A MICROFABRICATED

DEEP BRAIN STIMULATION ELECTRODE

by

CHIA-HUA LIN

Submitted in partial fulfillment of the requirements

For the degree of Master of Science

Thesis Advisor: Prof. Mehran Mehregany

Department of Electrical Engineering and Computer Science

CASE WESTERN RESERVE UNIVERSITY

August, 2009 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

______ candidate for the ______degree *.

(signed)______. (chair of the committee)

______

______

______

______

______

(date) ______.

*We also certify that written approval has been obtained for any proprietary material contained therein. Table of Contents

CHAPTER 1: INTRODUCTION ...... 1

1.1 BACKGROUND ...... 1

1.2 THESIS OVERVIEW ...... 8

CHAPTER 2: MICROFABRICATED DBS ELECTRODE ...... 9

2.1 ELECTRODE ARCHITECTURE...... 9

2.2 ELECTRODE FABRICATION PROCESS ...... 11

2.3 ASSEMBLY PROCESS ...... 15

CHAPTER 3: EXTENSION LEAD ...... 21

3.1 ATTACHMENT METHOD ASSESSMENT ...... 21

3.1.1 Wedge/ball Bonding ...... 21

3.1.2 Laser Spot Welding ...... 22

3.1.3 Manual Epoxy Bonding ...... 25

3.1.4 Conclusions ...... 26

3.2 MICROFLEX EXTENSION LEAD ...... 27

3.2.1 Introduction ...... 27

3.2.2 Flexible Lead Design and Fabrication Process ...... 28

3.2.3 Attachment Process ...... 30

CHAPTER 4: ELECTRICAL AND MECHANICAL PROPERTIES, AND IN-VITRO TESTING ...... 32

4.1 ELECTRICAL TESTING ...... 32

4.1.1 Introduction ...... 32

4.1.2 Test Process ...... 32

4.1.3 Data Analysis and Comparison with Design ...... 35

4.1.4 Conclusion ...... 38

i

4.2 TESTING FOR THE MECHANICAL ROBUSTNESS OF LEAD ATTACHMENTS ...... 39

4.2.1 Introduction ...... 39

4.2.2 Test Environment and Preparation ...... 39

4.2.3 Test Process ...... 40

4.2.4 Data Analysis ...... 42

4.2.5 Conclusion ...... 44

4.3 ELECTRODE MODELING AND IN-VITRO TEST ...... 45

4.3.1 Electrode Modeling and Electric Potential Field Simulation ...... 45

4.3.2 In-vitro Test Environment ...... 46

4.3.3 In-vitro Test and Simulation Result and Discussion ...... 50

CHAPTER 5: CONCLUSION AND FUTURE WORK ...... 56

5.1 CONCLUSION ...... 56

5.2 FUTURE WORK ...... 57

APPENDIX 1: ELECTRODE FABRICATION PROCESS DETAILS ...... 58

REFERENCE ...... 64

ii

List of Tables

TABLE 4.1: THE MEASURED AND DESIGN RESISTANCES...... 35

TABLE 4.2: ESTIMATED EPOXY RESISTANCE...... 37

TABLE 4.3: RESISTANCES BETWEEN EACH CHANNEL...... 38

TABLE 4.4: THE DIAMETER OF THE CONTACT AREA BETWEEN THE EPOXY STUD AND THE

BOND PAD FOR THE SIX SAMPLES PULLED MANUALLY...... 44

TABLE A1.1: THE PR RECIPES USED IN THE PROCESS...... 58

TABLE A1.2: THE RECIPES FOR BOTTOM INSULATOR DEPOSITION...... 59

TABLE A1.3: THE RECIPES FOR METAL SPUTTERING...... 59

TABLE A1.4: THE RECIPE OF THE TOP TEOS LAYER DEPOSITION...... 60

TABLE A1.5: THE RECIPE FOR THE 3000 Å-THICK OXIDE ETCHING BY RIE...... 60

TABLE A1.6: THE RECIPE FOR THE 1500 Å-THICK NITRIDE ETCHING BY RIE...... 61

TABLE A1.7: PARAMETERS USED FOR DRIE ETCHING OF THE SIDE PIECES...... 62

iii

List of Figures

FIGURE 1.1: MODEL 7482A EXTENSION LEAD CONNECTED TO A MEDTRONIC, INC. IPG AND

THE INTRACRANIAL LEAD. (FROM MEDTRONIC, INC.’S DBS EXTENSION KIT

MANUAL [1.6]) ...... 2

FIGURE 1.2: IMPLANTED DBS SYSTEM MADE BY MEDTRONIC, INC. (FROM

WWW.MEDTRONIC.COM) ...... 2

FIGURE 1.3: THE DIMENSIONS OF DBS INTRACRANIAL LEAD MODEL 3389 AND 3387 (FROM

MEDTRONIC, INC.’S DBS LEAD KIT MANUAL [1.4]) ...... 3

FIGURE 1.4: OPTICAL PHOTO OF A DETACHED ELECTRODE CONTACT ON MEDTRONIC,

INC.’S DBS INTRACRANIAL LEAD. THE INSET SHOWS THE DISTAL ENDS OF

THE MEDTRONIC INTRACRANIAL LEAD “3387 (LEFT)” AND “3389 (RIGHT).” ..... 4

FIGURE 2.1: ELECTRODE DESIGNS: (A) FIRST; AND (B) SECOND, WHICH IS A REVISION OF

THE FIRST. METAL TRACES CONNECTING THE ELECTRODE SITE TO

CORRESPONDING BOND PADS ARE NOT SHOWN IN (A) AND (B). (C)

DIMENSIONS OF THE SIDE PIECES OF THE SECOND ELECTRODE DESIGN. .... 10

FIGURE 2.2: FABRICATION PROCESS FLOW FOR THE 3-D ELECTRODE SIDE PIECES (NOT

SHOWN TO SCALE): (A) SUBSTRATE ISOLATION OXIDE/NITRIDE/OXIDE LAYER

DEPOSITION; (B) TITANIUM AND PLATINUM METAL DEPOSITION; (C) TRACE

PASSIVATION OXIDE DEPOSITION; (D) OPENING OF WINDOWS IN THE TRACE

PASSIVATION LAYER TO EXPOSE THE ELECTRODE SITES AND BOND PADS; (E)

PATTERNING OF THE ISOLATION/PASSIVATION LAYERS TO EXPOSE THE SIDE

PIECE OUTLINE; AND (F) ETCHING OUT THE SIDE PIECE FROM THE WAFER. 12

FIGURE 2.3: PATTERN MISALIGNMENT IN PR DUE TO REPETITION OF MASK 3...... 14

FIGURE 2.4: OPTICAL PHOTO OF THE FABRICATED ELECTRODE PARTS: ELECTRODE SIDE

PIECES, DISTAL CAP, AND PROXIMAL CAP...... 15

iv

FIGURE 2.5: (A) OPTICAL PHOTO OF THE ELECTRODE ASSEMBLY JIG STRUCTURE

CONSTRUCTED FROM TWO IDENTICAL SILICON PIECES AND FOUR SPACERS.

(B) THE LAYOUT OF THE ALIGNMENT HOLE SET...... 17

FIGURE 2.6: THE ELECTRODE ASSEMBLY PROCESS FLOW USING A CUSTOM DESIGNED

JIG AND MICROMANIPULATORS: (A) FOUR ELECTRODE SIDE PIECES ARE

ALIGNED AND INSERTED INTO THE HOLES ON THE JIG; (B) A PROXIMAL CAP

AND A DISTAL CAP ARE MOUNTED ONTO THE ELECTRODE; AND (C) SILASTIC

IS APPLIED ONTO THE CAPS AND CURED FOR 24 HOURS AT ROOM

TEMPERATURE...... 18

FIGURE 2.7: OPTICAL PHOTO SHOWING ELECTRODE ASSEMBLY PROCESS FOR THE FIRST

ELECTRODE DESIGN...... 19

FIGURE 2.8: OPTICAL PHOTO OF THE IMPRESSION WITH AN ASSEMBLED ELECTRODE. .. 19

FIGURE 2.9: THE ELECTRODE ASSEMBLY PROCESS USING A SILASTIC IMPRESSION OF AN

ALREADY ASSEMBLED ELECTRODE: (A) A PAIR OF ELECTRODE SIDE PIECES B

IS PLACED IN THE CORRESPONDING POSITION OF THE SILASTIC IMPRESSION;

(B) A PAIR OF SIDE PIECES A IS THEN INSTALLED FROM ABOVE; AND (C) A

PROXIMAL CAP AND A DISTAL CAP ARE MOUNTED. SILASTIC IS USED TO GLUE

THE PROXIMAL AND DISTAL CAPS AND CURED ON A HOTPLATE FOR 5 MIN. .. 20

FIGURE 2.10: THE PICTURE OF AN ASSEMBLED MICROFABRICATED DBS ELECTRODE. ... 20

FIGURE 3.1: OPTICAL PHOTOS OF METALLIZATION BOND PADS USED IN LASER SPOT

WELDING 30 M-THICK WIRES: (A) BEFORE; AND (B) AFTER...... 23

FIGURE 3.2: SCHEMATICS OF THE WIRE/PAD CONFIGURATION FROM LASER SPOT

WELDING EVALUATION REPORT: (A) ACTUAL; AND (B) RECOMMENDED...... 24

FIGURE 3.3: OPTICAL PHOTO OF THE JIG DESIGNED FOR LASER SPOT WELDING / EPOXY

BONDING. THE SILICON MOVABLE PART CONTAINS TWO TRENCH SETS FOR

SPACING AND ALIGNMENT OF THE FINE WIRE...... 26 v

FIGURE 3.4: CROSS-SECTION SCHEMATIC OF THE ATTACHED ELECTRODE SIDE PIECE

AND FLEXIBLE LEAD. THE FIGURE IS NOT DRAWN TO SCALE AND DETAILS OF

THE ELECTRODE SIDE PIECE ARE NOT SHOWN...... 27

FIGURE 3.5: (A) THE FLEXIBLE LEAD MADE BY MICROCONNEX, INC. (B) THE DISTAL END

FOR ATTACHMENT TO THE ELECTRODE. THE DIAMETER OF THE HOLES IS 200

M...... 29

FIGURE 3.6: AN OPTICAL PHOTO OF THE LEAD-ELECTRODE ATTACHMENTS UNDER THE

MICROSCOPE...... 31

FIGURE 3.7: OPTICAL PHOTO OF A DBS ELECTRODE ATTACHED TO A FLEXIBLE LEAD BY

EPOXY STUDS PROTECTED IN SILASTIC...... 31

FIGURE 4.1: (A) THE ROUTE MAP AND THE LOCATIONS PROBED IN THE ELECTRICAL

PROPERTY TEST. (B) A CROSS-SECTION SCHEMATIC OF THE ATTACHED

ELECTRODE SIDE PIECE AND EXTENSION LEAD. THE FIGURE IS NOT DRAWN

TO SCALE...... 34

FIGURE 4.2: OPTICAL PHOTO OF A TEST WAFER WITH 17 LCP LEADS ATTACHED...... 40

FIGURE 4.3: AN OPTICAL PHOTO OF THE PULL TESTER: UH 610 FROM ULTRON SYSTEMS,

INC...... 41

FIGURE 4.4: THE PULL FORCE ALONG THE LEAD DEPENDS ON THE ANGLE BETWEEN

THE LEAD AND THE BOND PAD SURFACE...... 42

FIGURE 4.5: OPTICAL PHOTO OF A MANUALLY PULLED BOND PAD. THE PATTERN IN THE

CENTER OF THE BOND PAD IS THE CONTOUR OF THE EPOXY CONTACT AREA.

