Development of Amorphous Silicon Carbide Ultramicroelectrode

DEVELOPMENT OF AMORPHOUS SILICON CARBIDE ULTRAMICROELECTRODE

ARRAYS FOR NEURAL STIMULATION AND RECORDING

Felix Deku

APPROVED BY SUPERVISORY COMMITTEE:

______Dr. Stuart Cogan, Chair

______Dr. Joseph Pancrazio

______Dr. Shalini Prasad

______Dr. Mario Romero-Ortega

______Dr. Timothy Gardner

Felix Deku

This work is dedicated to my parents, Florence Afi Kerker

and Stephen Kodzo Dzremegah

DEVELOPMENT OF AMORPHOUS SILICON CARBIDE ULTRAMICROELECTRODE

ARRAYS FOR NEURAL STIMULATION AND RECORDING

FELIX DEKU, BS, MS

DISSERTATION

Presented to the Faculty of

The University of Texas at Dallas

in Partial Fulfillment

of the Requirements

for the Degree of

DOCTOR OF PHILOSOPHY IN

BIOMEDICAL ENGINEERING

THE UNIVERSITY OF TEXAS AT DALLAS

December 2018

ACKNOWLEDGMENTS

For the incredible support I enjoyed from the talented group of students, faculty, support staff, family, and friends during my graduate school years, I now duly acknowledge. I will forever be indebted to my mentor, teacher, and advisor, Dr. Stuart F. Cogan, for giving me the opportunity to complete this original research in his lab. I have enjoyed exceptional mentorship, guidance and training throughout the years spent working with him. His open-door policy, all ears approach, readiness to be available physically or over the phone when needed, and his tireless effort to develop my technical abilities, and to instill in his students to ‘look at the big picture’ will stay with me eternally. I have become a better person, an excellent researcher, an engineer and a scientist thanks to his ever willingness to show me the path. I have personally enjoyed favors and kindness from him and his family for which I am incredibly grateful. There are truly not enough words to describe how honored and excited I am to be trained and guided by him. I will just say thank you.

I also want to thank Dr. Timothy J. Gardner for being the silent voice behind my training and for being there from the very beginning. I don’t think I would have gotten this far, within this short period, without the support I enjoyed from him and his lab. I acknowledge the contribution of Drs.

Yarden Cohen, Alket Mertiri and Ben Pearre and all members of Dr. Gardner’s lab in Boston

University for helping develop and advancing the carbon fiber and amorphous silicon carbide ultramicroelectrode array technology.

I also want to acknowledge the support of Dr. Joseph J. Pancrazio and contributions from his lab in evaluating the in vivo performance of the microelectrodes developed. I have enjoyed some thought-provoking conversations with him, and I want to thank him for his unwavering support,

comments, and insightful discussions. I acknowledge Dr. Christopher Frewin for helping with the rat cortical implantation, neural recording, and histology. Dr. Bryan Black is acknowledged for spike sorting and signal analyses. I also recognize Allison Stiller for the COMSOL modeling and the insertion force experiments.

I want to thank Dr. Shalini Prasad for being a good teacher and for sparking my interest in microfabrication. Her graduate class on biomedical microdevices provided a solid background and was one of the reasons why I joined Dr. Cogan’s lab. I want to thank Dr. Mario I. Romero-Ortega for keeping me on my toes and never accepting less from me. I have enjoyed collaborative work with him and his lab and thank him so much for the healthy discussions.

This work would never have seen the light of day without the collective effort of undergraduates, graduate students and post-doctoral fellows at the Neural Device Laboratory of which I am proud to be a part. The seriousness that the lab members take their work and the ability to disseminate findings to colleagues quickly is a success story that should make us proud. I want to thank Dr.

Alexandra Joshi-Imre, for all her efforts throughout the microelectrode array development process.

Thank you for your contribution to process development, SEM inspections, electrochemical characterization and your ability to help me find the path whenever I am stuck with no way forward. You know how to get the best out of me, and I appreciate your support. Drs. Jimin Maeng,

Aswini Kanneganti, and Shakil Mohammed are acknowledged for their varied contributions to the success of my project. I also want to acknowledge my good friends in the lab: Atefeh Ghazavi,

Rebecca Frederick, Negar Geramifard, Justin Abbott, and Bitan Chakraborty for helping me and building my confidence. To Atefeh, I say thank you for all the tough questions in process development and electrochemistry, and for the useful discussions, we have had over the last three

and a half years. I also want to acknowledge the contribution of the undergraduates I mentored during my graduate school years, especially those of Saher Aqeel and Benjamin Ting towards the development of the amorphous silicon carbide arrays.

I acknowledge the contribution of the Department of Bioengineering support staff for helping make Ph.D. life as easy as it should be. I recognize cleanroom staffers Dr. Gordon Pollock and

Ronald Scott Riekena for their support and maintenance of the PECVD tool.

Finally, I want to acknowledge the support of my father, Stephen Dzremegah, my better half Anna

Deku, my sister Anne-Marie Deku, and all my other siblings, and all my friends for their kindness and support throughout these years. They never wavered in their encouragements, love, and belief in me. I want to say thank you.

July 2018

vii

DEVELOPMENT OF AMORPHOUS SILICON CARBIDE ULTRAMICROELECTRODE

ARRAYS FOR NEURAL STIMULATION AND RECORDING

Felix Deku, PhD The University of Texas at Dallas, 2018

ABSTRACT

Supervising Professor: Dr. Stuart Cogan

Interest in restoring lost function using neuro-prosthetic devices and treating neurological disorders or neurodegenerative diseases through electrical stimulation of neural activity has increased in recent years. For example, implantable cortical neural interfaces allow investigation of sensorimotor learning, and control of both natural and prosthetic limbs through recording of volitional intent and stimulation of neural activity. However, these interfaces decline rapidly in performance over chronic timescales. Foreign body response is believed to limit their recording and stimulation reliability. The resulting glial scar isolates the indwelling microelectrodes from healthy neuronal cells. The consequence is recording from large populations of weak neural signals and the requirement for high current amplitudes to deliver the necessary charge for neural activation.

Recently, carbon fiber microelectrodes with small cross-sectional dimensions (below 10 µm) have been shown to reduce insertion damage to neurons and microvasculature, minimize adverse tissue reaction, and provide stable neural recording over chronic timescales. Despite these achievements,

viii

the development of carbon fiber MEAs faces the issue of micro-assembly, micromanipulation, and the general lack of control of the geometric surface area (GSA) of the active sites.

This dissertation addresses these issues by developing ultrathin cellular or sub-cellular scale microelectrode arrays (MEAs) based on amorphous silicon carbide. Amorphous silicon carbide

(a-SiC) deposited by plasma enhanced chemical vapor deposition has similar mechanical properties to carbon fiber but is amenable to thin-film microfabrication methods, thus permitting a wide variety of designs, control of GSA, and batch fabrication of microelectrode arrays.

Challenges associated with residual stress control in the a-SiC and those associated with metal patterning needs to be addressed to use the a-SiC in ultrathin MEA designs. Also, implantation strategies for ultrathin MEA shanks and the burden of using small contact sites for electrochemical measurement, electrical stimulation and electrophysiology need to be addressed.

In this dissertation, microelectrode arrays based on a-SiC were fabricated, characterized for their electrochemical properties in a saline model of the interstitial fluid, and evaluated functionally in songbird and rat brain. We describe stress engineering in the multilayered structure to regulate the curvature of the a-SiC MEAs. Engineering challenges associated with process controls to produce penetrating probes of reduced cross-sectional shank dimensions are discussed. We have developed implantation strategies to insert ultrathin a-SiC MEAs into rat motor cortex. We show that a minimum a-SiC thickness of 6 µm is required to insert 2 mm long a-SiC MEAs shanks into rat cortex without the need for insertion guides or temporary support structures. Below this thickness, we demonstrate that a-SiC MEAs will require temporary support structures such as polyethylene glycol or in situ designs that increases the critical buckling load of the implanted shanks.

With the reduced shank dimensions, the electrode sites on the a-SiC MEA are small with high electrode impedance and low charge injection properties. We investigated low impedance coatings such as titanium nitride, sputtered iridium oxide and electrodeposited iridium oxide films as a means of improving the electrochemical performance for neural stimulation and recording. We show that cathodal charge injection capacities greater than 17 mC/cm2 can be achieved with the coated ultramicroelectrode site with appropriate biasing.

TABLE OF CONTENTS

ACKNOWLEDGMENTS ...... v

ABSTRACT ...... viii

LIST OF FIGURES ...... xii

LIST OF TABLES ...... xviii

CHAPTER 1 INTRODUCTION ...... 1

1.1 Motivation for reducing shank dimensions of penetrating microelectrodes ...... 2 1.1.1 Creation of high density microelectrode arrays ...... 2 1.1.2 Reduced foreign body response to indwelling cortical implants ...... 3 1.2 Flexural rigidity of indwelling microelectrode shaft ...... 3 1.2.1 Compliant microelectrode arrays ...... 4 1.3 Amorphous silicon carbide films for biomedical applications ...... 7 1.3.1 Properties of silicon carbide ...... 7 1.3.2 Stability of amorphous silicon carbide ...... 7 1.3.3 Utility of a-SiC as substrate and encapsulation for neural interface devices ……………………………………………………………………………………..8 1.4 Ultrathin carbon fiber microelectrode arrays ...... 9 1.4.1 Insertion of ultrathin carbon fiber electrodes into brain tissues ...... 9 1.4.2 Carbon fiber microelectrodes and neural recordings ...... 10 1.5 Ultramicroelectrodes for neural stimulation ...... 11 1.5.1 Low impedance coatings for ultramicroelectrodes ...... 11 1.6 Justification and Goals ...... 12 1.7 Dissertation organization ...... 13

CHAPTER 2 ELECTRODEPOSITED IRIDIUM OXIDE ON CARBON FIBER ULTRAMICROELECTRODES FOR NEURAL RECORDING AND STIMULATION …….16 2.1 Author Contributions ...... 17 2.2 Abstract ...... 17 2.3 Introduction ...... 18

2.4 Experimental ...... 20 2.4.1 Carbon fiber ultramicroelectrode fabrication ...... 20 2.4.2 Test Electrolytes ...... 21 2.4.3 Cyclic voltammetry and electrochemical impedance spectroscopy ...... 21 2.4.4 Voltage transient measurements...... 23 2.5 Results and Discussion ...... 24 2.5.1 Electrochemical Characterization ...... 24 2.5.2 Stimulation charge injection capacity ...... 27 2.5.3 EIROF charge transfer characteristics...... 28 2.6 Conclusions ...... 34

CHAPTER 3 ENGINEERING CHALLENGES ASSOCIATED WITH FABRICATION OF AMORPHOUS SILICON CARBIDE MICROELECTRODE ARRAYS………………………35 3.1 Participation ...... 36 3.2 Introduction ...... 36 3.3 Methods...... 38 3.3.1 Multilayer stress engineering ...... 38 3.3.2 Accelerated aging and characterization ...... 38 3.3.2.1. Sample preparation ...... 38 3.3.2.2. Electrochemical measurements ...... 39 3.4 Results and Discussions ...... 40 3.4.1 Multilayer stress engineering ...... 40 3.4.1.1. Stress control during thin-film deposition……………………………..40 3.4.1.2. Stress engineering in multilayer stack by thermal annealing………….43 3.4.2 Microfabrication challenges in a-SiC MEA technology ...... 44 3.5 Conclusions ...... 51

CHAPTER 4 AMORPHOUS SILICON CARBIDE ULTRAMICROELECTRODE ARRAYS FOR NEURAL STIMULATION AND RECORDING…………………………………………52 4.1 Author Contributions ...... 53 4.2 Abstract ...... 53 4.3 Introduction ...... 54 4.4 Methods...... 56

xii

4.4.1 PECVD a-SiC deposition ...... 57 4.4.2 Fabrication of a-SiC MEAs ...... 57 4.4.3 Low impedance electrode coatings ...... 59 4.4.4 Electrochemical characterization ...... 60 4.4.5 Acute implantation and neural recording ...... 61 4.5 Results ...... 63 4.5.1 Residual film stress of the a-SiC ...... 63 4.5.2 Amorphous-SiC MEA fabrication ...... 64 4.5.3 Electrochemical characterization ...... 67 4.5.4 Neural recording ...... 69 4.6 Discussion ...... 72 4.7 Conclusions ...... 77

CHAPTER 5 AMORPHOUS SILICON CARBIDE PLATFORM FOR NEXT GENERATIONPENETRATING NEURAL INTERFACE DESIGNS…………………………79 5.1 Participation ...... 80 5.2 Introduction ...... 80 5.3 Methods...... 82 5.3.1 Thin film deposition and array fabrication ...... 82 5.3.2 Buckling and insertion mechanics...... 83 5.3.3 Surgery and a-SiC implantation ...... 83 5.3.4 In Vivo Recording and Analysis ...... 84 5.4 Results and Discussion ...... 85 5.4.1 Insertion strategies of ultrathin a-SiC shanks into cortex ...... 85 5.4.1.1. PEG-stabilized shanks ...... 85 5.4.1.2. Bundled shanks ...... 86 5.4.1.3. Reduction of effective shank length ...... 87 5.4.1.4. Insertion of individual shanks ...... 88 5.4.2 Neural Recording ...... 91 5.4.3 Electrochemical characterization ...... 94 5.4.4 Novel microelectrode designs enabled by a-SiC platform technology ...... 94 5.4.4.1. Barb Probes ...... 94

xiii

5.4.4.2. Flexible floating probes ...... 96 5.4.4.3. 32-channel MEAs ...... 97 5.4.4.4. High density Arrays - 3D matrix arrays ...... 98 5.4.4.5. High density Arrays - Wedge probe ...... 99 5.5 Summary ...... 100 CHAPTER 6 SUMMARY OF FINDINGS …………………………………………………...102

APPENDIX A SUPPLEMENTARY MATERIAL FOR CHAPTER 4 ...... 105

REFERENCES ...... 107

BIOGRAPHICAL SKETCH ...... 126

CURRICULUM VITAE ...... 127

xiv

LIST OF FIGURES

Figure 1.1. A comparison of structural stiffness of various materials used in neural probe designs as a function of shank length...... 6

Figure 2.1. Cyclic voltammetry of EIROF-coated carbon fiber UMEs. The onset of oxidative current at 0.7 V is presumed to be water oxidation. 1(b). Representative voltage transient in response to 120 µA, 500 µs pulse giving a 60 nC/ph charge. Voltage transient measurements were performed in response to monophasic current pulses in three controlled phases: in the interpulse period (I), the electrode potential is controlled. During the cathodal current pulse, the voltage transient is measured (II), and finally, in order to record the potential of the equilibrating electrode after a pulse, there is a brief 1 ms inter-phase delay, when the electrode is at open circuit (III)...... 21

Figure 2.2. Scanning electron micrographs taken at 2.0 kV acceleration voltage on a Zeiss Supra 40 microscope. (a) Uncoated CFUME with its Parylene insulation removed a distance of 40 m from the tip. (b) EIROF coating on a tip appears nodular and porous. (c) Detail of the surface morphology of the EIROF film...... 24

Figure 2.3. Representative cyclic voltammetry (a), impedance curve (b) and phase angle (c) of uncoated and EIROF-coated CFUMEs. GSA= 600 µm2. The EIROF-coated CFUME electrochemical behavior is compared in two different electrolytes: phosphate buffered saline (PBS) and model interstitial fluid (model-ISF). EIROF deposition: 50 CV cycles at 50 mV/s between -0.05 V and 0.5V...... 25

Figure 2.4. Charge injection limits as a function of (a) bias and (b) pulse width at 0 mV interpulse bias...... 26

Figure 2.5. Voltage transients recorded as the electrode is polarized to different cathodal potential excursions (Ec) starting from a 0.6 V bias to an Emc of -0.6V...... 28

Figure 2.6. A comparison of VAT and VAL as a function of Ec (pulse width= 0.5 ms, frequency= 50 Hz, bias= 0.6 V). The leading and trailing phases of the access voltage deviate at Ec values more negative than 0.1 V, with VAT increasing faster than VAL...... 29

Figure 2.7. Impedance of EIROF coated CFUMEs as a function of DC Bias (a) EIS spectra as a function of frequency and (b) comparison of impedance magnitude from (a) at 1, 103 and 105 Hz versus DC bias...... 30

Figure 2.8. The time integral of the cathodal current shown by the shaded region of the voltammogram between the bias potential (Vbias) and the end potential (Eend) represents the equilibrium charge (QE). The inset shows the location of Vbias and Eend on a voltage transient...... 31

Figure 2.9. Comparison of the end potential (Eend), charge injected (Qinj) and equilibrium charge (QE) of EIROF coated CFUMEs as a function electrode cathodal potential (Ec). Frequency = 50 Hz, Pulse width = 0.5 ms, Vbias = 600mV. Ec=-0.6 V= Emc...... 32

Figure 3.1. Electrochemical measurement set up for the accelerated aging experiment. A glass vial is clamped onto electrode sites on the silicon wafer which is then filled with PBS for the measurement...... 37

Figure 3.2. An example of a-SiC MEA under unbalanced residual stress...... 38

Figure 3.3. Stress evolution during thin film depositions. I represents deposition of bottom a-SiC layer. II represents deposition of the metallization on the bottom a-SiC layer (a-SiC + metal) or silicon control wafer (metal only). III represents deposition of top a-SiC layer on the metallization...... 39

Figure 3.4. Stress evolution in multilayered a-SiC/metal/a-SiC stack during thermal annealing at o 400 C in N2. The metallization was deposited either by evaporation or sputtering...... 42

Figure 3.5. SEM images taken at a tilted viewing angle show thin and tall free-standing metal film remnants, at the edges of the traces. These film remnants are prevalent defects when lift- off is used with sputtered metallization. (a) Shows rabbit ears often at both edges of metal traces following lift-off process. (b) Clean lift-off (red ellipse) and bad lift-off with rabbit ear (black ellipse) ...... 43

Figure 3.6. SEM images taken from a tilted viewing angle show rabbit ear defects conformably covered by a-SiC film creates bands of a-SiC at the edges of the metal traces...... 44

xvi

Figure 3.7. SEM and optical images obtained after 9 weeks of aging at 87 oC show delamination of a-SiC covering the metal traces. We associate this failure mode with the rabbit ear defects in a-SiC MEA fabrication...... 45

Figure 3.8. EIS of a-SiC MEA with uncoated electrode sites fabricated with 400 nm sputtered Au soaked in PBS at 87oC for 9 weeks (a). Inset is the impedance curve before soaking. Comparison of the average impedance curve of samples soaked at 37oC and 87oC (b) as a function of weeks soaked in PBS...... 46

Figure 3.9. The a-SiC MEA soaked for 20 weeks in PBS at 87oC (a) before and (b) after soaking. Metallization was by vapor deposition...... 47

Figure 3.10. Accelerated aging of SIROF-coated MEAs created with sputtered or evaporated films soaked in PBS at 87oC showing (a) the average impedance measured at 1 kHz and (b) impedance measured 5 weeks after the samples created with sputtered Au were soaked. Measurement from the samples measured in (b) was discontinued after week 5...... 48

Figure 3.11. Comparison of (a) CV and (b) EIS of a-SiC MEAs soaked in PBS at 87oC for 17 weeks...... 49

Figure 4.1. Microfabrication of amorphous silicon carbide microelectrode arrays (a-SiC MEAs). The process flow features at least three photolithography steps: one for defining the metal traces and electrodes, a second for patterning the top a-SiC layer for electrode site and bond pad openings, and a third photolithography step to singulate the a-SiC device geometries. A fourth lift-off lithography step is used (not shown) to restrict deposition of SIROF or porous TiN low impedance coatings to the electrode sites...... 57

Figure 4.2. Surface morphology and topography of a-SiC films. (a) AFM image (2µm x 2µm) and (b) SEM images show the surface morphology of 2 µm thick PECVD amorphous-SiC deposited on a silicon wafer. The surface roughness estimate from the AFM is below 4 nm rms...... 62

Figure 4.3. SEM images show released bundles of a-SiC MEAs (a, c, and e) taken at 5kV, and shanks still attached to the carrier silicon wafer (b, d, f, and g) taken at 2kV acceleration voltage. (a) The shanks of a released a-SiC MEA form a bundle when drawn out of water. The tip of the bundle is shaped by the layout design of the shank arrays, as shown in (b-c and d-e), (f) shows the exposed electrode tip at the distal end of the shank and (g) shows the side wall profile of the exposed electrode site at a 25-degree viewing angle...... 64

xvii

Figure 4.4. Optical micrographs show that the 16 shanks naturally bundle when the as-fabricated device is pulled out of the deionized water. Omnetics connectors were mounted on the arrays using a solder reflow process and medical grade epoxy. The Figure shows (a) the as-fabricated a-SiC MEA after release from deionized water, (b) after an Omnetics connector is soldered onto the bond pads and (c) a packaged device for implantation or in vitro electrochemical characterization...... 65

Figure 4.5. Electrochemical properties of a-SiC UMEs coated with SIROF, TiN and Pt compared with Au. (a) Cyclic voltammogram measured at 50mV/s between -0.6V and 0.8V limits. (b) Electrochemical impedance spectroscopy measured using a 10 mV rms AC sinusoid...... 67

Figure 4.6. Electrical stimulation capabilities of ultramicroelectrodes (a) Voltage transient response to current waveforms for TiN and SIROF electrodes biased at 0.6V vs Ag|AgCl (solid lines) and without anodic bias (dash lines). The electrodes were polarized to a cathodal potential limit of -0.6 V. The average current passed across the interface within the ‘safe’ electrochemical limit is 86.4 µA for SIROF (0.6 V bias) and 40.2 µA for TiN (0.6 V bias). (b) Maximum charge injection capacity and charge per phase as a function of interpulse bias. (Frequency= 50 pps, Pulse width = 200 µs)...... 68

Figure 4.7. Acute neural recording immediately following implantation in basal ganglia of Zebra Finch brain. The 16 recorded channels showed no strong coupling between contacts (a) single channel acute voltage trace with a subcutaneous reference on the head and (b) an overlay of a neuronal spike waveforms, detected by setting the threshold of the trace in (a) at – 50 µV...... 69

Figure 4.8. Spontaneous neural activity in rat motor cortex: (a) simultaneous spike activity recorded across three channels using the a-SiC MEA; (b) sorted single units on CH1 with average peak-to-peak amplitudes of 45 µV (Unit A), 87 µV (Unit B) and 114 µV (Unit C); and (c) the corresponding autocorrelograms processed with a bin size of 2 ms...... 70

Figure 4.9. A SEM image of the distal tip of an a-SiC UME shank with two electrode sites located on the same shank. The GSA of the exposed Au electrode sites is 100 µm2 but with unequal perimeter. The perimeter of the square electrode site is 40 µm versus 104 µm for the rectangular site...... 72

Figure 4.10. A SEM image of a-SiC MEA with a built-in shank curvature. The intrinsic curvature is expected to improve splaying capabilities of the a-SiC MEAs which form bundles when drawn out of water...... 73

xviii

Figure 4.11. Raw traces showing simultaneously recorded spontaneous spike activity on channel 10 and 11 (CH10 and CH11) and their corresponding local field potentials (FP10 and FP11). (Spike and local field potential amplitudes in mV)...... 75

Figure 5.1. Insertion of PEG-stabilized a-SiC MEA into rat motor cortex. The PEG temporarily provides mechanical support to the 4 µm thick a-SiC shanks prior to insertion. An insertion rate of 50 µm/s ensures that the PEG completely dissolves are the array is advanced into the brain...... 84

Figure 5.2. Bundles of 16-channel a-SiC MEA when drawn out of water (a) showing (b) tip geometry. Insertion of a bundled 8-channel a-SiC array into rat cortex (c)...... 85

Figure 5.3. Webbed a-SiC MEA with an effective shank length of 1 mm and features a return electrode as part of the MEA structure ...... 86

Figure 5.4. Buckling test. Insertion force measured when a single shank a-SiC probe is lowered against a glass surface (a). An image of the buckled state of the shank (b) and a COMSOL prediction of the buckled state (c)...... 87

Figure 5.5. Insertion forces recorded during the insertion of a 6 µm x 7 µm a-SIC shank into rat cortex. An insertion force of 0.35 mN was recorded at the point of insertion. Inset shows forces experienced by the indwelling shank inserted 2 mm into the brain surface...... 88

Figure 5.6. Acute extracellular action potentials recorded using 6 µm a-SiC MEAs. (A) Filtered continuous data traces from three representative electrodes on Array 1. Vertical and horizontal scale bar represents 125 µV and 1.75 s, respectively. (B) Left - Representative 2D principal component space indicating clear separation from the noise (central gray cluster). Right – Associated single units, indicating characteristic extracellular waveform shape. Vertical and horizontal scale bar represents 175 µV and 0.6 ms, respectively...... 90

Figure 5.7. Average waveforms of sorted single units detected on SIROF-coated a-SiC arrays 14 days post-implantation in rat cortex ...... 91

Figure 5.8. Electrochemical properties of a-SiC UMEs coated with SIROF (a) Electrochemical impedance spectra and (b) Cyclic voltammograms measured at 50 mV/s between -0.6 V and 0.8 V limits 1day post implantation ...... 92

xix

Figure 5.9. Barb Probe showing linear arrangement of electrode sites on hooks attached to a central spine. Electrode sites are located on the tips...... 93

Figure 5.10. Flexible floating probes designed to isolate the active sites from the probe body ....95

Figure 5.11. 32-channel a-SiC MEA has eight electrode sites on each shank...... 96

Figure 5.12. Development of high density a-SiC MEAs. A 16-channel, 4-shank a-SiC MEA (a) developed for intracortical neural interfacing with 4 electrodes distributed in a tetrode configuration towards the distal tips (b). 400 µm thick SU-8 spacers are used to stack multiple devices on each other creating a 3D geometry (c-d)...... 97

Figure 5.13. A single shank 64-channel a-SiC probe with electrode sites distributed along the edge of the probe in a diagonal fashion. The electrode site is SIROF-coated and 200 µm2 in geometric surface area...... 98

Figure A.1. Process flow for the application of Indium –Tin solder paste prior to the solder reflow process...... 103

Figure A.2. Possible device failure due to solder reflow processing. (a) Shows no delamination of metal pads and (b) shows delamination after the solder reflow process...... 104

LIST OF TABLES

Table 1.1. Comparison of flexural rigidity for different probe designs recently reported in the literature ...... 5

Table 3.1. Effect of sputtering pressure on metal stress and a-SiC MEA curvature post-fabrication. The deposition times were adjusted so that a nominal metal thickness of 400 nm was achieved...... 39

Table 5.1. Active electrode yield (AEY) percentage, total number of units, mean peak-to-peak amplitude, RMS noise, and SNR per array and cumulative values across all arrays...... 88

xxi

CHAPTER 1

INTRODUCTION

Brain-computer interfaces (BCIs) generally exploit the electrical properties of the nervous system.

