USF Libraries
USF Digital Collections

The neuron-silicon carbide interface


Material Information

The neuron-silicon carbide interface biocompatibility study and BMI device development
Physical Description:
Frewin, Christopher L
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:


Subjects / Keywords:
Cubic silicon carbide
Nanocrystalline diamond
Implantable neuronal prosthetics
Mammalian cell biocompatability
Neural action potential
Dissertations, Academic -- Electrical Engineering -- Doctoral -- USF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


ABSTRACT: Damage to the central nervous system (CNS) leads to the generation of an immune response which culminates with the encapsulation of the damaged area. The encapsulation, known as a glial scar, essentially breaks neural signal pathways and blocks signal transmissions to and from the CNS. The effect is the loss of motor and sensory control for the damaged individual. One method that has been used successfully to treat this problem is the use of a brain-machine interface (BMI) which can intercept signals from the brain and use these signals to control a machine. Although there are many types of BMI devices, implantable devices show the greatest promise with the ability to target specific areas of the CNS, with reduced noise levels and faster signal interception, and the fact that they can also be used to send signals to neurons.The largest problem that has plagued this type of BMI device is that the materials that have been used for their construction are not chemically resilient, elicit a negative biological response, or have difficulty functioning for extended periods of time in the harsh body environment. Many of these implantable devices experience catastrophic failure within weeks to months because of these negative factors. New materials must be examined to advance the future utilization of BMI devices to assist people with CNS damage or disease. We have proposed that two semiconductor materials, cubic silicon carbide (3C-SiC) and nanocrystalline diamond (NCD), which should provide solutions to the material biocompatibility problems experienced by implantable BMI devices.We have shown in this study that these two materials show chemical resilience to neuronal cellular processes, and we show evidence which indicates that these materials possess good biocompatibility with neural cell lines that, in the worst case, is comparable to celltreated polystyrene and, in most cases, even surpasses polystyrene. We have utilized 3C-SiC within an electrode device and activated the action potential of differentiated PC12 cells. This work details our initial efforts to modify the surfaces of these materials in order to improve cellular interaction and biocompatibility, and we examine our current and future work on improving our implantable BMI devices.
Dissertation (Ph.D.)--University of South Florida, 2009.
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Christopher L. Frewin.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 183 pages.
General Note:
Includes vita.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002317626
oclc - 660842532
usfldc doi - E14-SFE0003079
usfldc handle - e14.3079
System ID:

This item is only available as the following downloads:

Full Text


The Neuron-Silicon Carbide Interface: Biocompatibi lity Study and BMI Device Development by Christopher L. Frewin A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering Department of Electrical Engineering College of Engineering University of South Florida Major Professor: Stephen E. Saddow, Ph.D. Edwin Weeber, Ph.D. Andrew M. Hoff, Ph.D. Mark J. Jaroszeski, Ph.D. Ashok Kumar, Ph.D. Karl Muffly, Ph.D. Jing Wang, Ph.D. Date of Approval: May 28, 2009 Keywords: cubic silicon carbide, nanocrystalline di amond, implantable neuronal prosthetics, mammalian cell biocompatability, neura l action potential Copyright 2009, Christopher L. Frewin


Dedication I would like to dedicate this work to my whole com mittee. Each and every one of them has been involved closely with this research, and will continue to do so as we enter the next stage. Your knowledge, assistance, mentor ship and friendship have been invaluable to me. Secondly, I would like to thank my current associates and my friends: Chris Locke, Alexandra Oliveros, Norelli Schettini, Gene Short, and Eduardo MurphyPerez. Your help, patience and all the good times after work have kept me sane and moving forward. I also thank my many associates wh o have moved on or have returned to their countries: Dr. Jeremy “the raging’ Cajun” Walker, Dr. Andrea Severino, Patrizia Pappalardo, Dr. Massimo Camarda, Dr. Ulrich Starke, and Sebastian Schoell. Your hospitality and friendships are much appreciated. I would like to give a special thank you to the one who means a lot to me: Dr. Camilla Colet ti. She started our group onto the path of biotechnology, and was a very good friend o f mine. Thank for being there when I needed you. To my parents and family, thanks for h elping me get through the years of school and putting up with me not being around. Th anks also to my many fiends: Steve Bradshaw, Frank Pytel, Chris Coules, Steve Short, a nd the whole Hollis Clan. Now you have read this, perhaps you understand what I was d oing all the time. Finally, thank you to the original ‘Dr. Evil’, Ste phen Saddow. Thanks for giving a misfit a chance to change the world, and for push ing me harder each year (can it get a little better now?). I will be forever in debt for your mentorship and support. We will carry on and seize the day until the bitter end.


Acknowledgements I would like to thank the Max Planck Institut in S tuttgart, DE, and Dr. U. Starke for the opportunity to work alongside them on two o ccasions. Specifically, I would like to thank M. Hetzel and Dr. C. Virojanadara, and C. Riedl for their expertise with LEED, STM, and XPS and our work in (§ 2.3). Next, I woul d like to thank IMM-CNR in Catania, IT, and Dr. F. La Via for hosting me twice during our SiC collaboration. I would like to thank Dott. G. D'Arrigo for teaching me how to generate masks, C. Bongiorno for his excellent TEM work [(§ 2.2), (§ 5 .2)], and M. Italia for his SEM imaging (§ 2.2). I would like to thank the Walter Schottky Institut in Mnchen, DE, and Dr. M. S. Brandt, for our collaborative work into s urface modification of SiC. I would like to thank Dr I. D. Sharp and M. Hoeb for their work on surface functionalization and ATR-FTIR measurements in (§ 4.2). A special thanks goes to Dr. P. Spagnol and Dr. J. Bumgarner of SRI International in Largo FL for thei r collaboration work in (§ 2.2.3) and their financial support. At USF, I would like to t hank the NNRC facilities and staff for their assistance over the years of SiC film growth and characterization and A. Oliveros and Dr. Saddow for fabricating the first SiC NAD de vices (§4.4). At the Johnnie B. Byrd Sr. Alzheimer’s Center & Research Institute, I woul d like to thank Dr. J. Rogers and Dr. M. Peters for their assistance with the cell lines in (§3) and the NAD device activation in (§4.4). I also thank Dr. M. L. Selenica at the Byr d Institute for her primary mouse cortical cells and our work in (§4.3). I thank H. G omez and Dr. J. Sathyaharish for their NCD deposition on 3C-SiC substrates (§2.4). I wish to acknowledge the Office of Naval Research, Dr. C. Wood, for many years of financial support. Finally, I thank the USF FMMD fellowship and Dr. T. Vickerson for her ki ndness and support.


i Table of Contents List of Tables .................................... ................................................... .............................. iii List of Figures ................................... ................................................... .............................. iv Abstract……. ....................................... ................................................... .......................... vii Chapter 1: Introduction ........................... ................................................... ..........................11.1 Research Objective and Motivation ............. ................................................... 11.2 The Basic Central Nervous System .............. ..................................................61.2.1 The Neuron Cell ............................. ..................................................81.2.2 The Glia Cell ............................... ................................................... 101.2.3 Glial Scarring .............................. ................................................... 121.3 Theory of Bioelectricity ...................... ................................................... .......141.3.1 Cellular Membrane Theory .................... ........................................151.3.2 Parallel Conductance Model .................. ........................................161.3.3 Hodgkin’s and Huxley’s Equations ............ ...................................171.3.4 Microelectrode Device Interaction ........... .....................................201.3.5 Field Effect Device Interaction ............. .........................................241.4 Brain Machine Interface (BMI) ................. ................................................... 271.4.1 Non-invasive Neural Prosthetics.............. ......................................281.4.2 Invasive Implantable Prosthetics ............ .......................................301.5 Cubic Silicon Carbide (3C-SiC) ................ ................................................... 331.6 Nanocrystalline Diamond (NCD) ................. ................................................371.7 Summary of Chapter ............................ ................................................... ......39 Chapter 2: Cubic Silicon Carbide and Nanocrystallin e Diamond Material Development ....................................... ................................................... ..........412.1 Introduction .................................. ................................................... ..............412.2 Chemical Vapor Deposition Growth of 3C-SiC .... .......................................442.2.1 Establishing a Low Pressure Process for CVD 3 C-SiC Growth ............................................ ...............................................452.2.2 Low to Atmospheric Pressure Carbonization Com parison ............502.2.3 Growth of 3C-SiC on Poly-Si/ SiO2/ (111)Si ................................562.2.4 Characterization of the 3C-SiC on the Poly-Si Stack ....................612.3 Hydrogen Etching of SiC Polytypes ............. ................................................682.3.1 Low Pressure Hydrogen Etching Process ....... ...............................702.3.2 Characterization of Hydrogen Etched SiC Surfa ces ......................71


ii 2.4 NCD Deposition on 3C-SiC/ Si .................. ..................................................7 62.4.1 NCD Deposition Process ...................... .........................................772.4.2 Material Characterization.................... ...........................................782.5 Summary of Chapter ............................ ................................................... ......85 Chapter 3: Neural Biocompatibility of SiC and NCD ................................................... ...863.1 Chapter Introduction .......................... ................................................... ........863.2 Foreign Material and Living Cell Surface Intera ction ..................................883.2.1 SiC Biocompatibility ........................ .............................................893.2.2 Sample and Cleaning Protocol ................ .......................................933.2.3 Cell Culture and MTT Assay Protocol ......... .................................953.2.4 AFM Methodology for Cell Morphology and Subst rate Permissiveness .................................... ...........................................963.3 Experimental Results .......................... ................................................... .......983.4 Discussion of Results ......................... ................................................... ......1083.5 Summary of Chapter ............................ ................................................... ....118 Chapter 4: Biocompatibility Improvement and Device Interactions ...............................1204.1 Introduction .................................. ................................................... ............1204.2 Semiconductor Surface Modification and Explorat ion...............................1224.2.1 6H-SiC Surface Modification Characterization ...........................1244.2.2 Biocompatibility Testing of Modified Surfaces ..........................1314.3 Mouse Primary Cortical Neurons (CTX) on 3C-SiC ..................................1354.3.1 Cortical Cell Biocompatibility Testing ...... ..................................1364.4 Neural Activation Device (NAD): Generation 0 .. ......................................1404.4.1 NAD0 Activation of a Neural Action Potential ...........................1424.5 Summary of Chapter ............................ ................................................... ....147 Chapter 5: Future Work ............................ ................................................... ....................1495.1 Introduction .................................. ................................................... ............1495.2 Cell to Substrate Interaction Study ........... ..................................................1 505.2.1 Surface Modification and Biocompatibility Imp rovement ..........1515.3 Primary Mouse Cortical and Astrocyte Cell Bioco mpatibility Study ........1535.4 NAD Experimental Approach...................... ...............................................1555.4.1 NAD Signal Reception from Neuron Action Poten tials ..............1575.4.2 NAD2 Interaction with a Hippocampus Slice ... ..........................1605.5 Conclusion .................................... ................................................... ...........163 References… ....................................... ................................................... ..........................166 Bibliography ...................................... ................................................... ...........................183 About the Author .................................. ................................................... .............. End Page


iii List of Tables Table 1.1: Electrical, thermal and physical propert ies of SiC and Si [70-72]. ...........35 Table 1.2: Electrical, thermal and physical propert ies of diamond and SiO2 [78-82]............................................ ................................................... .........38 Table 2.1: XPS data of 3C-SiC films grown on variou s Si surfaces after 400 Torr carbonization. ............................... ................................................... ...51 Table 2.2: Summary of hydrogen etching parameters f or SiC surfaces [110113]. ............................................. ................................................... ...........71 Table 3.1: Compiled values representing the cell in teraction with the tested substrates as measured through AFM techniques. .... ...............................108


iv List of Figures Figure 1.1: A cutaway cartoon representation of one hemisphere of the human brain. ............................................ ................................................... .............7 Figure 1.2: A representation of a neuron displaying the soma, dendrites, and axon .............................................. ................................................... ...........10 Figure 1.3: Computer simulations of the action pote ntial using the HodgkinHuxley equations for (a) a polarizing current and ( b) a hyperpolarization current. ........................ ..................................................1 9 Figure 1.4: A depiction of the capacitive double la yer formed at the surface of a metal in an electrolyte [13] .................... .................................................21 Figure 1.5: A circuit model of the electrode-cell i nterface .......................................... 22 Figure 1.6: A cartoon representation of a neural ce ll on a MOSFET device where the cell membrane forms the gate metal ...... ...................................25 Figure 2.1: A generalized graph displaying the gene ral process schedule developed at USF for the growth of 3C-SiC on Si sub strates ....................47 Figure 2.2: AFM micrographs of 3C-SiC on Si surface s after the carbonization growth stage ...................................... ................................................... ......50 Figure 2.3: AFM micrographs of 3C-SiC on Si surface s after 1 minute of the growth stage ...................................... ................................................... ......53 Figure 2.4: TEM plan view micrographs displaying gr owth of 3C-SiC on (100)Si using carbonization at (a) atmospheric pres sure, and (b) 400 Torr .......................................... ................................................... ........55 Figure 2.5: (a) An 5 x 5 m AFM micrograph of the s urface of the poly-Si/ SiO2/ (111)Si stack showing the pinholes in the thin fi lm, and (b) shows an image from optical microscope obtained at 25X displaying the 3C-SiC growth on the poly-Si substra te stack using a growth temperature of 1250 C ................................................. ..............60


v Figure 2.6: AFM micrographs showing a comparison of 3C-SiC grown on the surfaces of (a) poly-Si/ SiO2/ (111)Si, (b) (111)Si, and (100)Si ............... .62 Figure 2.7: Cross-section SEM micrograph of the 3CSiC film on the poly-Si/ SiO2/ (111)Si stack (20 min growth time). ............. ...................................63 Figure 2.8: XRD -2 diffraction surveys for the 3C-SiC films grown on (a-b) poly-Si/ SiO2/ (111)Si, (c) (111)Si, and (d) (100)Si substrates .................65 Figure 2.9: TEM data taken in plan-view from sample USF2-07-170B, which is a 3C-SiC film grown on the 20 nm thick poly-Si s ubstrate stack. .........67 Figure 2.10: 10 x 10 m AFM micrographs of SiC surf aces displaying the surface before (left) and after (right) H2 etches using the parameters from Table 2.2 ......................... ................................................72 Figure 2.11: AFM micrographs of (100)3C-SiC surface s (a) epitaxially grown on (100)Si after subsequent H2 etches at the temperatures indicated, and (b) AFM micrographs of (111)3C-SiC s urfaces epitaxially grown on (111)Si after subsequent H2 etches ..........................74 Figure 2.12: Three dimensional AFM micrographs of N CD growth on various substrates. ....................................... ................................................... .........78 Figure 2.13: SEM plan view (column 1) and cross-sec tion (column 2) micrographs of the NCD deposited on the various tes ted substrates. ........80 Figure 2.14: Raman (a) and NEXAFS (b) absorption fi ne structure scans of NCD deposited on (100)3C-SiC ...................... ..........................................82 Figure 2.15: XRD -2 diffraction surveys for the NCD films deposited on (a) (100)3C-SiC, (b) (111)3C-SiC and (c) (100)Si substr ates .........................83 Figure 3.1: Depiction of various AFM measurements u tilized for the determination of cell permissiveness on the substra tes tested in this study ............................................. ................................................... ...........98 Figure 3.2: H4 and PC12 cell proliferation on Si an d SiC substrates measured via MTT on the fourth day of culture normalized to the PSt control and expressed as .................................................. ........................99 Figure 3.3: Selected 45 x 45 m AFM micrographs of H4 cells seeded on the substrates tested in this study ................... ................................................100


vi Figure 3.4: Selected 45 x 45 m AFM micrographs of PC12 cells deposited on the substrates tested in this study. .............. ..............................................105 Figure 3.5: 5 x 5 m AFM micrographs of the surface s of the three tested semiconductors .................................... ................................................... .108 Figure 4.1: ATR-FTIR measurements of (0001) and (00 0-1)6H-SiC surfaces that have been hydrogen etched and terminated ..... .................................127 Figure 4.2: ATR-FTIR measurements of (001) 3C-SiC t hat had been hydrogen etched and terminated ............................. .................................................12 9 Figure 4.3: A bar graph displaying the results of t he MTT analysis on 6H-SiC samples, normalized to the PSt control and then exp ressed as. ................................................. .............................................131 Figure 4.4: The results of single MTT assays perfor med with triplicate samples expressed as .................................................. ...............138 Figure 4.5: SEM micrographs of PLL coated PSt sampl e (a) with cortical cell imaged at 25KV and 8000X (scale bar 1m) and (b) a cortical cell at 25KV and 6000X (scale bar 2um) ................. ......................................139 Figure 4.6: Simplified cartoon representations of a n (a) electrode-electrolyte interface and a (b) metal-insulator-semiconductor ( MIS) capacitor device and an electrolyte which was covered in deta il within (§ 1.3.4 and 1.3.5) .................................. ................................................... ...141 Figure 4.7: The NAD0 device generated on 3C-SiC con sisting of gold leads and a polyethylene gasket ......................... ...............................................143 Figure 4.8: Electrical activation of cell action po tential via the SiC NAD0 ..............144 Figure 5.1: Cartoon representation of the proposed BMI that will be demonstrated by 2 NADs of the same material constru ction (NAD#1 and NAD#2) ................................. .............................................159


vii The Neuron-Silicon Carbide Interface: Biocompatibi lity Study and BMI Device Development Christopher L. Frewin ABSTRACT Damage to the central nervous system (CNS) leads t o the generation of an immune response which culminates with the encapsula tion of the damaged area. The encapsulation, known as a glial scar, essentially b reaks neural signal pathways and blocks signal transmissions to and from the CNS. The effe ct is the loss of motor and sensory control for the damaged individual. One method tha t has been used successfully to treat this problem is the use of a brain-machine interfac e (BMI) which can intercept signals from the brain and use these signals to control a m achine. Although there are many types of BMI devices, implantable devices show the greate st promise with the ability to target specific areas of the CNS, with reduced noise level s and faster signal interception, and the fact that they can also be used to send signals to neurons. The largest problem that has plagued this type of BMI device is that the mat erials that have been used for their construction are not chemically resilient, elicit a negative biological response, or have difficulty functioning for extended periods of time in the harsh body environment. Many of these implantable devices experience catastrophi c failure within weeks to months because of these negative factors. New materials m ust be examined to advance the future utilization of BMI devices to assist people with CN S damage or disease.


viii We have proposed that two semiconductor materials, cubic silicon carbide (3CSiC) and nanocrystalline diamond (NCD), which shoul d provide solutions to the material biocompatibility problems experienced by implantabl e BMI devices. We have shown in this study that these two materials show chemical r esilience to neuronal cellular processes, and we show evidence which indicates tha t these materials possess good biocompatibility with neural cell lines that, in th e worst case, is comparable to celltreated polystyrene and, in most cases, even surpas ses polystyrene. We have utilized 3CSiC within an electrode device and activated the ac tion potential of differentiated PC12 cells. This work details our initial efforts to mo dify the surfaces of these materials in order to improve cellular interaction and biocompat ibility, and we examine our current and future work on improving our implantable BMI de vices.


1 Chapter 1: Introduction 1.1 Research Objective and Motivation For humans and mammals, the central nervous system (CNS) processes incoming data from the senses, controls muscles and organs, and is a storehouse for acquired knowledge. Unfortunately, damage to this system fr om disease or trauma does not readily repair itself because scar tissue is formed in the damaged area that blocks the normal flow of CNS electrical signals. The effect of this scarring is to block signaling between the CNS and the organs/muscles it serves, l eaving the individual with decreased or no functionality in the blocked systems. Extrem e damage can cause paralysis of the motor system, blindness or deafness, while certain diseases, such as Alzheimer’s and Parkinson’s, cause malfunctions in the signaling ca pabilities of the brain. Brain machine interface (BMI) devices, also known as the brain co mputer interface (BCI) devices, is a burgeoning technology developed by engineers, neuro logists, and computer scientists that has the potential to offer hope and possible cures for many problems associated with the central nervous system (CNS) [1-3]. BMI devices al low communication between the signals generated by neurons in the brain and exter nal devices, and implantable BMI devices allow for the potential of two way communic ation [1-3]. A BMI with two way communication allows the device to act as a bridge that potentially restores neural pathways damaged by a glial scar, or intercepted si gnals are utilized to control external


2 devices which in turn could send information back t o the brain. This technology has the potential to allow motor restoration to paralyzed i ndividuals or those with diseases such as Parkinson’s. In particular, with so many affect ed by injuries from war, amputee victims will be able to use their mind to control a motorized prosthetic and receive sensory information back from this device. Interaction with biological processes at the cellu lar level is an important component of implantable BMI devices. Micro-electr omechanical systems (MEMS) are one method that opens an avenue for this interactio n to be realized. Electrical interaction with muscle and nerve cells is possible using the M EMS systems, and MEMS microfluidic devices potentially can allow targeted deli very of drugs or cellular enzymes to the brain [4, 5]. Although MEMS technology offers many new possibilities to interact with the cellular environment, current-generation neural prosthetic devices require indefinite long term chronic implantation to perform their des igned functions. Extended interaction with the body requires that the materials used have a high degree of biocompatibility which must be considered for the design of the devi ces. Biocompatibility generally refers to the ability of a device and its component materi als to perform their desired functions, maintaining beneficial cellular and tissue response s, and not eliciting undesirable local or systemic effects in the body [6]. This biocompatib ility requirement therefore places significant restrictions on the design of implantab le MEMS devices, and each part of the body possesses slightly different factors of biocom patibility. Silicon, Si, has been long used in the constructio n of MEMS devices as the material possesses the hardness of steel, is relati vely inexpensive to produce, and can be


3 easily fashioned into complex mechanical devices us ing techniques developed by the semiconductor industry [3-5]. Si is a primary cons tituent of many MEMS devices because it is a semiconductor, a material which can be electrically insulating or conductive depending on both its properties as well as the stimulus that is applied to the material. Hence semiconductor materials can be used to electronically sense stimuli which allows for the realization of complex electri cal devices. These devices allow for the precise control of electrical potential fields (voltages) and currents, and can be combined to make complex devices like sensors, micr oprocessors, and radio transmission devices. Complex electrical devices are the backbo ne for the functionality of many of the MEMS devices used for cellular interaction [3-5]. Unfortunately, the material properties of Si that allow for the construction of these comp lex devices provide a complex problem which complicates the biocompatibility of Si. Si i s not chemically resilient, possessing a high reactivity to numerous acids and bases as well as allowing chemical diffusion into the crystalline lattice, and it has been shown to a ctively react with many biological enzymes and fluids [7-11]. Silica, SiO2, is the native oxide of Si and is also the primary dielectric component of field effect devices, but i t is easily contaminated with metallic ions, many of which are found commonly within the b ody [12]. SiO2 also hydrolyses into silicic acid, Si(OH)4, a compound that is found as a trace element throu ghout the body in blood plasma and connective extracellular tissues; however, the exact purpose this compound serves with respect to cellular metabolic process is still being investigated [711]. Some studies show that silicic acid reacts wi th Al-containing compounds to produce substances that may contribute to Alzheimer’s and P arkinson’s disease [11]. Also,


4 minute silicate particulates have been known to be carcinogenic to many organs and cells and these materials usually cause initiation of the body’s inflammatory response which activates phagocytes and eventually escalates into material encapsulation [9, 10]. Once a Si-based sensor, for example, is encapsulated it cl early ceases to be a sensor and becomes a useless foreign object, and further body reaction s can lead to cell mortality [9-12]. Because of this bioreactivity, Si is not a material that can directly interact with the biological environment and must rely on other mater ials for this interaction as well for its own protection. Mechanical interaction can be achieved through phy sical coatings added over Si, such as polymers or ceramics. Electrical interacti on with the cells and the body’s constituents is achieved through extensions of cond uctive electrodes [3, 13, 14]. Although many of the physical coatings, especially polymers, allow the MEMS devices increase their biocompatibility and interact more s pecifically with the body environment, the electrodes are very susceptible to changes in i mpedance caused by biofouling or encapsulation, and many materials used for these el ectrodes, like platinum, iridium, iridium oxide, and conductive polymers have Faradic reactions at the current levels required for cellular interaction and, therefore, s uffer physical damage and degradation over time [14]. Field effect devices such as capac itors and field-effect transistors would be able to react with the environment through capac itive instead of Faradic reactions, but would require a biocompatible and bioresistive semi conductor material to achieve this interaction [14, 15].


5 One material that has been used in direct interact ions with the body environment is silicon carbide, SiC. Silicon carbide is a phys ically strong material and is chemically inert making it ideal for interaction with the hars h body environment. Amorphous and polycrystalline forms of silicon carbide, a-SiC and poly-SiC, have been used as coatings for many implantable biomedical devices, and clinic al studies for bone prosthetics and heart stents have confirmed the biocompatibility of these forms of SiC [16-19]. Single crystal SiC has an additional advantage in that it is a semiconducting material like Si, which enables the creation of complex electrical de vices and utilizes many of the device processing techniques available to Si. Two polytyp es of Si, 4H-SiC and 6H-SiC, have been available commercially as substrate wafers and have shown good biocompatibility, non-toxicity, and low biofouling [20]. Cubic silic on carbide 3C-SiC, the only polytype of silicon carbide with a zinc blended atomic structur e, can be heteroepitaxially grown on other zinc blende crystals like silicon and diamond or it can be heteroepitaxially grown on the hexagonal forms of SiC, namely 4H-SiC and 6H -SiC [21-24]. 3C-SiC possesses superior strength, excellent electrical properties, and chemical resilience when compared with other semiconductor materials, and would make an excellent platform for the creation of an implantable MEMS neural interface fo r BMI devices. This chapter is organized to serve as an introduct ion to biological systems and physical concepts involved with a man-made neural i nterface. First, the central nervous system (CNS), how it is organized and functions, an d its major cellular constituents are described and detailed (§ 1.2). Next, the concepts behind neural bioelectricity and how electrical based device could interact with them ar e explained (§ 1.3). BMI device


6 operation, both non-invasive and invasive operation forms, is explained in (§ 1.4). Finally, the semiconductor materials 3C-SiC (§ 1.5) and nanocrystalline diamond (NCD) (§ 1.6) are explored with respect to their unique m aterial properties and biocompatibility with mammalian cells as possible construction media for BMI devices. 1.2 The Basic Central Nervous System The nervous system serves as the control center an d signal transport for vertebrate organisms, and monitors all sensory input, motor fu nctions, and information processing. Anatomically, this system is composed of two distin ct subdivisions, the CNS and the peripheral nervous system (PNS). The CNS is compos ed of the brain, the brain stem, and the spinal cord, while the PNS contains all the con nection pathways from the CNS to sensory and motor actuators. The CNS is a complex organ system whose operation is just beginning to be understood. The human brain, shown in Figure 1.1, composes the majority of the CNS and has four primary systems each with different fu nctions. The first system is the cerebrum which is composed of two cerebral hemisphe res connected through the corpus callosum and represents 85% of the brain’s weight. This part of the brain controls voluntary actions, processes sensory information, a nd is involved in thinking, learning, and decisions. The second part of the brain, compo sing about 10% of the weight, is the cerebellum which processes sensory information, and is linked closely with the motor system. This system also maintains balance and lear ning associated directly with motor functions. The third part of the brain, the limbic system, links the higher brain functions


7 from the cerebrum with the automatic systems of the brain stem. The limbic system is composed of four parts: the amygdala processes stro ng emotions; the hippocampus processes short term memory into long term memory; the thalamus controls sensory information; and the hypothalamus regulates body se cretions and temperature. Finally, the brain stem is the fourth system of the brain wh ich controls all automatic functions, like breathing and heart rate, and provides a signa l pathway to the spinal cord. The second system of the CNS, the spinal cord, provides a pathway for signal transmission to sensory and maintenance organs as well as controls reflexive actions. Figure 1.1: A cutaway cartoon representation of one hemisphere of the human brain. The major components of the brain are indicated. [2 5].


8 Like other complex organs in the body, the CNS is composed of a multitude of specialized cells that allow it to complete its tas ks efficiently. This cellular diversity complicates interaction with this organ from an imp lantable device as each cell can potentially react differently. (§ 1.2.1) and (§ 1. 2.2) will explore the basic cells involved with an implantable device while (§ 1.2.3) details the cellular response to damage and foreign materials. 1.2.1 The Neuron Cell The main workhorse of the CNS is the neuron which is displayed in Figure 1.2. The neuron cell produces the signals that control b ody functions and is the target cell for an implantable BMI device. This cell is composed o f three main parts. The soma, or cell body, contains the nucleus and other cell organelle s common to all cells and regulates specialized enzymes and neurotransmitters. The sho rt, fibrous dendrites extend from the soma and contain specialized receptors that interce pt incoming chemical or electrochemical signals from other cells. Although the dendrites mainly are utilized for the reception of signals, some actually have two-wa y signaling functionality. The longer, fibrous axon is mainly utilized to transmit signals to other neurons or cells. Specialized protein-gated channels along the axon allow the tra nsmission of electrical signals, and the long microtubule-filled structure provides a pathwa y for the transport of organelles and vesicles to the synapse. Neurons are not directly connected to other cells in the neural network, but the axon termination is located near a receiving dendri te or cell and the space between them


9 is called the synapse. Two types of synapses are f ound within the CNS system, an electrical synapse and a chemical synapse. A neuro n can generate or transfer electrochemical signals called an action potential which is the product of electrical potentials created by the movement of ions through membrane channels, and these potentials in turn generate diffusion current that is propagated through the axon. As the AP reaches the terminal button in the axon containi ng an electrical synapse, cell pores open and ions travel freely between a very small sy naptic gap between the cells. The gap is very tight and isolated from the extracellular m edium as to prevent loss of the signal due to leakage, and some of the cell pores are acti vated electrically, allowing for some rectification of the signal. Neurotransmitters are special chemicals generated within the cell that are designed to either suppress or excite the electrica l excitation in a neighboring cell. Neurotransmitters are packaged in vesicles and tran sported down the axon. When an action potential excites a terminal button containi ng a chemical synapse, calcium permeable channels open allowing Ca2+ to enter the membrane. The high level of calcium activates proteins in the vesicles walls an d they move toward the synapse. As shown in the inset in Figure 1.2, the vesicles dock with the synaptic membrane and release the neurotransmitter into the gap. Recepto rs in the neighboring cell bind some of the neurotransmitters, which cause a change in ion permeable channels. The open ionic gates allow ions to move into or out of the cell an d modulate its transmembrane potential, and can depolarize the membrane if the transmitters are excitatory or hyperpolarize the membrane if the transmitters are inhibitory.


10 Figure 1.2: A representation of a neuron displayin g the soma, dendrites, and axon. The inset displays one synapse and details how neurotra nsmitter molecules are transmitted from one neuron to another [25]. 1.2.2 The Glia Cell The glia are a group of cells that outnumber the n eurons in the CNS by as much as 50 to 1, and they provide a regulatory role by m aintaining nutrition, structural support, aid signal transmission, and form the brain-blood b arrier. The first cell that a BMI implant would encounter is the astrocyte, also know n as astroglia. The astrocyte, a cell that is now beginning to be considered very importa nt in the CNS system, are the most abundant and variable glial cell in the CNS. First they provide structural support for the


11 other cells in the CNS [26]. Secondly, they mainta in the synapse and extracellular space by clearing expended neurotransmitters and excess i ons, and modulate signals through the release of neurotransmitters [27]. These cells also promote axonal guidance and myelination from oligodendrocyte cells [28, 29]. A strocytes help maintain the bloodbrain barrier, the barrier formed by endothelial ce lls that prevents blood from contacting CNS tissue directly [26]. Finally, astrocytes play an important role in CNS repair, which will be covered in more detail in (§ 1.2.3), a fact that places this cell in the interface path of an implantable device. Another very important glial cell is the oligodend rocyte which produces the myelin sheath around the axon. The myelin prevents the AP from being dissipated into the space surrounding the axon, allows the current to ‘hop’ down the system instead of traveling in a wave, and provides a pathway for reg eneration of the axon if it is severed. The myelin creates a depolarization at a node of Ra nvier which is sufficient to elevate the neighboring node to threshold potential and thus in creasing the speed of signal propagation. Microglia are a form of macrophage, or cell eater, which responds to injury in the CNS. These cells make up 20% of the glia population and are located throughout the CNS at all times unlike regular blood derived micro phages. They are vigilant sentinels which are inactive for the most part, but monitor t he CNS for damaged neurons, infectious agents, or foreign body invasions [30, 3 1]. Their exact role in the inflammatory process will be covered in more detail in (§ 1.2.3). Another important glial


12 cell that would encounter an implantable device is the ependyma cell which generates and transports cerebrospinal fluid throughout the C NS. Another group of cells that is encountered by impl antable devices comes from the meninges which is a protective membrane sheath cons isting of the dura matter, the arachnoid matter, and the pia matter. The Dura mat ter is a thick, fibrous, leather-tough layer which is infused with the major veins and art eries for the brain. This layer is the layer that must be pierced before a device can inte ract with the cells of the brain. The arachnoid matter possesses a spider-web appearance and acts as a shock absorber for the brain. The pia matter is firmly anchored to the br ain and is intertwined with capillaries that deliver the blood to the brain matter. An imp lantable device must prevent damage to this layer to preserve the blood-brain barrier. 1.2.3 Glial Scarring The reaction of the CNS to injury is a complex and little understood process. Injury, almost independent of source, causes a simi lar reaction that permanently leaves the recipient with decreased function. It has been shown by Santiago Ramn y Cajal in the late 19th century that neurons in vitro have the ability to regenerate damaged axons, but neurons in vivo do not regenerate and therefore do not reconnect s ignal pathways [32]. Instead, a process called reactive gliosis, more commonly referred to as glial scarring, is the reaction of the CNS to injury from trauma, disease, or long term inflammation [30, 31]. Although the exact purpose of this reaction is unknown, glial scarring essentially isolates and encapsulates the damaged portion of the brain and


13 restores the blood-brain barrier. Unfortunately th e glial scarring process also has a downside in that it prevents the regeneration of ax ons across the breach [31]. The final scar is created by the presence and injury response s of the glial cells, namely astrocytes, microglia, and oligodendrocytes, and if the meninge s was compromised during the process that initiated the glial scarring response, these cells are also react to the injury [31]. The process begins with the activation of the sent inel microglia cells, which begin to divide and migrate to the injury site, and if th e blood-barrier was broken, white blood cell macrophages will also migrate to the site to a ssist the microglia cells [31]. The activated microglia and macrophages work to clean t he site of myelin, which is inhibitory to axon growth, and necrotic tissue, but they also can produce cytotoxic responses for the purpose of destroying pathogenic organisms if they are present. The cytotoxic responses can, over time, poison the damaged area and inhibit regeneration of the axon [31]. If the meninges were damaged, meningeal cells rush to the site of the injury to reestablish a new continuous covering for the CNS, however these cells do not provide a substrate that is conducive to growth of new axons [31]. The olig odendrocytes and the myelin that they produce are present at the site, especially in the case of trauma, and the oligodendrocytes usually collapse regenerating growth of axon cones that contact with them [31]. Oligodendrocytes do not seem to promote axon growth or the generation of new myelin sheaths [31]. The majority of the glial scar is caused by the ac tions of the astrocytes. The injury changes the cell into a reactive astrocyte w hich has an increase in filament and


14 general cell size. The cell becomes phagocytic and ingests necrotic cellular material and it also releases inhibitive molecules that are non permissive to axon growth [31]. Although these cells seem to have the ability to in itiate neurogenesis in adult stem cells, the activated cells instead undergo cell division a nd fill the extracellular space around the damage [26]. The final scar is tissue composed of tightly packed hyperfilamentous astrocytes, many of which have processes that oppos e one another, and have tight junctions that forms a barrier to fluid. The tight junctions restore the blood-brain barrier, but the tight mesh of cells prevents all further at tempts from surviving neurons to form axons. 1.3 Theory of Bioelectricity Many cells in the body use electrical signals for communication or activation of functionality. Neurons are the major transport pat hway of these signals, and their main function is to generate and relay signals to actuat ors of the body. Actuator cells include muscle cells and many sensory associated cells. Th e generation of signals is a complex action derived by the difference of inorganic ionic concentrations outside and inside the cell, and the electric potential created by these i onic concentrations. (§ 1.3.1) begins with detailing the cellular membrane theory, followed by the parallel conductance model in (§ 1.3.2) and the quantification of the action potenti al in (§ 1.3.3). External stimulation of muscles and nerves was first observed and recorded by L. Galvani in 1783. (§ 1.3.4) and (§ 1.3.5) detail how an implantable device can inte ract electrically with a neuron at a cellular level.


