ease of
use. The new reservoir dome features two refill ports as oppose to one to facilitate filling
and flushing of the reservoir. The refill ports are robust enough to prevent leaks after high
number of punctures under high reservoir pressure. The new redesigned rigid reservoir
significantly decreased device footprint by decreasing dead volume by 91%. The total
filled encapsulated device mass including wireless power components was 3.8 g, slightly
exceeding the mass requirements for a mouse experiment (3 g or approximately 10% of an
adult mouse weight of 30 g). The encapsulation was identified as the heaviest component.
Therefore, the pump mass can be reduced by eliminating the encapsulation material and
instead using a low density polymer outer casing to enclose the current system. An outer
casing was designed to function as a fluid protective barrier for the system, a scaffold to
keep components in place, and to provide smoother surfaces to minimize damage to the
surrounding tissues. The casing was produced using SL in DSM Somos® WaterShed XC
11122 material (FineLine Prototyping, Inc., Raleigh, NC) as shown in Figure 2-27.

47
Components were designed to compression fit together to decrease the amount of epoxy
used in the assembly of the previous design.

Figure 2-27. Photograph of micropump outer casing produced using stereolithography.
Casing components shown (a) individually and (b) assembled.

Fully integrated wireless stationary micropumps successfully delivered 30 L daily
boluses at low flow rates (~ L/min) for several weeks, a regimen relevant for the
administration of siRNA therapy to a radiation resistance cancer. The 2.8 fold increase in
flow rate observed during the third week of the daily dosing study was attributed to a faulty
receiver. Due to this systematic error, flow rate variation was calculated after receiver was
replaced (4%). Since we previously demonstrated consistent actuator performance [23],
slight variations in mean flow rate were attributed to variations due to manual assembly of
the packaging and receiver coils and inconsistent performance in the commercially
produced valves. Also, valves permitted up to 33% of dose volume in back flow when the
pump was turned off.
Some delay in forward flow during current application was present in all
micropumps tested and is due to the low current application [45], and valve cracking
pressure. This delay was further increased in daily dosing regimens where the reservoir
was not refilled and higher pressure build up was required in order to deliver the same dose
volume [23]. As discussed in [46], the forward flow delay prior to refill follows a
predictable pattern and can be accounted for by increasing the current application time,
thus our flow rate variation of 4% does not include the forward flow delay. The forward

48
flow delay in our previous prototypes was minimized by replacing the compliant reservoir
material with a rigid material and utilizing higher efficiency electrolysis electrodes [24, 32].
The dosing regimen can be altered depending on the intended application. A diverse
assortment of liquid drug formulations can be delivered within a wide dynamic range of
dose volumes (nL – µL) and flow rates (0.33 – 141.9 µl/min) [23]. The bellows can safely
deliver 180 µl. Once this volume is reached the actuator should be allowed to fully
recombine prior to the next dosing event. Full recombination is achieved within several
hours [47]. While the pump can store medication for long term, it is important for the drug
to be stable when stored at body temperature between refill events.
Previous prototypes of this pump included commercial valves with high sealing
pressures [24] that showed significant reverse flow during recombination. For in vivo
studies reverse flow needs to be avoided in order to obtain safe and accurate dosing. The
check valve incorporated into this system was the only small commercial valve with low
sealing pressure and cracking pressure. Unfortunately, the valve manufacturing process led
to variability and inconsistency in valve performance (Figure 2-28). Thus, valves were
screened before incorporation into the pump and only 50% of the valves passed screening
(reverse leakage < 20% of dose volume). Variable and inconsistent performances of the
valve resulted in additional variability among pump performance. Therefore, a high quality
custom-made normally closed spring-loaded ball check valve was developed. Valves were
incorporated at the reservoir outlet of a wired pump to prevent backflow of fluids as a result
of the reverse pressure gradient caused by recombination. The check valve was
successfully characterized using a wired micropump with <1% reverse leakage. The
cracking pressure needs to be decreased to be overcome by the wireless system in a timely
fashion (within 5 min). Cracking pressure can be further decreased by utilizing springs with
lower spring rate and removal of the O-ring from the valve seat. Additionally, valve
housing components should be precision machined to achieve smoother surfaces and
prevent other sources of leaking.

49

Figure 2-28. Photographs of commercial Duckbill valve prior to incorporation into our
system. Valve slit shows characteristics of a normally opened valve and variability among
valves.

A class D wireless power system was developed to power an electrochemical
actuator in [23]. While it was simple in design and provided large currents, this class D
system had several drawbacks including difficulty in fine tuning the amplifier and inability
to generate the required power without overheating the transmitter circuit components. A
class E system combines the advantages of the class D system with its series-tuned output
and parallel-tuned amplifier tanks. Class E systems can cope with the rigorous demands of
weakly coupled links, produce smooth signals, be insensitive to low switching time, and
operate over several tens of MHz. This pump was powered with a class E powering system,
which allows higher efficacy when a considerable amount of power is to be transferred
[48]. Pump flow rate is linearly dependent on applied current to the actuator [32]. A current
regulator was chosen as part of the implanted receiver circuit to ensure the current applied
to the actuator to be relatively constant (within certain limits of distance and angular
misalignment between the transmitting and receiving coils) regardless of slight changes in
the power. Current regulators are inherently lossy components, as such, power efficiency
is partially sacrificed to ensure proper pump operation. Flow rate performance for an
individual pump varied when it was powered with different receivers. Variations in flow
rate remained within less than 6%. This slight variation in performance can be attributed
to the manual assembly of receiver coils, but can be alleviated by calibrating each pump
prior to in vivo studies. As expected, a decrease in power transferred due to increased axial
separation distance and angular misalignment between concentric transmitter and receiver

50
coils resulted in decreased in flow rate. Pump flow rate stability can be further improved
by increasing the transmitted power [36]. The addition of multiple coils at either the
transmitter or receiver side can improve power transfer to mitigate angle misalignment.
Another potential solution is to move the external receiving coil along with the moving
animal by tracking the movement of the internal transmitting coil with a similar setup to
the one proposed by Kilinc, et. al. [49].
The effects of environmental factors on pump performance were investigated. We
demonstrated that flow rate is not significantly affected by back pressures of up to 20
mmHg (when delivering solutions of 0 cP) or delivering solutions with viscosities of up to
6 cP. Changing the environmental temperature from room temperature (23 °C) to body
temperature (37 °C) for our wireless micropumps resulted in increased flow rate. Improved
electrolysis efficiency can be attributed to the increased permeability of Nafion® to
hydrogen and oxygen when the ambient temperature is increased [50]. This is in agreement
with [23], which found slight increase in flow rate at body temperature for electrochemical
actuators. Physiological environmental simulation test showed that all three pumps were
functioning consistently and as expected after 30 days of soaking in 1X PBS at body
temperature (37 °C) exceeding the intended in vivo study duration by 100%. This verifies
that the Parylene C and silicone rubber encapsulation layers provide adequate protection
for the electronics for the duration of the anticipated in vivo study. This packaging scheme
may also be suitable for longer term in vivo studies beyond 30 days.
Conclusion
We demonstrated the characterization of a fully implantable wireless infusion
micropump for use in chronic drug delivery therapies in small research animals. Wireless
control of operation permits a user to select the desired delivery regimen in a freely moving
animal. Wireless powering of the actuator with a class E inductive powering system
allowed repeatable delivery with less than 6% variation in flow rate. Pump performance
was affected by increased axial separation distance and angular misalignment between
concentric transmitter and receiver coils. If necessary, this can be alleviated by increasing

51
the power transmitted and inclusion of a second receiver coil [36]. Wireless powering with
an implantable device requires inductive coupling link through the skin and biological
tissue. As such, it is essential to characterize device performance in vivo. Future work will
include characterization of system operation in simulated tissue material to mimic device
operation in animals.
Stationary micropumps successfully delivered a single bolus daily for several
weeks and achieved flow rates on the order of L/min, an experimental regimen relevant
for an anti-cancer drug delivery application. Commercial valves decreased reverse flow
compared to a valve less system, but their performance has a lot of variability. A custom-
made normally closed spring-loaded ball check valve was designed and successfully
characterized with a wired system. The valve was easy to fabricate and significantly
decreased reverse flow. However, the cracking pressure was high and will need to
decreased prior to incorporation into the wireless system. We demonstrated that flow rate
is not significantly affected by back pressures of up to 20 mmHg (when delivering solutions
of 1 cP) or delivering solutions with viscosities of up to 6 cP. As expected, micropump
flow rate increased with increased environmental temperatures. Consistent micropump
performance was also demonstrated in simulated physiological environmental conditions
with less than 4% flow rate variation.
Drug delivery micropumps need to incorporate closed-loop feedback systems to
enable pump performance monitoring, and increase effective therapies. Control over
delivery profile can be achieved through the incorporation of physical sensors that provide
information on pressure, flow rate, delivered volumes, forward flow delay and real-time
state of the pump. A variety of sensing techniques are available for MEMS flow sensors,
such as electrical, nuclear, thermal, mechanical and optical acoustic. An electrochemical
impedance (EI) dose tracking system capable of tracking accidental release of reservoir
content, blockage and refill detection or lack of delivery in real time has been developed
and was presented on [51, 52].

52
References
[1] S. Reagan-Shaw, M. Nihal, and N. Ahmad, "Dose translation from animal to human
studies revisited," The FASEB Journal, vol. 22, pp. 659-661, 2008.
[2] C. C. Linder and M. T. Davisson, "Chapter 1.2 - Historical Foundations," in The
Laboratory Mouse (Second Edition), H. J. Hedrich, Ed., ed Boston: Academic Press,
2012, pp. 21-35.
[3] J. W. F. John Urquhart, and Kay L. Willis, "Rate-Controlled Delivery Systems in
Drug and Hormone Research," Ann. Rev. Pharmacol. Toxicol, vol. 24, pp. 199-236,
1984.
[4] D. Paolino, P. Sinha, M. Fresta, and M. Ferrari, "Drug delivery systems,"
Encyclopedia of Medical Devices and Instrumentation, 2006.
[5] R. Langer, "Drug delivery and targeting," Nature, vol. 392, pp. 5-10, 1998.
[6] M. H. Smolensky and N. A. Peppas, "Chronobiology, drug delivery, and
chronotherapeutics," Advanced Drug Delivery Reviews, vol. 59, pp. 828-851, 8/31/
2007.
[7] F. Lévi, C. Focan, A. Karaboué, V. de la Valette, D. Focan-Henrard, B. Baron, et al.,
"Implications of circadian clocks for the rhythmic delivery of cancer therapeutics,"
Advanced Drug Delivery Reviews, vol. 59, pp. 1015-1035, 8/31/ 2007.
[8] P. V. Turner, T. Brabb, C. Pekow, and M. A. Vasbinder, "Administration of
substances to laboratory animals: routes of administration and factors to consider,"
Journal of the American Association for Laboratory Animal Science: JAALAS, vol.
50, p. 600, 2011.
[9] P. V. Turner, C. Pekow, M. A. Vasbinder, and T. Brabb, "Administration of
substances to laboratory animals: equipment considerations, vehicle selection, and
solute preparation," Journal of the American Association for Laboratory Animal
Science: JAALAS, vol. 50, p. 614, 2011.
[10] A. Cobo, R. Sheybani, H. Tu, and E. Meng, "A wireless implantable micropump for
chronic drug infusion against cancer," Sensors and Actuators A: Physical, vol. 239,
pp. 18-25, 3/1/ 2016.
[11] Med-E-Cell. (2014, 3/25/2014). Welcome toMed-e-Cell: infusion pumps, fluid
delivery devices, oxygen generation/control and fuel cell technology for licensing and
development partnerships. Available: http://medecell.com
[12] iPrecio. (2014, 3/25/2014). iPRECIO®, Innovative Drug Infusion Technology for
Laboratory Animals. Available: http://www.iprecio.com

53
[13] Alzet. (2014, 3/25/2014). ALZET® Osmotic Pumps - Implantable pumps for research.
Available: http://www.alzet.com
[14] E. Meng and T. Hoang, "MEMS-enabled implantable drug infusion pumps for
laboratory animal research, preclinical, and clinical applications," Advanced Drug
Delivery Reviews, 2012.
[15] R. Sheybani, S. M. Schober, and E. Meng, "Drug Delivery Using Wireless MEMS,"
in Handbook of MEMS for wireless and mobile applications, Uttamchandani, Ed., ed:
Woodhead Publishing, 2013, pp. 489-517.
[16] A. Cobo, R. Sheybani, and E. Meng, "MEMS: Enabled Drug Delivery Systems,"
Advanced healthcare materials, vol. 4, pp. 969-982, 2015.
[17] P. Sun, Y. Zhang, F. Yu, E. Parks, A. Lyman, Q. Wu, et al., "Micro-
electrocardiograms to study post-ventricular amputation of zebrafish heart," Annals
of biomedical engineering, vol. 37, pp. 890-901, 2009.
[18] K. W. Oh and C. H. Ahn, "A review of microvalves," Journal of Micromechanics and
Microengineering, vol. 16, p. R13, 2006.
[19] A. K. Au, H. Lai, B. R. Utela, and A. Folch, "Microvalves and micropumps for
biomems," Micromachines, vol. 2, pp. 179-220, 2011.
[20] C. H. Ahn and J.-W. Choi, "Microfluidic devices and their applications to lab-on-a-
chip," in Springer Handbook of Nanotechnology, ed: Springer, 2010, pp. 503-530.
[21] SmartProducts. (2016, 08/20/2016). Check Valves Available:
http://www.smartproducts.com/check_valves_series_100_cartridge_specialty.php
[22] miniValve. (08/20/2016). Duckbill-Valves. Available:
http://www.minivalve.com/newsite/index.php/en/by-type/duckbill-
valves/components
[23] R. Sheybani, H. Gensler, and E. Meng, "A MEMS electrochemical bellows actuator
for fluid metering applications," Biomedical Microdevices, pp. 1-12, 2012/07/01 2012.
[24] H. Gensler, R. Sheybani, P.-Y. Li, R. L. Mann, and E. Meng, "An implantable MEMS
micropump system for drug delivery in small animals," Biomedical Microdevices, vol.
14, pp. 483-496, 2012.
[25] H. Gensler, R. Sheybani, L. Po-Ying, R. Lo, S. Zhu, Y. Ken-Tye, et al., "Implantable
MEMS drug delivery device for cancer radiation reduction," in 23rd IEEE
International Conference on Micro Electro Mechanical Systems, MEMS 2010, Hong
Kong, 2010, pp. 23-6.

54
[26] R. Masood, I. Roy, S. Zu, C. Hochstim, K.-T. Yong, W.-C. Law, et al., "Gold
nanorod–sphingosine kinase siRNA nanocomplexes: a novel therapeutic tool for
potent radiosensitization of head and neck cancer," Integrative Biology, vol. 4, pp.
132-141, 2012.
[27] A. M. Cobo, H. M. Tu, R. Sheybani, and E. Meng, "Characterization of a Wireless
Implantable Infusion Micropump for Small Animal Research Under Simulated In
Vivo Conditions," in IEEE Biomedical Circuits and Systems, Lausanne, Switzerland,
2014.
[28] R. Sheybani, A. Cobo, and E. Meng, "Wireless programmable electrochemical drug
delivery micropump with fully integrated electrochemical dosing sensors,"
Biomedical microdevices, vol. 17, p. 74, 2015.
[29] P.-Y. Li, R. Sheybani, C. A. Gutierrez, J. T. W. Kuo, and E. Meng, "A Parylene
bellows electrochemical actuator," Journal of Microelectromechanical Systems, vol.
19, pp. 215-28, 2010.
[30] S. Grigoriev, P. Millet, K. Dzhus, H. Middleton, T. Saetre, and V. Fateev, "Design
and characterization of bi-functional electrocatalytic layers for application in PEM
unitized regenerative fuel cells," international journal of hydrogen energy, vol. 35,
pp. 5070-5076, 2010.
[31] T. Ioroi, K. Yasuda, Z. Siroma, N. Fujiwara, and Y. Miyazaki, "Thin film
electrocatalyst layer for unitized regenerative polymer electrolyte fuel cells," Journal
of Power sources, vol. 112, pp. 583-587, 2002.
[32] R. Sheybani and E. Meng, "High-Efficiency MEMS Electrochemical Actuators and
Electrochemical Impedance Spectroscopy Characterization,"
Microelectromechanical Systems, Journal of, vol. 21, pp. 1197-1208, 2012.
[33] R. Ghodssi and P. Lin, MEMS materials and processes handbook vol. 1: Springer
Science & Business Media, 2011.
[34] P.-Y. Li, R. Sheybani, C. Gutierrez, J. T. Kuo, and E. Meng, "A parylene bellows
electrochemical actuator," Microelectromechanical Systems, Journal of, vol. 19, pp.
215-228, 2010.
[35] H. Gensler and E. Meng, "Rapid fabrication and characterization of MEMS Parylene
C bellows for large deflection applications," Journal of Micromechanics and
Microengineering, vol. 22, p. 115031, 2012.
[36] B. Lenaerts and R. Puers, Omnidirectional inductive powering for biomedical
implants: Springer, 2009.

55
[37] T. K. Givrad, Induction power microbolus infusion pump used for functional
neuroimaging applications in rodents: ProQuest, 2007.
[38] K. van Schuylenbergh and R. Puers, Inductive Powering: Basic Theory and
Application to Biomedical Systems: Springer Science & Business Media, 2009.
[39] H. Gensler, R. Sheybani, and E. Meng, "A MEMS Micropump System with One-Way
Valve for Chronic Drug Delivery," in Microtechnologies in Medicine and Biology,
Marina del Rey, CA, 2013.
[40] The Johns Hopkins University. (2015, 3/1/2015). Animal Care and Use Committee:
The Mouse [and] The Rat. Available:
http://web.jhu.edu.libproxy.usc.edu/animalcare/procedures/
[41] DSM Somos Materials Group, "Somos®WaterShed XC 11122," ed, 2012.
[42] D. Somos. (2008). DSM Somos WaterShed XC 11122 Passes USP Class VI Testing:
Now Suitable For Prototyping In Biomedical Or Skin Contact Applications.
Available: http://www.meddeviceonline.com/doc/dsm-somos-watershed-xc-11122-
passes-usp-0001
[43] M. E. Klingensmith, The Washington Manual of Surgery Philadelphia: Wolters
Kluwer Health/Lippincott Williams & Wilkins, 2008.
[44] T. A. Goers, "The Washington Manual of Surgery," ed: Lippincott Williams &
Wilkins, 2008.
[45] C. R. Neagu, A medical microactuator based on an electrochemical principle:
Universiteit Twente-Elektrotechniek, 1998.
[46] H. M. Gensler, A Wireless Implantable MEMS Micropump System for Site-specific
Anti-cancer Drug Delivery: University of Southern California, 2013.
[47] R. Sheybani and E. Meng, "Acceleration Techniques for Recombination of Gases in
Electrolysis Microactuators with Nafion®-Coated Electrocatalyst," Sensors and
Actuators B: Chemical, vol. 221, pp. 914-922, 2015.
[48] R. Puers, Omnidirectional inductive powering for biomedical implants: Springer,
2008.
[49] E. G. Kilinc, K. Kapucu, F. Maloberti, and C. Dehollain, "Servo-controlled remote
powering and low-power data communication of implantable bio-systems for freely
moving animals," in Biomedical Circuits and Systems Conference (BioCAS), 2014
IEEE, 2014, pp. 508-511.

56
[50] K. Broka and P. Ekdunge, "Oxygen and hydrogen permeation properties and water
uptake of Nafion® 117 membrane and recast film for PEM fuel cell," Journal of
Applied Electrochemistry, vol. 27, pp. 117-123, 1997.
[51] R. Sheybani, N. E. Cabrera-Munoz, T. Sanchez, and E. Meng, "Design, fabrication,
and characterization of an electrochemically-based dose tracking system for closed-
loop drug delivery," in Engineering in Medicine and Biology Society (EMBC), 2012
Annual International Conference of the IEEE, 2012, pp. 519-522.
[52] R. Sheybani, A. Cobo, and E. Meng, "Wireless programmable electrochemical drug
delivery micropump with fully integrated electrochemical dosing sensors,"
Biomedical microdevices, vol. 17, pp. 1-13, 2015.

57

The potential of peripheral nerve (PN) interfaces has been marred by the limited in
vivo lifetime of current technology due to biological and non-biological failure modes. If
PN interfaces functioned reliably under chronic conditions (in the range of multiple years),
the promise of restoring sensory and motor control in amputees can be realized. In an effort
to improve the reliability of the current state-of-the art, we present a novel Parylene-based
approach, leveraging Parylene’s thin, flexible substrate and compatibility with batch
microfabrication techniques with a unique drug delivery system to enhance implant-tissue
integration in vivo and prolong chronic performance.
Background
During the past several decades, research groups have been devoted to restoring
motor and sensory functions lost to amputation, injury, or degenerative diseases. Current
prosthesis substitute body parts (e.g. hand, leg) and help restore some degree of
functionality, but lack a mechanism for sensory feedback and intuitive control of the
prosthetic [1, 2]. Biological signals (e.g. electrical, mechanical, chemical) can be used to
couple prosthetics with the central nervous system (CNS) to achieve feedback control of
the prosthetic. The nervous system is a highly complex structure compromised of networks
of neurons that process and transmit information through electrical signals as action
potentials and through chemical signals such as neurotransmitters to control all conscious
and unconscious human functions [3]. The peripheral nervous system (PNS) is a division
of the nervous system that consists of axons of neural cell bodies located in the spinal cord.
Peripheral nerves contain efferent motor and afferent sensory fibers within fascicles
enclosed by layers of connective tissue (epineurium and perineurium, Figure 3-2) that
eventually innervate limbs and organs. The PNS relays sensory information to the CNS via

LYSE-AND-ATTRACT CUFF ELECTRODE

58
the afferent fibers and executes motor commands generated in the CNS via the efferent
fibers [3]. Electrical activity recorded along the nerve arises from multiple units and is
known as compound action potentials (CAPs) because it is the summation of action
potentials from individual axons. The magnitude of spontaneous PNS CAPs is usually in
the µV range, while the majority of evoked CAPs can reach amplitudes in the mV range
[4-8]. Figure 3-1 shows an example of a CAP recorded at the sciatic nerve of an adult cat.
Interfacing with the PNS allows direct neural signal recording and stimulation of intact
nerves or muscles in a less invasive and more physiologically based manner [9].
Additionally, interfacing the PNS allows other forms of sensory feedback besides visual,
such as tactile and proprioceptive sensation which are necessary to provide realistic and
fine motor control. [10] [11].