These interfaces rely on neural signals recorded to actuate external devices or prosthetic limbs. In laboratory settings, BCIs have allowed people with severe paralysis, including those with tetraplegia and locked-in syndrome, to physically interact or communicate with their environment

[1]–[6]. The majority of BCIs and mainly those available commercially are fabricated from silicon and leverage techniques like those used in integrated circuit microfabrication and packaging. The most common silicon-based multielectrode array (MEAs), the Blackrock arrays, are being investigated clinically in the BrainGate system to provide volitional control of prosthetic devices in patients with spinal cord injury, brainstem stroke, or ALS [6]. These devices are also used in clinical trials to investigate restoration of touch sensation by stimulation of the somatosensory cortex in individuals with prosthetic limbs [7].

Of considerable interest to the research and scientific community is the development of MEAs with larger numbers of recording channels and increased spatial selectivity of neural recordings, notably for small-animal research models [8] and for longer-term non-human primate studies [9].

Scaling up neural contacts allows an investigation into neural circuitry, brain connectivity and neuroplasticity which are pertinent to the fundamental understanding of neurological disorders, brain and spinal cord injuries, and neurodegenerative diseases. Increasing the number of electrode sites while minimizing their invasiveness increases the spatial resolution for recording and for stimulation with chronically implanted arrays.

These MEAs must consistently record neural activity with high fidelity or be able to inject functional levels of charge over chronic timescales or both. Other considerations include; the primary material for electrode construction must withstand the hostile in vivo microenvironment, endure mechanical deformations induced by localized stresses from surrounding tissue and, if insulated, must not delaminate from its encapsulation. Delamination creates additional leakage pathways that impact the viability and chronic functionality of neural interfaces [10].

1.2 Motivation for reducing shank dimensions of penetrating microelectrodes

1.2.1 Creation of high density microelectrode arrays

Penetrating microelectrodes are required to have adequate structural rigidity to penetrate the neural tissue but retain adequate flexibility to minimize mechanical mismatch with tissue when implanted. These requirements are compounded by the lack of chemical stability, and fragility of silicon, which makes developing extremely small geometries based on silicon exceptionally challenging. As a result, the state-of-the-art silicon based MEAs, such as the BlackRock arrays, are typically stiff with shank cross-sectional areas exceeding 500 µm2. The tip-to-tip separation between recording sites is about 200-400 µm. This device configuration limits the creation of high density arrays for neuroscience research and clinical neural stimulation and recording applications.

For example, a 0.5 mm3 volume of the motor cortex, containing approximately 5000 individual neurons (given an estimated neuronal density of 9.9 x 106 neurons/cm3) [11], [12] will be sampled using only 4 (or 8) electrodes on the BlackRock array (with 400 µm pitch).

1.2.2 Reduced foreign body response to indwelling cortical implants

The reduced cross-sectional dimensions also have physiological benefits. The cross-sectional area of the individual shanks generally corresponds to reduced damage to neural tissue and rupture of microvasculature during array implantation [13], [14]. Blood is a source of chemical factors such as interleukin -1 [15], tumor necrotic factor alpha [16], [17] and monocyte chemotactic protein -1

[16], [18]. In the brain, the release of these proteins due to insertion damage causes changes in the function and structure of two native glial cells – microglia and astroglia [19], [20]. Activated microglia and astroglia further recruit more glial cells resulting in a cascade of cell activation and the formation of glial scar that encapsulates and isolates the implanted array [16], [21] 50 to 200

µm from excitable neural tissue [22]–[25]. Astrocytes, identified by glial fibrillary acidic protein

(GFAP), at the initial stages of foreign body response, form cellular processes that infiltrate and surround the implanted array. As a result, microelectrode arrays tend to lose functionality within weeks of implantation, measurable by loss of readouts of brain activity by these arrays. Progress in understanding of the brain tissue response and the availability of advanced microfabrication techniques indicates that the next generation of chronically stable microelectrode arrays should have reduced geometric footprints to reduce damage during implantation and elicit little or no deleterious host response.

1.3 Flexural rigidity of indwelling microelectrode shaft

The use of a material with low degree of stiffness can moderate the impact of mechanical mismatch and micromotion on chronic inflammatory response [26]–[28]. Polymers such as polydimethylsiloxane (PDMS) [29], SU-8 [30], Parylene C [31]–[35] or polyimide [36]–[41] have

been explored as substrates for developing flexible neural probes because their Young’s modulus of ~2- 4 GPa is relatively closer to that of neural tissue than silicon. Other materials such as shape memory polymers or other nano-composites that soften after implantation are also being explored

[28], [42]–[44]. Since the flexural rigidity of an implant depends upon both geometry and material composition [45], [46], to penetrate the brain tissue, the polymer based MEAs typically require nominal thicknesses of about 15-300 µm. This thickness is above the minimum threshold estimated to elude a chronic foreign body response [47], [48].

1.3.1 Compliant microelectrode arrays

Compliant intracortical microelectrodes can be produced using extremely soft materials or ultrathin geometries. The material property is determined by the Young’s modulus (E) and the geometry determines the moment of inertia of the cross-section (I). The product of these quantities

(EI) is related to the structural stiffness (k) of the electrode shaft by the reciprocal of the compliance (kc) when the shaft is modelled using Euler- Bernoulli beam theory (classical beam theory) as in Equation 1.1. Literature suggests compliant microelectrodes can be created easily by reducing the cross-sectional dimensions. This is because the moment of inertia is directly proportional to the third power of the thickness in the bending direction for a shank with rectangular cross-section (Equation 1.2) and the fourth power of the radius for a shank with circular cross-section (Equation 1.3).

3퐸퐼 1 푘 = 3 = (Equation 1.1) 푙 푘푐

푤푡3 퐼 = (Equation 1.2) 12

휋푟4 퐼 = (Equation 1.3) 4 where k is the stiffness, kc is the compliance, E is the Young’s modulus, I is the moment of inertia of the cross-section, r is the radius, w is the width, l is the length and t is the thickness.

Thus, it is possible to create flexible neural implants using materials with very high Young’s modulus if the geometric cross-section is greatly reduced.

Table 1.1. Comparison of flexural rigidity for different probed designs recently reported in literature Substrate Material Young’s Thickness Width, (w), Moment of Flexural Shank Recording GSA, µm2 Reference

Modulus (t), µm µm inertia of rigidity cross- sites/ shank

(E), MPa cross- (EI), Nm2 section

section (I), (CS) µm2

Tungsten 390000 50 3.1 x 10-19 1.2 x 10-7 1964 1 1964 [1]

a-SiC 300000 4 10 5.3 x 10-23 1.6 x 10-11 40 1 100 [2]

Carbon fiber 241000 6.8 1.1 x 10-22 2.5 x 10-11 36 1 36 [3]

Utah array 165000 1 2000 [4]

Neuronexus probe 165000 15 123 3.5 x 10-20 5.7 x10-10 1845 16 177

Nanocomposite [5], [6]

before 5000 100 203 1.7 x 10-17 8.5 x 10-8 20300 na na

after 12 100 203 1.7 x 10-17 2.0 x 10-10 20300 na na

Parylene C 3778 25 100 1.3 x 10-19 4.9 x 10-10 2500 1 1256 [7]

Parylene C 3778 20 35 2.3 x 10-20 8.8 x 10-11 700 3 96 [8], [9]

Polyimide 2800 20 160 1.1 x 10-19 3.0 x 10-10 3200 3 900 [10]

SU8 2000 55 90 1.3 x 10-18 2.5 x 10-9 4950 4 314 [11]

Thio-ene/acrylate [12]

before 360 35 150 5.4 x 10-19 1.9 x 10-10 5250 4 177

after 4.7 35 150 5.4 x 10-19 2.5 x10-12 5250 4 177

For example, it is possible to create flexible neural interfaces using carbon fiber with a Young’s modulus of 241 GPa and diameter of approx. 6.8 um; or from amorphous silicon carbide with

estimated Young’s modulus of 300 GPa when the shank geometry is extremely small [50]. A review of common materials used for electrode construction, where the device dimensions are reported, were used to estimate the moments of inertia of the cross-section and the flexural rigidity

(EI), and the values shown in the Table 1.1. The data in Table 1.1 were used to estimate the stiffness, k, and the values are plotted as a function of shank length in Figure 1.1.

Figure 1.1. A comparison of structural stiffness of various materials used in neural probe designs as a function of shank length.

The flexural rigidity is also related to the critical buckling force by Equation 1.4 where Pcr is critical buckling force, K is the column effective length factor (one fixed end, one pinned end

=2.045). The critical buckling force is the maximum axial load a shank can experience that will not cause lateral deflections. For a microelectrode shank to successfully penetrate the pia mater of

a rat brain for example, it is generally expected that its buckling force exceeds the maximum recorded tissue force of 1 mN [53], [56].

휋2퐸퐼 푃 = (Equation 1.4) 푐푟 (퐾푙)2

1.4 Amorphous silicon carbide films for biomedical applications

1.4.1 Properties of silicon carbide

Crystalline silicon carbide (SiC) is a multifunctional, nontoxic and chemically stable semiconductor with bandgap between 2.3 and 3.4 eV, depending on the polytype (4H-, 6H-, 3C-) structure of SiC. It has outstanding electronic and mechanical properties and is currently used in high-frequency and high-power electronic devices. It has high Young’s modulus (about 700GPa), high saturation velocity (2x107 cm/s) and a high breakdown electric field (>2 MV cm-1). In comparison with silicon, SiC has a high thermal conductivity (3.2 - 4.9 W cm-1 K-1), high radiation and chemical hardness (2480 kg/mm2), and can be p- or n-type doped from insulating to semiconducting or metallic. A prototype neural implant that used doped-SiC as the electrode material was recently described [57].

1.4.2 Stability of amorphous silicon carbide

The amorphous form of SiC is equally attractive due to its relatively lower deposition temperature and it has been extensively investigated as a candidate encapsulation for biomedical devices [57]–

[61]. The dielectric encapsulation material exhibits robust long-term stability, high electronic resistivity and does not corrode [58], [62]–[64]. Deposited by plasma enhanced chemical vapor deposition (PECVD), amorphous silicon carbide (a-SiC) films soaked in physiological saline

models at elevated temperatures (87oC) maintained their encapsulation property by exhibiting extremely high impedance over a 6-month soak period [65]. At the same accelerated soak conditions, a dissolution rate of less than 0.1 nm/h was measured for a-SiC films as compared to

o 2 nm/h for LPCVD Si3N4 [58]. At 37 C, no dissolution of the a-SiC film was observed. The stability and wet etch tolerance of PECVD a-SiC films have also been demonstrated in acidic and basic etchants. Amorphous SiC is well tolerated in the cortex [58], [60] and has been evaluated clinically as a coating for coronary stents [66], [67]. PECVD a-SiC provides a dense and conformal coating of underlying substrates allowing complete trace coverage in devices where metal conductors are used.

1.4.3 Utility of a-SiC as substrate and encapsulation for neural interface devices

The obvious benefit of the a-SiC MEA technology is that it utilizes the a-SiC dielectric film as the sole insulator and substrate with the promise of extreme chronic longevity and a potential for clinical translation. The high mechanical modulus of a-SiC (300 GPa compared to 165 GPa of Si) provides the a-SiC MEAs with the intrinsic stiffness necessary for penetration of neural tissues.

Because the a-SiC processing is amenable to thin-film fabrication methods, there is precise control of the geometric surface area of the active electrode sites and design flexibility in meeting the varying needs of the neuromodulation community. The a-SiC MEA technology utilizes standard thin-film fabrication methods and patterning by UV lithographic techniques. Electrode sites, bond pads and the device superstructure are defined by reactive ion etching (RIE) in an SF6 plasma chemistry. Electrode sites as small as 20- 200 um2 can be enabled by the a-SiC technology. Design flexibility includes placing multiple electrode sites on the same shank, varying GSA, varying shank length or width, and control of the electrode tip geometries.

1.5 Ultrathin carbon fiber microelectrode arrays

The small cross-sectional dimensions of carbon fibers make them attractive in present microelectrode development because they are minimally invasive, and the risk of BBB disruption, tissue displacement and intracranial pressure is reduced substantially [13], [68]–[71]. Carbon fibers employed in microelectrode designs have a diameter of about 4-7 µm and Young’s modulus of 241 - 380 GPa [72], [73]. This stiffness however does not impact the in vivo performance because their small footprint makes them mechanically compliant with neuronal tissues [51].

Present carbon fiber MEAs are electrically isolated with 0.5 – 1 µm thick Parylene C, except the electrode site located at the distal end which is opened by using surgical scissors, focused ion- beam milling or a razor blade to blunt-cut the tips. This process limits the geometric surface area

(GSA) of the electrode site to the cross-sectional diameter of the carbon fiber creating electrodes with GSAs below 50 µm2. To increase the GSA and reduce impedance, one successful approach has been to fire-sharpen the electrode tips at a water/air interface with an oxygen torch [72]. This approach creates carbon fiber electrodes with exposed GSAs up to 2500 µm2 although much of the exposed electrode area is determined by the amount of Parylene C de-insulated along the length of the shaft.

1.5.1 Insertion of ultrathin carbon fiber electrodes into brain tissues

Various methods of deploying carbon fiber MEAs for chronic neural applications have been demonstrated. The assembly methods involve either mounting the individual fibers on printed circuit boards [51], mounting in grooved silicon support structures [73], or assembling the fibers into a 3D-printed micro funnel [72]. The later approach creates bundles of carbon fiber shafts that

provides mechanical support for implanting fibers ~3 - 5 mm long into brain tissue. The alternative is the use of a polyethylene glycol (PEG) coating to temporarily stiffen the shafts for implantation

[73]. Depending on the molecular weight of the PEG used, the fibers can be inserted at different insertion speeds determined by the dissolution rate of the PEG used.

1.5.2 Carbon fiber microelectrodes and neural recordings

Despite the challenges with micromanipulation, manual fiber assembly and control of electrode

GSA, carbon fibers MEAs have provided high quality single and multi-unit recordings in acute and chronic experiments [51], [72], [73]. Carbon fiber MEAs have been shown to reliably record from a variety of closely packed neuronal cell types, across a variety of brain regions at varying depth in different animal models. For example, Guitchounts et al. [72] demonstrated recording of neural extracellular potential in the high vocal center (HVC), basal ganglia nucleus area X and auditory area field L of Zebra Finches up to a depth of 3 mm. Patel et al. [73] and Kozai et al. [51], working in rat cortex, chronically recorded neural signals ~1.6 mm deep with signal amplitudes up to ~ 200 µV and a SNR as high as 12.7. For carbon fiber MEAs with GSAs less than 50 µm2, poly (3,4-ethylenedioxythiophene) (PEDOT) has been used to reduce the electrode impedance for recording. PEDOT coatings have also been shown to reduce impedance on small Au electrodes of diameter 15 µm [74], but the stability of PEDOT for chronic neural stimulation remains unexplored

[75]. Electrode coatings with electrodeposited iridium oxide films (EIROF) have been demonstrated to create stable neural interfaces on Au electrodes for neural stimulation [76]–[78].

1.6 Ultramicroelectrodes for neural stimulation

While a-SiC allows fabrication of very small chronically stable cellular or subcellular scale MEAs, the ultramicroelectrode dimensions result in increased electrode impedance and decreased charge injection capacity due to the reduced GSA of the active electrode sites. For ultrathin microelectrodes, the reduced shank dimensions result in active sites that are restricted in one linear dimension [79]–[81]. By exhibiting ultramicroelectrode behavior, these electrodes possess high charge densities, even though the current or charge are very small. Geometrically, electrodes whose characteristic dimension is less than 20 µm are usually classified as ultramicroelectrodes

[82]. The advantages of UMEs for chemical sensing includes reduced background charging currents at the electrode, enhanced mass transport to the electrode due to spherical or hemispherical rather than linear diffusion, and the low total current which reduces overpotentials and allows measurements at high potential sweep rates.

1.6.1 Low impedance coatings for ultramicroelectrodes

Stimulating with UMEs is challenging because electrical charge is typically injected within a pre- defined electrode potential window to avoid water electrolysis [83]. This window is typically -0.6

V to 0.8 V for iridium oxide and Pt, -0.8 V to 0.8 V for Au and -0.9 V to 0.9 V for TiN when measured against Ag|AgCl reference electrode in physiological electrolytes [75]. Considering the minimum charge requirement of 1-2 nC/ph in a 200 µs pulse (equivalent to 0.5-1 mC/cm2 for 200

µm2 site) to elicit a functional response with penetrating cortical microelectrodes [75], [84], this threshold may not be reached without exceeding the electrolysis potential limits for Au or Pt UMEs with a 20-200 µm2 geometric surface area [50]. Electrodeposition of low impedance materials like

EIROF or PEDOT, used previously on carbon fiber ultramicroelectrodes [51], [74], [85] to mitigate the issue of low charge injection capacity, may be applicable for reducing the electrode impedance of a-SiC UME devices.

1.7 Justification and Goals

Successfully interfacing the central nervous system (CNS) with external microelectronics presents the potential of improving lives of patients with neurological or neuro-degenerative disorders.

Specifically, intra-cortical neural interfacing allows investigating sensorimotor learning and control with both natural and prosthetic limbs through volitional control of neural activity.

However, the clinical implementation and long-term stability of these indwelling neural interfaces have been challenged by several device failure modes associated with biotic or abiotic factors.

Recent studies suggest that microelectrodes with shaft dimensions less than ~10 µm in the maximum transverse direction elicit low host immune response. Following observations of stable chronic neural recordings from songbird nucleus HVC using ~ 5 µm diameter carbon fiber ultramicroelectrodes, we proposed to investigate the development of UMEs of similar dimensions based on a-SiC for neural stimulation and recording. Silicon carbide was chosen for its mechanical strength and stiffness, high electronic resistivity and corrosion resistance, tolerance in the cortex and amenability to thin-film fabrication process. Films of a-SiC have excellent barrier properties and have been evaluated clinically as coatings for coronary stents. We hypothesize that a-SiC

MEAs will provide stable chronic interfaces for neural stimulation and recording. The aims of this dissertation are:

1. Develop and fabricate a-SiC based MEAs capable of intra-cortical stimulation and

recording.

2. Demonstrate the stability of a-SiC MEAs using in vitro bench testing.

3. Establish the in vivo performance of a-SiC UME arrays for chronic neural recording and

stimulation.

A significant challenge in the development MEAs based on thin films is modulation of residual stress within each layer to produce straight and planer shank curvature. Since a-SiC film is intrinsically compressive, we hypothesize that tensile stress engineered in the metallization will modulate the shank curvature in the a-SiC MEA technology.

This dissertation will address four challenges specifically associated with a-SiC based MEAs development and the development ultramicroelectrodes.

a. Stress control in multilayered a-SiC MEA structures.

b. Microfabrication challenges.

c. Implantation challenges associated with insertion of ultrathin shanks.

d. Challenges associated with high impedance ultramicroelectrode sites.

1.8 Dissertation organization

The aim of this work is to develop multielectrode arrays with ultrathin geometries using amorphous silicon carbide as the primary substrate material. Since the consequence of reduce electrode shank dimensions is very small electrode sites with high impedance, an extension of this work investigates methods of developing low impedance coatings on the ultramicroelectrode sites and characterizing their electrochemical properties in both in vitro and in vivo experimental models.

In Chapter 2, the electrochemical properties of electrodeposited iridium oxide films on ultramicroelectrodes are described. This was studied by developing methods to coat EIROF on

carbon fiber ultramicroelectrodes (CFUMEs) as a means of enhancing the charge injection capacity and reducing electrode impedance. It was shown that EIROF coatings produced low impedance electrodes with improved charge storage and charge injection capacities. The maximum charge injection capacity during current pulsing was strongly dependent on the interpulse bias and pulse width and reflected a potential-dependent impedance of the EIROF. The charge injection capacity of the EIROF-coated CFUMEs measured in an inorganic buffered saline model of interstitial fluid exceeded 17 mC/cm2, allowing charge-injection at levels well above charge/phase thresholds for intraneural stimulation.

Chapter 3 describes challenges associated with microfabrication of a-SiC MEAs. This chapter details intrinsic stress engineering in multilayered a-SiC films and identifies a possible failure mode in a-SiC MEA technology. Technical solutions to these challenges are presented.

Chapter 4 reports on the development, in vitro characterization, and in vivo evaluation of the first prototype a-SiC MEAs. The chapter details fabrication methods, electrochemical properties of a-

SiC MEA coated with SIROF and TiN and demonstrates the ability to record extracellular neural activity spontaneously in two animal models - songbird basal ganglia nucleus and rat motor cortex.

Chapter 5 demonstrates implantation strategies developed for inserting the ultrathin a-SiC MEAs into rat cortex. The chapter also highlights how the a-SiC platform enables a wide variety of innovative designs utilizing only a maximum of 6 µm thick amorphous silicon carbide (a-SiC) as the MEA superstructure. Insertion forces and tissue forces experienced by an indwelling shaft are reported in this chapter. Acute recording following successful implantation of 6 µm thick a-SiC shanks are presented.

Chapter 6 provides a summary of findings in this dissertation and suggests future directions for this work.

CHAPTER 2

ELECTRODEPOSITED IRIDIUM OXIDE ON CARBON FIBER

ULTRAMICROELECTRODES FOR NEURAL RECORDING AND STIMULATION1

Authors: Felix Deku1, Alexandra Joshi-Imre1, Alket Mertiri2, Timothy J. Gardner2, Stuart F. Cogan1

1Department of Biomedical Engineering

The University of Texas at Dallas

800 West Campbell Road

1Richardson, TX 75080-3021

2Department of Biology and Biomedical Engineering

Boston University

One Silber Way

Boston, MA 02215

1This work was published open access by the Journal of Electrochemical Society in June 2018.

1.9 Author Contributions

Felix Deku, Dr. Timothy J. Gardner and Dr. Stuart F. Cogan conceived and conceptualized the study. Dr. Alket Mertiri fabricated the carbon fiber ultramicroelectrodes. Felix Deku performed the electrode coatings and all electrochemical characterizations. Felix Deku, Dr. Alexandra Joshi-

Imre and Dr. Stuart F. Cogan interpreted the results and wrote the manuscript.

1.10 Abstract

Host encapsulation decreases the ability of chronically implanted microelectrodes to record or stimulate neural activity. The degree of foreign body response is thought to depend strongly on the cross-sectional dimensions of the electrode shaft penetrating neural tissue. Microelectrodes with cellular or sub-cellular scale shaft cross-sectional dimensions, such as carbon fiber ultramicroelectrodes have been previously demonstrated to elicit minimal tissue response, but their small geometric surface area results in high electrode impedances for neural recording, and reduced charge injection capacity during current pulsing for neural stimulation. We investigated electrodeposited iridium oxide films (EIROF) on carbon fiber ultramicroelectrodes as a means of enhancing the charge injection capacity and reducing electrode impedance. EIROF coatings reduced the electrode impedance measured at 1 kHz by a factor of 10 and improved charge storage and charge injection capacities. The maximum charge injection capacity was also strongly dependent on the interpulse bias and pulse width and reflected a potential-dependent EIROF impedance. The charge injection capacity of the EIROF-coated carbon fiber ultramicroelectrodes measured in an inorganic buffered saline model of interstitial fluid exceeded 17 mC/cm2 with

appropriate biasing, allowing charge-injection at levels well above reported charge/phase thresholds for intraneural microstimulation.

1.11 Introduction

The foreign-body response to indwelling cortical microelectrodes is believed to limit neural recording reliability, particularly for extended chronic studies [16], [21], [51], [86]. The extent to which this response compromises recording reliability depends on several factors including the materials used in electrode construction [87], the presence of micro-motion [22], [88], and the size and geometry of the indwelling structure [51], [89], [90]. Recent studies suggest that penetrating microelectrodes with shank cross-sectional dimensions of less than approximately 10 µm induce minimal tissue response in 4-5-week chronic cortical placements in rat [47], [51], [91]. This greatly reduced tissue response, compared with more conventional microelectrodes having shaft cross- sectional dimensions larger than about 30 µm, is attributed to reduced insertion trauma, particularly reduced disruption of the blood-brain-barrier, and an overall decrease in the surface area of the tissue-device interface [51], [92]. Since the active sites of these electrodes have at least one geometric dimension that is less than 10 µm, they are expected to exhibit ultramicroelectrode

(UME) behavior in which the electrode possesses high charge density even though the current or injected charge is small [79], [80], [82], [93]. There has been considerable interest previously in

UMEs for in vivo chemical sensing because of reduced background charging currents, enhanced mass transport to the electrode due to spherical or hemispherical rather than linear diffusion, and the low currents which reduce overpotentials and allow measurements at high potential sweep rates

[79], [80].