THE PAD SIZE IS 2 MM × 2 MM...... 43

FIGURE 4.6: THE IN VITRO EXPERIMENT SETUP (NOT SHOWN TO SCALE): (A) MONOPOLAR

STIMULATION; AND (B) BIPOLAR STIMULATION...... 48

FIGURE 4.7: THE STIMULUS (TRACE 1) AND RECORDED WAVEFORM (TRACE 2) IN THE

vi

IN-VITRO TEST. THE RECORDED SIGNAL WAS AMPLIFIED 100X IN THIS

FIGURE...... 49

FIGURE 4.8: THE MAPPING REGION IN THE MEASURING PLANE...... 49

FIGURE 4.9: BIPOLAR STIMULATION CIRCUIT FOR ELECTRODE IMPEDANCE

MEASUREMENT...... 50

FIGURE 4.10: THE X-Y CONTOURS OF THE ELECTRIC POTENTIAL DISTRIBUTION FROM

MONOPOLAR SIMULATION ...... 51

FIGURE 4.11: COMPARISON OF SIMULATION AND TEST RESULTS: (A) MONOPOLAR

STIMULATION IN REGION A; (B) MONOPOLAR STIMULATION IN REGION B; (C)

MONOPOLAR STIMULATION IN REGION C; (D) MONOPOLAR STIMULATION IN

REGION D; (E) BIPOLAR STIMULATION IN REGION A; (F) BIPOLAR

STIMULATION IN REGION B; (G) BIPOLAR STIMULATION IN REGION C; AND (H)

BIPOLAR STIMULATION IN REGION D...... 52

FIGURE 4.12: ELECTRIC POTENTIAL DISTRIBUTION ALONG THE Z-AXIS: (A) THE

MEASUREMENT POINTS WERE 1 MM AWAY FROM THE ELECTRODE SURFACE

AND SPACED 1 MM; (B) MEASURED PEAK CATHODIC VOLTAGE AND THE

SIMULATION DATA AT THE MEASUREMENT POINTS; (C) CORRESPONDING

POTENTIAL CONTOURS FROM SIMULATION...... 53

FIGURE 4.13: A SEM PHOTOGRAPH OF A BROKEN TRACE ON THE ELECTRODE...... 55

FIGURE A1.1: AN ELECTRODE OVER-ETCHED DURING THE DRIE STEP, RESULTING IN THE

LOSS OF THE MASKING PR AND DAMAGE TO THE SURFACE FEATURES...... 63

vii

Acknowledgement

I would like thank my advisor, Prof. Mehran Mehregany, for his guidance and support in the past two and half years. What I learned from him is not only knowledge about the topic, but also how to analyze and solve problems. Without his help, this thesis would not have been possible. Additionally, I would like to show my appreciation to

Professors Christian A. Zorman and Pedram Mohseni for technical guidance and access to their equipment. This work was part of a collaboration with Dr. Ali Rezai and Mr. Chip

Steiner of the Cleveland Clinic Foundation; their input and assistance are greatly appreciated.

I would also like to thank all my student friends, Grant McCallum, Noppasit

Laotaveerungrueng, Hari Rajgopal, Te-Hao Lee, Sheng Jin, Shih-Shian Ho, Man I Lei,

Cathy Soong and others in the MINO lab for their selfless help. I enjoyed every moment working with them.

Last but not least, I would like to express my sincere appreciation to my parents and my sister.

This work was supported in part by a grant from the Cleveland Clinic Foundation.

viii

A MICROFABRICATED DEEP

BRAIN STIMULATION ELECTRODE

Abstract

by

CHIA-HUA LIN

This work presents a novel 4-sided, 16-channel deep brain stimulation electrode with a flexible extension lead for connectivity with pulse generation electronics. The

3-dimensional electrode enables steering the current field circumferentially. The electrode is fabricated in pieces by microfabrication techniques; the pieces are then assembled mechanically, after which the lead is connected. The electrode is modeled by finite element analysis and tested in vitro to validate the design concept, i.e., targeted stimulation. With monopolar and bipolar configurations, the electric potential in front of the activated site is at least five times larger than that on the other not-activated sides within a 3 mm radius. The resistance of each channel is also measured. Finally, a pull test is performed to evaluate the mechanical strength of the epoxy used to attach the lead to the electrode. The samples withstood a force of 8.66 g at a 60° angle, thus exhibiting high robustness.

ix

Chapter 1: Introduction

1.1 Background

Implantable electrodes are widely used in biopotential measurement and the treatment of neural related disorders. Medical applications include cochlear implant, neural prosthesis, implants in the epidural space of the spinal cord (for the treatment of chronic pain or urinary control), and the subject of this thesis, deep brain stimulation

(DBS) [1.1]. DBS is currently an accepted therapy for Parkinson’s disease and essential tremor [1.1]. DBS’ efficacy in treating drug resistant depression, Tourette’s disorder, and epilepsy is also under research [1.1], [1.2]. A typical DBS system may be categorized into three parts (Fig. 1.1): the implantable pulse generator (IPG) for generating voltage or current pulses at a fixed frequency, the extension lead, and the intracranial lead (the electrode contacts and part of the wires).

The only DBS system currently approved by the US Food and Drug

Administration (FDA) for patient therapy is “Activa Tremor Control Therapy” from

Medtronic, Inc. (Minneapolis, MN). As a therapy for Parkinson’s disease for example, the electrode is placed near the subthalamic nucleus (STN) or pars interna of the globus pallidus (GPi) [1.3]. The electrical pulses are delivered from the IPG to the intracranial lead by the extension lead, which is placed along the neck and behind the ear (see Fig. 1.2).

1

Implanted Pulse Generator

Intracranial Lead

Extension Lead

Figure 1.1: Model 7482A extension lead connected to a Medtronic, Inc. IPG and the intracranial lead. (From Medtronic, Inc.’s DBS extension kit manual [1.6])

Figure 1.2: Implanted DBS system made by Medtronic, Inc. (From www.medtronic.com) 2

Figure 1.3 and 1.4 are the details of the DBS intracranial lead for two models.

The intracranial lead has four 1.5 mm-wide cylindrical electrode sites at the distal end.

The electrode sites are spaced by 0.5 mm in Model 3389 or 1.5 mm in Model 3387.

The diameter of the lead is 1.27 mm. The metal used for the electrode sites and the interconnect wires is an alloy of platinum and iridium. The coiled interconnect wires are insulated individually by Fluoropolymer coating and protected in an 80A Urethane tubing [1.4]. The attachment of the interconnect wires and the electrode sites is achieved by laser spot welding [1.5] as seen in Fig. 1.4.

Figure 1.3: The dimensions of DBS intracranial lead Model 3389 and 3387 (From Medtronic, Inc.’s DBS lead kit manual [1.4])

3

Laser Spot Welding Site

Figure 1.4: Optical photo of a detached electrode contact on Medtronic, Inc.’s DBS intracranial lead. The inset shows the distal ends of the Medtronic intracranial lead “3387 (left)” and “3389 (right).”

The IPG of the Medtronic, Inc.’s DBS is 55 mm × 60 mm and 5 mm-thick. It contains circuitry for square pulse generation and a non-rechargeable 3.7 V lithium-thionyl chloride battery. The IPG is encased in titanium and placed in a subcutaneous pocket on the patient’s chest [1.6].

Accurate positioning of the intracranial lead during the surgery is important as even a spatial deviation on the order of 1 mm will have a large impact on therapy efficacy. Side-effects like speech dysfunction are often observed when the system is activated [1.2]. McIntyre, et al., using finite element analysis, have studied the effect of size, shape, and aspect-ratio of electrode sites [1.7]. Infection and erosion, migration or misplacement of the leads, lead fractures, and battery failure are the most frequently

4

observed system failure causes [1.8].

In this work, a novel DBS electrode was fabricated leveraging

MicroElectroMechanical Systems (MEMS) technology to explore an alternative to

Medtronic, Inc.’s electrode design. Microfabricated electrode arrays can have higher electrode density and uniformity. MEMS technology is being used in biomedical applications such as blood pressure monitoring, infusion pumps, kidney dialysis, respirators, and neural interfaces [1.9].

In the Medtronic, Inc.’s DBS electrode design, the current distribution in the cylindrical electrode sites emanates uniformly about the circumference, which leads to stimulation of all areas surrounding the electrode. The design presented in this thesis has a total of 16 electrode sites distributed on four side pieces of an assembled electrode with a square cross-section. This approach provides the ability to steer the stimulation field circumferentially. Further, batch fabrication using standard integrated circuits (IC) processes provides a pathway for enhanced reproducibility of the shank and its features. Since the electrode is fabricated from silicon, it can eventually carry microsensors, microactuators, and electronics for new diagnostic and therapeutic capabilities, e.g., neural recording and neurochemical stimulation. A potential drawback is that the 3-D design of electrode requires assembly of silicon parts fabricated from a wafer; assembly of small parts with very small features is often challenging in the context of yield and cost. The electrode assembly has to then be connected with the IPG through an extension lead provision.

The first microfabricated silicon-based electrode array, also known as the

Michigan array, was made by Wise, Starr, and Angell in 1970 [1.10]. The electrode

5

consists of a titanium/iridium metal layer insulated by two sandwich insulators

(oxide/nitride/oxide for stress relief) on a shank defined by diffusing boron to the silicon substrate and then releasing the shank in ethylene diamine pyrocatechol.

Michigan array has evolved into a 64-site, 3-dimensional structure by assembling comb-like shanks into a frame. Later on, drug delivery trenches and active components were also added to the electrode [1.11].

Another microfabricated electrode is the Utah array. Unlike Michigan array, the

Utah array was fabricated to be a 3-D structure but not a planar shank. It contained 100 silicon-based conductive micro-needles which were insulated by polyimide and coated with platinum at the sharpened tip [1.12]. The shank length could be adjusted between

0.5 to 1.5 mm for stimulation or recording at different depths [1.13].

Judy, et al. presented a DBS electrode which was designed to minimize damage to surrounding tissue [1.14]. A novel 3-D plating technique shaped the shank to a tapered geometry similar to conventional microwires.