15 1.3.1 Cellular Membrane Theory The cell membrane is composed of a lipid bilayer w hich organizes itself so the hydrophilic polar head faces the water environment and the non-polar hydrophilic tails face away from the water. Within this lipid membra ne are various highly selective pumps and channels, some of which serve as transpor t for various ionic components. If these channels are opened, the ions within the extr acellular and intracellular matrix flow along both the ionic gradient and the changes in th e electric field generated by the ions on either side of the insulating lipid bilayer. The Ne rnst-Planck equation describes the current density vector in an electrochemical cell: F + = RT F C Z C FZ D Jp p p p p p (1.1) In this equation the electric current density, Jp, measured in coulombs per second per cm2, is a function of the diffusion coefficient Dp, the concentration Cp, the sign of the force, Zp, Faradays constant, F, the gas constant, R, and th e temperature, T. The Nernst-Planck equation can be used to derive the Nernst equilibri um which describes the electrical potential difference across a membrane, i e, or the interior potential minus the external potential. This difference simply equates to the membrane voltage, Vm, and is obtained through an integration of terms using the capacitance-voltage relationship as described by Ohms law. The resulting equation is a s follows:


16 [ ] [] = F F =e p i p p e i mC C F Z RT V ln (1.2) This model successfully explains the action of a si ngle ion concentration brought into equilibrium across the membrane, but does not take into consideration that all the other ionic species in the electrochemical solution have different concentrations and equilibrium levels. There are also many ionic chan nels with different conditions along a single membrane. A new model would have to be deve loped to describe the membrane voltage. 1.3.2 Parallel Conductance Model The parallel conductance model was developed by Da vid Goldman, Alan Hodgkin, and Bernard Katz to describe the dynamic m echanisms that are interacting across a cell membrane. The formulation of their m odel is a direct solution of the Nernst equation (1.2). The overall total potential across the membrane, Em, is a product of three main channels, potassium (K), sodium (N), and chlor ine (Cl). The membrane capacitance is placed in parallel with these three channel circuits to complete the model and the current through the membrane is then given by the Goldman-Hodgkin-Katz (GHK) equation: [ ] [ ] [ ] [][][] n + + + + =e Cl i Na i K i Cl e Na e K mCl P Na P K P Cl P Na P K P F RT V ln where d D Pq q q b = (1.3) ( ) q x m q qE V g I = (1.4)


17 t V C Im c = (1.5) Cl Na K C mI I I I I + + + = (1.6) For each of the equations, q is the index for the i on type, is the force sign constant, d is the thickness of the membrane, and D is Fick’s coef ficient. The parallel conductance model gave a mechanism for the combination of vario us ionic channels to explain the dynamic voltage across the membrane; however the co nductance’s for the model proved to perform in a nonlinear manner and also to change drastically over time. It was noticed that if a membrane was sufficiently excited, a sign ificant change would occur in the membrane channels that would elicit a large voltage response. At this voltage, known as the threshold voltage, Vt, most of the ionic channels would activate and the combination of their actions would create a large potential kno wn as the action potential, and below this value, the membrane channels would display no statistical change. 1.3.3 Hodgkin’s and Huxley’s Equations In 1952, Alan Hodgkin, Andrew Huxley, K. C. Cole, and the squid Loligo, investigated the nature of the action potential thr ough the non-linear conductance of K+ and Na+ [33-36]. Their work, which was rewarded with a No bel Prize, used a combination of a mechanism known as the space clamp to hold the voltage potential of the squid axon membrane at a experimentally determi ned level and modification of the ions in the media concentration [33-36]. The combi nation of these methods enabled the elimination of a particular channel’s influence fro m the Goldman-Hodgkin-Katz equation


18 so that each individual type of channel conductance could be evaluated separately [3336]. This experimental approach allowed the quanti fication, and later mathematical modeling as a function of time and of the membrane voltage, of the action potential with its relationship to the changing conductance of the K+ and Na+ channels and the effect of a leakage current [33-36]. The action potential wa s determined to be caused by a large change in the conductance of sodium followed by a s lower change in potassium conductance and a slow reduction in sodium conducta nce caused by channel inactivation [33-36]. The experimentally obtained conductance w as modeled statistically through curve fitting and correlated mathematically to the probability of the channel to be open or inactivated [33-36]. The probabilities were labele d (n) for potassium channel activation, and (m, h) for sodium activation and inactivation, respectively [33-36]. By using these probabilities, the total membrane voltage could be evaluated numerically through the use of the following equations [33-36]: ( ) ( ) ( ) [ ] i i m l Na i m i i Na K i m i K d m mE V g E V h m g E V n g I C t v D = D3 4 (1.7) t V C I I I Im l Na K d + + + =d (1.8) () ( ) () K m K KE V t I t g = (1.9) () ( ) () Na m Na NaE V t I t g = (1.10) ( ) ( ) m K m Kv t n g v t g ,4= (1.11)


19 ( ) ( ) ( ) m m Na m Nav t h v t m g v t g , ,3= (1.12) ( ) l m l lE V g I = (1.13) These equations relate the membrane voltage, Vm, to the exciting current, Id, the currents for the K+, Na+, and L (leakage or the Clchannel), and the membrane capacitance, Cm [33-36]. The action potential of a membrane can be modeled using these equations and a computer simulation based on this model is displaye d in Figure 1.3 for polarizing and hyperpolarizing exciting currents. (a) (b) Figure 1.3: Computer simulations of the action pot ential using the Hodgkin-Huxley equations for (a) a polarizing current and (b) a hy perpolarization current. (a) displays the polarizing stimulating current, Istim, the membrane voltage, Vm, and the gating variables (m, n, h) as a function of time variations as well as displaying the concept of absolute refractory period with the addition of a second exc itation current. (b) displays the temporal variations of the same factors as a respon se to a hyperpolarizing current [37].


20 The equation not only models the action potential as a function of the exciting current, but displays the concept of a refractory p eriod. If an action potential is activated, a period of time is required for the ionic channels to reset to resting state before another action potential can be executed no matter the leve l of the stimulus, and this time period is known as the absolute refractory period. Howeve r, there is a time at which sufficient channels have returned to the resting state and the action potential can be excited again with an abnormally high stimulus. This time period is known as the relative refractory condition and both of the refractory periods can be determined from the Hodgkin-Huxley equations. 1.3.4 Microelectrode Device Interaction The methods developed by Hodgkin, Huxley, and Cole to activate and measure the electrical activity of a neuron required a larg e instrument called the space clamp [3336]. Due to technological limitations of the late 1940’s, the wire electrodes required large membrane areas, like the one provided by the large nerves of the giant squid, and the signaling electronics were restricted to vacuum tube technology. In the late 1950’s, new fabrication techniques were enabling the creati on of glass microcapillary electrodes, later known as the patch clamp, that worked along s imilar principals to the space clamp to investigate the intracellular properties of smal l cells [38]. Although this method is excellent for understanding the neurophysiology of individual neuron cells, the method is limited to in vitro methodology and generally can only be utilized for a short time.


21 A parallel line of research developed around the s ame time utilized insulated metal electrodes to investigate populations of cell s [39]. This type of electrode was further refined in 1970 by utilizing techniques for microfabrication developed by the burgeoning Si semiconductor manufacturers. K. Wise et al. developed a planar microelectrode using a Si substrate, glass, and gol d [13]. An advantage of this electrode is it can interact and target specific cells instea d of generalized populations and had a very low profile of approximately 15 m [3, 4]. Al though other, more biologically compatible materials are utilized for the modern pl anar microelectrode, it remains the most popular style of electrode device as it is eas ily, inexpensively produced to exacting and repeatable standards and can be utilized in vivo for extended times [3, 4]. Figure 1.4: A depiction of the capacitive double l ayer formed at the surface of a metal in an electrolyte [13]. Solvated Cation Solvent (H 2 O) Adsorbed Anion Neutral Molecule Outer Helmholtz Plane Inner Helmholtz Plane Helmholtz Capacitive Double Layer Gouy-Chapman Diffuse Layer


22 Generally, a metal inserted into an electrolyte wi ll undergo oxidation, thereby extruding an electron and becoming a positively cha rged ion and is depicted in Figure 1.4 [13, 14]. The electrons are forced to the surface of the metal by the repulsive forces from the abundant electrons in the bulk, and the surface electrons in turn attract ionic species in the electrolytic solution. The ions gathered to th e surface form the Helmholtz plane, a double layer consisting of the inner Helmholtz plan e which is mainly solvated water molecules surrounding a cation, and outer Helmholtz plane which is composed of the solvated cations themselves. At surface equilibriu m, a Gouy-Chapman diffuse layer is generated beyond the Helmholtz layer by thermal mot ion. Figure 1.5: A circuit model of the electrode-cell interface. The cell membrane (a), coupling (b), and electrodes (c) are separated by d otted boxes [40, 41]. Rel Cel Rsprd Rseal CHD Rmetl Cm ENa EK ECl ENa RK RCl (a) Cellular Membrane (b) Cell / Electrode Junction (c) Electrode/ Electrolyte


23 The electrode-electrolyte to cell interface can be described by modeling the double layer interaction with the GHK membrane mode l. The combined circuit is shown in Figure 1.5 [40 -42]. The metal-electrode juncti on is modeled as a resistor, Rmetl, in parallel to a shunt capacitance, Csh, which accounts for all parasitic capacitances fro m the tip of the electrode to the external electronics [4 0]. The double layer capacitance from the electrolyte / metal junction is modeled as a pa rallel capacitor and resistor in series with the metal, Rel and Cel [40]. The cell-electrode junction is modeled as a resistor, Rsprd, or the voltage drop due to the distance of the ce ll from the electrode [41]. Rseal is the voltage drop across the cleft between the cell membrane and the electrode, and CHD is the capacitance effects of the double layer on the cellular membrane [41]. The cellular membrane can be electrically modeled as a capacitan ce generated across the lipid membrane in parallel with the major semipermeable c hannels. Each channel is modeled as a battery in series with a variable resistor whi ch is linked to the equations developed by Hodgkin and Huxley (§ Although this model explains both transmission and reception of signals by a planar microelectrode, there is a problem with thes e devices. Instead of working within the capacitive regime of the double layer, many of the materials used to date for electrode construction experience irreversible oxidation or r eduction reactions at high chargeinjection levels required for neuron activation, kn own commonly as Faradic electrode reactions [14]. Platinum, iridium, and conductive p olymers experience Faradic reactions at the charge injection levels required for neuron activation, and the resulting chemical reactions degrade the electrodes over time, produce chemical byproducts which can lead


24 to a foreign body response from the immune system, or produce hydrogen or oxygen that attack cells [14]. Another problem with electrode b ased neural implant devices is they require relatively large geometrical areas to gener ate the appropriate level of charge injection. One method to alleviate this difficulty is to use at least two electrodes to complete the circuit loop when signaling a neuron t o allow a balance in the ionic current flow required for neuron membrane depolarization [3 14]. Implantable neural devices must be designed as small as possible to decrease e vasiveness, and the space requirements for electrode wiring feed-through to o nboard electronics create major limitations in the amount of active interaction loc ations due to this small size. Finally, electrodes are highly sensitive to changes in imped ance levels, which elevate the resistances, Rsprd and Rseal. One of the causes of this factor originates from the chronic foreign body response, glial scarring, covered in ( § 1.2.3), which eventually produces device encapsulation or the electrodes become biofo uled with adsorbed biological material [14]. 1.3.5 Field Effect Device Interaction An alternative to electrode interaction with neuro ns was developed by P. Fromherz et al. from the Max-Planck-Institut which has shown that Si metal-oxidesemiconductor capacitors (MOScap) can stimulate neu ron membrane depolarization and Si metal-oxide-semiconductor field effect transisto rs (MOSFET) can receive action potential signals from neurons [15, 42-45]. The maj or advantage of MOS or metalinsulator-semiconductor (MIS) devices is they inter act with the cell through capacitive


25 field effects without using electroporation or elec trochemistry. Also, these devices are not as susceptible to changes in impedance as their electrode counterparts [15, 42-45]. Another advantage for FET devices is they can be in terlinked into arrays by combining common connections, thereby reducing the amount of feed-through wiring required and allowing more active interaction areas on the neura l interface [45, 46]. FET capacitor stimulation, unlike the electrode device, requires only one monophasic pulse to maintain charge balance, as charge dissipates gradually thro ugh RC parasitic losses [15, 43-44]. (a) (b) Figure 1.6: A cartoon representation of a neural c ell on a MOSFET device where the cell membrane forms the gate metal. (a) shows a represen tation of the neuron acting as a gate for the MOSFET biased in saturation. (b) shows a c ircuit representation of the neuron on a transistor based of the membrane effects modeled by Hodgkin and Huxley (§ [42]. To receive signals from the cell, Fromherz et al. utilized a MOSFET with no gate metal [42]. Figure 1.6(a) shows the concept that a neural cell essentially forms the gate metal through the coupling between the gate oxide a nd the cell membrane which they Silicon B Extracellular Medium (ECM) Neuron D S Id Vm(t) ECM Neuron Vm(t) Silicon CJG VJ(t) RJ CJM VJM RJM


26 called the transductive extracellular potential but is also known as the sealing resistance in (§ 1.3.4) [42]. The MOSFET driven into saturati on, or active mode, with Vsd > |Vgs Vth|, Vs = Vb = VDD, and VDD >> Vsd, provides a constant current flow through the device [42]. In saturation, the MOSFET drain curre nt is directly controlled by the gate voltage. Activation of the action potential within a neuron will generate an influx of ions into the space between the membrane and the gate ox ide, which leads to changes in ionic and displacement currents. This current modulates the transductive extracellular potential above the oxide, which in turn modulates the drain current within the MOSFET device [42]. Figure 1.6(b) displays a representati on of the electrical circuit formed by the neuron and the MOSFET with discrete elements indica ted [42]. The membrane voltage, VM(t), as a function of time is a product of the memb rane channel with capacitance, CJM, in parallel with the series combination of the vari able channel resistance, RJM, and channel potential, VJR, all factors derived from the Hodgkin and Huxley e quations. The membrane voltage is coupled to the specific capacit ance of the gate oxide, CJG, and the sheet resistance of the extracellular medium, RJ. The transductive extracellular voltage, VJ(t), is the time dependent response to changes in t he membrane voltage as influenced by sealing resistance RJ. The stimulation of an action potential in a neuron can also be achieved through a field effect device, the metal-oxide-semiconductor capacitor, MOScap, or metalinsulator-semiconductor capacitor, MIScap [43, 44]. Fromherz et al. used the neuron membrane as the device metal, which formed the junc tion over an oxide on a semiconductor as with his transistor experiments [4 3, 44]. The MOScap is stimulated


27 with a voltage, VST, which gives rise to ionic and displacement curren ts in the junction. The currents would modulate the transductive extrac ellular voltage, and if this voltage was greater than the threshold voltage of the neuro n, Vt, it would depolarize the membrane and generate an action potential [43, 44]. The advantage of using this stimulation over the electrode stimulation is that a single monophasic square pulse would lead to charge balance due to the charge dissipatio n of the capacitor [43, 44]. 1.4 Brain Machine Interface (BMI) As discussed in (§ 1.2.3), injury to the CNS does not readily regenerate but is instead isolated and encapsulated with a glial scar The inability of the brain to send and receive signals leads to loss of function of muscle s and sensory organs. However, undamaged portions of the CNS continue to function as before, and one proposed method to provide therapy is through the use of the brain machine interface (BMI). The broad concept of a BMI is to interface with undamaged ner ves and then intercept, interpret, and transmit these signals to a machine or even another neuron in the body [1, 2]. This section is dedicated to giving a brief explanation of the current technology involved in BMI devices. (§ 1.4.1) will cover the non-invasive BMI devices which are currently in use by many patients suffering from amyotrophic lat eral sclerosis (ALS), and (§ 1.4.2) details the BMI devices which utilize implantable n eural microelectrode devices and the difficulties associated with them.


28 1.4.1 Non-invasive Neural Prosthetics The indirect neural signal recording technology co nsists of electroencephalography (EEG), electrocortiography, (ECoG), galvanic skin response, near infrared recordings, and functional MRI [47]. EEG is a popular method of obtaining the input signals for a BMI device and has been use d for well over 20 years. EEG based BMI devices consist of electrodes placed on specifi c locations of the scalp of the patient or subdermally directly over the cerebral cortex [4 8, 49]. The uses of EEG in neurological medicine have generated many methods f or extracting signals related to action potentials. The device samples the electric al currents produced by post-synaptic potentials of many neurons, and can be adjusted to detect slow cortical potentials, , and rhythms, or responses to “oddball” stimulus, commo nly known as the P100 potential [49, 50]. The cortical area of interest is sampled electrically through current or voltage reception by the electrode, the sample is f iltered through the use of electrical filtering and common spatial patterns. Spatial pat tern filtering is the use of physical separation of electrodes, a method which weighs eac h electrode as to its importance to the signal discrimination, and correlations between neighboring electrodes [50]. The signals are then averaged and the appropriate patte rn is extracted from the signal, and if the EEG is used as a BMI interface, this pattern ta kes the form of a command [49]. The extracted information is then transferred to a comp uter or robotic interface which acts on the desired command. The use of EEG has many obvious advantages to peop le who would require the BMI device. First, it is purely non-invasive and d oes not require surgery, with the


29 exception of subdermally implanted EEG electrodes [ 49]. The device can extract many different types of potentials, and the positioning of the electrodes for collection of the potentials is well established [50, 51]. However t here are many drawbacks to the use of EEG BCI devices. They are usually cumbersome, requ ire multiple electrode attachments, require a long time for electrode to scalp applicat ion, and are uncomfortable for extended periods of time [49]. The potentials read by the s ystem are combinations of a large number of neurons which need to generate a signific ant spike in activity that will not be eliminated due to electronic filtering, signal aver aging or by the power drop across the highly resistive scalp [49]. Real-time execution o f actions is further complicated by the extraction paradigms which limit execution of comma nds to only a few times per minute [49]. Finally, the patient must learn to generate an arbitrary signal and relate it to the appropriate command for the BMI device to function [51]. The generation of this arbitrary brain signal requires a great deal of con scious effort which requires a long training period, and both emotional and external en vironmental situations can interrupt the patients concentration which can lead in errors in signal interpretation by the EEG BMI device [51]. Even with all of these problems, the EEG indirect BMI has provided many solutions for people who suffer from paralysis and its use has developed many invaluable methods for extracting information from neural activity. Individuals with neuron degradation due to disease have a reduced signal output which lowers to a level that can no longer be ascer tained with EEG devices. Galvanic skin response (GSR) is caused by a change in electr ical conductivity of the skin by increased activity in sweat glands and can be measu red by electrodes attached to the skin


30 [52]. The response was first documented by Carl Ju ng and is used as a biofeedback mechanism for evaluating emotional states, and has been used in the polygraph lie detection machine [52]. The GSR can only produce a binary response, and requires a long training period when used as a BMI device [52] Unlike the EEG, it requires only two electrodes for operation, thus increasing the c omfort level and decreasing application time. However, it suffers many of the same problem s associated with the EEG. Drastic changes in visual input can create artificial spike s in the extracted signal, but auditory stimulation seems to stabilize the GSR response [52 ]. Another problem is that the “off/ no” signal is easier to maintain 80% of the time, w hile raising the GSR above threshold for an “on/ yes” signal is only 30% effective [52]. Operator fatigue, anticipation of response, and GSR stabilization also provide diffic ulties to achieving the desired response 100% of the time [52]. 1.4.2 Invasive Implantable Prosthetics Whereas the non-invasive BMI is based on the post synaptic electrical current averaged from relatively large areas of neurons, im plantable devices can directly target single neurons to collect their action potential si gnals, or can collect local field potentials, which is a measure of dendritic synaptic activity f or all neurons in the immediate vicinity [48]. Implantable BMI devices possess many advanta ges, and disadvantages, that must be accounted for if these devices are to become a v iable therapeutic solution for restoration of the damaged CNS. Implantable BMI de vices can collect signals in realtime and manipulate the signal without long term av eraging required from non-invasive


31 techniques [49]. The ability to target specific ar eas of the brain, like motor neurons that control the movement of the arm, allows specific si gnals to be recorded so potentially the device needs no abstract signal generation or ‘oddb all’ stimulus that is required by indirect BMI devices [49]. Experimental BMI device s have been shown to allow rats to control mobile robots, monkeys to control a robotic arm, and a human with ALS could control a computer by thought alone [53-55]. The greatest advantage of the implantable device i s that they do not only allow reception of signals, but it can be used to induce action potentials in neurons allowing sensory input restoration for patients with somatic sensory loss [49]. This innovative technology has seen many successes such as the Coch lear implant, which assists people who have nerve damage associated with hearing, and the vagus nerve implant, which alleviates depression levels for people who are res istant to conventional antidepressant drug therapies [56, 57]. This factor also allows s ensors integrated with robotic devices to relay information back to the brain to allow improv ed operation and provide sensory awareness of the prosthetic [2]. The two-way commu nication ability of an implantable BMI would allow for the construction of a device th at could bridge signals over damaged neural areas thereby restoring natural functionalit y [2]. The implantable BMI device offers many solutions f or the repair of the CNS system, but there are a few problems that must be s urmounted. The selectivity of the devices is a major problem, and multiple signals ca ptured by the neural implant creates difficulties in real-time filtering, sorting, and d ecoding of the signal into usable commands [49]. Complex, computationally intensive algorithms are required to sort


32 these signals, and the speed at which signals are g enerated requires digitization speeds greater than 20 kHz [49]. The CNS response to chronic implantation is one of the greatest difficulties facing the widespread incorporation of BMI devices and pre vents its use in long term clinical studies [49]. The implantation of these devices, a nd subsequent activation of the healing process, lead to a “foreign body” response from the support cells in the CNS, the glia, whose actions escalate the levels of harsh chemical s in the implant area and lead to the encapsulation of the device [31]. Not only is this reaction medically unsuitable, but the BMI electrode devices are extremely vulnerable to c hanges in impedance and fail to function typically within months after encapsulatio n [31, 58-60]. Non-biocompatibility of the construction materials with the CNS tissue h as been linked to this adverse reaction [31]. Si has been a favored construction material for BMI devices, but this material has long been known to elicit a negative response from chronic implantation, and Si is not chemically resistant to many biological chemical pr ocesses [31, 58-60]. Also, as these devices require an insulation material to electrica lly isolate the electrodes, and this layer contacts more tissue area than the substrate and el ectrode materials, the biocompatibility of that layer is paramount to acceptance of the dev ice by the CNS [31]. Many of the devices using SiO2 and Si3N4 as insulators have shown a incompatibility with th e CNS, but polyimide, a chemically resistive polymer mater ial which also incorporates many existing microfabrication techniques used for the c onstruction of microelectrode devices, has shown the it has excellent biocompatibility [31 61-63]. This material has been used as an insulating coating and substrate material for many BMI device designs which have


33 worked for extended periods without total encapsula tion [61-63]. Although this material offers excellent biocompatibility, displaying a fib roblast adherence, growth, and spread comparable to polystyrene, it has major drawbacks t hat hinder its acceptance as the material of choice for BMI devices [64]. The mater ial has a relatively high water uptake which decreases electrical impedance, and it is a v ery flexible material [31]. Although this flexibility alleviated the issue of micromotio n, a problem associated with stiff BMI devices where the device’s relative motion causes a dditional penetration and injury response cycles, the material is difficult to impla nt requiring deposition on a stronger material or complex surgical implantation technique s using wires or scalpels [31]. As stated in (§ 1.1), a semiconductor material, si licon carbide, may offer a different approach to the generation of implantable neural MEMS devices. The advantages of using this material will be covered i n more detail in the next section. 1.5 Cubic Silicon Carbide (3C-SiC) Silicon carbide is a physically strong material an d is chemically inert making it ideal for interaction with the harsh body environme nt. Amorphous and polycrystalline forms of silicon carbide, a-SiC and poly-SiC, have been used as coatings for many implantable biomedical devices, and clinical studie s for bone prosthetics and heart stents have confirmed the biocompatibility of these forms of SiC [16-18]. A recent application for a-SiC was to use it as a thin film coating for the Utah Probe neural prosthetic constructed from silicon [65]. The advantage of th ese forms of SiC is that they can be deposited at lower temperatures than single crystal SiC, but have strictly insulating


34 properties because of the large amount of crystalli ne defects present in the material [1618, 21, 65]. Silicon carbide is a semiconducting material like Si, which enables the creation of complex electrical devices. SiC also is able to ut ilize many of the device processing techniques available to Si [21]. Unlike many semic onductors, SiC is a polymorphic material, having approximately 170 different crysta l organizations, but sharing a common stoichiometry (i.e., all forms contain bi-layers of Si and C) [21]. Table 1.1 summarizes some of the physical and electrical properties of a few of the SiC polytypes alongside with Si. Two of the hexagonal polytypes of SiC, 4H -SiC and 6H-SiC, have been available commercially as substrate wafers, and hav e been fashioned into devices [66]. Light emitting diodes, LEDs, high power Schottky Di odes, and metal-insulatorsemiconductor field effect transistor, MISFET, devi ces are all commercially available which utilize 4H-SiC and 6H-SiC [66]. These materi als have been formed into electrode myocardial biosensors, and have shown good biocompa tibility, non-toxicity, and low biofouling [19, 20]. However, 4H-SiC and 6H-SiC ar e much more difficult to fabricate into complex devices because of their chemical iner tness and physical strength. The hexagonal forms of SiC require temperatures upward of 1600 C for epitaxial growth, which makes incorporation with other materials very difficult [67]. As there is no ability to diffuse chemical dopants into the crystalline st ructure, SiC doping must either be done during the growth process or through ionic implanta tion and subsequent activation [21]. Etching of SiC is achieved using dry etching method s like reactive ion etching [21]. The effects of these difficulties mean that complicated integrated circuits that would be


35 necessary for the amplification, filtering, and rad io transmission of neural signals is difficult at best to reproduce in hexagonal SiC. Table 1.1: Electrical, thermal and physical proper ties of SiC and Si [70-72]. 3C-SiC 4H-SiC 6H-SiC Si Band Gap (eV) 2.36 3.23 3.0 1.12 Breakdown Field (MV cm 1 ) 1 3-5 3-5 0.25 Electron Mobility (cm 2 V 1 s 1 ) at 300K <800 <900 <400 1500 Hole Mobility (cm2 V-1s-1) at 300K <320 <120 <90 450 Thermal conductivity (W/cm-K) 5 4.9 4.9 1.5 Thermal expansion @300 K @1500-1600 K (10-6 K-1) 3.8 5.5 NA 4.3 c axis 4.7 c axis 2.6 4.56-4.6 Young’s Modulus (GPa) 300 to 700 47 Elastic Modulus (GPa) 433 424 170 Cubic silicon carbide, 3C-SiC, is the only polytyp e of silicon carbide with a zinc blended atomic structure [21]. With this cubic cry stal structure, it can be heteroepitaxially grown on other cubic structure su bstrates, like silicon or diamond. Although 3C-SiC has an overall cubic structure, it is still formed by stacking alternating hexagonal bilayers which gives this structure the a bility to be growth on the hexagonal forms of SiC [68, 69]. 3C-SiC offers an alternativ e to processing not available in other forms of SiC in that it can be grown on Si substrat es which are affected by cheaper wet chemical machining techniques. The generation of f ree-standing mechanical structures such as springs, diaphragms, and cantilevers is eas ily achieved by etching through the 3C-SiC epitaxial layer to the Si using reactive ion etching and then releasing the structure


36 with wet chemical etching of Si. Etch stops can be added to the Si by p++ doping, ion implantation and activation of oxygen to create imb edded SiO2, or the use of silicon on insulator (SOI) substrates [70]. 3C-SiC can also g rown at much lower temperatures, approximately 1200 C, which is much lower than the hexagonal forms of SiC and can allow integration with multiple material devices [7 1]. This material has many properties that make it ide al as a construction material for a neural interface device. SiC is much harder than Si and would offer the same strength with less material. This would mean that the ~ 15 m thicknesses limitation for planar electrode shanks constructed from Si could be drast ically reduced to a size limited by the amount of electrical components on the device [3]. As this material is a semiconductor and chemically resistant, interaction with neurons could be achieved through field effect devices instead of the electrode thereby ensuring c apacitive interaction and reducing or eliminating metal contacting the body. Electricall y, 3C-SiC possesses a wider band gap greater than twice that of Si and the highest hole mobility of the SiC polytypes which enables the production of field effect devices that have superior inversion characteristics and greater reliability than other SiC polytypes [7 0, 71]. 3C-SiC forms heterojunctions with Si and diamond which can be used to create bip olar transistors or MISFET devices with superior reliability and lower on-state voltag es. As stated previously, generation of complex interconnected circuits would be required f or an implantable neural interface, and 3C-SiC, like other forms of SiC, possesses the same doping limitations. However, as 3C-SiC is linked physically to Si through a heteroj unction, the integration of the two materials electrically would only require the addit ion of conductive interconnects. This


37 would mean that the complex devices could still be generated in the Si and the neural device would be constructed on the interconnected t hin film layer of 3C-SiC. 1.6 Nanocrystalline Diamond (NCD) One problem with the use of FET devices is that th ey require an insulator to function. Silicon dioxide, SiO2, has been the traditional insulator of choice as i t is the native oxide of the dominant semiconductor Si. Sil icon carbide also has SiO2 as a native oxide, although the quality of the oxide is less th an for Si [21]. This insulator has a large problem in that it is contaminated easily with meta llic ions, like sodium, Na, potassium, K, and iron, Fe [12, 73-75]. Within minutes from b eing immersed in an electrolytic solution containing solvated ions like Na or K, the mobile ions are absorbed onto the surface of SiO2 and begin to diffuse into the material [73, 74]. The speed of the diffusion is increased when an electric field is applied, and the ions are transported to the semiconductor surface where they are trapped [75]. Na or K ions trapped at the intersection between the semiconductor and the oxid e attract electrons and make devices turn on prematurely and change the threshold voltag e [76]. This problem would be even greater for a device that had no metal protection o ver the gate oxide, like one of Fromherz neural interaction FET’s, as the oxide is exposed constantly to an electrolytic environment. Coating the silicon dioxide with anot her material that resists diffusion, for instance polyimide or phosphosilicate glass, is a p ossible solution to this problem. Unfortunately addition of materials can modify capa citance and Fromherz has shown that thin, high K materials are most desirable for the d evices [15]. Also, it has been shown


38 that many types of glass do not possess a high leve l of biocompatible and can initiate foreign body immune response [77]. Table 1.2: Electrical, thermal and physical proper ties of diamond and SiO 2 [78-82]. Diamond SiO 2 Band Gap (eV) 5.47 ~ 9.0 Breakdown Field (MV cm-1) 10 60 Resistivity ( W cm) 10 3 to 10 14 10 13 Relative Dielectric Constant 5.68 3.9 Thermal conductivity (W/cm-K) 20 – 150max 0.014 Thermal expansion @300 K 0.9 5x10-7 Young’s Modulus (GPa) 110 14.4 – 18 Diamond has been drawing interest for many applica tions due to its chemical, mechanical, and electrical properties. As a semico nductor that is composed of only carbon, an element common to biology, this material has been shown to have good biocompatibility [78]. As an extreme wide band gap semiconductor, diamond has the possibility of being both a good insulator and a go od conductor depending on doping levels. Many of the properties of diamond are disp layed in Table 1.2 with silicon added as a contrast. Diamond has the best thermal proper ties of any material and it provides superior material strength being the hardest substa nce known to man [78, 79]. Diamond, unlike silicon dioxide, is virtually chemically ine rt would not suffer the trap incorporation by ionic diffusion experienced by silicon oxide whe n placed in an electrolyte, so it could be utilized as a possible insulator in a FET device Single crystalline diamond is expensive and difficult to produce, but nanocrystal line diamond, NCD, can be deposited


39 on many materials using plasma assisted chemical va por deposition at relatively low temperatures, approximately 700 C [78]. This would be a good insulator for a MISFE T device that would include NCD as the insulator and 3C-SiC as the substrate base. Although NCD provides potential solutions as an in sulator, its hardness and chemical resistivity make it difficult to process i nto devices. One method of etching diamond is to use oxygen plasma which also provides some selectivity as this would oxidize the SiC substrate which could be removed wi th HF [83]. The wide band gap of diamond makes it difficult to produce low resistanc e Ohmic contacts with metals, but this problem is not applicable to insulating NCD used fo r a Fromherz FET device which has a metal free gate [84]. To achieve a surface suitabl e for biological devices, the nucleation density must be greater than 1011 cm-2 to ensure that pinholes in the film are minimized and a low surface roughness is achieved [78]. 1.7 Summary of Chapter Damage to the CNS initiates the process known as g lial scarring that essentially isolates and encapsulates the damaged area with a t hick layer of cells (tissue). The effect of this is often the loss of motor or sensory abili ties, and without the prospect of regeneration, this loss is permanent. One proposed therapy is to utilize a BMI device which would read signals from undamaged portions of the CNS, process the signal, and relay the signal to either a mechanical or computer ized device or even to another undamaged neuron or electrically active cell. Two types of the devices have been developed. The first is a non-invasive device whic h reads signals through the scalp or


40 skin, but the reactions of these devices are slow a nd prone to error. Invasive, implantable devices have been developed that can use either con ductive or capacitive coupling to the neural membrane to receive and send electrical stim ulations called action potentials. These devices are much faster and targetable, but h ave a major biocompatibility issue that leads to glial scar encapsulation of the device. W e propose that the two materials, 3CSiC and NCD, may be possible biocompatible material platforms for the construction of an implantable neural device. In the next chapter, we will explore various methods of growth and deposition that were developed to achiev e this type of device.


41 Chapter 2: Cubic Silicon Carbide and Nanocrystalline Diamond M aterial Development 2.1 Introduction The construction of planar neural prosthetic probe s requires the formation of sharp shanks through processes which require sacrif icing large amounts of bulk material, or the epitaxial growth of a material upon bulk waf er materials and the subsequent removal of the thin film. Cubic Silicon carbide, a lso known as -SiC and 3C-SiC, is a thin film that can be grown epitaxially on Si throu gh chemical vapor deposition (CVD). Unfortunately, unlike other forms of SiC polytypes, mainly the hexagonal forms 4Hand 6H-SiC, material development for this particular po lytype is a work in progress and still requires more research in order to demonstrate elec tronic device viability. The reason for this situation is that the power and radio-frequenc y (RF) device industry has focused on the development of the 4H-SiC and 6H-SiC polytypes because the larger band gaps inherent with these materials was more desirable fo r their applications. Even with a lower band gap of 2.36 eV, 3C-SiC’s major strength is it is the only form of SiC that has a cubic lattice, can be directly grown on Si substr ates which are considerably less expensive than the hexagonal SiC substrates, and wo uld allow for a straight forward electrical connection between devices on the 3C-SiC film with devices constructed in the silicon substrate.


42 3C-SiC, like its hexagonal and rhombohedral counte rparts, is difficult to grow because of the high temperatures and lack of a stab le liquid form of SiC, but has an additional burden of being one of the metastable fo rms of SiC [70, 71, 85]. SiC has equilibrium states with 2H-SiC and 6H-SiC forms but not with the 3C-SiC polytype. 3CSiC is metastable over a large temperature range, f rom ~ 1000 C to temperatures greater than its sublimation point of 2700 C, which makes it difficult for single crystal epit axy [85]. Using Si substrates limits the temperature p rofile due to its melting point of 1410 C and limits many of the SiC polytype formations, b ut this substrate material adds another set of difficulties for the growth of cryst alline 3C-SiC. In 1959, W. G. Spitzer et al. was the first to document growth of 3C-SiC on Si by heating a Si substrate to 1300 C in a mixture of methane and argon [86]. This metho d of growth, known as the chemical vapor deposition (CVD) method, which was developed by Spitzer et al., is still the most widely used today for the heteroepitaxial growth of 3C-SiC on Si substrates [86-88]. The 3C-SiC heteroepitaxially grown on Si by Spitze r et al. included a wide array of crystalline defects due to differences in the at omic stacking structures and physical properties of the two materials [86]. Many of thes e crystalline defects still pose a problem to the quality of 3C-SiC grown using this m ethod. A large, 22% lattice mismatch (SiC = 0.436 nm; Si = 0.543 nm) and the 8% thermal expansion coefficie nt difference between Si and 3C-SiC leads to island gr owth formation which can generate many major planar and linear defects within the 3CSiC crystal structure [87]. A major planar defect is the stacking fault where one of th e 3 Si-C bi-layers that form 3C-SiC polytype does not follow the regular stacking order during crystal growth [90]. This


43 defect leads to the planar twin, where the stacking fault defect does not correct itself immediately but becomes the new stacking order over many crystalline layer depositions [90]. For a FCC oriented crystal, these features t end to align along the {111} direction, and can be eliminated or reduced by using that crys talline direction for growth [87]. Another planar defect common with island growth is the anti-phase domain (APD) boundary. These defects occur where growth islands deposit along the same planar orientations but deposition occurs rotated about th eir planar axis with respect to one another, and a boundary will form when the islands grow together [90]. The 22% lattice mismatch creates a major linear defect known as the edge misfit dislocation where a linear group of atoms does not align with the latti ce of the other crystal to relieve stress in the crystal [90]. This type of defect is surrounde d by areas of tensile stress in the smaller lattice, like the 3C-SiC crystal, and compressive s tress in the larger crystal lattice, like Si [90]. The temperature for the growth of 3C-SiC gives ris e to three dimensional defects, namely the interface hollow void and the ‘hillock’ defect. Within the temperature range of 1300 – 1400 C, which is close to the melting point of Si and co mmonly the optimal temperature for the growth of 3C-SiC, Si tends to d iffuse from the substrate and is either incorporated into the developing 3C-SiC film or los t due to evaporation [91-93]. The missing material, or void, can give rise to many li near and planar defects further degrading the quality of the 3C-SiC crystal [91-93] One major defect associated with the resulting interface void is the formation of de fective, polycrystalline material on the


44 edge of the interface voids [93, 94]. This defect, commonly known as the hillock defect, propagates through the crystal as it continues to g row and emerges at the surface [93-94]. Although many of these defects do not hinder the p hysical strength of the 3C-SiC crystal, they will affect the electrical performanc e of the device [72, 90]. This chapter is dedicated to detailing the work that was done to re duce the defects in 3C-SiC through novel methods, improve the surface quality of 3C-Si C, and grow NCD on 3C-SiC. (§ 2.2) is dedicated to detailing the methods and proc esses developed for the heteroepitaxial growth of 3C-SiC on Si and the use of a poly-silico n on oxide layer stack to assist in decreasing the defects caused at the interface. (§ 2.3) describes methods of SiC surface modifications using hydrogen etching which were dev eloped at the University of South Florida in collaboration with Max-Planck-Institut. Finally, (§ 2.4) will detail the deposition and characterization of NCD on 3C-SiC su bstrates. 2.2 Chemical Vapor Deposition Growth of 3C-SiC The growth of 3C-SiC on Si substrates performed at the University of South Florida (USF) occurs within a low pressure chemical vapor deposition (CVD) reactor called MF2 which is based off the design of Kordina et al. [94, 95]. The MF2 reactor is constructed with the reaction tube in a horizontal position which allows gasses to flow laterally over the substrate. Susceptor (and thus substrate) heating is provided through magnetic induction by an induction coil excited wit h high power RF current produced at a frequency of 10 KHz. The harsh conditions of the reaction chamber and the high temperatures require an infrared pyrometer for the measurement of the susceptor (and


45 thus growth) temperature. The hot-wall reaction ch amber is an amorphous SiC-coated graphite susceptor which is not actively cooled and instead supported carbon graphite foam insulation to maximize heating efficiency [95] The reactor nominally runs at pressures from 50 – 400 Torr under the flow of a pa lladium purified hydrogen carrier gas, H2, which transports the silane, SiH4, and propane, C3H8, precursor reactants to the growth zone [95]. Nitrogen, N2, hydrogen chloride, HCl, and methyl chloride, CH3Cl, gasses are also available for growth processing, wh ile argon is used as the purge and cooling gas [94]. 2.2.1 Establishing a Low Pressure Process for CVD 3C-SiC Growth Thin layers of high quality crystalline 3C-SiC are required for the generation of a neural prosthetic device, so our material research focused on the goal of improving the crystalline quality of 3C-SiC grown on Si by improv ement of the carbonization process. To achieve this purpose we examined growth of 3C-Si C on different orientations of the Si crystal, namely (100), (111) and (110), under bo th atmospheric and low process pressures [96]. Prior to 2007, the carbonization s tage of the epitaxial growth of 3C-SiC was conducted at 760 Torr which was followed by gro wth at low pressure (~ 100 Torr) [22]. Apart from the obvious hazards of performing this CVD process at atmospheric pressure, some improvements reported in the literat ure justified the investigation of lower pressure for carbonization [89, 91-93]. At atmosph eric pressure, the elevated partial pressure slows the evaporation of Si as the tempera ture is increased, decreasing the generation of interface voids, however the voids ar e much larger than found at lower


46 pressures as the growing 3C-SiC parasitically remov es more material from the substrate [92]. Although void generation is much greater at lower pressures as evaporation tries to establish a saturated vapor pressure, the voids are smaller than those generated at atmospheric pressure [92]. The slower removal of r eaction products at atmospheric pressure increases the generation of defective mate rials which deposit onto the surface of the crystal and become hillock defects, a defect th at is rarer at lower pressures [92]. Atmospheric carbonization allows for enhanced three dimensional growth, otherwise referred to as island nucleation that was covered i n (§ 2.1), whereas lower pressure allows for more two dimensional growth, known as step flow growth where the growth occurs in cohesive layers across the entire growth surface (p lane) [89]. Island growth produces a larger quantity of stacking faults, twins, and disl ocation defects which only begin decreasing in number with thick films greater than 17 m, whereas lower pressure carbonization and growth decreases the planar and l inear defects and reduction occurs within 5 m of the growth films [89]. Our group at USF uses a modification of the origin al three stage process created by Nishino et al. in 1987 which consisted of a separate surface etch process followed by a single carbonization and growth stage [88]. Figure 2.1 shows the current growth process developed at USF for the growth of 3C-SiC on Si sub strates. Removal of the native oxide is critical as 3C-SiC will not develop on any oxidized surface [86-88]. Nishino’s process used an initial HCl etching stage to remove the oxide and clean the surface of the Si substrate, but we have discovered that we could combine this stage with the carbonization stage without affecting crystal growt h [22, 88].