As recognized in previous studies, the greatly reduced geometric surface area of UMEs results in a high electrode impedance that can potentially compromise the recording of small neural signals

[45], [50], [51], [73], [74]. To address this concern, Kozai et al. used coatings of poly(styrenesulfonate)-doped poly(ethylenedioxythiophene) (PEDOT-PSS), electrodeposited on carbon-fiber ultramicroelectrodes (CFUMEs) to demonstrate a 100-fold decrease in electrode impedance at recording frequencies between 10 Hz and ~ 2000 Hz [51]. The electrodes, which had a nominal exposed geometric surface area (GSA) of ~38 µm2 before PEDOT-PSS deposition, recorded robust single-unit activity with peak-to-peak amplitudes of ~300 µV and signal-to-noise ratios (SNR) of >4. Similarly, Guitchounts et al. [72] demonstrated stable multi-unit neural recordings from motor nucleus HVC of the zebra finch for a period of 107 days with 5-µm diameter, Parylene-insulated carbon-fiber electrodes. Due the length of the exposed carbon fiber on the electrodes used by Guitchounts et al., 2013, their electrode GSAs are about 1000 µm2, obviating the need for a low impedance coating.

Although the ability of these UMEs to provide chronic neural recordings has been demonstrated

[51], [72], [73], electrical stimulation remains challenging because of the high charge densities required to achieve threshold levels of charge injection for eliciting a functional response [84],

[94]. While it is possible that the absence of significant foreign body response or the coordinated use of multiple UMEs might lead to lower charge thresholds for neural activation, present microelectrode thresholds of ~1 nC/phase would lead to charge densities at carbon-fiber UMEs that are beyond the reversible charge injection limit of the carbon fiber. Although electrode coatings with conducting polymers such as PEDOT have demonstrated reduced microelectrode impedance [51], [73], [74] and improved stimulation charge injection capacities of 3.6 mC/cm2

[95], PEDOT stability for chronic charge injection, especially at high current densities, remains unknown [73], [75]. Electrode coatings with electrodeposited iridium oxide films (EIROF) [76],

[96], [97] or sputtered iridium oxide films (SIROF) [98]–[100] have been proposed as stable chronic neural interfaces for neural microstimulation. EIROF substantially increased the charge injection capacity and reduced the impedance on platinum, PtIr-alloy, gold, and stainless-steel microelectrodes (GSA~2000 µm2) but the charge injection properties of EIROF UMEs have not been evaluated. EIROF-coated CFUMEs have been investigated as microscopic pH sensors [101] and more recently shown to reliably record neural activity from the peripheral nerve of song bird

[85]. In the present study, the electrochemical behavior and charge injection properties of EIROFs on CFUMEs were investigated as a means of achieving microelectrode levels of charge injection for neural stimulation.

1.12 Experimental

1.12.1 Carbon fiber ultramicroelectrode fabrication

Parylene-insulated CFUMEs with a diameter of approximately 5 µm and exposed fiber lengths varying from 70 µm to 300 µm, resulting in electrode GSAs of 350-1500 µm2, were fabricated by the fire-sharpening process previously described [72], [85]. The exposed electrode has a tip terminating in a cone with radius of curvature less than 0.5 µm. Electrical connection was made to the opposite end of the fiber by adhesively bonding a 5-cm long silver wire with carbon paste.

Iridium oxide was electrodeposited onto the exposed electrode tip following methods previously described by Meyer et al. [76]. Briefly, an electrodeposition solution was prepared by dissolving a 4 mM IrCl4 in 40 mM oxalic acid, followed by the slow addition of 340 mM K2CO3 to a final

pH of ~10.3. The solution equilibrated over a 14-day period before use. Although Meyer et al. [76] used a two-step electrodeposition process that included triangular and square potential waveforms, we found that a deposition process using only cyclic voltammetry (CV) with a triangular potential sweep was enough to coat the CFUMEs with EIROF suitable for charge injection. The electrodeposition was performed at room temperature using a potential-controlled triangular waveform, applied at a sweep rate of 50 mV/s between limits of -0.05 V and 0.50 V (vs. Ag|AgCl) for 50 cycles. The electrodeposition bath was air-equilibrated and unstirred.

1.12.2 Test Electrolytes

All electrochemical evaluation was performed in an inorganic model of interstitial fluid (ISF) that has similar ionic concentrations and buffering capacity to that of the physiological environment

[102], [103]. The model-ISF was prepared with NaCl (110 mM), Na2HPO4.7H2O (2 mM),

NaH2PO4.H2O (0.5 mM), NaHCO3 (28 mM), KHCO3 (7.5 mM), and 0.5 mM each of MgSO4,

o MgCl2, CaCl2. The model-ISF is maintained at 37 C and gently sparged with a gas mixture containing O2, CO2 and N2 in the ratio 5:6:89 to obtain a pH of 7.4-7.6. For comparison, CV and

EIS were also measured at laboratory temperature (~20oC) in argon de-aerated phosphate buffered saline (PBS) prepared with NaCl (130 mM), Na2HPO4.7H20 (81 mM), and NaH2PO4.H2O (22 mM) with a pH of 7.2. The ionic conductivities of the PBS and model-ISF are approximately 25 mScm-1 and 15 mScm-1, respectively, measured with a conductivity meter (Fisher Scientific).

1.12.3 Cyclic voltammetry and electrochemical impedance spectroscopy

The EIROF–coated electrodes were characterized by cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). A three-electrode cell comprising the CFUME

working electrode, a large surface area Pt counter electrode and Ag|AgCl reference electrode was used for all measurements. EIS and CV measurements were recorded with a Gamry Reference 600 potentiostat using vendor-supplied data acquisition software. The EIS measurements were performed by applying a 10 mV rms AC sinusoidal excitation over a 0.1 to 105 Hz frequency range about the open circuit potential (Eoc) and about a bias potential from -0.6 V to 0.6 V in 0.1 V increments

While the potential limits for water electrolysis on iridium oxide have been previously established with commonly used reduction and oxidation limits of -0.6 V and 0.8 V (Ag|AgCl), respectively

[75], [76], we observed a large oxidation current onset at 0.7 V (vs. Ag|AgCl) on the EIROF, which

we associated with water oxidation (Figure 2.1a). For this reason, a potential range of -0.6 V to

0.6 V was used for all CV scans and charge storage capacity (CSC) measurements. Cathodal charge storage capacity (CSCc) was calculated from the time integral of the cathodal current during

CV measurements from 0.6 V to -0.6 V (vs. Ag|AgCl) at a sweep rate of 50 mV/s [75].

1.12.4 Voltage transient measurements

Voltage transients in response to constant current pulsing were measured with a custom-built stimulator (Sigenics, Chicago IL) previously described [83], [104]. The stimulator is designed to generate monophasic current-controlled cathodal pulses and actively control the interpulse potential of the electrode using a voltage-controlled anodic recharge current. Between the cathodal current pulse and the anodic recharge phase there is an interphase delay, controllable from 0 - 1 ms, when the electrode is at open circuit. The use of a zero-current interphase period during current pulsing is also useful in assessing the reversibility of the charge-injection and the contribution of the Ir3+/Ir4+ redox reaction to charge injection.

Referring to Figure 2.1b, the access voltage arising from the ionic resistance of the electrolyte is estimated from the voltage transient as follows: VAL is the access voltage at the leading edge of the current pulse measured as the abrupt change in the electrode voltage when the current pulse is turned on; VAT is the trailing phase of the access voltage calculated as the difference between the cathodal electrode potential (Ec) and the maximum negative voltage (Vmin) in the transient. Vmin is observed at the end of the current pulse, and Ec is measured in the interphase period, 12 µs after the end of the current pulse when the applied current is zero. The maximum potential excursion

(Emc) is attained when Ec is equal to -0.6 V corresponding approximately to the potential at which water is reduced on EIROF (the reduction potential limit determined from the CV). In the

interphase period, the applied current is zero so there are no contributions to the measured electrode potential from either iR-drops in the electrolyte or activation overpotential, and if the delay is long enough concentration overpotentials should also approach zero [75]. The end potential (Eend) is the electrode potential just before the start of the anodic recharge current. The charge injection capacity Qinj, was calculated from the current and pulse width of the cathodal pulse. The maximum

Qinj was attained when the electrode was polarized to the Emc of -0.6 V. Charge injection capacities were measured for a range of cathodal pulse widths from 100 µs to 500 µs and interpulse bias levels from 0.0 to 0.6 V (Ag|AgCl), all at a frequency of 50 pulses per second.

1.13 Results and Discussion

1.13.1 Electrochemical Characterization

Electrodeposition of EIROF on CFUMEs produces a nodular and highly porous coating as shown in Figure 2.2, with some suggestion of higher deposition rates at the sharp tips of the electrodes where higher current densities are expected. Representative CV and EIS curves for an uncoated and EIROF-coated CFUME in model-ISF and PBS are shown in Figure 2.3. In general,

2 electrodeposition using 50 CV cycles yielded CSCc values of up to 50 mC/cm in model-ISF, which is a factor of 103 increase over the uncoated CFUME. Neglecting double-layer capacitance and minor contributions from side reactions such as oxygen reduction [105] the calculated CSCc estimates the total available charge for stimulation from the Ir3+/Ir4+ redox couple [76]. Deposition over a larger number of CV cycles increased the CSCc but resulted in the appearance of a sharp oxidation peak at 260-280 mV, not shown, that has been associated with cracking or delaminating iridium oxide coatings [104].

As shown in Figure 2.3a, there is a negative shift in the Ir3+/Ir4+ reduction wave from -50 mV in

PBS to -250 mV in model-ISF. This shift is like that observed with activated iridium oxide in PBS and model-ISF electrolytes and is related to the lower buffering capacity and conductivity of the model-ISF compared with that of the PBS [102]. The calculated CSCc from the CVs in Figure 2.3a

2 2 are 38 mC/cm for model-ISF and 43 mC/cm for PBS. The uncoated CFUME CSCc was ~ 0.01 mC/cm2 in model-ISF or PBS. The 1000-fold increase in charge storage capacity as a result of

EIROF coating is accompanied by a 10-fold decrease of electrode impedance at 1 kHz: 570 kΩ for uncoated (GSA=600 µm2) to 91 kΩ in model-ISF and 57 kΩ in PBS after EIROF deposition.

A similar magnitude in impedance reduction of microelectrodes after EIROF coating was observed by Han et al [78] and Meyer et al [76].

1.13.2 Stimulation charge injection capacity

The charge-injection capacity of EIROF-coated CFUMEs for neural stimulation was evaluated by measuring the voltage transient response of three electrodes in response to current pulsing as a function of interpulse bias from 0.0 to 0.6 V (Ag|AgCl). The average CSCc of the electrodes was

48.7± 1.2 mC/cm2 (mean ± s.d., n=3) with an average estimated GSA of 385±55 µm2 based on an average electrode diameter of 5 µm. Positively biasing EIROF electrodes in the interpulse period is a well-established means of increasing charge injection capacity for cathodal pulsing [75], [83].

The average maximum charge injection capacity and charge per phase for the EIROF electrodes in response to 500 µs current pulses as a function of interpulse bias are shown in Figure 2.4. The maximum charge per phase is calculated from the cathodal current amplitude that polarizes the electrode to its cathodal potential limit of -0.6 V Ag|AgCl.

Figure 2.4. Charge injection limits as a function of (a) bias and (b) pulse width at 0 mV interpulse bias.

The current output of the stimulator, which is an analog signal, is slowly increased until an Emc of

-0.6 V vs. AgCl, is observed on the oscilloscope. The electrode potential is measured through the stimulator using an isolated output to prevent external hardware such as the oscilloscope from affecting the measured potential. The experiment is repeated for each interpulse bias investigated.

2 2 The maximum Qinj at 0 V bias was 1 mC/cm and increased to 17 mC/cm as the interpulse bias was increased to 0.6 V. In the absence of an imposed bias, implanted iridium oxide electrodes adopt an equilibrium potential of about 0 V (vs. Ag|AgCl), demonstrating the advantage of the positive interpulse bias in increasing charge injection capacity [83], [106]. From Figure 2.4b, the charge injection capacity and charge per phase appears to increase with pulse width. A paired t- test shows a significant difference (p-value < 0.05) in charge delivered in the 500 µs pulse (1.0 ±

0.13 mC/cm2) to the 100 µs pulse (0.6 ± 0.14 mC/cm2) at 0 V bias.

1.13.3 EIROF charge transfer characteristics

Commonly used electrode materials for neural stimulation and recording such as titanium nitride

(TiN), platinum (Pt), or iridium oxide (IrOx) inject charge by two mechanisms; capacitive where charge is redistributed at the electrode-electrolyte interface or faradaic in which electrons are transferred across the interface by reduction-oxidation reactions [75], [107]. EIROF, like all iridium oxide coatings used for neural recording and stimulation, is a mixed conductor, exhibiting both electronic and ionic conductivity. On a subgroup of EIROF–coated CFUMEs (n=5), the voltage transient response was examined as a function of current pulse amplitude using an interpulse bias of 0.6 V.

Figure 2.5. Voltage transients recorded as the electrode is polarized to different cathodal potential excursions (Ec) starting from a 0.6 V bias to an Emc of -0.6V.

As the Ec shifts negative and the driving voltage (Vdrv) increases with increasing pulse amplitude, an inflection in the polarization is observed for pulse amplitudes high enough to drive the electrode cathodal potential (Ec) to a value more negative than 0.1 V, as shown in Figure 2.5. At the higher current pulse amplitudes associated with the inflection in the polarization, the magnitudes of the leading (VAL) and trailing (VAT) access voltages also differ, with VAT increasing to 2.8 V compared with 1.1 V for VAL when the EIROF is polarized to an Ec of -0.6 V Ag|AgCl (Emc).

This effect is shown in detail in Figure 2.6 in which the access voltages are shown as a function of

Ec with a representative EIROF CV plotted on the same potential axis.

2 2 At pulse amplitudes of 9 µA to 41 µA (1 mC/cm to 5 mC/cm ) and corresponding Ec of 0.5 V to

0.1 V, no inflection was noted in the transient and VAT was approximately equal to VAL. At Ec potentials more negative than 0.1 V, corresponding to the onset of the primary Ir3+/Ir4+ reduction wave in the voltammogram, a significant deviation in VAL and VAT is observed. Similar behavior in VAT and VAL has been previously noted for AIROF and PEDOT coated microelectrodes, and has been associated with a transition of the electrode coating to a high-impedance state [75].

Iridium oxide films have been shown to be electronically conducting in the oxidized state at potentials positive of 0.1 V vs Ag|AgCl but become increasing non-conductive at more negative potentials as the Ir3+/Ir4+ ratio increases [108].

To examine the effect of potential on the conductivity of the EIROF, EIS spectra were collected as a function of DC bias from 0.6 V to -0.6 V vs Ag|AgCl in a 0.1 V increment on the same subgroup of EIROF-coated CFUMEs (n=5). The impedance magnitude was normalized to the geometric surface area of each electrode and the resulting specific impedance spectra are presented in a Bode plot as shown in Figure 2.7(a). Figure 2.7(b) shows the specific impedance at 1 Hz, 1 kHz and 100 kHz plotted against bias. Both Figures demonstrate that the impedance of the EIROF- coated CFUME exhibits a significant potential dependence. Even the high frequency impedance, which is primarily associated with the electrolyte resistance, shows some increase as the Ir4+ is reduced to Ir3+ and the EIROF becomes less electronically conducting. A more pronounced effect is seen at frequencies below 10 kHz. The impedance increases monotonically across the 0.1-104

Hz frequency range by a factor of 10-100, as the DC potential is made increasingly negative from

0.2 V to -0.6 V (Figure 2.7(b)). The 0.2 V potential at which the impedance begins to increase

corresponds closely with the 0.1 V Ec at which the magnitude of VAT becomes noticeably larger than VAL. As shown by the superimposed EIROF CV in Figure 3 or 7b, the 0.2 V onset also corresponds to a potential that is just positive of the major Ir4+/Ir3+ reduction wave in the CV.

The stimulation waveform employed in this study has an interphase period in which the applied current is zero with the electrode at open-circuit (region III, Figure 2.1b). At open-circuit, the electrode potential changes due to the re-establishment of the equilibrium pH at the surface of the electrode and to the comparatively slow chemical reactions that act to return the EIROF to its equilibrium potential, which is about 0.2 V Ag|AgCl in model-ISF. The principal chemical reaction is probably oxygen reduction, resulting in oxidation of the EIROF and a consequent increase in the Ir4+/Ir3+ ratio and positive shift in electrode potential. Since the pH equilibrates relatively

rapidly [109] compared with the oxidation of the EIROF, the electrode potential at the end of the open-circuit period (or interphase period) closely corresponds to the potential that would be observed if the EIROF was reduced slowly, as in a 50 mV/s CV sweep, between the interpulse bias potential (Vbias) and the electrode end-potential (Eend).

The time-integral of the cathodal current between Vbias and Eend is used to estimate the total charge

4+ 3+ contribution from the Ir /Ir redox reaction (QE) to the total charge injected during a pulse as shown in the representative example in Figure 2.8. The relationship between Eend, QE, and the injected charge (Qinj) is shown in Figure 2.9 as a function of injected charge using 50 Hz, 0.5 ms pulses from an interpulse potential of 0.6 V. From Figure 2.9, it is apparent that there is very little difference between Qinj and QE, at all levels of stimulus intensity. This observation suggests that

the charge injected through the EIROF occurs substantially by reduction of Ir4+ to Ir3+ with minimal contribution from side reactions such as oxygen reduction or, at more negative potentials, water reduction. In addition, due to their small GSA, the EIROF UMEs have a substantially higher charge injection capacity, albeit at a much smaller charge per phase, than stimulation microelectrodes with a more typical GSA of 2000 µm2 [75], [76], [110]. The UME size also results in a high utilization of the EIROF. The utilization, defined as the Qinj/CSCc, is approximately 0.35 (35%) for the EIROF in Figure 2.9 pulsed to an Emc of -0.6 V, which compares with 5-20 % for microelectrode-sized EIROF [76], [78], as well as other microelectrode iridium oxide coatings

[98].

1.14 Conclusions

Due to their size, carbon fiber UMEs have high impedance and low charge-injection capacities.

EIROF coating the UMEs resulted in a large increase in charge storage capacity, by a factor of

103, and a 10-fold decrease in impedance over uncoated electrodes. The charge injection capacity of EIROF-coated UMEs was several times higher than conventional microelectrodes (GSA ~2000

µm2) coated with EIROF. Similar to activated iridium oxide (AIROF), the voltage transient data demonstrate that an anodic bias is desirable for cathodal-first pulsing with EIROF-coated electrodes at short pulse widths (< 500 µs). Our results suggest that under neural stimulation pulsing conditions, EIROF injects charge substantially by reversible oxidation and reduction of the Ir3+/Ir4+ redox couple. The in vivo reversibility of charge-injection with EIROF UMEs remains to be explored.

CHAPTER 3

ENGINEERING CHALLENGES ASSOCIATED WITH FABRICATION OF

AMORPHOUS SILICON CARBIDE MICROELECTRODE ARRAYS

Authors: Felix Deku, Negar Geramifard, Benjamin Ting, Alexandra Joshi-Imre, Stuart F. Cogan

Department of Biomedical Engineering

The University of Texas at Dallas

800 West Campbell Road

1Richardson, TX 75080-3021

1.15 Participation

Felix Deku and Dr. Stuart Cogan conceptualized the study. Felix Deku designed, fabricated MEAs, and developed multilayer stress engineering techniques. Felix Deku and Negar Geramifard performed thin film residual stress experiments. Felix Deku and Benjamin Ting performed accelerated aging experiments. Felix Deku, Drs. Alexandra Joshi-Imre and Stuart F. Cogan provided useful discussions, interpreted results and contributed to manuscript preparation.

1.16 Introduction

Amorphous silicon carbide (a-SiC) based microelectrode arrays (MEAs) are fabricated on silicon carrier wafers using a-SiC as the sole substrate and encapsulation [50]. A metal trace is sandwiched between two a-SiC layers connecting the electrode sites and bond pads. A significant challenge in developing a-SiC MEAs is achieving stress balance between individual layers in order to maintain the desired planar geometry once the MEAs are removed from the silicon carrier wafer. For penetrating MEAs, such as brain probes, shanks with none or minimal curvature are desired for successful implantation. Another significant challenge in a-SiC MEA fabrication lies in the development of thin film patterning processes. Because the quality of a patterned layer will influence subsequent thin-film growth of the next layer, patterning quality can be directly responsible for defects and for related failure modes of an MEA.

To achieve stress balance, we studied the residual stress in PECVD a-SiC films, where the residual film stress is determined by the chamber conditions such as deposition temperature, RF power density, deposition pressure and the flow rates of the SiH4 and CH4 reactive gases [111]–[114].

o -2 For example, a-SiC film deposited at 1000 mTorr, 325 C, 0.20 Wcm using a SiH4: CH4 gas ratio

of 1:3 results in a film with residual compressive stress between 80 and 120 MPa [50]. A similar stress can be obtained when films are deposited at 350 oC and 0.27 Wcm-2. A tensile stress is engineered in the metallization to counter-balance the compressive stress in the a-SiC layers. Most evaporated metal films deposited under high vacuum have relatively high tensile stress [115],

[116] whereas sputtered-deposited metal films may be tensile or compressive depending on the sputtering conditions and the film thickness [117]. Tensile stress is caused by grain boundary shrinkage (annealing) and compressive stress arises by lattice distortion or creation of interstitials produced by energetic particles striking the growing film [118], [119]. Consequently, one common way of controlling the intrinsic metal stress from tensile to compressive is by increasing the substrate bias [117], [120]. The application of bias or negative voltage to substrates during sputter deposition increases the energy delivered by the primary particle bombardment to the growing film and increases the purity of the resulting film [121]–[123]. Similar effects can also be achieved by increasing the sputtering cathode power resulting in an increased film deposition rate, or by reducing the argon pressure in the sputtering chamber [124].

We describe a method of engineering and controlling the residual stress within a-SiC MEAs by minimizing the residual a-SiC compressive film stress and by balancing it with the residual tensile stress in metallization that is deposited by either evaporation or by DC sputtering at low pressure.

A thermal annealing step is introduced to shift residual stresses towards tensile after the deposition of the second a-SiC layer, which is tuned based on measurements of the residual stress in the whole

MEA stack.

We also demonstrate that the long-term performance of the a-SiC ultramicroelectrodes is dependent on the integrity of the metallization. MEAs remain intact and functional when aged at

elevated temperature conditions for at least 20 weeks in phosphate buffered saline (PBS) when the metallization patterns are fabricated appropriately.

1.17 Methods

1.17.1 Multilayer stress engineering

The residual stress in the a-SiC films and the metallization was controlled during deposition by tuning the growth conditions. For the PECVD a-SiC, compressive stress control was achieved through regulating the deposition temperature, power density and gas ratio. For metal films obtained by sputtering, tensile stress was generated by depositing films at higher Ar pressure and lower sputtering power. Stress control in evaporated metal films was achieved by regulating the deposition rate. Stress control can also be achieved on a system level by measuring the overall stress of the multilayered structure and thermal annealing the thin-film stack to the desired final stress.

1.17.2 Accelerated aging and characterization

3.3.2.1. Sample preparation

A 100 mm diameter double-sided polished (dsp) silicon wafer was used as a substrate for the MEA microfabrication. A 2 µm thick a-SiC film was deposited on both sides of the dsp wafer. On one side, a patterned metal conducting trace was deposited to a nominal thickness of 400 nm followed by another 2 µm thick a-SiC film. Electrode sites (GSA= 1000 µm2) and contact pads were created by reactive ion etching in an SF6 plasma similar to methods previously used [50]. The electrode sites were either uncoated (Au) or coated with sputtered iridium oxide film (SIROF).

3.3.2.2. Electrochemical measurements

The stability of a-SiC thin-film as a dielectric encapsulation was evaluated in a series of accelerated saline soak test. Long-term stability was assessed by aging at an elevated temperature of 87oC in

PBS and compared with 37oC baseline measurements providing an approximate acceleration factor of 32, assuming an acceleration factor of 2∆푇/10 used for polymers in medical devices [125].

The test samples were completely immersed in the PBS and placed in a temperature-controlled oven for aging. Electrochemical impedance spectroscopy (EIS) was obtained weekly with a Gamry potentiostat Reference 600 using the vendor-supplied data acquisition software. A three-electrode cell comprising electrode sites on the a-SiC MEA as the working electrode, a large surface area Pt counter electrode and Ag|AgCl reference electrode, as shown in Figure 3.1, was used for all

measurements. The EIS measurements were performed by applying a 10 mV AC excitation over

5 a 0.1 to 10 Hz frequency range about the open circuit potential (Eoc).

1.18 Results and Discussions

1.18.1 Multilayer stress engineering

An example of a-SiC MEA under compressive stress is shown in Figure 3.2. It is nearly impossible to manipulate and perform post-fabrication processing on the curled superstructure or to insert the curled shanks into the cortex.

Figure 3.2. An example of a-SiC MEA under unbalanced residual stress.

3.4.1.1. Stress control during thin-film deposition

Metal stress control via argon pressure variation during sputtering was chosen to moderate the curvature of the a-SiC shanks to obtain planar and straight MEAs. Figure 3.3 shows residual stress evolution in multilayer stack during deposition of the first a-SiC layer (I), evaporated or sputtered metallization (II), and the deposition of the second a-SiC layer (III). With the

deposition conditions described above, the residual stress in a 2 or 4 µm thick a-SiC is approximately -100 MPa. The 400 nm thick metallization sputtered with an argon pressure of 12-

16 mTorr produces films with 40 to 100 MPa tensile stress as measured on a silicon control wafer (Refer to Table 3.1 for tensile metal stresses measured at different sputtering pressures).

When the metallization is deposited on the a-SiC, the measured average residual stress of the bilayer is -67 MPa. After deposition of the top a-SiC layer (~ -100 MPa), the overall stress in the tri-layer was measured to be -52.19 MPa. We identified a tensile stress evolution in the metallization during the top a-SiC deposition as being responsible for the reduction of the compressive stress in the layered stack even following deposition of a compressive a-SiC film. As

shown in Figure 3.3, the average residual tensile stress in the sputtered metallization measured on the silicon control wafer changed from ~89 MPa to ~402 MPa when placed in a PECVD condition.