Even though silicon-based microelectrodes have long been used for interfacing with the neuron, the mechanical mismatch between the stiff probe (silicon’s Young’s modulus is ~170 GPa) and soft biological tissue (brain tissue’s Young’s modulus is

~3 kPa) may cause inflammation at the implant site. The inflammation encourages the formation of a glial scar which encapsulates the probe and thus isolates the electrode

[1.1]. Therefore, implantable devices with a soft and flexible base structure are also used in some biomedical applications. Polyimide is chosen as the mechanical supporting and electrical insulation material for many such implantable electrode designs due to its biocompatibility, lower tensile strength, and ease of fabrication (by

6

spin-coating) [1.15]-[1.19]. However, polyimide suffers from high moisture absorption

(1-4% in 24 hrs, [1.20]), which can lead to metal delamination from the polyimide substrate. To overcome this problem, microelectrodes mechanically supported and electrically insulated by Liquid Crystal Polymer (LCP) [1.20], Parylene [1.21], [1.22], and PNB [1.20], [1.23] are currently under research. Besides silicon and flexible materials, glass [1.15], [1.24], metal [1.25], and sapphire [1.26] have also been used as the structural material, for example, for higher shank stiffness or better electrical isolation.

Various conductive materials like gold [1.17], platinum [1.12], [1.14], [1.15],

[1.17], [1.23], [1.26], platinum / iridium [1.27], iridium [1.28]-[1.30], iridium-oxide

[1.31], nickel-iron [16], silicon [11], tantalum [26], or phosphorus-doped polysilicon

[1.32] have been used for the electrode sites or traces connecting to them.

Iridium-oxide has been recently shown to be a more suitable metal for implantable electrodes due to its high charge injection limits (Au: 20 μC/cm2, Pt: 75 μC/cm2, IrOx:

3 mC/cm2) [1.33]. In this work, platinum is chosen for the electrode sites metallization, and gold is used for the traces.

Ultrasonic bonding [1.34], laser spot welding [1.6], and ball bonding [1.35] have been used to connect the implantable microfabricated electrodes with the signal processing circuitry. Another approach has been to integrate electrode array and the lead during the fabrication process [1.20]. However, this technique requires a large area on the device wafer and thus increases the cost for fabrication.

Stieglitz, et al. developed a technique named MicroFlex extension lead which is the approach in this thesis to connect the electrode to the IPG [1.18]. The approach

7

uses a flamed gold ball stud to attach the polyimide-based flexible lead to the silicon-based electrode shank or the PCB with signal processing circuit. The technique was initially designed for retina stimulation, but it has also been used in cerebral application later [1.19]. However, the packaging/interconnection methods are usually designed for specific applications. Thus, assessment and modification are necessary when incorporating these techniques into new electrode design.

1.2 Thesis Overview

The thesis outline is as follows. Chapter 2 discusses the electrode fabrication process and assembly. Chapter 3 presents the MicroFlex extension lead to connect the electrode to an IPG. An assessment of several interconnection strategies is discussed to understand the constrains on and requirements for the interconnect. Chapter 4 presents electrical testing, testing of the mechanical robustness of the extension lead and electrode attachment, and in-vitro testing along with modeling and simulation of the electrode stimulation field. Finally, Chapter 5 presents conclusions and future work.

8

Chapter 2: Microfabricated DBS Electrode

2.1 Electrode Architecture

We describe herein the microfabricated DBS electrode (Fig. 2.1). The electrode is designed to be assembled from four 824 μm × 13,600 μm silicon side pieces. In this thesis, the electrode sites are from platinum, i.e., similar to that used in the Utah array and other microelectrodes [2.1], [2.2], [2.3], [2.4]. Each side piece has four electrode sites. Each site has an area of 500 μm × 1500 μm, and the sites are separated by

1300 μm (edge-to edge) and connected to the bond pads by 10 μm-wide traces.

There are two design variations for the proximal end (i.e., the end with bond pads). In the first design, developed initially by Grant McCallum in our group, the four side pieces are identical, and the bond pads have an area of 100 m × 100 m and are separated by 30 m (edge-to-edge). A cap piece is used for assembly at each of the proximal and distal ends (Fig. 2.1(a)). Since this design led to assembly and attachment difficulty, the corresponding electrode side pieces are only used in Section 3.1 for attachment method assessment. The second design (Fig. 2.1(b)), which is a revision of the first, increases the smallest bonding pad area to 200 m × 200 m and pad-to-pad separation (center-to-center) to 400 m. Unlike the first design which uses four identical side pieces, the second design uses two pairs of different side pieces, complementary to each other, with a redesigned cap at the proximal end. The second design also reduces the assembly time (including epoxy curing time) from two days to one hour and reduces the difficulty of manual interconnect alignment and epoxy dispensing. The details of assembly will be further discussed in Section 2.3. This thesis focuses only on the second electrode design, except for in Section 3.1 as noted above.

9

Figure 2.1: Electrode designs: (a) first; and (b) second, which is a revision of the first. Metal traces connecting the electrode site to corresponding bond pads are not shown in (a) and (b). (c) Dimensions of the side pieces of the second electrode design.

10

2.2 Electrode Fabrication Process

An over view of the 3-mask fabrication process is presented here; the process traveler, containing the details, is included in Appendix 1 for further details. For the side pieces, the fabrication process began with a 250 m-thick, double-side polished, p-type,

(100) silicon wafer. A tri-layer insulator sandwich was deposited on the front side for substrate isolation using low pressure chemical vapor deposition (LPCVD): 3000 Å-thick low-temperature oxide (LTO), followed by 1500 Å-thick low stress nitride, followed by another 3000 Å-thick LTO (Fig. 2.2(a)) [2.5], [2.6]. Such a dielectric stack structure has two advantages: (i) improved electrical insulation; and (ii) compensation of the competing stresses from the nitride and oxide layers.

11

(a)

(b)

(c)

(d)

(e)

(f)

Low Stress Silicon Titanium Platinum Oxide Nitride Wafer

Figure 2.2: Fabrication process flow for the 3-D electrode side pieces (not shown to scale): (a) Substrate isolation oxide/nitride/oxide layer deposition; (b) titanium and platinum metal deposition; (c) trace passivation oxide deposition; (d) opening of windows in the trace passivation layer to expose the electrode sites and bond pads; (e) patterning of the isolation/passivation layers to expose the side piece outline; and (f) etching out the side piece from the wafer.

12

The metal features were then realized using the lift-off process. A 5 m-thick

AZ9245 photoresist (PR) was spun on and patterned (MASK 1). A 100 Å-thick titanium adhesion layer and a 3000 Å-thick platinum layer were then sequentially sputtered. The lift-off process was accomplished by dissolving the PR in acetone overnight (Fig. 2.2(b)).

After patterning the metal, a 3000 Å-thick SiO2 (Tetraethyl Orthosilicate or TEOS) was deposited by LPCVD to passivate the metal traces (Fig. 2.2(c)). A 10 m-thick AZ9260

PR (MASK 2) was patterned to serve as etch mask for opening windows in the passivation oxide to expose the electrode sites and bond pads; the oxide was etched by reactive ion etching (RIE) (Fig. 2.2(d)).

Next, two RIE steps were used to realize the side pieces (MASK 3). For this etch step, a 10 m-thick AZ9260 PR coats was applied twice to serve as the etch mask: The first time for the first RIE step to etch through the tri-layer isolation/passivation sandwich

(Fig. 2.2(e)); and the second time for the deep RIE (DRIE) step to etch through the

250 m-thick silicon wafer (Figs. 2.2(f)). However, reapplying PR for the same pattern caused a 5 m pattern shift as shown in Fig. 2.3. As a result, the smallest distance between the metal traces and the edge in the final fabricated electrode decreased to 35 m from 40 m.

13

Wafer Frame

Tab 5 μm Pattern Shift

Electrode Side Piece

50 μm

Figure 2.3: Pattern misalignment in PR due to repetition of Mask 3.

In order to assemble the four side pieces into a 3-D electrode, two microfabricated caps (distal and proximal) were used as shown in Fig. 2.4 to the right of the two electrode side pieces. The jig to assemble the electrode was also microfabricated from the same silicon wafer. The fabrication process for the caps and jig used one DRIE step with one mask.

14

Distal cap

Proximal cap

Side piece A Side piece B

Figure 2.4: Optical photo of the fabricated electrode parts: electrode side pieces, distal cap, and proximal cap.

2.3 Assembly process

For the electrode assembly, four micromanipulators were initially used to hold the electrode side pieces. However, this approach was time consuming for epoxy curing (at least 24 hrs at room temperature), since the micromanipulators could not be moved onto a hotplate along with the assembly jig and the electrode. Thus, the micromanipulators were later replaced by a silastic impression which was made by an assembled electrode to reduce the assembly and curing time.

In initial assembly attempts, a special jig was fabricated and assembled from two silicon pieces separated with four spacers. Figure 2.5(a) illustrates this jig structure

(for the second electrode design.) The silicon pieces of the jig (etched by DRIE as

15

described above) had five sets of five holes identical to the holes on the distal cap piece

(four square holes for holding the side pieces and one round hole for the alignment pin). as shown in Fig. 2.5(b). Next, four electrode side pieces (a pair of Part A and a pair of Part B) were held by micromanipulators with vacuum tips and moved to a position above the square holes on the jig as shown in Fig. 2.6(a) and Fig. 2.7. A pin for cap alignment was placed in the center of the electrode and held in place by the center round holes on the two silicon jig pieces. A proximal cap and a distal cap were slid onto the pin and the fixed side pieces from above in the order shown in Fig. 2.6(b).

Silastic MDX4-4210, a biomedical grade elastomer from Dow Corning (Midland, MI), was applied on the proximal and distal caps as glue (Fig. 2.6(c)). Silastic can be cured in 15 min at 100°C or 24 hrs at room temperature. Since the side pieces needed to be held by micromanipulators until Silastic was completely cured, room temperature curing was used at this time. The cap alignment pin was then removed after silastic was cured.

To simplify the assembly process thereafter, the assembled electrode can be used to create an impression by pouring silastic into a machined cavity containing an assembled electrode and then curing the silastic as shown in Fig. 2.8. To use the impression for faster electrode assembly, a pair of side pieces B (see Fig. 2.4) was moved to the corresponding position in the silastic impression by tweezers as shown in

Fig. 2.9(a). Next, a pair of side pieces A was then inserted into the impression from above as shown in Fig. 2.9(b). Next, a proximal cap and a distal cap were placed on the side pieces sequentially, and the entire stage was moved to a hot plate for a 15 min curing at 100°C as shown in Fig.2.9(c). Fig. 2.10 shows an assembled microfabricated

16

electrode.

Figure 2.5: (a) Optical photo of the electrode assembly jig structure constructed from two identical silicon pieces and four spacers. (b) The layout of the alignment hole set.