Figure 3-1. A compound action potential at the sciatic nerve of an adult cat recorded with
a silicone cuff electrode. The amplitude of the CAP is shown in voltage (modified from [7]
with permission from John Wiley and Sons).

Neural technologies are needed to interface with the PNS to create a fast and
reliable bidirectional link between the nervous system and prosthetics [12-15] (Figure 3-3).
Various approaches have been developed to create stable peripheral nerve devices, and the
most common type consists of microelectrodes placed around or inside nerve bundles,
enabling neural signal recording and stimulation from motor and afferent nerve fibers,
respectively. The ideal PNS interface provides selective stimulation of different muscle
motor units, allows access to sensory afferent information and satisfies material constraints,
including mechanical compliance to reduce soft-tissue damage and biocompatible
composition to avoid foreign body reaction [16]. Designs amenable to conventional batch-

59
scalable microfabrication offer increased reliability and design complexity while
maintaining a minimally invasive form-factor.

Figure 3-2. Schematic diagram of the anatomy of a peripheral nerve.

Figure 3-3. Peripheral nerve interfaces necessary for brain-controlled prostheses.

Electrical interfaces to the peripheral nerves are either extraneural or
intrafascicular/intraneural with each design having specific advantages and disadvantages
(Figure 3-4). Extraneural interfaces, also called cuff electrodes, encircle the circumference

60
of the nerve bringing electrodes in direct contact with the collagenous sheath (epineurium)
of the nerve. These minimally invasive interfaces consist of compliant sheath materials that
allow conformal fitting of the electrodes to the shape of the nerve, improving contact and
mitigating nerve damage. The signal to noise ratio (SNR) is significantly lower than
penetrating electrodes due to the epineurium that surrounds the nerve and limits access to
individual nerve fibers. Stimulation selectivity can be improved by incorporating multisite
electrodes and structures, novel stimulation parameters and by confining current
application to the inner space of the interface [17-19]. However, stimulation through these
interfaces still result in high crosstalk to adjacent fascicles. Self-sizing and locking
mechanisms allow ease of placement, prevent nerve damage while maintaining electrodes
close to the surface of nerve of varying diameters [17, 20].

Figure 3-4. (a) Extraneural interfaces wrap around the nerve and record/stimulate from the
nerve surface. (b) Intraneural interfaces penetrate the nerve gaining greater access to
individual nerve fibers.
Improved access to individual fascicles and reduced stimulus charge injection has
been achieved by extraneural interfaces that reshape the nerve, (i.e. flattening),
demonstrated by Durand et al. with the flat interface nerve electrode (FINE) [21]. However,
the SNR is still restricted by the impedance of the epineurium, and large nerve fibers
damage can be induced by the snug-fitting of the interface resulting in compression and
reshaping of the nerve [7, 22]. The simplicity of handling and the ability to record and
stimulate general nerve activity make extraneural electrodes ideal for chronic applications.
Still, additional work is required to improve their selectivity currently constrained to
superficial nerve fibers which limits their chances of providing fine control over prosthetics.

61
Intrafascicular/intraneural electrodes gain greater access to individual nerve fibers
by placing electrical contacts within the nerve. They are significantly more invasive,
resulting in improved SNR of recordings and stimulation selectively [23, 24]. Electrodes
may either be inserted longitudinally into the nerve and run parallel to nerve fibers, or
transversely, and run perpendicular to fibers. Longitudinally implanted intrafascicular
electrodes (LIFEs) have excellent selectivity compared to extraneural interfaces, but
require difficult implantation surgery, limiting its use and allow stimulations only of the
nerve fibers within the fascicle implanted. Transverse electrodes such as the transverse
intrafascicular multichannel electrode (TIME), penetrate the nerve fascicles achieving
single unit recordings from several nerve fibers in different fascicles, resulting in low
current stimulation and high spatial selectivity [25, 26]. Of the two methods, transverse
interfaces are associated with greater risk of nerve damage and inferior chronic-
performance. Invasively inserted electrodes are fabricated from materials with elastic
moduli several orders of magnitude greater than that of the nerve in order to penetrate the
tough connective tissue, contributing to chronic foreign-tissue response, nerve damage due
to electrode micromotion within the soft tissue of the nerve, and the degradation of
recording/stimulation stability over time due to electrode encapsulation with scar tissue
[27, 28]. Placement of intranueral interfaces is also difficult, due to the small size of both
the recording sites (~10-20 µm diameter) and the targeted nodes of Ranvier (~ 1 µm long)
[24, 27, 29, 30]. All of these contribute to the very poor in vivo chronic performance
observed in intrafascicular/intraneural electrodes. To our knowledge, chronic in vivo
implantation of transverse and longitudinal intrafascicular interfaces has only been
reported up to 2 to 6 months, respectively [31-34].
Neither approach has achieved the desired selectivity while minimizing
invasiveness, and all prior attempts suffered from a lack of chronic reliability (Figure 3-5).
To overcome these hurdles and achieve a long-term, stable interface with high fidelity
recordings, we developed a minimally invasive approach in which a polymer-based,
extraneural interface combines microelectronics with microfluidics to target individual
fascicles within a nerve by inducing the sprouting of axonal collaterals towards electrodes

62
embedded in microfluidic channels. This approach attempts to achieves high stability and
fascicle specificity without physically damaging or penetrating the nerve. The proposed
induction of collateral sprouting is based on the natural capacity of intact healthy axons to
grow outside the nerve, and is applied in surgical nerve reconstruction as the end-to-side
neurorrhaphy technique [35, 36]. Intact sensory and motor axons appear to have
comparable capacity for collateral sprouting [37]. The sprouting can be enhanced by
microsurgical removal of the connective sheaths (epineurial and perineurial layers) of the
nerve [38-40]. In contrast to the microsurgical removal of epineurium and perineurium,
we propose an enzymatic removal of collagen fibers to avoid the risk of a nerve injury [41].
Application of exogenous factors such as neurotrophic factors (e.g. NGF,
methylcobalamin) can significantly improve collateral sprouting from an intact nerve [42-
44]. Neurotrophic factors for collateral sprouting are commonly delivered via frequent
intraperitonially or epineural injections or silicone reservoir chambers [44-47]. Injections
result in additional mechanical injury and risk of infection, while the reservoir chambers
have a limited drug payload. Improved delivery of exogeneous factors can be achieve via
implantable microfabricated microfluidics to allow for chronic, localized, and aseptic
delivery.
Axonal sprouting will be selectively induced from the fascicles closest to the
microfluidic outlet ports via localized delivery of lysing agents to temporarily disrupt the
epineurium, followed by delivery of neurotrophic factors to promote axonal sprouting
towards electrodes within the microchannels. The design comprises a thin-film polymer
‘cuff’ with adjustable diameter to ensure localized diffusion of chemicals into the
epineurium of nerves of varying diameters. This Lyse-and-Attract Cuff Electrode (LACE)
aims to increase fascicular selectivity and signal-to-noise ratio for recording and/or
stimulation without traumatic invasiveness, while providing long-term stability beyond the
current state of the art.

63

Figure 3-5. The two types of electrodes applied to interface peripheral nerves classified
regarding invasivity and selectivity.

Design
3.2.1 Microfluidic channels
Embedding axons in microfluidic channels can improve proximity to recording
sites, amplify extracellular axonal signals, increase selectivity, and decrease cross-talk
between recording sites. Prior work has shown that the dimensions of the channel are
critical to this approach. The extracellular signal amplitude increases with decreasing
cross-sectional area and increasing channel length [48, 49]. Additionally, Loeb et al.
reports that axons will not regenerate in long, narrow channels (length > 1 mm) without a
blood supply, due to a lack of nutrients and oxygen [48]. Channels with greater cross-
sectional area (cross-sectional area ~ 10,000 µm
2
) allow for vascularization and axonal
regeneration in channels with up to 5 mm in length, as presented in [50]. Histology images
from this study indicate that axons and blood vessels were concentrated in the center third
of the channel cross-section. The remainder of the channel filled with fibrous tissue,
resulting from the induced sciatic nerve injury. The biochemically induced axonal
sprouting method, employed in the LACE design, will ensure minimal fibrous tissue
formation and encourage axonal regeneration and vascularization through long channels;
therefore, the dimensions for our channel include a minimum cross-sectional area of 5,000

64
µm
2
and a functional length of 5 mm.
The LACE is fabricated from thin-film polymer Parylene C, and incorporates 4
surface micromachined microfluidic channels (250 µm W × 20 µm H) for targeting
multiple fascicles in a rat sciatic nerve Figure 3-6. Parylene C is a USP class VI
biocompatible polymer, inert, and with a low Young’s modulus (~2-3 GPa) to more closely
match the mechanical properties of the surrounding tissue [51, 52]. In addition, the high
dielectric strength of Parylene C also provides excellent electric insulation which provides
isolation of recording channels. Microfluidic channels will each serve to deliver both lysing
agents and neurotrophic factors to the PN. Microfluidic channel outlet ports need to be
greater than 100 µm in diameter to encourage active and vascularized axons to grow into
the microfluidic channel for high signal to noise ratio (SNR) recording. Additionally,
staggered microfluidic outlet ports are necessary to reduce diffusion/interaction between
different channels. A microfluidic channel branch design was chosen to ensure equal
pressure drop, and hence flow rate, across all channels (Figure 3-7). The microfluidic
channels include a port for external fluidic connection.

Figure 3-6. (a) A photograph showing the branching pattern of the rat sciatic nerve with
the location of where the average diameter was measured indicated by the white arrow. (b)
Cross-section through the rat sciatic nerve, showing the four fascicles: sural (s), tibial (t),
peroneal (p), and cutaneous (c) (modified from [53] with permission from John Wiley and
Sons).

65

Figure 3-7. Schematic diagram of the microfluidic channels showing the branching design
and dimensions.
3.2.2 Electrodes
Each microfluidic channel contains a pair of platinum (Pt) electrodes (300 µm W ×
1500 µm L) for differential recording and/or stimulation. Pt was selected as the electrode
material due to high resistance to corrosion in in vivo environments, low decomposition
characteristics for a stimulating electrode, and high charge injection limit [54, 55].
Electrodes dimensions were chosen to obtain a low electrode impedance range of 2 – 5 kΩ
(for surface area of approximately 0.45 mm
2
) (Table 3-1). This low impedance will
improve recording and stimulation by improving the SNR and allowing the detection of
spontaneous compound action potentials (CAP) in unmyelinated nerve fibers. Our focus
will remain on spontaneous CAPs to avoid the possibility of mistaking the stimulation-
induced electromyography (EMG) artifact for CAP, as unmyelinated axons have a long
evoked potential latency similar to that of evoked EMG. Electrodes are rectangular shaped
to fit within limited cuff real estate and electrode corners are rounded to avoid high current
density at sharp edges [56]. Two electrodes are required at the bottom of each integrated
microfluidic channel (Figure 3-8), with an electrode pitch of 3 - 4 mm for electrodes within
the same channel to ensure differential recording and/or stimulation. A surface electrode,
outside the microfluidic channel, will be required as a reference. A single one ground
electrode is necessary and will be wired manually, external to the LACE device. A Parylene
ribbon cable (~2 cm long) will be incorporated for external electrical connection.

66

Table 3-1. Reference values for Pt electrode impedance as a function of electrode surface
area.
Electrode area
(mm
2
)
Impedance magnitude
@ 1 kHz
(k ꭥ)
Reference
0.7 0.9 - 4 [57]
0.07 8.3 [58]
0.009 35.5 [59]

Figure 3-8. Cross sectional view of Parylene-based PN cuff electrode showing electrode’s
placement within a microfluidic channel. Inter-electrode spacing is not to scale.
3.2.3 Interlocking mechanism

The cuff structure consists of an elongated tapered tab that loops around the sciatic
nerve with the microfluidic outlets in direct contact with the circumference of the
epineurium, inserts into an etched slit, and is held in position by etched metal reinforced
serrations (Figure 3-9 and Figure 3-10). A two-part locking mechanism is used to maintain
the loop around the nerve, and includes an additional slit through which the tab is threaded
to create a secure buckle structure. The guide needle hole is utilized to pass a suture from
the circular hole in the tab and facilitate guiding of the tab into the slits.
The cuff is adjustable (1.1-1.5 mm diameter, 0.1 mm increments) to provide close
fits to nerves of varying diameter. Device dimensions were chosen to fit the sciatic nerve
of a 340 g adult male rat with an average sciatic nerve diameter ranging from 1.1 to 1.4

67
mm. The horizontal diameter of the sciatic nerve increases at the fascicle separation zone,
but the nerve also flattens resulting in a consistent total diameter size. 12 mm length of the
sciatic nerve can be easily exposed to allow the insertion and placement of a cuff up to 10
mm in length.

Figure 3-9. Schematic of LACE for targeting individual fascicles within a nerve. Insert
shows the cross-sectional view.

Figure 3-10. SolidWorks design of LACE electrode for sciatic nerve recording and
stimulation. Units in mm.

68
Fabrication
3.3.1 Sham devices
Non-functional (sham) devices which we define as not having electrodes or any
other metal features, consisted of fluidic channels with varying inlet channel designs,
interlocking mechanism with varying gap between slits, and several guide holes at the tab.
The purpose of the sham devices was to confirm the fabrication process for creating the
microfluidic and interlocking features. Kic software was utilized to design the
photolithographic masks as shown in Figure 3-11. LinkCAD and CorelDRAW software
were utilized to generate each mask from the Kic file and arrange the masks into one image
file, subsequently printed on a transparency mask by Fineline Imaging (Appendix B). The
transparency images were then transferred to glass/chrome photomasks, which were used
for photolithographic fabrication.

Figure 3-11. Kic software masks. (a) Complete sham device, (b) locking mechanism
serenated teeth, (c) microfluidic channels with ports, and (d) final sham device mask layout
(100 mm wafer).

A fabrication process flow was developed and optimized for the fabrication of the
sham devices. Devices were fabricated on a bare silicon wafer for mechanical support

69
during the microfabrication process (Figure 3-12). A 10 m Parylene base layer was first
deposited by chemical vapor deposition (CVD). Sacrificial photoresist AZ 4620 was spun
to a height of 20 m and patterned to form the microfluidic channels. An 8 m Parylene
layer was then deposited over the photoresist. Oxygen plasma etching created openings for
microfluidic ports as well as the device cutout. Devices were released from the substrate
and sacrificial photoresist was removed with an acetone soak. No stiction of Parylene
structures was observed. Some fabricated sham devices were coated with Pt and imaged
with scanning electron microscopy (SEM) as shown in Figure 3-13.

Figure 3-12. (a) Cross sectional view of sham devices process flow, and (b) fabricated sham
device with photoresist in microfluidic channels.

70

Figure 3-13. SEM images of fabricated sham devices showing different side views of the
microfluidic channels.
3.3.2 Complete devices
Devices were fabricated on a bare silicon wafer for mechanical support during the
microfabrication process (Figure 3-14, Figure 3-15, Appendix C, Appendix D). A 10 m
Parylene base layer was first deposited by chemical vapor deposition. Sputtered or e-beam
deposited Pt electrodes and leads (2000 Å) were then patterned by a liftoff process with
negative photoresist AZ 5214 E-IR. A 10 m Parylene insulation layer was then deposited
and selectively plasma-etched to expose electrodes and contact pads. Sacrificial photoresist
AZ 4620 was spun to a height of 20 m and patterned to form the microfluidic channels.
An 8 m Parylene layer was then deposited over the photoresist. Oxygen plasma etching
created openings for microfluidic ports as well as the device cutout. Devices were released
from the substrate (Figure 3-16) and sacrificial photoresist was removed with an acetone
soak. Again no stiction of Parylene structures was observed.

71

Figure 3-14. Kic software masks. (a) Complete LACE device, (b) locking mechanism,
microfluidic channels with ports, and metal traces, and (c) final LACE devices mask layout
(100 mm wafer).

Figure 3-15. LACE fabrication process which utilizes standard surface micromachining
techniques for Parylene.

72

Figure 3-16. Fabricated LACE in which microfluidic channels are highlighted by the
presence of photoresist.
3.3.2.1 Metal deposition
Thin film Pt can be deposited using two physical vapor deposition methods: sputter
and evaporation (e-beam). Sputter metal deposition was used initially due to its superior
step coverage and excellent adhesion [60]. Sputtering was performed by LGA films, to a
final thickness of 200 nm (Figure 3-17). The deposited film was uniform with strong
adhesion to the Parylene substrate, however, the lift-off process was overly arduous (~ 3
hrs in 40°C acetone bath) due to damage to the lift-off photoresist during sputter deposition.
Device yield per wafer was low, as a result of mechanical damage to the soft Parylene from
brushing away unwanted metal during lift-off (Figure 3-18). Device curling upon release
of the devices from the silicon substrate was more apparent, and might be attributed to
increased internal material stress of sputtered devices. We made several changes to the AZ
5214 photoresist recipe resulting in an optimized process for sputtered Pt on Parylene
which reduced lift-off process time, and also decreased damage to the wafers.

73

Figure 3-17. Sputtered Pt features obtained from LGA process. Damaged lift-off
photoresist was identified as the wrinkled areas around the device metal features (right
image).

Figure 3-18. Devices following metal lift-off of sputtered metal.

E-beam driven physical vapor deposition results in higher quality films and
provides a higher deposition rate [60]. In an initial attempt, we observed stress-induced
cracking of deposited Pt likely due to excess heat (> 100 ºC) generated during the process
(Figure 3-19). Temperature was measured using temperature monitor stickers on the back
of the wafer (Omegalabel TM). Despite several attempts to fix or circumvent the issue with
our tool (Temescal BJD-1800), none were successful. Our theory is that a machine with a
wafer cooling system or a long throw distance (distance between metal source and
substrate) will help dissipate heat and prevent metal cracking during deposition.
Subsequent e-beam deposition was performed in a tool (CHA Industries MARK 40) with
a throw distance of 22 inches, 8 inches longer than the e-beam tool used for our first process
allowing for greater heat dissipation. Heat was measured with temperature monitor stickers
as previously described and was maintained below 77 ºC. No stress-induced cracking of
our Pt occurred as shown in Figure 3-20. The machine can hold up to 22 wafers, has a

74
turnaround time of 1 day, and coating is at a lower cost. The metal lift-off process was fast
(< 4 min in acetone) and no mechanical damage was induced to the soft Parylene resulting
in 100% device yield per wafer (Figure 3-21).

Figure 3-19. Cracked Pt features following e-beam deposition.

Figure 3-20. Successful e-beam driven physical vapor deposition.

75

Figure 3-21. Devices following metal lift-off of e-beam metal. (a) “UP” logo, (b) electrodes
and traces, (c) locking teeth and etch stop features, and (d) alignment crosses.
3.3.2.2 Parylene film anomalies
Parylene that was milky in appearance was observed when depositing the insulation
layer. Milky Parylene can result from a faster deposition rate (higher pressure) during the
chemical vapor deposition (CVD) of Parylene C as described by the tool manufacturer
(SCS) in a conversation. We identified a leak in our vacuum system and repaired it.
Unfortunately, milky Parylene continued to appear in subsequent runs. Improved metal
lift-off resulted in less appearance of milky Parylene. Magnified photographs reveal the
presence of small clusters in film areas that were intensely swabbed during metal lift-off
(Figure 3-22a-b). Fortunately, the areas affected were minimal and the overall quality of
the Parylene resulting film was acceptable.

76

Figure 3-22. Deposited Parylene insulation layer milky in appearance. (a-b) Magnified
photographs reveal the presence of small clusters in polymer areas that were intensely
swabbed during metal lift off. (c) Absence of milky Parylene in areas of the polymer that
were mildly swabbed.
3.3.2.3 Plasma etching of Parylene
Following the final Parylene deposition to define the microfluidic channels,
openings must be etched into the Parylene layers to expose the surface electrode, contact
pads, device cutout, and the fluidic outlet ports. Etching of thick Parylene layers (~ 10 µm)
can be performed with deep reactive ion etching (DRIE) in an oxygen plasma, using a
photoresist mask to define the pattern. Initially, plasma parameters of 20 W radio frequency
(RF) and 700 W inductive couple plasma (ICP) power were used during both etch and
passivation steps. Unfortunately, this resulted in very low selectivity of Parylene to
photoresist, leading to partially etched fluidics. Low selectivity can be attributed to DRIE
machine malfunction which required a recalibration of our etching recipe. During DRIE
we observed significant out-gassing of residual solvent in the sacrificial photoresist. This
solvent out-gassing pushed the Parylene film upwards, in some cases rupturing the film,
resulting in damaged fluidic channels (Figure 3-23). This residual solvent was likely a
result of incomplete baking during lithography.

77

Figure 3-23. Parylene bubbles near fluidic structures and low selectivity of Parylene to
photoresist etch rate during DRIE process resulted in damage to devices.
An etch selectivity of 1:1 Parylene to photoresist, and an etch rate of 0.147 µm per
loop was achieved with an 80 W RF and 900 W ICP power during etch and passivation
steps. With these parameters, a 30 µm thick photoresist etch mask was sufficient to protect
the microfluidic channels (20 µm photoresist and 8 µm Parylene) during DRIE. This etch
rate is 1.7x faster than our previous etch recipe, resulting in faster processing. However,
the photoresist etch mask was difficult to remove and the sacrificial photoresist defining
the fluidic channels was not removable after this processing step. Additionally, significant
out-gassing of remaining solvent in the sacrificial photoresist occurred during the high
power DRIE process and again resulted in the appearance of small gas bubbles in the resist
(Figure 3-24). A temperature of 99 ºC was reached during this process, which explains the
significant out-gassing in the sacrificial photoresist. There was also a concern of thermally
induced stress as a result of prolonged heating. The fabrication process was modified by
decreasing the plasma parameters to 20 W RF and 700 W ICP power. This recipe results
in an etch rate of 0.09 µm per loop. The etch mask was easily stripped with a 5 min acetone
soak, followed by 5 min IPA soak, and a rinse with DI water. The sacrificial photoresist
lithography recipe was modified by increasing the soft bake temperature to remove any
remaining solvent.