47 Figure 2.1: A generalized graph displaying the gen eral process schedule developed at USF for the growth of 3C-SiC on Si substrates. Two distinct stages are indicated, namely the carbonization and the growth stage. Gas flow, temperatures, time, and pressure for each stage are indicated. The USF process begins by cleaning the Si substrat e utilizing the well known RCA process with hydrofluoric acid, HF, immersion t o remove adsorbed surface metals, organics, and the defect filled native oxide. The carbonization stage begins by evacuating the reaction chamber of gasses and refil ling the chamber with H2 to a pressure of 400 Torr [96]. The RF induction is started and the substrate is heated under a flow of H2 and C3H8 [96]. Unlike the runs at atmospheric pressure, th e carbon precursor had to be drastically increased to maintain the proper par tial pressure above the substrate at the decreased carbonization pressure [96]. The etching provided by H2 in the initial heating 3CSiC Growth Cool 10 slm 120 slm 10 40 sccm ~45 35 slm 1 120 ~10 2-5 ~20 10% SiH 4 Ar Other Parameters: Carbonization: TC – 1135 C/ 100 – 400 Torr chamber pressure Growth: T G 1380 C/ 100 Torr chamber pressure 16 8 sccm 8sccm t (min) 16 sccm 10 slm H 2 C 3 H 8 T G T C T ( C) Carbonization


48 and the increased temperatures allow for the remova l of the few nanometers of oxide that have formed since the wafer chemical cleaning. Hea ting ramp speed can be modified by increasing or decreasing the power of the RF genera tor, but the normal time for this ramp is 20 minutes [95]. When the carbonization tempera ture is reached, which is normally 1135 C, gas flow, pressure and temperature are maintaine d for two to five minutes [22, 95, 96]. The carbonization temperature, which is m uch lower than that for the optimal growth, is optimized to prevent the evaporation of the Si surface but high enough to allow C3H8 to undergo sufficient pyrolysis (i.e., molecular c racking). Propane pyrolysis produces reactive carbon species that travel to the surface of the Si substrate, react with the Si atoms there, and forms the initial layer of 3C-SiC [22, 86, 95, 96]. After the carbonization step has been completed th e temperature is increased to the growth temperature, nominally around 1380 C. Our new process deviated from the previous process in a few aspects. Although SiH4 is still added to the reactor and the flow is slowly increased during the temperature ram p from the carbonization to growth temperature, our modification additionally regulate s propane by slowly decreasing the flow as the temperature increases [95, 96]. Our ra tionale for this change is that propane requires much more energy to pyrolyze than other ca rbon containing gasses. The higher flow required at the lower carbonization temperatur e which maintains the proper partial pressures and carbon molar species will produce exc ess carbon species as the temperature approaches the growth temperature. The excess carb on does not have enough silicon to react with, as Si is the rate limiting species of t he reaction, and will form C-C bonds to create graphitic carbon. Graphitic carbon becomes particulates which deposit on the


49 surface of the growing 3C-SiC crystal. The defects become incorporated within the crystal unless the flow, and therefore the amount o f available carbon, is reduced [95, 96]. The addition of silane and subsequent flow is incre ased to provide enough Si to react with the increase in carbon species. Void defects at the interface which are not coated by the growing 3C-SiC film become a source of Si from the substrate, as 3C-SiC does not allow diffusion of molecules [22, 95, 96]. The Si for the reactions in the second ramp would be parasitically provided from the void defec ts discussed in (§ 2.1) without the additional Si provided by SiH4 [22, 95, 96]. At 30 50 C before the growth temperature, the SiH4 flow is accelerated quickly to the final level req uired for optimal growth, the H2 precursor flow is increased to push the reactive sp ecies to the optimal growth zone located over the substrate, and the pressure is fur ther reduced to 100 Torr [22, 95, 96]. At growth temperature, optimally 1380 C, the gas flow rates and reactor temperature remain constant for 1 to 120 minutes, where the lat ter is a limitation due to the exhaustion of the H2 carrier gas bottle. At the end of the growth time all process and carrier gas flows are ceased and replaced by Ar and the RF gene rator is powered down [22, 95, 96]. The Ar gas serves as a cooling mechanism by pulling heat away from the hot zone, and the cooling process is inherently slow due to the h eat retention by the hot wall carbon foam insulation. The slow cool down is desirable a s the thermal expansion coefficient mismatch will cause stress between the films if the temperature is lowered quickly resulting in film cracking.


50 2.2.2 Low to Atmospheric Pressure Carbonization Compariso n The epitaxial growth process for 3C-SiC detailed i n (§ 2.4) was used for film growth on Si substrates with three different orient ations: (100), (111), and (110). The quality of the films was compared for the carboniza tion stage using a PSIA XE-100 scanning atomic force microscope, AFM, and X-ray ph otoelectron spectroscopy, XPS. (a) (b) (c) (d) (e) (f) Figure 2.2: AFM micrographs of 3C-SiC on Si surfac es after the carbonization growth stage. (a-c) were performed at atmospheric pressur e and (d-f) were performed at 400 Torr pressure. 3C-SiC samples (a) and (d) were gro wn on Si (100), (b) and (e) on Si (111), and (c) and (f) on Si (110). Surface roughn ess root mean square (RMS) values are located in the lower right corner. AFM performed i n tapping mode with Al coated silicon cantilevers (2 x 2 m). 0.10 nm RMS 0.075 nm RMS 0.12 nm RMS 0.22nm RMS 0.094 nm RMS 0.22 nm RMS


51 The growth stage was characterized using AFM and t ransmission electron microscopy, TEM. Figure 2.2 shows selected AFM mic rographs from the carbonization stage of the process. The upper row displays the r esults of atmospheric carbonization on Si (100), Si (111), and Si (110) from left to right and the lower row shows scans for the same substrates at the lower 400 Torr pressure. Al though the surfaces all have a similar surface roughness, rq, immediately noticeable is that the three substrat es have ~250 nm wide voids which are squared on the (100)Si and (11 0)Si surfaces and triangular on the (111)Si surface. The AFM micrographs of the low pr essure carbonized (100) surfaces do not appear to have large void features, but are als o not devoid of surface features. Small voids on the order of 25 – 50 nm are present on the surface and are difficult to see with these scans due to their shallow depth. The (111) and (110) substrates did not display a large change from carbonization between atmospheric and low pressure. Table 2.1: XPS data of 3C-SiC films grown on vario us Si surfaces after 400 Torr carbonization. Binding Energy (eV) (100)Si (111)Si (110)Si C 1s Survey 284.5 67.9% 42.2% 44.2% SiC (carbide) 283.5 15.8% 21.1% 18.1% Graphite 284.6 26.8% 14.5% 12.7% C-O complex 285-287 25.3% 6.6% 5.6% Si 2p3 Survey 103.3 16.1% 31.3% 28.3% SiC (carbide) 101.4 15.0% 23.3% 20.9% Silica 103.4 1.1% 5.8% 6.0% Si 99.7 --2.2% 1.4% O 1s Survey --16.0% 26.5% 27.5% Si/C ratio --0.95 1.10 1.15


52 XPS utilizing an aluminum source was used to gathe r the data shown in Table 2.1able 2.1. All samples show peaks associated wit h silicon carbide: C 1s, Si 2p3, and O 1s. The C 1s peak can be broken down into 3 compon ents, SiC, graphite, and C-O complexes, and the Si 2p3 peak is composed of SiC, silica, and Si. The O 1s peak is associated with oxides. The table suggests that th e 3C-SiC film grown on (100)Si contains much more graphite and carbon-oxygen compl exes on the surface as compared to the other surfaces which contain much less quant ities of these species. The (111) and (100) surfaces reveal a greater percentage of surfa ce atoms in SiC bonds, but also have considerably more silica and Si remaining on the su rface. As voids provide a pathway to the Si substrate, this result and the AFM micrograp hs confirm the presence of more voids on the surface. The (100) orientation displayed no appreciable value for natural silica on the surface and may indicate that the entire surfac e is covered in SiC or other carbon species. Finally, the oxide percentage is larger o n the (110) and (111) surfaces, than on the (100) surface. The XPS survey values were util ized to determine the optimal silicon to carbon ratio, Si/C, at the lower pressures. The ratio for the 3C-SiC on (100)Si is the same as utilized in the atmospheric growth process, and worked almost as successfully for subsequent growths using the low pressure proce ss obtaining a growth rate of 15 m/ hr. Later experimentation for the growth on (111)S i substrates show that the Si/C ratio of 1.10 is optimal for the growth on this substrate as well. A good process for growth on (110) Si wafers had never been developed before at USF. Figure 2.3 shows AFM micrographs of the surfaces o f the three Si substrates after one minute of processing at the growth stage. The (100) and (111) surfaces show an rq


53 (a) (b) (c) (d) (e) (f) Figure 2.3: AFM micrographs of 3C-SiC on Si surfac es after 1 minute of the growth stage. All growths were performed at 100 Torr foll owing carbonization of 760 Torr for (a-c) and 400 Torr for (d-f). 3C-SiC samples (a) a nd (d) were grown on Si (100), (b) and (e) on Si (111), and (c) and (f) on Si (110). Surf ace roughness root mean square (RMS) values are located in the lower right corner. AFM performed in tapping mode with Al coated silicon cantilevers (2 x 2 m). which is considerably less than that for the sample s grown at low pressure; however the surface morphology does not differ much between the growths from the two carbonization pressures. The (100) 3C-SiC shows a surface with a square mosaic growth pattern accompanied with the crack-like antiphase b oundaries arising from anti-phase domains in the crystal. The (111) 3C-SiC shows the beginning of the regular triangular step pattern with edges bounded by the {110} direct ion. Both films grown on (110) substrates show groups of rounded clusters indicati ng that the film is not single 0.40 nm RMS 0.33 nm RMS 17.61 nm RMS 1.51 nm RMS 1.83 nm RMS 8.84 nm RMS


54 crystalline 3C-SiC. As indicated by the XPS, the g rowth was performed at a Si/C ratio of 0.94, which is not the optimal ratio for 3C-SiC gro wth on Si (110). Although the AFM micrographs do not indicate many differences between the two processes, the TEM analysis confirms that the d efect levels are decreased with lower pressure carbonization processes, as indicated by t he literature in (§ 2.1). The plan view TEM micrograph shown in Figure 2.4(a) is for 3C-SiC grown using the atmospheric pressure carbonization process and displays very la rge and deep voids at the interface and the accompanying problem of increased defects in th e area surrounding the voids. The large dark area above the voids indicates a highly defective crystal there and could indicate hillock feature formation. Planar and lin ear defects, which are abundant at the interface, continue extending through the crystal w ithout a large level of reduction. On the other hand, the 3C-SiC crystal grown on (100)Si using the lower pressure carbonization process shows a different interface a nd defect concentration. Voids are much smaller at the interface, and large area defec tive material is not present around the void areas. The planar and linear defects are redu ced in concentration much faster, many within 100 nm, generating larger areas of almost de fect free material. Finally, the concentration of defects is confirmed by the select ed area diffraction images. Both images show that the films are 3C-SiC, but the film grown with the atmospheric carbonization process is more diffuse, indicating s ome scattering effects from defects, while the image obtained for the low pressure carbo nization process is very sharp and clear, indicating a higher quality crystal had been grown.


55 (a) (b) Figure 2.4: TEM plan view micrographs displaying g rowth of 3C-SiC on (100)Si using carbonization at (a) atmospheric pressure, and (b) 400 Torr. Interface voids on each substrate are circled in red. Selected area diffra ction patterns of the crystal structures located to the right of the plan view micrographs s how that the films are 3C-SiC, but that the growth associated with atmospheric pressure car bonization is more defective as evidenced by the more diffuse diffraction spots. T EM courtesy of Dr. Dr. C. Bongiorno of IMM-CNR in Catania, IT. The low pressure carbonization process has led to a decreased amount of defects within the thin films of 3C-SiC grown on Si substra tes. Crystal quality from the epitaxial growth using atmospheric carbonization, as measured by Phillips X’Pert Panalytical Xray diffractometer, XRD, had the lowest full width of half maximum (FWHM) value reported at the time. The X-ray rocking curve take n on the SiC (002) diffraction plane


56 had a FWHM of 360 arcsec [22]. Using modifications of the low pressure carbonization developed by Frewin et al. C. Locke et al has reduced the defect level further and has reported the lowest FWHM value for any 3C-SiC film presently of 219 arcsec [96, 97]. 2.2.3 Growth of 3C-SiC on Poly-Si/ SiO2/ (111)Si Growing a high quality thin film of 3C-SiC is only one of the necessary steps in the development of a planar neural interface system The necessary electronic devices must be incorporated into the 3C-SiC film, and thes e electronics are constructed on the sharp needle-like shank probes that will interface with the neurons in the CNS system. SiC’s resistance to wet chemical etching requires t he use of dry etching through deep reactive ion etching/ reactive ion etching (DRIE/RI E) to achieve the release of these neuronal shanks [98]. Utilizing the Si substrate a s a sacrificial layer for the creation of freestanding 3C-SiC structures is one common method to release the 3C-SiC layer [99101]. We have worked on the generation of cantilev ers constructed from 3C-SiC which were extremely resilient when released from the Si substrate, displaying only 9 m bow (bending) over a 700 m long, 20 m wide and 2 m t hick cantilever [100]. Unfortunately, there were many problems associated with the controlled release of the cantilevers as many remained connected to unetched Si material below the structure [99, 100]. Earlier, our group had looked into forming o ther mechanical devices using 3C-SiC grown into Si molds, but epitaxial growth at reason able growth rates within the mold and sidewalls would have eventually led to keyhole gene ration [101]. A further complication to this process is that recipes used to etch Si wit h DRIE/RIE have a similar etch rate with


57 SiC, thereby excluding selectivity and reducing acc uracy for the desired structure [98, 102]. SiO2 has been traditionally used as a release layer for Si MEMS technology and also serves as an etch stop for processing involvin g DRIE/RIE [103]. Unfortunately SiO2 inhibits the growth of 3C-SiC thus 3C-SiC cannot be directly formed on oxide release layers (indeed SiO2 has been used as a mask for selective 3C-SiC growt h). Therefore another method is required that would allow the inc lusion of an oxide release layer for 3C-SiC MEMS structures using Si substrates [86]. S ilicon on insulator, SOI, substrates provide an excellent media for the creation of free standing 3C-SiC devices by providing not only an oxide for the etch-stop for DRIE/RIE, b ut also a single Si crystal seed layer for the heteroepitaxial growth of the 3C-SiC. SOI provides some additional benefits for the grow th of 3C-SiC as shown in previous studies [104, 105]. The high temperatures required for the growth of singlecrystal 3C-SiC softens the SiO2 layer, allowing dispersion of stress caused by the 20% lattice mismatch, and the oxide provides mechanisms for the suppression of the formation of voids caused by Si evaporation at the 3C-SiC/Si interface [104]. The work at USF has shown that SOI seed layers (>50 nm) have produced 3 C-SiC films that are as defective when compared to 3C-SiC films grown on single-cryst al Si substrates [105]. The theorized benefits of the epitaxial growth of 3C-Si C on SOI were displayed when 3C-SiC is deposited on a thin (<50 nm) seed layer of Si, a nd the final product was excellent quality 3C-SiC [104, 105]. However, a major drawba ck of using SOI for the production


58 of 3C-SiC devices is the fact that it requires exte nsive processing techniques which add to the overall production cost of the device [104, 105] For our experiments, we replaced the expensive SOI wafer with a stack of polySi/ SiO2/ (111)Si, where the poly-Si would serve as the see d layer for the growth of polySiC. After the growth, we saw that our normal poly -SiC was not produced, which has a dull grey appearance, but Newton fringes were seen which usually indicated that we had achieved mono-crystalline 3C-SiC material. The res ulting growth was nearly singlecrystalline 3C-SiC that was comparable in quality t o 3C-SiC grown on single-crystal Si in both the <111> and <100> directions. We theoriz ed that the poly-Si was sufficiently thin and possessed an abundance of a particular cry stalline orientation which would allow the 3C-SiC to completely convert the poly-Si into 3 C-SiC. The oxide beneath it would serve as a stress release layer and allow the 3C-Si C to relieve defects at the interface level. This section details our new growth process using low process pressure and temperature and the resulting characterization of t he 3C-SiC film with optical microscopy, AFM, scanning electron microscopy (SEM) XRD, and TEM. Substrate preparation for the growth experiments w as as follows. A (111)Si wafer was RCA cleaned, followed by the deposition o f 5500 of thermally deposited silicon dioxide. After oxidation, a 20 to 50-nm-thi ck film of poly-Si was deposited by LPCVD at a temperature of 610C and a pressure of 3 00 mTorr [106]. This process was chosen from the various poly-Si recipes for many re asons. The first is that a compressive stress is produced between the resulting poly-Si fi lm and the oxide layer, which should help bring the Si crystal lattice into greater comp liance with the 3C-SiC crystal lattice


59 [107]. A secondary reason for the growth of poly-S i at this temperature is that it generates large columnar Si grains textured in the <110> direction with a minor presence of grains textured in the <111> and <311> direction s [106, 107]. The resulting thin polySi film was characterized by both AFM and XRD to as certain the starting growth surface properties. The AFM micrograph displayed in Figure 2.5(a) show s a surface with grains of average area ~5.5 nm2, having an average surface roughness of 0.49 nm rm s, but also indicated the presence of pinholes in the surface. The XRD measurements were performed on a Phillips X’Pert Panalytical Diffract ometer operating at the Cu-K line, and the measurements indicated alignment of the gra ins along the <110>, <111>, and <311> directions, as was expected from the literatu re [106, 107]. The pinholes created difficulties during the deposition of 3C-SiC by cre ating a pathway for the softened glass to flow onto the growth surface thus destroying the 3C-SiC film morphology. This problem, displayed in Figure 2.5(a), existed on all growth attempts until the growth temperature was reduced to eliminate this problem, resulting in a maximum growth temperature for 3C-SiC on the film stack of 1200C. The final optimized growth process is explained as follows. The poly-Si/ SiO2/ (111)Si wafer stack was placed in the MF2 horizonta l, hot-wall reactor heated by the RF induction of a SiC-coated graphite susceptor. The wafer was loaded into a molded polySiC plate to fix the position of the wafer within t he reactor hot zone. This polyplate was then seated into a recess in the susceptor and the chamber was sealed and evacuated of residual gases. The chamber was then filled with p alladium-purified hydrogen to a


pressure of 400 Torr. The remaining process is a m odification of the same process covered in (§ 2.2.1). The pressure for the carboni zation process was 400 Torr, and growth pressure was 100 Torr. The standard gases u sed for 3C palladiumpurified hydrogen, H which is the carbon precursor; and a 10% silane (Si H the silicon precursor. (a) Figure 2.5 : (a) An 5 x 5 m AFM micrograph of the surface of the poly stack showing the pinholes in the thin film, and (b ) shows an image from optical microscope obtained at 25X displaying the 3C using a growth temperature of 1250 flow through the pinholes in the poly The carbonization stage occurred while the sample t emperatu 1000C at a rate of ~ 35 C/min. Throughout the ram p a flow of 16 sccm of C maintained with a mass flow controller (MFC), and t he H maintained at 10 slm. Once the carbonization tempe rature was reached, the te was maintained for 3 min to attempt a full conversi on of the poly 60 pressure of 400 Torr. The remaining process is a m odification of the same process covered in (§ 2.2.1). The pressure for the carboni zation process was 400 Torr, and growth pressure was 100 Torr. The standard gases u sed for 3C SiC purified hydrogen, H 2 which is used as the transport gas; propane (C which is the carbon precursor; and a 10% silane (Si H 4 ) premixed in hydrogen, which is (b) : (a) An 5 x 5 m AFM micrograph of the surface of the poly stack showing the pinholes in the thin film, and (b ) shows an image from optical microscope obtained at 25X displaying the 3C -SiC growth on the polySi substrate stack using a growth temperature of 1250 C. (b) shows the glass that became soft enough to flow through the pinholes in the poly -film and onto the surface. The carbonization stage occurred while the sample t emperatu re was ramped to 1000C at a rate of ~ 35 C/min. Throughout the ram p a flow of 16 sccm of C maintained with a mass flow controller (MFC), and t he H 2 carrier gas flow was maintained at 10 slm. Once the carbonization tempe rature was reached, the te was maintained for 3 min to attempt a full conversi on of the poly Si surface into 3C pressure of 400 Torr. The remaining process is a m odification of the same process covered in (§ 2.2.1). The pressure for the carboni zation process was 400 Torr, and SiC growth were: which is used as the transport gas; propane (C 3H8), ) premixed in hydrogen, which is Si/ SiO2/ (111)Si stack showing the pinholes in the thin film, and (b ) shows an image from optical Si substrate stack C. (b) shows the glass that became soft enough to re was ramped to 1000C at a rate of ~ 35 C/min. Throughout the ram p a flow of 16 sccm of C 3H8 was carrier gas flow was maintained at 10 slm. Once the carbonization tempe rature was reached, the te mperature Si surface into 3C -SiC.


61 After carbonization and creation of the 3C-SiC temp late layer, the temperature was ramped a second time at a rate of 35C/min to the g rowth temperature of 1200C. During this ramp, we decreased the flow of C3H8 while simultaneously introducing and increasing the flow of SiH4. At the growth temperature the input gas silicon to carbon ratio, Si/C, for the growth stage was 1.2. H2 flow was maintained at 10 slm until 30C before the ramp was completed, where it was increas ed to 25 slm, and the pressure was reduced from 400 Torr to 100 Torr. The temperature and gas flows were then held constant, allowing the continued epitaxial growth o f 3C-SiC on the carbonized poly-Si template layer. The lower temperature process param eters produced a growth rate of 3.0 m/ h near the front (upstream side) of the wafer, and 2.5 m/ h at the rear (downstream side) of the wafer due to precursor depletion and t he lack of in-situ wafer rotation. The growth rate was measured using an Accent QS-1200 FT IR system. Using the same low temperature and pressure process, the growth rate w as also verified using samples of 3CSiC grown on single-crystal Si oriented in the <100 > and <111> directions. 2.2.4 Characterization of the 3C-SiC on the Poly-Si Stack AFM surface analysis was used to further character ize the films as shown in Figure 2.6. The growth of 3C-SiC on the poly-Si/ S iO2/ (111)Si substrate was compared with 3C-SiC grown directly on (100) and (111)Si sub strates. The morphology for the surface of the 3C-SiC on the poly-Si stack is simil ar to that of the 3C-SiC grown on Si (111), showing large triangular domains with bounda ries created by stacking fault and grain boundary defects. The AFM micrograph on the 3C-SiC grown on (100)Si has


62 smaller and more disassociated island growth, posse ssing square grain morphology. Improvements to the lower pressure process made by Locke et al improved the quality of the 3C-SiC grown on (111)Si at the lower 1200 C temperature [108]. The AFM data of this process show that 3C-SiC on (111)Si morphologi cally resembles 3C-SiC growth performed at 1380 C, but Figure 2.6(a) resembles the morphology obtai ned with Figure 2.6(e) unfortunately indicating that the process is not single crystalline. (a) (b) (c) (d) (e) Figure 2.6: AFM micrographs showing a comparison o f 3C-SiC grown on the surfaces of (a) poly-Si/ SiO2/ (111)Si, (b) (111)Si, and (100)Si. Data taken in contact mode using a silicon nitride tip and a scan area of 5 m by 5 m. The average surface roughness, rq, of the samples is: (a) 17.9 nm rms; (b) 26.5 nm rms ; (c) 6.57 nm rms. AFM micrographs of (d) 3C-SiC on (111)Si and (e) 3C-SiC growth on p -Si/ SiO2/ (111)Si utilizing the improved lower temperature process developed by Loc ke et al. [108]. .Note that the grown film on the poly-Si substrate stack exhibits triangular surface morphology typical for 3C-SiC on (111)Si films. Growth of films (d) a nd (e) is courtesy of C. Locke, USF SiC group.


63 A cross-section SEM micrograph, obtained from a Hi tachi 4800 microscope, displays the growth of 3C-SiC on the poly-Si stack is shown in Figure 2.7. This crosssection micrograph shows that the thickness of the 3C-SiC film grown for 20 min on the poly-Si stack is ~1.3 m, which supported the film thickness measurements made by FTIR. An important aspect of this growth process i s that the oxide remained perfectly intact and appeared to be unaffected during the gro wth of the 3C-SiC. Figure 2.7: Cross-section SEM micrograph of the 3C -SiC film on the poly-Si/ SiO2/ (111)Si stack (20 min growth time). The 3C-SiC gro wn has a thickness of ~1.3 m and the SiO2 a thickness of 0.55 m. The thickness of the poly-Si, measured by a Rud olph AutoELIII ellipsometer (prior to 3C-SiC growth), wa s 20 nm. Note that the growth of 3C-SiC on the thin poly-Si seed did not affect the oxide layer. Figure 2.8 shows the XRD -2 diffraction surveys and high-resolution rocking curves performed for the 3C-SiC films grown on poly -Si/ SiO2/ (111)Si, (111)Si and (100)Si substrates, and the resulting data is utili zed for determination of the crystal orientation and quality of the 3C-SiC layer. Two s cans were obtained for the 3C-SiC film grown on the poly-Si/ SiO2/ (111)Si stack. The first scan, shown in Figure 2 .8(a), is


64 from the film of the wafer grown near the gas inlet of the reactor where the film had a cloudy appearance. This film shows a strong peak o bserved at 35.6 and a weak peak at ~71.894. The peaks indicate the reflections from t he (111)3C-SiC and (311)3C-SiC planes, respectively. A comparison of the relative peak intensities suggests a preference for grain alignment in the <111> direction, little growth oriented along the <311> direction grains, and an absence of growth along th e <110> direction. Figure 2.8(b) shows a survey from the downstream h alf of the 50 mm wafer. This wafer has only the peak observed at 35.6 indicatin g the reflections are from the (111)3CSiC oriented grains. This film has no appreciable reflections from the <311> and <110> grains. The surveys of the 3C-SiC film grown on (1 11)Si and (100)Si, displayed in Figure 2.8(c) and Figure 2.8(d) respectively, show dominant peaks at 35.6 for the (111)3C-SiC and 41.4 for (100)3C-SiC which indicat e monocrystalline film growth. The rocking curves were taken at the primary Bragg peak of the 3C-SiC epitaxial films and can be used to determine general film cry stallinity. The insets displayed in Figure 2.8 show the results of the rocking curves o btained for each substrate type. The rocking curve for the 3C-SiC films on the poly-Si/ SiO2/ (111)Si substrate displayed a FWHM of 0.192 (691) upstream and 0.115 (414) downstream. The FWHM evaluated from the 3C-SiC on (111)Si was 0.134 (482). The rocking curve for the 3C-SiC on (100)Si displayed a slightly better result than the film grown on (111)Si, with a FWHM value of the 41.4 peak of 0.128 (460). The correlation of the FWHM values from the growth performed on poly-Si/ SiO2 / (111)Si versus the growth performed on singlecrystal Si appears to suggest that all films have r elatively comparable crystallinity. This


65 (a) (b) (c) (d) Figure 2.8: XRD -2 diffraction surveys for the 3C-SiC films grown on (a-b) poly-Si/ SiO2/ (111)Si, (c) (111)Si, and (d) (100)Si substrates. The XRD -2 diffraction surveys show that (a-c) possess a primary peak at 35.6, an d (d) possesses a primary peak at 41.4. Insets: rocking curves for each of the 3C-Si C films taken at their respective primary Bragg peaks. The FWHM values are 0.192 (6 91), 0.115 (414), 0.134 (482) and 0.128 (460), respectively. XRD courtesy of C. Locke, USF SiC group. correlation proves to be very interesting because t he growth of the 3C-SiC film on the poly-Si/ SiO2 / (111)Si stack began on a poly-Si seed layer. Whe n compared to 3C-SiC (311)SiC 71.894 (111)SiC 34.649 FWHM 0.192 FWHM 0.115 (111)SiC 34.649 (111)SiC 34.649 (222)SiC 75.448 FWHM 0.134 (400)SiC 90.063 (200)SiC 41.430 FWHM 0.128


66 films grown on SOI of similar Si seed layer thickne ss (50 nm) and 0.5 m of SiO2, the reported FWHM value was 0.20 (720) [104],which is almost double the FWHM of the 3C-SiC grown on the poly-Si seed reported in this w ork. One aspect of the XRD data opened up some questions about the results obtained The peak intensities of the 3C-SiC film grown on the poly-Si/ SiO2/ (111)Si substrate had unusually low counts as com pared to the 3C-SiC films grown on single crystal Si subs trates. TEM was consulted to examine the films in more detail so as to try and u nderstand the origin of this reduced XRD signal intensity. TEM analysis performed at IMM-CNR in Catania, Ital y by C. Bongiorno is displayed in Figure 2.9. The selected area electro n diffraction pattern shown in Figure 2.9(a) does not show the clear 3C-SiC diffraction p attern displayed in Figure 2.4(b), but shows multiple rings which is indicative of polycry stalline material. The bright field plan-view of Figure 2.9(b) indicates a large amount of different grains present at the surface of the sample. TEM indicates that the samp les are actually polycrystalline in nature, and explains the results of the AFM surface scan displayed in Figure 2.5(e). The data from all of the analysis made seems to in dicate that the growth of 3CSiC on the poly-Si/ SiO2/ (111)Si substrate is not single crystalline as fi rst thought, but polycrystalline. However, the results of growth on this substrate suggest some interesting dynamics that should be explored at a later date. First, the poly-Si utilized for this growth had an overwhelming amount of grains along the <110 > direction, with only minor grains along the <111> and <311> directions, however the g rowth was almost exclusively observed to be along the <111> direction and posses sed no <110> grains as indicated by


AFM and XRD. Even though TEM data indicates the film is polycrys talline, the XRD indicates a weak single peak for (111)3C that the {111} family when oriented about a fixed a xis has eight different directional positions. The 3CSiC crystals growing on the surface may align along any of these directions which explains the TEM bright field view and the AFM surface scan which show a surface full of many different individual cr ystal grains. (a) Figure 2.9 : TEM data taken in plan SiC film grown on the 20 nm thick poly area diffraction and (b) plan is polycrystalline in microstructure. Note this is not an issue for the MEMS application but for possible use in electronic devices. Data pr ovided by C. Bongiorno, IMM Catania, Sicily (IT). In summary, a s ingle crystalline film of 3C poly-Si/ SiO2/ (111)Si wafer and the process results verified multiple tim es. This process was developed to create a true 3C layer is co mpletely converted into SiC for use in MEMS applica tions. The resulting films 67 Even though TEM data indicates the film is polycrys talline, the XRD indicates a weak single peak for (111)3C SiC. This result may be explained by the fact that the {111} family when oriented about a fixed a xis has eight different directional SiC crystals growing on the surface may align along any of these directions which explains the TEM bright field view and the AFM surface scan which show a surface full of many different individual cr ystal grains. (b) : TEM data taken in plan -view from sample USF2-07170B, which is a 3C SiC film grown on the 20 nm thick poly Si substrate stack. The resulting (a) selected area diffraction and (b) plan -view are displaye d. Unfortunately this data shows the film is polycrystalline in microstructure. Note this is not an issue for the MEMS application but for possible use in electronic devices. Data pr ovided by C. Bongiorno, IMM ingle crystalline film of 3C SiC has been successfully grown on a wafer and the process results verified multiple tim es. This process was developed to create a true 3C SiC on oxide, or ‘SiCOI’ process, where the thin se ed mpletely converted into SiC for use in MEMS applica tions. The resulting films Even though TEM data indicates the film is polycrys talline, the XRD SiC. This result may be explained by the fact that the {111} family when oriented about a fixed a xis has eight different directional SiC crystals growing on the surface may align along any of these directions which explains the TEM bright field view and the AFM surface scan which 170B, which is a 3C Si substrate stack. The resulting (a) selected d. Unfortunately this data shows the film is polycrystalline in microstructure. Note this is not an issue for the MEMS application but for possible use in electronic devices. Data pr ovided by C. Bongiorno, IMM -CNR, SiC has been successfully grown on a wafer and the process results verified multiple tim es. This process SiC on oxide, or ‘SiCOI’ process, where the thin se ed mpletely converted into SiC for use in MEMS applica tions. The resulting films


68 were thought to be single crystalline in structure and their quality assessed using AFM, SEM, and XRD analysis. The analysis suggested that the films were single crystalline, but the addition of data from TEM analysis showed t he resulting films were polycrystalline. Although this process will not al low for the incorporation of the electronics required for an implantable neural inte rface which utilizes field effect devices, we believe this process will have great importance for implantable electrode MEMS applications utilizing the SiC material solely for mechanical purposes. The costeffectiveness and relative ease for the deposition of the oxide, poly-Si, and even poly-SiC layer make this process superior to the methods tha t utilize fabricated SOI substrates. Indeed if the neuronal implant electronics are real ized within the Si substrate itself, then there appear to be no drawbacks to using this techn ology to create a fully-functional BMI device in the near future. 2.3 Hydrogen Etching of SiC Polytypes Previously (§ 2.2) new techniques utilizing a CVD reactor for the epitaxial growth of 3C-SiC on Si substrates was discussed. These gr owth techniques were focused in generating a superior material for the fashioning o f MEMS devices. 3C-SiC is an excellent material for this application due to its superior material properties, but as a semiconductor, it offers the bonus of being an ‘all in one’ MEMS material. Unlike other semiconductor materials, like Si, this material has only been researched scientifically within the last 50 years. This section is dedicate d to the improvement of this extraordinary material for electrical devices.