17

(a) (b) (c)

Electrode side piece

Distal cap Silastic Proximal cap Vacuum tip on micromanipulator

Cap alignment pin Assembly jig

Figure 2.6: The electrode assembly process flow using a custom designed jig and micromanipulators: (a) four electrode side pieces are aligned and inserted into the holes on the jig; (b) a proximal cap and a distal cap are mounted onto the electrode; and (c) silastic is applied onto the caps and cured for 24 hours at room temperature.

18

Figure 2.7: Optical photo showing electrode assembly process for the first electrode design.

Figure 2.8: Optical photo of the impression with an assembled electrode.

19

Figure 2.9: The electrode assembly process using a silastic impression of an already assembled electrode: (a) a pair of electrode side pieces B is placed in the corresponding position of the silastic impression; (b) a pair of side pieces A is then installed from above; and (c) a proximal cap and a distal cap are mounted. Silastic is used to glue the proximal and distal caps and cured on a hotplate for 5 min.

Figure 2.10: The picture of an assembled microfabricated DBS electrode.

20

Chapter 3: Extension Lead

3.1 Attachment Method Assessment

Wedge/ball bonding, laser spot welding, and epoxy bonding have been considered as the extension lead attachment candidates. The first designed electrode was used for trial runs and evaluation. The results were then used to revise the electrode design and arrive at the second design described above.

3.1.1 Wedge/ball Bonding

Wedge/ball bonding is a common technique in integrated circuit (IC) packaging due to: (i) its ready adaptability to diverse metallization and packaging styles; (ii) its easy re-workability; (iii) its amenability to visual inspection for bond quality screening;

(iv) the ease with which nondestructive bond integrity tests may be performed; (v) its record of satisfactory field reliability; and (vi) the related base of technical expertise and installed equipment in the field [3.1].

Generally speaking, the bonding tool applies pressure, heat, ultrasonic power, or a combination to the wire and the bond pad metallization depending on the type of bonding technique. For example, wedge bonding is performed by pressure and ultrasonic power, and is also called ultrasonic bonding. On the other hand, ball bonding can be formed by heat and pressure (thermocompression bonding), or by also including ultrasonic power (thermosonic bonding) [3.1]. However, it is not possible to borrow any one of these techniques to attach the lead to our assembled electrode. Since the electrode structure is hollow, the silicon side pieces can break due to the pressure or

21

ultrasonic force applied. Further, the wire used for bonding is too thin (e.g., from 0.7 to

2 mils) to provide robust attachment strength in this application.

3.1.2 Laser Spot Welding

Laser spot welding is used by Medtronic, Inc. to attach the lead wires to the electrodes. There are three significant advantages in laser spot welding, the first being non-contact. Unlike wedge/ball bonding, there is no pressure applied to the wire or the bond pad during the process. The Nd: YAG laser system focuses a laser beam to the spot (where the wire is located on the bond pad), and melts the wire and pad metallization at the interface to achieve a robust connection, both mechanically and electrically. The second benefit is that there is no filler material between the wire and the bond pad. Dissimilar metals have different half-cell potentials. If the filler conductive material is in contact with electrolyte, brain tissue, or the body fluid in this case, the extra half-cell potential will affect the stimulation signal. An electrochemical reaction may result between the electrolyte and the two different metals, which can result in additional polarization and often in corrosion of one of the metals exposed to the electrolyte. Thus, the half-cell potential becomes less stable [3.2]. The third benefit of laser spot welding is that it is a rapid and cost-efficient process. As long as a mechanical system that can automatically align the electrode and the wire is built, rapid mass production is feasible.

However, laser spot welding has strict requirements on bond pad metallization thickness. Miyachi Unitek, Inc. (Monrovia, CA) provided us with a trial run to laser spot weld 1.2 mils 90% Pt / 10% Ir fine wires to the first electrode design. This trial

22

run was not successful. According to the evaluation report, the relative metallization thickness was the main reason for failure. The 300 nm-thick bond pad was damaged by the laser energy, which needs to be intense enough to melt the wire (30.5 m-thick) during the process (Fig. 3.1). One approach to use thick wire with a thinner pad is to flatten the wire to reduce its effective thickness over the pad (Fig. 3.2).

(a) (b)

Figure 3.1: Optical photos of metallization bond pads used in laser spot welding 30 m-thick wires: (a) before; and (b) after.

23

(a)

(b)

Figure 3.2: Schematics of the wire/pad configuration from laser spot welding evaluation report: (a) actual; and (b) recommended.

According to the laser welding guide [3.3], to have a good connection, the gap between the wire and the pad should not exceed 10% of the wire thickness. It is possible to deposit a 1 m-thick metal layer with conventional micromachining techniques, but wire flattening can only be done with a specifically designed automatic tool which was not available for this thesis.

24

3.1.3 Manual Epoxy Bonding

Epoxy bonding is an alternative to laser spot welding. The idea of epoxy bonding is simple: apply a drop of conductive epoxy onto the bond pads on the side piece; push the 1.2 mils wire into the epoxy drop; and cure the epoxy.

To evaluate the viability of this method, a micromanipulator with a 50 m-thick tip was used to deliver the epoxy. The epoxy used was EPO-TECK H20E from Epoxy

Technology, Inc. (Billerica, MA). The epoxy is an approved non-toxic material, complying with USP Class VI Biocompatibility Standards. In order to move the wires into the epoxy drops before curing, a two-piece jig was designed for the first electrode design (Fig. 3.3). The upper jig (the movable part) was fabricated by deep reactive ion etching (DRIE) of a 500 m-thick silicon wafer. The 1.2 mils 90% Pt / 10% Ir TEFLON coated wires could be aligned to the pads as expected. However, the epoxy amount applied onto each pad was critical. The wire could not be attached if the epoxy volume was not enough; the epoxy overflowed to adjacent pads during the 120°C 15 min curing step if excessive epoxy was dispensed. Inserting 16 fine wires into the fixture to connect the 16 electrodes of the probe was also a challenge.

25

1 cm

Space for the assembled electrode

Four trenches

Movable upper jig piece Aluminum base jig piece Four trenches

Figure 3.3: Optical photo of the Jig designed for laser spot welding / epoxy bonding. The silicon movable part contains two trench sets for spacing and alignment of the fine wire.

3.1.4 Conclusions

Even though the methods in Section 3.1 do not meet the requirements in the DBS electrode application here, we can conclude the following. First, the lead attachment must be straight forward, in addition to being reliable and robust. Since the assembled electrode has a total of 16 electrode sites, using a wire bundle for the lead is a better choice than individual wires. Secondly, the area and the pitch of the bond pads on the electrode can be relocated and enlarged to simplify wire connection. Thirdly, a new method for attachment of the lead wires which does not require a large external force is needed.

These conclusions were incorporated in the electrode redesign. The MicroFlex interconnect technique was then explored for a flexible lead incorporating a wire bundle and the related attachment of the wires to the electrode.

26

3.2 MicroFlex Extension Lead

3.2.1 Introduction

MicroFlex interconnect is a technique developed originally for retina stimulation by Stieglitz, et al. [3.4]. In their approach, the interconnect was a flexible polyimide-based lead with via holes and partially exposed bond pads. The holes (and therefore the partially exposed bond pads) were aligned to the bond pads on the microfabricated chip. Then, a melted gold ball stud with a diameter larger than the exposed bond pad area was dropped onto the hole for attachment. The ball stud was generated by a flame-off device installed on a standard ball bonder [3.4]. In this work, conductive epoxy was used instead of gold, and the materials used for the flexible lead were also changed. Figure 3.4 shows the cross-section of the modified MicroFlex extension lead structure.

Gold Parylene Epoxy Stud Trace Coating Electrode Site

Flexible Lead

LCP Electrode Side Piece

Figure 3.4: Cross-section schematic of the attached electrode side piece and flexible lead. The figure is not drawn to scale and details of the electrode side piece are not shown.

27

3.2.2 Flexible Lead Design and Fabrication Process

Liquid crystal polymer (LCP) and parylene were chosen as the materials for the base structure and protective coating, respectively, in this thesis because of their higher moisture rejection and the relative ease of fabrication [3.5]. The diameter of the via holes was 200 m. A stiffener was added to the proximal end of the flexible lead that interfaces with the micro-connector (FH12F-16S-0.5SH, Hirose) on the printed circuit board (PCB).

Figure 3.5 shows the flexible lead, which is fabricated by Microconnex, Inc

(Snoqualmie, WA). The fabrication process was generally as follows:

1) A bare 50 m-thick LCP was sheared into 6” x 6” square panels.

2) The panels were tooled for shop fixtures and were serialized.

3) The panels were plasma cleaned.

4) The panels were coated with a 1μm-thick dryfilm resist.

5) The circuit areas were exposed and developed using photolithography.

6) 10 nm-thick TiW and 400 nm-thick Au were sputtered onto the panels.

7) Dryfilm was chemically removed from the panels, leaving the interconnect

pattern sputtered onto LCP substrate.

8) Panels were plasma cleaned.

9) 30 μm-thick dryfilm resist was applied to panel.

10) Areas to be free of parylene were covered with dryfilm resist using

photolithography to protect the thin Au layer when removing the parylene coating.

11) The panels were plasma cleaned.

12) The panels were shipped to a parylene coating vendor for 10 m-thick coatings by

chemical vapor deposition (CVD).

28

13) The parylene over the dryfilm resist was removed with UV laser ablation.

14) The dryfilm resist was chemically removed, leaving openings in the Parylene

layer.

15) The panels were ultrasonically cleaned in sodium persulfate solution to remove

any laser debris.

16) The panels were plasma cleaned.

17) Stiffeners were laser cut.

18) Stiffeners were applied.

19) Individual circuits were laser trimmed from the panels.

Figure 3.5: (a) The flexible lead made by Microconnex, Inc. (b) The distal end for attachment to the electrode. The diameter of the holes is 200 m.

29

3.2.3 Attachment Process

The flexible extension lead can be attached to the electrode side piece of the second electrode design either before or after electrode assembly process. Each method has its own pros and cons. There is a higher possibility of attachment failure if the electrode is assembled first because the height of the assembled electrode causes an angle between the lead and the electrode surface. On the other hand, attaching the lead first makes the assembly process more complicated and time-consuming. In this work, we decided to attach the lead after electrode assembly.

At first, an assembled electrode was fixed on a metal holder by tape. Four via holes on the distal end of one of the leads were aligned to the bond pads on the electrode.

The flexible lead was also fixed by tape after alignment. In order to avoid attachment failure, the flexible lead was placed on a spacer to make it level with the electrode side piece. Epoxy studs (EPO -TECK H20E) were then applied onto the bond pads on the electrode through the via holes in the flexible lead. Care was exercised to avoid shorting any adjacent channels due to excessive epoxy application. The recommended epoxy curing recipe was 120°C and 15 min on a hotplate. Silastic was used to cover the epoxy studs and cured at 100°C for 15 min after all the four sides were attached. Figure 3.6 shows a photo of the attachment site under microscope, and Fig. 3.7 shows an attached electrode with a flexible lead.