Under similar condition, the a-SiC + metal bilayer evolved in residual stress from -67 MPa to -8

MPa. We define PECVD condition as when samples are placed in the PECVD chamber with platen temperature set to 325 - 350oC for the duration of a typical 2 µm thick a-SiC deposition without actual deposition of the a-SiC. Table 3.1 summarizes our findings for typical deposition pressures between 4 to 30 mTorr, and the resulting tensile stress as measured on silicon control wafers. Also shown is the final shank curvature of the a-SiC MEA (where both layers of a-SiC on silicon controls were measured to be approximately -100 MPa, and no thermal annealing was applied to the stack).

Table 3.1. Effect of sputtering pressure on metal stress and a-SiC MEA curvature post- fabrication. The deposition times were adjusted so that a nominal metal thickness of 400 nm was achieved. Argon Pressure (mTorr) Metal stress (MPa) MEA curvature

< 10 < 20 bends downwards

12-16 40-100 Straight

>20 >120 bends upwards

Stress control in evaporated metal films was achieved by controlling the deposition rate. A slow deposition rate of 1 Å/s is enough to produce MEAs with straight curvature although low deposition rates have been associated with more electrically resistive films [126]. The 1 Å/s deposition rate corresponds to an average tensile film stress of 9 MPa which evolves to 426 MPa when exposed to a-SiC PECVD deposition times and temperatures (measured on silicon control

wafer) without actual a-SiC deposition. From Figure 3.3, evaporated metals deposited onto a-SiC result in an average bilayer compressive stress of -81 MPa, which evolves to -5 MPa when exposed to PECVD deposition times and temperatures and then to -57 MPa when the top a-SiC layer is deposited over the metallization.

For the a-SiC/metal/a-SiC multilayered structures, we have determined that a straight and planar

MEA can be obtained if the overall multilayer residual stress is about 30 to 60 MPa compressive

(as measured on a silicon control wafer). This corresponds to depositing both a-SiC films to ~100

MPa in compression and sputtering the metallization at 12-16 mTorr or evaporating the metal films at 1 Å/s.

3.4.1.2. Stress engineering in multilayer stack by thermal annealing

Figure 3.4. Stress evolution in multilayered a-SiC/metal/a-SiC stack during thermal annealing at o 400 C in N2. The metallization was deposited either by evaporation or sputtering.

Thermal annealing is usually used for reducing intrinsic stress, structural modification, and surface roughness control in thin-film materials. Thermal annealing of PECVD a-SiC films has been shown to reduce the compressive stress, as identified by a decrease in the Si-H peak intensities in

FTIR spectrum [127]. Thermal annealing of Au films is expected to increase the grain size of both evaporated and sputtered films [126]. Thermal annealing of the a-SiC/metal/a-SiC stack (with film thicknesses of 4000 nm/400 nm/2000 nm) at 400 oC in a Thermco Mini Brute annealing tube under

N2 shifts the residual stress towards tensile as shown in Figure 3.4. The residual stress evolution under this condition appears to be higher when sputtered metallization was used in the stack as compared to evaporated metallization.

1.18.2 Microfabrication challenges in a-SiC MEA technology

Figure 3.5. SEM images taken at a tilted viewing angle show thin and tall free-standing metal film remnants, at the edges of the traces. These film remnants are prevalent defects when lift-off is used with sputtered metallization. (a) Shows rabbit ears often at both edges of metal traces following lift-off process. (b) Clean lift-off (red ellipse) and bad lift-off with rabbit ear (black ellipse)

Our microfabrication technology uses a bilayer lift-off photoresist process to produce patterned metal traces. The bilayer photoresist consists of a photo sensitive resist on the top (Microposit

S1800 series) and a non-photosensitive lift-off resist at the bottom (Microposit LOR) to generate an under-cutting lift-off profile. This bilayer photoresist lift-off process allows us to pattern consistently 400 nm thick metal traces with 2 µm line widths.

Figure 3.6. SEM images taken from a tilted viewing angle show rabbit ear defects conformably covered by a-SiC film creates bands of a-SiC at the edges of the metal traces.

However, thin and tall free-standing residual metal structures, commonly referred as rabbit ears, may remain at the edges of the traces when the metal layer is deposited by a sputtering process, as seen in Figure 3.5. Rabbit ears arise due to the high degree of angular distribution of matter

depositing and coating the open surfaces of the photoresist and the LOR. When the photoresist and

LOR are washed away, the thin metal film that deposited inside the undercut on the LOR can break away or remain.

The remaining film remnants produce the rabbit ears. The degree of the undercut depends on the soft-bake temperature of the LOR and the pattern development time. If the undercut is increased,

rabbit ears can be avoided, but at the expense of reduced resolution and consequent reduced spacing between metal traces. Because the PECVD a-SiC film conformably encapsulates its underlying substrate to at least some degree, a rabbit ear will project into the top a-SiC layer. As seen in Figure 3.6, the rabbit ears can form extra bands of a-SiC at the edges of the metal traces and possibly bury cavities in the process.

In the case of the samples depicted in Figure 3.6, a close inspection revealed continuity of the a-

SiC layer covering the rabbit ears and the metal traces, however under aging conditions, the top a-

SiC layer was observed to delaminate from the underlying substrate as indicated by the series of

SEM and optical images shown in Figure 3.7. Delamination of the metallization occurred at the traces and the electrode sites, and in some cases, complete removal from the underlying a-SiC

layer. These failed MEAs exhibited large reductions in impedance as the samples were aged. For example, for a-SiC MEA with 400 nm sputtered metallization (and exposed Au electrode sites) soaked in PBS at 87 oC, the average impedance remained steady for 6 weeks, but failed abruptly afterwards, as shown in Figure 3.8 (b).

Figure 3.9. The a-SiC MEA soaked for 20 weeks in PBS at 87oC (a) before and (b) after soaking. Metallization was by vapor deposition.

Most of the electrodes showed large impedance changes by week 9, and for those soaked at 37oC, impedance changed drastically at week 13. Similar observations were made for electrode sites coated with SIROF under elevated-temperature accelerated soaking conditions. The failure observed was associated with complete discontinuity of the metal traces and also delamination of the SIROF coatings from the electrode sites. This resulted in a drastic increase in the average electrode impedance after 5 weeks of soaking in in PBS at 87oC (Figure 3.10).

Evaporated metallization provided a solution to this problem. The evaporated metal flux is highly directional and is incident on the substrate with an extremely small angular distribution such that photoresist sidewall coatings are substantially reduced or eliminated completely. An example of an a-SiC MEA developed using evaporated metallization and soaked in PBS at 87oC for 20 weeks is shown in Figure 3.9. The a-SiC MEA layers remained intact without delamination after the soak

period. Assuming an acceleration factor of 32, the results predict that the a-SiC MEA layers will remain intact for a period of over 10 years at a physiological temperature of 37oC.

A comparison of average impedance at 1 kHz of aging samples soaked at 87oC is shown in Figure

3.10. The metal traces were either sputter-deposited or evaporated. The electrode sites were coated with SIROF to reduce the electrode impedance. From these results it can be seen that the average electrode impedance for MEAs created using sputter-deposited metal traces is substantially increased after week 5 whereas that for MEAs with evaporated metal traces only increases slightly over 17 weeks. A look at the individual impedance curves shows that about 50% of the MEAs fabricated with sputtered metallization have increased impedance at week 5 as shown in Figure

3.10 (b).

Figure 3.11. Comparison of (a) CV and (b) EIS of a-SiC MEAs soaked in PBS at 87oC for 17 weeks.

The increase in impedance is related to the exposure of the metal traces when the top layer of the a-SiC and gold delaminate from the bottom layer and fracture. In other cases, the SIROF

completely delaminated from the electrode sites. A comparison of the average CV curves measured at 50 mV/s, and the electrochemical impedance spectroscopy for a-SiC MEAs developed with evaporated metallization is shown in Figure 3.11. The figure compares CVs and EIS of the samples before soaking and after 17 weeks in PBS at 87oC. The slight increase in average impedance (Figure 3.11(b)) corresponds to a small tilt in the average CV curve (Figure 3.11(a)).

The average impedance increased from 8 ± 0.6 kΩ to 10.7± 1.6 kΩ whereas the cathodal charge storage capacity, estimated from the time integral of the cathodal current in the CVs, increased from 76.9 ± 1.0 mC/cm2 to 97.2 ± 36.7 mC/cm2.

1.19 Conclusions

With minimized residual stress in the a-SiC film and the use of stress engineering in the metallization, one can successfully produce the planarity of a-SiC MEAs necessary for inserting into tissue. The long-term performance of the a-SiC MEA relies on structural integrity of the layers and details of microfabrication and patterning processes. Rabbit ear defects in metallization result in delamination of the layers under accelerated soak conditions. Appropriate fabrication of the metallization by e-beam evaporation should ensure the robustness and long- term usability of a-SiC MEAs.

CHAPTER 4

AMORPHOUS SILICON CARBIDE ULTRAMICROELECTRODE ARRAYS FOR

NEURAL STIMULATION AND RECORDING1

Authors: Felix Deku1, Yarden Cohen2, Alexandra Joshi-Imre1, Aswini Kanneganti1, Timothy J. Gardner2, Stuart F. Cogan1

1Department of Biomedical Engineering

The University of Texas at Dallas

800 West Campbell Road

1Richardson, TX 75080-3021

2Department of Biology and Biomedical Engineering

Boston University

One Silber Way

Boston, MA 02215

2552/aa8f8b

1.20 Author Contributions

Felix Deku, Dr. Stuart Cogan, and Dr. Timothy Gardner designed and conceptualized the study.

Felix Deku designed and fabricated the ultramicroelectrode arrays and performed electrochemical characterizations. Felix Deku and Dr. Aswini Kanneganti performed the implantation and acute recording of extracellular neural signals from rat motor cortex. Dr. Yarden Cohen implanted and recorded neural activity from songbird brain. Felix Deku, Dr. Alexandra Joshi-Imre and Dr. Stuart

Cogan provided useful discussions, interpreted results and contributed to manuscript preparation.

This work was published in the Journal of Neural Engineering, 2018.

1.21 Abstract

Objective: Foreign body response to indwelling cortical microelectrodes limits the reliability of neural stimulation and recording, particularly for extended chronic applications in behaving animals. The extent to which this response compromises the chronic stability of neural devices depends on many factors including the materials used in the electrode construction, the size, and geometry of the indwelling structure. Here, we report on the development of microelectrode arrays

(MEAs) based on amorphous silicon carbide (a-SiC). Approach: This technology utilizes a-SiC for its chronic stability and employs semiconductor manufacturing processes to create MEAs with small shank dimensions. The a-SiC films were deposited by plasma enhanced chemical vapor deposition and patterned by thin-film photolithographic techniques. To improve stimulation and recording capabilities with small contact areas, we investigated low impedance coatings on the electrode sites. The assembled devices were characterized in phosphate buffered saline for their electrochemical properties. Main results: MEAs utilizing a-SiC as both the primary structural

element and encapsulation were fabricated successfully. These a-SiC MEAs had 16 penetrating shanks. Each shank has a cross-sectional area less than 60 µm2 and electrode sites with a geometric surface area varying from 20-200 µm2. Electrode coatings of TiN and SIROF reduced 1 kHz electrode impedance to less than 100 kΩ from ~2.8 MΩ for 100 µm2 Au electrode sites and increased the charge injection capacities to values greater than 3 mC/cm2. Finally, we demonstrated functionality by recording neural activity from basal ganglia nucleus of Zebra

Finches and motor cortex of rat. Significance: The a-SiC MEAs provide a significant advancement in the development of microelectrodes that over the years has relied on silicon platforms for device manufacture. These flexible a-SiC MEAs have the potential for decreased tissue damage and reduced foreign body response. The technique is promising and has potential for clinical translation and large-scale manufacturing.

1.22 Introduction

Chronically implanted microelectrode arrays (MEAs) for recording extracellular neural activity are central to scientific studies of neural circuit function [16], [21], [128]–[132]. These recordings help in understanding how neurons encode information and how neural signals can be decoded to provide insights into brain adaptation and learning. MEAs that penetrate the pial surface of the brain to provide cortical recordings are typically fabricated in silicon using thin-film processing techniques [130], [133], [134] or assembled as an array of polymer-insulated microwires [135],

[136]. The cross-sectional dimension of the intra-parenchymal shanks or wires on these arrays vary with each particular design, but generally cross-sectional areas exceed 500 µm2 with a minimum transverse dimension greater than 10 µm. The most common MEAs of this type are those based on the University of Michigan design (Michigan Probes) [130], [134], and those developed at the

University of Utah known as the Utah Array [137]. The ability of these multielectrode arrays to provide reliable chronic recordings is limited by the foreign body response and insertion trauma that results in gliosis, scar formation, and a substantial loss of viable neurons within about 50 µm of the electrode site [16], [24], [51], [138], [139]. In addition, degradation of polymer encapsulation, loss of low impedance coatings at electrode sites, and delamination in thin-film multilayer structures have been identified as likely contributors to the reduced reliability of these devices for chronic recording [10], [140].

Recently, it has been recognized that the severity of the foreign body response is greatly reduced with implanted microelectrodes that have transverse cross-sectional dimension less than approximately 10 µm [47], [48], [90], [138]. Polymer insulated carbon fiber microelectrodes with shank diameters of 8.4 µm or less have demonstrated minimal gliosis and provided high quality single-unit recordings in acute and chronic studies [51], [72], [73], [139], [141], [142]. However, it has proved challenging to manufacture carbon-fiber MEAs by processes that are amenable to fabricating large numbers of MEAs with a consistent electrode geometry. Here, we report on the development of MEAs based on amorphous silicon carbide (a-SiC). Amorphous SiC has emerged as a promising material for encapsulating implanted neural devices [58], [60], [61], [65]. Films of a-SiC are well-tolerated in the cortex and resistant to corrosion or dissolution in saline [58], [62],

[64], [66]. In addition, a-SiC exhibits high intrinsic stiffness and, as a thin-film, is highly flexibility, comparable to carbon fiber microelectrodes.

The a-SiC MEAs fabricated for this study have intra-parenchymal shanks with a maximum transverse dimension of 10 µm and cross-sectional area less than about 60 µm2. These extremely small dimensions result in electrode sites with geometric surface areas (GSAs) less than 200 µm2

and potentially as small as 20 µm2. Consequently, the electrodes have a higher impedance and more limited charge-injection capacity for stimulation than typical silicon-based microelectrodes.

However, the electrode sites are sufficiently small, in at least one dimension, to exhibit ultramicroelectrode (UME) behavior [80]–[82] and consequently higher charge densities, injectable charge per unit area, than larger microelectrodes with GSAs greater than 1000 µm2. The electrochemical behavior of both titanium nitride (TiN) and sputtered iridium oxide (SIROF)

UMEs on the a-SiC MEAs has been investigated in buffered physiological saline with a notable decrease in impedance and increase in charge-injection capacity compared with microelectrodes

(electrodes with typical GSA~2000 µm2) of the same electrode materials. Finally, the ability of a-

SiC UMEs to provide neural recordings was evaluated acutely in the basal ganglia nucleus of the

Zebra Finch and in the motor cortex of rat.

1.23 Methods

The a-SiC MEAs were developed using standard thin-film fabrication processes. The a-SiC films were deposited by plasma enhanced chemical vapor deposition (PECVD) and patterned by thin- film photolithographic techniques. Electrode sites, bond pads and the device geometry were defined by reactive ion etching using an SF6 plasma chemistry. Electrical connection to the a-SiC

MEAs was achieved by mounting an Omnetics connector directly on the array. The assembled devices were characterized in phosphate buffered saline (PBS) for their electrochemical properties and evaluated in the basal ganglia nucleus of Zebra Finches and in the motor cortex of rat for their neural recording capabilities.

1.23.1 PECVD a-SiC deposition

Amorphous SiC films were deposited in a PlasmaTherm Unaxis 790 Series PECVD system at a substrate temperature of 325°C, RF power density of 0.20 Wcm-2 (13.56 MHz), and pressure of

1000 mTorr using a reactive gas mixture of SiH4 and CH4 at flow rates of 12 sccm and 36 sccm respectively. The total gas flow rate into the reaction chamber was maintained at 800 sccm using

Ar as the carrier gas. Scanning electron microscopy (SEM) and atomic force microscopy (AFM) were used to determine the surface morphology of the films. Film thickness was measured with a

Nanometric NanoSpec 6100 analyzer using data collected from 20 randomly selected points across the a-SiC film. Film stress was measured using silicon control wafers with a Toho Technology

FLX-2320 stress analyzer and estimated using the Stoney equation [143].

1.23.2 Fabrication of a-SiC MEAs

The multilayered a-SiC MEAs were fabricated on prime-grade 100 mm Si (100) wafers using polyimide (HD Microsystems, PI 2610) as a release layer between the a-SiC MEA and silicon substrate. The detailed fabrication steps are summarized in Figure 4.1. Briefly, an approximately

1 µm polyimide film is spin-coated onto the Si wafer and cured at 350oC for one hour under nitrogen. A 2-µm thick a-SiC film is then deposited over the polyimide. Metal traces are formed on the a-SiC by sputter deposition using lift-off photolithography to define the metal pattern. To facilitate liftoff, a non-photosensitive LOR 5A (Microchem Inc) layer and a photosensitive Shipley

(S1813, Microposit) photoresist layer are spin-coated consecutively on the a-SiC in a process designed to create an undercut in the two-layer resist. The metallization is deposited by DC

sputtering and comprised either a three-layer coating of Ti/Au/Ti (30 nm/250 nm/30 nm, thickness) or a four-layer coating of Ti/Au/Pt/Ti (30 nm/150 nm/150 nm/30 nm, thickness).

After the metal deposition, the carrier wafers are immersed in EBR-PG (Microchem Inc.) for metal lift-off. A second 2-µm layer of a-SiC is deposited over the metal traces to provide complete

encapsulation of the metal in a-SiC. Vias for the electrode sites and bond pads are opened by removing the top a-SiC layer and upper titanium layer by reactive ion etching in an SF6 plasma at low pressure using an inductively coupled plasma (ICP) etcher (PlasmaTherm). The opening for the electrode sites was 2 µm by 50 µm, resulting in a site GSA of 100 µm2 in the absence of a low impedance electrode coating. Singulation of the a-SiC into individual MEAs is also accomplished in a second SF6 RIE step in which the a-SiC superstructure of the MEA is protected by a 6.5 µm thick layer of photoresist. After the RIE singulation process, the devices are soaked in AZ400T

(AZ Electronic Materials) at 70oC to remove the remaining photoresist and then thoroughly rinsed in deionized water to remove remaining residue. In the final step, the wafers with a-SiC MEAs are immersed in deionized water at 87ºC until the arrays release from the silicon wafer. Electrical connection to the a-SiC MEAs is obtained with an Omnetics connector mounted on the array using a solder reflow process and indium-based solder paste (Indium Corporation). A medical grade epoxy (Loctite) or dental acrylic (Flow-It ALC, Pentron) is then cured around the base of the

Omnetics connector to provide mechanical strength.

1.23.3 Low impedance electrode coatings

Low impedance coatings of sputtered iridium oxide (SIROF) or porous titanium nitride (TiN) were deposited onto electrode sites by reactive DC sputtering from iridium or titanium metal targets, respectively. Lift-off photolithography was used to restrict deposition to the electrode sites.

However, a narrow band of the electrode coatings is deposited on the rim of the a-SiC at the electrode site resulting in electrode coatings with approximately a 150 µm2 (3 µm x 50 µm) GSA.

The film thickness, determined by surface profilometry, was 300 nm for the SIROF coatings and

2 µm for the TiN coatings.

1.23.4 Electrochemical characterization

The electrochemical properties of the a-SiC UMEs were evaluated at room-temperature in inorganic phosphate buffered saline (PBS), having a composition of 126 mM NaCl, 22 mM

NaH2PO4.7H2O and 81 mM Na2HPO4.H2O at a pH of 7.2, that was purged with argon to remove dissolved oxygen gas [102], [144]. The electrochemical characterization included cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) and voltage transient measurements in response to constant current pulsing. The cathodal charge storage capacity

(CSCc) of the electrodes was calculated from the CV measurements that were made at a sweep rate of 50 mV/s between potential limits of -0.6 V and 0.8 V versus Ag|AgCl [75]. A three- electrode cell comprising the SIROF or Pt UME working electrode, a large surface area Pt counter electrode and a Ag|AgCl reference electrode was used for all measurements. EIS measurements were made with a 10 mV rms AC perturbation about the open circuit potential over a 1 Hz to 105

Hz frequency range. The CV and EIS data were acquired with a Gamry Reference 600plus potentiostat using vendor-supplied software. The stimulation charge-injection capacity of the electrodes was determined from voltage transient measurements using cathodal monophasic current pulsing combined with active control of the interpulse potential of the electrode to maintain charge-balance [104]. Anodic bias potentials of 0.0 V to 0.8 V versus Ag|AgCl in the interpulse period were investigated [75]. A pulse width of 200 µs and repetition rate of 50 pulses per second were used throughout the study. The maximum charge-injection capacity of the electrode coatings

was determined by increasing the current during pulsing until the iR-corrected electrochemical potential of the electrode reached a value of -0.6 V Ag|AgCl.

1.23.5 Acute implantation and neural recording

In the first study of neural recording with a-SiC MEAs, the arrays were implanted in the basal ganglia nucleus, area X, of Zebra Finch. All procedures were approved by the Institutional Animal

Care and Use Committee of Boston University and follow methods previously described [72].

Briefly, the zebra finches were anesthetized with 4 % isoflurane (mixed with pure oxygen at 0.5

L/min flow rate) and maintained at 1-2 % isoflurane during the surgical procedure. A 120 µL dose of analgesic Meloxican (1 % in PBS) was injected intramuscularly into the right breast at the start of the surgical procedure and the animal was placed into a stereotaxic instrument. An incision was made in the scalp anterior-posterior (AP) axis and the outer bone leaflet removed using a dental drill. Using an ophthalmic scalpel, the lower bone leaflet was carefully removed [145] exposing a 1 mm diameter portion of the dura. An incision of 1 mm was performed in the dura to facilitate the MEA penetration into the brain. The a-SiC MEAs were mounted on a manual manipulator attached to the stereotaxic instrument and slowly lowered through the brain. Prior to mounting the arrays, the a-SiC MEAs were embedded in a droplet of polyethylene glycol (PEG) following methods described elsewhere [73]. Recordings were made from spontaneously active cells using an Intan RHD 2000 system with a 16-channel unipolar input head stage. Data was sampled at 20 kHz with filter cutoffs set to 350 Hz and 7.5 kHz.

Neural recording with the a-SiC MEAs was also investigated acutely in rat motor cortex using the procedure described below and approved by the institutional animal care and use committee

(IACUC) at The University of Texas at Dallas. The MEAs were implanted in the motor cortex of

male Long-Evans (Charles River Labs, MA) rats (~300 gram) under anesthetized conditions. The animals were anesthetized by intraperitoneal injection of a mixture of 65mg Ketamine, 13.33 mg

Xylazine and 1.5 mg Acepromazine per kg body weight of the animal post isoflurane (2-3%) induction. Atrophine (0.05mg/kg) was also administered to prevent cardiovascular depression under anesthesia. The hair of the animals’ scalp was expunged, cleaned and disinfected with 10% povidone iodine solution and 70% ethanol. The head of the animal was then placed in a stereotaxic frame. Prior to the midline incision, dexamethasone was administered subcutaneously over the shoulders followed by 0.16 cc of 0.5% lidocaine under the incision site. The skull was exposed by retracting the surrounding skin and muscles and scraping the periosteum above the planned implant site. A dental drill (Fine Science Tools, Foster City, CA) was used to create holes on the contralateral side of the skull to secure two bone screws for grounding and securing of the

Omnetics connector. A 2mm by 2mm craniotomy was performed anterior to the bregma for array implantation. To prevent thermal damage due to drilling, the skull surface was periodically flushed with sterile saline at 37oC. To constrict local blood vessels and reduce bleeding, gauze moistened with epinephrine at 1 mg/ml was applied to the craniotomy for 1 minute, followed by an additional flush of the craniotomy with 0.9% saline. The dura (~80-100µm) was gently retracted with a 30- gauge needle tip. The a-SiC MEAs were mounted onto a custom-built array holder and aligned to the skull opening with the electrode tips facing the cortex using a micro-positioner (Kopf

Instruments, CA) attached to the stereotaxic frame. The PEG stabilized tips were advanced using the micro-positioner set at a speed of 50 µm/s to ensure dissolution of the PEG as the arrays are advanced into the brain. The arrays were inserted to a depth of ~2-3 mm from the cortical surface.

Spontaneous neural activity was recorded using a 32 channel OmniPlex data acquisition system

(Plexon Inc., USA). Post signal amplification (20 X), the signal band passed filtered between 50

Hz and 8 kHz was digitized at 40 kHz and recorded for 5 minutes.

1.24 Results

1.24.1 Residual film stress of the a-SiC

A significant challenge in developing the a-SiC MEAs was control of the intrinsic film stress.

Building multilayer structures requires stress balance between individual layers to avoid interlayer delamination, and to maintain the desired planar geometry once the MEAs are removed from the silicon carrier wafer. For PECVD a-SiC, the intrinsic film stress is dependent on deposition conditions such as RF power density, deposition temperature, deposition pressure and the flow rates of the SiH4 and CH4 reactive gases [111]–[114].

Using the deposition conditions described above, we obtained amorphous films with a compressive stress in the range of 80-120 MPa for a 2 µm thick film. To balance the stress across the multilayer device, the thickness and tensile stress in the Ti/Au/Ti or Ti/Au/Pt/Ti metallization was adjusted by controlling sputtering power and pressure during metal deposition. A metal stress of 40-100

MPa in tension produced planar MEAs with good interlayer adhesion.

As shown in Figure 4.2(a), the rms surface roughness for the 2 µm thick a-SiC was 3.75 nm, indicating a surface that is sufficiently smooth for subsequent photolithographic processing. SEM inspection of the 2 µm thick a-SiC films under high magnification revealed a nodular surface morphology as shown in Figure 4.2(b).