78

Figure 3-24. Out-gassing of sacrificial photoresist solvent during oxygen plasma etching
in the DRIE. Device from a) wafer 1 and (b) wafer 2.
3.3.3 Sacrificial photoresist release
Initially, the sacrificial layer of photoresist which defined the microfluidic channel
interior was removed with a 30 hour soak in room temperature acetone. Catheter insertion
tests into the microfluidic channel inlet showed increased delamination of Parylene layers
due to lengthy acetone soaks; see section 3.4.2.2. Several parameters were evaluated to
accelerate dissolution of the sacrificial layer such as: temperature, solvent choice, solution
volume, and stirring (Table 3-2). We discovered that the combination of increased acetone
volume, heating (40 °C) and agitation (170 rpm magnetic stir-bar) significantly reduced
soak time, mitigating risk of Parylene delamination and reducing production time. Several
release fixtures were tested and are presented in Appendix E.
Table 3-2. Microfluidic channel sacrificial photoresist removal method and dissolution
times.
Device Photoresist release method Total soak time
(HH:mm)
A Warm acetone (40 °C) 47:50
B Warm acetone (40 °C) 45:15
C Warm acetone (40 °C) & Remover PG 69:05
D Warm acetone (40 °C) - Stirred 29:40

The final fixture was designed and fabricated to accommodate multiple devices
while submersed in a heated beaker of agitated acetone. The fixture involved hanging each

79
LACE device from a wire by inserting the wire through the hole in the tab and attaching
the other end to a scaffold suspended above the solvent (Figure 3-25a). Also shown in
Figure 3-25b-c, this method avoids any abrasion during the soak and also allows easy
access to each device to check dissolution progress. Potential for damaging the devices
during handling is minimized since the wire provides a convenient handle and removes the
need to grasp the device with tweezers. Furthermore, this method does not require
additional components in the bath (e.g. polypropylene sheet) to separate the stir bar and the
devices, thereby maximizing the agitating action of the solvent in dissolving the
photoresist. With this fixture, we have achieved a consistent and satisfactory soak time of
approximately 26 hours (n=13, 40 ºC, 170 rpm) and can release more than 20 devices at a
time with a microfluidic channel length of 26.3 mm per device. Once the sacrificial
photoresist is fully removed, the devices are cleaned with a 30 min IPA soak at room
temperature, followed by 1 hr water soak under vacuum at room temperature, and finally
air-dried in an oven at 80 ºC for 1 hr.

Figure 3-25. (a) Sacrificial photoresist release fixture involved hanging each device from
a wire attached to a scaffold above the acetone. Before (b) and after (c) photos of a device
soaked via the wire fixture show no significant damage.
3.3.4 Residual oxygen plasma exposed photoresist
Residual AZ 4620 photoresist from the last etch mask was observed on the surface
electrode and contact pads after sacrificial photoresist release once the devices were

80
inspected under the microscope (Figure 3-26). Photoresist subjected to prolonged oxygen
plasma can become crosslinked and difficult to strip requiring a strong stripper chemistry
for full removal. Several solutions were evaluated to remove this oxygen plasma exposed
photoresist. A brief mention of the solutions and methods are presented in Table 3-3 and
detailed information can be found in Appendix F. We decided to make a version of a
popular commercial stripper used to remove oxygen plasma exposed photoresist in order
to improve photoresist removal at lower temperatures. The solution consisted of 1 part
Remover PG (N-methyl-2-pyrrolidone based) and 1 part of AZ 726 (< 3%
tetramethylammonium hydroxide based developer). Photoresist residue was successfully
removed after a 15 min soak at 50 °C for 15 min as shown in Figure 3-26. Stripper solution
was cleaned with a 30 min isopropyl alcohol (IPA) soak followed by a 1 hour water soak.
Table 3-3. Oxygen plasma exposed photoresist removal methods.
Solution Chemical composition Soak
temperature
(°)
Total soak
time
(mm)
Result
RCA-1 5:1:1 H 2O:NH 4OH:H 2O 2 70 15 Failed
Piranha 4:1 H 2SO 4:H 2O 2 40 3 Successful*
Remover PG NMP 60 60 Failed
AZ KWIK Stripper • 60%:35% DPGME:C 4H 8O 2S 60 60 Failed
Homemade stripper 97:3 NMP:TMAH 50 15 Successful

Figure 3-26. Oxygen plasma treated AZ 4620 photoresist residue on LACE metal features.
*Solution too aggressive and resulted in oxidized Parylene after thermal treatment

81

Figure 3-27. Oxygen plasma exposed AZ 4620 photoresist cleaned with homemade
stripper. Photographs show no photoresist residue.

3.3.5 Post-process annealing treatment
Multi-layered Parylene devices tend to show interface delamination under
prolonged soak conditions. Delamination results from weakening of layer interfaces such
as Parylene-Parylene or Parylene-metal by liquid penetration. Post-process annealing
treatment can increase electrode’s in vivo lifetime by improving adhesion between Parylene
layers and decreasing permeability to water vapor. Annealing process takes place at 200
°C (temperature above Parylene’s glass transition of ~90 °C [61]) and results in increased
entanglement between polymer chains of each Parylene layer and polymer crystallinity
[62]. Devices were annealed under vacuum to prevent Parylene oxidative degradation.
Oxygen content in the oven can be further minimized by nitrogen purging. The LACE
devices were annealed at 200°C for 48 hours followed by a slow cooling step (>12 hrs)
also under vacuum [63]. Devices were annealed on a jig consisting of Teflon®
(polytetrafluoroethylene) blocks and sheets (part no. 8735K59 and 8545K201; McMaster-
Carr Supply Co., Santa Fe Springs, CA) to prevent device movement or adhesion to the
oven surface (Figure 3-28). No materials were placed over the fluidic channel regions to
prevent the channels from collapsing.

82

Figure 3-28. Side and top view photographs of Teflon® PTFE jig used to anneal LACE
devices.

Fourier transform-infrared spectroscopy (FTIR) was performed on both untreated
and annealed Parylene samples. The results indicated no effects to the chemical
composition of Parylene following 30 hrs acetone soak for sacrificial photoresist removal
and the post-process annealing process as shown in Figure 3-29.

Figure 3-29. Absorbance measurements from Fourier transform infrared spectroscopy
(FTIR) analysis of Parylene untreated, and acetone soak and thermoformed under vacuum.

83
Packaging
3.4.1 Electrical packaging
3.4.1.1 Bench top and acute animal studies
Packaging was designed to withstand surgical manipulation during placement of
LACE and acute electrical recording and stimulation (1 day). Electrical connection directly
to the Parylene cuff (device level) consisted of a 10 channel zero insertion force (ZIF, part
no. 0514411072 Molex Inc.) connector. The ZIF connector connected with a commercially
available 10 channel flat flexible cable (FFC, part no. 050R10-102B, Parlex USA LLC)
and the connection can be coated with biocompatible epoxy to smooth the corners and
avoid potential muscle irritation. The FFC was connected to a ZIF connector mounted on
a printed circuit board (PCB). This PCB included a 10-contact Preci-Dip female socket
(part no. 801-87-010-10-012101) for connections to a Plexon electrophysiology recording
setup that terminates in 18 channel Omnetics with an Omnetics/Preci-Dip adaptor. Figure
3-30 shows the overall system components that have been identified for the acute studies.

Figure 3-30. Cuff electrode with integrated Parylene cable is attached to a zero insertion
force (ZIF) connector.
3.4.1.2 Chronic animal studies
Packaging was designed to withstand extensive surgical manipulation during
placement of LACE and chronic implantation (2 weeks). Electrical connection directly to

84
the Parylene cuff (device level) consisted of a 10 channel zero insertion force (ZIF)
connector. The ZIF connector was mounted on a flexible printed circuit board (FlexPCB),
to facilitate soldering of a ten-conductor Cooner cable (pat no. CW8227, Cooner Wire Co.,
Chatsworth, CA). The connection can be coated with biocompatible epoxy to smooth the
corners and avoid potential muscle irritation. The FlexPCB included several 1 mm ID holes
for suturing to the muscle and prevent connection damages during animal movement. The
Cooner cable consisted of 5 twisted pairs of 10 copper conductors insulated by a silicone
rubber jacket. The Cooner cable terminated at the second (headmounted) FlexPCB, which
mated the cable to a 10-contact Preci-Dip female socket connector as shown in Figure 3-31.
This connection was designed to be routed in a 10 mm OD tube to the head for skull
mounting of a percutaneous connector as it is less prone to dislodging than skin mounting
of percutaneous connectors (Figure 3-32). A Preci-Dip connector was selected as it is more
robust than 10-contact Omnetics for multiple connects and disconnects. The Preci-Dip
connector was designed to connect to a Plexon electrophysiology recording setup that
terminates in 18- channel Omnetics with an Omnetics/Preci-Dip adaptor.

Figure 3-31. Electrical packaging for chronic in vivo animal studies.

85

Figure 3-32. Percutaneous connector inside surgical tunnel.
3.4.2 Fluidic packaging
3.4.2.1 Bench top studies
Device level fluidic connection was established by way of a section of small (100
µm ID; 360 µm) tubing made from polyether ether ketone (PEEK:1571, IDEX Health &
Science, Oak harbor, WA). First, a stainless steel microwire (200 µm OD, Cooner Wire
Co., Chatsworth, CA) was inserted into the inlet port of the sham device to widen the
channel opening. Once the inlet channel was propped open, the PEEK tubing was inserted,
and fixed in place with silicone adhesive (MED -4210, Factor II). The silicone was cured
for 1 hour at 60 °C. The device level fluidic connection process is show in Figure 3-33.
The fluidic connection from the infusion pump to the LACE device required a 25 µL glass
syringe (Hamilton®), 22 G coring needle, and polyurethane catheter (305 µm ID), which
attached to the LACE microfluidics by way of the PEEK insert.

Figure 3-33. Fabrication of fluidic connection at the device level. (a) Microwire insertion
into fluidic channel inlet, followed by (b) PEEK insertion and coating with Silicone MED-
4210.
3.4.2.2 Acute animal studies
Packaging was designed to withstand surgical manipulation during placement of
LACE and acute fluid infusions (1 day). A custom-made polyurethane (PU) catheter (SAI

86
Infusion Technologies) was designed to connect the device fluidic inlet directly to syringe
pump (Ultramicropump UMP3 with SYS-Micro4 controller) (Figure 3-34). During initial
attempts to insert the PU catheter to the LACE microfluidic channel inlet, we encountered
delamination between the Parylene layers of the device. Additionally, the friction between
the PU and Parylene made insertion difficult. Three parallel strategies were pursued to
mitigate delamination: thermoforming the LACE microfluidic channel inlet, reduction of
acetone soak time for releasing of the sacrificial photoresist, and increasing the width of
the fluidic inlet channel to facilitate alignment and insertion. A steel microwire (200 µm
OD) was inserted into the fluidic inlet channel of the device to widen the inlet and maintain
it open during annealing. The devices were then annealed as previously described.
Annealing resulted in simple and successful insertion of the PU catheter into the
thermoformed microfluidic channel inlet (Figure 3-35).

Figure 3-34. SAI custom-made fluidic catheter.

87

Figure 3-35. PU catheter inserted into thermoformed LACE device microfluidic channel
inlet.

The PU catheter is secured in place to the device fluidic channel with a flexible
adhesive to provide mechanical support and prevent leaks. Several medical grade adhesives
were tested and the results are shown in Table 3-4. Loctite 4902 is sufficiently flexible and
with the help of activators (Loctite 7701 Primer, Loctite 713 Accelerator, Loctite,
Westlake, OH) applied to the PU catheter and inside of the channel, cures in approximately
5 seconds with good bond strength. While it does have a relatively low viscosity that allows
it to wick into the channel quickly, the fast cure time prevents it from blocking the channel
as shown in Figure 3-37. MED-4210 silicone (Factor II, Inc., Lakeside, AZ) has superior
flexibility, but requires partial pre-curing to prevent blocking of the channel. The preferred
method for securing the catheter was to affix and seal the catheter into the channel with a
small amount of Loctite 4902, then pot the entire connection with MED-4210 silicone for
reinforcement.

88
Table 3-4. Medical grade adhesives tested for securing LACE fluidic connection.
Solution Characteristics Failure mode
Epo-Tek 301 Low viscosity and
long cure time
Blocked fluidic inlet channel
Sylgard 184 silicone Low viscosity and
long cure time
Blocked fluidic inlet channel (Figure 3-36)
Epo-Tek 353 ND-T High viscosity Breaks away from Parylene due to brittleness
Loctite 5240 • Low viscosity and
UV curable
Parylene degradation [64-66]
MED 4210 silicone Flexible Weak adhesion to Parylene
Loctite 4902 Low viscosity with
fast cure time
Accelerator solution for fast cure time results
in mild Parylene delamination

Figure 3-36. Sylgard 184 silicone wicked into the channel, blocking flow completely.
When dye was flowed into the device, the blockage caused it to leak out of the inlet
opening.

Figure 3-37. The fast cure of Loctite 4902 prevents it from blocking the channel.

The final process of gluing the PU catheter in place with Loctite 4902 begins with
inserting a small section of PEEK tubing (360 µm OD) into the inlet. With the PEEK tubing
in place, a small drop of Loctite 713 Accelerator is applied to the inlet opening and quickly

89
covers the interior surface of the inlet. A small amount of Loctite 7701 Primer is brushed
on to the end of the PU catheter and allowed to dry. Once the accelerator inside the inlet is
nearly dry the PEEK tubing is removed and the PU catheter is inserted. The PU catheter
stylet assists with insertion and is removed once the catheter is in place. Next, a small drop
of Loctite 4902 is applied at the inlet opening, wicking into the inlet but curing in several
seconds. Additional drops are added successively as needed to seal the inlet. Any excess
glue can be cured by adding a drop of accelerator. MED-4210 silicone is then added
generously to the outside of the inlet and the catheter-LACE connection and cured at 60 ºC
for one hour. The schematic representation of the LACE acute implantation and
experimental setup is shown in Figure 3-38.

Figure 3-38. The schematic representation of the LACE acute implantation and
experimental setup.
3.4.2.3 Chronic animal studies
Chronic fluidic packaging was designed to withstand extensive surgical
manipulation during placement of LACE and chronic drug delivery (2 weeks). A custom-
made PU interfaced with the microfluidic channel in the LACE device. The PU catheter
was secured to the LACE fluidic channel with Loctile 4902 and then the connection was

90
potted with a silicone elastomer as previously described. The catheter was designed to be
surgically tunneled under the skin and exit at the back of the animal to connect to a port
housed in a vascular access harness (part no. VAH95AB, Instech, Plymouth Meeting,
PA). The port's septum is pierced by a connector attached to the spring tether to make the
fluid connection. A syringe pump will be used to infuse directly into the harness as needed.
The tether connector can be seal off when the animal is disconnected. This infusion method
allows quick and aseptic connections of the catheterized rat and infusion tether. The
schematic representation of the LACE chronic implantation and experimental setup is
shown in Figure 3-39.

Figure 3-39 Schematic representation of the LACE chronic implantation and experimental
setup.

Experimental Methods
3.5.1 Interlocking mechanism
Initial proof of concept of an earlier version of the locking mechanism shown in
Figure 3-40a was demonstrated using a scaled up design in 180 µm thick paper (Figure

91
3-40b). The paper was precisely cut using a vinyl cutter (Graphtec cutting plotter CE6000-
40). A 12 µm thick Parylene sheet was similarly cut with a vinyl cutter (Figure 3-40c). In
both mock ups, the locking mechanisms were simple to use and provided adequate holding
strength under gentle handling conditions.

Figure 3-40. Proof of concept testing of adjustable interlocking mechanisms. (a) Buckle
serrated mechanism, (b) scaled 180 µm thick paper cuff, and (c) 12 µm thick Parylene cuff.

The interlocking mechanism was tested with prototype and complete LACE
devices (Figure 3-41 and Figure 3-42). Devices were wrapped and unwrapped around
simulated agarose nerve phantoms and holding strength was evaluated under gentle
handling conditions (n > 10). The agarose nerve was cored out of an agarose block using a
16 ga syringe tip with inner diameter of 0.053” (1.3462 mm), which falls in the sciatic
nerve diameter range of 1.2 – 1.4 mm. We used an agarose concentration of 1.37 % wt/wt
in water, which was selected to produce agarose with a compressive modulus equal to that
of rabbit sciatic nerve [67, 68].

92

Figure 3-41. Complete LACE interlocking mechanism. (a) Suture needle passed through
the needle hole, (b-c) tab threaded through the first slit and kept in place by serrated teeth,
and (d) tab secured by threading through second slit to form the buckle locked structure.

Figure 3-42. Photograph of the LACE interlocking mechanism serrated teeth.

93
3.5.2 Microfluidic channel evaluation
3.5.2.1 Fluidic connection
The fluidic inlet channel connection intended for in vivo animal studies was
evaluated by driving colored dye through the channels using an infusion pump (Harvard
apparatus Infuse/Withdraw PHD 2000) and imaging the progression of the dye with a
microscope (Caltex Scientific HD60T). The LACE device was submerged in distilled
water under the microscope, and primed by flushing the channels with isopropyl alcohol
and distilled water for 5 minutes each. A schematic of the testing setup is depicted in Figure
3-43. The fluidic connection from the infusion pump to the LACE device required a 25
µL glass syringe (Hamilton®), 22 ga coring needle, and the custom-made PU catheter (254
µm OD), which attached to the LACE microfluidics. Flow rates were tested starting at 800
nL/min and increased by 200 nL/min with 1 minute of test duration at each rate. Any leaks
or Parylene layers delamination around the fluidic connection were considered as failure.

Figure 3-43. Microfluidics evaluation experimental setup.
3.5.2.2 Uniform fluidic channel flow
Evaluation of the microfluidic channels was conducted with the previously
mentioned setup. With the LACE device in a flat or curled orientation, a solution of dilute
green dye was infused into the LACE device at varied flow rates ranging from 5 to 2000
nL/min. The flow rate range was selected based on the typical epineural space drug delivery

94
rate of 17 nL/min that will be employed in our studies [69]. Evaluation of the microfluidic
channel in the curled configuration was conducted by standing prototype devices upright
into a custom jig while imaging the transverse progression of the dye with a microscope as
shown in Figure 3-44. The fluidic connection from the infusion pump to the LACE device
required a custom-made PU catheter (584 µm OD), which attached to the LACE
microfluidics by way of the PEEK insert.

Figure 3-44. Standing curled prototype LACE upright into a custom acrylic jig to image
the transverse progression of the dye with a microscope.
3.5.2.3 Localized drug delivery
Localization of fluid delivery was evaluated by locking a LACE device around a
simulated sciatic nerve phantom produced from molded 1.37% agarose, then driving
sodium fluorescein dye (Sigma F6377) through the microfluidic channels using the
infusion method previously described at a flow rate of 50 nL/min. The degree of
localization of the infused fluid will be quantified by the location and dimensions of the
dyed area. The agarose nerve was imaged using a fluorescent microscope (Nikon LV100;
mercury 100 W lamp, Chiu Technical Corporation; B-2A filter cube) and digitally
processed in ImageJ (NIH) in order to correlate fluorescent signal with dye concentration.
3.5.2.4 Maximum flow rate
Maximum possible flow rate of the LACE device was determined by pumping dye
solution through a LACE device wrapped around an agarose nerve phantom (1.37 % wt/wt
in water) as previously described and observing which flow rates induced leaks or
structural failure. Flow rates were increased from 300 nL/min at increments of 100

95
nL/minute. Between each increment, the device was uncurled and infused with dye in a flat
orientation under the microscope in order to detect any small leaks that may have formed
in the interior of the channels, then re-curled for the next flow rate.
3.5.3 Electrode evaluation
Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) were
used to assess the stimulation and recording capabilities respectively of the cuff electrodes.
The electrochemical characteristics of LACE electrodes were measured in devices without
microfluidic channels (surface electrode), with microfluidic channels, and with the LACE
device curled as if encircling a nerve. Electrochemical characterization of the electrodes
was performed with a Gamry Reference 600 potentiostat (Gamry Instruments, Warminster,
PA). Prior to EIS measurement, CV was performed in 0.05 M H2SO4 to electrochemically
clean the electrode surface. The stimulation characteristics of the electrodes were measured
with CV in a solution of phosphate buffer saline (1× PBS). LACE devices were immersed
in solution with a constant stream of N2 in a three-electrode cell. The working electrode
was cycled between −0.2 to 1.2 V for 0.05 M H2SO4 or −0.6 to 0.8 V for 1× PBS with
respect to an Ag/AgCl (3M NaCl) reference, a potential range that is within the water
window [70, 71]. A 1 cm
2
Pt plate served as a counter electrode. Using a scan rate of 250
mV/s, each electrode was cycled for 30 cycles. EIS was performed in 1× PBS at room
temperature with an AC perturbation signal of 25 mV (rms) in the frequency range of 1-
10
5
Hz. As with CV, an Ag/AgCl (3M NaCl) reference and 1 cm
2
Pt plate counter were
used.
Results
3.6.1 Interlocking mechanism
The locking mechanism was simple to implement and robust. A suture needle
facilitated wrapping the device around an agarose nerve phantom of 1.2 mm diameter. The
LACE was able to wrap tightly around the phantom without sliding of the device or damage
to the simulated nerve. Once locked in place, the LACE did not disengage during fluidic

96
and electrical testing, evidence of a secure lock. No damage to the cuff was observed
despite repeatedly engaging and releasing the lock and curling and uncurling the LACE
around the nerve phantom.
3.6.2 Microfluidic channel evaluation
3.6.2.1 Fluidic connection
The fluidic connection was successfully secured with Loctite 4902 and MED 4210
silicone. Two devices handled flow rates up to 9000 nL/min with no signs of failure. Three
other devices with known fabrication issues failed at lower rates (~2500 nL/min), but not
at the fluidic connection.
3.6.2.2 Uniform fluidic channel flow
No leakage or delamination was observed up to a flow rate of 2000 nL/min, with
uniform flow through all fluidic channels and outlet ports in the flat configuration (n = 4).
Figure 3-45a displays a sequence of images showing simultaneous and equally distributed
flow out of all four outlet ports in the flat orientation. Image analysis confirmed that flow
rate through channels was uniform within 6% standard error. In the curled configuration,
flow was achieved in 3 out of 4 channels which is sufficient for the in vivo application
(Figure 3-45b). The sequence of images shows simultaneous and equally distributed flow
out of the 3 outlet ports. Flow through the distal channel (channel 4) was blocked by
pinching of the horizontal feeder channel. Uniform flow through all fluidic channels and
outlet ports was observed after uncurling the device. Attempts to mitigate the pinching
were made by thermoforming flat devices both with and without photoresist. Unfortunately
following thermoforming, photoresist proved impervious to solvent, and could not be
removed despite lengthy soaks in heated acetone and other solutions. Devices without
photoresist were curled during thermoforming and evaluated, but pinching still occurred
preventing flow through the distal channel (Figure 3-46).