30

Figure 3.6: An optical photo of the lead-electrode attachments under the microscope.

Figure 3.7: Optical photo of a DBS electrode attached to a flexible lead by epoxy studs protected in Silastic.

31

Chapter 4: Electrical and Mechanical Properties, and In-vitro

Testing

4.1 Electrical Testing

4.1.1 Introduction

A first set of tests was to test the electrical performance of the lead and its attachment to the electrode. Stimulation signals transmitted along the path from a PCB through the PCB-lead connection, the traces on the flexible lead, the epoxy studs, the bond pads and the traces on the electrode side pieces to the electrode sites. The resistance of each segment/element along the path was measured and analyzed.

4.1.2 Test Process

Testing began with the electrode side pieces and extension lead separately.

Figure 4.1 defines the position and channel number in the measurement. At first, the resistances of the traces between the electrode sites (F) and the bond pads (E) on the electrode side pieces were measured. Since there were two different side pieces in an electrode and each of them had four traces, the measurement was performed eight times to acquire the resistance for all the eight different traces.

Next, the resistance of the traces on the flexible lead was measured. Since there were only two different distal ends in a flexible lead, only the eight traces in the middle two distal ends were chosen for testing. The probes were placed on the exposed trace (B) at the proximal end and the exposed bond pad (C) around the via hole.

Next, epoxy studs were used to attach the flexible lead to the electrode side pieces.

Assembled electrodes were not used because flat pieces were easier for measurement and

32

handling. Next, after the epoxy studs were cured, the resistance between the exposed traces at the proximal end of the flexible lead (B) and the electrode sites (F) was measured. In order to make sure the channels were not shorted with each other, the resistance between each electrode site (F) was also measured. The flexible lead was then connected to a PCB with a 0.5 mm-pitch micro-connector, FH12F-16S-0.5SH from

Hirose, Inc. (Simi Valley, CA), and the resistance between the traces on the test PCB (A) and the electrode site (F) was measured.

33

Figure 4.1: (a) The route map and the locations probed in the electrical property test. (b) A cross-section schematic of the attached electrode side piece and extension lead. The figure is not drawn to scale.

34

4.1.3 Data Analysis and Comparison with Design

Tables 4.1 and 4.2 present the data from the electrical testing. RE-F (5 μm) is the resistance measured between the bond pad and the electrode site on the electrode side piece, whose metal layer was patterned by 5 μm PR. RE-F (calc.) is the calculated resistance

l value from equation: R   , where  is the resistivity of sputtered platinum (15.8 × wt

10-6 -cm) [4.1], l is the trace length, w is the trace width, and t is the metal layer thickness (3000 Å). It was observed that the ratio between RE-F (5 um) and RE-F (calc.) was between 1.4 and 1.6. Since the metal thickness was measured as 3000 Å as designed, the result indicated the resistivity is not that which was expected.

RE-F (5 RE-F (2.6 RE-F (5 μm) : RE-F RE-F (2.6 μm) : RE-F RB-C Channel RE-F (calc.) RB-C RB-F RA-F μm) μm) (calc.) (calc.) (calc.)

1 503 369 434 1.4 1.2 255 778 776 2 770 479 561 1.6 1.2 247 1146 1143 3 291 184 225 1.6 1.2 243 531 530 4 100 68.5 88 1.5 1.3 242 362 361 150 5 612 439 510 1.4 1.2 249 969 967 6 862 552 641 1.6 1.2 240 1317 1229 7 394 256 311 1.5 1.2 239 686 685 8 209 142 166 1.5 1.2 236 475 475 Unit:  Table 4.1: The measured and design resistances.

Sputtering chamber contamination from the 5 m-thick PR needed for lift-off is considered to be one contributor to the higher measured platinum resistivity. A

2.6 m-thick PR layer is usually sufficient for 3000 Å-thick metal lift-off. Thicker PR resulted in the chamber pressure increase as observed during the sputtering process. To

35

examine this further, 100 Å-thick titanium followed by 3000 Å-thick platinum layers were sputtered onto two silicon monitor wafers with 3000 Å-thick oxide layer on the surface. One wafer had patterns (same traces, pads, stimulating sites as on the electrode side piece) defined by 2.6 m-thick Shipley 1818 PR, and the other one did not. The resistance of the traces on the patterned monitor wafer was measured (marked as (2.6 μm) in

Table 4-1). The ratio between RE-F (2.6 μm) on the monitor wafer and RE-F (calc.) decreased from 1.5 to 1.2.

The other monitor wafer which had no pattern was used for four-point probe test to find the exact metal layer resistivity. The measured resistivity of the platinum layer was 1.91 × 10-5 -cm. The value was derived by multiplying the mean sheet resistance,

0.63681 /, by the metal thickness, which is 3 × 10-5 cm. This value was 1.2 times higher than the expected sputtered Pt resistivity and matched the result of the patterned monitor wafer. The 20% resistivity difference may come from different sputtering parameters and different thin film thickness. The film thickness in [4.1] was 5000 Å and the thickness of the Pt layer used in this work was 3000 Å.

Table 4.1 also shows that the trace resistance on the flexible lead (B-C) was larger than the calculated value. In other words, the sputtered gold resistivity was also higher than the expected value, 3 × 10-6 -cm [4.1]. Thus, two conclusions are made: (i) using thicker PR for lift-off may increase the resistivity of the metal layer; (ii) the sputtered platinum or gold has higher resistivity than expected due to different sputtering processes.

As expected, the measured resistance of the traces on the electrode is proportional to the length of the traces on the side piece. The values are in the following order: 2 > 1 >

36

3 > 4 and 6 > 5 > 7 > 8. The resistance of the epoxy between the lead and the pads on the side piece can be derived by subtracting RE-F and RB-C from RB-F. The result is shown in

Table 4-2. The mean value in Table 4.2 is 71 Ω. However, the range is from -3 Ω to

215 Ω, and the standard deviation of mean is 74. The result indicates that applying the epoxy manually without an automatic dispenser is not repeatable. Table 4.3 also indicates that the resistance measured is not accurate since the estimated RC-E is negative in one case. One possible reason is that the parylene coating on the exposed gold trace was not ablated completely. Therefore, the measured RB-C is larger than the actual value due to the parylene residue between the probe and the trace. The actual RB-C as a part of RB-F is not affected because there is no parylene on the side wall of the via hole.

Channel RC-E (estimated) ()

1 20.0 2 129 3 -3.00 4 20.0 5 108. 6 215 7 53.0 8 30.0 Table 4.2: Estimated epoxy resistance.

The last column in Table 4.1 is the total resistance from the traces on the PCB to the electrode site (RA-F). The value is almost the same as RB-F. The stiffener and the micro-connector, FH12F, worked well and provided good contact between the flexible lead and the PCB.

Table 4.3 shows the resistance between each electrode site. Channels 1, 2, and 3 were electrically connected and so were channels 5 and 6. This is because of the epoxy

37

overflow and may be solved by increasing the pitch or dispensing the epoxy by an automatic dispenser.

Channel-Channel RF-F () 1-2 1469 1-3 863 1-4 open 2-3 1183 2-4 open 3-4 open 5-6 1785 5-7 open 5-8 open 6-7 open 6-8 open 7-8 open Table 4.3: Resistances between each channel.

4.1.4 Conclusion

In Section 4.1, the resistance of the entire signal path and each segment/element was measured and analyzed. The resistance of the epoxy was derived from the experiment data as well. The test results provide three conclusions:

1. The MicroFlex lead is electrically effective in this application.

2. To make the electrical performance of the lead and its attachment more repeatable,

an automatic dispenser should be used.

3. Platinum and gold traces are the main resistance contributors along the path. In

future, the resistance can be improved by increasing the metal thickness or the

trace width on both the electrode and the flexible lead.

38

4.2 Testing for the Mechanical Robustness of Lead Attachments

4.2.1 Introduction

Mechanical robustness is very important for any implantable biomedical device.

Any unexpected fragments or materials detached may lead to unpredictable consequences.

Assuming the interface between the epoxy stud and the bond pad on the electrode is a potential breaking point, a non-destructive pull test was performed. By immersing the attached parts in a saline tank (to mimic the tissue fluid) for 5 days and repeating the pull test after every 24 hours, the change in the strength of the attachment was examined.

4.2.2 Test Environment and Preparation

Instead of using the electrodes as test devices, a test wafer was designed and fabricated for this pull test. There were 88, 2 mm × 2 mm bond pads on the wafer; therefore, 44 test leads could be attached at most. The distance between the paired bond pads was 15 mm (center-to-center). A 4” silicon wafer was covered with a 3000 Å-thick oxide layer. A 100 Å -thick titanium layer followed by a 3000 Å-thick platinum layer were sputtered on the wafer. All parameters for sputtering were the same as those used for the electrode metallization. The pulled leads were cut from a 50 m-thick LCP sheet by laser. Each LCP test lead has two ends matching the outline of the bond pads on the test wafer and a 200 m-diameter via hole in the center of each end. The distance between the holes on the lead was 30 mm.

One end of the test LCP lead was attached to the pad by an epoxy stud, and the other end was then fixed to the other pad by tape. The epoxy attached end worked as the

39

test end and was expected to break before the taped end. 18 leads were pulled in this test.

Figure 4.2 shows the test wafer with 18 leads attached.

Figure 4.2: Optical photo of a test wafer with 17 LCP leads attached.

4.2.3 Test Process

The pull tester used was UH 610 from Ultron Systems, Inc. (Moorpark, CA)

(Fig. 4.3). The pull tester contained a base, a cantilever, and a force gauge with pressure control. The transparent tube at the right side of Fig. 4.3 was for air pressure control to slow down the cantilever’s moving speed. The maximum recordable force was

15 gram-weight. The total test time was 5 days. At first, 24 hours after the epoxy was applied and cured, three leads (a, b, c as defined in Fig. 4.2) were pulled. All the three

40

attachments withstood the maximum pulling force of the pull tester. Then the wafer was immersed into a tank filled with 0.9% sodium chloride solution. Lead b was observed broken after 2 hours immersion in saline, but leads a and c were still firmly attached.

After 24 hours in saline, the wafer was removed from the tank, and the two pulled leads were pulled again. Then three other leads were pulled. The process was repeated until 18 leads were pulled in total. All the leads, except for lead b, did not show any sign of breakage with the maximum pulling force. After that, six test leads were pulled by tweezers and the diameters of the epoxy residues on the bond pads were recorded.

Figure 4.3: An optical photo of the pull tester: UH 610 from Ultron Systems, Inc.

41

4.2.4 Data Analysis

The results showed that the epoxy-bond pad interface can survive at least

8.7 gram-force from a 60° angle above the pad. According to the equation and simplification given by Schafft [4.2], the force along the lead (Fl) was derived from Fup =

F / 2sin, where Fup is the vertical pulling force (15 gram-force maximum), and  is the angle between the lead and the bond pad (60°) (Fig. 4.4). Note that the pulling force is meaningful only when the pulling angle is given.