69 An important consideration for electronic devices is the maintenance of the integrity of the material during processing so as t o gain optimal yield from the finished device [109]. The surface of the semiconductor pla ys an integral role in device statistics, as it is the interconnection point of a device in c ontact with other materials, like epitaxial layers or metal conductors required for the device. Defects present at the surface must be reduced or eliminated for maintenance of electrical integrity and to avoid electrical traps generation, and atomically flat surfaces present a surface with available bonds ready for material integration [109]. However, these danglin g bonds actively want to react with other materials present in the environment, which a dds parasitic charge to the surface/interface or potentially traps defective ma terial within layers of the device during fabrication and decreases the device efficiency [10 9]. 3C-SiC and Si will actively react with oxygen in the environment to create an oxide o n the surface of the semiconductor. Surface passivation is a technique that places anot her atom or molecule on these dangling surface bonds to prevent oxide formation thus leavi ng the surface pristine for subsequent layer formation [109]. More importantly for this w ork, surface passivation has been a method used to attach complex organic molecules to the surface of semiconductors [20]. Crystals grown using the methods discussed in (§ 2 .2) show mesa or terraced structures that are not completely flat. The atomi c steps are produced by the alternating layers of carbon and silicon molecules on the surfa ce. One method utilized by the semiconductor industry to maintain the flat surface required for integration of layers within a multiple layer device is chemical-mechanic al polishing. CMP uses mechanical grinding, usually with diamond grit, alongside a ch emical etching to equalize the levels


70 of features on a surface. Polishing has been used for the hexagonal forms of SiC for many years to achieve a mirror like surface, but as there is no wet chemical etchant for SiC, the surfaces remain scratched and highly defec tive. One method of removing polishing damage and creating an atomically flat, w ell ordered and passivated surface suitable for device fabrication is to etch the surf ace of the semiconductor using hydrogen gas etching [110]. Our group has also recently dis covered that (100)3C-SiC can be passivated for long periods of time using hydrogen etching [111]. This section presents a comprehensive etching stud y, part of a collaborative effort between the University of South Florida and the Max -Planck-Institut in Stuttgart, Germany, that covers the three major polytypes of S iC. This work examines the previous techniques developed for 4Hand 6H-SiC, at both at mospheric and low pressure [112], as well as the work on (100)3C-SiC at atmospheric p ressure [113], and reports a new low pressure etching regime for (100) and (111)3C-SiC. 2.3.1 Low Pressure Hydrogen Etching Process We have developed processes which successfully etc h the surface of the three commonly used polytypes of SiC through the use of a n inductively heated hot-wall CVD reactor. First, all the samples were cleaned using the “piranha” clean followed by an HF dip and DI rinse. The CVD reactor was loaded with a hot zone specifically dedicated for H2 etching a requirement which was found to be crit ical in eliminating particulates from depositing on the substrate surface during the etch which originated from hot zone components previously used for growth processes [11 3]. The specifics for the etching


71 processes developed are summarized in Table 2.2. After etching, the samples were allowed to cool in a hydrogen flow of 10 slm until the temperature was well below 400 C, which had been previously established as a necessity for stable surface passivation [111]. Table 2.2: Summary of hydrogen etching parameters for SiC surfaces [110-113]. SiC Polytype Crystal Orient. Wafer Axis Polishing Type Etch T [C] Etch P [Torr] H2 [SLM] T [min]* 4H-SiC (0001) On(Off-) Optical 1400 760 10 20 (30 ) 4H-SiC (000-1) On(Off-) Optical 1500 760 10 20 (3 0) 6H-SiC (0001) On(Off-) Optical 1550 760 10 20 (30 ) 6H-SiC (000-1) On(Off-) Optical 1500 760 10 20 (3 0) 4H-SiC (0001) On(Off-) Optical 1400 150 3 5 (15) 4H-SiC (000-1) On(Of) Optical 1500 150 3 5 (15) 6H-SiC (0001) On(Off-) Optical 1550 150 3 10 (15) 6H-SiC (000-1) On(Off-) Optical 1500 150 3 10 (15 ) 3C-SiC (100) OnAs grown 1200 760 10 30 3C-SiC (100) OnAs grown 1200 100 5 10 3C-SiC (111) OnAs grown 1275 100 5 10 First number listed is for on-axis substrates and number in parenthesis is for off-axis substrates. 2.3.2 Characterization of Hydrogen Etched SiC Surfaces The experiments show that the surface damage cause d by the mechanical polishing of 4Hand 6H-SiC wafers can be removed a nd an ordered surface of atomic steps revealed. Work on these substrates has been previously reported and Figure 2.10 shows two examples of etching using the parameters in Table 2.2 [110, 112]. For all hexagonal SiC samples, polishing damage was removed and the resulting surface displayed atomic steps of ~ 10 heights for 4Hand ~15 for 6H-SiC.


72 (a) (b) Figure 2.10: 10 x 10 m AFM micrographs of SiC sur faces displaying the surface before (left) and after (right) H2 etches using the parameters from Table 2.2. Inclu ded is a measurement of the step height after etching. (a) On-axis 4H-SiC (0001) sample with a roughness of rq = 0.396 nm. (b) On-axis (000-1) 6H-SiC sample wit h a roughness of rq = 0.502 nm. AFM was performed in tapping mode on a P ark Systems XE-100 AFM. Note that there are no morphological differences between atmospheric etching and LP etching. Experiments were also performed on 3C-SiC epilayer s which were grown on (100) and (111)Si using the CVD process described i n (§ 2.2.1). The surfaces of the 3CSiC crystals exhibited atomic-scale steps directly after growth. The (100)3C-SiC morphology displayed an inhomogeneous surface chara cterized by mesa growth within antiphase domains (APD), however within the APD’s t here were atomically flat terraces with small, wavy steps loosely aligned along the [1 10] and [-110] low-Miller indices. Figure 2.11(a) shows an AFM micrograph of the surf ace of as-grown (100)3CSiC which possesses steps of ~ 100 nm length and hei ghts of ~2.5 i.e. layers corresponding to the bi-atomic (2.18 ) lattice of Si-C bonds, indicating a polar surface termination with either C or Si atoms. The APD’s a re measured to fall within 2 4 nm in height and 25 50 nm in width, and the surface rou ghness, rq, was 6.38 nm. The surface of (111)3C-SiC displays a much more homogenous surf ace as seen in Figure 2.11(b). The regular terrace steps are loosely aligned along the [110] and equivalent directions.


73 One major difference of the surface is that, unlike (100)3C-SiC, APD’s are not generated during epitaxial growth because (111)Si is the pola r face of the 3C-SiC, although there are many linear defects that occur from twinning al ong the {111} crystal plane [114]. As noted before with (100)3C-SiC, the steps are not sh arply aligned along the [110] Miller indices direction, but appear wavy and disorganized The step heights measure ~ 8 or 16 along the tips of the terraces triangles and measu re ~ 16 or 24 along the boundaries indicating step bunching occurring during the growt h of the crystal. The surface roughness is rq = 6.80 nm. As we have previously developed an atmospheric pre ssure 3C-SiC H2 etching process, we dedicated the final part of this study to develop and characterize a low pressure etching process [113]. Utilizing the prev ious study as a starting point for process etching temperature, (100) and (111)3C-SiC samples were etched at temperatures of 1200 C, 1250 C, and 1300 C and the sample sur faces characterized via AFM to determine the effect of the etching. Figure 2.11(a ) shows the 2.5 m x 2.5 m AFM scans of the (100)3C-SiC etched surfaces performed at each of these temperatures. Etching at 1200 C shows steps that have increased order along the low-Miller indices, and the step edges are more squared and straight th an the as-grown steps. The step heights on the surface were measured at ~ 1.8 – 2.5 and terraces ~ 75 100 nm long indicating that the surface still displays the atom ic bi-layer surface characteristic of the as grown surface. Over etching of 3C-SiC surface defe cts is a large problem for H2 etching, but the APD trenches were not affected significantl y at this temperature, measuring ~ 2 – 4 nm in depth and ~ 25 75 nm in width. Overall th e surface became more flat between


74 (a) (b) Figure 2.11: AFM micrographs of (100)3C-SiC surfac es (a) epitaxially grown on (100)Si after subsequent H2 etches at the temperatures indicated, and (b) AFM micrographs of (111)3C-SiC surfaces epitaxially grown on (111)Si a fter subsequent H2 etches. All scans are 2.5 x 2.5 m, and were done with silicon nitrid e tips using contact mode. the varying surfaces as the rq has been reduced to 5.7 nm. The (100)3C-SiC etch at a temperature of 1250 C resulted in even more square d edges with steps along the same Miller indices; however many interior steps were me asured at ~ 4 – 5 intermingled with steps of 2.5 – 3.0 , indicating that step bun ching of the bi-layer steps was beginning to occur. Also noted was the appearance of inverted square pyramidal pits 3 5 nm deep, and the APD trenches have a slightly inc reased depth of 3 – 4 nm and width of 60 – 100 nm. However the etching further flatte ned the surface reducing the surface roughness to 4.93 nm. The (100)3C-SiC etched at a temperature of 1300 C became more mosaic, was completely squared, and possessed step heights of ~ 4 – 12 nm and lengths of 200 – 400 nm. The etch pits increased i n size and depth, many measuring over 100 nm wide and 7 nm deep. The APD trenches increa sed in width to 100 – 300 nm, but As Grown SiC H2 Etch 1300 C H2 Etch 1250 C H 2 Etch 1200 C H2 Etch 1300 C H2 Etch 1250 C H 2 Etch 1200 C As Grown SiC


75 have a reduction in depth to ~ 0.5 – 2.5 nm. This o ver-etched surface displays a great reduction in the surface roughness to 1.45 nm. The H2 etching of (111)3C-SiC results are shown in Figure 2.11(b). At 1200 C, the steps on the surface still appear wavy and loos ely aligned along the low-Miller indices. Steps along the apex of the step edge sho w a reduction in step height to ~ 7 – 8 or ~ 10.5 – 12 and steps along Miller index boun daries were reduced to ~ 11 – 12 or ~ 16.5 to 18 . Most terraces have a width of 25 – 50 nm, but there are some larger terraces present with widths of 75 – 200 nm. The s urface roughness was only slightly improved with an rq of 6.37 nm RMS. The H2 etch at 1250 C reveals a great number of steps measuring ~ 2.5 – 3.7 intermingled with ~ 4.8 to 6.7 steps along both the triangular apex and the edges of the terraces; howe ver the step morphology still remains wavy and loosely aligned to the Miller-index bounda ries. The terraces have relatively uniform with widths of 10 – 30 nm and the rq is 3.77 nm RMS. At a temperature of 1300 C, the morphology changed to straight steps w ith low-Miller indices alignment. Steps of heights ~ 6.5 – 8.2 , ~15.2 – 19.4 , and 2 2.3 – 26.6 appeared on all edges of the terraces, indicating a return to the step bu nching as seen in the as-grown material. The terrace widths were increased along the Millerindex borders to 20 – 40 nm, and many of the widths along the triangular apex were i ncreased to 150 – 400 nm in width. The surface roughness also increased to 4.78 nm RMS The results of this study show that the three majo r polytypes of SiC can be etched successfully at low pressure, which establishes a s afe and more controlled process. The study did not electrically evaluate the surface ban d bending as was done by Coletti et al .;


76 however this process will eventually be quantified in the same manner by new members of the SiC group at USF [111]. 2.4 NCD Deposition on 3C-SiC/ Si The unique material characteristics of the two mat erials, 3C-SiC and NCD, present some unique solutions for the generation of high temperature, power, and high frequency MEMS devices. As semiconductor materials 3C-SiC and NCD have very good electrical properties that can be utilized for the creation of electrical devices. The band gap difference and high carrier mobility for b oth of these materials generates a heterojunction that can become part of an extremely efficient and fast switching bipolar junction device [24]. As both of these materials a re potentially biocompatible, the generation of permanently implantable MEMS devices is also a possibility. Single crystal 3C-SiC is normally heteroepitaxially grown on silicon using CVD methods previously discussed in (§ 2.1 2.2) and the proce ss is extremely sensitive to both growth and substrate conditions. NCD, on the other hand, can be deposited on any material which is stable at the deposition temperature by us ing microwave plasma enhanced chemical vapor deposition (MPECVD) [115]. NCD depo sited on Si substrates have many additional defects from the stress caused by t he relatively large lattice mismatch [115]. Because the lattice mismatch is slightly le ss between NCD and 3C-SiC, deposition of NCD on 3C-SiC substrates has been pro posed to help alleviate stress defects caused by the lattice mismatch (Si = 0.543 nm, SiC = 0.436 nm, and NCD = 0.357 nm).


77 The focus of this section is to characterize the d eposition of NCD on 3C-SiC which has been homoepitaxially grown on (100) and ( 111)Si substrates and to evaluate the quality of the interface between the different layers. The characterization of the materials was performed using FTIR, SEM, TEM, AFM, XRD, Raman spectroscopy, and near edge x-ray absorption spectroscopy (NEXAFS). 2.4.1 NCD Deposition Process The deposition of NCD on another material using a commercial IPLAS Cyrannus I MPECVD reactor requires some preparation stages. First the Si substrate is seeded with diamond nanoparticles. This seeding involves addin g the deposition substrate to a slurry solution containing 1.2 g of titanium nanopowder, 1 .2 g of diamond nanopowder, and 100 ml of methanol or acetone. The solution and su bstrate are left in an ultrasound bath for 20 minutes followed by two 20 minute cleaning c ycles within only methanol. The samples are then removed and died with a N2 gun. The diamond seeded substrates are placed within the MPECVD reactor where an Ar plasma is struck. The pressure of the deposition is 135 Torr, a temperature of 750C and the microwave power is maintained at 1.8 KW. Methane, CH4, is used as the carbon precursor and H2 is used to control the size of the NCD grains. To grow intrinsic NCD, the CH4 is 0.5%, the H2 is1%, and Ar provides the balance of the gas used for the deposi tion which has a rate of ~0.17 m/ hr. Samples of (100) and (111)Si as well as (100) and ( 111)3C-SiC were seeded and NCD was deposited for 3 hours. Measurements using FTIR showed an uneven deposition with film thickness ranging from 400 – 600 nm.


78 2.4.2 Material Characterization (a) (b) Figure 2.12: Three dimensional AFM micrographs of NCD growth on various substrates. The micrographs in (a) depict from left to right NC D on (100)Si, 3C-SiC grown on (100)Si, and (c) NCD deposited on (100)3C-SiC. The micrographs in (b) depict from left to right (111)3C-SiC grown on (111)Si, 3C-SiC grown on (111)Si, and NCD grown on (111)3C-SiC. Note that the NCD on (100)Si has larg er micro-grain defects that are not present within the NCD film deposited on (100)3C-Si C and that the rough mesa surface of as-grown 3C-SiC does not affect the structures o f the NCD whereas the defective clusters are about the same on the (111) face. The triangular morphology of the (111)3CSiC is apparent through the NCD deposited on that f ace. Figure 2.12 depicts AFM analysis of the surface mo rphology of NCD deposited on (100) and (111)Si, the initial surface of the 3C -SiC grown on these two Si faces, and NCD deposited on the two 3C-SiC surfaces. NCD depo sited on (100)Si shows a regular, cubic grains ranging from 100 – 500 nm in perimeter and 1,000 nm2 – 10,000 nm2 in area. However, the surface is marked with craters and the presence of many large, oval shaped defects which measure approximately 5 m in perimeter and 1m2 in area. The NCD/ (100)3C-SiC NCD/ (100)Si (100)3C-SiC NCD/ (111)Si (111)3C-SiC NCD/ (111)3C-SiC


79 difference between these features on the NCD surfac e increased the roughness, rq, to 17.0 nm RMS. The initial (100)3C-SiC, which was 3.45 m thick measured by FTIR, shows the characteristic mesa growth pattern, atomic step terraces, and APB crevices. The initial surface has an rq, of 7.25 nm RMS. The NCD grown on the (100)SiC su rface shows a relatively even, flat surface with many sma ll cubic grains 75 300 nm in perimeter and 500 nm2 – 7,500 nm2 in area. This surface does not show the large gra in inclusions displayed by the NCD on (100)SiC but ins tead has round inclusions with approximately 2 m perimeters and 75,000 nm2 area. The relatively flat surface and similar grain have surface roughness of 7.30 nm RMS which is a little over half the roughness of the NCD on (100)Si. NCD grown on (111)Si displays a much different sur face morphology than that seen with NCD on (100)Si. The grains on this surfa ce possess rounded surfaces and show a wider range of size differences with measure ments ranging from 500 nm – 2 m in perimeter and 15,000 nm2 – 100,000 nm2 in area. The surface is littered with large circular clump-like inclusions 28 m in perimeter and areas up to 1.8 m2. The large defects and differences in grain size increase the rq to 37.9 nm RMS. 3.35 m of 3C-SiC grown on (111)Si shows a regular atomic step, trian gular terraced surface and possesses an rq of 10.6 nm RMS. NCD deposited on this surface is similar to the NCD deposited on (111)Si in that there are irregularly sized grains and many large circular defects. The difference is the grains on this surface are triang ular and much larger than the NCD grains on (111)Si with an average perimeter of 1.5 – 3 m and area of 150,000 nm2 – 500,000 nm2. The larger, circular defects are measured approx imately possessing 5m


perimeters and 1.5 m2 areas. The roughness of this surface is deposited on (111)Si which was (a) (c) (e) Figure 2.13 : SEM plan view NCD deposited on the various tested substrates. NC D deposited on (100)3C displayed in (ab), on (111)3C the NCD and 3CSiC inter courtesy of M. Italia, IMM 80 areas. The roughness of this surface is almost equivalent to NCD deposited on (111)Si which was 27.7 nm RMS. (b) (d) (f) : SEM plan view (column 1) and crosssection (column 2) micrographs of the NCD deposited on the various tested substrates. NC D deposited on (100)3C b), on (111)3C -SiC in (c-d), and on (100)Si in (e-f). 3CSiC that indicates SiC inter face is not completely integrated. SEM data (a) an d (b) courtesy of M. Italia, IMM -CNR, Catania (IT). almost equivalent to NCD section (column 2) micrographs of the NCD deposited on the various tested substrates. NC D deposited on (100)3C -SiC is SiC that indicates face is not completely integrated. SEM data (a) an d (b)


81 Figure 2.13 shows further examination of the morph ology of the samples through the use of SEM. (100)3C-SiC, shown in Figure 2.13( a-b), shows a relatively smooth surface which has small grain and very small large grain defects. The cross section shows that the film is highly granular, possessing many random grain boundaries. Figure 2.13(c-d) show the morphology of the NCD film depos ited on (111)3C-SiC. This film is very rough and has many large, dark defects, indica ting that these defects are more conductive than the film around them. The cross se ctional view indicates that these mound-like defects extend far above the nominal NCD surface. The cross section also shows that the NCD film is not as granular as the N CD deposited on (100)3C-SiC, and tends to align in sheets 45 normal to the surface. Figure 2.13(e-f) show NCD deposited on (100)Si. This film shows a surface which posses ses small grains with intermixed larger grains, much like was seen with the AFM. A difference in this film when compared to the other NCD films is film is very def ective close to the NCD/ Si interface, but the defects decrease with increasing film thick ness. NCD does not appear to have good adhesion with any of the films, as indicated w ithin the cross section views. The film is broken back and uneven as compared to the S i and 3C-SiC which are even with the cleaved surface. The crystal quality of the NCD deposited on these various substrates was examined next. Figure 2.14 shows Raman and NEXAFS measurements of the NCD film deposited on (100)3C-SiC. Raman peaks indicated at 1333 cm-1 and 1380 cm-1 in the D band as well as 1580 cm-1 in the G band are indications of sp3 bonds of diam ond. The band at 1172 cm-1 is an indicator of the presence of NCD. The Raman measurement in


82 (a) (b) Figure 2.14: Raman (a) and NEXAFS (b) absorption f ine structure scans of NCD deposited on (100)3C-SiC. Raman courtesy of H. Gom ez, USF, and NEXAFS courtesy of Dr. J. Sathyaharish SRI International. Figure 2.14(a) indicates that the NCD film possesse s 95 – 98% sp3 bonded carbon atoms. The NEXAFS spectrum obtained for the NCD deposited on 3C-SiC indicates a similar type of graph as single crystal diamond, suggesting NCD film is phase diamond.


83 (a) (b) (c) Figure 2.15: XRD -2 diffraction surveys for the NCD films deposited on (a) (100)3CSiC, (b) (111)3C-SiC and (c) (100)Si substrates. X RD measurements courtesy of H. Gomez of USF. XRD -2 diffraction surveys were utilized to collaborate t he presence of diamond on the surface of the different substrates and to quantify the crystal orientation of the NCD film. Figure 2.15(a) shows the -2 diffraction surveys measured at CuK emission from 25 80 for the NCD grown on (100)3C-SiC. The (100)3C-SiC and (100)Si substrate beneath the NCD film are detected as indicated by the peaks at 41.5 for the (200)3C-SiC, and 32.95 and 69.1 for (200)Si and (400)Si respectively. The peaks normally associated with (100) diamond, 51.17 for (200)C, are not present within this measurement, and any (111) diamond, 43.9 would be lost within the broad 3C-SiC peak. Figure 2.15(b) shows the -2 diffraction surveys measured at CuK emission 1 10 100 1000 10000 100000 254565 Position [2 Q ] NCD on (100)3C-SiC 305070 Position [ 2 Q ] NCD on 3C-SiC(111)(200)Si (200)3C-SiC (111)C (111)3C-SiC (111)C (200)Si 25 35 45 400 200 Position [ 2 Q ] (400)Si (222)3C-SiC


84 from 30 85 for the NCD grown on (111)3C-SiC. Two major peaks which are associated with (111)3C-SiC are located at 35.7 and 75.7 A (111) diamond peak is present in this measurement, and its low, broad pea k suggests the polycrystalline alignment of the nanoparticle diamonds on the surfa ce. The (111) Si peaks located at 58.85 were not present with this measurement and more in vestigation into the cause of this is needed. Figure 2.15(c) shows the -2 diffraction surveys measured at CuK emission from 20 50 for the NCD grown on (100) Si. A large (100)Si pe ak is present at 32.95 and the (111) diamond peak at 43.9 is much more sharp in comparison to the peak of the NCD on (111)3C-SiC. A repeat of this m easurement, increasing the sweep of Q to at least 80 should be done to examine if (200) diamond is pres ent in the NCD. Our initial probing into the deposition of NCD on 3C-SiC has shown us that the resulting material is different than that deposited on Si using the same process. The grains and defective clusters are smaller with the NCD deposited on 3C-SiC than that of the NCD deposited on Si. The difficulties of the ( 200)3C-SiC peak overlapping the (111)C diamond peak requires the use of another mea surement for crystal quality. Selected area electron diffraction (SAED) TEM shoul d be performed on the samples to quantify the crystal orientation. Dark field and b right field imaging techniques can be used to obtain information on the grain quality as well. We are currently examining the films with the TEM, but have not generated enough q uality evidence as of yet to present in this work.


85 2.5 Summary of Chapter 3C-SiC, unlike other forms of SiC, is a material w ith little material development. Our group at USF has been recognized as a world lea der with the growth of 3C-SiC and has developed methods to decrease defects inherent with its epitaxial growth. We have developed a low pressure method which decreased the interfacial defects present in 3CSiC, and hydrogen etching treatments which reduce d efects and create atomically flat terraces on the surface of the semiconductor. Our improvements in the crystal quality of 3C-SiC will improve the quality of a MEMS device th at would be required to interact with the CNS. Another requirement of the implantab le neural prosthetic would require a high quality, biocompatible insulating material to enable the generation of field effect devices. NCD may be the biocompatible material tha t would achieve this purpose, and our investigation into the integration of this mate rial with Si and 3C-SiC revealed defects that would decrease electrical operation. Continue d study of the integration of these materials is required for the generation of a final neural implantation device and shall be discussed in chapter 5. The focus of the next chap ter is to evaluate the biocompatibility of monocrystalline 3C-SiC and NCD using in vitro methodology, which is clearly integral to the development of an eventual in vivo neuronal implant device.


86 Chapter 3: Neural Biocompatibility of SiC and NCD 3.1 Chapter Introduction The central nervous system, or CNS, is one of the most complicated systems in the body and consists of a dense network of neurons and glial cells. Due to the very nature of this system it is very difficult to inter act externally with the CNS, either through the use of drugs or externally interfaced biomedica l devices. The blood-brain barrier and the actions of glial cells, more specifically the a strocytes, effectively prevent antibodies and many drugs treatments from reaching their inten ded targets within the CNS [26]. Fortunately implantation of a biomedical device int o the CNS makes direct interaction with the cells possible and effectively bypasses th e blood-brain barrier, but the construction materials can activate microglia and a strocytes which react with the foreign material to induce gliosis, which is the glial scar response [30, 31, 116, 117]. Devices which can interact electrically with the cells were discussed in (§ 1.4), but these devices could also be utilized as a platform for drug deliv ery to specific CNS areas or to transport neural factors or even other cells to the CNS to re pair the system damaged by glial scarring [116, 117]. These devices can offer solut ions to many problems that are currently untreatable by conventional medicine, but their implantation has led to the initiation of glial scarring instead of providing t he therapies that they were designed for.


87 Studies have shown that damage caused by the inser tion of devices into the brain can be minimized by a carefully designed device; th e smaller in dimension and sharper the device is, the lower the probability of initiat ing glial scarring immediately after insertion [118]. The damage from the insertion its elf is not the source of glial cell activation, but is caused instead by another factor which leads to long term inflammatory response [118]. The greatest factor that prevents the use of proposed CNS interaction devices arises from the interaction between the dev ice materials and the multitude of cells, fluids, and chemicals within the CNS, which is simply an effect commonly known as material biocompatibility [6, 31]. Of the many materials used to date and incorporated within the CNS system, few have shown the level of biocompatibility necessary for incorporation or for proper long term successful op eration of the device. In (§ 1.5-1.6), two materials were introduced as possible candidate s for the construction of a neural implantation device based on the fact that amorphou s or polycrystalline forms of these materials had shown good biocompatibility along wit h chemical and mechanical resilience. Within this section, we will cover the research we have done to investigate the biocompatibility of 3C-SiC and NCD with immortalize d neural and CNS cell lines. (§ 3.2) will explain how the living cell can interact at a molecular level with inanimate materials, methods to test for biocompatibility of the materials, and the experiments that were used to determine the degree of biocompatibili ty. (§ 3.3) details the results obtained from these experiments, and (§ 3.4) will discuss th e results and offer some possible explanations for these results.


88 3.2 Foreign Material and Living Cell Surface Interactio n The cell to substrate interaction is a very import ant consideration for determining a material’s performance for biomedical implant app lications. Neurons and many glial cells, like many cells in the mammalian body, must attach to an external surface or they will experience programmed cell death or apoptosis. Generally a cell will attach to a substrate surface through physiochemical interactio n between the cells cytoskeleton, cell membrane, extracellular proteins, substrate ions, h ydration compounds, and molecules [119, 120]. Macroscopically cells on biocompatible surfaces will interact with the substrate by increasing their attached area and wil l have greater viability than cells interacting with unsuitable or toxic surfaces [119] Thus, cell morphology and viability can be used as measures of surface biocompatibility Another measurement of surface biocompatibility is to evaluate the extension of lamellipodia or filopodia from motile cells onto a substrate surface [121]. Neurons and many glial cells experience different levels of mot ility during their lifetimes [122, 123]. During the developmental stage of the brain the cel ls are fully motile and travel down radial glial cells to predetermined locations where their somata firmly attach to the extracellular matrix [122, 123]. The attached soma ta of the neurons extend microtubulefilled growth cones that are the backbone for the d ynamic formation of dendritic spines and axons [121]. The growth cone is guided by peri pheral extensions of actin filled cytoplasmic lamellipodia and spike-like filopodia [ 121]. Astrocytes and other glial cells also extrude lamellipodia or filopodia which active ly seek to interact with neurons, endothelial cells, or other glial cells [122]. The lamellipodia or filopodia are attracted

PAGE 100

89 and repulsed by chemical and contact-mediated cues which influence how the structures connect to the outside world [121, 122]. If an ext ernal surface is repulsive to the cells, the lamellipodia or filopodia will not couple with the surface and actin filaments flow back into the central core, but if the substance is attractive, the lamellipodia or filopodia couple strongly with the surface and new actin fibe rs flow to the leading edge of the cell extension [121]. Using the actions of lamellipodia or filopodia structures, a surface can be considered permissive to the cells if it produce s an attraction of the lamellipodia and filopodia, and non-permissive if it repels them [12 1]. 3.2.1 SiC Biocompatibility One method for increasing the chronic implantation time for the implantable neural device is to increase the level of biocompat ibility of the device. One method of improving the biocompatibility of a device is to ad d surface coatings [31, 116, 117]. Many coatings have been used such as biocompatible material, inorganic or organic polymers, or chemical complexes and enzymes [31, 11 6, 117]. Although these coatings can make the device much more biocompatible, many m ajor difficulties arise from this approach. Devices that utilize electrical intercon nections with cells and have coatings added to their surfaces also have additional comple x electrical impedance factors which decrease the electrical operational capabilities of the device [3, 4, 13, 14, 40]. Many of the proposed coatings are also temporary, and when the base material is exposed to the body these coatings fail. What is clearly preferre d is a material that is biocompatible without the use of coatings, particularly for devic es which must sense the biological

PAGE 101

90 system electrically. 3C-SiC, which our group has a ctively tested in vitro with skin and bone cell lines to evaluate its general biocompatib ility with the goal of demonstrating if the material is a suitable candidate for use in ele ctrically active biological sensors, may be just that material [124]. The biocompatibility of 3C-SiC was determined by u sing the methods for testing of biomedical devices which are outlined in ISO 109 93. Initial testing techniques outlined in this standard follow in vitro testing of bulk and surface materials involved in the creation of the device with cultures of cells. Specifically, for Coletti et al. the growth of selected cell lines on specific tested materials was monitored using fluorescence microscopy. Optical microscopy allows for qualitati ve determination of cell morphology and cell attachment over populations of cells. Cel l viability on these surfaces was determined quantitatively using yellow MTT, (3-(4,5 -Dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide), assays [125]. Colett i et al. compared the three major polytypes of SiC, namely 3C-SiC, 4H-SiC, and 6H-SiC with Si and used polystyrene, PSt, as a control. The immortalized mammalian cell s cultured on the semiconducting substrates were: B16-F10 mouse melanoma cells (ATCC CRL6475); BJ human fibroblasts (ATCC CRL2522); and human keratinocytes cells (HaCaT). All cells were cultured initially into 75 mm2 flasks and incubated at 37 C in an air atmosphere containing 5% CO2 and 95% relative humidity. For each test, cells w ere collected from the flasks through trypsinization, and enumerated u sing a Hemocytometer to determine the number of cells per ml.

PAGE 102

91 The results obtained clearly show that SiC is, in all of its tested phases, a highquality surface for cell culture with significantly better performance than Si and nearly as good a performance as the PSt control. No statisti cally significant differences were found among the cell proliferation values on the differen t SiC polytypes [124]. An extremely satisfying proliferation was observed for BJ fibrob last cells on SiC, which were on average the same obtained for the culture-well read ings, and a slight bioactivity of the SiC substrate could even be hypothesized from the r esults [124]. A smaller difference in B16 cell proliferation was observed between Si and SiC substrates, but can be easily justified by the nature of the cell line [124]. B1 6 melanoma cells are, in fact, extremely aggressive cancer cells capable of indifferently ad hering to substrates of diverse biocompatibility: therefore a reduced selectivity w as expected. A secondary set of MTT assays was performed on only the Si, 3C-SiC, and 4H -SiC samples using the HaCaT cell line which was evaluated on the first and eighth da y of HaCaT cell culture [126]. Although proliferation values for each substrate ar e comparable on the first day after seeding, cell proliferation was greater on both pol ytypes of SiC and significantly decreased on Si surfaces after eight days of cultur e [126]. The lowered viability of SiC during the short term suggests a difficulty of the material to absorb proteins, but the study by Coletti et al was not designed to verify this result [126]. Cell proliferation on the tested substrates was qu antified using MTT assays. For this proliferation assay, cells were cultured on th e testing substrates and yellow MTT was added after a predetermined time [125]. MTT is onl y metabolized by living mitochondria to produce purple formazan, insoluble in water but soluble in dimethyl sulfoxide, DMSO,

PAGE 103

92 and subsequently quantified the assay by using a sp ectrophotometer measuring the index of refraction with a wavelength in the range of 500 to 600 nm [125]. Testing cellular morphology and its permissiveness to a particular s ubstrate has many well established methods, one of which is fluorescence microscopy. Another very selective measurement technique for cell attachment is total internal ref lection fluorescence (TIRF), which can image the cell interface within 200 nm from the sur face; however this technique nominally requires a substrate that allows evanesce nt and internal reflection wave generation, such as silica, and substrates like SiC Si, and NCD are either opaque or translucent and will not easily allow for the inter nal reflection [127]. Confocal microscopy, a technique which is gaining prevalence in biological examination, generates point illumination of a sample through a combinatio n of filters, beam splitters, and gratings to eliminate out-of-focus light thereby de tecting light within a focal plane, and raster scanning can then be used to generate detail ed two and three dimensional images [128]. Unfortunately, confocal microscopes are ver y expensive, and many of the models used for biology are designed with inverted microsc opes and are not suitable for opaque substrates. AFM uses a laser reflected off a canti lever to gather quantitative morphological surface information. This technique can image living cells on opaque surfaces, and costs much less than a confocal micro scope. Although it cannot provide information about the surface interaction underneat h a cell, or information from inside the cell, it can provide detailed information on ce ll spread and morphology on almost any surface. The methodology for AFM measurements will be covered in detail in (§ 3.2.4).

PAGE 104

93 3.2.2 Sample and Cleaning Protocol In this study we report on research to evaluate th e biocompatibility of SiC, NCD, and Si with neuron cell lines and utilize polystyre ne (PSt) and amorphous glass from glass microscope slides as control materials. Unli ke Coletti et al. the growth of 3C-SiC had progressed onto another crystalline face, so we tested (100)Si as well as (111)Si. Using the methods covered in (§2.2), (100)3C-SiC an d (111)3C-SiC was homoepitaxially grown on (100)Si and (111)Si wafers in the MF2 CVD reactor at the University of South Florida. 4H-SiC and 6H-SiC samples were obtained f rom wafers purchased from Cree Inc. NCD was deposited on (100)Si wafers using a C yrannus Iplas MPECVD NCD Reactor utilizing growth methods covered in (§2.4). Amorphous glass cover slips and sterile Corning Cellbind" treated polystyrene Petri dishes (Fisher 07-202-51 4) were utilized for the control materials. The semiconductor samples were further subdivided by bulk doping levels. The Si samples were intentionally negatively doped with phosphorus, while the 3C-SiC was unintentionally negatively doped due to a small pre sence of nitrogen during growth. The NCD, which was deposited at a much lower pressure t han the 3C-SiC, would be intrinsic with hydrogen surface termination. There was n-type p-type, and intrinsic samples available of both the hexagonal forms of SiC. All samples used in the study were diced into dimensions of 8 mm x 10 mm to facilitate inser tion into 12-well and 24-well tissue culture treated polystyrene plates (USA Scientific # 5666-5180 and 5666-2160). After dicing, the semiconductor and glass samples were ul trasonically cleaned in solvent baths of acetone, methanol, and isopropanol for 5 minutes each to remove dicing particulates

PAGE 105

94 and disinfect the surfaces. The semiconductor samp les were then cleaned using the standard RCA cleaning process. The cleaning steps o f this procedure were as follows: a 1:1:6 mixture of NH4OH:H2O2:H2O heated to 65oC for 10 minutes; followed by a 50:1 H2O:HF for 30 seconds and by a 1:1:6 mixture of HCl:H2O2:H2O heated to 65oC for 10 minutes [124]. All H2O was deionized (DI) with r > 16 M W cm. The diced PSt samples were used only for the AFM analysis and were diced within a class II A/B3 biological safety cabinet followed with brief immersion in 70% (v/v) ethanol to remove any particulates. Before cell plating, the semiconductor samples wer e cleaned using a piranha solution (2:1 H2SO4:H2O2) with 10 minutes of immersion and then rinsed thre e times in DI H2O. This was followed by a 5 minute immersion in a solution of HCl (2:1 HCl:H2O) to remove any remnant sulfur, a known cytotoxic ele ment to neurons, from the surface. After rinsing the sample once again in DI H2O, the substrates were dipped in 50:1 H2O:HF for 2 minutes to remove any native oxides and rinsed three times in DI H2O. The samples were dried using pressurized N2 and placed into ethanol to prevent both bacterial growth and surface oxidation. The glass substrates were autoclaved in 121 C water steam for 30 minutes prior to cell deposition Due to the high cost of CVD crystal growth and time required to generate these films, t he same SiC and NCD samples were reused for multiple experiments. The results of th e study by [124] showed that 3C-SiC becomes silicon rich from loss of carbon due to mul tiple cleanings in piranha solution, so to maintain surfaces representative with ‘as grown’ characteristics, which contains both

PAGE 106

95 silicon and carbon on the surface, we replaced the 3C-SiC and NCD with fresh samples after 3 cleanings [124]. 3.2.3 Cell Culture and MTT Assay Protocol Two neural cell lines obtained from American Type Culture Collection, ATCC, in Manassas, MD were used for this study; H4 human neu roglioma (ATCC #HTB-148), and PC12 Rat pheochromocytoma (ATCC #CRL-1721). The H4 cell line is neuron derived from a CNS glial tumor in the frontal cortex, and t he PC12 is a peripheral neural modeling cell derived from the adrenal gland [129]. The H4 cell line was cultured in advanced Dulbecco’s Modified Eagle’s Medium (DMEM) (Invitrogen #12491-015) without L-glutamine supplemented with 10% fetal bov ine serum, FBS (Invitrogen #26140-087), 2.2 mM L-glutamine GlutaMAX-1 suppleme nt (Invitrogen #35050-061), and 1% penicillin-streptomycin (Sigma #P4458-100ML) The PC12 line was grown in Kaighn’s F-12K Media (Invitrogen # 21127-030) suppl emented with 2.5 % FBS, 15% horse serum (Invitrogen #16050-122), and 1% penicil lin-streptomycin. The cells were cultured in Corning Cellbind 75 mm2 flasks (Fisher # 07-202-515) to approximately 90% confluence, collected through trypsinization, a nd enumerated using a Hemocytometer to determine the number of cells per ml. Cleaned substrate samples were seeded with cells b y first drying the ethanol on the samples using a hand held bulb. Each sample wa s placed into one well of a 12-well plate. Initial experimentation had determined that 1x105 H4 cells and 5x105 PC12 cells would be seeded in each well, followed with the add ition of 3 ml of the appropriate

PAGE 107

96 media, to achieve the level of cell coverage necess ary for the assay. After 96 hours, the MTT assay was performed in accordance with the proc edures outlined in [124, 125], with each assay repeated in triplicate for three individ ual samples from each tested substrate along with the PSt control. The results from these tests were normalized with respect to the PSt control readings and expressed as the sampl ing distribution of the mean ( x) standard error of the mean (M s ) as reported in the next section. Amorphous glass was not tested in the MTT analysis due to the fact that it already has been determined to have a low biocompatibility level [10]. 3.2.4 AFM Methodology for Cell Morphology and Substrate P ermissiveness AFM measurement, using a Park Systems XE-100 AFM, for each substrate and cell line was conducted to provide quantitative ana lysis of cell morphology and permissiveness. Cells were plated on each of the s ubstrates using the plating procedures described in (§3.2.3) and allowed to incubate for a t least 48 hours. The AFM of living cells was performed in phosphate buffered solution, PBS, provided the basis for a quantification of cell height, total surface area o f the cell, angle of attachment, and the full extension of the lamellipodia/filopodia. Afte r the living cell scans, the cells were then fixed using 4% paraformaldehyde for 20 minutes to immobilize the lamellipodia and filopodia structures so we could measure an overall progression or regression over the surface of the substrate. As each cell studied in the AFM had slightly different physical dimensions, the measurements were therefore express ed as ratios or percentages to provide a basis for comparison. The ratio of cell height to cell area and the angle of

PAGE 108

97 attachment, which is the slope of the membrane to t he substrate surface, were used to evaluate the attachment level of the cell to the su bstrate. Analysis of the AFM images was performed with Park Systems XEI image analysis software. Figure 3.1(a) shows an example of line scan analysis which was used to det ermine maximal cell height and angle of attachment. Figure 3.1(a) also shows that area analysis is possible by selecting the image pixel height and was used to determine the su rface area of the cell. Although the topographical measurement provided by AFM will not image the internal actin meshwork associated with lamellipodia and filopodia, these s tructures can still be identified and measured. It is the consensus view that filopodia are the cell vanguard for probing the local environment which is then followed by the gen eration of lamellipodia [130]. By examining the outer extremities of the cell, we ide ntified the spike-like filopodia structures, as shown in Figure 3.1(b), and used the m to identify the lamellipodia structure through line scan analysis since the lamellipodia a re known to be a thin membrane layer 100 – 200 nm thick and parallel to the substrate su rface [130]. Once identified, these features were measured using XEI area analysis on b oth the living cell and the fixed cell observations, normalized as a percentage of total c ell attachment area, and the subsequent loss (or gain) percentage in area was used to quant itatively express the cell permissiveness to the substrate. Due to the fact t hat many hours were required for the analysis and measurement of the cells, three cells were randomly selected from each surface studied. The results were then analyzed and mean results used to make the final comparison between surfaces.