97

Figure 3-45. Sequential photographs of the infusion experiment at 830 nL/min flow rate in
a (a) flat device and (b) curled device. Image analysis using ImageJ software confirmed
that flow rate through channels was uniform within 6% standard error.

Figure 3-46. Microfluidic channel functionality test in thermoformed sham devices in the
curled orientation. (a) Sham device indicating fluidic channel number. (b) Pinching
occurred along the horizontal channel feeder blocking flow to channel 4, and (c) uniform
flow achieved through channels 1 to 3.

3.6.2.3 Localized drug delivery
Localized drug delivery was achieved with the LACE as shown in Figure 3-47.
Localized fluorescein dye delivery with minimal longitudinal diffusion on a nerve phantom
was achieved. Unfortunately, the fluorescein dye completely diffused through the entire
agarose nerve phantom making it difficult to observe the dye localization. Additionally,
the fluorescent microscope can only image sections of the nerve phantom preventing

98
imaging and analyzing of the dye diffusion in the entire phantom. Consequently, the
fluorescent signal could not be correlated with dye concentration.

Figure 3-47. 1.37% (w/w) agarose nerve phantom after localization testing with fluorescein
dye.
3.6.2.4 Maximum flow rate
During maximum flow rate testing, microfluidic channel inlet failure occurred at a
rate of 1000 nL/min. The leak appeared to be from delamination between Parylene layers
rather than bursting through a Parylene channel (Figure 3-48). A safe flow rate of 900
nL/min can be achieved by the LACE device.

Figure 3-48. Bursting occurred at a maximum flow rate of 1 μL/min. This leak occurred
near the edge of the inlet, signifying it was likely due to delamination of Parylene layers.

99
3.6.3 Electrode evaluation
The expected standard voltammogram for Pt immersed in sulfuric acid and saline
solution was achieved for all electrodes. Distinct peaks of current in the cyclic
voltammogram result from the process of hydrogen atoms adsorption to and desorption
from the electrode surface were observed as shown in Figure 3-49. These peaks can be
integrated with respect to the scan rate to calculate the electroactive surface area (ESA). In
equation (1), QH is the hydrogen desorption charged measured via CV and 210 µC/cm
2
is
the value for the characteristic charge density associated with the monolayer of hydrogen
atoms adsorbed to polycrystalline Pt [72]. Stimulation electrodes can be characterized by
their cathodal charge storage capacity (CSC), which is calculated from the time integral of
the cathodic current in a cyclic voltammogram per unit geometric surface area (GSA) [71]
as shown in equation (2).
𝐸𝑆𝐴 =
𝑄 𝐻 210 𝜇𝐶
𝑐𝑚
2
(1)
𝐶𝑆𝐶 =
𝑄 𝑐𝑎𝑡 ℎ𝑜𝑑𝑖𝑐 𝐺𝑆𝐴
(2)
The cyclic voltammogram obtained for all electrodes tested indicated a clean
electrode surface (Figure 3-49). The ESA calculated was 688,810 ± 67,116 µm
2
(n=4),
which is higher that the GSA of 475,396 µm
2
due to surface topology. The cathodal CSC
in 0.05 M H2SO4 and 1× PBS were not significantly different as confirmed by statistical
analysis (ANOVA, p > 0.05) with a calculated value of 1.64 ± 0.4 mC/ cm
2
and 1.09 ± 0.1
mC/ cm
2
, respectively. The roughness factor, defined as the ratio of ESA to GSA was 1.4
(n = 4). No bubble generation was observed during the CV, which is important for safe
electrical stimulation.

100

Figure 3-49. Cyclic voltammetry in H2SO4 of single recording site yielded standard Pt
electrode characteristics. The integration of the hydrogen desorption peaks to find QH and
the integration of the cathodic current (Qcathodic) for ESA and CSC are highlighted,
respectively.
Impedance spectroscopy was used to investigate electrode properties. Electrical
impedance is commonly recorded at 1 kHz since action potential signals are in the
frequencies of 100 Hz to 10 kHz [3, 73]. EIS measurements on cuff surface electrodes
yielded an electrical impedance of 1.8 ± 0.11 kΩ and a phase of -56.4 ± 1.0º at 1 kHz
demonstrating low variation between electrodes. EC measurements on surface electrodes
helped establish a baseline impedance values (n = 6). A representative EIS measurement
for a single electrode site post CV-clean is shown in Figure 3-50. EIS performed on a
LACE device in flat, curled, and uncurled configuration showed minimal variation in
impedance magnitude and phase at 1 kHz indicating curling does not affect electrode
performance (Table 3-5). As expected, electrodes that were embedded in the channels
exhibited different CV curves and higher impedance values, which are attributed to the
limited ion diffusion imposed by the surrounding fluidic channels (Figure 3-51, Figure
3-52, Figure 3-53, and Table 3-6).

101

Figure 3-50. EIS in PBS of single recording site showed a low impedance (< 2 kΩ) at 1
kHz).
Table 3-5. Electrochemical impedance spectroscopy in 1x PBS at 1 kHz (mean ± SE) for
a LACE device in flat, curled, and uncurled configuration.
Device configuration n Impedance magnitude
(k Ω)
Impedance phase
(°)
Flat 3 1.61 ± 0.1 -54.70 ± 1.4
Curled 3 1.99 ± 0.1 -49.60 ± 1.0
Uncurled 3 1.70 ± 0.1 -53.93 ± 1.4

Figure 3-51. Solidworks diagram of LACE with fluidic channels.

102
Table 3-6. Electrochemical impedance spectroscopy in 1x PBS at 1 kHz (mean ± SE).
Partially embedded electrodes are located underneath the fluidic outlet port (reprinted with
permission from [74] © 2017 IEEE).
Electrode n Impedance magnitude
(k Ω)
Impedance phase
(°)
No fluidics 6 1.80 ± 0.11 -56.4 ± 1.0º
Partially embedded 8 14.27 ± 1.38 -35.16 ± 3.81
Fully embedded 8 44.94 ± 8.50 -18.10 ± 3.08

Figure 3-52. Representative cyclic voltammogram of LACE Pt electrodes in sulfuric acid
following priming. (a) surface electrode with no fluidics, (b) partially embedded electrode
(fluidic outlet port), and (c) fully embedded electrode.

Figure 3-53. Representative electrochemical impedance spectroscopy plots of impedance
magnitude and phase in 1×PBS for a (a) surface electrode with no fluidics, (b) partially
embedded electrode (fluidic outlet port), and (c) fully embedded electrode. EC
measurements were taken with a Gamry Reference 600 potentiostat (reprinted with
permission from [74] © 2017 IEEE).
Discussion
A fabrication process was developed in which Parylene microfluidic channels were
produced with embedded Pt recording and stimulating electrodes using surface

103
micromachining processes (Appendix C). Several challenges were overcome during the
fabrication and packaging processes to successfully develop a PN cuff electrode. Pt
electrode sites, traces, and contact pads were sputtered or e-beam deposited on Parylene.
Sputtered Pt films were uniform with strong adhesion to Parylene; however, lift-off
photoresist was damaged during deposition resulting in an arduous lift-off process and poor
device yield. E-beam deposition of Pt yielded high quality films with good adhesion. The
lift-off process was fast with 100% device yield. Parylene milky in appearance was
observed after CVD with a fast deposition rate or on areas of the wafer with mechanically
induced damage (e.g. sputter metal lift-off process). This type of anomaly did not seem to
affect film quality or device performance. DRIE in an oxygen plasma was chosen to expose
the recording sites, contact pads, and define the device outline. DRIE etching of Parylene
resulted in smooth and vertical walls. RF and ICP power values of the etching recipe were
adjusted to improve the selectivity of Parylene to photoresist and decrease process
temperature. Improved selectivity prevented etching of structures protected by the
photoresist mask during etching such as the microfluidic channels. Lower process
temperature minimized photoresist solvent out-gassing and improved removal of the
photoresist etch mask. Photoresist can become crosslinked when subjected to prolonged
oxygen plasma at high temperatures becoming impervious to solvent. Removal of this
photoresist was achieved with the use of a strong stripper solution such as the homemade
stripper (97:3 NMP:TMAH) developed.
It is necessary to release the sacrificial photoresist contained within the microfluidic
channels to prepare the device for use. The sacrificial photoresist removal process is
diffusion driven and can be accelerated by increasing the diffusion rate of acetone through
heating and agitation. Fast sacrificial photoresist release was achieved with agitation in a
warm acetone bath (170 rpm, 40 °C). Devices were annealed at 200°C for 48 hours under
vacuum once the sacrificial photoresist was fully released. The post-process annealing
improved adhesion between Parylene layers and prevented delamination during fluidic
packaging. The fluidic connection between the custom-made PU catheter and the LACE
fluidic inlet was successfully secured with medical grade flexible epoxies. The Loctite

104
4902 glue was sufficiently flexible and formed a robust bond to Parylene allowing for a
strong and secure fluidic connection. Potting of the fluidic connection with MED 4210
silicone provided additional reinforcement and a soft interface for better device-tissue
integration. The fluidic connection proved to be robust and leak-free up to flow rates 500x
higher than the expected flow rate for future in vivo infusion experiments. Electrical
connection to the LACE was established with a zero insertion force (ZIF) connector.
The locking mechanism was simple to implement and provided a secured hold
around the never phantom. No damage was observed during repeated engaging and
disengaging of the cuff which will allow readjustment of the device around the nerve
during implantation to achieve adequate fit. The LACE adjustable locking mechanism will
allow for a secured and tight fit around the nerve while minimizing mechanical damage to
the delicate nerve tissue commonly seem in cuff interfaces surgically secured by suturing
to the epineurium [12, 20]. Additionally, this microfabricated thin-film locking mechanism
avoids the use of bulky handmade external fixation components and the manual dexterity
to assemble them [17].
The LACE successfully achieved uniform flow through all microfluidic channels
up to a flow rate of 2000 nL/min in the flat configuration. The stiction forces between the
top and bottom Parylene layers of the microfluidic channels were strong and induced
channel collapsed in the curled configuration. Attempts to prevent (e.g. thermoforming) or
overcome collapsed channels (e.g. increased fluid pressure) were investigated but proved
ineffective. Clever microfluidic channel designs will be incorporated in future LACE
iterations to maintain channel integrity during deformation and allow for fluid flow through
all the microfluidic channels to ultimately target multiple fascicles inside the nerve.
The LACE can safely handle a flow rate of up to 900 nL/min when curled around
a nerve phantom, which greatly exceeds the 17 nL/min epineural space drug delivery rate
that will be employed in our in vivo studies. Initial fluid delivery on the agarose nerve
phantom demonstrated the potential of the LACE for localized drug delivery with minimal
longitudinal diffusion which is important to induce precise axonal sprouting. The agarose

105
nerve phantom proved difficult to image preventing analysis of the dye diffusion. A
different nerve phantom model is currently being developed to quantify the degree of
localization.
EC techniques were used to assess the recording and stimulation capabilities of the
electrodes. CV on LACE electrodes demonstrated high CSC (> 1 mC/ cm
2
) values suitable
for stimulation of neural tissue. If necessary, the CSC can be further increased by coating
the electrode site with porous materials such as iridium oxide films or Pt black [71, 75].
CV proved to be a useful EC technique for cleaning of the electrode surface, and
determining true ESA and CSC. EIS of LACE electrodes in devices without microfluidic
channels (surface electrode) and with the LACE device curled demonstrated low electrical
impedance (~ 2 kΩ) values at 1 kHz with minimal variation. Recorded impedances are
expected to help reduce susceptibility to noise pickup from sources such as muscle outside
the cuff or recording equipment [17] to provide adequate neural recording potential. High
impedance values of embedded electrodes are attributed to the limited ion diffusion
imposed by the Parylene microfluidic channels. As these electrodes are intended to be used
on axons that have grown into the microchannels, the increased impedance is not expected
to adversely affect recordings. Instead, the close proximity of a specific axon is expected
to improve nerve selectivity and signal fidelity. The choice of electrode material,
dimensions, and design resulted on cuff electrodes with suitable electrode properties for
peripheral nerve recording and stimulation.
Conclusions
A peripheral nerve interface that combines both electrodes and microfluidic
channels in an adjustable cuff sized to interface with rat sciatic nerve was developed. The
LACE device uses a robust and simple locking mechanism that is adjustable for close
contact with nerves of varying diameters. Parylene C was selected as the structural material
for its biocompatibility and improved mechanical matching to tissue. Devices were batch
fabricated using standard Parylene micromachining processes. Convenient and reusable
electrical connections with flexible Parylene ribbon cables were achieved with ZIF

106
connectors and FFCs. Leak-free external microfluidic connection to the LACE fluidic
channels was achieved with a custom-made PU catheter and medical grade adhesives. This
connection proved to withstand flow rates 500x higher than those expected for drug
delivery in our animal studies. Fully functional cuffs having controlled microfluidic
infusion at < µL/min flow rates, and localized drug delivery, which is important to generate
specific drug gradients necessary for guiding the axons into the microchannels, were
demonstrated. Electrochemical characterization of the electrodes demonstrated low
electrode impedances of < 2 kΩ (1 kHz). Electrochemical characterization of the LACE
electrodes demonstrated low impedances < 2 kΩ (1 kHz) and high CSC > 1 mC/ cm
2

suitable for recording and stimulation of neural tissue. High EIS impedance values obtained
for the embedded electrodes are attributed to the limited ion diffusion of the channels. The
close proximity of the axon to the electrode is expected to yield high quality neural
recordings. To prepare for animal testing, the next chapter describes the redesign of fluidic
channels to improve their robustness and allow uniform flow through all channels in the
curled orientation in order to target each fascicle inside the sciatic nerve, long-term
characterization of electrodes in simulated in vivo conditions, and initial animal work.
References
[1] A. E. Schultz and T. A. Kuiken, "Neural interfaces for control of upper limb
prostheses: the state of the art and future possibilities," PM&R, vol. 3, pp. 55-67, 2011.
[2] M. R. Popovic, D. B. Popovic, and T. Keller, "Neuroprostheses for grasping,"
Neurological research, 2013.
[3] E. R. Kandel, J. H. Schwartz, T. M. Jessell, S. A. Siegelbaum, and A. J. Hudspeth,
Principles of neural science vol. 4: McGraw-hill New York, 2000.
[4] D. T. Plachta, N. Espinosa, M. Gierthmuehlen, O. Cota, T. C. Herrera, and T. Stieglitz,
"Detection of baroreceptor activity in rat vagal nerve recording using a multi-channel
cuff-electrode and real-time coherent averaging," in Engineering in Medicine and
Biology Society (EMBC), 2012 Annual International Conference of the IEEE, 2012,
pp. 3416-3419.

107
[5] R. J. Ignelzi and J. K. Nyquist, "Direct effect of electrical stimulation on peripheral
nerve evoked activity: implications in pain relief," Journal of neurosurgery, vol. 45,
pp. 159-165, 1976.
[6] B. N. Van Vliet and N. H. West, "Responses to circulatory pressures, and conduction
velocity, of pulmocutaneous baroreceptors in Bufo marinus," The Journal of
physiology, vol. 388, pp. 41-53, 1987.
[7] C. Krarup, G. E. Loeb, and G. H. Pezeshkpour, "Conduction studies in peripheral cat
nerve using implanted electrodes: II. The effects of prolonged constriction on
regeneration of crushed nerve fibers," Muscle & nerve, vol. 11, pp. 933-944, 1988.
[8] P. K. Stys, B. R. Ransom, and S. G. Waxman, "Compound action potential of nerve
recorded by suction electrode: a theoretical and experimental analysis," Brain
research, vol. 546, pp. 18-32, 1991.
[9] S. Micera and X. Navarro, "Bidirectional interfaces with the peripheral nervous
system," International review of neurobiology, vol. 86, pp. 23-38, 2009.
[10] N. G. Hatsopoulos and J. P. Donoghue, "The science of neural interface systems,"
Annual review of neuroscience, vol. 32, pp. 249-266, 2009.
[11] M. A. Lebedev and M. A. Nicolelis, "Brain–machine interfaces: past, present and
future," TRENDS in Neurosciences, vol. 29, pp. 536-546, 2006.
[12] X. Navarro, T. B. Krueger, N. Lago, S. Micera, T. Stieglitz, and P. Dario, "A critical
review of interfaces with the peripheral nervous system for the control of
neuroprostheses and hybrid bionic systems," Journal of the Peripheral Nervous
System, vol. 10, pp. 229-258, 2005.
[13] T. Stieglitz, M. Schuetter, and K. P. Koch, "Implantable biomedical microsystems for
neural prostheses," Engineering in Medicine and Biology Magazine, IEEE, vol. 24,
pp. 58-65, 2005.
[14] S. Micera, M. C. Carrozza, L. Beccai, F. Vecchi, and P. Dario, "Hybrid bionic systems
for the replacement of hand function," Proceedings of the IEEE, vol. 94, pp. 1752-
1762, 2006.
[15] P. M. Rossini, S. Micera, A. Benvenuto, J. Carpaneto, G. Cavallo, L. Citi, et al.,
"Double nerve intraneural interface implant on a human amputee for robotic hand
control," Clinical neurophysiology, vol. 121, pp. 777-783, 2010.
[16] A. Branner, R. B. Stein, and R. A. Normann, "Selective stimulation of cat sciatic
nerve using an array of varying-length microelectrodes," Journal of neurophysiology,
vol. 85, pp. 1585-1594, 2001.

108
[17] G. Loeb and R. Peck, "Cuff electrodes for chronic stimulation and recording of
peripheral nerve activity," Journal of neuroscience methods, vol. 64, pp. 95-103, 1996.
[18] M. D. Tarler and J. T. Mortimer, "Selective and independent activation of four motor
fascicles using a four contact nerve-cuff electrode," IEEE Transactions on neural
systems and rehabilitation engineering, vol. 12, pp. 251-257, 2004.
[19] Z. Lertmanorat, K. J. Gustafson, and D. M. Durand, "Electrode array for reversing
the recruitment order of peripheral nerve stimulation: experimental studies," Annals
of biomedical engineering, vol. 34, pp. 152-160, 2006.
[20] G. G. Naples, J. T. Mortimer, A. Scheiner, and J. D. Sweeney, "A spiral nerve cuff
electrode for peripheral nerve stimulation," IEEE transactions on biomedical
engineering, vol. 35, pp. 905-916, 1988.
[21] D. J. Tyler and D. M. Durand, "Functionally selective peripheral nerve stimulation
with a flat interface nerve electrode," Neural Systems and Rehabilitation Engineering,
IEEE Transactions on, vol. 10, pp. 294-303, 2002.
[22] J. O. Larsen, M. Thomsen, M. Haugland, and T. Sinkjær, "Degeneration and
regeneration in rabbit peripheral nerve with long-term nerve cuff electrode implant: a
stereological study of myelinated and unmyelinated axons," Acta neuropathologica,
vol. 96, pp. 365-378, 1998.
[23] T. Suzuki, N. Kotake, K. Mabuchi, and S. Takeuchi, "Flexible Regeneration-type
Nerve Electrode with Integrated Microfluidic Channels," in Microtechnologies in
Medicine and Biology, 2006 International Conference on, 2006, pp. 303-305.
[24] T. Stieglitz, H. Beutel, and J.-U. Meyer, "A flexible, light-weight multichannel sieve
electrode with integrated cables for interfacing regenerating peripheral nerves,"
Sensors and Actuators A: Physical, vol. 60, pp. 240-243, 1997.
[25] M. Frankel, "Peripheral Nerve Interface, Intraneural Electrode," Encyclopedia of
Computational Neuroscience, pp. 2297-2299, 2015.
[26] T. Boretius, J. Badia, A. Pascual-Font, M. Schuettler, X. Navarro, K. Yoshida, et al.,
"A transverse intrafascicular multichannel electrode (TIME) to interface with the
peripheral nerve," Biosensors and Bioelectronics, vol. 26, pp. 62-69, 2010.
[27] A. Branner and R. A. Normann, "A multielectrode array for intrafascicular recording
and stimulation in sciatic nerve of cats," Brain research bulletin, vol. 51, pp. 293-306,
2000.