Fup

Fθ

θ

Figure 4.4: The pull force along the lead depends on the angle between the lead and the bond pad surface.

The result also showed that the mechanical robustness of the epoxy stud was not affected (within the measurement limit) by the environment in the short-term considered.

During the five days in the 0.9% NaCl solution, among all the tested leads, only one (b in

Fig. 4.2) broke. All the rest remained firmly attached and did not show any sign of degeneration. The one detached pad was inspected, and the diameter of contact epoxy was 420 m.

42

Figure 4.5 shows the picture of one of the six bond pads which was manually detached from the lead; Table 4.4 presents the diameter of the epoxy contacts. The average contact diameter was 1107 m. It was observed the well-attached epoxy stud contact area was at least two times larger than the contact area of the naturally broken stud. It showed that a single 420 m stud may not be enough to attach the lead firmly.

Comparing the epoxy contact area (1107 m) with the via hole diameter (200 m), the result shows that the lateral epoxy spread was significant with no vertical force applied to the lead to minimize the gap between the lead and the bond pad. To prevent shorting on the device, the distance between the via holes needs to be larger than 1 mm or the epoxy stud formation must be refined further.

Figure 4.5: Optical photo of a manually pulled bond pad. The pattern in the center of the bond pad is the contour of the epoxy contact area. The pad size is 2 mm × 2 mm.

43

Test Lead Epoxy Contact Diameter(m) E 841 F 1161 G 1337 H 1088 I 950 j 1264 Average 1107 Table 4.4: The diameter of the contact area between the epoxy stud and the bond pad for the six samples pulled manually.

4.2.5 Conclusion

The mechanical robustness of the attachment between the epoxy stud and the bond pad was tested in this section. The result can be generalized into two conclusions:

1. An epoxy contact area larger than 800 m is strong enough to survive at least

8.7 gram-force pulling along a 60° angle.

2. A 200 m-wide via hole makes an 1107 m-wide contact area when using

epoxy H20E without any pressure to minimize the gap between the lead and the

bond pad.

44

4.3 Electrode Modeling and In-vitro Test

4.3.1 Electrode Modeling and Electric Potential Field Simulation

To validate the concept that the 16 sites on the four side pieces can steer the stimulation field circumferentially, COMSOL Multiphysics (v3.4) software was used to simulate the in-vitro potential field in a saline tank. The electrode was vertically placed in a saline tank modeled by a cylinder 7 cm in diameter and 7 cm in height. In the simulation, the conductivity of 0.9% sodium chloride solution was 1.5 S/m, mimicking the tissue fluid of the brain [4.3]. The electrode was tested in both monopolar and bipolar configurations. For monopolar stimulation, one of the electrode sites was excited with a

1 V-DC signal; the boundaries of the saline tank were grounded. For bipolar stimulation, an electrode site near the distal end was excited with a 1 V-DC signal; another site near the proximal end on the same side piece was grounded. In order to model the double layer impedance at the interface between the electrode surface and the electrolyte, an additional 0.1 mm-thick film was placed on the surface of the electrode sites. It was found that only the ratio of the electrode-saline interface layer thickness and conductivity needed to be estimated to find a unique solution [4.4]. Therefore, in this work, the interface layer thickness was fixed to 0.1 mm and the interface conductivity was adjusted to fit the test data and the simulation result. These results are presented in the next section.

The bipolar model was also used to estimate the electrode impedance between two electrode sites by dividing the applied voltage by the applied current calculated by integrating the current density over the electrode contact surface area.

45

4.3.2 In-vitro Test Environment

For the in-vitro experiment, the electrode was inserted from the bottom of a cylindrical glass container and placed vertically, approximately centered, as shown in

Fig. 4.6. The container was 7 cm in diameter and 7 cm in height, and was filled with

0.9% sodium chloride. The recording electrode, a 170 mm-long Epoxylite-coated tungsten microelectrode (FHC Inc., Bowdoin, ME), was placed from the top. Translation of the electrode was accomplished using a hydraulic oil micromanipulator (MO-95,

Narishige International USA, East Meadow, NY). A biphasic square wave pulse generator

(BPG-1, Bak Electronics, Mount Airy, MD) and a biphasic stimulus isolator (BSI-1, Bak

Electronics) were used to generate stimulation pulses. The stimulation waveform was 1 V amplitude, 500 µs cathodic pulse, followed by 500 µs pulse-interval delay, and then followed by 500 µs anodic pulse, with a 7 ms period as Fig. 4.7, Trace 1 shows. For monopolar stimulation, a stimulus was applied across an electrode site and a stainless steel wire wound around the beaker wall. The stainless steel wire served as the return electrode. A silver/silver chloride wire (reference recording electrode) was placed near the stainless steel wire as shown in Fig. 4.6(a). For the bipolar setting, the stimulus was applied from electrode Site #1 to Site #3. A stainless steel wire was wound around the beaker wall and served as a reference electrode for signal recording as shown in

Fig 4.6(b). On the recording side, a differential amplifier with a headstage (Model 3000,

A-M Systems, Carlsborg, WA) was used to amplify and filter the recording signal with

100x gain and 0.1 Hz – 20 kHz band-pass filter. The recording electrode was brought to the vicinity of each electrode site by moving in the x-y plane to map the potential fields in proximity of the four sides of the 3D electrode as shown in Fig. 4.8. The electric

46

potential was also recorded along the z-axis over the active site, 1 mm away from it.

The electrode impedance was measured in the bipolar stimulation test. Since the applied voltage was a known value, the impedance could be derived as long as the total current is measured. Thus, a current monitor resistor was added to the bipolar stimulation circuit in series with the electrode as shown in Fig. 4.9. No recording electrode or amplifier was needed.

47

Recording Recording + Electrode Signal - Differential Amplifier

Reference Active Electrode Electrode site Return Electrode Biphasic Assembled Voltage Stimulator Electrode z

+– x Flexible Lead y (a)

Recording Recording + Electrode Signal - Reference Electrode Differential Amplifier

Electrode Site #1(active)

Electrode Site #3(return) Assembled Electrode Biphasic Voltage Stimulator z

+ – x y Flexible Lead

(b)

Figure 4.6: The in vitro experiment setup (not shown to scale): (a) Monopolar stimulation; and (b) bipolar stimulation.

48

Trace 1

Trace 2

Figure 4.7: The stimulus (Trace 1) and recorded waveform (Trace 2) in the in-vitro test. The recorded signal was amplified 100x in this figure.

x

y

Figure 4.8: The mapping region in the measuring plane.

49

Current Monitor Biphasic Oscilloscope Resistor Voltage Stimulator + –

Electrode Impedance

Figure 4.9: Bipolar stimulation circuit for electrode impedance measurement.

4.3.3 In-vitro Test and Simulation Result and Discussion

Trace 2 in Fig. 4.7 shows the recorded signal from bipolar test. Since the interface impedance is not purely resistive, the recorded signal is not identical in pulse-shape to the stimulus. Indeed the recorded signal was parabolic and decayed toward 0 V with time constant equal to 140 µs (estimated from the curve of Trace 2 in Fig. 4.7). The peak of the cathodic phase was recorded and is analyzed in this section. Figure 4.10 shows the electric potential distribution in the x-y measurement plane from monopolar simulation.

Comparisons of the monopolar and bipolar testing results are plotted in Fig. 4.11. The results of both the monopolar and bipolar stimulation show that, within a 3 mm radius, the electric potential in front of the activated side (Fig. 4.11(a) and (e)) was at least 5 times larger than the three other non-activated sides (Fig. 4.11(b)-(d) and (f)-(h)).

However, the recorded voltage was much lower than the applied stimulus due to the large frequency-dependent, electrode-electrolyte interface impedance between platinum and saline [4.3] [4.4]. As stated before, the interface was modeled as a 0.1 mm-thick layer in

COMSOL Multiphysics and its conductivity was adjusted to be 0.025 S/m for monopolar

50

stimulation and 0.05 S/m for bipolar stimulation to fit the testing result. The misalignment of the electrode to vertical orientation may be the reason of the measured potential difference in Regions C and D in bipolar stimulation.

Figure 4.12 shows the electric potential distribution along the z-axis. The measured potential 1 mm away above the center of the active electrode site in Fig. 4.12 is slightly lower than the data at the same point in Fig. 4.11(a). Recording electrode positioning and impedance changing may be reasons for the difference since these measurements were not performed on the same day.

y

x

Figure 4.10: The x-y contours of the electric potential distribution from monopolar simulation

51

(a) Simulation and Test Results in Region A (e) Simulation and Test Results in Region A

80 70 70 Test Result (Monopolar) 60 Test Result (bipolar) 60 Simulation (0.025 S/m) 50 Simulation (0.05 S/m) 50 40 40 30 30 20 Voltage (mV) Voltage 20 (mV) Voltage 10 10 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Distance From the electrode surface (mm) Distance From the electrode surface (mm) (b) Simulation and Test Results in Region B (f) Simulation and Test Results in Region B

4.5 3.5 4.0 3.0 Test Result (Bipolar) 3.5 2.5 3.0 Simulation (0.05 S/m) 2.5 2.0 2.0 1.5 1.5

Voltage (mV) Voltage (mV) Voltage 1.0 1.0 Test Result (Monopolar) 0.5 0.5 Simulation (0.025 S/m) 0.0 0.0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Distance From the electrode surface (mm) Distance From the electrode surface (mm) (c) (g) Simulation and Test Results in Region C Simulation and Test Results in Region C

9 7 8 Test Result (Monopolar) 6 Test Result (Bipolar) 7 Simulation (0.025 S/m) 5 Simulation (0.05 S/m) 6 5 4 4 3 3 2 Voltage (mV) Voltage Voltage (mV) Voltage 2 1 1 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Distance From the electrode surface (mm) Distance From the electrode surface (mm) (d) (h) Simulation and Test Results in Region D Simulation and Test Results in Region D

9 7 8 Test Result (Monopolar) 6 Test Result (Bipolar) 7 Simulation (0.025 S/m) 5 Simulation (0.05 S/m) 6 5 4 4 3 3 2 Voltage (mV) Voltage 2 (mV) Voltage 1 1 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Distance From the electrode surface (mm) Distance From the electrode surface (mm)

Figure 4.11: Comparison of simulation and test results: (a) monopolar stimulation in region A; (b) monopolar stimulation in region B; (c) monopolar stimulation in region C; (d) monopolar stimulation in region D; (e) bipolar stimulation in region A; (f) bipolar stimulation in region B; (g) bipolar stimulation in region C; and (h) bipolar stimulation in region D.