1.24.2 Amorphous-SiC MEA fabrication

For intracortical studies, the a-SiC MEAs were fabricated with 16 penetrating shanks with one electrode per shank. Each shank was 4 - 5 µm thick, 9 µm wide and terminated in a sharply pointed tip with an angle of 10 -14˚. The shank length for this study was 4 mm and the GSA of the Pt and

Au electrode sites was 100 µm2 and 150 µm2 (3 µm x 50 µm) for the SIROF or TiN coated electrodes. The electrode sites were located at the distal end of the shank as shown in Fig. 3. The metal interconnects, either gold or a gold/platinum bilayer, run approximately along the neutral axis of each shank, connecting the electrode sites at the distal end and bond pads at the proximal end. A top layer a-SiC was deposited over the metallization and the first layer of a-SiC to completely encapsulate the metal traces. Openings in the top a-SiC were formed by reactive ion etching to expose the electrode sites and bond pads. Figures 4.3(f) and 4.3(g) show an electrode site opened through the top a-SiC layer. The photolithographic patterning process ensures control of the GSA of the electrode site which is recessed approximately 2 µm. As seen in Figure 4.3(g),

the ICP etching process produces a near-vertical side wall profile. The depth of the recess is such that the current distribution during stimulation should be approximately uniform over the surface of the recessed electrode [146], [147].

To release the devices from the carrier wafer, the a-SiC MEAs are immersed in DI water. When carefully withdrawn from the DI water, the 16 individual shanks form a shank bundle with an approximate diameter of 40 µm (Figure 4.3(a)). The tip of the bundle is shaped by the layout design of the shank arrays. Figure 4.3(b) and 4.3(d) shows two different tip configurations of as-fabricated devices prior to release and their resultant bundle-profiles (Figure (4.3(c) and 4.3(e)) after release.

Bonding pads, 750 x 500 µm in dimension, with a 635 µm pitch designed to mate with a 16- channel Omnetics connector (Omnetics, A79040-001), are located at the proximal end of the array.

An indium-tin eutectic solder paste (Indium Corporation) reflowed at 200 °C was used in creating electrical connection between the bond pads and the connector. Figure 4.4 shows the a-SiC MEA after fabrication (4.4(a)), after the Omnetics connecter has been soldered (4.4(b)) and after the connector is securely fixed with epoxy (4.4(c)).

1.24.3 Electrochemical characterization

Cyclic voltammograms (CVs) of gold, platinum, SIROF and porous TiN on the a-SiC MEAs are compared in Figure 4.5(a). Each CV is an average of 10 CVs from one MEA. The CSCc of the

SIROF and TiN-coated UMEs was 35±2.2 mC/cm2 and 12±2.8 mC/cm2 (Mean±SD, n=10),

2 respectively. The uncoated platinum and gold UMEs had lower CSCc’s of 10±1.7 mC/cm and

2 2±0.3 mC/cm , respectively. These CSCc’s were calculated over a nominal water window range of -0.6 V to 0.8 V Ag|AgCl and a higher CSCc would be expected for the TiN if the negative potential limit was extended to the TiN water reduction potential which is close to -0.9 V Ag|AgCl

[75], [110]. Impedance spectra for each UME, normalized to the electrode GSA, are compared in

Figure 4.5(b). As expected, the UME impedance measured at 1 kHz was reduced by at least two orders of magnitude for both TiN and SIROF coatings compared to the gold UMEs. The average impedance values measured at 1 kHz for SIROF, TiN and Pt are 90.2±26.1 kΩ, 31.1±7.3 kΩ and

481.3±67.9 kΩ respectively, compared to 2.86±0.3 MΩ for Au UME sites. The ability of the

SIROF and TiN-coated UMEs to deliver charge for stimulation was investigated by voltage transient measurements in response to constant current pulsing. Figure 4.6(a) shows representative voltage transient responses for anodically biased and unbiased electrodes at the current amplitude that drives the iR-corrected electrode potential (Emc) to the –0.6 V water electrolysis limit [75],

[83], [110]. The average maximum driving current for TiN-coated UMEs was 24.2 µA (0 V bias)

and 46.9 µA (0.8 V bias), and for SIROF-coated UMEs, 25.4 µA (0 V bias) and 114.6 µA (0.8 V bias). The maximum charge injection capacity (Qinj) of the TiN and SIROF UMEs as a function of anodic bias is shown in Figure 4.6(b).

The charge injection capacity for SIROF increased with anodic bias from ~ 3.4 mC/cm2 at a 0.0 V bias to 15.3 mC/cm2 at a 0.8 V bias. A similar trend was observed for TiN whose charge injection capacity increased monotonically from 3.2 mC/cm2 to 6.2 mC/cm2 over the same anodic bias range. The Qinj values reported here are generally higher than those reported for microelectrode arrays (electrode GSAs ~ 2000 µm2) with the same electrode coatings [75], [110].

1.24.4 Neural recording

To evaluate the functionality of the a-SiC MEAs, the arrays were implanted in Zebra Finch basal ganglia nucleus Area X. Following the method introduced previously [73], the shanks of a-SiC

MEAs were coated with poly-ethylene glycol (PEG). PEG-coatings proved reliable in creating stiff assemblies in which individual shanks are held apart prior to implanting. During insertion, the

PEG dissolved from the comb allowing the shafts to enter the tissue without buckling. We determined an insertion speed of 50 µm/s to be adequate to ensure dissolution of the PEG during implantation. It is however possible that some PEG may remain on the penetrating shanks and the effect of the PEG-residue on the chronic functionality of the devices is yet to be evaluated. The

data shown in Figure 4.7 were recorded in an acute preparation immediately after surgery using a

Pt a-SiC UME with 100 µm2 GSA. Figure 4.7(a) shows the recorded voltage trace and 7 (b) is an overlay of neuronal spike waveforms from a single channel demonstrating single unit spiking activity. The feasibility of recording spontaneous activity with PEG-stabilized a-SiC MEAs arrays was also tested in anesthetized rat motor cortex.

For this study, a 16-shank a-SiC MEA with SIROF-coated UMEs was mounted on a motorized drive and inserted into the cortex. Figure 4.8(a) shows spontaneous neural activity recorded simultaneously on three channels in a single trial. Distinct spiking activity across multiple channels confirms that the electrode array recorded from spatially selective neuronal population. Post- processing of the recorded activity revealed depth dependent single unit activity on 9 of the 16 electrodes, as the electrodes were inserted incrementally to 2-3 mm below the cortical surface.

Array yield quantified as number of electrodes with identified single unit spikes was 56% for the acute trial

The single unit mean peak-to-peak amplitudes across all recorded electrodes were in the 45-200

µV range. Fig 8(b) shows 3 single unit signals from Channel 1 (CH1) identified after offline processing of the neural spikes. The waveforms of the units show a mean peak to peak amplitudes of 45 µV, 87 µV and 114 µV for Unit A, Unit B and Unit C respectively. The spike shapes shown here are typical to previously reported cortical single unit recording [148]–[150]. Figure 4.8(c)

shows the corresponding autocorrelograms of the sorted units in 4.8(b) computed with a bin size of 2 ms and smoothing with a Gaussian filter (filter width = 3 bins). The autocorrelation function shows a bimodal distribution which clearly indicate distinct single unit spikes. While Units B and

C have a stronger bimodal distribution with refractory periods >1ms, Unit A has a weaker distribution with some multi-unit contamination in the sorted units, indicated from the firing frequency distribution.

1.25 Discussion

Minimizing insertion trauma and the foreign body response to implanted multielectrode arrays is an important objective in the development of recording and stimulation devices for brain-machine interfaces in both research and clinical settings. The insertion trauma and foreign body response lead to electrode encapsulation and death or damage of neurons at distances up to approximately

100 µm from the implant site [16], [21], [23], [25], [151]. The length scale of this tissue reaction is detrimental to neural recordings, since action potential amplitudes decay exponentially with distance from an electrode [152]–[154]. As a result, chronically implanted electrodes are typically limited to recording from a larger population of weak signals, limiting spike sorting and single neuron resolution - a problem that is particularly severe for small animal models with small, densely packed neurons. Approaches to minimizing the foreign body response have included the development of polymer-based electrodes that reduce the elastic modulus mismatch between the electrode and neural tissue [30], [31], [37], [38], [41], the development of wireless interfaces that eliminate wired connections and reduce electrode micromotion [155]–[157], and the use of stiff but highly flexible electrodes with cross-sectional dimensions typically smaller than 10 µm [51],

[72], [73], [142]. The a-SiC MEAs, described here, are designed to address insertion trauma and

foreign body response by employing the latter strategy with the maximum transverse dimension of individual shanks under 10 µm and the cross-sectional area of individual shanks under 60 µm2.

One obvious benefit of employing a-SiC as the primary material of array construction is the ability to fabricate devices by established thin-film deposition and photolithographic patterning processes. The thin-film fabrication of a-SiC MEAs is highly reproducible with inherent flexibility in the design of the arrays.

It is possible to fabricate arrays with individual shanks having multiple electrode sites along the length of the shanks and optionally with different shapes and GSAs. An example of a two-electrode combination of gold electrodes with a GSA of 100 µm2 in a 2 x 50 µm and 10 x 10 µm configuration at the distal end of a shank is shown in Figure 4.9.

This UME geometry is part of an ongoing study to investigate the effect of perimeter-to-area ratio on the in vivo and in vitro electrochemical properties of ultramicroelectrodes. Although arrays with individual shanks having a tetrode layout have been fabricated, we have yet to evaluate the possibility of using this configuration for spike sorting.

Similarly, it is possible to build in geometries that promote splaying when using the bundling properties of the shanks to insert the MEAs into tissue. A 16-channel MEA with shanks having intrinsic curvature intended to direct the splaying action during insertion is shown in Figure 4.10.

An array with the splayed geometry forms a monolithic and tight bundle of shanks when withdrawn from water so that insertion into tissue can be initiated without the shanks buckling.

While we observed the bundled shanks penetrate cortex in the absence of PEG coatings, we have yet to demonstrate that the intrinsic curvature promotes splaying while minimizing insertion trauma. Nonetheless, the geometries shown in Figs. 4.9 and 4.10 give some indication of the options available in designing UME arrays to address different study needs. Irrespective of these benefits, a consequence of reducing the shank cross-sectional dimensions is that at least one linear dimension of the electrode site is in the ultramicroelectrode range [80]–[82] and the reduced GSA results in an electrode impedance that is higher than that of a typical microelectrode used for neural recording [74].

However, UME-dimensioned electrodes have provided high quality neural recordings in both acute and chronic preparations [51], [72], [73]. In previous studies, Gardner and colleagues developed methods of assembling carbon fiber UME arrays and evaluated these electrodes in chronic songbird preparations [72].

These studies confirmed that UMEs can provide stable long-term chronic neural recordings in behaving animals and that small shank dimensions result in minimal foreign body response.

Similar studies also produced important findings regarding the nature of song stability at the level of neural coding in the high vocal center (HVC) of Zebra Finch pre-motor cortex [158], [159].

Implanted a-SiC MEAs successfully recorded simultaneous neural activity in the two animal models evaluated. Since the in vivo studies were intended for pilot feasibility evaluation, the analysis of the recorded signals was limited to the observations of single unit spikes from spontaneous neural activity. Figure 4.11 shows the raw traces for both spike activity obtained from high-pass filtered data and the corresponding local field potentials (low-pass filtered data at

<350Hz) showing the correlations in the temporal distribution similar to those reported elsewhere

[160], [161]. Detailed analysis of the local field potentials, especially in different low frequency bands like delta, spindle and gamma among others will be evaluated in future studies.

Electrical stimulation of neural activity is more challenging with UMEs. The use of SIROF or TiN partially mitigates the impact of reduced electrode size. Charge-injection capacities of 11.5 mC/cm2 and 5.4 mC/cm2, respectively, were observed for 200 µs pulses in buffered saline when the UMEs are biased to 0.6 V, and as high as 15.3 mC/cm2 and 6.3 mC/cm2 respectively at 0.8 V bias. These charge-injection capacities are notably larger than those obtained with similar coatings

on larger microelectrodes [75], [98], [106], [110]. For example, using similar pulse parameters, charge injection limits of approximately 4 mC/cm2 and 0.87 mC/cm2 were estimated for 4000 µm2 iridium oxide and TiN microelectrodes respectively [110]. While the small size of UMEs results in an increase in charge-injection capacity the total deliverable charge remains low even with electrode coatings. The maximum charge that could be delivered in a 200 µs pulse was 23 nC/ph for SIROF and 9 nC/ph for TiN UMEs. The uncoated Pt and Au UMEs exhibited charge per phase values less than 1 nC/ph, likely below the microelectrode charge threshold for neural activation

[84]. It remains to be seen whether reduced tissue damage and the likely presence of healthy neurons in much closer proximity to UME sites compared with conventional microelectrodes will result in reduced charge thresholds for neural activation. The UME concept originated in the context of chemical sensing and quantification using voltammetric electrodes and the extension of this concept to stimulation electrodes has limitations [162]. An ultramicroelectrode may be broadly defined as an electrode in which transport of the electrolyzed species to the electrode occurs via spherical or hemispherical diffusion [79], [80]. Geometrically, at least one dimension of the UME must be less than the diffusion layer thickness of the species involved in the electrode reactions

[80]. For neural stimulation electrodes the electroactive species are the counterions that preserve electrode neutrality and these species are charged.

1.26 Conclusions

MEAs have been fabricated using a-SiC as the primary material of construction. The arrays are designed with shanks that penetrate into target neural tissue with cross-sectional dimensions that are expected to minimize insertion trauma and reduce foreign body response. By employing conventional thin-film processing techniques, a wide variety of array geometries are possible

including the placement of multiple electrodes on a single shank. Since the cross-sectional dimensions of the shanks are typically 5 µm by 10 µm or less, the electrode sites have ultramicroelectrode dimensions with geometric surface areas between 20 µm2 and 200 µm2. These electrodes recorded spontaneous neural activity in acute Zebra Finch nucleus and rat cortical preparations. Implantation into cortex using bundled a-SiC shanks was also demonstrated. Low- impedance coatings of SIROF and porous TiN deposited on the UME sites were investigated as a means of providing sufficient charge for neural stimulation. These coatings provided more than 1 nC/ph within water electrolysis limits in a 200 µs pulse, which is comparable to neural activation thresholds reported in some microelectrode studies [84]. However, additional studies are needed to determine the extent to which the a-SiC MEAs minimize foreign body response and whether the UMEs can provide functionally useful levels of charge-injection for stimulation.

CHAPTER 5

AMORPHOUS SILICON CARBIDE PLATFORM FOR NEXT GENERATION

PENETRATING NEURAL INTERFACE DESIGNS

Authors: Felix Deku1, Christopher L. Frewin1, Allison Stiller1, Yarden Cohen2, Saher Aqeel1, Alexandra Joshi-Imre1, Bryan Black1, Timothy J. Gardner2, Joseph J. Pancrazio1 Stuart F. Cogan1

1Department of Biomedical Engineering

The University of Texas at Dallas

800 West Campbell Road

1Richardson, TX 75080-3021

2Department of Biology and Biomedical Engineering

Boston University

One Silber Way

Boston, MA 02215

1.27 Participation

Felix Deku, Dr. Joseph Pancrazio and Dr. Stuart F. Cogan conceived and conceptualized the study.

Felix Deku designed and fabricated the a-SiC MEAs. Felix Deku, Dr. Bryan Black and Dr.

Christopher Frewin performed the array implantations, neural recording and analyses. COMSOL modelling was performed by Allison Stiller. Insertion force measurements were performed by

Allison Stiller and Felix Deku. Felix Deku, Drs. Christopher Frewin, Alexandra Joshi-Imre,

Yarden Cohen, Timothy J. Gardner, Joseph J. Pancrazio and Stuart F. Cogan provided useful discussions and interpreted results.

1.28 Introduction

Penetrating microelectrode arrays (MEAs) that stimulate or record neural activity usually consist of a base substrate material which may be an insulator or conductor. Typical conducting substrates include silicon [163], tungsten or iridium wire [151], [164] or carbon fiber [72], [73], [85], [165], which provide the backbone and structural stiffness necessary to penetrate the neural tissue. Silicon is typically doped to provide conductivity [166] and is usually insulated so that current conduction is restricted to the doped silicon. A common polymeric coating used to isolate the conducting substrate from surrounding electrolyte is Parylene-C. It is also common practice to use thin-film dielectric materials such as low-pressure chemical vapor deposited (LPCVD) SiO2 to encapsulate polycrystalline silicon traces [167]. In most cases another dielectric material, such as Si3N4, is deposited over the SiO2 to control the intrinsic compressive stress in the SiO2 [168], [169] or to create a multi-layer passivation stack of PECVD SiO2/Si3N4/SiO2 over the conducting trace [170].

The silicon-based microelectrodes, however, have been shown to deteriorate when chronically

implanted [10], [52], [140]. Failure modes associated with silicon-based MEA degradation was recently described following array implantation in non-human primates [52].

Recent studies have shown that flexible neural interfaces may provide an alternative to traditional silicon-based implants and have the potential to greatly improve the chronic longevity of the implanted microelectrodes [151], [171]. Polymers such as polyimide [40], [55], Parylene-C [31],

[34], SU-8 [30], polydimethylsiloxane (PDMS) [29] have been investigated as substrates for neural stimulation and recording microelectrodes. Their low Young’s modulus reduces the mechanical mismatch between neural tissue and implants. Thin-film metal conducting traces such as gold or platinum are used between layers of the polymer substrate connecting electrode sites and bond pads. The insulating layers effectively sandwich the conducting traces. Electrode sites are then created by removing or etching the top layer through a precise and controlled microfabrication process.

Implantation of penetrating polymer-based MEAs have been aided by a delivery vehicle [73],

[172]–[175] or temporary support structure [73], [176]–[178] to minimize buckling during insertion. These strategies increase the critical buckling load [179]. To penetrate the neural tissue without the assistance of support structures, a minimum cross-sectional dimension of the shank

(the part that penetrates neural tissue) is typically greater than 20 µm [30], [32], [55].

Unfortunately, this cross-sectional dimension may still be higher than that required to elude a foreign body reaction, noting that the prevailing thought has been that the minimum geometric dimension required to evade the FBR, at least in one dimension, should be under 10µm [47]. We recently described the development of multielectrode arrays based on PECVD amorphous silicon carbide (a-SiC)[50]. The 16 channel MEAs were developed with two a-SiC layers sandwiching a

thin-film Au conducting trace. Each shank was 10 µm wide and 2 mm long with a shank cross- sectional area below 45 µm2. The greatly reduced shank cross-sectional dimensions may promote mechanical compliance with neural tissue when implanted [45]. The electrode sites were opened at the distal tips by removing the top a-SiC layer and were coated with sputtered iridium oxide films (SIROF) or titanium nitride (TiN) to reduce the electrode impedance [50].

Here, we evaluate different approaches of reducing the critical buckling load of a-SiC MEAs having individual shank cross-sectional areas below 45 µm2 and demonstrate insertion of multiple a-SiC MEA shanks into rat cortex. Acute extracellular neural recording from the a-SiC MEAs following array insertion is also presented. Finally, we describe a variety of novel microelectrode designs that are enabled by the a-SiC platform when a 6 µm thick amorphous silicon carbide (a-

SiC) is used as the MEA superstructure. This thickness was determined to be the minimum necessary to insert a 2 mm long shank or multiple co-linear shanks into rat brain with the dura reflected.

1.29 Methods

1.29.1 Thin film deposition and array fabrication

Plasma enhanced chemical vapor deposited a-SiC films using the Plasmatherm Unaxis 790 series deposition system are used as substrates for MEA development. The a-SiC films are deposited at

o -2 1000 mTorr, 350 C, and 0.27 Wcm using a SiH4: CH4 gas ratio of 1:3. A 2 µm or 4 µm thick a-

SiC forms the bottom layer of the MEA. The bottom a-SiC layer is followed by the deposition of approximately 400 nm thick gold layer that forms the interconnecting traces. Another 2 µm a-SiC layer is deposited over the metal trace and the bottom a-SiC producing an overall thickness of 4

µm or 6 µm thick a-SiC superstructure. The detailed fabrication methodology has been previously reported [50] and is also described in Chapter 4.

1.29.2 Buckling and insertion mechanics

Force measurements were made using a 20g S-Beam load cell (Futek Advanced Sensor

Technology, Inc., Irvine, CA) mounted to a pneumatically controlled micro-positioner (Kopf

Instruments, Tujunga, CA). The sample probe was mounted on a screw which was directly threaded into the bottom of the load cell so that compression forces could be measured as the MEA was inserted into the brain tissue. An insertion speed of 50 μm/s was used for all studies. Before implantation, the probe was lowered until it was directly above the surface of the brain. The load cell was then tared, and the probe inserted 2 mm into the brain at 50 μm/s.

1.29.3 Surgery and a-SiC implantation

All surgical procedures were approved by The University of Texas at Dallas Institutional Animal

Care and Use Committee (IACUC). Long Evans rats were deeply anaesthetized with 5% isoflurane vapor and administered an intraperitoneal KXA cohort consisting of ketamine (65 mg/ kg), xylazine (13.33 mg/kg), and acepromazine (1.5 mg/ kg) cocktail. The anesthesia was maintained with 0.5-1.5% isoflurane throughout the remainder of the procedure. A 1-2 mm square craniotomy was centered 2.5 mm rostral and 2.5 mm lateral to bregma, and bone debris was carefully removed using sterile phosphate buffered solution (PBS). The dura was reflected using a dura pick and kept moist with sterile PBS. The Omnetics 18 pin male connector attached to the a-SiC cortical implant was placed within a NeuroNexus IST implantation tool (NeuroNexus, Ann Arbor, MI USA) and loaded onto a Kopf Model 2650 hydraulic micropositioner (David Kopf Instruments, USA). The

implant was inserted into the cortex at 50 µm/s at a location at the center of the craniotomy, deviating only enough to avoid large surface vasculature. The dura was sealed using Kwik Cast silicone elastomer (World Precision Instruments, USA), followed by a layer of GLUture

Octyl/Butyl cyanoacrylate glue (World Precision Instruments, USA). A protective head cap was constructed using two-part dental cement (Stoelting Co., USA) which served to secure and support the implant as well as protect the surgical site. The scalp wound was sutured, and the animal was administered an intramuscular injection of Cefazolin (5mg/Kg), a subcutaneous injection of sustained release Buprenorphine (0.15 mg/Kg), and 2 mL of 0.9% saline. The rat was individually housed following implantation. Clavamox was administered orally and buprenorphine was administered every 72 hours for one week.

1.29.4 In Vivo Recording and Analysis

Following construction and curing of the surgical head cap, 10-minute spontaneous recordings were collected using an OmniPlex Neural Acquisition System (Plexon Inc., USA) connected to the a-SiC array via the Omnetics connector and 16-channel digital headstage. Wideband signals

(0.1 – 7000 Hz) were recorded simultaneously from all 16 electrodes at 40 kHz sampling frequency and later filtered offline using a 4-pole Butterworth high pass filter (250 Hz). A -4σ threshold based on RMS noise calculations was applied to filtered continuous data to identify potential waveforms (or spikes). Single units were identified manually based on 2D principal component clustering using Plexon’s Offline Sorter software (Plexon, USA). Sorted units which were not comprised of at least 100 individual spikes or which exhibited greater than 0.5% spike refractory period violations were excluded from analysis. Signal-to-noise ratios (SNR) were calculated by

dividing the mean peak-to-peak amplitude of each unit by the adjusted RMS noise of the associated channel, which excluded values greater or less than ±4σ of the filtered continuous signal.

1.30 Results and Discussion

1.30.1 Insertion strategies of ultrathin a-SiC shanks into cortex

5.4.1.1. PEG-stabilized shanks

While shanks with very small cross-sectional area promise reduced foreign body reaction, insertion of individual shanks into the neural tissue proved challenging. One approach successfully employed was coating the shanks with polyethylene glycol (PEG) which temporarily stiffened the

shanks while leaving a small portion of the tips exposed [73]. This approach increases the buckling threshold of the shanks and allows the arrays to be implanted without buckling. A similar method of array implantation was employed and successfully inserted a 4 µm thick versions of the a-SiC arrays into rat brain. An example of an array coated with PEG (MW 2000) prior to implantation is shown in Figure 5.1. An insertion rate of 50 µm/s was used to insert the shanks so that, as the array is slowly advanced into the neural tissue, the PEG coatings dissolves away.

5.4.1.2. Bundled shanks

Another successful approach introduced by Guitchounts et al. when working with carbon fiber ultramicroelectrodes was to draw the fibers into a bundle allowing the individual fibers to provide mechanical support to each other during array insertion [72]. This approach also increases the overall cross-sectional area of the bundled fibers and increases the buckling threshold for insertion.

Since the shanks on the bundled array are held together by weak Van-der-Waals forces, they separate upon insertion and spread out into the brain following the path of least resistance defined by the mechanical heterogeneity of the brain [72].

The 4 µm thick a-SiC arrays were successfully inserted using this approach, however unlike carbon fibers, we observed that the shanks of the a-SiC MEA twisted together or intertwined when drawn out of water. The tangled shanks prevented the individual shanks from separating and splaying completely when implanted. Figure 5.2(a) shows a bundled a-SiC arrays formed when the shanks are drawn out of water. Figure 5.2(b) shows the tip geometry of the bundle and Figure 5.2(c) shows a bundled 8-channel a-SiC array prior to rat cortical implantation.

Figure 5.2. Bundles of 16-channel a-SiC MEA when drawn out of water (a) showing (b) tip geometry. Insertion of a bundled 8-channel a-SiC array into rat cortex (c).

5.4.1.3. Reduction of effective shank length

Another factor that influences the critical buckling load is the effective length of the shanks. The buckling threshold decreases with increasing length of the shank making insertion of arrays with ultrathin geometries into deep structures of the brain extremely difficult. For carbon fiber arrays, a maximum length of 0.5 mm can be inserted in rat brain without buckling [73]. Patel et al. [73] developed a silicon support structures, 0.75 -1 mm long, to insert 0.5 mm long carbon fibers to deeper cortical layers by securing the fibers within silicon grooves using epoxy. One obvious advantage of the a-SiC technology over carbon fiber is that structures that will reduce the effective length of the shanks can be designed in situ. We designed and developed the webbed arrays as

shown in Figure 5.3 with an effective shank length of 1 mm. The individual shanks are fused in pairs by a-SiC film interconnects as the shanks approach the base of the MEA. A-SiC MEAs with ultrathin shank geometries (4 µm thick x 10 µm wide) have been successfully implanted when the shanks are webbed.