109
[28] S. M. Lawrence, G. S. Dhillon, and K. W. Horch, "Fabrication and characteristics of
an implantable, polymer-based, intrafascicular electrode," Journal of neuroscience
methods, vol. 131, pp. 9-26, 2003.
[29] J. J. FitzGerald, S. P. Lacour, S. B. McMahon, and J. W. Fawcett, "Microchannels as
axonal amplifiers," IEEE Transactions on Biomedical Engineering, vol. 55, pp. 1136-
1146, 2008.
[30] A. F. Mensinger, D. J. Anderson, C. J. Buchko, M. A. Johnson, D. C. Martin, P. A.
Tresco, et al., "Chronic recording of regenerating VIIIth nerve axons with a sieve
electrode," Journal of Neurophysiology, vol. 83, pp. 611-615, 2000.
[31] J. Badia, T. Boretius, A. Pascual-Font, E. Udina, T. Stieglitz, and X. Navarro,
"Biocompatibility of chronically implanted transverse intrafascicular multichannel
electrode (TIME) in the rat sciatic nerve," IEEE Transactions on Biomedical
Engineering, vol. 58, pp. 2324-2332, 2011.
[32] T. Lefurge, E. Goodall, K. Horch, L. Stensaas, and A. Schoenberg, "Chronically
implanted intrafascicular recording electrodes," Annals of biomedical engineering,
vol. 19, pp. 197-207, 1991.
[33] S. M. Lawrence, J. O. Larsen, K. W. Horch, R. Riso, and T. Sinkjær, "Long ‐term
biocompatibility of implanted polymer ‐based intrafascicular electrodes," Journal of
biomedical materials research, vol. 63, pp. 501-506, 2002.
[34] N. Lago, K. Yoshida, K. P. Koch, and X. Navarro, "Assessment of biocompatibility
of chronically implanted polyimide and platinum intrafascicular electrodes," IEEE
Transactions on Biomedical Engineering, vol. 54, pp. 281-290, 2007.
[35] F. Viterbo, J. C. Trindade, K. Hoshino, and A. Mazzoni, "Two end-to-side
neurorrhaphies and nerve graft with removal of the epineural sheath: experimental
study in rats," British Journal of Plastic Surgery, vol. 47, pp. 75-80, 1994/01/01/ 1994.
[36] P. Tos, G. Colzani, D. Ciclamini, P. Titolo, P. Pugliese, and S. Artiaco, "Clinical
Applications of End-to-Side Neurorrhaphy: An Update," BioMed Research
International, vol. 2014, p. 646128, 07/20
[37] F. Šámal, P. Haninec, O. Raška, and P. Dubový, "Quantitative assessment of the
ability of collateral sprouting of the motor and primary sensory neurons after the end-
to-side neurorrhaphy of the rat musculocutaneous nerve with the ulnar nerve," Annals
of Anatomy - Anatomischer Anzeiger, vol. 188, pp. 337-344, 7/3/ 2006.
[38] U. Kovačič, T. Žele, M. Tomšič, J. Sketelj, and F. F. Bajrović, "Influence of breaching
the connective sheaths of the donor nerve on its myelinated sensory axons and on

110
their sprouting into the end-to-side coapted nerve in the rat," Journal of neurotrauma,
vol. 29, pp. 2805-2815, 2012.
[39] P. Haninec, R. Kaiser, and P. Dubový, "A Comparison of collateral sprouting of
sensory and motor axons after end-to-side neurorrhaphy with and without the
perineurial window," Plastic and reconstructive surgery, vol. 130, pp. 609-614, 2012.
[40] H. F. Liu, Z. G. Chen, T. L. Fang, P. Arnold, W. C. Lineaweaver, and J. Zhang,
"Changes of the donor nerve in end ‐ to ‐ side neurorrhaphies with epineurial
window and partial neurectomy: a long ‐ term evaluation in the rat model,"
Microsurgery, vol. 34, pp. 136-144, 2014.
[41] B. Rydevik, M. D. Brown, T. Ehira, and C. Nordborg, "Effects of collagenase on
nerve tissue. An experimental study on acute and long-term effects in rabbits," Spine
(Phila Pa 1976), vol. 10, pp. 562-6, Jul-Aug 1985.
[42] W. V. McCallister, P. Tang, J. Smith, and T. E. Trumble, "Axonal regeneration
stimulated by the combination of nerve growth factor and ciliary neurotrophic factor
in an end-to-side model," J Hand Surg Am, vol. 26, pp. 478-88, May 2001.
[43] W. C. Liao, Y. J. Wang, M. C. Huang, and G. F. Tseng, "Methylcobalamin facilitates
collateral sprouting of donor axons and innervation of recipient muscle in end-to-side
neurorrhaphy in rats," PLoS One, vol. 8, p. e76302, 2013.
[44] L. Isaacson, B. Saffran, and K. Crutcher, "Nerve growth factor ‐induced sprouting
of mature, uninjured sympathetic axons," Journal of Comparative Neurology, vol.
326, pp. 327-336, 1992.
[45] W. V. McCallister, P. Tang, J. Smith, and T. E. Trumble, "Axonal regeneration
stimulated by the combination of nerve growth factor and ciliary neurotrophic factor
in an end-to-side model," The Journal of hand surgery, vol. 26, pp. 478-488, 2001.
[46] W.-C. Liao, Y.-J. Wang, M.-C. Huang, and G.-F. Tseng, "Methylcobalamin facilitates
collateral sprouting of donor axons and innervation of recipient muscle in end-to-side
neurorrhaphy in rats," PloS one, vol. 8, p. e76302, 2013.
[47] K. M. Rich, J. R. Luszczynski, P. A. Osborne, and E. M. Johnson, "Nerve growth
factor protects adult sensory neurons from cell death and atrophy caused by nerve
injury," Journal of neurocytology, vol. 16, pp. 261-268, 1987.
[48] G. Loeb, W. Marks, and P. Beatty, "Analysis and microelectronic design of tubular
electrode arrays intended for chronic, multiple singleunit recording from captured
nerve fibres," Medical and Biological Engineering and Computing, vol. 15, pp. 195-
201, 1977.

111
[49] J. J. FitzGerald, S. P. Lacour, S. B. McMahon, and J. W. Fawcett, "Microchannel
electrodes for recording and stimulation: in vitro evaluation," IEEE Transactions on
Biomedical Engineering, vol. 56, pp. 1524-1534, 2009.
[50] S. P. Lacour, J. J. Fitzgerald, N. Lago, E. Tarte, S. McMahon, and J. Fawcett, "Long
micro-channel electrode arrays: a novel type of regenerative peripheral nerve
interface," IEEE transactions on neural systems and rehabilitation engineering, vol.
17, pp. 454-460, 2009.
[51] E. Meng, Biomedical microsystems: CRC Press, 2011.
[52] B. J. Kim and E. Meng, "Micromachining of Parylene C for bioMEMS," Polymers
for Advanced Technologies, 2015.
[53] H. Schmalbruch, "Fiber composition of the rat sciatic nerve," The Anatomical Record,
vol. 215, pp. 71-81, 1986.
[54] L. Geddes and R. Roeder, "Criteria for the selection of materials for implanted
electrodes," Annals of biomedical engineering, vol. 31, pp. 879-890, 2003.
[55] Y. Nam, "Material considerations for in vitro neural interface technology," MRS
bulletin, vol. 37, pp. 566-572, 2012.
[56] B. Wang and J. Weiland, "Reduction of Edge Effect on Disk Electrodes by Optimized
Current Waveform," in Fred S. Grodins Graduate Research Symposium Los Angeles,
CA, 2015, p. crucibles and materials.
[57] J. S. Ordonez, V. Pikov, H. Wiggins, C. Patten, T. Stieglitz, J. Rickert, et al., "Cuff
electrodes for very small diameter nerves—Prototyping and first recordings in vivo,"
in Engineering in Medicine and Biology Society (EMBC), 2014 36th Annual
International Conference of the IEEE, 2014, pp. 6846-6849.
[58] H. Yu, W. Xiong, H. Zhang, W. Wang, and Z. Li, "A Parylene Self-Locking Cuff
Electrode for Peripheral Nerve Stimulation and Recording," Journal of
Micoelectromechanical Systems, vol. 23, 2014.
[59] N. Xue, T. Sun, W. M. Tsang, I. Delgado-Martinez, S.-H. Lee, S. Sheshadri, et al.,
"Polymeric C-shaped Cuff Electrode for Recording of Peripheral Nerve Signal,"
Sensors and Actuators B: Chemical, 2015.
[60] M. J. Madou, Manufacturing techniques for microfabrication and nanotechnology
vol. 2: CRC Press, 2011.

112
[61] H.-S. Noh, Y. Huang, and P. J. Hesketh, "Parylene micromolding, a rapid and low-
cost fabrication method for parylene microchannel," Sensors and Actuators B:
Chemical, vol. 102, pp. 78-85, 2004.
[62] E. M. Davis, N. M. Benetatos, W. F. Regnault, K. I. Winey, and Y. A. Elabd, "The
influence of thermal history on structure and water transport in Parylene C coatings,"
Polymer, vol. 52, pp. 5378-5386, 2011.
[63] B. Kim, B. Chen, M. Gupta, and E. Meng, "Formation of three-dimensional Parylene
C structures via thermoforming," Journal of Micromechanics and Microengineering,
vol. 24, p. 065003, 2014.
[64] K. Pruden, K. Sinclair, and S. Beaudoin, "Characterization of parylene ‐N and
parylene ‐ C photooxidation," Journal of Polymer Science Part A: Polymer
Chemistry, vol. 41, pp. 1486-1496, 2003.
[65] M. Bera, A. Rivaton, C. Gandon, and J. Gardette, "Comparison of the
photodegradation of parylene C and parylene N," European Polymer Journal, vol. 36,
pp. 1765-1777, 2000.
[66] J. Fortin and T.-M. Lu, "Ultraviolet radiation induced degradation of poly-para-
xylylene (parylene) thin films," Thin Solid Films, vol. 397, pp. 223-228, 2001.
[67] M.-S. Ju, C.-C. K. Lin, J.-L. Fan, and R.-J. Chen, "Transverse elasticity and blood
perfusion of sciatic nerves under in situ circular compression," Journal of
biomechanics, vol. 39, pp. 97-102, 2006.
[68] V. Normand, D. L. Lootens, E. Amici, K. P. Plucknett, and P. Aymard, "New insight
into agarose gel mechanical properties," Biomacromolecules, vol. 1, pp. 730-738,
2000.
[69] G. Cirillo, M. R. Bianco, A. M. Colangelo, C. Cavaliere, L. Zaccaro, L. Alberghina,
et al., "Reactive astrocytosis-induced perturbation of synaptic homeostasis is restored
by nerve growth factor," Neurobiology of disease, vol. 41, pp. 630-639, 2011.
[70] D. Zhan, J. Velmurugan, and M. V. Mirkin, "Adsorption/desorption of hydrogen on
Pt nanoelectrodes: Evidence of surface diffusion and spillover," Journal of the
American Chemical Society, vol. 131, pp. 14756-14760, 2009.
[71] S. F. Cogan, "Neural stimulation and recording electrodes," Annu. Rev. Biomed. Eng.,
vol. 10, pp. 275-309, 2008.
[72] A. N. Frumkin, "Hydrogen overvoltage and adsorption phenomena -- 2,," Advances
in Electrochemistry and Electrochemical Engineering, vol. 3, pp. 287-391, 1962.

113
[73] J. Wolpaw and E. W. Wolpaw, Brain-computer interfaces: principles and practice:
OUP USA, 2012.
[74] A. M. Cobo, B. Boyajian, C. Larson, K. Schotten, V. Pikov, and E. Meng, "A parylene
cuff electrode for peripheral nerve recording and drug delivery," in Micro Electro
Mechanical Systems (MEMS), 2017 IEEE 30th International Conference on, 2017,
pp. 506-509.
[75] X. Kang, J.-Q. Liu, H. Tian, B. Yang, Y. Nuli, and C. Yang, "Self-Closed Parylene
Cuff Electrode for Peripheral Nerve Recording," Microelectromechanical Systems,
Journal of, vol. 24, pp. 319-332, 2015.

114

Ample evidence has been provided in the previous chapters on the suitability of
Parylene-based MEMS devices for drug delivery and neural applications. But still,
additional work is required on the characterization and validation of the Lyse-and-Attract
Cuff Electrode (LACE) as a peripheral nerve interface. Here we present two studies in an
effort to further advance the LACE device: (1) measuring long-term (2 weeks) probe
integrity and performance under simulated in vivo conditions and (2) improving the LACE
microfluidic channel design to feature improved mechanical robustness for succesfull drug
delivery in acute in vivo animal testing.
Background
The LACE is an extraneural peripheral nerve interface designed to be minimally
invasive while providing high fascicle selectivity and reliable chronic neural recording and
stimulation. Improved selectivity will be achieved by inducing axons to grow into
microfluidic channels with embedded electrodes. Collateral sprouting will be selectively
and atraumatically induced from the fascicles closest to the microchannel outlet via a
highly localized, intra-neural infusion of collagenase (for transient digestion of collagen
fibers in the epineurium and perineurium that mechanically impede the sprouting process)
and the neurotrophic factors NGF and methylcobalamin (for inducing, promoting, and
spatially guiding the axonal sprouting toward the microchannel and associated electrode)
[1-4]. The LACE is a multi-layered, thin-film polymer (Parylene C) device consisting of a
pair of recording and stimulating platinum electrodes embedded within microfluidic
channels (Figure 4-1). Four microfluidic channels were incorporated into the LACE to
target multiple fascicles found in the rat sciatic nerve.

RELIABILITY TESTING OF THE LYSE-AND-ATTRACT CUFF
ELECTRODE TOWARDS IN VIVO IMPLEMENTATION

115

Figure 4-1. Optical micrograph of fabricated LACE (reprinted with permission from [5] ©
2017 IEEE).

The electrical, fluidic, and mechanical features of the LACE must be tested under
simulated and acute in vivo conditions to demonstrate the viability and reliability of the
device as an implantable neural-interface for chronic in vivo applications. The major areas
of concern are the integrity of the electrical insulation, performance of the microfluidic
channel, and efficacy of the interlocking mechanism, which holds the device in place
around a nerve. Under in vivo conditions the polymer insulation will be in constant
exposure to the harsh physiological fluid, and there is a risk of electrical or electrochemical
failure due to moisture intrusion or the action of saline. Polymers are known to be
permeable to water vapor and ions to some degree [6, 7]. The condensation of water vapor
between layers of a polymer-based device can increase ion diffusion into the layers
generating pressure that leads to delamination and compromising the functional integrity
of the electrical features (electrodes and traces) resulting in leakage current. Post-process
thermal annealing treatment has been shown to increase the in vivo lifetime of multi-
layered Parylene devices by decreasing water vapor permeation and improving adhesion
between polymer layers [8]. Electrical and electrochemical testing were performed on
LACE devices subjected to a wet and saline environment to determine device lifetime for
the planned in vivo study (2 weeks).
Results of initial fluidic channel characterization presented in Chapter 3 revealed
that the original Parylene microfluidic channel configuration was prone to collapse along

116
the horizontal feeder channel when the LACE is curled as if encircling a nerve. The stiction
forces sealing the channel could not be overcome by increasing the fluid pressure without
damaging the channel, as such resulting in blocked fluid flow. As the design for the LACE
calls for fluid flow through all the microfluidic channels to ultimately induce axonal
sprouting from multiple fascicles inside the nerve, a new microfluidic channel design is
required to prevent such obstructions. Microfluidic channel were redesigned, fabricated,
and tested in simulated and acute in vivo conditions.
Long-term Simulated In Vivo Characterization
A fundamental part of a reliable peripheral nerve interface is performance of the
insulation material and the electrodes. Simulated in vivo testing was performed to assess
the risk of insulation failure, moisture intrusion, and electrode oxidation due to chronic
moisture and saline exposure. LACE devices were submerged in a simulated in vivo
environment (phosphate buffered saline, 1×PBS at room and body temperature) for the
lifetime of the planned in vivo study (2 weeks). Devices underwent electrical and
electrochemical measurement; cross-talk measurements were performed to assess
insulation integrity, and electrochemical impedance spectroscopy (EIS) was performed to
characterize changes in the electrochemical properties of electrodes and delamination of
the polymer insulation.
LACE devices were fabricated using standard microfabrication techniques as
described in Chapter 3 and Appendix D. As LACE electrodes are intended to record and
stimulate axons which have grown into the microfluidic channels, the primary concern is
the electrochemical properties of the electrode surface; therefore, long-term soak
experiments were performed on LACE devices without microfluidic channels. Thermally
treated devices were annealed at 200 °C for 48 hrs under vacuum. Cyclic voltammetry
(CV) in 0.05M H2SO4 was performed on all the devices to electrochemically clean the
electrode surface under a constant stream of N2 in a three-electrode cell. The working
electrode was cycled between −0.2 to 1.2 V with respect to an Ag/AgCl (3M NaCl)
reference, a potential range that is within the water window [9]. A 1 cm
2
Pt plate served as

117
a counter electrode. Each electrode was cycled 30 times using a scan rate of 250 mV/s.
LACE devices were then epoxied to vial caps and soaked in 1× PBS to simulate in vivo
conditions (Figure 4-2). The PBS solution was replaced weekly to prevent concentration
changes due to evaporation. Cross-talk and EIS measurements were taken at room
temperature (RT) for day zero. The rest of the cross-talk and EIS measurements were taken
periodically in 1× PBS at room (23 °C) or body (37 °C) temperature depending on the
study.

Figure 4-2. LACE devices were epoxied into vial caps for long-term soaking experiments.
4.2.1 Inter-site cross-talk
An automated testing system was developed in our laboratory to measure inter-
electrode cross-talk. The system was designed to measure the voltage bias induced on an
electrode due to stimulation of an adjacent electrode, as resulting from compromised
insulation. The system consists of a VirtualBench tool (National Instruments, Austin, TX),
two relay modules, and a LabVIEW program. The protocol entails applying a 1 kHz
voltage signal to one electrode while measuring the potential of an adjacent electrode with
respect to ground. The program analyzes the power spectrum of the signal and uses the
amplitude at 1 kHz for calculating cross-talk. This effectively isolates the signal of interest
and filters out other noise sources (e.g. thermal noise and 60 Hz wall noise). Cross-talk is
defined as the ratio Vread,n /Vsend, and is measured for each pair of electrodes n. Calibration

118
steps were incorporated to account for the baseline cross-talk of the system itself. Detailed
information about the cross-talk system and testing protocol can be found in Appendix G.
During each measurement, the LACE device was removed from the saline solution,
the surface was dried, and the measurement performed. We worked under the assumption
that moisture soaked into the Parylene film, or trapped between insulation layers, would
not evaporate before the measurement was taken; this assumption was supported by a series
of validation measurements on intentionally compromised devices. Once the
measurements were taken the device was re-submerged to resume soaking. This method
allowed the exclusion of the conductive path between electrodes in solution while still
detecting cross-talk caused by moisture trapped in areas of compromised insulation.
The results of each cross-talk test are represented as a grid, as shown in Figure 4-3.
Each row index represents the stimulating electrode, while the column index represents the
electrode that was measured. There are several notable features about this representation.
First, the diagonal positions correspond to stimulation and measurement of the same
electrode; we expect an output of 100% in these cases. Second, we should expect values to
be mirrored across the diagonal. Lastly, the more proximal a value is to the diagonal, the
closer the corresponding electrodes’ metal traces are to each other. Because of this layout,
as water intrusion increases over time and induces cross-talk, we expect values near the
diagonal to increase first, followed later by values that are farther from the diagonal.

119

Figure 4-3. (a) Data output from the automated cross-talk testing system is represented as
a grid. The values shown here represent ideal results from a perfectly-insulated device. (b)
Electrode numbering for the LACE. SE is the surface electrode not contained within a
microfluidic channel.
4.2.1.1 Cross-talk: effects of temperature and heat treatment
The cross-talk system was used to reveal differences in cross-talk development in
relation to soak temperature and thermal annealing. Cross-talk was evaluated in several
LACE devices that had been soaking in 1× PBS for approximately three weeks.
The cross-talk data is shown in Figure 4-4. The un-annealed device soaked at body
temperature (37 °C) exhibited catastrophic insulation failure and very high levels of cross-
talk. Similar un-annealed devices soaked at room temperature did not exhibit the same
failure. Very little cross-talk was observed in annealed devices, even after 20+ days of
soaking at 37 °C. This result was consistent across multiple devices.
High cross-talk values in the un-annealed devices soaked at 37 °C suggest they
would likely fail following implantation, before the completion of the proposed chronic
recording experiments. Good electrical insulation of the un-annealed devices soaked at
room temperature suggests that the rate of moisture diffusion through the Parylene bulk,
and between the Parylene layers, decreases at lower temperatures. The minimal cross-talk
values seen in annealed devices soaked at 37 °C indicates improved electrical insulation
and reduced rates of moisture intrusion due to increased Parylene adhesion and crystallinity

120
after the thermal treatment [10]. This data suggests that annealed LACE device will have
sufficient electrical insulation during chronic in vivo studies.

Figure 4-4. Cross-talk testing was performed on several devices that had been soaking for
approximately 3 weeks to reveal differences in cross-talk development in relation to soak
temperature and thermal annealing. For comparison, cross-talk results for an un-soaked
device are also shown (Dry). Values that are negative, greater than 100, or asymmetrical
across the diagonal are attributed to noise.
4.2.1.2 Cross-talk: effects of current application
In a subsequent experiment, annealed LACE devices (n=5) were soaked for four-
weeks (1× PBS at 37 °C) with cross-talk and EIS measurements performed periodically. A
single un-annealed device served as an experimental control. Results of the EIS testing is
presented in the following section (4.2.2). Cross-talk measurements were first collected
prior to initial CV cleaning to ensure good electrical insulation before soaking experiments.
Cross-talk testing was performed periodically throughout the four-week duration, in
parallel with EIS data collection. Cross-talk measurements were collected as previously
described, alongside microscopy imaging. Once the measurements were taken, the device
was re-submerged to continue soaking.
Inter-electrode cross-talk testing pre and post CV cleaning in 0.05M H2SO4
indicated no fabrication defeats (e.g. metal short) and good electrical insulation in all

121
devices (n=6), as shown in Figure 4-5. However, in contrast with the prior experiment
(section 4.2.1.1), cross-talk was observed almost immediately after the beginning of this
experiment. Values as high as 23% cross-talk were observed after only 2 days of soaking
as shown in Figure 4-6. As may be expected, cross-talk was initially highest between
adjacent electrodes, followed by increasing cross-talk between spatially separated
electrodes. This pattern suggests gradual failure of the insulation and increasing water
intrusion, creating a continuous short throughout the entire device by the end of the four-
week period.
The only change between this experiment and the one described in 4.2.1.1 is the
inclusion of the EIS (17 EIS measurements were taken per electrode during the 1 month
soak experiment). It is hypothesized that the application of current to the LACE devices,
a feature of EIS testing, accelerated insulation failure seen in this study.