52

(a)

Depth (mm) 3 1mm 2 1 0 1mm -1 -2 -3

z

x y (b)

18 Test Data 16 14 Simulation Data 12 10 8 6 4 2 0 z -3 -2 -1 0 1 2 3 Distance Along Z-axis from the Center of the Activated Electrode (mm) (c)

x

z

Figure 4.12: Electric potential distribution along the z-axis: (a) the measurement points were 1 mm away from the electrode surface and spaced 1 mm; (b) measured peak cathodic voltage and the simulation data at the measurement points; (c) corresponding potential contours from simulation.

53

The impedance between two adjacent electrode sites was calculated by measuring the voltage across the monitor resistor in Fig. 4.9. The applied voltage was 1 V, the measured monitor voltage was 300 mV, and the monitor resistor was 998 Ω. Thus, the total current and the electrode impedance were calculated to be 0.3 mA and 2,329 Ω. The electrode impedance was almost twice larger than the Medtronic DBS electrode, which is around 1 kΩ [4.5]. This is contributed from the smaller electrode site area and thinner interconnect metal layer used in this design.

The stimulus signal diminished after several cycles of on and off or several hours of operation, for almost every activated site on the test electrode. As Fig. 4.13 shows, there is a break in the 10 µm-wide metal trace on the electrode side piece.

Electromigration and surge spike were considered to be the reasons. Electromigration may occur with high current density or temperature gradient. The designed maximum current density of the platinum trace during the testing process was 104 A/cm2 which was lower than the normal IC operating value, 106 A/cm2 [4.6]. However, a large current spike may happen while changing the pulse direction or switching on and off. Further research on electromigration and surge prevention is not included in this thesis.

54

Figure 4.13: A SEM photograph of a broken trace on the electrode.

55

Chapter 5: Conclusion and Future Work

5.1 Conclusion

A novel DBS electrode with the ability to steer the stimulation field circumferentially was designed and fabricated in this thesis. A unique assembly technique is used for this 3D electrode, which enables multi-surface stimulation. Since the implanted electrode will be in direct contact with soft tissue of the brain, a flexible lead was used to connect the electrode and the signal processing circuits. The mechanical strength of the epoxy stud used for lead attachment is verified by a pull test, where the epoxy stud withstood at least 8.7 gram-force applied along a 60 angle with reference to the bond pad surface. Resistance measurements of individual electrodes showed that the resistances of the metal traces on the electrode side pieces and the flexible lead were the main contributors to the resistance of the electrical path. The measured data also indicated manual epoxy dispensing lacked electrical repeatability. Thus, an automatic dispenser is recommended for repeatable epoxy volume dispensation.

Simulation and measurement results demonstrate that the electric potential in vicinity of the activated side is at least 5 times larger than the other three not-activated sides within a 3 mm radius. This further demonstrates the ability of the device to concentrate the stimulation field in a targeted direction, while minimizing the stimulation of non-targeted areas. Using this platform, partial selective stimulation study of the subthalamic nucleus (STN) is also possible considering the size of electrode and applying a simple current steering technique. This will enable direct and improved therapeutic stimulation with reduced side effects, thereby greatly enhancing patient quality-of-life.

56

5.2 Future Work

For better characterization and modeling of the electrode including the electrode-electrolyte interface, a test setup should be designed. This will provide a guideline for any future design, since this interface contains both capacitive and resistive components which their value dependent on the frequency and the shape and size of the electrode. On the other hand, the current handling capability of the metal traces on the electrode should be investigated to produce enough stimulation at each site while producing a long lifetime for the electrode. Since small electrodes will require a large current density which in turn reduce the life time of the electrode due to electromigration or delamination. Future work must focus on fabrication repeatability, operational reliability, and efficacy demonstration of the electrode. The reactions and compatibilities with the surrounding tissue must also be studied for future in-vivo tests.

57

Appendix 1: Electrode Fabrication Process Details

The mask aligner for exposure was a Karl Suss MA6/BA6. The settings were channel 2, 15 mWatt, and hard-contact with a 10 m-wide gap. HMDS was performed before spinning on PR. (Program #1 for bare silicon; #3 for other substrates.) Two different PR processes were used as noted in Table A1.1 below.

Name AZ9260 AZ9245 Thickness (m) 10 5 Spin rate (rpm) 2000 3500 Spin time (sec) 60 30 Soft bake temperature 95 110 (°C) Soft bake time (sec) 60 60 (on hot plate) Exposure time (sec) 30 22 Developer 400K 400K Develop time (min : sec) 4 : 30 2 : 30 Hard bake none none Table A1.1: The PR recipes used in the process.

Table A1.2 presents the settings for the bottom insulation layers deposition. RCA clean was performed before the furnace runs. However, since the processes of oxide/nitride/oxide are continuous, thus, no cleaning steps were performed between each step.

58

Material Low Temperature Oxide Low Stress Nitride Thickness (Å) 3000 1500 Recipe number #400 #140

Gas SiH4, O2 SiH2Cl2, NH3 Temperature (°C) 450 850 Pressure (mTorr) 5000 5000 Deposition rate (Å/min) 179 20 Deposition time (min) 17 70 Table A1.2: The recipes for bottom insulator deposition.

Table A1.3 presents the parameters for the metal deposition. A Discovery 18 from

Denton Vacuum was used for sputtering. The surface was roughened by plasma for better adhesion. Titanium was used as the adhesion layer between the oxide layer and the platinum. 5 m AZ9245 PR was applied and patterned by mask 1 before sputtering, and the lift-off process was accomplished by removing the PR after immersing the wafer into acetone over night.

Metal layer Titanium Platinum RF power (Watt) 125 NONE RF time (sec) 30 NONE Sputtering power (Watt) 250 250

Gas 30% O2, 70% Ar 30% O2, 70% Ar Deposition pressure (mTorr) 2 2 Deposition rate (Å/sec) 3 5.4 Deposition time (sec) 33 556 Thickness (Å) 100 3000 Table A1.3: The recipes for metal sputtering.

59

Table A1.4 presents the parameters for the top insulation layer deposition. This layer was oxide from TEOS by PECVD. The PECVD tool used in this work was Ultra

Dep 1000 from TYMKON, Inc. PECVD is a low temperature process and does not harm the existing metal layers. The deposition rate of TEOS was roughly 77 Å/sec. Thus, the deposition time was 40 sec for a 3000 Å-thick oxide layer. The deposition temperature was 365 °C.

Range Set value (%) Working condition

N2 2 L/min 50 1 L/min

O2 1 L/min 90 0.9 L/min TEOS 1.35 mL/min 50 0.675 mL/min HF Gen-ON 1000 W 30 300 W (13.56 MHz) LF GEN-ON 1000 W 10 100 W (400 kHz) Table A1.4: The recipe of the top TEOS layer deposition.

Helium 10 sccm Fluoroform (Freon-23) 4 sccm RF power supply 450 Watt Chamber pressure 200 mTorr Etching time 8 min Etching rate 176.3 Å/min to 431.6 Å/min Table A1.5: The recipe for the 3000 Å-thick oxide etching by RIE.

60

The final fabrication step was etching the outline. Here, 10 m-thick AZ9260 PR was used again and defined by Mask 3. To etch through the 250 m-thick silicon wafer by DRIE, it was necessary to pattern the top insulation layers (6000 Å-thick oxide /

1500 Å-thick nitride / 3000 Å-thick oxide) first by RIE. The top 6000 Å-oxide layer took

16 min to etch. The etch time was split (7 min, 7 min, 2 min) to improve uniformity of etching. The 1500 Å-thick nitride layer was etched by a LAM etcher with 35 sec etch time. The parameters of nitride etching are listed in Table A1.6. Due to poor etch uniformity and selectivity, the etch time for the bottom 3000 Å-thick oxide layer was only 6 min.

Recipe number #002 Pressure 300 mTorr RF power 200 Watt Gap 1.5 cm Helium 150 sccm

SF6 150 sccm Etching rate 3476 Å/min Etching time 35 sec Table A1.6: The recipe for the 1500 Å-thick nitride etching by RIE.

Due to lack of good selectivity during the insulation layer patterning, the PR on the wafer was re-applied and defined by mask 3 again before the final DRIE step. The

DRIE used was an Inductively Coupled Plasma (ICP) tool from Surface Technology

Systems, Inc (Newport, UK). The etch process must be monitored closely because the etch rate near the center was higher than the edge. Over-etching damages the surface features if the masking PR is etched through (Fig. A1.1). The DRIE step to define the

61

shanks takes 121 min, using a 12 sec etch and 8 sec passivation cycle.

Recipe Cwru_mil_200 Time 121 min cycles 363 Etching time 12 sec Passivation time 8 sec Start Etch End Passivation Type Discrete Base Pressure 0 mTorr Pressure Trip 94 mTorr He flow pressure 9500 mTorr

Etch Passivation Gas name Flow (sccm) Tol (%) Flow (sccm) Tol (%)

C4F8 0 5 85 10

SF6 130 10 0 5

O2 13 10 0 5 Ar 0 5 0 5 Table A1.7: Parameters used for DRIE etching of the side pieces.

62

Figure A1.1: An electrode over-etched during the DRIE step, resulting in the loss of the masking PR and damage to the surface features.

63

Reference

[1.1] K. C. Cheung, ”Implantable microscale neural interfaces,” J. Biomed. Microdevices,

vol. 9, no. 6, pp. 923-938, Dec. 2007.

[1.2] D. Chen, et al., “Advances in neural interfaces: report from the 2006 NIH Neural

Interfaces Workshop,” J. Neural Eng., vol. 4, pp. S137-S142, Sep. 2007.

[1.3] S. Miocinovic, “Theoretical and experimental predictions of neural elements

activated by deep brain stimulation”, Ph.D. thesis, Case Western Reserve University,

August 2007.

[1.4] Datasheet of Lead kit 3387/3389 for Deep Brain Stimulation from Medtronic, Inc.

[1.5] United States Patent 5843148, “High resolution brain stimulation lead and method

of use”

[1.6] Datasheet of Extension kit 7482A for Deep Brain Stimulation from Medtronic, Inc.

[1.7] C. R. Butson and C. C. McIntyre, “Role of electrode design on the volume of tissue

activated during deep brain stimulation”, J. Neural Eng., vol. 3, no. 1, pp. 1-8, Mar.

2006.

[1.8] C. Hamani, A. M. Lozano, “Hardware-related complications of deep brain

stimulation: a review of the published literature”, J. Stereotactic and Functional

Neurosurgery, vol. 84, no. 5-6, pp. 248-251, Nov. 2006.

[1.9] N. I. Maluf, D. A. Gee, K. E. Petersen, and G. T. A. Kovacs, “Medical applications

of MEMS”, WESCON/'95. Conference record. 'Microelectronics Communications

Technology Producing Quality Products Mobile and Portable Power Emerging

Technologies', pp.300-306, Nov. 1995.

[1.10] K. D. Wise and J. B. Angell, “A low-capacitance multielectrode probe for use in

64

extracellular neurophysiology,” IEEE Trans. Biomed. Eng., vol. BME-22, pp.