Figure 5.3. Webbed a-SiC MEA with an effective shank length of 1 mm and features a return electrode as part of the MEA structure

5.4.1.4. Insertion of individual shanks

A trade-off between flexibility and stiffness is required when developing compliant microelectrode arrays for cortical applications [179]. Insertion of ultrathin flexible microelectrodes into neural tissue usually fails during implantation. The flexural rigidity (a product of the Young’s modulus of the material and moments of inertia of the cross-section) of the shank is related to the critical

buckling load by Equation 1 (see also Section 1.2.1) where Pcr is the critical buckling load, E is the Young’s modulus, I is the moment of inertia of cross-section, l is the length and K is the column effective length factor (one fixed end, one pinned end =2.045).

휋2퐸퐼 푃 = …………………………………………………………. Equation 1 푐푟 (퐾푙)2

The critical buckling load is the maximum axial load a shank can experience that will not cause lateral deflections. For a microelectrode shank to successfully penetrate the pia mater of a rat brain it is generally expected that its critical buckling load has to be larger than the tissue insertion force, which has been estimated to be around 0.5- 2 mN [53], [56], [180]–[182].

Since the moment of inertia of the cross-section, which influences the critical buckling load, depends greatly on the thickness of the shank, COMSOL finite element modeling was used to predict the thickness necessary to insert a 2 mm long a-SiC shank into neural tissue.

Force values during a buckling test with a single-shank dummy a-SiC probe with a 6 µm thick and

7 µm wide cross-section are shown in Figure 5.4(a). The probe was lowered against a glass surface at a speed of 50 µm/s. Lowering was paused upon visible observation of buckling to accurately record the force at buckling. The recorded buckling force of 0.69 mN was in agreement with the

COMSOL modeling prediction of 0.61 mN. Upon visible inspection, the shape of buckled shank

(Figure 5.4(b)) was also in good agreement with predictions from the COMSOL model (Figure

5.4(c)).

The single a-SiC shank was implanted into rat cortex. Force-displacement curve during implantation of the single shank with 6 x 7 µm cross-section at 50 µm/s is shown in Figure 5.5.

From the curve, the point of penetration of the probe corresponds to an insertion force of 0.35 mN.

Forces due to brain micromotion (inset) after the probe was implanted to full 2 mm depth shows that the indwelling shank experience an extremely low tissue force which relaxed at a rate of ~2.2

µN/s. We have successfully implanted single shank and multiple co-linear a-SiC arrays with a thickness of 6 µm into a 0.6 % agarose gel phantom and into rat brain at an insertion rate of 50

µm/s. To prevent the a-SiC arrays from forming bundles, a minimum inter-shank distance of 100

µm is required for the 7-10 µm wide shanks investigated.

1.30.2 Neural Recording

To determine whether 6 µm a-SiC MEAs with SIROF electrode coatings could be used for in vivo single-unit extracellular recordings, we performed 10-minute electrophysiological recordings immediately following implantation. Figure 5.6(a) shows three representative filtered continuous recordings from a single a-SiC array. Extracellular spikes were well-resolved and sorted based on characteristic waveform shape and 2D principal-component space clustering into single units

(Figure 5.6(b)). We observed distinguishable single units on between 25 and 75% of electrode sites, with the total number of units ranging from 4 to 16. These units had mean peak-to-peak amplitudes ranging from 118.5 to 287.7 µV, with a cumulative mean amplitude of 179.4 ± 18.4

µV and SNR of 24.1 ± 2.2. Table 5.1 contains RMS noise, mean amplitude, and SNR values for all 3 implanted arrays as well as cumulative means. These data suggest that 6 µm a-SiC MEAs are stiff enough to penetrate into the cortex without compromising their mechanical or electrical stability and their ability to record single-unit activity.

Figure 5.6. Acute extracellular action potentials recorded using 6 µm a-SiC MEAs. (a) Filtered continuous data traces from three representative electrodes on Array 1. Vertical and horizontal scale bar represents 125 µV and 1.75 s, respectively. (b) Left - Representative 2D principal component space indicating clear separation from the noise (central gray cluster). Right – Associated single units, indicating characteristic extracellular waveform shape. Vertical and horizontal scale bar represents 175 µV and 0.6 ms, respectively.

Table 5.1. Active electrode yield (AEY) percentage, total number of units, mean peak-to-peak amplitude, RMS noise, and SNR per array and cumulative values across all arrays. AEY (%) # of units Mean Vpp (μV) RMS noise (μV) SNR Array 1 75 16 179.0 ± 19.8 10.2 ± 1.8 25.6 ± 2.9 Array 2 25 4 287.7 ± 64.4 8.8 ± 0.2 30.8 ± 6.8 Array 3 31.3 7 118.5 ± 12.2 7.8 ± 0.4 16.7 ± 1.7 Cumulative 43.75 % 27 179.4 ± 18.4 8.9 ± 0.6 24.1 ± 2.2

An example of average waveforms of sorted single units detected on an indwelling a-SiC MEA 14 days post-implantation is shown in Figure 5.7.

Figure 5.7. Average waveforms of sorted single units detected on SIROF-coated a-SiC arrays 14 days post-implantation in rat cortex

1.30.3 Electrochemical characterization

A comparison of the electrochemical impedance spectroscopy of 100 µm2 SIROF-coated sites before (in PBS) and 1 day after implantation is shown in Figure 5.8(a). The average impedance magnitude at 1 kHz increased from 88.5 kΩ in PBS (pH=7.14) to 355 kΩ in vivo. Figure 5.8(b) shows the average CV curves from 10 electrode sites measured one day post-implantation. The average charge storage capacity, calculated from the time integral of the cathodal current from each CV curve, is 19.4 ± 2.4 mC/cm2 (mean ± s.d., n=10).

1.30.4 Novel microelectrode designs enabled by a-SiC platform technology

5.4.4.1. Barb Probes

A large percentage of intracortical microelectrode arrays designs have their active electrode site directly on the shank tip penetrating the neural tissue. For example, the Blackrock microelectrode

array, carbon fibers, or microwires have the active sites located at the tips of their penetrating shanks. On the other hand, the standard Michigan probe has electrode sites located linearly on the shank and recessed slightly from the top surface. The a-SiC arrays have electrode sites on the penetrating shank recessed 2 µm from the top of the MEA and about 50 µm from the distal tips. It may however be advantageous to decouple the shank bearing the sites from the primary shank penetrating the neural tissue in which case the arrays appear to be drawn through tissue rather than being push through tissue during insertion. Seymour et al. [91] described the fabrication of neural probes with a conventional stiff, 42 µm thick SU-8 and Parylene C backbone but with an adjoining

Parylene-C lattice structure.

Figure 5.9. Barb Probe showing linear arrangement of electrode sites on hooks attached to a central spine. Electrode sites are located on the tips.

The electrode sites were located on the lattice structure and the stiff backbone ensured the MEA could penetrate the tissue successfully. Inspired by this idea, and by the shape of a bee stinger, we designed the barb probe shown in Figure 5.9. The overall a-SiC thickness used for the MEA is 6

µm. The barbed probe has a central spine connecting the shanks positioned at a 35o angle to the spine. The shank bearing the electrode site is 100 µm long from the edge of the spine with a 50

µm2 (2 x 25 µm) electrode site. Inter-barb distance is 50 µm.

5.4.4.2. Flexible floating probes

The flexible floating (flexfloat) microelectrodes have a similar overall design to a 16-channel single shank Michigan probe as shown in Figure 5.10. The flexfloat array is created from a 6 µm thick a-SiC with the electrode sites recessed 2 µm from the top surface. Additional extraneous a-

SiC around the electrode site is removed so that the electrode floats on the bottom 4 µm SiC layer

(upper left). Multiple versions of the flexfloat probes have been developed. In one version, a total of 6 m thick silicon carbide is removed from the electrode site creating a fluid passage around the electrode site (upper right). The cut-outs allow free movement of ions and dissolved molecules around the electrode site. It may also allow the electrode site to more effectively record neural activity from both sides of the probe. To further improve the circulation, a version of the flexfloat probe was designed to have a hollow circular electrode site (right bottom). The a-SiC within and around the hollow circle is completely etched out. All flexfloat a-SiC probes have GSA of 250

µm2.

Figure 5.10. Flexible floating probes designed to isolate the active sites from the probe body

5.4.4.3. 32-channel MEAs

The a-SiC MEA technology permits a wide variety of electrode designs and can be adapted to conditions where large channel counts are required. To demonstrate design flexibility, a 32- channel probe with four penetrating shanks and 8 electrode sites on each shank was designed and fabricated. To reduce the width of the shank at the base, metal traces running along the length of the shank are 2 µm wide and with 3 µm spacing. This results in eight electrode sites on each shank,

50 µm wide at the base of the shank and tapering to less than 1 µm at the distal tip. The 200 µm2 electrode sites are elliptical in shape and distributed so that at least two electrode sites lie in each

rat cortical layer except layers 1 and 2 that have one electrode site each. Figure 5.11 shows the shanks of the 32-channel a-SiC MEA developed.

Figure 5.11. 32-channel a-SiC MEA has eight electrode sites on each shank.

5.4.4.4. High density Arrays - 3D matrix arrays

Figure 5.12(a) shows a 4-shank, 16-channel cortical probe (pitch= 400 µm) developed using amorphous silicon carbide films as the substrate material. Each shank has four electrode sites

(GSA= 100 µm2) located towards the distal tip of the shank (Figure 5.12(b)). Using 400 µm thick

SU-8 spacers, a 3D array geometry was attained (Figures 5.12(c-d)). The 3D geometry results in a 64 contact sites occupying about 0.22 mm3 volume of tissue. By reducing the pitch between the shanks and the thickness of the SU-8 by to 200 µm, the number of contacts within an equivalent volume of tissue can be increased to 256.

5.4.4.5. High density Arrays - Wedge probe

It is also possible to increase the channel counts on a single shank. We have developed a single shank electrode having 64 electrode sites located diagonally across the width of the shank. This results in a distribution of 64 electrode sites in a 2 mm long rat cortical layer and 400 µm cortical column. Each electrode site is 200 µm2 (10 x 20 µm) and is distributed 3 µm horizontally and 20

µm vertically from the nearest electrode site. Figure 5.13 shows the wedge probe and the electrode site distribution.

1.31 Summary

The a-SiC platform allowed innovation in neural interface designs. Additionally, electrode sites could also be coated with commonly used low impedance coating materials, such as TiN or SIROF.

For MEAs developed with 4 µm thick a-SiC, we have described various techniques which reduce the critical buckling load of the individual shanks and enable penetration of the shanks without buckling. These methods include the addition of a temporary stiffening structure, bundling the individual shanks, or through in situ designs which reduced the effective length of the shanks while

100

allowing for targeted depth penetration. With just the addition of a minimal amount of a-SiC material to a thickness of 6 µm, individual single shanks or co-linear 2 mm long a-SiC fibers were successfully implanted into rat cortex without buckling. We have also demonstrated the ability to record neural signals using 6 µm thick a-SiC MEAs acutely in rat motor cortex. Our results have also indicated that SIROF-coated sites showed high amplitude and high SNR of the recorded neural signals. Design flexible was demonstrated by the development of MEAs with the electrode site carrying shanks pushed or pulled through tissue, development of single or multiple shanks with one or more electrode sites on each shank or by creating perforations in the MEA superstructure to allow free movement of ions around the sites.

101

CHAPTER 6

SUMMARY OF FINDINGS

As an original and unique contribution to biomedical engineering, this dissertation demonstrates the development of penetrating amorphous silicon carbide (a-SiC) brain probes with greatly reduced cross-sectional shank dimensions as compared to commercially available silicon probes.

Amorphous silicon carbide is expected to exhibit exceptional chronic stability, it possesses attractive mechanical and electrochemical properties, and is amenable to thin-film microfabrication processes. Also, the reduced shank dimensions are expected to minimize microvasculature damage during implantation and subsequently reduce foreign body response.

With the reduced shank dimensions, however, the available geometric surface area (GSA) of the active electrode sites is greatly reduced and restricted to one linear dimension (along the length of the shank). Small electrode sites generally have high impedance and low charge injection capacity.

Consequently, this dissertation also investigates low impedance electrode coatings on a-SiC MEA sites to improve the electrochemical properties of the ultramicroelectrodes. Microelectrode arrays

(MEAs) based on a-SiC were fabricated, characterized for their electrochemical properties in a saline model of the interstitial fluid, and evaluated functionally in songbird and rat brain.

For the first time, we demonstrate the fabrication of microelectrodes that uses amorphous silicon carbide as the sole substrate and encapsulation. The detailed description of the patterning and etching processes is described in Chapter 4. We fabricated and evaluated MEAs with a minimum shank cross-sectional area of 40 µm2 (4 x 10 µm) and electrode GSA of 100 µm2 (4 x 25 µm) or

50 µm2 (2 x 25 µm). A bilayer lift-off patterning process for a metal trace of 2 µm and line spacing of 3 µm was also developed. Beyond the design and fabrication of very thin shank geometries, we

102

demonstrated that the a-SiC platform permits a wide variety of MEA designs and geometries.

Chapter 5 describes various novel design configurations enabled by the a-SiC.

The fabrication of the ultrathin a-SiC MEAs was benchmarked against the ability of the shanks to penetrate neural tissue. Insertion mechanics of the a-SiC MEAs was introduced in Chapter 4 and detailed in Chapter 5. The 4 µm thick a-SiC MEAs successfully penetrated rat and songbird’s brain when the shanks were temporarily stabilized with polyethylene glycol (MW 2000) or when the shanks were bundled together. Insertion of a single shank or multiple co-linear shanks into rat brain was achieved with a minimum a-SiC thickness of 6 µm. We demonstrated the insertion of shanks with 42 µm2 (6 x 7 µm) cross-sectional area into rat cortex. With this cross-sectional area, an insertion force of 0.35 mN was measured, and the indwelling shank experienced an average tissue relaxation force of 2.2 µN/s.

A critical challenge associated with the a-SiC MEA technology is stress control in the multilayer structure. We engineered tensile stress in the metallization to counterbalance the compressive stress in the a-SiC films in the multilayer MEA structure. Stress engineering in sputtered and evaporated metallization is described in Chapter 3. Chapter 3 also addresses how metal patterning influences the integrity and stability of a-SiC films under accelerated aging conditions. Fabricated with evaporated metallization, the a-SiC MEA remained intact after aging when soaked in PBS at

87oC for at least 20 weeks.

Finally, we examined electrode coatings on the ultramicroelectrode sites as a means of reducing electrode impedance for recording and increasing charge injection capacity for neural stimulation.

Electrode coatings such as sputtered iridium oxide films (SIROF) and titanium nitride (TiN) were investigated. These coatings reduced the electrode impedance from 2.5 MΩ (Au sites) to values

103

below 100 kΩ in PBS. The electrodes with reduced impedance were shown to record extracellular neural activity acutely in Chapter 4 and Chapter 5. Charge injection properties were also significantly improved. The maximum charge that was delivered in a 200 µs pulse was 23 nC/ph for SIROF and 9 nC/ph for TiN UMEs.

The electrochemical properties of electrodeposited iridium oxide films (EIROF) on carbon fiber ultramicroelectrodes were extensively reportedin Chapter 2. We showed that the electrode impedance is reduced by at least an order of magnitude, and a maximum charge injection capacity of 17 mC/cm2 was achieved with appropriate biasing in physiological saline. The coated electrode delivers much higher charge than a bare carbon fiber electrode before reaching a -0.6 V water reduction limit. We demonstrated that charge delivery is predominantly due to Ir4+/Ir3+ redox reactions and not due to faradaic reactions external to the film. EIROF films become resistive as the Ir3+/Ir4+ ratio increases during cathodal pulsing, which later drastically increases the driving voltage during the charge delivery due to an increased ohmic overpotential.

104

APPENDIX A

SUPPLEMENTARY MATERIAL FOR CHAPTER 4

Back-end processing

Electrical connection to the a-SiC MEAs is obtained with an Omnetics connector mounted on the array using a solder reflow process and indium-based solder paste (Indium Corporation). A solder reflow process is employed and the step-by-step guide in preparing the array prior to the reflow process is summarized in Figure A.1.

Figure A.1. Process flow for the application of Indium –Tin solder paste prior to the solder reflow process.

After the reflow process, it is important to inspect the back side of the device for possible delamination. Figure A.2 shows whose metallization remained intact after the reflow process (a) and one that exhibit metal delamination at the connector side post- reflow (b). Delamination is

105

eliminated when a dense titanium adhesion layer is deposited for instance under high power- low pressure sputtering conditions (e.g. 200 W and <4 mTorr) or at a very slow deposition rate during metal evaporation (e.g. 0.5 Å/s).

Figure A.2. Possible device failure due to solder reflow processing. (a) Shows no delamination of metal pads and (b) shows delamination after the solder reflow process.

To facilitate easy release of the arrays after the soldering process, a PDMS-coated glass slide is used as the substrate on which the device assembly occurs. A medical grade epoxy (Loctite) or dental acrylic (Flow-It ALC, Pentron) is then cured around the base of the Omnetics connector to provide mechanical strength.

106

REFERENCES

[1] M. J. Vansteensel et al., “Fully Implanted Brain-Computer Interface in a Locked-In Patient with ALS.,” N. Engl. J. Med., vol. 375, no. 21, pp. 2060–2066, 2016.

[2] M. L. Homer, A. V Nurmikko, J. P. Donoghue, and L. R. Hochberg, “Sensors and Decoding for Intracortical Brain Computer Interfaces,” Annu. Rev. Biomed. Eng., vol. 15, no. 1, pp. 383–405, Jul. 2013.

[3] T. Aflalo et al., “Neurophysiology. Decoding motor imagery from the posterior parietal cortex of a tetraplegic human.,” Science, vol. 348, no. 6237, pp. 906–10, May 2015.

[4] B. Jarosiewicz, A. A. Sarma, J. Saab, B. Franco, S. S. Cash, E. N. Eskandar, and L. R. Hochberg, “Retrospectively supervised click decoder calibration for self-calibrating point- and-click brain-computer interfaces.,” J. Physiol. Paris, vol. 110, no. 4 Pt A, pp. 382–391, Nov. 2016.

[5] J. C. Kao, P. Nuyujukian, S. I. Ryu, M. M. Churchland, J. P. Cunningham, and K. V. Shenoy, “Single-trial dynamics of motor cortex and their applications to brain-machine interfaces,” Nat. Commun., vol. 6, no. 1, p. 7759, Dec. 2015.

[6] A. B. Ajiboye et al., “Restoration of reaching and grasping movements through brain- controlled muscle stimulation in a person with tetraplegia: a proof-of-concept demonstration,” Lancet, vol. 389, no. 10081, pp. 1821–1830, May 2017.

[7] S. N. Flesher, J. L. Collinger, S. T. Foldes, J. M. Weiss, J. E. Downey, E. C. Tyler-Kabara, S. J. Bensmaia, A. B. Schwartz, M. L. Boninger, and R. A. Gaunt, “Intracortical microstimulation of human somatosensory cortex,” Sci. Transl. Med., vol. 8, no. 361, p. 361ra141-361ra141, Oct. 2016.

[8] W. A. Liberti, J. E. Markowitz, L. N. Perkins, D. C. Liberti, D. P. Leman, G. Guitchounts, T. Velho, D. N. Kotton, C. Lois, and T. J. Gardner, “Unstable neurons underlie a stable learned behavior,” Nat. Neurosci., vol. 19, no. 12, pp. 1665–1671, Dec. 2016.

[9] V. R. Athalye, K. Ganguly, R. M. Costa, and J. M. Carmena, “Emergence of Coordinated Neural Dynamics Underlies Neuroprosthetic Learning and Skillful Control.,” Neuron, vol. 93, no. 4, p. 955–970.e5, Feb. 2017.

107

[10] S. R. Kane, S. F. Cogan, J. Ehrlich, T. D. Plante, D. B. McCreery, and P. R. Troyk, “Electrical performance of penetrating microelectrodes chronically implanted in cat cortex,” IEEE Trans. Biomed. Eng., vol. 60, no. 8, pp. 2153–2160, 2013.

[11] B. G. Cragg, “The density of synapses and neurones in the motor and visual areas of the cerebral cortex.,” J. Anat., vol. 101, no. Pt 4, pp. 639–54, Sep. 1967.

[12] S. F. Lempka, M. D. Johnson, M. A. Moffitt, K. J. Otto, D. R. Kipke, and C. C. McIntyre, “Theoretical analysis of intracortical microelectrode recordings,” J. Neural Eng., vol. 8, no. 4, p. 045006, Aug. 2011.

[13] T. D. Y. Kozai, T. C. Marzullo, F. Hooi, N. B. Langhals, A. K. Majewska, E. B. Brown, and D. R. Kipke, “Reduction of neurovascular damage resulting from microelectrode insertion into the cerebral cortex using in vivo two-photon mapping.,” J. Neural Eng., vol. 7, no. 4, p. 046011, Aug. 2010.

[14] T. Saxena, L. Karumbaiah, E. A. Gaupp, R. Patkar, K. Patil, M. Betancur, G. B. Stanley, and R. V. Bellamkonda, “The impact of chronic blood–brain barrier breach on intracortical electrode function,” Biomaterials, vol. 34, no. 20, pp. 4703–4713, Jul. 2013.

[15] T. D. Y. Kozai, X. Li, L. M. Bodily, E. M. Caparosa, G. A. Zenonos, D. L. Carlisle, R. M. Friedlander, and X. T. Cui, “Effects of caspase-1 knockout on chronic neural recording quality and longevity: Insight into cellular and molecular mechanisms of the reactive tissue response,” Biomaterials, vol. 35, no. 36, pp. 9620–9634, Dec. 2014.

[16] R. Biran, D. C. Martin, and P. A. Tresco, “Neuronal cell loss accompanies the brain tissue response to chronically implanted silicon microelectrode arrays,” Exp. Neurol., vol. 195, no. 1, pp. 115–126, 2005.

[17] J. A. Smith, A. Das, S. K. Ray, and N. L. Banik, “Role of pro-inflammatory cytokines released from microglia in neurodegenerative diseases,” Brain Res. Bull., vol. 87, no. 1, pp. 10–20, Jan. 2012.

[18] A. J. Sawyer, W. Tian, J. K. Saucier-Sawyer, P. J. Rizk, W. M. Saltzman, R. V. Bellamkonda, and T. R. Kyriakides, “The effect of inflammatory cell-derived MCP-1 loss on neuronal survival during chronic neuroinflammation,” Biomaterials, vol. 35, no. 25, pp. 6698–6706, Aug. 2014.

108

[19] D. J. Edell, V. Van Toi, V. M. McNeil, and L. D. Clark, “Factors Influencing the Biocompatibility of Insertable Silicon Microshafts in Cerebral Cortex,” IEEE Trans. Biomed. Eng., vol. 39, no. 6, pp. 635–643, 1992.

[20] J. N. Turner, W. Shain, D. H. Szarowski, M. Andersen, S. Martins, M. Isaacson, and H. Craighead, “Cerebral astrocyte response to micromachined silicon implants.,” Exp. Neurol., vol. 156, no. 1, pp. 33–49, 1999.

[21] V. S. Polikov, P. A. Tresco, and W. M. Reichert, “Response of brain tissue to chronically implanted neural electrodes,” J. Neurosci. Methods, vol. 148, no. 1, pp. 1–18, 2005.

[22] A. Ersen, S. Elkabes, D. S. Freedman, and M. Sahin, “Chronic tissue response to untethered microelectrode implants in the rat brain and spinal cord,” J. Neural Eng., vol. 12, no. 1, p. 016019, Feb. 2015.

[23] G. C. McConnell, H. D. Rees, A. I. Levey, C.-A. Gutekunst, R. E. Gross, and R. V Bellamkonda, “Implanted neural electrodes cause chronic, local inflammation that is correlated with local neurodegeneration,” J. Neural Eng., vol. 6, no. 5, p. 056003, 2009.

[24] D. H. Szarowski, M. D. Andersen, S. Retterer, A. J. Spence, M. Isaacson, H. G. Craighead, J. N. Turner, and W. Shain, “Brain responses to micro-machined silicon devices,” Brain Res., vol. 983, no. 1–2, pp. 23–35, 2003.

[25] B. D. Winslow and P. A. Tresco, “Quantitative analysis of the tissue response to chronically implanted microwire electrodes in rat cortex,” Biomaterials, vol. 31, pp. 1558–1567, 2009.

[26] J. Subbaroyan, D. C. Martin, and D. R. Kipke, “A finite-element model of the mechanical effects of implantable microelectrodes in the cerebral cortex,” J Neural Eng, vol. 2, no. 4, pp. 103–113, 2005.

[27] A. Sridharan, J. K. Nguyen, J. R. Capadona, and J. Muthuswamy, “Compliant intracortical implants reduce strains and strain rates in brain tissue in vivo,” J. Neural Eng., vol. 12, no. 3, p. 036002, 2015.

[28] J. K. Nguyen, D. J. Park, J. L. Skousen, A. E. Hess-Dunning, D. J. Tyler, S. J. Rowan, C. Weder, and J. R. Capadona, “Mechanically-compliant intracortical implants reduce the neuroinflammatory response,” J. Neural Eng., vol. 11, no. 5, p. 056014, Oct. 2014.

109

[29] Liang Guo and S. P. DeWeerth, “PDMS-based conformable microelectrode arrays with selectable novel 3-D microelectrode geometries for surface stimulation and recording,” in 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2009, vol. 2009, pp. 1623–1626.

[30] A. Altuna et al., “SU-8 based microprobes for simultaneous neural depth recording and drug delivery in the brain,” Lab Chip, vol. 13, no. 7, p. 1422, 2013.