PAGE 109

98 (a) (b) Figure 3.1: Depiction of various AFM measurements utilized for the determination of cell permissiveness on the substrates tested in thi s study. (a) From left to right: a 45 x 45 m AFM micrograph of a cell displayed in enhanced c olor, a line evaluation of the same cell to determine cell height and the angle of atta chment to the substrate, and a total surface area measurement. (b) From left to right: a cropped section of the original AFM micrograph with (a) 10 x 10 m size depicting filop odia/ lamellipodia, an area measurement of the filopodia/ lamellipodia includin g the cell body, and an area measurement of the cell body only eliminating the f ilopodia/ lamellipodia to determine the percentage of total cell surface area that is c omposed of these features. 3.3 Experimental Results The MTT analysis results for both the H4 and PC12 cell lines are displayed in Figure 3.2. The H4 cell line shows a viability lev el of approximately 80% 5% as compared to the PSt control for all samples excludi ng (111)Si which performed at 108% 9%, and 6H-SiC p-type 47.4% 5%. This result is confirmed through the use of Tukey’s pairs of means test on a sample population, which shows that all surfaces are statistically similar, except (111)Si which is simi lar to PSt, and 6H-SiC p-type which is different from all surfaces. The PC12 line display s results with the (100)Si substrate that were very similar to Coletti et al. [124], however the (111)Si has excellent viability which is equal to that of the 3C-SiC samples. Tuke y’s pairs of means test indicates that the 3C-SiC samples and (111)Si possess a statistica lly similar level of viability to the PSt control. This large discrepancy between the Si sub strates is very noteworthy, and will be discussed in more detail within (§3.2.4). Line Scan Full Cell Area Measure Cropped Area Area with lamellipodia Area without lamellipodi a

PAGE 110

Figure 3.2 : H4 and PC12 cell proliferation on Si and SiC sub strates measured via MTT on the fourth day of culture normalized to the PSt control and expressed as Unlike the previous MTT experiments with fibroblast s and skin cells, Si maintains high level s of viability with the H4 cell line and a low leve l with the PC12 cell line. There is a similar difference between (100) and (111) face mat erials. But surprisingly, 4H 6HSiC do not show the same level of viability as 3C due to the doping of the semiconductors. The hexagonal polytypes of SiC viability with the PC12 cell line viability for these materials is for n 4HSiC sample. Viability increases slightly for intri nsic material and shows a greater increase with ptype material. As we did not have the ability to g row p did not possess ptype Si at t patterns due to semiconductor material doping conti nued, but as the trend was followed Si (100)Si (111) 3C (100) Proliferation (Ratio of PSt Control)99 : H4 and PC12 cell proliferation on Si and SiC sub strates measured via MTT on the fourth day of culture normalized to the PSt control and expressed as Unlike the previous MTT experiments with fibroblast s and skin cells, Si maintains high s of viability with the H4 cell line and a low leve l with the PC12 cell line. There is a similar difference between (100) and (111) face mat erials. But surprisingly, 4H SiC do not show the same level of viability as 3C SiC, and definite patter due to the doping of the semiconductors. The hexagonal polytypes of SiC along with the NCD samples show a very poor viability with the PC12 cell line compared to less than 60% of the PSt. The lowest viability for these materials is for n -type bulk doping, with 6HSiC sample lower than the SiC sample. Viability increases slightly for intri nsic material and shows a greater type material. As we did not have the ability to g row p type 3C type Si at t he time of this study, we could not determine if th e viability patterns due to semiconductor material doping conti nued, but as the trend was followed 3C -SiC (100) 3C-SiC (111) 6H-SiC n-type 6H-SiC p-type 6H-SiC intrinsic 4H-SiC n-type 4H-SiC p-type 4H intrinsic Substrate Tested MTT Assay: 96 Hours H4 : H4 and PC12 cell proliferation on Si and SiC sub strates measured via MTT on the fourth day of culture normalized to the PSt control and expressed as Unlike the previous MTT experiments with fibroblast s and skin cells, Si maintains high s of viability with the H4 cell line and a low leve l with the PC12 cell line. There is a similar difference between (100) and (111) face mat erials. But surprisingly, 4H -SiC and SiC, and definite patter ns emerge the NCD samples show a very poor 60% of the PSt. The lowest SiC sample lower than the SiC sample. Viability increases slightly for intri nsic material and shows a greater type 3C -SiC, and he time of this study, we could not determine if th e viability patterns due to semiconductor material doping conti nued, but as the trend was followed SiC intrinsic NCD H4 PC12

PAGE 111

100 by both hexagonal substrates, it is likely that thi s trend would be consistent for p-type Si and 3C-SiC as well. PSt – Live Cell PSt – Fixed Cell Glass – Live Cell Glass – Fixed Cell Si – Live Cell Si – Fixed Cell NCD – Live Cell NCD – Fixed Cell 4H-SiC – Live Cell 4H-SiC – Fixed Cell 6H-SiC – Live Cell 6H-SiC – Fixed Cell 3C-SiC – Live Cell 3C-SiC – Fixed Cell Figure 3.3: Selected 45 x 45 m AFM micrographs of H4 cells seeded on the substrates tested in this study. Each micrograph possesses a pseudo three dimentional insert in the upper right hand corner to give some perspective on cell morphology in relation to the surface of the substrate. Underneath each micrgrap h is the type of substrate and the condition of the cell. AFM data taken in contact mo de.

PAGE 112

101 The morphology and lamellipodia attachment quality of the H4 cells were evaluated with AFM analysis. Figure 3.3 displays t wo selected AFM micrograph scans of the H4 cell line incubated on each of the tested substrates, with one scan showing a living cell and the other a fixed cell. The H4 cel ls on PSt show a low level of attachment to the surface, possessing an average height to sur face area ratio of 5.40x10-3 and a steep angle of attachment of 30.6. The micrograph shows that the H4 cells on PSt have microtubule extensions which are connecting with ne ighboring cells, and the average lamellipodia extensions from three micrographs cons ist of about 28.9% of the total cell area while the cell was measured alive. The fixed c ells show an average decrease in the lamellipodia area, constituting only 15.4% of the c ell area, which is a reduction of 13.5% as compared to the living cell measurements. The H 4 cells on glass react in a similar manner when compared to the cells on PSt. The aver age height to surface area ratio is 7.04x10-3 and the angle of attachment is 23.8, indicating t hat the cells are retracting from the surface. Although there are large expanses of lamellipodia present with many of the cells on the glass substrate, they show very little attachment to the surface, and therefore little permissiveness, constituting an average of 4 5.5% of the cell area while measured alive and only an average of 8.43% of the cell afte r fixation which is a 27% reduction. The H4 cells on the surface of Si and on NCD posses s similar levels of cell attachment to the surface, but they show differences in the lamel lipodia permissiveness. The H4 cells flatten and expand over a larger surface area on th ese substrates, with ratios of 3.11x10-3, and 3.82x10-3 for Si and NCD, respectively. The angles of attac hment, measured at 15.2 and 13.8 respectively, also indicate that the cell cytoskeletons attach to the surface at

PAGE 113

102 many points causing the cell to spread and flatten over the surface. The H4 cell on Si has small microtubule extensions and shows the lamellip odia extending from multiple parts of the cell. These extensions account for an avera ge of 36.4% of the cell area when measured alive and 26.4% after fixation. NCD shows a slightly better lamellipodia extension, possessing an attachment level with 30.5 % of the cell area before fixation and 22.4% cell area after fixation, and display some ve ry large microtubule extensions which suggest a good permissiveness to this surface. Ove rall, the lamellipodia reduction level for Si and NCD collected from the averaged readings was 10% and 8.1%, respectively. The AFM scans for H4 cells on the surfaces of 3C-S iC, 4H-SiC, and 6H-SiC show very different morphologies, indicating that t he biocompatibility of SiC cannot be treated as a single material as was previously sugg ested, but must be examined on the basis of the different polytypes [18-20]. The H4 c ells on 3C-SiC flatten, expand and elongate over a large surface area showing an avera ge height to area ratio of 4.58x10-3 m-1 and an angle of attachment of 16.3. For 3C-SiC l arge microtubule extensions on many of the cells were evident with multiple smalle r extensions present. The lamellipodia, which measured 15.1% of the living ce llular area, was the only sample group tested with the H4 cell line that had an aver age increase in area percentage with an average area measurement of 15.4% after the cell wa s fixed. This is indicative of a lightly permissive substrate and explains the prese nce of the multiple microtubule extensions, which are a normal function of the H4 c ell, and suggests that the surface of 3C-SiC possesses innate molecular characteristics r equired by these cells. The H4 cells seeded on 4H-SiC are morphologically rounded or sli ghtly triangular with an average

PAGE 114

103 height to area ratio of 4.21x10-3 m-1 and an angle of attachment of 17.3. Microtubule extensions are either non-existent or of very short length from the soma, and the lamellipodia that measured 17.3% of the living cell ular area reduced to 8.4% after the cell was fixed, showing a decrease of 8.9% in area. Thi s indicates that the surface of 4H-SiC is not as permissive to these cells as was seen wit h the 3C-SiC polytype, and surface science investigation of the differences between th e surface could shed light on the physical cues that the cells use for membrane attac hment. The cells on 6H-SiC show slightly different characteristics than their 4H-Si C counterpart with a more elongated, ellipsoidal morphology and multiple small microtubu le extensions. The cells are elevated from the substrate, possessing an average height to area ratio of 4.78x10-3 m-1 and an angle of attachment of 19.9, but the lamellipodia, which have smaller extensions, show an attachment measuring 5.8% of the living cellular area while it was 2.35% after the cell was fixed, which is a small decrease of 3.5% in are a. The presence of the microtubules and the smaller lamellipodia retreat indicate that the surface of 6H-SiC possesses better surface characteristics than that of 4H-SiC, but th e lack of large microtubule extensions and the better lamellipodia extensions show that 3C -SiC has better biocompatibility than 6H-SiC. The PC12 cells have much different attachment char acteristics than the H4 cells which may be most likely due to the fact that they are derived from adrenal cells and are a neuronal modeling line. This cell does not chang e into a peripheral neuron without the addition of neural growth factor (NGF), an expensiv e neural factor that would have made this study cost prohibitive. The Glass, Si, and NC D substrates show lower attachment

PAGE 115

104 characteristics for this cell line. The cell heigh t to area ratios are 6.85x10-3, 5.14x10-3, and 5.88x10-3 for the glass, Si, and NCD, and their angles of at tachment are 31.2, 31.5 and 29.9, indicating the cells are trying to minim ize their contact with these substrates. This analysis has further confirmation as each of t hese substrates show poor lamellipodia attachment. Glass had few or no microtubule extens ions but extensive lamellipodia extensions covering 45.9% to cell area when measure d alive and only 21.1% after fixation, a reduction of 24.8%. NCD was similar to glass in that it had few microtubule extensions and also had many extensions of lamellip odia covering 37.6% area for the living cell and 18.1% cell area for the fixed cell with a reduction of 19.5%. Although Si displayed good extensions of microtubules, it has a lamellipodia area reduction of 10.2 % as indicated for the 25.0% and 14.8% live cell and fixed cell lamellipodia measurements. The PC12 cells showed a good permissiveness with PS t having a height to surface area ratio of 2.65x10-3 and an angle of attachment of 22.6. This materia l allows the cell to project multiple microtubule extensions and had a 3 .3% increase in lamellipodia attachment area due to the living cell area lamelli podia percentages were 25.9% and 29.2% for the living and fixed cells, respectively. As with the H4 cells, the PC12 cell line had very different attachment and morphology statistics over the three tested polytyp es of SiC. The PC12 cells on 3C-SiC have a slightly lower height to surface area ratio of 4.80x10-3 m-1 and a higher angle of attachment of 25.8 when compared to the PSt, but t he morphologies of the cells were very similar. The cells show elongated ellipsoidal shape with two or more extensions of microtubules. The lamellipodia showed an attachmen t factor composing 29.3% of the

PAGE 116

105 PSt –Live Cell PSt –Fixed Cell Glass –Live Cell Glass –Fixed Cell Si –Live Cell Si –Fixed Cell NCD –Live Cell NCD –Fixed Cell 4H –Live Cell 4H –Fixed Cell 6H –Live Cell 6H –Fixed Cell 3C –Live Cell 3C –Fixed Cell Figure 3.4: Selected 45 x 45 m AFM micrographs of PC12 cells deposited on the substrates tested in this study. Each micrograph p ossesses a pseudo three dimentional insert in the upper right hand corner to give some perspective on cell morphology in relation to the surface of the substrate. Undernea th each micrgraph is the type of substrate and the condition of the cell. cell area for the living cell and 24.8% of the cell area for the fixed cells which amounts to a reduction of only 4.5%. Although this is not as good as the permissiveness shown with the H4 cell line, it is still relatively good and i ndicates that 3C-SiC is a good biomaterial with good permissiveness. The AFM measurements for 4H-SiC show that the PC12 cell line does not have good biocompatibility with this surface. The cells show a similar

PAGE 117

106 attachment to this surface when compared to Si and 3C-SiC, possessing a height to area ratio of 4.71x10-3 m-1 and an angle of attachment of 24.2. The cells sh ow an extremely rounded morphology and possess smaller microtubule extensions than 3C-SiC. The lamellipodia show similar reactions to this surface with Si, composing 19.2% of the cell area for the living cell and 8.8% of the cell area for the fixed cells. This reduction of 10.4% shows that the surface is not as permissive t o this cell line as PSt and may not be suitable for long term implants. PC12 cells seeded on 6H-SiC show generally rounded cell morphologies with few microtubule extensions. The cells show an average height to area ratio of 7.40x10-3 m-1 and an angle of attachment of 42.4 which indicate s that they do attach well to the surface of this material. Th e PC12 cells show extensions of lamellipodia from many parts of the cell, but they are small in comparison with the entire cell body. The living cells show lamellipodia that composed about 4.24% of the cell area when measured alive, and this value reduced to 3.51 % when the cell had been fixed. The values measured by the AFM and the MTT viability va lues show that 6H-SiC is not a suitable material for long term implants. During the AFM measurement of both cell lines depo sited on the various substrates, the Si surface was observed to suffer p hysical damage. Figure 3.5 displays AFM micrographs of the surfaces of three tested sem iconductor substrates before cell deposition and their morphology after the cells had been deposited for a time of 48 hours. For each of the substrates seeded with cells, any c ellular remnants and extracellular proteins were completely removed through the piranh a cleaning method described in the material and methods section (§ 3.2.2). Figure 3.5 (a) displays an AFM micrograph of the

PAGE 118

107 Si surface after the initial RCA cleaning and subse quent piranha cleaning that occurs before the seeding of the cells. Commercial Si sur faces are chemical-mechanical polished (CMP) to an atomically flat surface, indic ated in Figure 3.5(a), which shows a surface morphology marked by atomic-scale deep poli shing scratches, and it possesses a surface root mean square roughness, rq, within a 5 x 5 m scan size of 0.137 nm RMS. The surface of the Si after cell seeding and remova l shows many large elliptical depressions approximately 100 – 200 m long which r oughly correlate to the shapes of the H4 or PC12 cells as seen in Figure 3.3 and Figu re 3.4 The surface within these depressions, which is displayed in Figure 3.5(b), s hows numerous small, rounded surface pits with diameters of 500 to 700 nm and 30 to 70 n m deep. The 5 x 5 m scan size rq had increased to 18 nm RMS from 0.137 nm RMS. 3C-SiC and NCD are known to be chemically inert ma terials and they therefore do not display the changes seen on the surface of t he Si substrates. The ‘as-grown’ surface of 3C-SiC, displayed in Figure 3.5(c), show s a rough, mosaic surface due to defects caused by the 20% lattice mismatch between 3C-SiC and Si, and it possesses a 5 x 5 m scan size rq of 2.46 nm RMS. The deposited NCD on Si surface d isplayed in Figure 3.5(e) has many approximately 50 – 100 nm di ameter crystallites marked with larger 1 m long boulder defects. The NCD 5 x 5 m scan size rq of 17.79 nm RMS was a direct result of the uneven nature of this surfac e. Figure 3.5(d) and Figure 3.5(f) show the 3C-SiC and NCD surfaces after cell removal, and the surface seemed to have little effect on the cells. The 5 x 5 m scan size rq of the 3C-SiC was 2.46 nm RMS and of the NCD was 13.42 nm RMS, showing little change in roug hness, and was only different as

PAGE 119

108 it is difficult to find the same area of measuremen t after the initial scan has been performed and the sample removed from the AFM stage .. (a) (b) (c) (d) (e) (f) Figure 3.5: 5 x 5 m AFM micrographs of the surfac es of the three tested semiconductors. The micrographs are presented with the first column displaying the morphology of the semiconductors before cell seedin g and the second column represents the morphology of the semiconductors after cell rem oval and subsequent cleaning. (a) and (b) are the Si surfaces, (c) and (d) represent the 3C-SiC surfaces, and (e) and (f) are the NCD surfaces. 3.4 Discussion of Results Table 3.1: Compiled values representing the cell i nteraction with the tested substrates as measured through AFM techniques. Each cell line is listed followed by a qualitative ranking of each substrate for negative and positive cell interaction. Cell Interaction with Substrate Negative Interaction Positive Interaction H4 Cell Line Glass PSt Si 4H-SiC NCD 6H-SiC 3C-SiC Live Cell Height/ Area (10-3 m2) 7.04 5.40 3.11 4.21 3.82 4.78 4.58 Angle of Attachment ( ) 23.8 30.6 15.2 17.3 13.8 19.9 16.3 % Area: Lamellipodia and Filopodia Reduction 27 13.5 10 8.9 8.1 3.5 -0.3* PC12 Cell Line Glass NCD 6H-SiC 4H-SiC Si 3C-SiC PS t Live Cell Height/ Area (10-3 m2) 6.85 5.88 7.40 4.71 5.14 4.80 2.65 Angle of Attachment ( ) 31.2 29.9 42.4 24.2 31.5 25.8 22.6 % Area: Lamellipodia and Filopodia Reduction 24.8 19.5 12.7 10.4 10.2 4.5 -3.3* Note: Negative numbers indicate an increase in l amellipodia and filopodia area.

PAGE 120

109 This study focused on gaining insight into the bio compatibility of bare, untreated substrates with neuronal cell lines, and primarily used two factors to determine the level of biocompatibility. MTT assays, shown in Figure 3. 2, were used to determine cell proliferation and viability, and AFM measurements, presented in (§ 3.3) and summarized in Table 3.1, were used to determine cell permissiv eness to the substrate. AFM measurements quantified both the cell to substrate interaction and the permissiveness of the substrate. Cell proliferation and subsequent membrane attachm ent on substrates is dependent on a complex combination of physical, ele ctrical, and chemical reactions which are influenced by both surface and bulk subst rate properties and the cell membrane structure [119, 120]. Cellular attachment to a sub strate involves the absorption of proteins contained within the media serum to its su rface followed by receptors within the cellular membrane binding to the absorbed proteins [131]. Subsequent membrane attachment, membrane spread, cytoskeleton polymeriz ation, and cellular growth are controlled by the internal cellular cues initiated by these intracellular receptors involved in substrate-cell binding, and it has been demonstr ated that this is heavily mediated by surface wet-ability and charge [131, 132]. Hydroph ilic and charged surfaces allow extracellular proteins, like fibronectin (Fn) which are involved in cell membrane to surface binding, to unravel and elongate as it is a bsorbed onto the surface, whereas hydrophobic surfaces show Fn is absorbed in natural soluble form which is compact and rounded [133]. Cells show little reactivity with F n in solution, so the chain elongation is one important factor for cell membrane reactions to occur [133]. Furthermore, cell

PAGE 121

110 charge and surface wet-ability have been observed t o be very influential factors for both surface interaction and neurite growth and guidance Growth cone formation and filopodia/ lamellipodia guidance have been shown to favor a more hydrophobic surface [134]. Neurite formation and extension are also in fluenced by substrate charge, with this growth favoring positive charge over negative charg e, and neutral charges produces almost no neurite outgrowth [135-138]. PSt treated through oxygenation for optimal cell a ttachment was used as a normalizing control for the MTT assay test and as a comparison for cell attachment and permissiveness in the AFM analysis. Proliferation of the H4 and PC12 cells was higher than the tested substrates with the exception of (1 11)Si for the H4 cells. The level of attachment of the cells to the PSt substrate showed some variation in that the AFM analysis indicated that the PC12s showed a low surf ace profile and large surface area spread for the cell membrane, while the H4 cells ha ve a large angle of attachment, a subsequent high surface profile, and less surface a rea membrane attachment. The lamellipodia permissiveness was also much better fo r the PC12 cells than the H4 cells on this surface. This material has been shown to be n ormally hydrophilic, as indicated by Kim et al ., and the material data sheet provided by the supp lier, Corning Inc., specifies a H20 contact angle of 12.3 16.3 [132]. This material has been also shown to often possess a negative surface charge [133]. The hydro philic surface and the negative charge of this material would allow for elongation of atta ched proteins, and therefore could be a reason the cells have high proliferation to this pa rticular surface. The negative charge could also be a reason for the improved cell attach ment and lamellipodia extensions for

PAGE 122

111 the PC12 cells, as was also found by Fine et al. and Schmidt et al. [133, 138] However Clark et al showed evidence that CNS neurite growth was more sensitive to charge when compared to membrane attachment and favored a more positively charged environment, and this may be a reason for the decreased H4 cell and lamellipodia attachment [134]. The untreated amorphous glass substrate surface wa s evaluated in this study as it has been used in traditional neural MEMS devices an d is more hydrophobic than 3C-SiC and slightly less hydrophobic than Si, PSt and NCD, respectively, as seen by Sklodowska et al. which reports a contact angle of 51.05 [139]. Am orphous glass would also not provide a net substrate charge as it is an insulati ng material, but any charges that contact the surface would potentially remain in place as gl ass is a good insulator. AFM analysis showed that both cell lines tended to greatly minim ize their contact with this substrate and have little lamellipodia surface attachment, al though cells on this substrate displayed the largest areas of overall lamellipodia probing. The performance of the cells on this substrate is consistent with the previously discuss ed study results in that the extracellular proteins would probably remain in a natural form as they are absorbed onto the surface, therefore providing a lower number of membrane bind ing points [10, 133]. The neutral charged surface could be a large factor in inhibiti ng the neurite expansion, as was shown by previous studies [135-138]. Both faces of Si, which had been cleaned using the protocols discussed in the methods section, have been shown by Coletti et al to have a measured contact angle of 65.85, indicating a weakly hydrophobic surface [12 4]. The samples used in this study were doped with phosphorus which provides a negativ e charge potential within the bulk

PAGE 123

112 of the material. The H4 cell line showed good viab ility for both Si surfaces with (111)Si showing a higher viability level than (100)Si to bo th cell lines. The H4 cells displayed a good level of cell membrane attachment but only a m oderate lamellipodia attachment. A contradiction in the results was seen with the PC12 cell line interactions with the Si surfaces which indicated poor cell viability for (1 00)Si, but excellent cell viability for (111)Si. The AFM results support the MTT findings for (100)Si with a high cell profile, a large angle of attachment, and moderate lamellipo dia attachment, but these same results does not support the MTT assay findings for Si(111) As Si is nearly as hydrophobic as natural PSt, and the samples would possess the same negative charge potential, the differences between the reactions of the cells to t hese two substrates may be due instead to the chemical properties of their surfaces. The evidence that this may be a plausible assumption is displayed in Figure 3.5(b), which ind icates some chemical reaction between the cell membrane and the Si which has pitt ed the surface. Both of the Si substrates tested have different surface elemental Si concentrations, where (111)Si has three exposed but bonded Si atoms compared to the s ingle bonded Si atom on (100)Si. Also, studies have indicated the presence of Si com pounds present in the extracellular matrix of the CNS, and Si compounds are also found in blood plasma [11, 77]. The results of this study indicate that both cell lines may have a preference to bond with Si compounds, which may be the reason for the much hig her proliferation on the (111)Si substrate than the (100)Si substrate, and the H4 ce ll line shows a preference to the (111)Si substrate that rivals PSt with higher viabi lity and tight surface attachment. The lower viability of the PC12s on (100)Si is not so e asily explained, but a lower membrane

PAGE 124

113 attachment was shown by the AFM analysis which indi cates that this substrate is not entirely favorable for this particular cell line. It is well known in the semiconductor industry that (100)Si reacts at much higher rates w ith many reagents than (111)Si, and the Si substrate immersed in solutes like water form si licic acid, which can have a negative effect on the cell viability when present in large quantities [9-11, 77]. The moderate lamellipodia attachment to these substrates is most likely due to the negative charge of the substrate, and mirrors the results seen by Fine Makohliso, Young, and Schmidt et al. but attachment of the lamellipodia may have been al tered due to the hydrophobic substrate surface [134-138]. Intrinsic NCD showed a similar performance profile with the H4 cells as (100)Si by possessing a good level of viability, and the AF M scans showed a good level of cellular membrane expansion and attachment to this substrate. Alternatively, also like Si (100), there was a poor level of viability with the PC12 line and the high cell profile shows it does not favor this surface. The results of the lamellipodia attachment show this material has similarities to (100)Si as well. H4 c ells had a moderate level of lamellipodia attachment while the PC12s showed a very poor level of lamellipodia attachment. Hydrogen treated NCD has been shown by Ostrovskaya et al (via contact angle measurements of 75 95) to be a hydrophobic surf ace which is mainly due to the large amount of sp2 bonded graphitic defects along the crystal grain b oundaries [140]. As this material is intrinsic like glass, there should be n o appreciable surface charge, but some random charges may be present along defect boundari es and may be attracted by capacitive forces to the surface from charges in th e Si semiconductor substrate. The

PAGE 125

114 PC12s follow the previous studies of Rouhi and Kim et al which show a low level of cellular viability and attachment due to a hydrophi lic surface [131, 132]. The low level of lamellipodia attachment, almost equal to the lev el of amorphous glass, may be due to the intrinsic nature of the substrate as was seen i n previous studies [135-138]. These conclusions do not account for the cellular attachm ent and viability seen by the H4 cell line. These cells should not be attracted to a sur face as hydrophobic as NCD, but as the surface was grown in hydrogen ambient, the proteins may actually be attaching to the hydrogen on the surface instead of the carbon benea th. Protein and polymer surface modification has used hydrogen or O-H bonded surfac e terminations to attach specific proteins to the surface of semiconductors, and it m ay be that proteins in the media are attaching to the H-bonded surface of the NCD [31, 1 16, 117]. Further exploration into the surface properties of NCD and how this material interacts with extracellular proteins would be necessary to confirm this hypothesis. Both crystal orientations of 3C-SiC, unlike Si, ha ve high proliferation values with both cell lines when compared with the PSt control and the surface, shown in Figure 3.5(d), showed no adverse effects or surface degrad ation through cellular interaction. AFM indicated that the H4 cells have good cellular spread and low profile when deposited on the surface, and excellent lamellipodi a permissiveness. A low angle of attachment and moderate level of cell spread was ob served with the PC12 cells, and a relatively good level of lamellipodia attachment wa s also observed. The same discrepancy in the level of viability that was obse rved between the different crystal orientations of Si samples also exists between (100 ) 3C-SiC and (111) 3C-SiC, although

PAGE 126

115 it is not as pronounced. Coletti et al has shown the USF ‘as grown’ 3C-SiC has a contact angle of 52.53, and Yakimova et al indicated that 3C-SiC contains both hydrophilic a nd hydrophobic sites due to its alternating elemental surface from both island nucleation and defect structures [20, 124]. C rich 3C-SiC surface s are extremely hydrophobic and Si rich 3C-SiC is hydrophilic [124]. It should be not ed that these substrates were negatively charged throughout their bulk by the incorporation of nitrogen present during the growth process, although at a level much lower than the in tentionally doped Si substrates. The fact that this substance is a hydrophilic material does not alter the viability level significantly from Si, which was hydrophobic; along with the observation that there is a similar viability contrast between the crystal orie ntations suggest that the cellular membrane may have some competing bonding principals Our earlier hypothesis is that Si provides greater protein absorption and is utilized in the extracellular matrix [9-11]. 3C-SiC, like S i, has more atoms per unit cell on the (111) face than the (100) face, and if the cell mem brane prefers to attach to Si, the viability will be higher on the (111) face. As ind icated by Figure 3.2, both Si and 3C-SiC possess this pattern difference in the viability be tween the two crystal orientations, but 3C-SiC has slightly lower viability than Si. NCD h as lower viability than both 3C-SiC and Si, and is also a cubic faced material. NCD po ssesses wither C or C-H bonds on the surface, and a lower viability level than the (100) Si and (100) 3C-SiC faces, suggesting that membrane affinity is lower for elemental C tha n Si. As 3C-SiC falls between the two mono-elemental surfaces, the viability differen ce may be due to the different membrane affinities to the alternating Si-C bonds o n the surface. 3C-SiC, like the PSt

PAGE 127

116 and Si tested in this study, possesses a negative m aterial charge, and presence of hydrophobic sites from Si on 3C-SiC surface would g enerate good lamellipodia extension, but the greater attachment of the lamell ipodia may be due to the hydrophobic C sites on the surface. The hexagonal polytypes of SiC, 4H-SiC and 6H-SiC, do not behave like their 3C-SiC counterpart with either cell line. Although 4H-SiC has a statistically equivalent viability level when compared to the 3C-SiC polytyp e for the H4 cell line, the AFM analysis shows circular to slightly elongated cells with a slightly elevated angle of attachment. The cells possessed small microtubule extensions and had a moderate surface permissiveness when compared to 3C-SiC. Th e PC12 cell line had a much lower viability with 4H-SiC than with 3C-SiC, which was m ore statistically comparable with the NCD and (100)Si substrates. Cells show more me mbrane attachment to this surface than many of the other substrates, but the lamellip odia attachment to the surface was poor. The other hexagonal substrate, 6H-SiC, had t he worst performance of all the materials tested, which is contrary to Yakimova’s et al findings which shows this material had a good biocompatibility level [20]. T he viability of H4 cells was lower than that on 4H-SiC and this material had the lowest via bility with PC12 cells. AFM shows that the cells on both of these substrates were rou nded to elliptical in shape with small microtubule extensions. The H4 cells had few exten sions of the lamellipodia, but they were relatively firmly attached to the surface. Th e PC12 cell line showed little lamellipodia extensions as well, but this surface d id not have the same level of permissiveness as the H4 cells did.

PAGE 128

117 Although 4H-SiC and 6H-SiC did not show the same l evel of biocompatibility as was displayed by 3C-SiC, they display the effects o f bulk charge on each of the cell lines. As seen in Figure 3.2, both substrates follow the s ame patterns when viewed by charge in respect to viability, and both cell lines have very different preferences. The H4 Cells show little difference between surfaces charged neg atively or which are neutral, but have a definite dislike of positively charged materials. PC12s, on the other hand, favor positively charged material, and to a slightly less er degree, intrinsic (neutral) material, but has a dislike of negative charged surfaces. It is interesting to note that both the Si and 3C-SiC substrates are negatively charged and still possess high viability. Later experimentation should be conducted to determine th e charge effect with 3C-SiC and Si surfaces. The fact that 4H-SiC and 6H-SiC react with the neu ral lines much differently than 3C-SiC is very difficult to explain. Ideally, 3C-S iC, 4H-SiC, and 6H-SiC all possess the same surface characteristics with the exception of unit cell spacing which would only affect the atomic step spacing as detailed in (§ 2. 2). The Si-C bi-layer stacking of all the polytypes ideally create on the on-axis surface of the semiconductor a pure layer of Si atoms only, commonly designated the Si-face, and th e opposite surface is C atoms only, or the C-face. Commercially available SiC, 4H-SiC and 6H-SiC, indicate these surfaces as both faces have different properties and process ing techniques [141]. The samples utilized in this study are Si-face 4H-SiC and 6H-Si C, which was the same type covered in Coletti et al ., and should have a very hydrophilic surface with contact angles in the range of 20.3 for 4H-SiC and 25.2 for 6H-SiC [124]. The reaction of the SiC n-type surfaces

PAGE 129

118 is similar to the reaction of both (100)Si and NCD, two materials much less hydrophilic than SiC. The contradictions of these two material s as compared with the previous findings used for this study warrant more study int o the nature of the semiconductor surface in relation to the surface of a cellular me mbrane. As n-type 6H-SiC produced the lowest reaction, this material is an excellent cand idate for future studies. 3.5 Summary of Chapter This initial study of neural cell interaction with 3C-SiC and NCD has provided evidence that these materials may be suitable for u se in an implantable BMI device. The results of this study indicate that 3C-SiC is very biocompatible to both cell lines and showed a relatively high level of cellular viabilit y as compared to PSt treated for optimal cell growth. Cell morphology results showed that t he cells attach very well to the surface of 3C-SiC and displayed excellent lamellipodia perm issiveness, which is an important quality for BMI devices. Furthermore, the chemical resilience of 3C-SiC demonstrates that this material resists the surface degradation that was observed on the Si surfaces which resulted from cellular interaction with its s urface. The as-grown NCD showed good viability and moderate permissiveness for the H4 cell line, but has a poor performance with the PC12 line. This result may be due to the fact that the NCD is hydrogen rich at the surface, but further investiga tion is needed to examine the neuron cell reaction on this substrate. Many of the neura l reactions we observed to these substrates are similar to previous studies; however many of the explanations utilized to explain the results are conjuncture and will have t o be verified.

PAGE 130

119 Future exploration which is proposed in (§ 4.2) wi ll explore the surfaces of SiC and NCD and (§ 4.3) will propose an exploration of neural molecular interaction with 3C-SiC surfaces by examining the bonding of extrace llular proteins to substrates. Experiments with alternative substrates or the util ization of surface modification with different processes will be used to explain the sur face interaction. Verification of the charge effect on the cellular attachment using intr insic, n-type and p-type doped Si and 3C-SiC substrates would provide more data for the r eaction of the cells membrane attachment and lamellipodia attachment to different charge factors within the bulk substrate. Ostrovskaya et al shows that the hydrophobic / hydrophilic nature o f NCD can be modified through hydrogen and oxygen treatme nt after growth, so establishing an oxidation method for our NCD without etching would help give further evidence for the cell membrane reactivity with [140]. Finally, an e xamination of the extracellular remnants that remain after cell removal by AFM, XPS and FTIR, then removal by centrifuge for examination using Western blot, or a nother protein separation test, could provide evidence as to the molecular interactions o f the cell membrane and the substrate.