212–219, May 1975.

[1.11] K. D. Wise, D. J. Anderson, J. F. Hetke, D. R. Kipke, and K. Najafi, “Wireless

implantable microsystems: high-density electronic interfaces to the nervous

system”, Proc. IEEE., vol. 92, no. 1, pp. 76-97, Jan. 2004.

[1.12] 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 Trans. Biomed. Eng., vol. 38, no. 8, pp.

758-768, Aug. 1991.

[1.13] A. Branner, R. B. Stein, and R. A. Normann, “Selective stimulation of cat sciatic

nerve using an array of varying-length microelectrodes”, J. Neurophysiology, vol.

85, no. 4, pp. 1585-1594, Apr. 2001.

[1.14] P. S. Motta and J. W. Judy, “Micromachined probes for peep brain stimulation”,

Microtechnologies in Medicine & Biology 2nd Annual Int. IEEE-EMB Special

Topic Conf., pp. 251-254, May 2002.

[1.15] S.Metz, M.O. Heuschkel, B. V. Avila, R. Holzer, D. Bertrand and Ph. Renaud,

“Microelectrodes with three-dimensional structure for improved neural

interfacing”, Proc. 23rd Annual Int. Conf. EMBS, pp. 765-768, 2001.

[1.16] R. S. Pickard, P. L. Joseph, A. J. Collins, and R. C. J. Hicks ,”Flexible

printed-circuit probe for electrophysiology”, J. Med. & Biol. Eng. & Computing.,

vol. 17, pp. 261–267, Mar. 1979.

[1.17] A. Hung, D. Zhou, R. Greenberg, and J. W. Judy, “Micromachined electrodes for

high density neural stimulation systems”, Micro Electro Mechanical Systems,

65

2002. the 15th IEEE Int. Conf., pp.56-59, Jan. 2002.

[1.18] T. Stieglitz, H. Beutel, M. Schuettler, and J.-Uwe Meyer, “Micromachined,

polyimide-based devices for flexible neural interfaces”, J. Biomed. Microdevices,

vol. 2, no. 4, pp. 283-294, Dec. 2000.

[1.19] H. P. Neves, G. A. Orban, M. Koudelka-Hep, T. Stieglitz, and P. Ruther,

“Development of modular multifunctional probe arrays for cerebral applications”,

Proc. 3rd Int. IEEE EMBS Conf. on Neural Eng., May 2007.

[1.20] A. E. Hess, J. Dunning, D. Tyler and C. A. Zorman, “Development of a

microfabricated flat interface nerve electrode based on liquid crystal polymer and

polynorbornene multilayered structures”, Neural Engineering, 2007. CNE '07. 3rd

Int. IEEE/EMBS Conf., pp. 32-35, May 2007.

[1.21] D. McCreery, A. Lossinsky, V. Pikov, and X. Liu, “Microelectrode array for

chronic deep-brain microstimulation and Recording”, IEEE Trans. Biomed. Eng.,

vol. 53, no. 4, pp. 726-737, Apr. 2006.

[1.22] C. Pang, J. G. Cham, Z. Nenadic, S. Musallam, Y. C. Tai, J. W. Burdick, and R. A.

Andersen, “A new multi-site probe array with monolithically integrated parylene

flexible cable for neural prostheses”, Engineering in Medicine and Biology

Society, 2005. IEEE-EMBS 2005. 27th Annual Int. Conf., vol., no., pp.7114-7117,

2005.

[1.23] A. E. Hess, J. L. Dunning, D. J. Tyler and C. A. Zorman, “A Polynorbornene-based

microelectrode array for neural interfacing”, Solid-State Sensors, Actuators and

Microsystems Conf., 2007. Trans. 2007, Int., pp. 1235-1238, June 2007.

[1.24] O. J. Prohaska, F. Olcaytug, P. Pfundner, and H. Dragaun ,”Thin-film multiple

66

electrode probes: Possibilities and limitations”, IEEE Trans. Biomed. Eng., vol.

BME-33, pp. 223–229, Feb. 1986.

[1.25] C. Xu, W. Lemon, C. Liu, “Design and fabrication of a high density metal

microelectrode array for neural recording”, J. Sensors and Actuators, vol. A 96,

no. 1, pp. 78-85, Jan. 2002.

[1.26] G. A. May, S. A. Shamma, and R. L. White, “Atantalum-on-sapphire

microelectrode array”, IEEE Trans. Electron Devices, vol. ED-26, pp. 1932–1939,

Dec. 1979.

[1.27] S. K. An, S. Park, S. B. Jun, C. J. Lee, K. M. Byun, J. H. Sung, L. S. Wilson, S. J.

Rebscher, S. H. Oh, and S. J. Kim, “Design for a simplified cochlear implant

system”, IEEE Trans. Biomed. Eng., vol. 54, no. 6, pp. 973-982, June 2007.

[1.28] D. J. Anderson, K. Najafi, S. J. Tanghe, D. A. Evans, K. L. Levy, J. F. Hetke, X.

Xue, J. J. Zappia, and K. D. Wise, “Batch-fabricated thin-film electrodes for

stimulation of the central auditory system”, IEEE Trans. Biomed. Eng., vol. 36,

no. 7, July 1989.

[1.29] J. Chen, K. D. Wise, J. F. Hetke, and S. C. Bledsoe, Jr., “A multichannel neural

probe for selective chemical delivery at the cellular level”, IEEE Trans. Biomed.

Eng., vol. 44, no. 8, Aug. 1997.

[1.30] D. McCreery, V. Pikov, A. Lossinsky, L. Bullara, and W. Agnew, “Arrays for

chronic functional microstimulation of the lumbosacral spinal cord”, IEEE Trans.

Neural Systems & Rehabilitation Eng., vol. 12, no. 2, June 2004.

[1.31] S. J. Tanghe, “Micromachined silicon stimulating probes with CMOS circuitry for

use in the central nervous system”, Ph.D. thesis, the university of Michigan, Ann

67

Arbor, 1992.

[1.32] J. F. Hetke, J. L. Lund, K. Najafi, K. D. Wise, and D. J. Anderso, “Silicon ribbon

cables for chronically implantable microelectrode arrays”, IEEE Trans. Biomed.

Eng., vol. 41, no. 4, April 1994.

[1.33] L. S. Robblee and T. L. Rose, “Electrochemical guidelines for selection of

protocols and electrode materials for neural stimulation” in “Neural Prosthesis:

Fundamental Studies”, W. F. Agnew and D. B. McCreery, Englewood Cliffs, NJ:

Prentice-Hall, Inc., 1990.

[1.34] Q. Bai, K. D. Wise, and D. J. Anderson, “A high-yield microassembly structure for

three-dimensional microelectrode arrays”, IEEE Trans. Biomed. Eng., vol. 47, no.

3, March 2000.

[1.35] J-Uwe Meyer, T. Stieglitz, O. Scholz, W. Haberer, and H. Beutel, ”High density

interconnects and flexible hybrid assemblies for active biomedical implants”,

IEEE Trans. Advanced Packaging, vol. 24, no. 3, Aug. 2001.

[2.1] 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 Trans. Biomed. Eng., vol. 38, no. 8, pp.

758-768, Aug. 1991.

[2.2] P. S. Motta and J. W. Judy, “Micromachined probes for peep brain stimulation”,

Microtechnologies in Medicine & Biology 2nd Annual Int. IEEE-EMB Special

Topic Conf., pp. 251-254, May 2002.

[2.3] S.Metz, M.O. Heuschkel, B. V. Avila, R. Holzer, D. Bertrand and Ph. Renaud,

“Microelectrodes with three-dimensional structure for improved neural interfacing”,

68

Proc. 23rd Annual Int. Conf. EMBS, pp. 765-768, 2001.

[2.4] G. A. May, S. A. Shamma, and R. L. White, “Atantalum-on-sapphire microelectrode

array”, IEEE Trans. Electron Devices, vol. ED-26, pp. 1932–1939, Dec. 1979.

[2.5] K. D. Wise, D. J. Anderson, J. F. Hetke, D. R. Kipke, and K. Najafi, “Wireless

implantable microsystems: high-density electronic interfaces to the nervous system”,

Proc. IEEE., vol. 92, no. 1, pp. 76-97, Jan. 2004.

[2.6] J Wand, M. Gulari, P. T. Bhatti, B. Y. Arcand, K. Beach, C. R. Friedrich and K. D.

Wise, “A cochlear electrode array with built-in position sensing”, MEMS 2005. 18th

IEEE Int. Conf., pp. 786 – 789, 2005.

[3.1] B. L. Gehman, “Bonding wire microelectronic interconnections”, IEEE Trans.

Components, Hybrids & Manufacturing Technology, vol. CHMT-3, no. 3, Sept.

1980.

[3.2] Editor: John G. Webster, “Medical Instrumentation, Third Edition”, Chapter 5, John

Wiley & Sons, Inc. New York, 1998.

[3.3] Unitek Miyachi Corporation, “Nd: Yag Laser Welding Guide”.

[3.4] J-Uwe Meyer, T. Stieglitz, O. Scholz, W. Haberer, and H. Beutel, ”High density

interconnects and flexible hybrid assemblies for active biomedical implants”, IEEE

Trans. Advanced Packaging, vol. 24, no. 3, Aug. 2001.

[3.5] A. E. Hess, J. Dunning, D. Tyler and C. A. Zorman, “Development of a

microfabricated flat interface nerve electrode based on liquid crystal polymer and

polynorbornene multilayered structures”, Neural Engineering, 2007. CNE '07. 3rd

Int. IEEE/EMBS Conf., pp. 32-35, May 2007.

[4.1] H. Kwon, et al., “Contact materials and reliability for high power RF-MEMS

69

switches”, MEMS 2007. 20th IEEE Int. Conf., pp. 231 – 234, Jan. 2007.

[4.2] G. G. Harman and C. A. Cannon, “The microelectronic wire bond pull test – how to

use it, how to abuse it”, IEEE Trans. Components, Hybrids & Manufacturing

Technology, vol. CHMT-1, no. 3, Sept. 1978.

[4.3] S. Miocinovic, et al., “Experimental and theoretical characterization of the voltage

distribution generated by deep brain stimulation,” Experimental Neurology, vol.

216, no. 1, pp. 166-176, Mar. 2009.

[4.4] R .W. de Boer and A. van Oosterom, “Electrical Properties of Platinum Electrodes:

Impedance Measurements and Time-domain Analysis”, J. Medical & Biological

Engineering & Computing, vol. 16, no. 1, pp. 1-10, Jan. 1978.

[4.5] S. Miocinovic, “Theoretical and experimental predictions of neural elements

activated by deep brain stimulation”, Ph.D. thesis, Case Western Reserve University,

2007.

[4.6] F. M. D’Heurle, “Electromigration and failure in electronics: An introduction”, Proc.

IEEE, vol. 59, no. 10, pp. 1409-1416, Oct. 1971.

70