[31] B. J. Kim, J. T. W. Kuo, S. A. Hara, C. D. Lee, L. Yu, C. A. Gutierrez, T. Q. Hoang, V. Pikov, and E. Meng, “3D Parylene sheath neural probe for chronic recordings,” J. Neural Eng., vol. 10, no. 4, p. 045002, Aug. 2013.

[32] H. S. Sohal, G. J. Clowry, A. Jackson, A. O’Neill, and S. N. Baker, “Mechanical Flexibility Reduces the Foreign Body Response to Long-Term Implanted Microelectrodes in Rabbit Cortex,” PLoS One, vol. 11, no. 10, p. e0165606, Oct. 2016.

[33] C. D. Lee and E. Meng, “Mechanical Properties of Thin-Film Parylene–Metal–Parylene Devices,” Front. Mech. Eng., vol. 1, no. 1, Sep. 2015.

[34] S. Takeuchi, D. Ziegler, Y. Yoshida, K. Mabuchi, and T. Suzuki, “Parylene flexible neural probes integrated with microfluidic channels,” Lab Chip, vol. 5, no. 5, p. 519, Apr. 2005.

[35] J. T. W. Kuo, B. J. Kim, S. A. Hara, C. D. Lee, C. A. Gutierrez, T. Q. Hoang, and E. Meng, “Novel flexible Parylene neural probe with 3D sheath structure for enhancing tissue integration,” Lab Chip, vol. 13, no. 4, pp. 554–561, Feb. 2013.

[36] K. C. Cheung, “Implantable microscale neural interfaces,” Biomed. Microdevices, vol. 9, no. 6, pp. 923–938, Oct. 2007.

[37] A. Mercanzini, K. Cheung, D. L. Buhl, M. Boers, A. Maillard, P. Colin, J.-C. Bensadoun, A. Bertsch, and P. Renaud, “Demonstration of cortical recording using novel flexible polymer neural probes,” Sensors Actuators A Phys., vol. 143, no. 1, pp. 90–96, May 2008.

[38] Y.-Y. Chen, H.-Y. Lai, S.-H. Lin, C.-W. Cho, W.-H. Chao, C.-H. Liao, S. Tsang, Y.-F. Chen, and S.-Y. Lin, “Design and fabrication of a polyimide-based microelectrode array: Application in neural recording and repeatable electrolytic lesion in rat brain,” J. Neurosci. Methods, vol. 182, no. 1, pp. 6–16, Aug. 2009.

110

[39] W. Poppendieck, A. Sossalla, M.-O. Krob, C. Welsch, T. A. K. Nguyen, W. Gong, J. DiGiovanna, S. Micera, D. M. Merfeld, and K.-P. Hoffmann, “Development, manufacturing and application of double-sided flexible implantable microelectrodes,” Biomed. Microdevices, vol. 16, no. 6, pp. 837–850, Dec. 2014.

[40] S. Metz, A. Bertsch, D. Bertrand, and P. Renaud, “Flexible polyimide probes with microelectrodes and embedded microfluidic channels for simultaneous drug delivery and multi-channel monitoring of bioelectric activity,” Biosens. Bioelectron., vol. 19, no. 10, pp. 1309–1318, May 2004.

[41] G. Márton, G. Orbán, M. Kiss, R. Fiáth, A. Pongrácz, and I. Ulbert, “A Multimodal, SU-8 - Platinum - Polyimide Microelectrode Array for Chronic In Vivo Neurophysiology,” PLoS One, vol. 10, no. 12, p. e0145307, Dec. 2015.

[42] J. P. Harris, J. R. Capadona, R. H. Miller, B. C. Healy, K. Shanmuganathan, S. J. Rowan, C. Weder, and D. J. Tyler, “Mechanically adaptive intracortical implants improve the proximity of neuronal cell bodies.,” J. Neural Eng., vol. 8, no. 6, p. 066011, Dec. 2011.

[43] J. P. Harris, A. E. Hess, S. J. Rowan, C. Weder, C. A. Zorman, D. J. Tyler, and J. R. Capadona, “In vivo deployment of mechanically adaptive nanocomposites for intracortical microelectrodes.,” J. Neural Eng., vol. 8, no. 4, p. 046010, Aug. 2011.

[44] T. Ware, D. Simon, C. Liu, T. Musa, S. Vasudevan, A. Sloan, E. W. Keefer, R. L. Rennaker, and W. Voit, “Thiol-ene/acrylate substrates for softening intracortical electrodes,” J. Biomed. Mater. Res. Part B Appl. Biomater., vol. 102, no. 1, pp. 1–11, Jan. 2014.

[45] J. J. Pancrazio, F. Deku, A. Ghazavi, A. M. Stiller, R. Rihani, C. L. Frewin, V. D. Varner, T. J. Gardner, and S. F. Cogan, “Thinking Small: Progress on Microscale Neurostimulation Technology,” Neuromodulation Technol. Neural Interface, vol. 20, no. 8, pp. 745–752, Dec. 2017.

[46] K. C. Spencer, J. C. Sy, K. B. Ramadi, A. M. Graybiel, R. Langer, and M. J. Cima, “Characterization of Mechanically Matched Hydrogel Coatings to Improve the Biocompatibility of Neural Implants.,” Sci. Rep., vol. 7, no. 1, p. 1952, May 2017.

[47] J. P. Seymour and D. R. Kipke, “Neural probe design for reduced tissue encapsulation in CNS,” Biomaterials, vol. 28, no. 25, pp. 3594–3607, Sep. 2007.

111

[48] P. Stice, A. Gilletti, A. Panitch, and J. Muthuswamy, “Thin microelectrodes reduce GFAP expression in the implant site in rodent somatosensory cortex.,” J. Neural Eng., vol. 4, no. 2, pp. 42–53, Jun. 2007.

[49] M. P. Ward, P. Rajdev, C. Ellison, and P. P. Irazoqui, “Toward a comparison of microelectrodes for acute and chronic recordings,” Brain Res., vol. 1282, pp. 183–200, Jul. 2009.

[50] F. Deku, Y. Cohen, A. Joshi-Imre, A. Kanneganti, T. J. Gardner, and S. F. Cogan, “Amorphous silicon carbide ultramicroelectrode arrays for neural stimulation and recording,” J. Neural Eng., vol. 15, no. 1, p. 016007, Feb. 2018.

[51] T. D. Yoshida Kozai, N. B. Langhals, P. R. Patel, X. Deng, H. Zhang, K. L. Smith, J. Lahann, N. A. Kotov, and D. R. Kipke, “Ultrasmall implantable composite microelectrodes with bioactive surfaces for chronic neural interfaces,” Nat. Mater., vol. 11, no. 12, pp. 1065– 1073, Nov. 2012.

[52] J. C. Barrese, N. Rao, K. Paroo, C. Triebwasser, C. Vargas-Irwin, L. Franquemont, and J. P. Donoghue, “Failure mode analysis of silicon-based intracortical microelectrode arrays in non-human primates,” J. Neural Eng., vol. 10, no. 6, p. 066014, Dec. 2013.

[53] B. A. Wester, R. H. Lee, and M. C. LaPlaca, “Development and characterization of in vivo flexible electrodes compatible with large tissue displacements,” J. Neural Eng., vol. 6, no. 2, p. 024002, Apr. 2009.

[54] H. S. Sohal, A. Jackson, R. Jackson, G. J. Clowry, K. Vassilevski, A. O’Neill, and S. N. Baker, “The sinusoidal probe: a new approach to improve electrode longevity.,” Front. Neuroeng., vol. 7, p. 10, 2014.

[55] P. J. Rousche, D. S. Pellinen, D. P. Pivin, J. C. Williams, R. J. Vetter, and D. R. Kipke, “Flexible polyimide-based intracortical electrode arrays with bioactive capability.,” IEEE Trans. Biomed. Eng., vol. 48, no. 3, pp. 361–71, Mar. 2001.

[56] W. Jensen, K. Yoshida, and U. G. Hofmann, “In-vivo implant mechanics of flexible, silicon- based ACREO microelectrode arrays in rat cerebral cortex,” IEEE Trans. Biomed. Eng., vol. 53, no. 5, pp. 934–940, May 2006.

112

[57] C. A. Diaz-Botia, L. E. Luna, R. M. Neely, M. Chamanzar, C. Carraro, J. M. Carmena, P. N. Sabes, R. Maboudian, and M. M. Maharbiz, “A silicon carbide array for electrocorticography and peripheral nerve recording,” J. Neural Eng., vol. 14, no. 5, p. 056006, Oct. 2017.

[58] S. F. Cogan, D. J. Edell, A. a Guzelian, Y. Ping Liu, and R. Edell, “Plasma-enhanced chemical vapor deposited silicon carbide as an implantable dielectric coating.,” J. Biomed. Mater. Res. A, vol. 67, no. 3, pp. 856–867, 2003.

[59] J.-M. Hsu, P. Tathireddy, L. Rieth, A. R. Normann, and F. Solzbacher, “Characterization of a-SiCx:H thin films as an encapsulation material for integrated silicon based neural interface devices,” Thin Solid Films, vol. 516, no. 1, pp. 34–41, Nov. 2007.

[60] G. L. Knaack, D. G. McHail, G. Borda, B. S. Koo, N. Peixoto, S. F. Cogan, T. C. Dumas, and J. J. Pancrazio, “In vivo Characterization of Amorphous Silicon Carbide as a Biomaterial for Chronic Neural Interfaces,” Front. Neurosci., vol. 10, no. June, p. 301, 2016.

[61] X. Lei, S. Kane, S. Cogan, H. Lorach, L. Galambos, P. Huie, K. Mathieson, T. Kamins, J. Harris, and D. Palanker, “SiC protective coating for photovoltaic retinal prosthesis,” J. Neural Eng., vol. 13, no. 4, p. 046016, 2016.

[62] R. G. Azevedo, Jingchun Zhang, D. G. Jones, D. R. Myers, A. V Jog, B. Jamshidi, M. B. J. Wijesundara, R. Maboudian, and A. P. Pisano, “Silicon carbide coated MEMS strain sensor for harsh environment applications,” in 2007 IEEE 20th International Conference on Micro Electro Mechanical Systems (MEMS), 2007, no. January, pp. 643–646.

[63] M. Mehregany and C. A. Zorman, “SiC MEMS: opportunities and challenges for applications in harsh environments,” Thin Solid Films, vol. 355–356, pp. 518–524, Nov. 1999.

[64] C. A. Zorman, “Silicon Carbide as a Material for Biomedical Microsystems,” EDA Publ., pp. 1–7, 2009.

[65] J.-M. Hsu, P. Tathireddy, L. Rieth, A. R. Normann, and F. Solzbacher, “Characterization of a-SiCx:H thin films as an encapsulation material for integrated silicon based neural interface devices,” Thin Solid Films, vol. 516, no. 1, pp. 34–41, Nov. 2007.

113

[66] U. Kalnins, A. Erglis, I. Dinne, I. Kumsars, and S. Jegere, “Clinical outcomes of silicon carbide coated stents in patients with coronary artery disease.,” Med. Sci. Monit., vol. 8, no. 2, pp. PI16-20, Feb. 2002.

[67] C. W. Hamm and P. G. Hugenholtz, “Silicon carbide-coated stents in patients with acute coronary syndrome,” Catheter. Cardiovasc. Interv., vol. 60, no. 3, pp. 375–381, Nov. 2003.

[68] P. Barzó, A. Marmarou, P. Fatouros, K. Hayasaki, and F. Corwin, “Contribution of vasogenic and cellular edema to traumatic brain swelling measured by diffusion-weighted imaging,” J. Neurosurg., vol. 87, no. 6, pp. 900–907, Dec. 1997.

[69] C. S. Bjornsson, S. J. Oh, Y. a Al-Kofahi, Y. J. Lim, K. L. Smith, J. N. Turner, S. De, B. Roysam, W. Shain, and S. J. Kim, “Effects of insertion conditions on tissue strain and vascular damage during neuroprosthetic device insertion.,” J. Neural Eng., vol. 3, no. 3, pp. 196–207, 2006.

[70] T. D. Y. Kozai, A. L. Vazquez, C. L. Weaver, S. Kim, and X. T. Cui, “In vivo two-photon microscopy reveals immediate microglial reaction to implantation of microelectrode through extension of processes,” J. Neural Eng., vol. 9, no. 6, p. 066001, Dec. 2012.

[71] A. W. Unterberg, J. Stover, B. Kress, and K. L. Kiening, “Edema and brain trauma,” Neuroscience, vol. 129, no. 4, pp. 1019–1027, Jan. 2004.

[72] G. Guitchounts, J. E. Markowitz, W. A. Liberti, and T. J. Gardner, “A carbon-fiber electrode array for long-term neural recording,” J. Neural Eng., vol. 10, no. 4, p. 046016, Aug. 2013.

[73] P. R. Patel, K. Na, H. Zhang, T. D. Y. Y. Kozai, N. A. Kotov, E. Yoon, and C. A. Chestek, “Insertion of linear 8.4 μ m diameter 16 channel carbon fiber electrode arrays for single unit recordings,” J. Neural Eng., vol. 12, no. 4, p. 046009, Aug. 2015.

[74] K. A. Ludwig, N. B. Langhals, M. D. Joseph, S. M. Richardson-Burns, J. L. Hendricks, and D. R. Kipke, “Poly(3,4-ethylenedioxythiophene) (PEDOT) polymer coatings facilitate smaller neural recording electrodes,” J. Neural Eng., vol. 8, no. 1, p. 014001, Feb. 2011.

[75] S. F. Cogan, “Neural stimulation and recording electrodes.,” Annu. Rev. Biomed. Eng., vol. 10, pp. 275–309, 2008.

114

[76] R. D. Meyer, S. F. Cogan, T. H. Nguyen, and R. D. Rauh, “Electrodeposited iridium oxide for neural stimulation and recording electrodes,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 9, no. 1, pp. 2–11, 2001.

[77] X. Liu, D. B. McCreery, L. A. Bullara, and W. F. Agnew, “Evaluation of the stability of intracortical microelectrode arrays,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 14, no. 1, pp. 91–100, 2006.

[78] M. Han, P. S. Manoonkitiwongsa, C. X. Wang, and D. B. McCreery, “In vivo validation of custom-designed silicon-based microelectrode arrays for long-term neural recording and stimulation,” IEEE Trans. Biomed. Eng., vol. 59, no. 2, pp. 346–354, 2012.

[79] S. Ching, R. Dudek, and E. Tabet, “Cyclic Voltammetry with Ultramicroelectrodes,” J. Chem. Educ., vol. 71, no. 7, p. 602, 1994.

[80] J. J. Heinze, “Ultramicroelectrodes in Electrochemistry,” Angew. Chemie Int. Ed. English, vol. 32, no. 9, pp. 1268–1288, Sep. 1993.

[81] C. G. Zoski, “Ultramicroelectrodes: Design, fabrication, and characterization,” Electroanalysis, vol. 14, no. 15–16, pp. 1041–1051, 2002.

[82] K. Aoki, “Theory of Ultramicroelectrodes,” Electroanalysis, vol. 5, pp. 627–639, 1993.

[83] S. F. Cogan, P. R. Troyk, J. Ehrlich, and T. D. Plante, “In Vitro Comparison of the Charge- Injection Limits of Activated Iridium Oxide (AIROF) and Platinum-Iridium Microelectrodes,” IEEE Trans. Biomed. Eng., vol. 52, no. 9, pp. 1612–1614, Sep. 2005.

[84] D. B. McCreery, “Cochlear nucleus auditory prostheses,” Hear. Res., vol. 242, no. 1–2, pp. 64–73, 2008.

[85] W. F. Gillis et al., “Carbon fiber on polyimide ultra-microelectrodes,” J. Neural Eng., vol. 15, no. 1, p. 016010, Feb. 2018.

[86] N. F. Nolta, M. B. Christensen, P. D. Crane, J. L. Skousen, and P. A. Tresco, “BBB leakage, astrogliosis, and tissue loss correlate with silicon microelectrode array recording performance,” Biomaterials, vol. 53, pp. 753–762, 2015.

115

[87] M. Jorfi, J. L. Skousen, C. Weder, and J. R. Capadona, “Progress towards biocompatible intracortical microelectrodes for neural interfacing applications,” J. Neural Eng., vol. 12, no. 1, p. 011001, Feb. 2015.

[88] R. Biran, D. C. Martin, and P. A. Tresco, “The brain tissue response to implanted silicon microelectrode arrays is increased when the device is tethered to the skull,” J. Biomed. Mater. Res. Part A, vol. 82A, no. 1, pp. 169–178, Jul. 2007.

[89] J. L. Skousen, M. J. Bridge, and P. A. Tresco, “A strategy to passively reduce neuroinflammation surrounding devices implanted chronically in brain tissue by manipulating device surface permeability,” Biomaterials, vol. 36, pp. 33–43, 2015.

[90] J. E. Sanders, C. E. Stiles, and C. L. Hayes, “Tissue response to single-polymer fibers of varying diameters: Evaluation of fibrous encapsulation and macrophage density,” J. Biomed. Mater. Res., vol. 52, no. 1, pp. 231–237, 2000.

[91] J. P. Seymour and D. R. Kipke, “Fabrication of Polymer Neural Probes with Sub-cellular Features for Reduced Tissue Encapsulation,” in 2006 International Conference of the IEEE Engineering in Medicine and Biology Society, 2006, no. 1, pp. 4606–4609.

[92] C. Bennett, M. Samikkannu, F. Mohammed, W. D. Dietrich, S. M. Rajguru, and A. Prasad, “Blood brain barrier (BBB)-disruption in intracortical silicon microelectrode implants,” Biomaterials, vol. 164, pp. 1–10, May 2018.

[93] D. a Walsh, K. R. J. Lovelock, and P. Licence, “Ultramicroelectrode voltammetry and scanning electrochemical microscopy in room-temperature ionic liquid electrolytes.,” Chem. Soc. Rev., vol. 39, no. 11, pp. 4185–4194, 2010.

[94] S. F. Cogan, K. A. Ludwig, C. G. Welle, and P. Takmakov, “Tissue damage thresholds during therapeutic electrical stimulation,” J. Neural Eng., vol. 13, no. 2, p. 021001, 2016.

[95] T. Nyberg, A. Shimada, and K. Torimitsu, “Ion conducting polymer microelectrodes for interfacing with neural networks,” J. Neurosci. Methods, vol. 160, no. 1, pp. 16–25, 2007.

[96] S. J. Wilks, S. M. Richardson-Burns, J. L. Hendricks, D. C. Martin, and K. J. Otto, “Poly(3,4- ethylenedioxythiophene) as a Micro-Neural Interface Material for Electrostimulation.,” Front. Neuroeng., vol. 2, no. June, p. 7, 2009.

116

[97] N. Peixoto, K. Jackson, R. Samiyi, and S. Minnikanti, “Charge storage: Stability measures in implantable electrodes,” Proc. 31st Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. Eng. Futur. Biomed. EMBC 2009, pp. 658–661, 2009.

[98] S. F. Cogan, J. Ehrlich, T. D. Plante, A. Smirnov, D. B. Shire, M. Gingerich, and J. F. Rizzo, “Sputtered iridium oxide films for neural stimulation electrodes,” J. Biomed. Mater. Res. - Part B Appl. Biomater., vol. 89, no. 2, pp. 353–361, 2009.

[99] E. Slavcheva, R. Vitushinsky, W. Mokwa, and U. Schnakenberg, “Sputtered Iridium Oxide Films as Charge Injection Material for Functional Electrostimulation,” J. Electrochem. Soc., vol. 151, no. 7, p. E226, 2004.

[100] B. Wessling, W. Mokwa, and U. Schnakenberg, “RF-sputtering of iridium oxide to be used as stimulation material in functional medical implants,” J. Micromechanics Microengineering, vol. 16, no. 6, pp. S142–S148, 2006.

[101] D. O. Wipf, F. Ge, T. W. Spaine, and J. E. Baur, “Microscopic Measurement of pH with Iridium Oxide Microelectrodes,” Anal. Chem., vol. 72, no. 20, pp. 4921–4927, Oct. 2000.

[102] S. F. Cogan, P. R. Troyk, J. Ehrlich, C. M. Gasbarro, and T. D. Plante, “The influence of electrolyte composition on the in vitro charge-injection limits of activated iridium oxide (AIROF) stimulation electrodes,” J. Neural Eng., vol. 4, no. 2, pp. 79–86, Jun. 2007.

[103] S. B. Baumann, D. R. Wozny, S. K. Kelly, and F. M. Meno, “The electrical conductivity of human cerebrospinal fluid at body temperature,” IEEE Trans. Biomed. Eng., vol. 44, no. 3, pp. 220–223, Mar. 1997.

[104] P. R. Troyk, D. E. Detlefsen, S. F. Cogan, J. Ehrlich, M. Bak, D. B. McCreery, L. Bullara, and E. Schmidt, “‘Safe’ charge-injection waveforms for iridium oxide (AIROF) microelectrodes,” in The 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2004, vol. 4, pp. 4141–4144.

[105] S. F. Cogan, J. Ehrlich, T. D. Plante, M. D. Gingerich, and D. B. Shire, “Contribution of oxygen reduction to charge injection on platinum and sputtered iridium oxide neural stimulation electrodes,” IEEE Trans. Biomed. Eng., vol. 57, no. 9, pp. 2313–2321, 2010.

117

[106] X. Beebe and T. L. Rose, “Charge injection limits of activated iridium oxide electrodes with 0.2 ms pulses in bicarbonate buffered saline (neurological stimulation application),” IEEE Trans. Biomed. Eng., vol. 35, no. 6, pp. 494–495, Jun. 1988.

[107] D. R. Merrill, M. Bikson, and J. G. R. Jefferys, “Electrical stimulation of excitable tissue: Design of efficacious and safe protocols,” J. Neurosci. Methods, vol. 141, no. 2, pp. 171– 198, 2005.

[108] K.-F. Chow, T. M. Carducci, and R. W. Murray, “Electronic Conductivity of Films of Electroflocculated 2 nm Iridium Oxide Nanoparticles,” J. Am. Chem. Soc., vol. 136, no. 9, pp. 3385–3387, Mar. 2014.

[109] C. L. Ballestrasse, R. T. Ruggeri, and T. R. Beck, “Calculations of the pH changes produced in body tissue by a spherical stimulation electrode,” Ann. Biomed. Eng., vol. 13, pp. 405– 424, 1985.

[110] J. D. Weiland, D. J. Anderson, and M. S. Humayun, “In vitro electrical properties for iridium oxide versus titanium nitride stimulating electrodes,” IEEE Trans. Biomed. Eng., vol. 49, no. 12, pp. 1574–1579, Dec. 2002.

[111] M. Avram, A. Avram, A. Bragaru, Bangtao Chen, D. P. Poenar, and C. Iliescu, “Low stress PECVD amorphous silicon carbide for MEMS applications,” in CAS 2010 Proceedings (International Semiconductor Conference), 2010, vol. 01, pp. 239–242.

[112] C. Iliescu, B. Chen, D. P. Poenar, and Y. Y. Lee, “PECVD amorphous silicon carbide membranes for cell culturing,” Sensors Actuators, B Chem., vol. 129, no. 1, pp. 404–411, 2008.

[113] C. Iliescu, M. Avram, B. Chen, A. Popescu, V. Dumitrescu, D. P. Poenar, A. Sterian, D. Vrtacnik, S. Amon, and P. Sterian, “Residual stress in thin films PECVD depositions: A review,” J. Optoelectron. Adv. Mater., vol. 13, no. 4, pp. 387–394, 2011.

[114] P. M. Sarro, C. R. Deboer, E. Korkmaz, and J. M. W. Laros, “Low-stress PECVD SiC thin films for IC-compatible microstructures,” Sensors Actuators, A Phys., vol. 67, no. 1–3 pt 1, pp. 175–180, 1998.

118

[115] R. W. Hoffman, R. D. Daniels, and E. C. Crittenden, “The Cause of Stress in Evaporated Metal Films,” Proc. Phys. Soc. Sect. B, vol. 67, no. 6, pp. 497–500, Jun. 1954.

[116] H. P. Murbach and H. Wilman, “The Origin of Stress in Metal Layers Condensed from the Vapour in High Vacuum,” Proc. Phys. Soc. Sect. B, vol. 66, no. 11, pp. 905–910, Nov. 1953.

[117] G. C. A. M. Janssen and J.-D. Kamminga, “Stress in hard metal films,” Appl. Phys. Lett., vol. 85, no. 15, pp. 3086–3088, Oct. 2004.

[118] F. M. D’Heurle, “Aluminum films deposited by rf sputtering,” Metall. Mater. Trans. B, vol. 1, no. 3, pp. 725–732, 1970.

[119] H. Windischmann, “Intrinsic stress in sputter-deposited thin films,” Crit. Rev. Solid State Mater. Sci., vol. 17, no. 6, pp. 547–596, Jan. 1992.

[120] A. G. Blachman, “Stress and resistivity control in sputtered molybdenum films and comparison with sputtered gold,” Metall. Trans., vol. 2, no. 3, pp. 699–709, Mar. 1971.

[121] D. W. Hoffman and J. A. Thornton, “Internal stresses in Cr, Mo, Ta, and Pt films deposited by sputtering from a planar magnetron source,” J. Vac. Sci. Technol. , vol. 205790731, no. 10, 1982.

[122] P. Ziemann and E. Kay, “Correlation between the ion bombardment during film growth of Pd films and their structural and electrical properties,” J. Vac. Sci. Technol. A Vacuum, Surfaces, Film., vol. 1, no. 2, pp. 512–516, Apr. 1983.

[123] Y. H. Cheng, B. K. Tay, S. P. Lau, X. Shi, and H. S. Tan, “Influence of substrate bias on the microstructure and internal stress in Cu films deposited by filtered cathodic vacuum arc,” J. Vac. Sci. Technol. A Vacuum, Surfaces, Film., vol. 19, no. 5, pp. 2102–2108, Sep. 2001.

[124] D. W. Hoffman, “Stress and property control in sputtered metal films without substrate bias,” Thin Solid Films, vol. 107, no. 4, pp. 353–358, Sep. 1983.

[125] D. W. L. Hukins, A. Mahomed, and S. N. Kukureka, “Accelerated aging for testing polymeric biomaterials and medical devices,” Med. Eng. Phys., vol. 30, no. 10, pp. 1270– 1274, Dec. 2008.