PAGE 131

120 Chapter 4: Biocompatibility Improvement and Device Interaction s 4.1 Introduction Thousands of people are suffering from irreparable damage to the CNS which has one response to damage regardless of the cause th e glial scar. The glial scar restores the blood-brain barrier, the system that protects the C NS from external viral attacks. Unfortunately, the scarring process does not repair the neural signal pathways, inevitably leading to loss of body functionality and leaving t he individual permanently handicapped. A proven therapy is available for the treatment of some of these conditions, the BMI, and implantable systems allow two-way communication wit h targeted individual neurons. Implantable BMI devices have shown that this techno logy can provide treatment for many problems with the CNS, but unfortunately it ha s one major Achilles heel. The implantable neural interface has limited lifetime d ue to the very process it is trying to repair, the glial scar. Science has agreed that th e major cause of this problem is that the biocompatibility of the device and the materials of construction lead to a foreign body response from microglia and astrocyte cells. Findi ng a biocompatible material for these devices has become paramount for their continued ap plication in the treatment of problems with the CNS system and this dissertation research has been focused on this goal.

PAGE 132

121 3C-SiC, or cubic silicon carbide, and NCD, nanocry stalline diamond, are two possible materials for BMI construction. Both of t hese materials have properties that make them ideal for use as a permanent neural impla nt for the BMI system. The materials are mechanically resilient, chemically in ert, and can be deposited on many different substrates. These materials are semicond uctors which allow the customization of bulk electrical charge. One problem with the us e of these materials is that they possess many crystalline defects that limit the electrical performance of electronic devices such as transistors and diodes. We have shown in (§ 2.2) t hat the quality of 3C-SiC was drastically increased utilizing low pressure CVD me thods we developed. More importantly, we have also shown in (§ 3) through bi ocompatibility tests suggested by ISO 10993 that 3C-SiC shows an innate biocompatibility which is statistically comparable to cell treated polystyrene, PSt, with human CNS glial cells, and rat cells derived from the adrenal gland that correspond with cells that conne ct the CNS to the PNS. The same tests show that NCD is a good material for interacting wi th the CNS glial cells, but has some problems interacting with the adrenal cells. These same tests revealed some unexpected cellular behavior on the Si substrate and the alter native forms of SiC, 4H-SiC and 6HSiC. Our conclusions from the tests have suggested that future investigations must be conducted before an implantable 3C-SiC/ NCD device can be constructed and tested in vivo The first set of proposed experiments will involve a substrate which showed the lowest biocompatibility, 6H-SiC. 6H-SiC was tested using the same protocols as placed upon the materials tested in (§ 3). (§ 4.2) will d etail methodology to modify the surface

PAGE 133

122 of 6H-SiC, and utilizing methodology covered in (§ 3), we will determine if any of the proposed modifications improves biocompatibility. (§ 4.3) will describe our work with primary cultures of mouse cortical neurons, which w ill be the first direct CNS neuron we have evaluated. The cells used for the work presen ted in (§ 3) were derived from cancerous cells, and thus have the ability to proli ferate, but it is well known that these cells react slightly differently to electrical stim ulus than primary, animal derived noncancerous cells. Finally, in order to show that 3C -SiC and NCD are materials that can provide interaction with neurons, (§ 4.4) will deta il our current experiments involving the neural activation device, or NAD. The NAD will uti lize 3C-SiC as the base of an electrode-based device modeled after the Michigan p lanar probe as a basis, and will be used to send signals to neurons in order to elicit an action potential. The data obtained from the initial NAD, namely the electrical charact erization and interaction statistics with neural cells, will allow for the design of a new ge neration of NAD’s utilizing field effect devices instead of the electrode devices which is t he subject of future work in (§ 5). 4.2 Semiconductor Surface Modification and Exploration (§ 3.3) detailed the results of the MTT assay viab ility tests along with the cell morphology and lamellipodia permissiveness to the s ubstrate evaluation as assessed by AFM. The results for many of the substrates tested were inconclusive as to the exact mechanisms that led to the observed results. Clear ly, more exacting measures were needed to evaluate how proteins were interacting wi th the surface of the semiconductor and, in turn, how the cells themselves interacted w ith the absorbed proteins. The results

PAGE 134

123 contained in (§ 3.3) and the literature supported e xplanations in (§ 3.4) detail that proteins absorb onto the surface of a foreign material based on its chemical composition, which dictates reactivity and surface wettability, and wa ter retention. Another important factor for protein absorption and activation is the influe nce of electrical charge. As the materials we were testing are semiconductors, it is possible to alter the electrical characteristics by changing the doping impurity lev els either during growth or through the addition of dopants during epitaxial growth. T he goal of the next aspect of this work was to modify the surface chemical composition of t he tested substrates to better understand how cells attach to the surface of these materials. With better knowledge of how the surface state influences cell attachment an d viability, we can design specific modifications that will improve the biocompatibilit y of SiC with neural cells. For many years, it has been recognized that the es tablishment of an inorganic/ organic interface is necessary for biological senso r applications, and this environment is more important still with any implantable device wh ich will interact with the body for an extended period [4 – 6]. Thus studying the behavio r of cells and proteins at a molecular level is crucial to our understanding of this inter face. One method that has been gaining interest is the use of modified surfaces to test th e cell or protein interaction. Specific covalent attachment and subsequent surface function alization of organic molecules on inorganic semiconductor devices enable the targetin g of specific factors that cause cancer, lead to heart attacks, and a myriad of othe r biomarkers that would allow early detection and prevention of many diseases. Stutzma nn et al. at the Walter Schottky Institut (WSI) within the Technische Universitt M nchen in Germany has been

PAGE 135

124 functionalizing semiconductor surfaces using self a ligning organic molecules [142]. Stutzmann et al. utilizes a hydrogen-terminated semiconductor to co valently bond organic molecules in solution to the surface [142]. Predominately, Si has been utilized due to the ease of terminating the surface with hyd rogen; however Stutzmann et al theorized that SiC is electrically matched to organ ic systems because of its band gap size and position, and would make an interesting inorgan ic/ organic heterojunction [142]. Beginning in January 2009, we started a research co llaboration with this WSI group in order to study the functionalization and subsequent improvement of the biocompatibility of SiC. (§ 4.2.1) explains the current work on the functionalization of 6H-SiC using covalent bonding of organic molecules, while (§ 4.2 .2) will detail the results of the initial biocompatibility testing of these surface modificat ions. 4.2.1 6H-SiC Surface Modification Characterization 6H-SiC was tested using the same methodology as pr esented in (§ 3). Along with n-type semiconductor samples of the 6H-SiC, we were able to secure CMP polished ptype, n-type and intrinsic 4H-SiC samples for this study. The results, presented in Figure 3.2, were quite surprising in that 6H-SiC and 4H-Si C had either lower or extremely low viability levels with the two tested cell lines. A nother interesting difference was the effect of semiconductor bulk charge which produced different results with the two cell lines. The H4 cell line seems to be in favor of ntype doped and intrinsic SiC, while the PC-12 cell line clearly favors p-type over intrinsi c or n-type material. These trends were similar with both 4H-SiC and 6H-SiC, but at this ti me we have not correlated the viability

PAGE 136

125 trend with Si or 3C-SiC; however that work is being conducted presently by A. Oliveros in our laboratory with S. Schoell at WSI and will b e reported when complete.. Negatively doped (i.e., n-type) 6H-SiC was shown i n our MTT assays and AFM analysis to have the lowest viability with both of our neuronal cell lines and very poor cell attachment, and virtually no lamellipodia exte nsion for the PC12 cell line. This material is under study for surface modification by Schoell et al at WSI, and the results of their recent research was presented at the Europ ean Conference on Silicon Carbide and Related Materials (ECSCRM) conference in Barcelona, Spain in September 2008 [143]. Our two groups decided to functionalize the surface of 6H-SiC and test it with these two cell lines. To functionalize the surface of a semiconductor wi th a covalently bonded organic molecule, one needs either a hydrogen or hydroxide terminated surface. Hydrogen terminated Si surfaces have been achieved through s urface treatment with diluted hydrofluoric acid or through high purity hydrogen e tching [144, 145]. A monolayer of self assembled molecules (SAM) can be added covalen tly to the surface of the semiconductor using processes known as hydrosilylat ion or silanization. Hydrosilylation involves reacting long chain alkenes to the surface of the semiconductor through reactions with the hydrogen terminations at the sur face, which breaks the alkyne double bond into an alkyl radical, and this radical forms a Si-C covalent bond with the surface [143, 146]. Silanization is a similar process but instead involves the surface and is terminated with a hydroxide which then interacts wi th organosilicon molecule chains which has one end of the chain containing Si bonded to alkoxyl groups. O-H

PAGE 137

126 terminations react with the bonded alkoxyl groups, releasing the hydrogen to the alkoxyl groups, reducing them to Si-O-SI covalent bonds at the surface and between the end groups [143, 147]. These layers are relatively sta ble and can be further manipulated using lithographic techniques [143]. Oxygen plasma can be used to remove the molecules from the surface, and other biological mo lecules can be attached to the end groups of the SAM [143]. The methods of hydrosilylation or silanization can be used to terminate a SiC surface with a SAM, but certain differences are evi dent. First, treating (0001) 6H-SiC (i.e., the Si face) with HF does not create a hydro gen terminated surface as seen with Si, but creates a weakly O-H terminated surface which d eteriorates in a matter of hours [148]. (000-1)6H-SiC (i.e., the C face) surfaces s how a mix of O-H terminated and H terminated surfaces [148]. O-H termination is idea l for silanization, and hydrosilylation is still possible, but the reaction leads to a less stable, non-interfacial Si-O-C bonding condition, and can lead to possible defective bondi ng at the C site from CHx species [143]. A more ideal condition for hydrosilylation is achieved through the H termination that emerges from hydrogen etching [110-113]. This process first generates a well ordered, atomically flat surface free of defects fr om CMP scratches through the hydrogen etching process covered in (§2.3), and subsequent c ooling of the sample in ultrapurified hydrogen [110-113] terminates dangling bonds on the ordered surface with hydrogen bonds. 6H-SiC(0001) terminated in this manner has been shown to be relatively stable for at least 2 days, however Si(000-1) have shown i ncreasing hydrocarbon contamination after only a few hours [148].

PAGE 138

127 Schoell et al have used diluted HF to O-H terminate the surface of (0001) 6HSiC, and hydrosilylation the surface was achieved u sing an alkyl molecule, 1-octadecene (C18H36)which is based off their previous work with Ge [14 9]. Our hydrogen etching process and subsequent H termination of 3C-SiC shou ld provide an optimal surface for the hydrosilylation of 6H-SiC [110-113]. Schoell s ent us samples of (0001) 6H-SiC and (000-1) for hydrogen etching. We etched the sample s using the parameters listed in Table 2.2, and cooled them in purified H2 at 150 Torr to well below 400 C (which is our pyrometer’s lowest temperature reading value) befor e performing an Ar purge and repressurization of the system to 760 Torr. Figure 4.1: ATR-FTIR measurements of (0001) and (0 00-1)6H-SiC surfaces that have been hydrogen etched and terminated. After two wee ks the (0001)6H-SiC shows oxide formation, however, the (000-1)6H-SiC shows an orde red hydrocarbon termination and no oxide formation. ATR-FTIR courtesy of M. Hoeb o f the Walter Schottky Institut.

PAGE 139

128 The samples were immediately removed, carefully st ored in individual sample cases, and placed in an air-tight container which w as purged with N2 for 1 minute before sealing. The samples were delivered to WSI and mea sured using attenuated total reflection Fourier transform infrared (ATR-FTIR) sp ectroscopy. The measured results in Figure 4.1 show some very interesting results that differ from that presented in the work of Tsuchida et al [148]. First, the (0001)6H-SiC surface, or the S i face, shows a definite oxide peak and shows considerable contamination fro m hydrocarbons as seen from the CH3 vibrational peak. The (000-1)6H-SiC surface, or C face, has much different results and seems to present a more stable H terminated sur face. This surface does not have a visible oxide peak, and does not have an appreciabl e CH3 peak which would indicate hydrocarbon contamination. The CH2 peaks, one indicator of C-H bonding, are present instead. This is significant because the samples w ere in transit for over 2 weeks, and it was unknown if the container was opened in transit by customs authorities which would have resulted in air exposure of the surface. The electrical contact potential difference, or CP D, measured on hydrogen etched 3C-SiC surfaces by Coletti et al shows a surface potential flat-band condition aft er our etching process which suggested that we had achieve d H termination. In addition this surface electrical potential was completely stable for ~ 10 hours [111]. The surface potential was observed to slowly degenerate over ~ 1 000 hours after termination, stabilizing at a surface potential of ~ -30 mV inste ad of returning to the surface potential of -160 to -191 mV present after HF treatment [111] We had inconclusive XPS evidence leading us to believe that the surface was not Si-H terminated but instead C-H terminated.

PAGE 140

We etched a (100) process developed in (§2.3), and transported the wa fer to Walter Schottky Institut in the same manner as mentioned for the 6H in Figure 4.2, and show completely nonexistent, indicating that all on the surface. There are the characteristic CH also present is the CH3 peak of contaminating hydrocarbons. The middle gra ph shows the peak wavenumber for C hydrogen termination, but this peak is absent. Alt hough we cannot be sure that the surface is terminated with hydrogen bonds, the surf ace no Figure 4.2: ATRFTIR measurements of (001) 3C and terminated. The 3C some CH3 contamination. ATR Institut. 129 (100) 3C-SiC wafer, USF1-09007, using the new low pressure process developed in (§2.3), and transported the wa fer to Walter Schottky Institut in the same manner as mentioned for the 6H -SiC samples. The ATR-FTIR results are displayed similar results to the (000-1)6HSiC surface. The SiO completely nonexistent, indicating that there is either a thin layer or no oxide present on the surface. There are the characteristic CH 2 peaks as seen in the 6H peak of contaminating hydrocarbons. The middle gra ph shows the peak wavenumber for C -SiH bonds which provide an indication of the degree o f hydrogen termination, but this peak is absent. Alt hough we cannot be sure that the surface is terminated with hydrogen bonds, the surf ace no netheless remained oxide free. FTIR measurements of (001) 3C SiC that had been hydrogen etched and terminated. The 3C -SiC shows no oxide formation and CH termination, but has contamination. ATR -FTIR c ourtesy of M. Hoeb of the CH3 CH2 a-sym. CH2 sym. 007, using the new low pressure process developed in (§2.3), and transported the wa fer to Walter Schottky Institut in the results are displayed SiC surface. The SiO x peak is there is either a thin layer or no oxide present at peaks as seen in the 6H -SiC sample; peak of contaminating hydrocarbons. The middle gra ph shows H bonds which provide an indication of the degree o f hydrogen termination, but this peak is absent. Alt hough we cannot be sure that the netheless remained oxide free. SiC that had been hydrogen etched H termination, but has ourtesy of M. Hoeb of the Walter Schottky

PAGE 141

130 One notable difference indicated by this experimen t was due to difficulties arising from sample transportation. The terminated 3C-SiC wafer was sitting for 40 days before ATR-FTIR measurement could be conducted, which puts this surface into the 1000 hour measurement time period in the work of Coletti et al. [111]. The ATR-FTIR measurements indicating a hydrocarbon terminated su rface instead of an oxidized surface may be a reason why the original tested surfaces di d not return to their original surface potential value long after hydrogen etching and ter mination. We have obtained two Cree (0001)6H-SiC n-type wafe rs (3.43 off-axis, n = 0.055 W -cm, 420 m thick) single side polished on the Si f ace. The Si face wafer was diced into 5 x 5 cm squares and hydrogen etched and terminated, but we sent the other wafer to NovaSiC Inc. for CMP polishing of the C-fa ce side. The goal was to hydrogen etch and terminate both wafers at the University of South Florida and send the wafers to WSI where they will be subjected to hydrosilylation Although it is unlikely that the hydrosilylation process will work properly with our H terminated (0001)6H-SiC surface due to the oxide as shown in Figure 4.1, we took ex tra precautions to try and deliver the samples with the termination intact to the WSI. Th e samples were placed in sealed boxes within zip-lock bags inside air-tight containers, a nd all were filled with N2 gas. When the samples were received, they were immersed in neat 1 -octadecence for 60 min at 200C and subsequently cleaned in hexane, chloroform and methanol for 10 min each. The samples were then transported back to USF for cell viability testing via MTT assays. Contact angles, measured by Schoell, on the samples were slightly lower than expected, falling within the 80 range instead of ranges meas ured previously that were greater than

PAGE 142

90 [149]. To see the affect tested the functionalized s viability of the cells on CMP polished 6H 6HSiC samples alongside the functionalized samples wh ich then served as controls. 4.2.2 Biocompatibility Testing Figure 4.3 : A bar graph displaying the results of the MTT an alysis on 6H normalized to the PSt control and then expressed as of PSt and 3CSiC are included as benchmarks for comparison. We repeated the MTT assay viability tests terminated 6H-SiC samples as well as the As both surfaces of the semiconductors were specifical ly functionalized, we transferred the H2 etched surfaces directly from the reactor to the MT T testing well and rinsed the hydrosilylation treated surface with ethanol before placing 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 Polystyrine Well 6H-SiC Not Etched 6H Proliferation (Ratio of PSt Control)131 affect of an increased hydrophobic surface on cell viabili ty, we tested the functionalized s amples using MTT assays. As we had information abo ut the viability of the cells on CMP polished 6H SiC, we tested hydrogen etched and terminated SiC samples alongside the functionalized samples wh ich then served as controls. Biocompatibility Testing of Modified Surfaces : A bar graph displaying the results of the MTT an alysis on 6H normalized to the PSt control and then expressed as The results for the viability SiC are included as benchmarks for comparison. We repeated the MTT assay viability tests from (§3.2) on H samples as well as the 6H-SiC samples modified by both surfaces of the semiconductors were specifical ly functionalized, we transferred etched surfaces directly from the reactor to the MT T testing well and rinsed the surface with ethanol before placing the sample in the PSt t 6H -SiC H2 Etched 6H-SiC H etched Oct 6H-SiC HF Oct 6H-SiC HF APDEMS 6H-SiC HF APTES 3C-SiC (100) not etched 3C (100) H etched Surface Tested MTT Analyxix on 6H-SiC of an increased hydrophobic surface on cell viabili ty, we amples using MTT assays. As we had information abo ut the SiC, we tested hydrogen etched and terminated SiC samples alongside the functionalized samples wh ich then served as controls. : A bar graph displaying the results of the MTT an alysis on 6H -SiC samples, The results for the viability on H 2 etched and samples modified by hydrosilylation. both surfaces of the semiconductors were specifical ly functionalized, we transferred etched surfaces directly from the reactor to the MT T testing well and rinsed the in the PSt t esting 3C -SiC (100) H etched 3C-SiC (111) not etched H4 PC12

PAGE 143

132 well. After cell plating and the 96 hour prolifera tion time period, the MTT assay was performed. Figure 4.3 shows the results from the o riginal 6H-SiC surface, the 6H-SiC surface that was H2 etched and terminated and the 6H-SiC hydrosilylati on processed surface treated with 1-octadecence on the H2 etched and terminated surfaces. The results were not incredibly different for the H4 cell line, but there was a definite change in the viability level with the PC12 cell line. The H4 ce ll line prefers an un-etched surface and 1-octadecence treated surfaces over the H-terminate d surface, while the PC 12 has greater viability on the 1-octadecence surface and just sli ghtly less on the H-terminated surface. As was theorized before, proteins normally attach b etter to a hydrophobic surface, so a dramatic difference should have existed between the H etched surface, which would have a static water contact angle of 10 – 20 C and the 1-octadecence static water contact angle of 80 The effects were not dramatic, and mainly inconc lusive, indicating that the surfaces of the 6H-SiC that were treated may not ha ve been ideal. As we specified before, the surface may have had pinhole coverage d ue to the deterioration of the Htermination on the 6H-SiC Si face, which would have created patches of both H terminated and O2 surfaces which would have prevented the surface re action from occurring over the entire surface. Next the 6H-SiC samples were cleaned using the pir anha cleaning method and returned to WSI for the 1-octadecence treatment, ag ain using a different surface termination structure. The surface was O2 plasma etched two times followed by immersion in 5% HF solution. This creates the OH t erminated surface that is immediately processed with the 1-octadecence. Thes e samples were returned to USF and

PAGE 144

133 evaluated using the MTT assay method. Figure 4.3 s hows that this surface performed as anticipated. The levels of viability increased to almost to the same level as seen with the untreated (100) and (111)3C-SiC surfaces. This was almost a 50% increase in viability for both lines, which suggests that an extremely hy drophobic surface is better for cell viability. Silanization of the (0001)6H-SiC samples was the n ext logical surface modification due to the inability to successfully m aintain a long-term H surface termination. The process utilizes the easily obtai nable OH terminated surface that was used for the previous functionalization experiments using 1-octadecence. The process uses our standard cleaning procedure explained thor oughly in (§3.2.2). The samples are then placed in 5% HF solution for five minutes foll owed by a thorough rinse with DI water to ensure removal of any fluorine radicals. The samples were then immersed in a solution of toluene and one of two aminosilane mole cules. The first aminosilane molecule is (3-aminopropyl)-diethoxy-methylsilane, known commonly as APDEMS, and the second is (3-aminopropyl)-triethoxysilane, know n commonly as APTES. For APTES, one end of the molecule’s chain has multiple methoxy radicals, CH3O, molecules bound to a Si atom, while an amine, NH2, is bonded to the last electron site available on the center Si atom. APDEMS has almost the same structure possessing two methoxy molecules and a single methyl, CH3, molecule linked to the Si atom through a Si-C bond. The coating solution is mixed at (100:1 ) toluene to aminosilane. The samples must be prepared inside of a fume hood as the react ion releases gaseous methanol, H3COH, which is both a flammable hydrocarbon and can cause blindness and skin

PAGE 145

134 irritation. After 90 minutes, the samples were rem oved and cleaned in toluene and isopropanol for 20 minutes each within an ultrasoni c bath to remove physisorbed molecules from the surface. The samples were then dried with nitrogen and immersed in ethanol until cell seeding was initiated. The results from the MTT assay were considerably b etter than for any of the surfaces tested to date. The H4 Cell line had a vi ability of a little better than two times that of cell treated PSt with the APDEMS, and was a pproaching three times that of APTES. The PC12 line was a little less volatile, d isplaying a viability of 1.2 and 1.9 times that of cell treated PSt. These results were anticipated as these functionalization molecules have an amine molecule on the end of the chain which reacts readily with reactive amino acids and proteins. The MTT suggests the amine better viability and proliferation. The surface is much less hydrophobi c than the 1-octadecence treated surface, and static water contact angles measured a t WSI were approximately 50 [143]. Another observation is the APDEMS, which contains o ne less oxygen bound to the Si atom, has lower viability with both cell lines than the surface of APTES. It should also be noted that these results are preliminary and not conclusive. The previous experiment detailed in (§3) requires that other independent me asurements must be performed to assess a cell reaction to the surface. As we have performed measurements of the cell’s surface morphology and lamellipodia permissiveness on previous samples, we will repeat the AFM measurements again to allow comparison with the earlier study. Also, the tests only evaluated chemical changes to the surface, and neglected the charge differences caused by the capacitance of the new layers or char ges introduced by the layers

PAGE 146

135 themselves. It has been shown that these modified surfaces do alter the surface charge, and we need to measure and account for these change s before making any real conclusions [143, 149]. 4.3 Mouse Primary Cortical Neurons (CTX) on 3C-SiC Experimentation for the determination of the bulk biocompatibility of neural cell lines has been performed by our group; however the cells utilized were from immortalized cell lines. Generally, immortalized c ell lines do not behave in the same manner as their in vivo counterparts due to the fact that many of these ce lls are derived from cancer cells. Immortalized cells have an adva ntage in the fact that many of them are easy to maintain, and they have the ability to proliferate quickly on promoting surfaces. The H4 and PC12 cell lines are no except ion to this fact, as both were derived from cancerous tissue, and unlike healthy neurons, they do proliferate. For completeness to this research, and to further our understanding of the cell-semiconductor interaction, neural and glial cellular attachment to our surface s needs to be evaluated with cells obtained from primary sources. We have begun a col laboration with Dr. Maj-Linda B Selenica, a biochemist at USF who is currently extr acting cortical neural cell cultures from mice for modeling chemicals and enzymes in the CNS. Nominally, the cells are collected from embryonic mice from approximately th e 15th to 18th day of gestation (E1518) and plated for 7 days to allow for cell adhesio n and neural differentiation. The eighth day marks the first day of neural maturity and test ing nominally begins at that point. Dr. Selenica normally has more cells than is necessary for her experimental needs, so we

PAGE 147

136 proposed to repeat the MTT and cell attachment expe riments from (§3.2) using mouse primary cortical neurons, CTX. 4.3.1 Cortical Cell Biocompatibility Testing Primary neural cultures are much more delicate tha n the immortalized cell lines utilized in our previous studies. The initial expe riment involved cells plated in a 96 well plate that was left over from a previous experiment They had been plated with a count of 125,000 in poly-L-lysine wells, and at day 7 we attempted to collect them using 5% trypsin. The cells were re-plated on our semicondu ctor surfaces in accordance with the processes described earlier in (§3.2). The bare su bstrates that were compared in this study were the (100) and (111) faces of Si and 3C-S iC/ Si along with deposited NCD. The MTT assays were not readable for any of the sub strates tested including PSt, indicating that the trypsin process most likely kil led the delicate cells. The decision was made to plate the embryonic cells directly on the s emiconductor samples for the next experiment. The second attempt for gathering the CTX cells beg an with an unpleasant revelation. The pregnant mouse that was to be used for the experiment had given birth to pups. We only had the E15-18 protocol, so it was u tilized to collect primary 1-2 day postnatal CTX cells. After 12 days (8 for the cell maturation, attachment and differentiation plus 4 days for the assay), we perf ormed an MTT analysis. The resulting data collected, seen in Figure 4.4(a), indicates th at (111)3C-SiC has great viability with the CTX cells and equal to that of the PSt control. (100)3C-SiC however was just above

PAGE 148

137 40%, and the rest of the samples were at less than 20%. These cells were not usable for Dr. Selenica, as the statistical population was too low to make any legitimate claims, and replication of this experiment will not be possible in the near future. The results served as motivation only to develop a protocol for the CT X cells that allows experimental repetition. As a point of reference pure neuron cu ltures of post natal cells have a limited lifetime potential of about two weeks as reported i n the literature [153]. Neuronal plating media contains horse serum, but t his media is replaced 2-4 hours after plating, at which time the media is without s erum. As we were using non-coated 3C-SiC, serum was thought to be critical for cell s urface adhesion and viability. We repeated the MTT experiment on the semiconductor su rface and added a new testing surface, Kapton tape, a brand name for polyimide. As polyimide is the current leading construction material for neural interface devices, inclusion and comparison of its biocompatibility is paramount to establishing the b iocompatibility level of 3C-SiC. The third plating, using E15-18 CTX cells, was maintain ed for the maturation and MTT time requirement, and the results of the assay are displ ayed in Figure 4.4(b). A noticeable improvement is seen between the Si and NCD surfaces which are double than the previous experiment, but still around half that of 3C-SiC. However, there are viability differences between the crystal faces, as the (100) face has greater viability than (111) for both 3C-SiC and Si substrates. Also observed is th at the new additional tested substrate, Kapton tape, shows slightly more viability than (10 0)3C-SiC.

PAGE 149

138 (a) (b) Figure 4.4: The results of single MTT assays perfo rmed with triplicate samples expressed as The graph of (a) displays the resulting MTT of D 1-2 post-natal CTX cells plated in 96 well plates and seeded on se miconductor samples. (b) shows the MTT results of E15-18 prenatal cells plated directl y on the samples after their collection and dissolution from the mouse cortex. (a) is norm alized to a polystyrene control, whereas (b) did not have enough cells for a complet e MTT and the PSt was not seeded. One major discovery is that after the eight days t he cells were not showing signs of neural differentiation. Although a few cells sh ow spontaneous axon or dendritic extensions, many still appear rounded with no exten sions whatsoever. Literature research points to two factors that cause differentiation in cortex cells. One is a surface with the proper chemical factors, such as the amino acid pol y-lysine, and the other is a surface 0 0.2 0.4 0.6 0.8 1 1.2 1.4Proliferation (Ratio of PSt Control)96 hour MTT Assay Mouse CXT D1-2 Post-natal Control Well Si (100) Si (111) 3C-SiC (100) as grown 3C-SiC (111) NCD Proliferation (Ratio of PSt Control)96 Hour MTT Assay Mouse CXT D15-18 Pre-natal Si (100) Si (111) 3C-SiC (100 3C-SiC (111) NCD Polyimide

PAGE 150

from actual plated astrocytes [154, 155]. The next preliminary experiment investigating the neural differentiation on the sur faces of the semiconductor. We plated half of our samples with poly surface chemistry We then examine adhesion, and the differentiation process through t he 8 day period after seeding. AFM and SEM were to be utilized to visualize the cells, quantify their morphologies, examine the cells for lamellipodia extension, and evaluate dendritic images from the SEM in fixed cells on the surface of PLL (a) (c) Figure 4.5: SEM micrographs of PLL coated PSt sample (a) with c ortical cell imaged at 25KV and 8000X (scale bar 1um) and (b) a cortical c ell at 25KV and 6000X (scale bar 2um). (c) shows a three dimensional image of CTX o n PSt, bare substrate at day 6 and (d) shows t he CTX on a (100)3C courtesy of S. Saddow and A. Oliveros, SiC group at USF, respectively. 139 from actual plated astrocytes [154, 155]. The next preliminary experiment investigating the neural differentiation on the sur faces of the semiconductor. We plated our samples with poly -Llysine, PLL, and the others were left with their na tural We then examine d the cells every day to monitor cell morphology, adhesion, and the differentiation process through t he 8 day period after seeding. AFM and SEM were to be utilized to visualize the cells, quantify their morphologies, examine the cells for lamellipodia extension, and evaluate dendritic or axon, images from the SEM in Figure 4.5(a, b) and AFM in Figure 4.5 (c, d) show examples of fixed cells on the surface of PLL treated PSt. (b) (d) SEM micrographs of PLL coated PSt sample (a) with c ortical cell imaged at 25KV and 8000X (scale bar 1um) and (b) a cortical c ell at 25KV and 6000X (scale bar 2um). (c) shows a three dimensional image of CTX o n PSt, bare substrate at day 6 and he CTX on a (100)3C SiC PLL coated substrate at day 8. SEM and AFM courtesy of S. Saddow and A. Oliveros, SiC group at USF, respectively. from actual plated astrocytes [154, 155]. The next preliminary experiment involved investigating the neural differentiation on the sur faces of the semiconductor. We plated lysine, PLL, and the others were left with their na tural cell morphology, adhesion, and the differentiation process through t he 8 day period after seeding. AFM and SEM were to be utilized to visualize the cells, quantify their morphologies, examine extensions. The (c, d) show examples of SEM micrographs of PLL coated PSt sample (a) with c ortical cell imaged at 25KV and 8000X (scale bar 1um) and (b) a cortical c ell at 25KV and 6000X (scale bar 2um). (c) shows a three dimensional image of CTX o n PSt, bare substrate at day 6 and SiC PLL coated substrate at day 8. SEM and AFM

PAGE 151

140 The resulting section has shown preliminary result s of this ongoing research. Our first goal is to create protocols that produce the same CTX cell quality, with repeatability between harvests, so as to produce statistically re levant results. The future work for this portion of this research will be covered in (§5.4). 4.4 Neural Activation Device (NAD): Generation 0 The ultimate goal of this research is to demonstra te action potential stimulation in neuronal cells via a SiC-based electrical stimulati on device, which we have called a neuronal activation device, or simply NAD. Before discussing details of this preliminary work, it is instructive to look at the general prin ciples involved in neuronal cell/electrical stimulation and for this discussion we divide the s tory into two parts. The first part is what is called a conductive or electrode-based devi ce, which is the principal behind devices like the Michigan probe. The second type o f device relies on the use of an electric field that is applied via a capacitor, wit h no metal necessarily contacting the intracellular fluid (aka media). Figure 4.6 depict s the electrical situation for these two types of stimulation by first showing the semicondu ctor/media model (left side of figure) depicting ion position, the Helmholtz electrical la yer, etc. Next, the figure depicts the equivalent electrical circuit followed by a plot of the required stimulation waveform that must be applied to induce an action potential in a neuron. Figure 4.6(a) outlines the situation for an electrode based device and Figure 4.6 (b) for a capacitive based device. This research has been primarily focused on the gr owth improvement, as seen in (§2), and biocompatibility of 3C-SiC and NCD with n eural cells, as explained in (§3) and

PAGE 152

141 (a) (b) Figure 4.6: Simplified cartoon representations of an (a) electrode-electrolyte interface and a (b) metal-insulator-semiconductor (MIS) capac itor device and an electrolyte which was covered in detail within (§ 1.3.4 and 1.3.5). Included for reference are the simplified interactions occurring between the interface of the metal or insulation and the electrolyte. Insets show representation of a time current graph for the signals necessary to safely initiate an action potential in a neuron. Ct

PAGE 153

142 the previous sections of this chapter, but the goal of these experiments was to establish the use of these materials for eventual utilization in an implantable neural interface that would be the basis of a BMI device. The NAD device was designed as a testing platform to explore the electrical interface between the sem iconductor and the neuron. The first NAD device, called NAD Generation 0, or NAD0, is a simple electrode based device which has been electrically characterized since the initial Michigan probe was created in 1970 [1-5, 13, 14]. Because 3C-SiC contains defect s from the growth process, and normal growth at low pressure generates highly resi stive material, this device will consist simply of 3C-SiC base layer, metal electrodes, and an SU8 insulation layer to prevent electrical cross-talk between adjacent conductors. This device will interact with the cell through the extracellular media through complicated electrochemical reactions of a metal electrode in electrolytes as explained in (§ 1.3.4) and displayed in Figure 4.6(a). NAD0 will use gold as the metal contacts with a thin lay er of titanium underneath which provides contact adhesion to the semiconductor. 4.4.1 NAD0 Activation of a Neural Action Potential The generation of the NAD0 device began in Decembe r 2008. The circular contact and wire dimensions were based on the PC12 and H4 average cell sizes obtained by AFM measurements in (§3.3) which were observed t o be on average typically 30 m wide by 60 m long. Unfortunately, as money was an issue, we had to settle for a larger mask design on an acrylic film which had a minimum feature size of 50 m. The final NAD0 device is an array of 64 cell interaction pads 100 m in diameter, connected with

PAGE 154

143 (a) (b) Figure 4.7: The NAD0 device generated on 3C-SiC co nsisting of gold leads and a polyethylene gasket. (a) displays a 50 mm wafer wi th multiple devices and a dime to indicate the size of the device. (b) shows a photo graph of the completed NAD0 device with the polyethylene well added and a cartoon repr esentation of the device with media within the well. Device fabrication performed by A Oliveros and S. Saddow, USF SiC Group. PDMS provided by S. Dupont, USF Chemical Eng ineering Dept. 50 m leads to 500 m square contact pads for event ual wire bonding to a package. Signatone SLM micropositioner metal probes would in itially provide electrical connection between the contacts and the signal gene rator which would be utilized to send the biphasic square-wave pulse between two adjacent electrodes. The device was simply created from two masks. The first was the mask for positive photoresist designed for the metal lift off process to define the metal electrod e contact geometry. The second mask was for the negative SU8 photoresist layer to provi de electrical insulation and isolation between adjacent wires once the cell media was depo sited on the structure. The mask was simply 300 m diameter holes centered on the ce ll interaction pads to isolate each electrode. Figure 4.7 shows the completed NAD0 dev ice layout which was fabricated from a two (2) inch 3C-SiC film grown on (100)Si. A piece of PDMS was used as a cell

PAGE 155

media containment well, and the PDMS was prov Toomey’s group in the USF Chemical Engineering Depa rtment. (a) (d) Figure 4.8 : Electrical activation of cell action potential v ia the SiC NAD0. microscope images showing undifferentiated PC12 cel ls seeded on t formaldehyde Image (a) shows proliferation cells p roliferate across the entire device after 5 days. (c) shows differentiated PC12 which have become peripheral neurons. (d) and (e) show the device before biphasic wave activation and after the pulsing had began. A s few cells fired, we have added a red ci rcle to indicate the same, less populated area on e ach image, and a green oval to indicate the area where the AP activated cells orig inated. The NAD0 die were diced by C. Locke of our about 7 complete die with a total of 12 experimentation on which we electrodes. Figure 4.8 (a) seed. 200,000 cells were seeded in 144 media containment well, and the PDMS was prov ided by Samuel Dupont of Dr. R Toomey’s group in the USF Chemical Engineering Depa rtment. (b) (e) : Electrical activation of cell action potential v ia the SiC NAD0. microscope images showing undifferentiated PC12 cel ls seeded on t he NAD Image (a) shows proliferation after 1 day and (b) indicates that the PC12 roliferate across the entire device after 5 days. (c) shows differentiated PC12 which have become peripheral neurons. (d) and (e) show the device before biphasic wave activation and after the pulsing had began. A s few cells fired, we have added a red rcle to indicate the same, less populated area on e ach image, and a green oval to indicate the area where the AP activated cells orig inated. die were diced by C. Locke of our USF SiC group, which yielded about 7 complete die with a total of 12 usable die. A NAD0 die on which we seeded cells to evaluate their size in relation to the (a) and (b) show the difficulty in estimating the amoun t of cells to seed. 200,000 cells were seeded in Figure 4.8 (a) and measured the next day with lit Dupont of Dr. R (c) : Electrical activation of cell action potential v ia the SiC NAD0. Optical he NAD fixed with and (b) indicates that the PC12 roliferate across the entire device after 5 days. (c) shows differentiated PC12 which have become peripheral neurons. (d) and (e) show the device before biphasic wave activation and after the pulsing had began. A s few cells fired, we have added a red rcle to indicate the same, less populated area on e ach image, and a green oval to USF SiC group, which yielded was selected for seeded cells to evaluate their size in relation to the and (b) show the difficulty in estimating the amoun t of cells to (a) and measured the next day with lit tle

PAGE 156

145 change in cell population; however, after 5-6 days, the PC12 and H4 cells proliferated across the surface of the device. We then seeded 5 0,000 cells in a 2” Corning Cellbind" Petri dish and added NGF at 20 ng/ ml to the media to induce differentiation of the cells into peripheral neurons. The cells were allowed to mature for approximately seven days in an incubator until they differentiated, then the y were collected and plated on the device. As the cells continue to proliferate until differentiation, the result was a tissuelike mass of interconnected peripheral neuron cells which would not disassociate even with 100% trypsin. After the best disassociation p ossible, the cells were plated on the NAD0 device and a Ca2+ sensing dye, Rhod-2, was added to the media and in cubated for 60 min to allow dye uptake and cell surface attachm ent. Figure 4.8(d) shows the cells on the surface of the NAD0 device, and the cloudy appe arance of the photograph is due to the dense population of cells on the surface. Following the incubation uptake period, the NAD0 w as placed under a microscope in Dr. Weeber’s laboratory at the USF By rd center equipped with fluorescent filters that was housed in a Faraday cage for elect rical noise reduction, and electrical connection to a waveform generator located outside the cage was made. We activated the NAD0 with an alternating, balanced, biphasic square wave electrical pulse consisting of a cathode phase followed by a slight hold sequence an d an anodal pulse. This stimulus was made between two electrodes. The pulse frequency w as set to 1 KHz, and pulse amplitude started at 10 mV and was increased manual ly in 10 mV increments approximately every second. During action potentia l firing, voltage-gated calcium channels (VGCC) will open and allow Ca+2 exchange into the cell, which is then

PAGE 157

146 hydrolyzed to Rhod-2 by endogenous esterases produc ing a change in fluorescence. The increase in fluorescent excitation was captured by a camera attached to the microscope. At peak to peak voltages of ~300 mV, the neurons wen t from a non-active status seen in Figure 4.8(d), to active action potential as seen i n Figure 4.8(e). It should be noted that only the cells nearest or in actual contact with th e edge of the electrodes experienced action potential activation and only a few cells ex perienced the activation that were in contact with the electrode. This may be due to the fact that cells above and adjacent to the activated cells were not synaptically connected and would not have the proper neurotropic factors delivered to them from the adja cent neurons. As of the time of this work, we are still investigating the activation of neural action potentials by the NAD0. The success and failure of NAD0, our first attempt to activate a neuronal action potential with a microelectronic device; has noneth eless shown that 3C-SiC can be utilized as a support for an electrode device for n eural activation. Device redesign with smaller electrodes, easier electrical interface cap abilities to external equipment, and better biological cellular protocols are all necessary and have been planned. A field effect NAD and a control electrode NAD constructed from Si wil l also be constructed to provide alternative and comparative data. Eventually, we w ill utilize primary cortical and hippocampal cell cultures, which contain cells dire ctly from the CNS, instead of the immortalized PC12 cell line. We will also use (via the USF electrophysiology core; Dr. E. Weeber, Director) whole-cell current clamp metho ds so we can quantitatively measure the cell membrane current and voltage changes [156] Different pulse widths and current densities will be evaluated to determine the comple te device characterization and cell

PAGE 158

147 reactions for eventual computer and biological mode ling purposes. Cross talk, noise levels, signal loss across the ionic media, distanc e of signal transmission, and the levels of electrical excitation that create irreversible F aradic reactions at the metal-electrolyte interface will be explored for the NADs. Through th e use of the whole cell current clamp, we can examine and quantify the reactions of the ne uron to the electrical stimulations. The results of this future study will determine if 3C-SiC is a material that can act as the basis for a neural prosthetic, and more impo rtantly, it can show that this material can be used for a capacitive based device instead o f an electrode device. The potential of this demonstration is enormous in that not only can we extend the life of neuronal implants using 3C-SiC, a biocompatible material, bu t with the MIScap field effect device which we can tailor specific and controlled electri cal fields, allowing for more efficient and controlled activation of neural action potentia ls. Not only is the cost of patient care likely to be reduced with successful BMI applicatio ns, but the quality of life will be significantly improved. No longer will non-neurona l compatible materials be surgically implanted in the brain only to wait for device fail ure in the near term. 4.5 Summary of Chapter While in (§2) the focus was on material developmen t of both 3C-SiC and the combination of NCD on 3C-SiC, and (§3) specifically focused on the in vitro biocompatibility of the bulk materials, this chapte r shows the division of our research into three new parallel tracts. These new areas of rese arch are focused on the generation of a device that will interact with neurons indefinitely within the human body.