Figure 4-5. Representative cross-talk data for a device (a) pre, and (b) post cyclic
voltammetry cleaning in 0.05M H2SO4.

122

Figure 4-6. Representative long-term cross-talk data for a LACE device after EIS
measurements (except for "Dry"). Percentage cross-talk value is the average across all
measurements (n = 56).
4.2.2 Electrochemical impedance spectroscopy
EIS measurements were collected on annealed LACE devices (n=5) throughout
four-weeks of soaking in 1× PBS at 37 °C. EIS measurements were performed in 1× PBS
with an AC perturbation signal of 25 mV (rms) in the frequency range of 1-10
5
Hz. An
Ag/AgCl (3M NaCl) reference and 1 cm
2
Pt plate counter were used. EIS was performed
at room temperature for day zero. The rest of the EIS measurements were taken in 1× PBS
at 37 °C.
Long-term EIS measurements showed a decrease in impedance magnitude and an
increase in phase at low frequencies with increased soak time, indicating water absorption.
The changes at lower frequencies became more evident after 2 weeks of soaking as shown
in Figure 4-7. A decrease in impedance magnitude at higher frequencies was observed after
1 day of soaking. Insulation integrity for the un-annealed device was compromised earlier
than annealed devices, as can be seen by the large drop in impedance magnitude and the
phase shift after 1 day of soaking (Figure 4-8 and Figure 4-9). The phase of the un-annealed
device also exhibited a second time constant after 1 day of soaking suggesting
compromised Parylene insulation layer along the traces of the LACE (Figure 4-9d). This

123
was supported by visual delamination observed at the small metal traces as shown in Figure
4-10.

Figure 4-7. Representative EIS (a) magnitude and (b) phase of an annealed LACE device
long-term soak test at 37° C in 1× PBS. Mean ± SE, n = 8 electrodes.

Figure 4-8. EIS (a) magnitude and (b) phase of the un-annealed LACE device (control)
undergoing long-term soak test at 37° C in 1× PBS. Mean ± SE, n = 8 electrodes.

124

Figure 4-9. EIS magnitude and phase for an annealed and un-annealed LACE device
undergoing soak experiments in 1× PBS at 37 °C. Mean ± SE, n = 8 electrodes. (a-b)
annealed device, and (c-d) un-annealed device.

Figure 4-10. Optical micrograph of an un-annealed LACE device after 1 day of soaking in
1× PBS at 37 °C shows visual delamination at the small metal traces but not other features.

125

It is important to analyze the insulation performance of neural interfaces at 1 kHz
since this is the frequency of action potentials [11, 12]. EIS measurements of annealed
devices during the first 2 weeks of soak testing yielded an electrical impedance magnitude
of 2.2 ± 0.1 kΩ and a phase of -59.5 ± 0.1º at 1 kHz, demonstrating low variation between
devices (mean ± SE, n = 40 electrodes across 5 probes). Figure 4-11 shows a significant
(ANOVA, p < 0.05) drop in impedance magnitude after 1 day of soaking followed by an
increase in impedance magnitude at day 2. Visual delamination was observed as early as
day 9, as indicated in Figure 4-11 by the colored vertical lines and corresponding optical
micrographs.

Figure 4-11. Representative EIS magnitude at 1 kHz for an annealed LACE device soaked
in 1× PBS at 37°C. Colored lines and optical micrographs show the beginning of
observable delamination.

Several research groups report similar changes in the impedance magnitude and
phase at low frequencies when soaking un-annealed Parylene coated electrodes in PBS,
and most claimed the changes were attributed to water penetration after one day of soaking
[13, 14]. In some cases, the dramatic changes occurred within 1 hour of soaking [14]. Water

126
penetration can led to delamination of un-annealed and annealed Parylene films resulting
in a decrease in magnitude and an increase in phase at low frequencies. The decrease in
impedance magnitude at higher frequencies suggests a decrease in solution resistance
resulting from an increase in electrode area. Electrodes were fabricated with a small border
of Parylene insulation (designed width: 20 µm and width after RIE: 13 µm) around the
periphery to prevent the isotropic nature of plasma etching or any slight photolithography
misalignment from changing the electroactive area of the electrode (Figure 4-12). The
geometric surface area of the exposed electrode is approximately 0.47 mm
2
and the area of
the same electrode with compromised insulation can be approximately 0.52 mm
2
, which
represents a 10% increase in electrode area. Superior electrochemical performance of
annealed devices compared to un-annealed can be attributed to the improved Parylene
insulation common to thermally threated devices.
Recorded impedances at 1 kHz are expected to provide adequate neural recording
potential. The drop in impedance magnitude at 1 kHz after 1 day of soaking indicates water
saturation of the Parylene layers which was followed by an increase in impedance
magnitude suggesting salt intrusion [15]. Impedance magnitude values continue to drop as
a function of soak time, approaching the impedance magnitude of an un-insulated electrode
(~ 0.19 kΩ, n = 4). The drop in impedance magnitude can be attributed to increased
electrode area due to compromised insulation along the metal trace. The hypothesis that
application of electrical current accelerates device failure was supported by local
delamination observed at the metal features subjected to current (Figure 4-13). Metal
features not subjected to current (e.g. locking slit, serrated teeth) did not exhibit any visible
delamination. Additionally, no Parylene-Parylene interface delamination was observed.

127

Figure 4-12. Electrodes are exposed using a plasma etching process and leaving a thin
border of Parylene defining the electrode boundary.

Figure 4-13. Representative optical micrographs of a long-term tested device at day 25
showing delamination limited to metal features subjected to current.
4.2.3 Cyclic voltammetry
Cross-talk and EIS data suggested that the application of current led to accelerated
device failure during the long-term soak experiment. To further test this hypothesis, new
devices were CV cleaned in 0.05M H2SO4 with low scan rates (50 mV/sec) to prolong the
exposure of the electrodes and metal traces to current. No delamination was observed. The
next step was CV in 1× PBS at the same slow scan rate (50 mV/sec). Several electrodes on
a LACE device were tested (n=5) and all of them exhibited severe delamination as shown
in Figure 4-14. Electrode delamination seen during the CV test (as opposed to the metal
traces delamination observed in cross-talk and EIS testing) suggests that the Parylene
insulation was not fully compromised; therefore, current application in the saline solution
primarily attacked the adhesion between the Parylene and exposed metal electrode
interface. This suggests that the combination of a saline solution and current are the main

128
contributors to the accelerated failure seen on the long-term electrical and electrochemical
characterization of LACE devices. This hypothesis was additionally supported by
mechanical testing data on multi-layer Parylene devices (unpublished data by Jessica
Ortigoza). In this study, the integrity of the devices was assessed by soaking devices in 1×
PBS at 37°C and evaluating the Parylene-metal bond via visual inspection and peel test.
Delamination was observed after 4 days for un-annealed devices and 3 weeks for annealed
devices, demonstrating prolonged device integrity in the absence of current.

Figure 4-14. Metal delamination (metal wrinkling) after CV in 1× PBS with a scan rate of
50 mV/sec. Only the tested electrodes showed delamination (n=5 electrodes).
4.2.4 Conclusion
Improved electrical insulation in simulated chronic (20+ days) in vivo conditions
was achieved with annealed LACE devices. Low impedance values at 1 kHz were recorded
and are expected to provide adequate neural recording potential. However, poor insulation
integrity after 2 days of soaking and delamination after 2 weeks of soaking were observed
in devices tested with the cross-talk system in conjugation with EIS. It is hypothesized that
application of electrical current in a saline solution accelerates LACE device failure which
will limit the recording and stimulation capabilities of our neural interface. Failed
insulation will prevent electrode isolation from other electrodes as well as the surrounding
environment. Compromised insulation between electrode traces will result in neural signals
carried by one electrode to interfere with the signals carried by the neighboring electrodes.
Compromised insulation between the electrode and the surrounding environment will
diminish the electrode’s ability to isolate targeted neural signals. As far as stimulation,
compromised insulation will result in current from one electrode being carried by

129
neighboring electrodes leading to stimulation of non-targeted areas and current attenuation
at targeted area. Ultimately, decreasing stimulation selectivity and spatial resolution of
functional responses.
Fortunately, water and salt intrusion through the bulk Parylene and their penetration
between the Parylene layers interface can be minimized to improve insulation integrity,
and recording and stimulating performance in vivo. Diffusion through the bulk Parylene
can be decreased by surface modification of the Parylene and metal to a more hydrophobic
surface as shown in [13, 16]. The interfacial adhesion of Parylene can be improved by the
addition of adhesion promotes between Parylene layers such as AdPro Plus
TM
(Specialty
Coating Systems, Indiana), and ethylene glycol diacrylate (EGDA) ([17] and unpublished
data by Jessica Ortigoza).
Simulated and Acute In vivo Infusion
4.3.1 Microfluidic channels redesign
Initial fluidic channel characterization showed that in the curled configuration, flow
was achieved in 3 out of 4 microfluidic channels. Flow through the distal channel (channel
4) was blocked by pinching of the horizontal feeder channel. Two separate fluidic channel
designs were investigated to improve mechanical robustness, particularly during
deformation, and allow uniform flow through all channels in the curled orientation. Design
2 incorporates polymer support walls (20 µm H × 20 µm W × 390 µm L) along the
horizontal channel as shown in Figure 4-15b and Figure 4-16a [18, 19]. Mid-channel
support walls were fabricated from Parylene along the horizontal feeder channels at sites
where channel collapse was previously observed. Design 3 incorporates microfluidic
channels with narrower widths and lower aspect ratio (Figure 4-15c, Figure 4-16b).
Standard Parylene microfabrication techniques were employed as described in detailed in
chapter 3.

130

Figure 4-15. Schematic of LACE showing three different microfluidic channel
configurations to support channel integrity when the cuff is curled: (a) Design 1: standard
channels (20 µm H x 250 µm W), (b) design 2: incorporation of mid-channel supporting
walls (20 µm H × 20 µm W × 390 µm L), and (c) design 3: narrower channels (150 µm
and 250 µm W sections) (modified from [5] © 2017 IEEE).

Figure 4-16. Optical micrographs of LACE showing the two new microfluidic channel
configurations: (a) design 2: incorporation of mid-channel supporting walls, and (b) design
3: narrower channels.
4.3.2 Simulated in vivo infusion
4.3.2.1 Uniform fluidic channel flow
Microfluidic channels were evaluated in curled devices by infusing colored dye at

131
a 500 nL/min flow rate using an external syringe pump (Harvard Apparatus
Infuse/Withdraw PHD 2000) and imaging the transverse progression of the dye with a
microscope (Caltex Scientific HD60T). Flat LACE devices were submerged in distilled
water under the microscope, and primed by flushing the channels with isopropyl alcohol
and distilled water for 5 minutes each. Devices were then curled, placed upright in an
acrylic jig, and submerged under distilled water while infusing the microfluidic channels
with blue dye.
Uniform flow through all fluidic channels and outlet ports was achieved with design
2 (n = 3) (Figure 4-17b). Flow through some channels was blocked by pinching of the
horizontal feeder channel in designs 1 and 3 (Figure 4-17a & c). Design 2 was selected for
future animal testing.

Figure 4-17. Representative optical micrographs of the infusion experiment in curled
devices at 500 nL/min flow rate (transverse view). The channels were primed with the dye
introduced at the microfluidic inlet channel. (a) Design 1, (b) design 2, and (c) design 3.
Scale bar is 1 mm (reprinted with permission from [5] © 2017 IEEE).
4.3.2.2 Localized drug delivery
Localization of fluid delivery was demonstrated by encircling a prototype LACE
around a simulated sciatic nerve, produced from a molded polydimethylsiloxane (PDMS)
core covered with an absorbent wood fiber wipe (Kimtech Science Kimwipe, Kimberly-
Clark, Irving, TX) then driving blue dye through the microfluidic channels using a syringe
pump. The nerve phantom was selected to mimic the size of a rat sciatic nerve for the
purpose of facilitating visualization during infusion studies. First, a PDMS core was cut
from a block using an 18G coring needle (0.97 mm ID). This core was then

132
circumferentially wrapped with two layers of the wipe, increasing the diameter of the nerve
phantom to approximately 1.3 mm, which falls in the measured sciatic nerve diameter
range of 1.2 – 1.4 mm for a 340 g male rat. Devices were primed as previously described
and then curled around the nerve phantom. Dye was infused at 100 nL/min until it reached
the fluidic outlet ports. The nerve phantom was imaged using an optical microscope and
digitally processed using ImageJ (NIH) software to correlate the color signal with
localization. The degree of localization of the infused fluid was quantified by the location
and dimensions of the dyed area.
Localized drug delivery on nerve phantoms was confirmed and is shown in Figure
4-18. Diffusion area was characterized as previously described and the results are shown
in Table 4-1. Localized delivery was achieved through at least three microfluidic channels
with differing flow rates among individual channels. Consequently, the diffusion area for
each channel is slightly different even within the same device. Minimal transverse and
longitudinal dye diffusion observed in this study demonstrates the LACE potential for
highly localized in vivo delivery of lysing agents and neurotrophic growth factors.

Figure 4-18. (a) Transverse view of LACE wrapped around a nerve phantom. Insert shows
side-view. (b) Representative nerve phantom after localized dye delivery (n =3). Insert
scale bar is 2 mm (reprinted with permission from [5] © 2017 IEEE).

133
Table 4-1. Average diameter of outlet port stain calculated using ImageJ (NIH) software
(n = 3) (modified from [5] © 2017 IEEE).
Diameter of fluidic port stain
(mean ± SE)
Device 1
0.209 ± 0.022 mm
Device 2
0.455 ± 0.058 mm
Device 3
0.310 ± 0.056 mm

4.3.3 Acute in vivo infusion
A series of acute in vivo infusion experiments were performed to determine the
optimal infusion regimen through the LACE microfluidic channels to achieve localized
drug delivery on the nerve. These experiments are necessary to determine the correct flow
rate for delivering the lysing agent in future chronic experiments. The LACE microfluidic
channels were primed ex vivo by flushing the channels with isopropyl alcohol and distilled
water for 5 minutes each using a syringe pump (maximum flow 30 µL/hr) with the device
submerged in either solution. The LACE polyurethane catheter was then switched to a
Hamilton syringe loaded with methylene blue dye. The dye was infused using a syringe
pump until it reached the LACE microfluidic channel. Devices with at least three
functioning microfluidic channels were selected for in vivo infusion experiments. All
procedures for the animal experiments were in accordance with the animal protocol
approved by the Animal Care and Use Review Office (ACURO).
The LACE device was surgically placed around the sciatic nerve of anesthetized
rats as shown in Figure 4-19. The nerve was exposed and the tab of the LACE device was
fed under the nerve with the fluidic channel outlet ports in contact with the nerve. The tab
was then threaded through the locking slit and secured in place at the appropriate serrated
tooth, making sure the device fit securely around the nerve. With the LACE tight around
the nerve, dye was infused at various flow rates and durations (up to 5 hours). Following
dye infusion experiments, the animals were sacrificed and the nerves dissected. The nerves
were imaged for evidence of dye diffusion on the surface and penetration into the nerve via

134
optical micrographs and histology, respectively. 10 to 15 µm sections were cut on a cryostat
for histology evaluation.

Figure 4-19. Sequence of photographs illustrating the in vivo implantation for the LACE
around the sciatic nerve of an anesthetized rat. (a) Exposed sciatic nerve, (b) device fed
under the nerve with fluidic ports in contact with the nerve, (c-d) tab threaded through
locking slit, (e) tab kept in place by serrated teeth, and (f) LACE securely wrapped around
the sciatic nerve. Scale bar is 1 mm.
The tested flow rates and durations, and the resulting observations are presented in
Table 4-2. After implantation, some non-functioning ports became functional indicating
that the implant procedure might have dislodged air bubbles from some channels. In some
cases, only one or two ports seemed functional suggesting that device handling could have
caused other channels to become occluded. The 0.3 µL/hr flow rate test showed dye
diffusion along the surface of the nerve, but was attributed to damage to the fluidic channels
accrued during implantation (Figure 4-20). No histology images were obtained for that
experiment since the nerve was damaged during the fixation process. A 1 µL/hr flow rate
provided discrete dye diffusion spots on the surface of the nerve as shown in Figure 4-21a.

135
The histology showed no dye penetration, but given that the lysing agent will disrupt the
connective tissue, such a flow rate may lead the solution to penetrate the nerve and result
in highly localized axonal sprouting (Figure 4-21b). A flow rate between 3 – 4 µL/hr for
more than 2 hours resulted in non-discrete coverage of the nerve circumference. A faster
infusion rate (15 µL/hr) for a shorter period of time (5 minutes) showed non-localized
coverage of the nerve surface with localized dye diffusion into the smallest fascicles of the
nerve as shown in Figure 4-22. It is hypothesized that application of lysing agents and
growth factors could induce somewhat selective sprouting from small (sensory) fascicles,
potentially giving the cuff higher selectivity.
Table 4-2. LACE in vivo infusion of methylene blue dye on the sciatic nerve of anesthetized
rats. *No functional channels post implantation.
Device Dye diffusion Functional
channels
Infusion
protocol
Optical micrograph Histology
1 Longitudinal Sample damaged
during fixation
1 0.3 µL/hr for 3.5hrs
2 Localized None 2 1 µL/hr for 4hrs
3 None* Not processed 0 2 µL/hr for 4hrs
4 Non-discrete Non-discrete 1 3 µL/hr for 4hrs
5 Non-discrete Non-discrete 1 4 µL/hr for 2-3hrs
6 Non-discrete Localized 1 or 2 15 µL/hr for 5 min

136

Figure 4-20. Dye diffused longitudinally along the nerve at a 0.3 µL/hr flow rate.

Figure 4-21. (a) Localized dye delivery shown as discrete spots of dye on the surface of
the nerve with a flow rate of 1 µL/hr. (b) Histology showed no dye diffusion into the nerve.

137

Figure 4-22. (a) Sciatic nerve infused with 15 µL/hr flow rate for 5 minutes showed non-
localized dye coverage of the nerve. (b) Histology results showed dye penetration into the
smallest fascicles.
4.3.4 Conclusion
Redesigned microfluidic channels with incorporated Parylene mid-channel
supporting walls achieved uniform flow through all channels in a curled configuration and
localized fluid delivery on a nerve phantom at low flow rates. Successful fluid flow through
all channels and localized drug delivery are important for targeting multiple fascicles
within the nerve and achieve high selectivity. Infusion variability in acute in vivo
experiments was attributed to the loss of microfluidic channel patency during implantation,
not device damage. Future in vivo implantations can implement the needle suture locking
procedure described in Chapter 3 to minimize forceful device handling with tweezers and
avoid device damage. Device placement around the nerve can be facilitated by
thermoforming the LACE device in a semi curled shape to ease threading of the device tab
through the locking slit. Additionally, the incorporation of tweezer holding areas on the
LACE device can help avoid damage to the fluidic channels and metal features. Lysing
agents infusion experiments are required to determine their appropriate flow rate and

138
infusion duration. If short infusions are adequate, fast flow rates (~ 15 µL/hr) may prove
useful, but if long infusions are required, then lower flow rates (~ 1 µL/hr) might be suitable.
Interlocking Mechanism Acute In Vivo Testing
The cuff incorporates an adjustable locking mechanism which allows the device to
securely fasten around nerves of varying diameters, keeping the microfluidic ports in direct
contact with the epineurium for highly localized drug delivery. This proximity is important
to ensure high signal-to-noise ratio (SNR) recordings. The interlocking mechanism
consists of an elongated tapered tab that loops around the sciatic nerve, inserts into an
etched slit, and is held in place by a set of serrated teeth which set the diameter of the loop
to between 1.1 to 1.5 mm in 0.1 mm increments (Figure 4-23). An additional “buckle” slit
is used to maintain the loop around the nerve. The locking mechanism was tested in vivo
by wrapping a LACE device around the sciatic nerve of an anesthetized 260g rat, and
assessing its holding strength under handling conditions expected during implantation.
(Figure 4-24). The sciatic nerve was then dissected with the LACE device wrapped around
it.

Figure 4-23. Optical micrograph of a LACE device indicating the adjustable interlocking
mechanism features.

139

Figure 4-24. Photographs of the LACE placement on the sciatic nerve. (a) Unlocked LACE
on the sciatic nerve, (b) locked LACE on the sciatic nerve, and (c-d) locked LACE on the
dissected sciatic nerve (reprinted with permission from [5] © 2017 IEEE).
The locking mechanism proved simple to implement and robust; no damage to the
nerve or cuff was observed when engaging and releasing the lock or after curling. Minor
improvements are required to facilitate and improve the locking mechanism for future
implantations. Perforations can be added to the serrated pads to make the cuff more flexible
and compliant with the nerve, and additional serrated teeth can accommodate smaller nerve
diameters.
Summary
Reliability of the LACE in simulated acute and chronic in vivo conditions was
investigated. The insulation integrity was assessed though electrical and electrochemical
testing, the mechanical behavior of the microfluidic channels was evaluated, and initial
acute in vivo experiments were successfully carried out to determine the appropriate
infusion regimen and evaluate the interlocking mechanism.