119

[126] M. Adamov, B. Perović, and T. Nenadović, “Electrical and structural properties of thin gold films obtained by vacuum evaporation and sputtering,” Thin Solid Films, vol. 24, no. 1, pp. 89–100, Nov. 1974.

[127] H. Windischmann, “Intrinsic stress and mechanical properties of hydrogenated silicon carbide produced by plasma-enhanced chemical vapor deposition,” J. Vac. Sci. Technol. A Vacuum, Surfaces, Film., vol. 9, no. 4, pp. 2459–2463, 1991.

[128] P. J. Rousche and R. A. Normann, “Chronic recording capability of the utah intracortical electrode array in cat sensory cortex,” J. Neurosci. Methods, vol. 82, no. 1, pp. 1–15, 1998.

[129] S. Suner, M. R. Fellows, C. Vargas-Irwin, G. K. Nakata, and J. P. Donoghue, “Reliability of signals from a chronically implanted, silicon-based electrode array in non-human primate primary motor cortex,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 13, no. 4, pp. 524–541, 2005.

[130] R. J. Vetter, J. C. Williams, J. F. Hetke, E. A. Nunamaker, and D. R. Kipke, “Chronic neural recording using silicon-substrate microelectrode arrays implanted in cerebral cortex,” IEEE Trans. Biomed. Eng., vol. 51, no. 6, pp. 896–904, 2004.

[131] J. C. Williams, R. L. Rennaker, and D. R. Kipke, “Long-term neural recording characteristics of wire microelectrode arrays implanted in cerebral cortex,” Brain Res. Protoc., vol. 4, no. 3, pp. 303–313, 1999.

[132] K. A. Ludwig, J. D. Uram, J. Yang, D. C. Martin, and D. R. Kipke, “Chronic neural recordings using silicon microelectrode arrays electrochemically deposited with a poly(3,4- ethylenedioxythiophene) (PEDOT) film,” J. Neural Eng., vol. 3, no. 1, pp. 59–70, 2006.

[133] K. D. Wise, J. B. Angell, and A. Starr, “An Integrated-Circuit Approach to Extracellular Microelectrodes,” IEEE Trans. Biomed. Eng., vol. BME-17, no. 3, pp. 238–247, 1970.

[134] D. R. Kipke, R. J. Vetter, J. C. Williams, and J. F. Hetke, “Silicon-substrate intracortical microelectrode arrays for long-term recording of neuronal spike activity in cerebral cortex,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 11, no. 2, pp. 151–155, Jun. 2003.

120

[135] M. A. L. Nicolelis, D. Dimitrov, J. M. Carmena, R. Crist, G. Lehew, J. D. Kralik, and S. P. Wise, “Chronic, multisite, multielectrode recordings in macaque monkeys,” Proc. Natl. Acad. Sci., vol. 100, no. 19, pp. 11041–11046, Sep. 2003.

[136] L. Etemadi, M. Mohammed, P. T. Thorbergsson, J. Ekstrand, A. Friberg, M. Granmo, L. M. E. Pettersson, and J. Schouenborg, “Embedded ultrathin cluster electrodes for long-term recordings in deep brain centers,” PLoS One, vol. 11, no. 5, pp. 1–18, 2016.

[137] C. T. Nordhausen, E. M. Maynard, and R. A. Normann, “Single unit recording capabilities of a 100 microelectrode array,” Brain Res., vol. 726, no. 1–2, pp. 129–140, 1996.

[138] J. L. Skousen, S. M. E. Merriam, O. Srivannavit, G. Perlin, K. D. Wise, and P. A. Tresco, “Reducing surface area while maintaining implant penetrating profile lowers the brain foreign body response to chronically implanted planar silicon microelectrode arrays,” in Progress in Brain Research, 1st ed., vol. 194, no. December, Elsevier B.V., 2011, pp. 167– 180.

[139] T. D. Y. Kozai, A. S. Jaquins-Gerstl, A. L. Vazquez, A. C. Michael, and X. T. Cui, “Brain tissue responses to neural implants impact signal sensitivity and intervention strategies,” ACS Chem. Neurosci., vol. 6, no. 1, pp. 48–67, 2015.

[140] J. C. Barrese, J. Aceros, and J. P. Donoghue, “Scanning electron microscopy of chronically implanted intracortical microelectrode arrays in non-human primates,” J. Neural Eng., vol. 13, no. 2, p. 026003, 2016.

[141] P. J. Gilgunn, R. Khilwani, T. D. Y. Kozai, D. J. Weber, X. T. Cui, G. Erdos, O. B. Ozdoganlar, and G. K. Fedder, “An ultra-compliant, scalable neural probe with molded biodissolvable delivery vehicle,” in 2012 IEEE 25th International Conference on Micro Electro Mechanical Systems (MEMS), 2012, no. February, pp. 56–59.

[142]P. R. Patel, H. Zhang, M. T. Robbins, J. B. Nofar, S. P. Marshall, M. J. Kobylarek, T. D. Y. Kozai, N. A. Kotov, and C. A. Chestek, “Chronic in vivo stability assessment of carbon fiber microelectrode arrays,” J. Neural Eng., vol. 13, no. 6, pp. 1–15, 2016.

[143] G. G. Stoney, “The Tension of Metallic Films Deposited by Electrolysis,” Proc. R. Soc. A Math. Phys. Eng. Sci., vol. 82, no. 553, pp. 172–175, May 1909.

121

[144] S. F. Cogan, J. Ehrlich, and T. D. Plante, “The effect of electrode geometry on electrochemical properties measured in saline,” in 2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2014, vol. 2014, pp. 6850–6853.

[145] M. A. Long, D. Z. Jin, and M. S. Fee, “Support for a synaptic chain model of neuronal sequence generation,” Nature, vol. 468, no. 7322, pp. 394–399, Nov. 2010.

[146] J. T. Rubinstein, F. A. Spelman, M. Soma, and M. F. Suesserman, “Current Density Profiles of Surface Mounted and Recessed Electrodes for Neural Prostheses,” IEEE Trans. Biomed. Eng., vol. BME-34, no. 11, pp. 864–875, 1987.

[147] M. F. Suesserman, F. A. Spelman, and J. T. Rubinstein, “In vitro measurement and characterization of current density profiles produced by nonrecessed, simple recessed, and radially varying recessed stimulating electrodes,” IEEE Trans. Biomed. Eng., vol. 38, no. 5, pp. 401–408, May 1991.

[148] A. Eleryan, M. Vaidya, J. Southerland, I. S. Badreldin, K. Balasubramanian, A. H. Fagg, N. Hatsopoulos, and K. Oweiss, “Tracking single units in chronic, large scale, neural recordings for brain machine interface applications,” Front. Neuroeng., vol. 7, Jul. 2014.

[149] M. S. Fee, P. P. Mitra, and D. Kleinfeld, “Automatic sorting of multiple unit neuronal signals in the presence of anisotropic and non-Gaussian variability,” J. Neurosci. Methods, vol. 69, no. 2, pp. 175–188, Nov. 1996.

[150] C. Pouzat, O. Mazor, and G. Laurent, “Using noise signature to optimize spike-sorting and to assess neuronal classification quality.,” J. Neurosci. Methods, vol. 122, no. 1, pp. 43–57, Dec. 2002.

[151] M. A. M. Freire, E. Morya, J. Faber, J. R. Santos, J. S. Guimaraes, N. A. M. Lemos, K. Sameshima, A. Pereira, S. Ribeiro, and M. A. L. Nicolelis, “Comprehensive Analysis of Tissue Preservation and Recording Quality from Chronic Multielectrode Implants,” PLoS One, vol. 6, no. 11, p. e27554, Nov. 2011.

[152] G. Buzsáki, “Large-scale recording of neuronal ensembles,” Nat. Neurosci., vol. 7, no. 5, pp. 446–451, May 2004.

122

[153] M. A. Moffitt and C. C. McIntyre, “Model-based analysis of cortical recording with silicon microelectrodes,” Clin. Neurophysiol., vol. 116, no. 9, pp. 2240–2250, Sep. 2005.

[154] A. H. Marblestone et al., “Physical principles for scalable neural recording,” Front. Comput. Neurosci., vol. 7, no. October, pp. 1–34, 2013.

[155] M. Romero-Ortega, A. Kanneganti, G. Bendale, J. Seifert, S. Bredeson, P. Troyk, F. Deku, and S. Cogan, “Chronic and low charge injection wireless intraneural stimulation in vivo,” in 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2015, vol. 2015–Novem, pp. 1013–1016.

[156] S. Bredeson, A. Kanneganti, F. Deku, S. Cogan, M. Romero-Ortega, and P. Troyk, “Chronic in-vivo testing of a 16-channel implantable wireless neural stimulator,” in 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2015, vol. 2015–Novem, pp. 1017–1020.

[157] M. Ghovanloo, K. J. Otto, D. R. Kipke, and K. Najafi, “In vitro and in vivo testing of a wireless multichannel stimulating telemetry microsystem,” in The 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, 2004, vol. 4, no. August 2015, pp. 4294–4297.

[158] L. C. Katz and M. E. Gurney, “Auditory responses in the zebra finch’s motor system for song,” Brain Res., vol. 21, pp. 192–197, 1981.

[159] M. S. Lewicki, “Intracellular Characterization of Song-Specific Neurons in the Zebra Finch Auditory Forebrain,” J. Neurosci., vol. 16, no. 18, p. 5854 LP-5863, Sep. 1996.

[160] M. V. Puig, M. Ushimaru, and Y. Kawaguchi, “Two distinct activity patterns of fast-spiking interneurons during neocortical UP states,” Proc. Natl. Acad. Sci., vol. 105, no. 24, pp. 8428– 8433, Jun. 2008.

[161] A. Destexhe, D. Contreras, and M. Steriade, “Spatiotemporal analysis of local field potentials and unit discharges in cat cerebral cortex during natural wake and sleep states.,” J. Neurosci., vol. 19, no. 11, pp. 4595–608, Jun. 1999.

[162] M. A. C. Wightman Mark R., May Leslie J., “Detection of Dopamine Dynamics in the Brain,” Anal. Chem., vol. 60, no. 13, p. 769A, 1988.

123

[163] J. J. Jun et al., “Fully integrated silicon probes for high-density recording of neural activity,” Nature, vol. 551, no. 7679, pp. 232–236, Nov. 2017.

[164] D. B. McCreery, T. G. H. Yuen, W. F. Agnew, and L. A. Bullara, “Stimulation with chronically implanted microelectrodes in the cochlear nucleus of the cat: Histologic and physiologic effects,” Hear. Res., vol. 62, no. 1, pp. 42–56, 1992.

[165] F. Deku, A. Joshi-Imre, A. Mertiri, T. J. Gardner, and S. F. Cogan, “Electrodeposited Iridium Oxide on Carbon Fiber Ultramicroelectrodes for Neural Recording and Stimulation,” J. Electrochem. Soc., vol. 165, no. 9, pp. D375–D380, Jun. 2018.

[166] J. Y. W. Seto, “The electrical properties of polycrystalline silicon films,” J. Appl. Phys., vol. 46, no. 12, pp. 5247–5254, Dec. 1975.

[167] J. Batey and E. Tierney, “Low-temperature deposition of high-quality silicon dioxide by plasma-enhanced chemical vapor deposition,” J. Appl. Phys., vol. 60, no. 3136, 1986.

[168] T.-C. Yang and K. C. Saraswat, “Effect of physical stress on the degradation of thin SiO/sub 2/ films under electrical stress,” IEEE Trans. Electron Devices, vol. 47, no. 4, pp. 746–755, Apr. 2000.

[169] A. Sheikholeslami, F. Parhami, H. Puchner, and S. Selberherr, “Planarization of Silicon Dioxide and Silicon Nitride Passivation Layers,” J. Phys. Conf. Ser., vol. 61, pp. 1051–1055, Apr. 2007.

[170] J. F. Hetke, J. L. Lund, K. Najafi, K. D. Wise, and D. J. Anderson, “Silicon Ribbon Cables for Chronically Implantable Microelectrode Arrays,” IEEE Trans. Biomed. Eng., vol. 41, no. 4, pp. 314–321, 1994.

[171] C. Hassler, T. Boretius, and T. Stieglitz, “Polymers for neural implants,” J. Polym. Sci. Part B Polym. Phys., vol. 49, no. 1, pp. 18–33, Jan. 2011.

[172] X. Wei, L. Luan, Z. Zhao, X. Li, H. Zhu, O. Potnis, and C. Xie, “Nanofabricated Ultraflexible Electrode Arrays for High-Density Intracortical Recording,” Adv. Sci., vol. 1700625, p. 1700625, 2018.

124

[173] S. K. Gopalakrishnaiah, K. Joseph, and U. G. Hofmann, “Microfluidic drive for flexible brain implants,” Curr. Dir. Biomed. Eng., vol. 3, no. 2, pp. 675–678, Jan. 2017.

[174] T. D. Y. Kozai and D. R. Kipke, “Insertion shuttle with carboxyl terminated self-assembled monolayer coatings for implanting flexible polymer neural probes in the brain,” J. Neurosci. Methods, vol. 184, no. 2, pp. 199–205, Nov. 2009.

[175] A. J. Shoffstall, S. Srinivasan, M. Willis, A. M. Stiller, M. Ecker, W. E. Voit, J. J. Pancrazio, and J. R. Capadona, “A Mosquito Inspired Strategy to Implant Microprobes into the Brain,” Sci. Rep., vol. 8, no. 1, p. 122, Dec. 2018.

[176] M. Lo, S. Wang, S. Singh, V. B. Damodaran, H. M. Kaplan, J. Kohn, D. I. Shreiber, and J. D. Zahn, “Coating flexible probes with an ultra fast degrading polymer to aid in tissue insertion,” Biomed. Microdevices, vol. 17, no. 2, p. 34, Apr. 2015.

[177] D. Lewitus, K. L. Smith, W. Shain, and J. Kohn, “Ultrafast resorbing polymers for use as carriers for cortical neural probes,” Acta Biomater., vol. 7, no. 6, pp. 2483–2491, Jun. 2011.

[178] G. Lind, C. E. Linsmeier, J. Thelin, and J. Schouenborg, “Gelatine-embedded electrodes— a novel biocompatible vehicle allowing implantation of highly flexible microelectrodes,” J. Neural Eng., vol. 7, no. 4, p. 046005, Aug. 2010.

[179] A. Lecomte, E. Descamps, and C. Bergaud, “A review on mechanical considerations for chronically-implanted neural probes,” J. Neural Eng., vol. 15, no. 3, p. 031001, Jun. 2018.

[180] M. A. Howard, B. A. Abkes, M. C. Ollendieck, M. D. Noh, C. Ritter, and G. T. Gillies, “Measurement of the force required to move a neurosurgical probe through in vivo human brain tissue,” IEEE Trans. Biomed. Eng., vol. 46, no. 7, pp. 891–894, Jul. 1999.

[181] J. A. Molloy, R. C. Ritter, M. S. Grady, M. A. Howard, E. G. Quate, and G. T. Gillies, “Experimental determination of the force required for insertion of a thermoseed into deep brain tissues,” Ann. Biomed. Eng., vol. 18, no. 3, pp. 299–313, May 1990.

[182] ChunXiang Tian and Jiping He, “Monitoring Insertion Force and Electrode Impedance during Implantation of Microwire Electrodes,” in 2005 IEEE Engineering in Medicine and Biology 27th Annual Conference, 2005, pp. 7333–7336.

125

BIOGRAPHICAL SKETCH

Felix Deku was born in Mangoase, Ghana, and received a BS degree in molecular biology and biotechnology with First Class Honors from The University of Cape Coast, Cape Coast, Ghana in

2008. He received a MS degree in biomedical engineering from The University of Texas at Dallas,

Richardson, Texas in 2017. His PhD work was supervised by Dr. Stuart F. Cogan where he gained extensive experience working with amorphous silicon carbide films and developing microelectrode arrays for neural recording and stimulation. During his doctoral work, Felix discovered the relationship between residual stress variation in amorphous silicon carbide films and post-deposition oxidation and demonstrated the influence of chamber conditions on the long- term stress stability of amorphous silicon carbide films. He also demonstrated the fabrication of microelectrode arrays using amorphous silicon carbide films as sole substrate and encapsulation.

His research interest is in thin-film fabrication, neural stimulation and recording, electrochemical processes at neural interfaces, electrodes and electrode coating development, and the development and characterization of novel implantable materials. Felix has presented posters and talks at several international conferences including Biomedical Engineering Society (BMES), the Society for

Neuroscience (SfN), the Neural Interfaces Conference (NIC), North American Neuromodulation

Society (NANS) and the Gordon Research Conference for Bioelectronics Interfaces. He was one of a few diversity travel awardees to attend NANS/NIC and the GRC. He was also a Founders

Fellowship Finalist at The University of Texas at Dallas and enjoyed a handsome stipend supplement during his doctoral work. Felix is a member of the Electrochemical Society (ECS),

BMES and the SfN.

126

CURRICULUM VITAE

Felix Deku Address: 800 W. Campbell Road, Richardson TX 75080 USA Email: [email protected]

EDUCATION PhD, Biomedical Engineering January 2015 – Aug 2018 University of Texas at Dallas Richardson Texas, USA

MS, Biomedical Engineering January 2014 - May 2017 University of Texas at Dallas, Richardson TX, USA

BS, Molecular Biology and Biotechnology August 2004 - May 2008 University of Cape Coast, Cape Coast, Ghana

RESEARCH ASSISTANT EXPERIENCE Neural Interfaces Laboratory, Department of Bioengineering, University of Texas at Dallas, Richardson TX. USA The neural interface laboratory focuses on developing materials and devices for stimulation and recording in the nervous system to treat neurological diseases and sensory deficits. We focus on thin-film fabrication of microelectrode arrays and implantation in neural tissues. Our lab also investigates the electrochemical properties and stability of electrodes and devices in bench-top and animal studies. My research focuses on the development of intracortical ultramicroelectrodes utilizing amorphous silicon carbide as the sole substrate and encapsulation. Experience in utilizing thin-film photolithography techniques to develop microelectrode arrays with subcellular or cellular scale dimensions (e.g. devices with shank cross-sectional area less than 60 square microns) to minimize the effects of foreign body response. The resultant electrode sites are of ultramicroelectrode

dimensions (research space yet to be explored for neural stimulation). Extensive experience in microfabrication, material characterization, electrochemical evaluation and electrophysiology.

RESEARCH INTERESTS Thin-film fabrication of microelectrodes Neural stimulation and recording Electrochemical processes at neural interfaces Electrodes and electrode coating development Development and characterization of novel implantable materials

PUBLICATIONS 1. F. Deku, Y. Cohen, A. Joshi-Imre, A. Kanneganti, T.J. Gardner, S.F. Cogan, Amorphous silicon carbide ultramicroelectrode arrays for neural stimulation and recording, J. Neural Eng. 15 (2018) 016007. doi:10.1088/1741-2552/aa8f8b. 2. F. Deku, A. Joshi-Imre, A. Mertiri, T.J. Gardner, S.F. Cogan, Electrodeposited Iridium Oxide on Carbon Fiber Ultramicroelectrodes for Neural Recording and Stimulation, J. Electrochem. Soc. 165 (2018) D375–D380. doi:10.1149/2.0401809jes. 3. W.F. Gillis, C.A. Lissandrello, J. Shen, B.W. Pearre, A. Mertiri, F. Deku, S. F. Cogan, B.J. Holinski, D.J. Chew, A.E. White, T.M. Otchy, T.J. Gardner, Carbon fiber on polyimide ultramicroelectrodes, J. Neural Eng. 15 (2018) 016010. doi:10.1088/1741-2552/aa8c88. 4. J.J. Pancrazio, F. Deku, A. Ghazavi, A.M. Stiller, R. Rihani, C.L. Frewin, V.D. Varner, T.J. Gardner, S.F. Cogan, Thinking Small: Progress on Microscale Neurostimulation Technology, Neuromodulation Technol. Neural Interface. 20 (2017) 745–752. doi:10.1111/ner.12716. 5. C.L. Frewin, E.E. Bernardin, F. Deku, R. Everly, J. Hassan, J.J. Pancrazio, S.E. Saddow, (Invited) Silicon Carbide as a Robust Neural Interface, ECS Trans. 75 (2016) 39–45. doi:10.1149/07512.0039ecst. 6. S. Bredeson, A. Kanneganti, F. Deku, S. Cogan, M. I. Romero-Ortega, P. Troyk, Chronic in-vivo testing of a 16-channel implantable wireless neural stimulator, in: 2015 37th Annu. Int.

Conf. IEEE Eng. Med. Biol. Soc., IEEE, 2015: pp. 1017–1020. doi:10.1109/EMBC.2015.7318537. 7. M. I. Romero-Ortega, A. Kanneganti, G. Bendale, J. Seifert, S. Bredeson, P. Troyk, F. Deku, S. Cogan, Chronic and low charge injection wireless intraneural stimulation in vivo, in: 2015 37th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., IEEE, 2015: pp. 1013–1016. doi:10.1109/EMBC.2015.7318536.

CONFERENCE PRESENTATIONS 1. Felix Deku, Alison Stiller, Christopher Frewin, Alexandra Joshi-Imre, Victor Varner, Joseph Pancrazio, Stuart Cogan Progress on amorphous silicon carbide ultramicroelectrode array development (GRC Neuroelectronic Interfaces 2018, Poster) 2. Felix Deku, Ellen Shih, Aswini Kanneganti, Alexandra Joshi-Imre, Joseph J. Pancrazio, Stuart F. Cogan. Evaluation of amorphous silicon carbide multielectrode arrays in rodent motor cortex (SFN 2017, Poster) 3. Bitan Chakraborty, Aswini Kanneganti, Rashed Rihani, Bryan J. Black, Felix Deku, Rafi Ayub, Christopher Frewin, Alexandra Joshi-Imre, Joseph J. Pancrazio, and Stuart F. Cogan. Investigation of Blackrock Microelectrode Arrays Chronic Electrical and Recording Performance in Rat Motor Cortex (SFN 2017, Poster) 4. E. J. Welle, P. R. Patel, F. Pourdanesh, F. Deku, D. Egert, A. Ghazavi, A. Joshi-imre, J. D. Weiland, J. D. Berke, C. A. Chestek. Characterization of carbon fiber electrode tip preparation and coatings to increase longevity. (SFN 2017, Poster) 5. Felix Deku, Yarden Cohen, Ben Pearre, Alexandra Joshi-Imre, Atefeh Ghazavi, Winthrop Gillis, Timothy Gardner, Stuart Cogan. Evaluation of Novel Amorphous Silicon Carbide Ultramicroelectrodes for Neural Interfacing (BMES 2016, Podium) 6. Felix Deku, Alexandra Joshi-Imre, Yarden Cohen, Ben Pearre, Timothy Gardner, Stuart F. Cogan. Fully-Integrated Amorphous Silicon Carbide Ultramicroelectrode Array for Neural Stimulation and Recording (NANS/NIC 2016, Poster)

7. Christopher L. Frewin, Felix Deku, Evans Bernardin, Richard Everly, Jawad Ul Hassan, Erik Janzén, Joseph J. Pancrazio, and Stephen E. Saddow. Electrical Performance of Single Material Silicon Carbide (SiC) Microelectrodes (NANS/NIC 2016, Poster) 8. A.Kanneganti, F.Deku, S.Bredeson, P.Tryok, S.Cogan, M.I.Romero. A 14-month chronic study of Wireless Microelectrode Array (WFMA) shows stable and selective graded stimulation of rat sciatic nerve by non-invasive quantification of hind limb motor recruitment 9. Felix Deku, Alket Mertiri, Atefeh Ghazavi, Stuart F. Cogan, Timothy J. Gardner. EIROF- coated Carbon Fiber Ultramicroelectrodes for neural Stimulation and Recording (BMES 2015, Poster) 10. G. S. Bendale, A. Kanneganti, F. Deku, S. Bredeson, J. L. Seifert, P. Troyk, S. Cogan, M. I. Romero-ortega. A Novel Wireless Microelectrode Array Implanted Chronically Provides Reliable and Selective Stimulation (SFN 2015, Poster)

TECHNICAL SKILLS o Extensive cleanroom experience and deep expertise in thin-film fabrication. o Process engineering, process development and quality refinement. o Experience in designing implantable medical devices, thin-film microelectrode array design, development and fabrication. o PECVD and Plasma RIE, wet etch chemistry. o Thin-film metallization and hand-on experience developing encapsulation/ barrier processes and photo-patterning. o Extensive experience in developing medical devices based on amorphous silicon carbide and common polymers such as polyimide, Parylene C etc. o Expertise in performing inspection and utilizing characterization techniques including FTIR, XPS, SEM, AFM, Ellipsometry, Profilometry, Microscopy, Stress analysis. o Deep understanding of electrochemical characterization techniques such as electrochemical impedance spectroscopy, voltammetry and voltage transient measurements.

o Extensive understanding of stimulating parameters including choice of waveforms; biphasic vs monophasic, cathodal first vs anodal first pulsing; frequency (pps), pulse width, duty cycle, interface delays, interpulse bias etc. o Extensive experience with operating potentiostats, two or three electrode systems and picoamp current measurements. o Experience in packaging, wire bonding, soldering etc. o Proficiency with Autosketch or Autocad CAD tools with working knowledge of k-layout. o Experience with rat surgical procedures and rat cortical implantation. o Experience with electrophysiological recording

AWARDS & HONORS o Bachelor of Science degree in Molecular Biology and Biotechnology with First Class Degree Honors (May 2008) o Founders Fellowship Finalist Stipend Supplement Award, University of Texas at Dallas, Richardson TX USA (2015-2018) o Diversity Travel Award, NANS/NIC joint conference (2016) o Diversity Travel Award, GRC Bioelectronics Interfaces Conference (2018)

LANGUAGUES English, Twi, Ewe,

PROFESSIONAL MEMBERSHIPS Electrochemical Society (ECS) Biomedical Engineering Society (BMES) Society for Neuroscience (SfN)