PAGE 159

148 The first new investigation was the modification o f our surfaces to generate better interaction between the biological cells and manmad e semiconductor materials. Through silanization of 6H-SiC surfaces, this goal was achi eved, and that material showed drastic increased biocompatibility. Our second investigati on was to begin to work with cells directly obtained from an animal donor instead of f rom immortalized, cancerous cell lines. The initial results of this work are even m ore promising, showing that 3C-SiC has good biocompatibility comparable to Kapton" polyimide, which is utilized in long term neural implants currently to solve the biocompatibi lity problem of silicon implants. Our third investigation begins to demonstrate 3C-SiC’s ability to be utilized as a platform for neural communication. The NAD0 device was able to activate differentiated PC12 cells, although many improvements to both the device and c ell protocol must be made before action potential activation is truly confirmed The chapter (§5) will detail the future work our group is proposing to accomplish.

PAGE 160

149 Chapter 5: Future Work 5.1 Introduction The future work outlined in this chapter details t he problems that must be overcome for the development of a successful neural implant and later the development of an entire BMI device system. The first hurdle i s the ever present biocompatibility issue. We recently began to study biocompatibility by utilizing surface functionalization and will continue to work along this route. We wil l include not only the immortalized cell lines we have used in the past studies, but mo ve on to a more relevant primary cell line derived from the brain itself. (§5.2) will de tail future work and exploration of various surface modifications, and we will be perfo rming these modifications in conjunction with WSI. The combination of the new c ell lines and surfaces should assist in quantifying and modeling cellular response to in organic surfaces, a benefit that will help not only neural implants but all implantable b iomedical devices. (§5.3) will detail future plans for continuing biocompatibility testin g with not only primary mouse cortical cells, CTX, but the addition of primary mouse astro cytes, microglia, and the eventual culture of hippocampal tissue slices. NAD0 success fully activated an action potential of a cultured neuron. (§5.4) will detail our plans fo r new NAD generations which will eventually provide true two-way communication betwe en electronic semiconductor devices and CNS neurons. More importantly, we will begin utilizing a higher order

PAGE 161

150 tissue for these new experiments which not only can provide additional electrical data, but also indicate the level of bioreactivity from m ultiple integrated cells. The exploration outlined in this section will be exciting and illum inating for the ‘dark horse’ of the silicon carbide polytypes, cubic silicon carbide, and will have the ultimate goal of producing a BMI interface which will be suitable for extended i mplantation in vivo 5.2 Cell to Substrate Interaction Study The results of the MTT assays from the silanizatio n of the surface of 6H-SiC shows that we can increase the biocompatibility of a substance using SAM, but extensive research is required before we can incorporate this technique into a device, or use the method to better understand the mechanics behind th e chemical interactions involved in cell membrane attachment to a surface. The next st ep in understanding the ability of the SAM to coat a surface uniformly is to use (000-1) 6 H-SiC (i.e., C-face) substrates with the process. The C-face of SiC has been shown to h ave more defects and surface abnormalities, and the OH termination generated by HF immersion has been shown to possess interdispersals of H terminations [143]. T his will decrease the SAM monolayer generation from silanization, but the results from our H2 etching displayed in Figure 4.1 show promising results for the long term hydrogen t ermination, which may lead to a better surface from hydrosilylation. Testing the v iability of the H4 and PC12 cells using MTT assay on the (000-1)6H-SiC samples is the next stage of this exploration. We need to include unetched H terminated, 1-octadecence hyd rosilylation and silanization samples using APDEMS and APTES.

PAGE 162

151 5.2.1 Surface Modification and Biocompatibility Improveme nt Surface charge has been shown to affect the activa tion of proteins absorbed on the surface of inorganic substances, and we have shown that the viability of cells is affected by bulk charge within a semiconductor as was eviden t with the results displayed within Figure 3.2. The advantage of using SAM termination of the surface is that it has been shown to prevent native oxidation of a substrate, a nd therefore modify the charging characteristics of the surface [149]. Silicon carb ide has a thin layer of native oxide, but it tends to be of less quality than the oxide on Si du e to the effects of C which either incorporate at the oxide interface as a defect or b ecomes a vacancy by forming CO2 [150]. We need to perform charge profiles of the s urface for the SAM coated substrates in air and within the cellular media, characterizin g the material using CPD measurements, capacitance voltage (CV), and current voltage ( IV) profiling. The final substrates that need to be included to complete the charge profilin g study are intrinsic and positively bulk doped (i.e., p-type) 6H-SiC. Understanding th e changes that these surfaces make to the charge level will help us understand the protei n and viability changes seen with the MTT assays and develop methods to use the SAM layer s to manipulate not only the surface wettability, but surface charge as well. The MTT analysis and surface characterization are not complete without inspection of the cell morphology and lamellipodia extension permissiveness analysis. As with our earlier experiment, AFM of both living and fixed cells on the evaluated substrates provided quantitative information that c an be used directly for this type of inspection. The difficulty that AFM has is that it can only look at a single cell or part of

PAGE 163

152 two cells. As it was difficult for us to obtain fl uorescence microscopy measurements due to the substrates being opaque, we will use the SEM to investigate multiple cell populations. Coating the sample with a few nanomet ers of gold after measurement of the fixed cells in the AFM allows us to utilize the ele ctron microscope (SEM) to examine the surface of the semiconductor on a larger scale. Th e arrangement of cell populations will show if cells are establishing connections between one another, which for neural and glial cells is another measurement of the natural cell ev olution. Cell adhesion to foreign materials is not well doc umented or quantified, and much of what is known has been gathered from the cell ad hesion studies to glass substrates possessing surface modifications [120-124, 127]. U sing both the SAM layers and untreated substrates, we will attempt to examine th e adhesion of proteins that enable cells to adhere to materials, and we will attempt to iden tify specific adhesion proteins used by the cellular membranes to attach to the substrates. The first experiment is to look at surfaces that have been immersed in media with seru m with the AFM and ATR-FTIR. The AFM not only measures surface morphological dif ferences, but can be used to detect changes in surface friction, and can be used to per form force microscopy [151, 152]. Bergkvist et al has shown that the AFM can be used to detect chan ges in proteins absorbed on a surface, and Dufrne et al has shown that force microscopy using surface modified cantilevers can identify certain protein c haracteristics [151, 152]. We will utilize similar AFM techniques to attempt to identi fy the proteins that absorb to the surface of the 6H-SiC from the serum media, and we will attempt to determine if the proteins are activated and elongated, or compact an d unreactive. We will then utilize the

PAGE 164

153 ATR-FTIR and XPS to examine and characterize the ch emical composition of the protein absorbed surfaces. The next stage will be to remov e the cells by scraping and trypsinization to examine the protein complexes lef t on the surface. Using the same techniques, we can attempt to characterize which pr oteins and cellular complexes have been left behind, and correlate information from th e different surfaces to observe changes in cell adhesion. Another method for cellular atta chment is to utilize fluorescent molecules attached to the aminosilanization modifie d surfaces. Schoell attached an amino acid, glutaric dialdehyde, to the amine at th e end of the silanization chain, and utilized the amino acid to attach Alexa488 fluoresc ent labeled BSA proteins [143]. We can utilize the same process and allow the cells to attach to the surface. Potentially, the cell will displace the BSA proteins and attach to t he surface, and we will be able to utilize fluorescence microscopy to quantify the cytoskeleto n focal points. The utilization of the cell morphology, protein attachment, and viability will enable us to develop a more concrete vision of the attachment of cell membranes to SiC and SAM modified surfaces. This knowledge can be used to help formulate a devi ce which can be recognized as part of the extracellular matrix instead of a foreign ob ject. 5.3 Primary Mouse Cortical and Astrocyte Cell Biocompat ibility Study Clearly, the results obtained in (§4.3.1), while s howing lots of promise, are still within the developmental stage. Although we can pr oduce the primary neural cell lines, they have not reacted in the same manner regarding the same stimulus as the immortalized PC12 and H4 cell lines. The first ord er of business will be to establish a

PAGE 165

154 proper protocol for the generation of CTX cells whi ch produces comparable cells between different experiments, and these CTX cells will experience differentiation into neurons after 6 days. This type of experimentation unfortunately requires more empirical experimentation than calculations, and may require multiple experiments to confirm. Once a repeatable cell line protocol has been deve loped, the second experiment will be to determine the reaction of the CTX cells to both poly-lysine treated and natural or untreated surfaces. The cell reaction to the un treated surface is paramount for baseline biocompatibility comparisons, while treatment allow s options for improving the biocompatibility. Ideally, we can include some of the new surfaces covered in (§4.2) and (§5.2). The experiment will begin by plating the E 16-18 CTX cells on the tested substrates. We will then perform live and fixed ce ll AFM measurements each day of the test in order to monitor changes in the cell morpho logy, attachment, and differentiation quality. This research would be used to characteri ze the interaction of the neuron with the semiconductor surface or chemical treatments so as best to determine coatings for possible long term neural implants. Another proposed biocompatibility test should incl ude primary astrocytes and microglia in order to evaluate their reactivity wit h the surface of the semiconductors. As these are the two primary cells involved in glial s carring, establishing their reaction to the device material is paramount in determining the out come of long term implantation within the CNS. Although we have found that the H4 cell line, which has been derived from a glial tumor, shows good biocompatibility and viability with 3C-SiC and NCD, this cell is derived from a cancerous tumor. We need to evaluate how normal, non-cancerous

PAGE 166

155 cells would react to these materials. As was discu ssed in (§1.2.3), normal astrocytes and microglia cells experience physical and chemical ch anges when they are activated by damage to the CNS system or when they identify a fo reign material. These changes eventually lead to the formation of the glial scar encapsulation around the material. Testing our material to evaluate if the glial cells actually react positively or negatively is paramount for the continuation of our biocompatibil ity study. We will use the primary glial cells for experiments similar to the experime nts explained in (§3), and the discoveries made with these cells, will dictate if 3C-SiC or NCD are materials that potentially can be tested in vivo. 5.4 NAD Experimental Approach Our first generation of NAD devices, NAD0, proved to be successful by activating neuron action potentials, but it is nece ssary to redesign the device to address many problems. NAD generation 1, or NAD1, will add ress these issues and involves the creation of three separate devices. The devices wi ll be constructed and tested as part of the future research effort to achieve two-way commu nication between CNS neural cells and an electronic device. NAD1 will have smaller c ontacts 20 m in diameter that are insulated from one another. Device packaging will be added to improve interface to both the signaling and measurement equipment. The first NAD1 is an electrode device using Si as the base substrate material (SiNAD). The device will consist of highly resistive Si, a layer of oxide to provide insulation, metal electrodes, and a secondary insul ation layer, SU8, which is a

PAGE 167

156 photosensitive biocompatible polymer. If necessary we can try to utilize another suitable insulating polymer as insulation between electrodes if SU8 is unavailable or does not fit the device requirements. The second NAD1 will be a nother electrode device similar in make to the first, but will instead utilize highly resistive 3C-SiC as a base material (SiCNAD). Because 3C-SiC contains defects from the gro wth process, and normal growth at low pressure generates highly resistive material, t his device will consist simply of 3C-SiC base, metal electrodes, and the SU8 insulation laye r. A semiconductor material, like 3CSiC, allows for another device with a different int erface which is the metal-insulatorsemiconductor capacitor device (MIScap or C-NAD). MIScaps interact with the neurons using electric fields to interact with ions in the extracellular media instead of through the complicated electrochemical reactions of a metal el ectrode in electrolytes as seen explained in (§ 1.3.4 and 1.3.5) and in Figure 4.6. The C-NAD device will be a MIScap device with conductive, n-type 3C-SiC as the base m aterial which is then coated in electrically insulating NCD material. Metal is adde d to the backside of the device for connection of the MIScaps devices to the electronic s, but the insulator remains uncoated at the cellular interface. All three devices will use gold, or platinum if available, as metal contacts. A thin layer of titanium will provide fo r contact adhesion. The NAD1’s will have gold wires bound to the contact pads, which wi ll be insulated by a coating of SU8. All device wiring will be gold bonded and soldered to insulated copper wiring which can be easily connected to external electronics which w ill be required for neural activation and signal processing. After the devices are creat ed, they will be fully characterized

PAGE 168

157 electrically in both air and within a buffered salt solution environment to confirm their functionality. 5.4.1 NAD Signal Reception from Neuron Action Potentials The advantage of implantable neuronal prosthetics for the realization of BMI interfaces over non-invasive EEG type neuronal inte rfaces is the ability of implantable devices to have two-way communication with the CNS via the sending and receiving of electrical signals. Unlike signal transmission, th e current density for signals transmitted by cells is in the tens of microvolt range; therefo re electrical noise becomes a very important factor. Current microelectrode devices f rom Si or polyimide can receive signals from up to 100 m away [14]. We would like to use the new NAD1 device as signal receiving as well as transmitting devices. An important consideration that may need to be included is external amplification and f iltering of the signals due to the small 50–100 V initial signal amplitudes. To determine these levels and set a baseline for the experiments, we will plate cortical cells onto the surfaces of the NAD1 and a control set within Petri dishes. A whole-cell voltage clamp wil l then be used to depolarize the membrane. Once depolarized, VGCC’s will open allow ing Ca2+ into the cell and the resulting currents will be recorded. Fortunately n euronal VGCC properties sitting on Petri dishes have been well characterized, and we h ave much to compare our results with [157-167]. The current characteristics of the neur ons on the dishes will be used as a control and VGCC currents obtained from the neurons plated on the NADs will be compared to determine if the neuronal properties di ffer. The current characteristics will

PAGE 169

158 also be utilized to determine the levels of signal amplification and filtration that would be required by the NAD1s to intercept signals from neu rons. The next stage for the NAD1 will be to combine the receiving and sending of signals to achieve two-way communication with the n eural cell. A whole cell current clamp will again be used and the electrodes nearest to the cells will be connected to the amplification electronics and an oscilloscope. The current clamp will be used to elicit an action potential which will be subsequently receive d by the nearest electrode(s). The signals received on all of the NAD1s will be evalua ted against the recorded patch clamp signals to establish signal integrity. The NAD1s wi ll then be used to both elicit and receive a signal from a neuron simultaneously. For the electrode NAD1s (Si-NAD and SiC-NAD), two electrodes are selected as the cathod e and anode signaling electrodes while a third electrode in the immediate area will be utilized as the signal recording site. The C-NAD device will require the use of only 2 sit es, one capacitor for stimulation and the second for signal reception. The whole-cell cu rrent clamp will again be used with a neuron within the targeted signaling field to verif y the action potentials. For the Si-NAD and SiC-NAD, a biphasic signal will be passed betwe en the two electrodes, and the C-NAD will use a monophasic signal on its single ca pacitor. Once an action potential has been produced, the resulting signal will be rec orded by both the clamp and the receiving electrode. The two signals will then be compared. The generation of the proof of concept for a BMI d evice will be the next goal to evaluate the NAD1’s BMI potentiality. The configur ation, displayed in Figure 5.1, will use two NAD1s of similar electrical characteristics which will be selected and

PAGE 170

159 Figure 5.1: Cartoon representation of the proposed BMI that will be demonstrated by 2 NADs of the same material construction (NAD#1 and N AD#2). The NADs will be seeded with primary neuron cultures or organotypic hippocampal slice cultures and a signal generated from the wave generator (1) will b e delivered to the NAD device to elicit an action potential from the cultured cells. An electrode intercepts the resulting action potential signal and transmits it to filteri ng and amplification electronics (2). An oscilloscope (3) is utilized to analyze and send th e resulting signal to the control computer (4). The computer analyzes the input and a ctivates another wave generator (5) to elicit an action potential on the second device. Whole-cell current clamp using glass electrodes will provide conformation of action pote ntial activation (6). interconnected through computer controlled electron ics. The first NAD1 device will be connected to a wave generator, and the signal recei ving NAD1 will be connected to amplification and filtration electronics. The ampl ification electronics will be connected to an oscilloscope which is controlled by a compute r. Computer programs will be used to detect the signal spikes generated during an action potential, and then the signals will be characterized. The computer will send a signal to a second wave generator connected to the second NAD1 device. Neuron cultures will be cu ltured on both of the separated NAD1s, and whole-cell current clamps will be attach ed to target neurons on both devices. NAD#2 NAD#1 (1) (2) (3) (4) (5) Neuron Neuron Hippocampal Slice Hippocampal Slice

PAGE 171

160 The first NAD1 will be used to stimulate neurons, r eceive the subsequent signals and send signals to the computer. The computer will the n send a corresponding signal to the second NAD1, which will stimulate the cultured neur ons. The results will be verified through the whole-cell current clamps on both devic es. The NAD1s will then be compared to determine differences between them. The strength of implantable neuronal devices is fa st two-way connection with target neurons, and the device mode for the NAD1s m ust be proven before clinical trials can be started. The final experiment detailed in ( §.5.4.2) will also establish the ability of this type of platform to act as a bridge BMI device an important factor for the restoration of individuals with brain damage and the resulting loss of motility. Also, this work will serve a secondary purpose in that they allow the el ectrical characterization of a MIScap device constructed from diamond and cubic silicon c arbide which, to date, has only been proposed theoretically. 5.4.2 NAD2 Interaction with a Hippocampus Slice The CNS medial temporal lobe, which includes the h ippocampus, is important for cognitive function and memory formation. The under lying mechanisms believed to affect memory formation in this area of the brain i nvolve changes in synaptic plasticity, specifically long term plasticity (LTP). The cornu ammonis (CA region), which is filled with densely packed pyramidal neurons not unlike th ose located within the cortex, is divided into four regions and is regularly used to examine LTP [168-170]. The CA1 region of the hippocampus displays several types of synaptic plasticity, including two

PAGE 172

161 forms of LTP and at least one form of long-term dep ression (LTD) [168-171]. All forms can be initiated by electrical stimulation of varyi ng degrees and are measured in the CA1 as an increase (LTP) or decrease (LTD) in the slope of field excitatory postsynaptic potentials (fEPSPs). Not surprisingly, lesions to the hippocampus impair performance on hippocampal-dependent tasks [172, 173]. Because le sions to the hippocampus disrupt normal learning and memory, a NAD that has the abil ity to be inserted into the hippocampus and reconnect a broken circuit would be of great benefit not only to the scientific community, but in neurology and neurosci ence. The eventual long-term goal is to build man-made n eural networks to repair large portions of damaged brain and CNS tissue, and clear ly the first step is to interact with a well established brain network model, such as organ otypic hippocampal slice cultures. These hippocampal cultures maintain their tri-synap tic circuitry following plating and are electrically plastic, making them an appropriate ti ssue model for restoring brain function. Also, this higher level tissue contains both neuron s and glial cells and will give us the opportunity to test the biocompatibility of our mat erials, specifically 3C-SiC and NCD, with more accuracy than the single cell in vitro methodology. A new NAD construction, NAD generation 2 which is based on the previous NAD generations, will be designed and fabricated to specifically interact with the hippoc ampal tri-synaptic neuronal network so as to provide a model for ex vivo communication. To establish a baseline comparison, we will plate hippocampal slices on the NAD2 and stimu late the CA2/3 Schaffer’s collaterals using electrodes or capacitive devices on the NAD2. Subsequent fEPSPs will be

PAGE 173

162 recorded from the stratum radiatum CA1 region of th e hippocampal slice using the NAD2s and glass recording electrodes, which will ac t as a control for signal comparison. The NAD2s will then be utilized to demonstrate the principal of a bridge BMI device. Hippocampal slices will be cut along the Sc haffer’s collaterals and through the dentate gyrus, creating two halves. The principal BMI operation will be utilized to allow communication of signals between the separated netw orks of the same hippocampus. Each part of the hippocampus will be placed on sepa rate devices which are electrically connected through computer controlled electronics. A glass electrode will be placed in the CA1 region of the hippocampal slice to provide signal verification. The hippocampal slice containing the CA2/3 region will be stimulate d with a biphasic stimulating electrode and the subsequent potentials will be recorded by t he NAD2. The NAD2 will transmit the signals to the computer where they will be proc essed using the spike detection software. A secondary signal will be transmitted f rom the computer to the second NAD2 device. The second NAD2 device will then be used t o excite a signal in the CA1 region of the separated hippocampal slice, and this signal will be confirmed through the glass electrode. The final demonstration will use the NAD2 device n ot only to bridge between separated neurons, but to show that these devices c an be utilized to perform physiological actions by eliciting LTP in organotypic hippocampal slice cultures. Hippocampal slices will be separated as previously described and plate d on dual NAD2s connected to computer controlled electronics. First, we will es tablish baseline fEPSP responses to stimulation from a bipolar stimulating electrode. F ollowing baseline, we will induce LTP

PAGE 174

163 in the CA1 region of the separated hippocampal slic e halves through high frequency stimulation (HFS) of the CA2/3 regions in the other half of the hippocampus. Following HFS, we will record fEPSPs again at baseline protoc ols for an additional 60 min. The NAD2 will be used to detect and transmit excitatory signals from the CA2/3 regions of the hippocampus to the computer. The second NAD2 w ill be used to receive signals from the computer and deliver those signals to the CA1 region of the hippocampus. This will not only demonstrate that NAD2s can act as a g ate between severed circuits, but also become a physiological circuit itself in the CNS. This is clearly the last critical hurdle before fu lly-functional NADs based on SiC are developed for clinical trials. Therefore, it is essential that a demonstration of device interaction with brain slices, such as the hippocam pus, be made thus clearing the way for a continuation of this work. Successful demonstrati on of the therapeutic and two way communication ability of the NAD will pave the way for long-term animal implantation trials. 5.5 Conclusion Many people currently are suffering from many prob lems that are the direct result of damage to their central nervous system. This is due to the fact that glial scarring, although protecting the undamaged part of the brain does not readily allow the regeneration of synaptic pathways. However, it has been shown that microelectronic devices can be implanted that allow the reconnectio n of these synaptic pathways, thereby restoring functionality of the lost bodily function s. The major problem with these devices

PAGE 175

164 is that they cause the generation of the very probl em they are trying to resolve. The material of these devices activate astrocyes and mi croglia cells, the primary cells of the CNS immune system, and therefore cause the generati on of another glial scar. Different materials need to be tested that will allow these d evices to work indefinitely in the CNS. This dissertation concentrated on two materials th at may be the solution of this problem. 3C-SiC and NCD are two semiconductor mate rials that we have tested using in vitro methodology and have shown an excellent level of b iocompatibility. One of these materials, 3C-SiC, was utilized to activate an acti on potential in neural cells. These two initial results show great promise for these materi als for the development of a long term implantable neural prosthetic, however much work re mains to be done before an in vivo device can be tested. We have divided the research toward a neural impla nt device into three distinct experiments. The first set of proposed experiments is to further explore the interaction of the surface of SiC and NCD with the CNS cells throu gh chemical modification. The modifications will inevitably lead to greater under standing of the molecular interactions between the proteins of the cells membranes and the molecular and atoms on the surface of the substrate. The second set of proposed exper iments expands the biocompatibility testing to including primary cell lines. The immor talized cell lines used in the previous studies have been derived from cancerous tumors and may not be perfect models for cellular reactions to these materials. Healthy neu rons, astrocytes, and microglia need to be tested with 3C-SiC and NCD before in vivo methods can be attempted. Finally, we propose using these materials to construct both ele ctrode and field effect devices that will

PAGE 176

165 interact with neurons. These new devices will not only activate action potentials, which we have already accomplished, but allow true two-wa y communication through the interception of action potential signals as well. The data collected from these devices will be used to model another generation of more co mplicated devices that can interact with a hippocampal tissue slice. Not only will thi s new device allow two-way communication, but will induce plasticity in the sy naptic pathways of the hippocampal slices. If this research is successful, the materi al will be ready for in vivo testing and may be ready for clinical trials in the near future.

PAGE 177

166 References [1] J. Vidal, "Toward direct brain-computer communicati on", in Annual Review of Biophysics and Bioengineering L. J. Mullins, Ed. Palo Alto, CA, USA: Annual Reviews, Inc., 1973, pp. 157 – 180. [2] M. A. Lebedev, and M. A. L. Nicolelis, “Brain-machi ne interfaces: past, present and future”, Trends Neurosci. vol. 29, no. 9, pp. 536 – 546, 2006. [3] K. D. Wise, A. M. Sodagar, Y. Yao, M. N. Gulari, G. E. Perlin, and K. Najafi, “Microelectrodes, microelectronics, and implantable neural microsystems”, Proceedings of the IEEE vol. 96,no. 7, pp. 1184 – 1202, 2008. [4] C. Richards-Grayson, R. S. Shawgo, and A. M. Johnso n, “A bioMEMS review: MEMS technology for physiologically integrated devi ces”, Proceedings of the IEEE vol. 92, no.1, pp. 6 – 21, 2004. [5] T. James, M. S, Mannoor, and D. V. Ivanov, “BioMEMS – advancing the frontiers of medicine”, Sensors vol. 8, no. 9, pp. 6077 – 6107, 2008. [6] D. F. Williams, “On the mechanisms of biocompatibil ity”, Biomaterials vol. 29, no. 20, pp. 2941 – 2953, 2008. [7] S. C. Bayliss, L. D. Buckberry, P. J. Harris, and M Tobin, “Nature of the siliconanimal cell interface”, Journal of Porous Materials vol. 7, no. 1 – 3, pp. 191 – 195, 2000. [8] J. D. Birshall and A. W. Espie, “Biological implica tions of the interaction (via silanol groups) of silicon with metal ions” in Ciba Foundation Symposium 121 Silicon Biochemistry D. Evered and M. O’Connor, Eds. Chichester, UK: J ohn Wiley & Sons, 1986, pp. 140 – 159. [9] C. O’Neil, P. Jordan, T. Bhatt, and R. Newman, “Sil ica and oesophageal cancer” ” in Ciba Foundation Symposium 121 Silicon Biochemistr y D. Evered and M. O’Connor, Eds. Chichester, UK: John Wiley & Sons, 1 986, pp. 214 – 230. [10] L. L. Hench and J. Wilson, “Biocompatibility of sil icates for medical use” ” in Ciba Foundation Symposium 121 Silicon Biochemistr y D. Evered and M. O’Connor, Eds. Chichester, UK: John Wiley & Sons, 1 986, pp. 231 – 253.

PAGE 178

167 [11] J. D. Birchall, and J. S. Chappel, “The chemistry o f aluminum and silicon in relation to Alzheimer’s disease”, Clin. Chem. vol. 34, no. 2, pp. 265 – 267, 1988. [12] F. Beaudoin, M. Simard-Normandin, and M. Meunier, “ Metallic contamination from wafer handling” in Silicon Recombination Lifetime Characterization Methods D. C. Gupta, F. Bacher, and W. H. Hugh, Eds. West Conshohocken, PA, USA: American Society for Testing Materials, 19 98, pp. 219 – 225. [13] K. D. Wise, J. B. Angell, and A. Starr, “An integra ted-circuit approach to extracellular microelectrodes”, IEEE Transactions on Bio-medical engineering vol. BME-17, no. 3, pp. 238 – 247, 1970. [14] S. F. Cogan, “Neural stimulation and recording elec trodes”, Annu. Rev. Biomed ., vol. 10, pp. 275 – 309, August 2008. [15] P. Fromherz, “Neuroelectronic interfacing: semicond uctor chips with ion channels, nerve cells, and brain” in Nanoelectronics and Information Technology: Advanced Electronic Materials and Novel Devices R. Waser, Ed., Berlin, DE: Wiley-VCH, 2003, pp. 781 – 810. [16] X. Li, X. Wang, R. Bondokov, J. Morris, Y. H. An, a nd T. S. Sudarshan, “ Micro/nanoscale mechanical and tribological charact erization of SiC for orthopedic applications”, J. Biomed. Mater. Res. B Appl. Biomater., vol. 72, no. 2, pp. 353 – 361, 2005. [17] U. Kalnins, A. Erglis, I. Dinne, I. Kumsars, and S. Jegere, “Clinical outcomes of silicon carbide coated stents in patient with coron ary disease”, Med. Sci. Monit ., vol. 8, no. 2, pp. 16 – 20, 2002. [18] S. Santavirta, M. Takagi, L. Nordsletten, A. Anttil a, R. Lappalainen, and Y. T. Konttinen, “Biocompatibility of silicon carbide in colony formation test in vitro : A promising new ceramic THR implant coating materia l”, Arch. Orthop. Trauma Surg. vol. 118, no. 1 – 2, pp. 89 – 91, 1998. [19] A. Naji and M. Harmand, “Cytocompatibility of two c oating materials, amorphous alumina and silicon carbide, using human differentiated cell cultures”, Biomaterials vol. 12, no. 7, pp. 690 – 694, 1991. [20] R. Yakimova, R. M. Petoral Jr, R. M. Yazdi, G. R. Vahlberg, C. L. Spetz, and A. K. Uvdal, “Surface functionalization and biomedical applications based on SiC” J. Phys. D: Appl. Phys. vol. 40, pp. 6435 – 6442, October 2007.

PAGE 179

168 [21] O. Kordina and S. E. Saddow. “Silicon carbide overv iew” in Advances in Silicon Carbide Processing and Applications S. E. Saddow and A. Agarwal, Eds., Boston, MA, USA :Artech House, Inc., 2004, pp. 2 – 3, 7 – 8, 18. [22] M. Reyes, Y. Shishkin, S. Harvey, and S. E. Saddow, “Development of a highgrowth rate 3C-SiC on Si CVD process”, in Mater. Res. Soc. Symp. Proc. vol. 911, 2006, pp. 79 – 84. [23] S. E. Saddow, M. Lang, T. Dalibor, and G. Pensl. "T hermal capacitance spectroscopy of epitaxial 3C and 6H-SiC pn junction diodes grown side by side on a 6H-SiC substrate", Appl. Phys. Lett. vol. 66, no. 26, pp. 3612 – 3614, 1995. [24] K. Das, “Diamond and silicon carbide heterojunction bipolar transistor”, U. S. Patent 5,285,089, February 8, 1994. [25] National Institute on Aging, “Inside the human brai n”. [Online]. Available: veling/Part1/inside.htm [Accessed: May 20, 2009]. [26] C. N. Svendsen, "The amazing astrocyte", Nature vol. 417, no. 2, pp. 29 – 32, 2002. [27] M. Santello and A. Volterra, "Synaptic modulation b y astrocytes via Ca2+ dependent glutamate release", Neuroscience vol. 158, no. 1, pp. 253 – 259, 2009. [28] K. J. Tomaselli, K. M. Neugebauer, J. L. Bixby, J. Lilien, and L. F. Reichardt, "Ncadherin and integrins: two receptor systems that m ediate neuronal process outgrowth on astrocyte surfaces", Neuron vol. 60, no. 3, pp. 398 – 399, 2008. [29] T. Ishibashi, K. Dakin, B. Stevens, P. Lee, S. Kozl ov, C. Stewart, and R. Fields, "Astrocytes promote myelination in response to elec trical impulses". Neuron vol. 49, no. 6, pp. 823 – 832, 2006. [30] J. W. Fawcett and R. A. Asher, “The glial scar and central nervous system repair”, Brain Research Bulletin vol. 49, no. 6, pp. 377 – 391, 1999. [31] V. S. Polikov, P. A. Tresco, and W. M. Reichert, “R esponse of brain tissue to chronically implanted neural electrodes”, J. of Neuroscience Methods, vol. 148, pp. 1 – 18, August 2005. [32] L. Olsen, “Regeneration in the adult nervous system : experimental repair strategies”, Nature Medicine vol. 3, no. 12, pp. 1329 – 1335, 1997.

PAGE 180

169 [33] A. L. Hodgkin and A. F. Huxley, “Currents carried b y sodium and potassium ions through the membrane of the giant axon of Loligo”, J. Phusiol. (Lond.) vol. 116, no. 4, pp. 449 – 472, 1952. [34] A. L. Hodgkin and A. F. Huxley, “The components of membrane conductance in the giant axon of Loligo”, J. Phusiol (Lond.) vol. 116, no. 4, pp. 473 – 496, 1952. [35] A. L. Hodgkin and A. F. Huxley, “The dual effect of membrane potential on sodium conductance in the giant axon of Loligo”, J. Phusiol (Lond.) vol. 116, no. 4, pp. 497 – 506, 1952. [36] A. L. Hodgkin and A. F. Huxley, “A quantitative des cription of membrane current and its application to conduction and excitation in nerve”, J. Phusiol (Lond.) vol. 117, no. 4, pp. 500 – 544, 1952. [37] R. Plonsey, and R. C. Barr, “Action potential chara cteristics” in Bioelectricity: A quantitative approach 2nd edition, New York, NY, USA: Kluwer Academic/ Plenum Publishers, 2000, pp. 125 – 127. [38] D. W. Kennard, "Glass microcapillary electrodes use d for measuring potential in living tissue," in Electronic Apparatus for Biological Research P. E. K. Donaldson, Ed. New York, NY, USA: Academic Press, 1 958, pp. 534 – 567. [39] D. H. Hubel, “Tungsten microelectrode for recording from single units”, Science vol. 125, no. 3247, pp. 549 – 550, 1957. [40] D. A. Robinson, “The electrical properties of metal microelectrodes”, Proc. IEEE vol. 56, no. 6, pp. 1065 – 1071, 1968. [41] S. Martinoia, P. Massobrio, M. Bove, and G. Massobr io, "Cultured neurons coupled to microelectrode arrays: circuit models, s imulations and experimental data", IEEE Transactions on Biomedical Engineering vol. 51, no. 5, pp. 859 – 864, 2004. [42] P. Fromherz, C.O. Muller, and R. Weis, "Neuron tran sistor: electrical transfer function measured by the patch-clamp technique," Physical Review Letters vol. 71, no. 24, pp. 4079 – 4082, 1993. [43] D. Braun, and P. Fromherz, “Fast voltage transients in capacitive silicon-to-cell stimulation detected with a luminescent molecular e lectronic probe”, Phys. Rev. Lett ., vol. 86, no. 13, pp. 2905 – 2908, 2001.