140
Long-term electrical and electrochemical characterization of LACE devices in
simulated in vivo conditions was performed to evaluate the time-to-failure of the LACE
device. The LACE device exhibited signs of insulation failure with 2 days of soaking for
cross-talk measurements and after 2 weeks of soaking for EIS, raising suspicious about the
characterization methods. It is hypothesized that the application of electrical current to a
multi-layer Parylene C device with salt intrusion leads to accelerated device failure.
Fortunately, several Parylene treatments can be incorporated to improve insulation
integrity and prolog the device in vivo performance. We anticipate the LACE in vivo
recording performance to be adequate for a chronic peripheral interface given that a
previously multi-layer Parylene-based neural probe developed in our laboratory
demonstrated stable electrophysiological recordings for 12 months of implantation in vivo
[20]. The lifetime performance of the stimulating and recording LACE device should be
determined with a long-term simulated in vivo study where the desired in vivo stimulation
parameters (e.g. amplitude, pulse width, frequency) are applied.
Redesigned microfluidic channel achieved sufficient mechanical robustness to
avoid channel collapse during deformation. Uniform flow through all channels and
localized fluid delivery on a nerve phantom was achieved which will allow targeting of
multiple fascicles within the nerve. Initial dye infusion tests in vivo demonstrated that the
LACE device can provide localized drug delivery to sciatic nerve fascicles. Fast flow rates
were adequate for short infusion durations, while slow flow rate were suitable for long
infusions. Acute in vivo studies proved the adjustable locking mechanism to be simple to
implement and provided adequate holding strength when wrapping around the rat sciatic
nerve. Modifications to the interlocking mechanism design and surgical placement
protocol will improve drug delivery by protecting the LACE microfluidic channels and
electrical features during implantation. Initial acute in vivo infusion tests demonstrated
targeted drug delivery that could induce selective sprouting from multiple nerve fascicles
resulting in higher device selectivity.

141
References
[1] B. Rydevik, M. D. Brown, T. Ehira, and C. Nordborg, "Effects of Collagenase on
Nerve Tissue: An Experimental Study on Acute and Long-Term Effects in Rabbits,"
Spine, vol. 10, pp. 562-566, 1985.
[2] L. Isaacson, B. Saffran, and K. Crutcher, "Nerve growth factor ‐induced sprouting
of mature, uninjured sympathetic axons," Journal of Comparative Neurology, vol.
326, pp. 327-336, 1992.
[3] P. Haninec, R. Kaiser, and P. Dubový, "A Comparison of collateral sprouting of
sensory and motor axons after end-to-side neurorrhaphy with and without the
perineurial window," Plastic and reconstructive surgery, vol. 130, pp. 609-614, 2012.
[4] W.-C. Liao, Y.-J. Wang, M.-C. Huang, and G.-F. Tseng, "Methylcobalamin facilitates
collateral sprouting of donor axons and innervation of recipient muscle in end-to-side
neurorrhaphy in rats," PloS one, vol. 8, p. e76302, 2013.
[5] A. M. Cobo, B. Boyajian, C. Larson, K. Schotten, V. Pikov, and E. Meng, "A parylene
cuff electrode for peripheral nerve recording and drug delivery," in Micro Electro
Mechanical Systems (MEMS), 2017 IEEE 30th International Conference on, 2017,
pp. 506-509.
[6] S. Kirsten, M. Schubert, J. Uhlemann, and K.-J. Wolter, "Characterization of ionic
permeability and water vapor transmission rate of polymers used for implantable
electronics," in 2014 36th Annual International Conference of the IEEE Engineering
in Medicine and Biology Society, 2014, pp. 6561-6564.
[7] P. Menon, W. Li, A. Tooker, and Y. Tai, "Characterization of water vapor permeation
through thin film Parylene C," in TRANSDUCERS 2009-2009 International Solid-
State Sensors, Actuators and Microsystems Conference, 2009, pp. 1892-1895.
[8] S. A. Hara, B. J. Kim, J. T. Kuo, and E. Meng, "An electrochemical investigation of
the impact of microfabrication techniques on polymer-based microelectrode neural
interfaces," Journal of Microelectromechanical Systems, vol. 24, pp. 801-809, 2015.
[9] D. Zhan, J. Velmurugan, and M. V. Mirkin, "Adsorption/desorption of hydrogen on
Pt nanoelectrodes: Evidence of surface diffusion and spillover," Journal of the
American Chemical Society, vol. 131, pp. 14756-14760, 2009.
[10] E. M. Davis, N. M. Benetatos, W. F. Regnault, K. I. Winey, and Y. A. Elabd, "The
influence of thermal history on structure and water transport in Parylene C coatings,"
Polymer, vol. 52, pp. 5378-5386, 2011.

142
[11] J. Wolpaw and E. W. Wolpaw, Brain-computer interfaces: principles and practice:
OUP USA, 2012.
[12] E. R. Kandel, J. H. Schwartz, T. M. Jessell, S. A. Siegelbaum, and A. J. Hudspeth,
Principles of neural science vol. 4: McGraw-hill New York, 2000.
[13] H. Yasuda, Q. Yu, and M. Chen, "Interfacial factors in corrosion protection: an EIS
study of model systems," Progress in organic Coatings, vol. 41, pp. 273-279, 2001.
[14] X. Xie, L. Rieth, P. Tathireddy, and F. Solzbacher, "Long-term in-vivo investigation
of parylene-C as encapsulation material for neural interfaces," Procedia Engineering,
vol. 25, pp. 483-486, 2011.
[15] H. K. Yasuda, "Some important aspects of plasma polymerization," Plasma
Processes and Polymers, vol. 2, pp. 293-304, 2005.
[16] J. H.-C. Chang, B. Lu, and Y.-C. Tai, "Adhesion-enhancing surface treatments for
parylene deposition," in Solid-State Sensors, Actuators and Microsystems Conference
(TRANSDUCERS), 2011 16th International, 2011, pp. 390-393.
[17] M. M. De Luna, B. Chen, L. C. Bradley, R. Bhandia, and M. Gupta, "Solventless
grafting of functional polymer coatings onto Parylene C," Journal of Vacuum Science
& Technology A: Vacuum, Surfaces, and Films, vol. 34, p. 041403, 2016.
[18] C. H. Mastrangelo and G. Saloka, "A dry-release method based on polymer columns
for microstructure fabrication," in Micro Electro Mechanical Systems, 1993,
MEMS'93, Proceedings An Investigation of Micro Structures, Sensors, Actuators,
Machines and Systems. IEEE., 1993, pp. 77-81.
[19] T.-J. Yao, X. Yang, and Y.-C. Tai, "BrF 3 dry release technology for large
freestanding parylene microstructures and electrostatic actuators," Sensors and
Actuators A: Physical, vol. 97, pp. 771-775, 2002.
[20] S. A. Hara, B. J. Kim, J. T. Kuo, C. D. Lee, E. Meng, and V. Pikov, "Long-term
stability of intracortical recordings using perforated and arrayed Parylene sheath
electrodes," Journal of Neural Engineering, vol. 13, p. 066020, 2016.

143

Implantable medical devices make possible the treatment of chronic conditions by
providing continuous treatment at or near the affected area. BioMEMS technologies allow
for the development of complex, smaller, and versatile implantable devices. However,
reliable chronic performance in vivo has been limited by poor biocompatibility, mechanical
mismatch to the soft biological tissue, inappropriate packaging, and lack of combined
functionalities within a device (e.g. electrodes, drug delivery, sensors). Fortunately, these
challenges can be overcome by utilizing Parylene C as an encapsulation or structural
material. Parylene-based implantable devices hold promise due to high biocompatibility,
flexibility, good water barrier properties, and compatibility with standard MEMS
fabrication techniques. Presented in this work are two bioMEMS devices for clinical
applications designed to achieve reliable chronic performance in vivo.
Chapter 2 presented an implantable micropump that enables chronic drug
administration intended for evaluation and development of drug therapies in freely moving
small research animals such as rodents. A MEMS approach achieved miniature form factor
and completely wireless operation of the micropump. Low power-consumption
electrochemical pumps were developed with polymer microfabrication techniques and
characterized for a model anti-cancer application. A custom-made normally closed spring-
loaded ball check valve was designed and significantly decreased reverse flow. Micropumps
delivered an assortment of liquid drug formulations with a range of dose volumes and flow
rates (µL/min) suitable for drug administration in the treatment of chronic conditions.
Chapter 3 presented a novel polymer-based peripheral nerve interface for restoring
sensory and fine motor control in amputees. The interface combines both electrodes and

CONCLUSION

144
Parylene microfluidic channels in an adjustable cuff sized to be initially tested in rat sciatic
nerves. Close-fitting of the cuff around the nerve ensures localized drug delivery of lysing
agents and neurotrophic factors to induce axonal sprouting from the fascicles closest to the
nerve surface towards the embedded electrodes. The multifunctional approach of the cuff
electrode with integrated drug delivery system allows recording and stimulation of
peripheral nerves and aims to enhance implant-tissue integration in vivo and prolong
chronic performance beyond the current state-of-the-art. Lyse-and-Attract Cuff Electrodes
(LACE) were batch fabricated using standard Parylene micromachining processes. Several
challenges were overcome including the fabrication methods necessary to develop a thin-
film peripheral nerve cuff electrode with an integrated drug delivery system. Electrical and
fluidic packaging were designed for the acute and chronic in vivo implantation of the
LACE. Fully functional cuffs demonstrated controlled microfluidic infusion at low flow
rates, localized fluid delivery, high charge storage capacities, and low electrode
impedances which are necessary for achieving a successful chronic nerve interface.
Chapter 4 focused on reliability testing and characterization of the electrical,
fluidic, and mechanical functions of the LACE under simulated and acute in vivo
conditions in preparation for chronic in vivo implementation. Electrical and
electrochemical testing was conducted to access the integrity of the insulation and
electrodes. Cross-talk testing demonstrated adequate electrical insulation (20+ days) in
annealed devices. However, application of current in a saline solution, a feature of
electrochemical impedance spectroscopy testing, resulted in compromised insulation
within 2 days of soaking for cross-talk measurements and after 2 weeks of soaking for EIS.
Further investigation is necessary to understand the mechanisms of accelerated insulation
failure and electrode delamination, and evaluate strategies to improve barrier properties
and layer adhesion in multi-layer Parylene-based devices. The microfluidic channels were
redesigned to improve mechanical robustness of the thin-film channels and maintain device
integrity in vivo. Improved fluidic channel integrity during deformation allowed fluid flow
through all channels and localized fluid delivery. The interlocking mechanism was simple
to implement in vivo and maintained a secured lock around the rat sciatic nerve without

145
inducing tissue damage. Acute in vivo infusion tests proved targeted drug delivery
demonstrating the potential for precise axonal sprouting from multiple fascicles within the
nerve. While initial results are promising, the safety and efficacy of the LACE still needs
to be validated in vivo through delivery of lysing and neurotrophic factors,
electrophysiological testing, and verification of axonal sprouting through
immunohistochemistry and ultrastructural microscopy. Successful neural integration and
reliable electrophysiological performance of the LACE under chronic conditions will help
realize the promise of restoring sensory and fine motor control in amputees.
The aim of this work was to develop multifunctional Parylene-based implantable
technologies with reliable chronic performance to treat critical medical challenges
otherwise unresolvable with current technologies. It is my hope that the work presented
here will inspire the use of bioMEMS technologies and polymer materials to develop
sophisticated multifunctional implantable devices that are key prerequisites in the
development of future tailored drug delivery systems and neural implants.

146
APPENDIX A Pediatric Drug Delivery Pump for the
Treatment of Leptomeningeal Metastases
Medical Application
Leptomeningeal metastases (LM) is a deadly secondary cancer that affects the
membrane surrounding the central nervous system (meninges and spinal cord) [1, 2]. This
type of cancer is mainly encountered in children with approximately 1600 children being
diagnosed yearly. Untreated pediatric patients have a median survival time of three weeks
after diagnosis and with current treatment, six months. LM remains clinically difficult to
treat [3]. Systemic chemotherapy delivery is a potential treatment but it is limited by the
blood-brain barrier (BBB) and induces severe systemic side effects for marginal efficacy
due to low therapeutic drug concentration. Intrathecal (IT) chemotherapy by lumbar
puncture has suboptimal efficacy partially because it is not feasible to administer IT at high
injection frequency to sustain therapeutic concentration and has high risk of infection.
Effective localized chemotherapy and improved clinical outcomes can be achieved using
metronomic IT drug infusion pumps. However, there is currently no approved implantable
pediatric pump. Previously, we have successfully developed wirelessly-operated
micropumping technology suitable for use in mouse. Now we will demonstrate scaling up
of the electrochemical actuator to develop pumps suitable for use in pediatric patients.
Approach
Previous efforts in our laboratory led to the use of a sacrificial lost wax-like process
to fabricate the bellows. Here, we developed new processes that incorporate injection
molding and casting to achieve high-throughput fabrication and decrease variability among
bellows. Thin film polymer bellows have complex geometry and highly nonlinear behavior,
not adequately described by an analytical closed-form solution [4]. Instead, finite element
modeling (FEM) can be employed to model deflection and stress of complex polymer
structures under various loads. FEM and simulations of various designs provide insight
into the bellows design parameters and their relationship to load-deflection performance.
Finally, mechanical characterization of bellows and electrochemical actuators was

147
performed to demonstrate that our technology is capable of scaling to clinically-relevant
drug volumes (5 and 10 cc pump actuator).
Fabrication
Parylene bellows were fabricated using a lost-wax two-part molding process, where
a mold of the bellows is constructed out of a soluble material and then coated with 15 µm United States
Pharmacopeia (USP) class VI Parylene C (Specialty Coating Systems, Indianapolis, IN). Once coated,
the bellows were soaked in deionized water to dissolve the sacrificial mold. Bellows design
features for the LM application are limited by the desire pump dimensions (19 mm H × 86
mm ID). The bellows designs consisted of three convolutions each having a 16 mm inner
diameter, 28 mm outer diameter (design 1) or 23 mm inner diameter, 36 mm outer diameter
(design 2), 0.5 mm height and 15 m thickness (Figure. A-1).

Figure. A-1. Standard profile of a bellows with design parameters labeled. 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 [5].

Three different methods were developed to fabricate the soluble bellows molds.
Method 1 was developed by Gensler, et al. [5] and presented in Chapter 2. Methods 2 and
3 focus on the fabrication of hollow bellows sacrificial molds using injection molding and
casting techniques, respectively. These fabrication methods can accelerate the fabrication
process and reduce variability among bellows. A custom-made multipurpose aluminum
(Al) lost-wax bellows mold was designed and precision machined with the collaboration
of Jeffrey Field Design, Inc. (Figure. A-2). This mold was designed to inject or cast large
bellows (design 1). It is important to find materials compatibles with these fabrication
techniques and at the same time soluble in water or mild solvents in order to be incorporated
into our molding process. Several soluble materials were tested by injecting or gravity

148
casting into the mold. Materials tested include: freeman optical soluble wax (Freeman
Manufacturing & Supply Company, Avon, OH), riace hydroresin water-soluble injection
wax (Rio Grande, Albuquerque, NM), water-soluble injection wax (Rio Grande,
Albuquerque, NM), parowax household wax (Royal Oak Enterprises, LLC., Roswell, GA),
and polyethylene glycol (PEG; mw 1000). Fabrication methods and materials presented
below were optimized to obtained clean wax release from the mold and complete filling of
the bellows fins.

Figure. A-2. Custom-made multipurpose aluminum lost-wax bellows mold. (a) Solidworks
drawing showing a cross-sectional view of the mold, (b) injection molding mold consists
of part A and B, and (c) casting mold consists of part A only.
Method 1 – PDMS sheets mold
Polydimethylsiloxane (PDMS, Sylgard 184; Dow Corning Corp., Midland, MI)
sheets (0.5 mm thick) were perforated with 16 mm, 28 mm, 23 mm, 36 mm holes to mold
the inner and outer diameters of the two different designs. The sheets were visually aligned
and stacked together on glass slides to form mold modules. These modules were filled
with molten (50°C) PEG (mw 1000). Once the PEG solidified, PDMS sheets were peeled.
The modules were stacked and fused together by moistening connecting interfaces with
water as shown in Figure. A-3.

149

Figure. A-3. Photographs of the (a-b) design 1 and (c-d) design 2 PEG bellows prior to
sacrificial Parylene C coating.

Figure. A-4. Photograph of design 1 and 2 Parylene C bellows after sacrificial PEG
removal.
Method 2 – injection molding
An AB-100 manual plastic injector (A.B. Plastic Injectors, Canada) was used to
inject the riace hydroresin water-soluble injection wax (Rio Grande, Albuquerque, NM)
into the Al mold. During initial runs, oxygen was entrapped in the resins during injection
and curing resulting in incomplete mold filling and cavities within the mold. A degassing
method was developed prior to injecting the resin since the AB injector does not have a
material degassing feature (e.g. vacuum). The hydroresin was placed in a glass beaker,
covered, and heated at 90 C in the oven for 1 hour. The beaker was covered with aluminum
foil to minimize the effects of evaporation during resin heating. The covered beaker was

150
then transferred to a hot plate set to 120 C next to the injection molder to prevent the resin
from solidifying. The liquid hydroresin was poured directly into the injection molding
funnel using a plastic dropper until the barrel was filled. The hydroresin was injected into
the multipurpose Al mold at a pressure of 10 psi and a temperature of 75 C. The bellows
mold was released from the Al mold after a 5 min cool down period. The IM process is
summarized in Figure. A-5. This method yielded complete and reliable bellows molds with
automatic filling of the Al mold. A higher bellows molds throughput (~ 1 hour/bellows
mold) was achieved with this method as it requires less material preparation time (Figure.
A-6).

Figure. A-5. Schematic diagram of injection molding process for hydroresin.

Figure. A-6. Photographs of hydroresin bellows mold fabricated by injection molding. (a)
Side view and (b) top view.
Method 3 - casting

151
Several soluble materials were tested by gravity casting into the Al mold. Best
bellows molds were obtained with a 5 parts PEG-1000 and 2 parts hydroresin mixture. The
5:2 PEG-1000:hydroresin composite was placed in a glass beaker, covered, and heated at
70C in the oven for 2 hours. The Al mold was cleaned with DI water, dried, and brushed
with cutting fluid to help release the bellows mold. The Al mold was also heated for 10
min to delay the material cure time and allow a complete filling of the mold. Inside the
oven, the liquid mixture was transferred to the Al mold with a plastic dropper until the
liquid resin was flushed with the top of the Al mold. A custom-made acrylic slide with a
plug was placed over and screwed into the Al mold with the plug inserting into the liquid
resin (Figure. A-7). This acrylic jig aided in the release of the bellows mold. The mixture
was then cooled down for 10 min at - 4C. Once cooled, the fixture was placed in the
acrylic release frame where the Al mold halves were separated leaving the bellows mold
attached to the acrylic slide. A summary of the casting method for 5:2 PEG-
1000:Hydroresin is shown in Figure. A-8. The casting method yielded complete and
reliable bellows molds (Figure. A-9), but the process proved to be more time consuming
(~ 3 hour/bellows mold) than IM.

Figure. A-7. Machined acrylic release jig consisting of a slide and a frame.

152

Figure. A-8. Schematic diagram of simplified casting process for the 5:2 PEG-1000:
hydroresin composite.

Figure. A-9. Photographs of 5:2 PEG-1000 and hydroresin bellows mold fabricated by
casting. (a) Side view and (b) top view.
Experimental Methods
Finite element models
Three-dimensional finite element models (FEM) were developed for nonlinear
static simulations (Solidworks Simulation 2014, Dassault Systèmes SolidWorks Corp.,
Concord, MA) of different bellows designs to determine which could displace a minimum
of 5 mL of fluid contained within the adjacent drug reservoir without experiencing plastic
deformation. Parylene C material properties used in the FEM simulations are shown in
Table. A-1. Quarter models of the symmetrical bellows were used to minimize FEM
simulations 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
0.01 to 0.15 psi (0.069 to 1 kPa) were applied and the resulting deflection and von Mises
stress values were recorded.

153
Table. A-1. Material properties for Parylene C used in the FEM simulations [5].
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

Mechanical characterization
Bellows were clamped in a custom acrylic test fixture connected to a custom
pressure setup (Figure. A-10). Loads were applied at room temperature to the bellows
mounted in the fixture with pressurized gas and deflection of the center of the top of the
bellows was recorded using a CCD camera. For mechanical characterization in the elastic
range, loads from 0.01 to 0.35 psi (0.069 to 2.4 kPa) were applied in discrete steps of 0.05
psi (0.34 kPa) and the resulting bellows deflection was calculated by measuring bellows
displacement with a ruler.

Figure. A-10. Schematic overview of pressure testing setup for load-deflection testing of
the bellows.
Bellows demonstration in an electrochemical actuator
Bellows were integrated with an electrochemical (EC) actuator as described in
Chapter 2. The bellows EC actuator was placed within custom machined acrylic reservoirs
during maximum dose volume testing at room temperature. A constant current of 0.5 mA

154
was applied (2400 Sourcemeter; Keithley Instruments Inc., Cleveland, OH) to the
electrodes to induce electrolysis within the bellows. This deflects the bellows and displaces
the surrounding fluid out of the rigid reservoir through a PEEK tubing into a collection
reservoir. The experimental setup is shown in Figure. A-11. After 24 hours of continuous
actuation, the water in the collection reservoir was weighed to determine the maximum
dose volume of the bellows.

Figure. A-11. Electrochemical actuators mounted in a reservoir for maximum dose volume
testing.
Results and Discussion
Finite element models
FEM simulations showed that the two bellows designs do not exceed the yield
strength of Parylene C (59 MPa [6]) when a 0.1 psi pressure was applied (Figure. A-12).
The maximum stresses were observed in the 90-degree corners. Simulations showed that
increasing the outer diameter of the bellows from 28 mm to 36 mm resulted in an upwards
shift in the load- displacement curve and the stresses observed.