PAGE 181

170 [44] I. Schoen, and P. Fromherz, “Extracellular stimulat ion of mammalian neurons through repetitive activation of Na+ channels by weak capacitive currents on a silicon chip”, J Neurophysiol, vol. 100, pp. 346 – 357, May 2008. [45] A. Stett, B. Mller, and P. Fromherz, "Two-way sili con-neuron interface by electrical induction," Physical Review E vol. 55, no. 2, pp. 1779 – 1782, 1997. [46] D. Feili, M. Schuettler, and T. Stieglitz, “Matrixaddressable, active electrode arrays for neural stimulation using organic semicon ductors-cytotoxicity and pilot experiments in vivo ”, J. Neural Eng. vol. 5, no. 1, pp. 68 – 74, 2008. [47] M. M. Moore-Jackson, S. G. Mason, and G. E Birch, “ Analyzing trends in brain interface technology: A method to compare studies”, Annals of Biomedical Engineering vol. 34, no. 5, pp. 859 – 878, 2006. [48] R. Patrick, C. Ryan, C. Visser, J. Cruz, and J. Si, “Neurabotics: an integration of neurons, rats, and robots for advanced studies of b rain-computer interface systems”, in Proceedings of the 1st International IEEE EMBS Conf erence on Neural Engineering 2003, pp. 462 – 465. [49] J. P Donoghue, “Connecting cortex to machines: rece nt advances in brain interfaces”, Nature Neuroscience Supplement, vol. 5 Suppl., pp. 1085 – 1088, November 2002. [50] C. Guger, H. Ramoser, and G. Pfurtscheller, “Real-t ime EEG analysis with subject-specific spatial patterns for a brain-compu ter interface (BCI)”, IEEE Transaction on Rehabilitation Engineering vol. 8, no. 4., pp. 447 – 456, 2000. [51] M. Shiraishi and H. Sumiya, “Manipulator operation by using brain-wave signals” in IEEE/ RSJ International Conference on Intelligent R obots and Systems 2005, pp. 3105 – 3110. [52] M. M. Moore-Jackson and U. Dua, “A galvanic skin re sponse interface for people with severe motor disabilities”, in ACM SIGACCESS Access and Computing vol. 77 – 78, 2003, pp. 48 – 54. [53] J. Wessberg, C. R. Stambaugh, J. D. Kralik, P. D. B eck, M. Laubach, J. K. Chapin, J. Kim, S. J. Biggs, M. A. Srinivasan, and M. A. L. Nicolelis, “Real-time prediction of hand trajectory by ensembles of corti cal neurons in primates”, Nature vol. 408, pp. 361 – 365, November 2000. [54] W. Jensen and P. J. Rousche, “On variability and us e of rat primary motor cortex responses in behavioral task discrimination”, J. Neural Eng. vol. 3, no. 1, pp. L7 – L13, 2006.

PAGE 182

171 [55] L. R. Hochberg, M. D. Serruya, G. M. Friehs, J. A. Mukand, M. Saleh, A. H. Caplan, A. Branner, D. Chen, R. D. Penn, and J. P. Donoghue, “ Neuronal ensemble control of prosthetic devices by a human w ith tetraplegia”, Nature vol. 422, pp. 164 – 171, June 2006. [56] R. J. S. Briggs, H. C. Eder, P. M. Seligman, R. S. C. Cowan, K. L. Plant, J. Dalton, D. K. Money, and J. F. Patrick, “ Initial c linical experience with a totally implantable cochlear implant research device”, Otology & Neurotology vol. 29, no. 2, pp. 114 – 119, 2008. [57] A. Y. Mogilner, A. L. Benabid, and A. R. Rezai, “Ch ronic therapeutic brain stimulation: history, current clinical indications, and future prospects”, in Bioelectromagnetic Medicine Paul J. Rosh, Ed. New York, NY, USA: Marcel Dekker, 2004, pp. 133 – 151. [58] P. J. Rousche and R. A. Normann, “Chronic recording capability of the Utah intracortical electrode array in cat sensory cortex ”, Journal of Neuroscience Methods vol. 82, no. 1, pp. 1 – 15, 1998. [59] J. C. Williams, R. L. Rennaker, and D. R. Kipke, “L ong-term neural recording characteristics of wire microelectrode arrays impla nted in cerebral cortex” Brain Res. Protoc ., vol. 4, no. 3, pp. 303 – 313, 1999. [60] J. C. Williams, R. L. Rennaker, and D. R. Kipke, “S tability of chronic multichannel neural recordings: implications for a long-term neural interface”, Neurocomputing vol. 26 – 27, pp. 1069 – 1076, 1999. [61] S. A. Boppart, B. C. Wheeler, and C. S. Wallace, “A flexible perforated microelectrode array for extended neural recordings ”, IEEE Trans. Biomed. Eng. vol. 39, no. 1, pp. 37 – 42, 1992. [62] P. J. Rousche, D. S. Pellinen, D. P. Pivin Jr., J. C. Williams, R. J. Vetter, and D. R. Kipke, “Flexible polyimide-based intracortical e lectrode arrays with bioactive capability”, IEEE Trans. Biomed. Eng. vol. 48, no. 3, pp. 361 – 371, 2001. [63] A. Jackson and E. E. Fetz, "Compact movable microwi re array for long-term chronic unit recording in cerebral cortex of primat es", J. Neurophysiol. vol. 98, pp. 3109 – 3118, September 2007. [64] K. K. Lee, J. P. He, A. Singh, S. Massia, G. Ehtesh ami, B. Kim, B. Kima, and G. Rauppc, “Polyimide based intracortical neural impla nt with improved structural stiffness”, J. Micromech. Microeng. vol. 14, no. 1, pp. 32 – 37, 2004.

PAGE 183

172 [65] J-M. Hsua, 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, 2007. [66] Cree Inc., [Online]. Available: [Accessed: May 20, 2009]. [67] C. M. Tanner, C. Frewin, S. E. Saddow, and J. P. Ch ang, “Electrical performance of Al2O3 gate dielectric films deposited by atomic layer de position on 4H-SiC”, Appl. Phys. Lett. vol. 91, no. 20, pp. 203510-1 – 203510-3, 2007. [68] P. G. Neudeck, D. J. Larkin, J. E. Starr, J. A. Pow ell, C. S. Salupo, and L. G. Matus, "Electrical properties of epitaxial 3Cand 6H-SiC p-n junction diodes produced side-by-side on 6H-SiC substrates", IEEE Transactions on Electron Devices vol. 41, no. 5, pp. 826 – 835, 1994. [69] P. G. Neudeck, A. J. Trunek, D. J. Spry, J. A. Powe ll, H. Du, M. Skowronski, X. R. Huang, and M. Dudley, "CVD growth of 3C-SiC on 4 H/6H mesas", Chemical Vapor Deposition vol. 12, no. 8 – 9, pp. 531 – 540, 2006. [70] C. A. Zorman, M. Mehregany, “Materials for microele ctromechanical systems”, in The MEMS Handbook: Introduction and Fundamentals M. Gad-el-Hak, Ed., Boca Raton, FL, USA:CRC Press, 2001, pp. 15 – 23. [71] Y. Goldberg, M. Levinshtein, and S. Rumyantsev, "Si licon carbide (SiC)", in Properties of Advanced Semiconductor Materials: GaN AIN, InN, BN, SiC, SiGe M E. Levinshtein, S. L. Rumyantsev, M. S. Shur, Eds ., New York, NY, USA: John Wiley and Sons, 2001, pp. 93 – 147. [72] J. Deva Reddy, A. A. Volinsky, C. Frewin, C. Locke, and S. E. Saddow, “Mechanical properties of single and polycrystallin e SiC thin films” in Mat. Res. Soc. Symp. Proc. vol. 1049, 2008, pp. AA03 – AA06. [73] M. Kuhn and D. Silversmith, “Ionic contamination an d transport of mobile ions in MOS structures", J. Electrochem. Soc ., vol. 118, no. 6, pp. 966 – 970, 1971. [74] S. Raider, L. Gregor, and R. Flitsch, “Transfer of mobile ions from aqueous solutions to the silicon dioxide surface”, J Electrochem. Soc ., vol. 120, no. 3, pp. 425 – 431, 1973. [75] E. H. Snow, A. S. Grove, B. E. Deal, and C. T. Sah, "Ion transport phenomena in insulating films", J. Appl. Phys. vol. 36, no. 5, pp. 1664 – 1673, 1965.

PAGE 184

173 [76] B. G. Streetman and S. Banerjee, Solid State Electronic Devices Upper Saddle River, NJ, USA: Prentice Hall, 2000, pp. 274 – 275. [77] E. M. Carlisle, “Silicon as an essential trace elem ent in animal nutrition” in Ciba Foundation Symposium 121 Silicon Biochemistry D. Evered and M. O’Connor, Eds., Chichester, UK: John Wiley & Sons, 1986, pp. 123 – 139. [78] M. Amaral, A. G. Dias, P. S. Gomes, M. A. Lopes, R. F. Silva, J. D. Santos, and M. H. Fernandes, “Nanocrystalline diamond: In vitro biocompatibility assessment by MG63 and human bone marrow cells cultures”, J. Biomed. Mater. Res. A vol. 87, no. 1, pp. 91 – 99, 2008. [79] Diamond at Workr, “Properties of Diamond”, Knowledge Base: Properti es. [Online]. Available: [Accessed: May 20, 2009]. [80] Kobelco, Kobe Steel, Ltd” Comparison of materials p roperties between diamond and other semiconducting materials”, Properties of Diamond. [Online]. Available: 022e.htm [Accessed: May 20, 2009]. [81] Diamond Engineering, “Material properties for MOS c ap EM simulation”, Microwave Library: MOS Cap EM Models [Online]. Avai lable: f [Accessed: May 20, 2009]. [82] MEMSnet", “Material: Silicon Dioxide (SiO2), film”, Materia l: Silicon Dioxide (SiO2), film. [Online]. Available: esio2film [Accessed: May 20, 2009]. [83] F. K. de Theije, O. Roy, N. J. van der Laagand, and W. J. P. van Enckevort, "Oxidative etching of diamond", Diamond and Related Materials vol. 9, no. 3 – 6, pp. 929 – 934, 2000. [84] F. Fang, C. A. Hewett, M. G. Fernandez, and S. S. L au, “Ohmic contacts formed by ion mixing in the Si-Diamond system”, IEEE Transactions on Electronic Devices vol. 36, no. 9, pp. 1783 – 1786,1989. [85] Y. M. Tairov, V. F. Tsvetkov, M. A. Chernov, and V. A. Tarantes, "Investigation of phase transformations and polytype stability of -SiC", Phys. Stat. Sol. (a) vol. 43, no. 1, pp. 363 – 369, 1977. [86] W. G. Spitzer, D. A. Kleinman, and C. J. Frosch, "I nfrared properties of cubic silicon carbide films", Physical Review vol. 113, no. 1, pp. 133 – 136, 1959.

PAGE 185

174 [87] H. Nakashima, T. Sugano, and H. Yanai, "Epitaxial g rowth of SiC film on silicon substrate and its crystal structure", Jap. J. of Appl. Phys. vol. 5, no. 10, pp. 874 – 878, 1966. [88] S. Nishino, H. Suhara, H. Ona, and H. Matsunami, "E pitaxial growth and electric characteristics of cubic SiC on silicon", J. Appl. Phys. vol. 61, no. 10, pp 4889 – 4893, 1987. [89] J. Yun, T. Takahashi, Y. Ishida, and H. Okumu, "Dep endence of stacking fault and twin densities on deposition conditions during 3C-SiC heteroepitaxial growth on on-axis Si(001) substrates", Journal of Crystal Growth vol. 291, no. 1, pp. 140 – 147, 2006. [90] S. Mahajan and K. S. Sree Harsha, “3.3 Structural c haracteristics of defects” in Principals of Growth and Processing of Semiconducto rs Boston, MA, USA: McGraw Hill Companies Inc., 1999, pp. 83 – 113. [91] Y. Sun, T. Miyasato, J. K. Wigmore, N. Sonoda, and Y. Watari, "Characterization of 3C-SiC films grown on monocrystalline Si by reac tive hydrogen plasma sputtering", J. Appl. Phys. vol. 82, no. 5, pp 2234 – 2341, 1997. [92] J. P. Li and A. J. Steckl, "Nucleation and void for mation mechanisms in SiC thin film growth on Si by carbonization", J. Electrochem. Soc. vol. 142, no. 2, pp. 634 – 641, 1995. [93] V. Papaioannou, H. Moller, M. Rapp, L. Vogelmeier, M. Eickhoff, G. Krotz, and J. Stoemenos, "The evolution of cavities in Si at t he 3C-SiC:Si interface during 3C-SiC deposition by LPCVD", Materials Science and Engineering B vol.61 – 62, pp. 539 – 543, July 1999. [94] O. Kordina, L.-O. Bjorketun, A. Henry, C. Hallin, R C. Glass, L. Hultman, J.-E. Sundgren, and E. Janen, "Growth of 3C-SiC on on-axi s (100)Si substrates by chemical vapor deposition", Journal of Crystal Growth vol. 154, no. 3, pp. 303 – 314, 1995. [95] C. Frewin, “Design and Implementation of a 200 mm 3 C-SiC CVD Reactor”, M.S.E.E. Thesis, University of South Florida, Tampa FL, 2006. [96] A. Severino, C. Frewin, R. Anzalone, C. Bongiorno, P. Fiorenza, G. D’Arrigo, F. Giannazzo, G. Foti, F. La Via and S. E. Saddow, “Gr owth of 3C-SiC on Si: Influence of process pressure,” in Materials Science Forum vols. 600 – 603, 2009, pp 211 – 214.

PAGE 186

175 [97] C. Locke, R. Anzalone, A. Severino, C. Bongiorno, G Litrico, F. La Via, and S. E. Saddow, “High quality single crystal 3C-SiC(111) films grown on Si(111)”, presented in European Conf. on SiC and Related Materials September 2008, Tu1-2. [98] G. F. McLane and J. R. Flemish, “High etch rates of SiC in magnetron enhanced SF plasmas”, Appl. Phys. Lett. vol. 68, no. 26, pp. 3755 – 3757, 1996. [99] G. Carter, J. B. Casady, M. Okhuysen, J. D. Scofiel d, and S. E. Saddow, “Preliminary investigation of 3C-SiC on silicon for biomedical applications,” in Materials Science Forum vols. 338 – 342, 2000, pp. 1149 – 1152. [100] C. Locke, G. Kravchenko, P. Waters, J. Deva Reddy, A. A. Volinsky, C. L. Frewin, and S. E. Saddow, “3C-SiC films on Si for M EMS applications: Mechanical Properties,” in European Conf. on SiC and Related Materials September 2008, Late News. [101] M. Reyes, M. Waits, S. Harvey, Y. Shishkin, B. Geil J.T. Wolan and S.E. Saddow, “Growth of 3C-SiC on Si Molds for MEMS appl ications,” in Materials Science Forum vols. 527 – 529, 2006, pp. 307 – 313. [102] S. A. Rosli, A. A. Aziz, and H. A. Hamid, “Characte ristics of RIE SF6/ O2/ Ar plasmas on n-silicon etching”, in IEEE International Conference on Semiconductor Electronics 2006, pp. 851 – 859. [103] S. Frederico, C. Hilbert, R. Fritschi, P. Fluckiger P. Renaud, and A. M. Ionescu, “Silicon sacrificial layer dry etching (SSLDE) for free-standing RF MEMS architectures”, in The Sixteenth Annual IEEE International Conference on Micro Electro Mechanical Systems 2003, pp 570 – 573. [104] T. Shimizu, Y. Ishikawa, and N. Shibata, “Epitaxial growth of 3C-SiC on thin silicon-on-insulator substrate by chemical vapor de position using alternating gas supply”, Jpn. J. Appl. Phys. vol. 39, no. 6B, pp. L617 – L619, 2000. [105] R. L. Myers, S. E. Saddow, S. Rao, K. D. Hobart, M. Fatemi, and F. J. Kub, “Development of 3C-SiC SOI structures using Si on p olycrystalline SiC wafer bonded substrates,” in Materials Science Forum vols. 457 – 460, 2004, pp. 1511 – 1514. [106] G. Harbeke, L. Krausbauer, E. F. Steigmeier, A. E. Widmer, H. F. Kappert, G. Neugebauer, "Growth and physical properties of LPCV D polycrystalline silicon films”, J. Electrochem. Soc. vol. 131, no. 1 – 2, pp. 675 – 682, 1984.

PAGE 187

176 [107] J. Yang, H. Kahn, A.-Q. He, S. M. Phillips, and A. H. Heuer, “A new technique for producing large-area as-deposited zero-stress L PCVD polysilicon films: the MultiPoly process”, IEEE Journal of Microelectromechanical Systems vol. 9, no. 4, pp. 485 – 494, 2000. [108] C. Locke, C. Frewin, J. Wang, and S. E. Saddow, “Gr owth of single crystal 3CSiC(111) on a poly-Si seed layer,” in European Conf. on SiC and Related Materials, 2008, WeP-5. [109] S. Dogan, D. Johnstone, F. Yun, S. Sabuktagin, J. L each, A. A. Baski, H. Morko, G. Li, and B. Ganguly, “The effect of hydrogen etch ing on 6H-SiC studied by temperature-dependent current-voltage and atomic fo rce microscopy”, Appl. Phys. Lett. vol. 85, no. 9, pp. 1547 – 1549, 2004. [110] U. Starke, “Atomic structure of SiC surfaces” in Silicon Carbide, Recent Major Advances W. J. Choyke, H. Matsunami, and G. Pensl, Eds., B erlin, DE: Springer Verlag, 2004, pp. 281 – 316. [111] C. Coletti, C. L. Frewin, A. M. Hoff and S. E. Sadd ow, “Electronic passivation of 3C-SiC(001) via hydrogen treatment”, Electrochemical and Solid-State Letters vol. 11, no. 10, pp. H285 – H287, 2008. [112] S. Soubatch, S. E. Saddow, S. P. Rao, W. Y. Lee, M. Konuma, and U. Starke, “Structure and morphology of 4H-SiC wafer surfaces after H2-etching”, Mater. Sci. Forum v ols. 483 – 485, 2005, pp. 761 – 764. [113] C. Coletti, S. E. Saddow, C. L. Frewin, M. Hetzel, C. Virojanadara, and U. Starke, "Surface studies of hydrogen etched 3C-SiC(001) on Si (001)”, Appl. Phys. Lett 91, no. 6, pp. 061914 –060916, 2007. [114] A. Severino, G. D'Arrigo, C. Bongiorno, S. Scalese, F. La Via, and G. Foti, “Thin crystalline 3C-SiC layer growth through carbonizati on of differently oriented Si substrates”, J. of Applied Physics, vol. 102, no. 2, pp. 23518 – 23520, 2007. [115] H. Yoshikawaa, C. Morela, and Y. Kogab, “Synthesis of nanocrystalline diamond films using microwave plasma CVD”, Diamond and Related Materials vol. 10, no. 9 – 10, pp. 1588 – 1591, 2001. [116] L. Olsen, “Regeneration in the adult nervous system : experimental repair strategies”, Nature Medicine vol. 3, no. 12, pp. 1329 – 1335, 1997. [117] C. C. Stichel and H. W. Mller, “The CNS lesion sca r: new vistas on an old regeneration barrier”, Cell Tissue Res. vol. 294, no. 1, pp. 1 – 9, 1998.

PAGE 188

177 [118] C. S. Bjornsson, S. J. Oh, Y. A. Al-Kofahi, Y. J. L im, K. L. Smith, J. N. Turner, S. De, B Roysa, W Shain, S. J. Kim, “Effects of ins ertion conditions on tissue strain and vascular damage during neuroprosthetic d evice insertion”, J. Neural Eng. 3, pp. 196 – 207, September 2006. [119] R. G. Richards, “The effect of surface roughness on fibroblast adhesion in vitro ”, Injury vol. 27, no. Supl. 3, pp. S-C38 – S-C48, 1996. [120] T. V. Kumari, U. Vasudev, A. Kumar, and B. Menon, ” Cell surface interactions in the study of biocompatibility”, Trend Biomater. Artif. Organs vol. 15, no. 2, pp. 37 – 41, 2002. [121] C. S. Goodman, “Mechanisms and molecules that contr ol growth cone guidance”, Annu. Rev. Neurosci. vol. 19, pp. 341 – 77, March 1996. [122] J. Hirrlinger, S. Hulsmann, and F. Kirchhoff, “Astr oglial processes show spontaneous motility at active synaptic terminals i n situ”, European Journal of Neuroscience vol. 20, no. 8, pp. 2235 – 2239, 2004. [123] D. J. Solecki, E.-E. Govek, and M. E. Hatten, “mPar 6 controls neuronal migration”, The Journal of Neuroscience vol. 26, no. 42, pp. 10624 – 10625, 2006. [124] C. Coletti, M. J. Jaroszeski, A. Pallaoro, A. M. Ho ff, S. Iannotta, and S. E. Saddow, “Biocompatibility and wettability of crysta lline SiC and Si surfaces”, in IEEE EMBC Proceedings vol. 22 – 26, 2007, pp.5849 – 5852. [125] T. Mosmann, “Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assay s”, Journal of immunological methods vol. 65, no. 1 – 2, pp. 55 – 63, 1983. [126] C. Coletti, “Silicon carbide biocompatibility, surf ace control and electronic cellular interaction for biosensing applications”, Ph.D.E.E. dissertation, University of South Florida, Tampa, FL, 2007. [127] C. A. Easley IV, C. M. Brown, A. F. Horwitz, and R. M. Tombes, “CaMK-II promotes focal adhesion turnover and cell motility by inducing tyrosine dephosphorylation of FAK and paxillin”, Cell Motility and the Cytoskeleton vol. 65, no. 8, pp.662 – 674, 2008. [128] M. J. Booth, R. Ju, and T. Wilson, “Spectral confoc al reflection microscopy using a white light source”, Journal of the European Optical Society vol. 3, pp. 080261 – 08026-6, August 2008.

PAGE 189

178 [129] Y. Wakamatsu, X. Zhao, C. Jin, N. Day, M. Shibahara N. Nomura, T. Nakahara, T. Murata, K. K. Yokoyama, “Mannosylerythritol lipi d induces characteristics of neuronal differentiation in PC12 cells through an E RK-related signal cascade”, Eur. J. Biochem., vol. 268, no. 2, pp. 374 – 383, 2001. [130] P. R. Gordon-Weeks, “Microtubules and growth cone f unction”, J. Neurobiol. vol. 58, no. 1, pp. 70 – 83, 2003. [131] A. Rouhi, “Contemporary biomaterials”, C & EN, vol. 77, no. 3, pp. 51 – 59, 1999. [132] K. H. Kim, J.-S. Cho, D.-J. Choi, S.-K Koh, “Hydrop hilic group formation and cell culturing on polystyrene Petri-dish modified b y ion-assisted reaction”, Nuclear Instruments and Methods in Physics Research B vol. 175 – 177, pp. 542 – 547, 2001. [133] M. Bergkvist, J. Carlsson, J. Oscarsson, “Surface-d ependent conformations of human plasma fibronectin adsorbed to silica, mica, and hydrophobic surfaces, studied with use of Atomic Force Microscopy”, J. Biomed. Mater. Res. A ., vol. 64, no. 2, pp. 349 – 356, 2003. [134] P. Clark, S. Britland, and P. Connolly, "Growth con e guidance and neuron morphology on micropatterned laminin surfaces", Journal of Cell Science vol. 105, no. 1, pp. 203-212, 1993. [135] E. G. Fine, R. F. Valentini, R. Bellamkonda, and P. Aebischer, “Improved nerve regeneration tooth piezoelectric vinylidenefluoride -truoroetlene copolymer guidance channels”, Biomaterials vol. 12, no. 8, pp. 775 – 780, 1991. [136] S. A. Makohliso, R. F. Valentini, and P. Aebischer, “Magnitude and polarity of a fluoroethylene propylene electrical substrate charg e influences neurite outgrowth in vitro” J. Biomed. Mater. Res. vol. 27, no. 8, pp. 1075 – 1085, 1993. [137] T.-H. Young and C.-H. Hung, “Change in electrophore tic mobility of PC12 cells after culturing on PVA membranes modified with diff erent diamines”, J. Biomed. Mater. Res. A ., vol. 67, no. 4, pp. 1238 – 1244, 2003. [138] C. E. Schmidt, V. R., Shastri, J. P. Vacanti, and R Langer, “Stimulation of neurite outgrowth using an electrically conducting polymer” Proc. Natl. Acad. Sci. vol. 94, pp. 8948 – 8953, 1997. [139] A. Sklodowska, M. Wozniak, R. Matlakowska, “The met hod of contact angle measurements and estimation of work of adhesion in bioleaching of metals”, Biol. Proc. vol. 1, no. 3, pp. 114 – 121, 1999.

PAGE 190

179 [140] L. Ostrovskaya, V. Perevertailo, V. Ralchenko, A. S aveliev, and V. Zhuravlev, “Wettability of nanocrystalline diamond films”, Diamond and Related Materials, vol. 16, no. 12, pp. 2109 – 2113, 2007. [141] Cree Inc, “MAT-CATALOG”, Materials: Products. [Onli ne]. Available: http:// [Accessed: May 20, 2009]. [142] M. Stutzmann, J. A. Garrido, M. Eickhoff, and M. S. Brandt, "Direct biofunctionalization of semiconductors: A survey", Physica Status Solidi (a) vol. 203, no. 14, pp. 3424 – 3437, 2006. [143] S. J. Schoell, M. Hoeb, I. D. Sharp, W. Steins, M. Eickhoff, M. Stutzmann, and M. S. Brandt, "Functionalization of 6H-SiC surfaces with organosilanes", Appl. Phys. Lett. vol. 92, p. 153301-1 – 153301-3, 2008. [144] G. W. Trucks, K. Raghavachari, G. S. Higashi, and Y J. Chabal, "Mechanism of HF etching of silicon surfaces: A theoretical under standing of hydrogen passivation", Phys. Rev. Lett ., vol. 65, no. 4, pp. 504 – 507, 1990. [145] H. Bender, S. Verhaverbeke, M. Caymax, O. Vatel, an d M. M. Heyns, J., "Surface reconstruction of hydrogen annealed (100) silicon", J. Appl. Phys ., vol. 75, no. 2, pp. 1207 – 1209, 1994. [146] M. R. Linford, C. E. D. Chidsey, "Alkyl monolayers covalently bonded to silicon surfaces", J. Am. Chem. Soc ., vol. 115, no. 26, pp. 12631 – 12632, 1993. [147] M. Hu, S. Noda, T. Okubo, Y. Yamaguchi and H. Komiy ama, "Structure and morphology of self-assembled 3-mercaptopropyltrimet hoxysilane layers on silicon oxide", Applied Surface Science vol. 181, no. 4, pp. 307 – 316, 2001. [148] H. Tsuchida, I. Kamata, and K. Izumi, "Infrared att enuated total reflection spectroscopy of 6H–SiC(0001) and (000-1) surfaces", J. Appl. Phys ., vol. 85, no. 7, pp. 3569 – 3575, 1999. [149] I. D. Sharp, S. J. Schoell, M. Hoeb, M. S. Brandt, and M. Stutzmann, "Electronic properties of self-assembled alkyl monolayers on Ge surfaces” Appl. Phys. Lett ., vol. 92, pp. 223306-1 – 223306-3, 2008. [150] L. A. Lipkin and J. W. Palmour, "Insulator investig ation on SiC for improved reliability," IEEE Trans. Electron Devices, vol. 46, no. 3, pp. 525 – 532, 1999.

PAGE 191

180 [151] P.-H. Puech1, A. Taubenberger, F. Ulrich, M. Krieg, D. J. Muller, and C.-P. Heisenberg, "Measuring cell adhesion forces of prim ary gastrulating cells from zebrafish using atomic force microscopy", Journal of Cell Science vol. 118, no. 18, pp. 4199 – 4206, 2005. [152] Y. F. Dufrne and P. Hinterdorfer, "Recent progress in AFM molecular recognition studies", Pflugers Arch. Eur J Physiol vol. 456, no. 1, pp. 237 – 245, 2008. [153] J. E. Huettner and R. W. Baughman, "Primary culture of identified neurons from the visual cortex of postnatal rats", The Journal of Neuroscience vol. 6, no. 10, pp. 3044 – 3060, 1986. [154] S. Cronier, H. Laude, and J.-M. Peyrin, "Prions can infect primary cultured neurons and astrocytes and promote neuronal cell de ath", Proc. Natl. Acad. Sci. USA vol. 101, no. 33, pp. 12271 – 12276, 2004. [155] T. B. Toha, M. J. Chena, A. Armugama, Z. F. Pengb, Q.-T. Lia, K. Jeyaseelana, and N. S. Cheung, "Antioxidants: promising neuropro tection against cardiotoxin4b-induced cell death which triggers oxidative stre ss with early calpain activation", Toxicon vol. 51, no. 6, pp. 964 – 973, 2008. [156] O. P. Hamill, A. Marty, E. Neher, B. Sakmann, and F J. Sigworth, “Improved patch-clamp techniques for high-resolution current recording from cells and cellfree membrane patches”, Pflugers Arch. Eur. J. Physiol. vol. 391, no. 2, pp. 85 – 100, 1981. [157] P. W. Landfield, L. W. Campbell, S. Y. Hao, and D. S. Kerr, “Aging-related increases in voltage-sensitive, inactivating calciu m currents in rat hippocampus. Implications for mechanisms of brain aging and Alzh eimer's disease,” Ann. N.Y. Acad. Sci. vol. 568, pp. 95 – 105, December 1989. [158] O. Thibault, N. M. Porter, P. and W. Landfield, “Lo w Ba2+ and Ca2+ induce a sustained high probability of repolarization openin gs of L-type Ca2+ channels in hippocampal neurons: physiological implications”, Proc. Natl. Acad. Sci. USA vol. 90, pp. 11792 – 11796, December 1993. [159] L. W. Campbell, Su-Yang Hao, Olivier Thibault, Eric M. Blalock, and Philip W. Landfield, “Aging changes in voltage-gated calcium currents in hippocampal CA1 neurons”, J. Neurosci. vol. 16, no. 19, pp. 6286 – 6295, 1996. [160] O. Thibault and P. W. Landfield, “Increase in singl e L-type calcium channels in hippocampal neurons during aging”, Science vol. 272, no 5264, pp. 1017 – 1020, 1996.

PAGE 192

181 [161] N. M. Porter, O. Thibault, V. Thibault, K.-C. Chen, and P. W. Landfield, “Calcium channel density and hippocampal cell death with age in long-term culture”, J. Neurosci., vol. 17, no. 14, pp. 5629 – 5639, 1997. [162] O. Thibault, R. Hadley, and P. W. Landfield, “Eleva ted Postsynaptic [Ca2+]i and L-type calcium channel activity in aged hippocampal neurons: relationship to impaired synaptic plasticity”, J. Neurosci. vol. 21, no. 24, pp. 9744 – 9756, 2001. [163] O. Thibault, M. L. Mazzanti, E. M. Blalock, N. M. P orter and P. W. Landfield, "Single-channel and whole-cell studies of calcium c urrents in young and aged rat hippocampal slice neurons", J. of Neuroscience Methods vol. 59, no. 1, pp. 77 – 83, 1995. [164] R. W. Tsien, D. Lipscombe, D. V. Madison, K. R. Ble y, and A. P. Fox, "Multiple types of neuronal calcium channels and their select ive modulation", Trends Neurosci., vol. 11, no. 10, pp. 431 – 438, 1988. [165] B. P. Bean, “Classes of calcium channels in vertebr ate cells”, Annu Rev Physiol vol. 51, pp. 367 – 384, March 1989. [166] W. A. Catterall, M. J. Seagar, M. Takahashi, and K. Nunoki, ”Molecular properties of voltage-sensitive calcium channels”, Adv. Exp. Med. Biol. vol. 255, no. 1, pp. 101 – 109, 1989. [167] W. A. Catterall, M. J. Seagar, M. Takahashi, and K. Nunoki, “Molecular properties of dihydropyridine-sensitive calcium cha nnels“, Ann N Y Acad Sci vol. 560, no. 1, pp. 1 – 14, 1989. [168] L. M. Grover and T. J. Teyler, “Two components of l ong-term potentiation induced by different patterns of afferent activatio n”, Nature vol. 347, no. 6292, pp. 477 – 479, 1990. [169] C. M. Norris, S. Halpain, and T. C. Foster, “Revers al of age-related alterations in synaptic plasticity by blockade of L-type Ca2+ channels”, J. Neurosci. vol. 18, no. 9, pp. 3171 – 3179, 1998. [170] T. V. Bliss and T. Lomo, “Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit follow ing stimulation of the perforant path”, J. Physiol., vol. 232, pp. 331 – 356, July 1973. [171] C. M. Norris, D. L. Korol, and T. C. Foster, “Incre ased susceptibility to induction of long-term depression and long-term potentiation reversal during aging”, J. Neurosci. vol. 16, no. 17, pp. 5382 – 5392, 1996.

PAGE 193

182 [172] G. Winocur and M. Moscovitch, “Hippocampal and pref rontal cortex contributions to learning and memory: analysis of l esion and aging effects on maze learning in rats “, Behav. Neurosci. vol. 104, no. 4, pp. 544 – 551, 1990. [173] G. Winocur, “A comparison of normal old rats and yo ung adult rats with lesions to the hippocampus or prefrontal cortex on a test o f matching-to-sample”, Neuropsychologia vol. 30, no. 9, pp. 769 – 781, 1992.

PAGE 194

183 Bibliography I. B. Levitan and L. K. Kaczmarek, The Neuron: Cell and Molecular Biology 3rd edition, New York, NY, USA: Oxford University Press, 2002. D. Purves, G. J. Augustine, D. Fitzpatrick, W. C. H all, A-S. LaMarta, J. O. McNamara, and L. E. White, Eds., Neuroscience fourth edition, Sunderland, MA, USA: Sinauer Associates, Inc., 2008. R. Plonsey and R. C. Barr, Bioelectricity: A Quantitative Approach New York, NY, USA: Kluwer Academic / Plenum Publishers, 2000.

PAGE 195

1 About the Author Christopher Leroy Frewin was born on November 25th, 1970 in Salt Lake City, Utah. He started at the University of South Florid a in the summer of 2002 and earned both his bachelor’s and master’s degree December 20 07. During his graduate work, he has collaborated and worked at the Max-Planck-Insti tut (MPI) of Stuttgart (DE) and the Instituto per la Microelettronica e Microsistemi Consiglio Nazionale Delle Ricerche (IMM-CNR) in Catania, Italy. He traveled to and pe rformed research at both of these institutions two times each. He is the first autho r of two peer-reviewed journal papers (one in the Journal of Molecular Recognition and the other is in Journal of Crystal Growth ), is co-author of three other journal papers, and has presented and written articles in proceedings for numerous conferences. A provisi onal patent, serial no. 61/187808, based on the results of this work has been filed wi th the US patent office.

xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200385Ka 4500
controlfield tag 001 002317626
005 20100903143133.0
007 cr cnu|||uuuuu
008 100903s2009 flua ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0003079
TK145 (Online)
1 100
Frewin, Christopher L.
4 245
The neuron-silicon carbide interface
h [electronic resource] :
b biocompatibility study and BMI device development /
by Christopher L. Frewin.
[Tampa, Fla] :
University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains 183 pages.
Includes vita.
Dissertation (Ph.D.)--University of South Florida, 2009.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
ABSTRACT: Damage to the central nervous system (CNS) leads to the generation of an immune response which culminates with the encapsulation of the damaged area. The encapsulation, known as a glial scar, essentially breaks neural signal pathways and blocks signal transmissions to and from the CNS. The effect is the loss of motor and sensory control for the damaged individual. One method that has been used successfully to treat this problem is the use of a brain-machine interface (BMI) which can intercept signals from the brain and use these signals to control a machine. Although there are many types of BMI devices, implantable devices show the greatest promise with the ability to target specific areas of the CNS, with reduced noise levels and faster signal interception, and the fact that they can also be used to send signals to neurons.The largest problem that has plagued this type of BMI device is that the materials that have been used for their construction are not chemically resilient, elicit a negative biological response, or have difficulty functioning for extended periods of time in the harsh body environment. Many of these implantable devices experience catastrophic failure within weeks to months because of these negative factors. New materials must be examined to advance the future utilization of BMI devices to assist people with CNS damage or disease. We have proposed that two semiconductor materials, cubic silicon carbide (3C-SiC) and nanocrystalline diamond (NCD), which should provide solutions to the material biocompatibility problems experienced by implantable BMI devices.We have shown in this study that these two materials show chemical resilience to neuronal cellular processes, and we show evidence which indicates that these materials possess good biocompatibility with neural cell lines that, in the worst case, is comparable to celltreated polystyrene and, in most cases, even surpasses polystyrene. We have utilized 3C-SiC within an electrode device and activated the action potential of differentiated PC12 cells. This work details our initial efforts to modify the surfaces of these materials in order to improve cellular interaction and biocompatibility, and we examine our current and future work on improving our implantable BMI devices.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
Advisor: Stephen E. Saddow, Ph.D.
Cubic silicon carbide
Nanocrystalline diamond
Implantable neuronal prosthetics
Mammalian cell biocompatability
Neural action potential
0 690
Dissertations, Academic
x Electrical Engineering
t USF Electronic Theses and Dissertations.