155

Figure. A-12. Finite element model simulation of the (a-b) von Mises stress (maximum
value – Design 1: 32.1 MPa, Design 2: 50.2MPa), and (c) displacement for bellows designs
1 and 2 with a 0.1 psi load.
Mechanical characterization
FEM simulations of deflection for design 1 bellows were slightly larger than the
mechanical tetsing results, but within an order of magnitude (Figure. A-13). Design 2
mechanical results were noticeably underestimated by the simulation. No plastic
deformation was observed in either design up to a load of 0.35 psi. Mechanical testing
predicts that design 1 bellows will deliver more than 6 mL dosed volume without exceeding
the 14 mm maximum displacement height with an applied pressure of 0.05 psi.
Consequently, only design 1 bellows were tested with the EC actuator. Variability in the
characterization results was attributed to the manual fabrication process of the bellows
(Method 1).

156

Figure. A-13. Mechanical characterization of design 1 and 2.
Bellows demonstration in an electrochemical actuator
EC actuator test demonstrated the capability of design 1 bellows to deliver a
maximum dosed volume of 4.9 mL after a 24 hours period of continuous actuation with a
0.5 mA applied current. Bellows were fully extended at the end of the experiments with no
water remaining inside the bellow. Thus, no recombination of gases into liquid was
observed.
Conclusion
Through mechanical characterization and finite element simulations the dimensions
favorable for the leptomeningeal metastases intrathecal chemotherapy delivery application
were determined. Large diameter bellows were successfully designed, modeled, fabricated,
and characterized. We developed reliable and repeatable fabrication methods for the
soluble bellows molds. Bellows for electrolysis-based actuation displaced the desired
volume required for LM treatment. Implantable micropump packaging was designed and

157
prototyped through stereolithography (SL) in USP class VI DSM Somos® WaterShed XC
11122 material as shown in Figure. A-14 (FineLine Prototyping, Inc., Raleigh, NC).

Figure. A-14. Prototyped pediatric drug pump.
However, the scaled up bellows were only able to deliver the required daily dose
volume once, limiting their use for an application that requires chronic dosing. A new drug
delivery system was developed and preliminary characterization was performed. The
system consists of design 2 bellows as a collapsible drug reservoir, a modified Fluid
Synchronic implantable drug delivery micropump to drive the drug out of the reservoir to
the target tissue, and two spring-loaded check valves for fluid regulation (Figure. A-15).
Initial characterization of the system, demonstrated it was capable of providing a dosed
volume of almost 4 mL per day with a flow rate of 3 L/min. This initial results show great
potential of the new system for the chronic treatment of LM in pediatric patients.

Figure. A-15. New drug delivery system for the treatment of leptomeningeal malignancy
in pediatric patients.

158
References
[1] D. Gangji, "Treatment of leptomeningeal metastases," in Management in Neuro-
Oncology, ed: Springer, 1992, pp. 41-62.
[2] M. C. Chamberlain, "Topical Review: A Review of Leptomeningeal Metastases in
Pediatrics," Journal of child neurology, vol. 10, pp. 191-199, 1995.
[3] M. C. Chamberlain, "New approaches to and current treatment of leptomeningeal
metastases," Current opinion in neurology, vol. 7, pp. 492-500, 1994.
[4] A. C. Ugural, Stresses in beams, plates, and shells: CRC Press, 2009.
[5] H. Gensler and E. Meng, "Rapid fabrication and characterization of MEMS Parylene
C bellows for large deflection applications," Journal of Micromechanics and
Microengineering, vol. 22, p. 115031, 2012.
[6] C. Shih, T. A. Harder, and Y.-C. Tai, "Yield strength of thin-film parylene-C,"
Microsystem Technologies, vol. 10, pp. 407-411, 2004.

159
APPENDIX B Lyse-and-Attract Cuff Electrode Sham Device
Process Flow

Figure. B-1. Sham devices masks transparency.

1. Bake clean 4” silicon wafer to remove moisture 110 °C, > 10 mins

2. Deposit Parylene (10 µm)

3. Pattern AZ 4620 (double layer) sacrificial channel mold (20.1 µm thick) (Mask 1 –
Sac PR)
LAYER 1
Pre spin 5 sec, 500 rpm
Spin 45sec, 1700 rpm
Softbake 90 °C, 6 minutes
LAYER 2
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1900 rpm
Softbake 90 °C, 7 minutes
Hydration 60 minutes
Exposure 600 mJ/cm
2
(20 mW/cm
2
, 30 sec)
Development 3 minutes using two baths
Hard bake 1 90 °C, 15 minutes, hotplate
Hard bake 2 90 °C, 15 minutes, under vacuum

4. Descum, O2 plasma 100 W, 100 mTorr, 1 min

5. Deposit Parylene (8 µm)

160

6. Pattern AZ 4620 (double layer) etch mask (32 µm thick) (Mask 2 – Outline Release)
LAYER 1
Pre spin 5 sec, 500 rpm
Spin 45sec, 1000 rpm
Softbake 90 °C, 12 minutes
LAYER 2
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1000 rpm
Softbake 90 °C, 14 minutes
Hydration 60 minutes
Exposure 750 mJ/cm
2
(20 mW/cm
2
, 37.5 sec)
Development 3.5 minutes using two baths
Hard bake 1 100 °C, 20 minutes, hotplate
Hard bake 2 100 °C, 15 minutes, under vacuum

7. Deep Reactive Ion Etching 115 loops, rotate wafer 90° every 25
(Oxygen plasma) loops

Table. B-1. DRIE etch and deposition parameters for each loop.
Parameter Deposition Etch
ICP Power (W) 700 700
RF Power (W) 10 20
O 2 (ccm) 1 60
C 4F 8 (ccm) 35 1
Ar (ccm) 40 40
SF 6 (ccm) 0 0

Pressure (mTorr) 23 23

Time (s) 3 10

8. Strip remaining photoresist mask with acetone, IPA, and DI water

9. Pattern AZ 4620 (double layer) etch mask (32 µm thick) (Mask 3 – Port Etch)
LAYER 1
Pre spin 5 sec, 500 rpm
Spin 45sec, 1000 rpm

161
Softbake 90 °C, 12 minutes
LAYER 2
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1000 rpm
Softbake 90 °C, 14 minutes
Hydration 60 minutes
Exposure 750 mJ/cm
2
(20 mW/cm
2
, 37.5 sec)
Development 3.5 minutes using two baths
Hard bake 1 100 °C, 20 minutes, hotplate
Hard bake 2 100 °C, 15 minutes, under vacuum

10. Deep Reactive Ion Etching 115 loops, rotate wafer 90° every 25
(Oxygen plasma) loops

Release Clean surface with acetone and IPA.
Peel carefully while immersed in water

162
APPENDIX C Lyse-and-Attract Cuff Electrode Complete
Device Process Flow

Figure. C-1. Complete LACE devices photomasks transparency.

1. Bake clean 4” silicon wafer to remove moisture 110 °C, > 10 mins

2. Deposit Parylene (10 µm)

3. Pattern AZ 5214-IR for lift-off (2 µm thick) (Mask 1 - Metal)
Pre spin 8 sec, 500 rpm
Spin 45 sec, 1800 rpm
Softbake 90 °C , 70 seconds
Exposure 37.5 mJ/cm
2
(25 mW/cm
2
, 1.5 sec)
IR bake 110 °C, 55 sec
Hydration 3 minutes
Global exposure 1000 mJ/cm
2
(25 mW/cm
2
, 40
seconds)
Development (AZ 351 1:4 dilution) 18 seconds

163
4. Descum, O2 plasma 100 W, 100 mTorr, 1 min

5. Metal deposition (Pt) 2000 Å (in 4 runs of 500 Å )

6. Lift-off in acetone (gentle scrub if necessary) In warm acetone 50 °C

7. Descum, O2 plasma 100 W, 100 mTorr, 1 min

8. Deposit Parylene (10 µm)

9. Pattern AZ 4620 etch mask (12.8 – 13.5 µm thick) (Mask 2 – Insulation Cutout)
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1000 rpm
Softbake 90 °C, 5 minutes
Hydration 45 minutes
Exposure 550 mJ/cm
2
(20 mW/cm
2
, 27.5 sec)
Development 1.5 minutes
Hard bake 1 90 °C, 15 minutes, hotplate
Hard bake 2 90 °C, 15 minutes, under vacuum

10. Deep Reactive Ion Etching 100 loops, rotate wafer 90° every 25
(Oxygen plasma) loops

Table. C-1. DRIE etch and deposition parameters for each loop
Parameter Deposition Etch
ICP Power (W) 700 700
RF Power (W) 10 20
O 2 (ccm) 1 60
C 4F 8 (ccm) 35 1
Ar (ccm) 40 40
SF 6 (ccm) 0 0

Pressure (mTorr) 23 23

Time (s) 3 10

11. Strip remaining photoresist mas with acetone, IPA, and DI water

164
12. Pattern AZ 4620 (double layer) sacrificial channel mold (20.1 µm thick) (Mask 3 –
Sac PR)
LAYER 1
Pre spin 5 sec, 500 rpm
Spin 45sec, 1700 rpm
Softbake 90 °C, 6 minutes
LAYER 2
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1900 rpm
Softbake 90 °C, 7 minutes
Hydration 60 minutes
Exposure 600 mJ/cm
2
(20 mW/cm
2
, 30 sec)
Development 3 minutes using two baths
Hard bake 1 90 °C, 15 minutes, hotplate
Hard bake 2 90 °C, 15 minutes, under vacuum

13. Descum, O2 plasma 100 W, 100 mTorr, 1 min

14. Deposit Parylene (8 µm)

15. Pattern AZ 4620 (double layer) etch mask (32 µm thick) (Mask 4 – Pads Etch)
LAYER 1
Pre spin 5 sec, 500 rpm
Spin 45sec, 1000 rpm
Softbake 90 °C, 12 minutes
LAYER 2
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1000 rpm
Softbake 90 °C, 14 minutes
Hydration 60 minutes
Exposure 750 mJ/cm
2
(20 mW/cm
2
, 37.5 sec)
Development 3.5 minutes using two baths
Hard bake 1 100 °C, 20 minutes, hotplate
Hard bake 2 100 °C, 15 minutes, under vacuum

16. Deep Reactive Ion Etching 140 loops, rotate wafer 90° every 25
(Oxygen plasma) loops

165
17. Strip remaining photoresist mas with acetone, IPA, and DI water

18. Pattern AZ 4620 (double layer) etch mask (32 µm thick) (Mask 5 – Ports Etch)
LAYER 1
Pre spin 5 sec, 500 rpm
Spin 45sec, 1000 rpm
Softbake 90 °C, 12 minutes
LAYER 2
Pre spin 5 sec, 500 rpm
Spin 45 sec, 1000 rpm
Softbake 90 °C, 14 minutes
Hydration 60 minutes
Exposure 750 mJ/cm
2
(20 mW/cm
2
, 37.5 sec)
Development 3.5 minutes using two baths
Hard bake 1 100 °C, 20 minutes, hotplate
Hard bake 2 100 °C, 15 minutes, under vacuum

19. Deep Reactive Ion Etching 140 loops, rotate wafer 90° every 25
(Oxygen plasma) loops

Release Clean surface with acetone and IPA.
Peel carefully while immersed in water

166
APPENDIX D LACE Fabrication Batch
Several runs were completed to fabricate LACE devices for various experiments.
Major fabrication processes such as metal deposition and/or oxygen plasma etching
changed among the runs due to process optimization or challenges encountered. Parylene
films were visually inspected to identify any film anomalies. The table below presents the
runs from which devices were chosen for electrochemical and electrical testing:
Table. D-1. LACE fabrication runs from which devices were chosen for electrochemical
and electrical testing. The table highlights the type of metal deposition and Parylene
etching, as well as Parylene film anomalies.
Fabrication Metal
deposition
Insulation
etch
Release etch Parylene film
anomalies
Experimental
results
Batch 1 Sputter DRIE: 20 W
RF power
DRIE: 20 W
RF power
Milky Figure 3-38,
Figure 3-50, &
Table 3-5
Batch 2 Sputter DRIE: 80 W
RF power
RIE: 100 W
power
Milky - mild Figure 3-52 &
Figure 3-53
Batch 3 Sputter RIE: 100 W
power
RIE: 100 W
power
Milky - mild Figure 4-4
Batch 4 E-beam RIE: 100 W
power
RIE: 100 W
power
None Figure 4-5 to
Figure 4-14

167
APPENDIX E Sacrificial Photoresist Removal Post-Oxygen
Plasma Exposure

Several fixtures were designed and fabricated to dissolve the sacrificial layer of
photoresist which defined the microfluidic channel interior. The fixtures accommodate
multiple devices while submersed in a heated beaker of agitated acetone
Fixture 1
The fixture consisted of 4 computer numerical controlled (CNC) milled
Polypropylene sheets (each 0.5 in thick), and corrosion-resistant stainless steel mesh and
screws (Figure. E-1). Initial testing showed that PR dissolution was achieved in less than
30 hours of soaking. During testing, we found that devices were scratched when trying to
remove them from the fixture to check if the PR was dissolved. Therefore, we modified
the fixture to improve ease of device removal from fixture.

Figure. E-1. Sacrificial photoresist removal fixture. SolidWorks diagram showing (a) top
view and (b) bottom view. (c) Machined Polypropylene fixture.

Fixture 2
As shown in Figure. E-2a, the second design for the acetone soaking fixture
involved placing each device in a spherical stainless steel tea infuser and placing these
infusers on top of the CNC milled polypropylene sheet in the acetone beaker to keep them
from interfering with the magnetic stir bar. As demonstrated in Figure. E-2b-c, this method

168
tended to scratch the devices as they brushed against the inside of the tea infusers and
required an unsatisfactory soak time of approximately 48 hours (40 ºC, 250 rpm).

Figure. E-2. (a) The second acetone soak fixture involved placing each device in a spherical
stainless steel tea infuser (tea ball) above a polypropylene sheet. Before (b) and after (c)
photographs of a device that was soaked via the fixture 2 showing that the tea infuser
resulted in significant abrasions to the devices.

169
APPENDIX F Removal of Oxygen Plasma Exposed
Photoresist

RCA-1
RCA-1 cleaned was performed on one of the released LACE devices. RCA-1
consists of 5 parts water (H2O), 1 part 27% ammonium hydroxide (NH4OH), and 1 part
30% hydrogen peroxide (H2O2). H2O and NH4OH were combined and heated to 70 °C on
a hotplate. H2SO4 was added to the solution once the temperature was reached. The LACE
device was then soaked in the solution for 15 minutes but the photoresist was not removed.
Piranha
A more aggressive approach was used by cleaning a device with a cool piranha
solution consisting of: 4 parts sulfuric acid (H2SO4) and 1 part 30% hydrogen peroxide
(H2O2). Upon mixing the solution, an exothermic reaction occurs which quickly raises the
solution temperature to above 100 °C. Soaking the LACE device in a piranha solution at
temperature above 60 °C resulted in metal delamination. Consequently, devices were
soaked only after the solution cooled down to 40 °C and for an average of 3 minutes. This
method successfully removed all of the photoresist residue as shown on Figure. F-1.
Although piranha was successful, we believe that it is too aggressive for our devices.

Figure. F-1. Piranha cleaned device shows no photoresist residue.

Remover PG
We also tried a N-methyl-2-pyrrolidone (NMP) based stripper such as Remover PG
to remove the photoresist residue on other LACE devices. Some of the photoresist was
removed after a 1 hour soak of a LACE device in Remover PG at 60 °C. The temperature

170
was increased to 80 °C as suggested by the photoresist supplier and a 30 min soak removed
all of the remaining photoresist. No metal delamination was observed at this temperature,
but the device curled significantly while retaining its flexibility (Figure. F-2).

Figure. F-2. Curled LACE device after soaking in in Remover PG at 80 °C.

AZ KWIK Stripper
We tried a mild commercial stripper use to remove oxygen plasma exposed
photoresist. It has a pH neutral and is metal safe. The solution consists of 59% dipropylene
glycol monomethyl ether and 34% tetrahydrothiophene-1,1-dioxide. A LACE device was
soaked in the solution for 60 minutes at 60 °C, but the photoresist was not removed.

171
APPENDIX G Electrical Cross-talk System and Testing
Protocol

System and algorithm
The cross-talk system uses LabVIEW graphical user interface (GUI) to control a
function generator, oscilloscope, DC power source, and digital outputs, all of which are
contained in a NI VirtualBench tool (Figure. F- 1). Also part of the system are two 8-
channel relay modules. The complete crosstalk system is shown in Figure. F- 2. The send
module is used to route a 1 kHz sinusoidal signal from the signal generator to a selected
electrode on the LACE, while the read module simultaneously routes the signal of an
adjacent electrode to the oscilloscope for sampling. In order to accommodate the NI
VirtualBench’s limit of only eight digital outputs, the relays of each module were cascaded
to form an 8-to-1 multiplexer, thus requiring a total of only six digital control lines.

Figure. F- 1. Diagram of the automated cross-talk testing setup.
NI VirtualBench
LabVIEW
Function
Generator
Oscilloscope
DC Power
Digital I/O
Send
Mod
ule
Read
Mod
ule
LACE

172

Figure. F- 2. The cross-talk system consists of a computer with LabVIEW software, NI
VirtualBench tool, and two 8-channel relay modules.
The LabVIEW script obtains the cross-talk measurements by first driving an
electrode A while sampling simultaneously from all electrodes, then driving B while
sampling from all electrodes and so on. Cross-talk between each pair of electrodes n is then
understood as the ratio Vread,n /Vsend, where V is the amplitude at 1 kHz.

𝑐𝑟𝑜𝑠𝑠𝑡𝑎𝑙𝑘 𝑛 =
𝑉 𝑟𝑒𝑎𝑑 ,𝑛 𝑉 𝑠𝑒𝑛𝑑 ×100

Baseline cross-talk of the system itself is approximately 8%. To calibrate, the
baseline measurement is collected before each testing session with an open system (no
device connected). The calibration values are obtained with:

𝑝 𝑛 =
𝑉 𝑟𝑒𝑎𝑑 ,𝑛 ,ℎ𝑎𝑟𝑑𝑤𝑎𝑟𝑒 𝑉 𝑠𝑒𝑛𝑑

173
The calibrated cross-talk values are then calculated as:

𝑐𝑟𝑜𝑠𝑠𝑡𝑎𝑙𝑘 𝑛 =
𝑉 𝑟𝑒𝑎𝑑 ,𝑛 − 𝑝 𝑛 (𝑉 𝑠𝑒𝑛𝑑 )
𝑉 𝑠𝑒𝑛𝑑 (1 − 𝑝 𝑛 )
×100

The LabVIEW program displays the cross-talk values as shown in Figure. F- 3.

Figure. F- 3. LabVIEW user interface for the LACE cross-talk testing. Calibrated cross-
talk values are displayed.

Protocol
1. Open file “XTalk.lvproj”.
2. Open “Main.vi” from the project explorer in LabVIEW.
3. In the cross-talk LabVIEW interface, type in device information in “Device ID”,
and select stimulation voltage and file path.

174
4. Click on “Calibrate” without a device connected, you should hear clicking of the
relays as the test runs. The calibration is only necessary for the first device to be
tested.
5. Connect device to cross-talk system wiring connection as shown in Figure. F- 4.
6. Click on “Get Crosstalk Data”, you should hear clicking of the relays as the test
runs.
7. Click “Save Data” to export the data to a spreadsheet and save the waveforms.
8. Continue testing more devices, or click “Exit” to finish.

Figure. F- 4. LACE device connection to the cross-talk system

Abstract (if available)

Abstract Implantable medical devices make possible the treatment of chronic conditions by providing continuous treatment at or near the affected area. BioMEMS technologies allow for the development of complex, smaller, and versatile implantable devices. However, reliable chronic performance in vivo has been limited by poor biocompatibility, mechanical mismatch to the soft biological tissue, inappropriate packaging, and lack of combined functionalities within a device (e.g. electrodes, drug delivery, sensors). Fortunately, these challenges can be overcome by utilizing Parylene C as an encapsulation or structural material. Parylene-based implantable devices hold promise due to high biocompatibility, flexibility, good water barrier properties, and compatibility with standard MEMS fabrication techniques. Presented in this work are two bioMEMS devices for clinical applications designed to achieve reliable chronic performance in vivo. ❧ Human drug therapies are initially evaluated in small animal models such as rodents

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Development of implantable Parylene-based MEMS technologies for cortical applications

PDF

Parylene C bioMEMS for implantable devices with electrochemical interfaces

PDF

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

PDF

Parylene-based biomems sensors for multiple physiological systems

PDF

Development of fabrication technologies for robust Parylene medical implants

PDF

Fabrication and packaging of three-dimensional Parylene C neural interfaces

PDF

The electrochemical evaluation of Parylene-based electrodes for neural applications

PDF

Development of micromachined technologies for neural interfaces

PDF

A percutaneously implantable wireless neurostimulator for treatment of stress urinary incontinence

PDF

Next generation neural interfaces for neural modulation and prosthesis

PDF

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

PDF

Towards a high resolution retinal implant

PDF

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

Asset Metadata

Creator Cobo, Angelica Maria (author)

Core Title Parylene-based implantable interfaces for biomedical applications

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 05/09/2018

Defense Date 07/25/2017

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

Tag check valve,drug delivery,electrochemical bellows actuator,flexible cuff electrode,implantable micropump,MEMS,microchannels,neural interface,neuroprosthese,OAI-PMH Harvest,parylene C, electrode,peripheral nerve,polymers,wireless powering

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Meng, Ellis (committee chair), D'Argenio, David Z. (committee member), Gupta, Malancha (committee member)

Creator Email acobo@usc.edu,acobo001@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-452886

Unique identifier UC11263871

Identifier etd-CoboAngeli-5883.pdf (filename),usctheses-c40-452886 (legacy record id)

Legacy Identifier etd-CoboAngeli-5883.pdf

Dmrecord 452886

Document Type Dissertation

Rights Cobo, Angelica Maria

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