Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Dimensional modification of vertically aligned carbon nanotubes using miniaturized arc and glow discharges Sarwar, Mirza Saquib us 2014

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2014_spring_sarwar_mirza.pdf [ 3.16MB ]
Metadata
JSON: 24-1.0166888.json
JSON-LD: 24-1.0166888-ld.json
RDF/XML (Pretty): 24-1.0166888-rdf.xml
RDF/JSON: 24-1.0166888-rdf.json
Turtle: 24-1.0166888-turtle.txt
N-Triples: 24-1.0166888-rdf-ntriples.txt
Original Record: 24-1.0166888-source.json
Full Text
24-1.0166888-fulltext.txt
Citation
24-1.0166888.ris

Full Text

DIMENSIONAL MODIFICATION OF VERTICALLY ALIGNED CARBON NANOTUBES USING MINIATURIZED ARC AND GLOW DISCHARGES  by Mirza Saquib us Sarwar  B.Sc., Bangladesh University of Engineering and Technology, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE  in The Faculty of Graduate and Postdoctoral Studies (Electrical and Computer Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) March 2014 ? Mirza Saquib us Sarwar, 2014   ii  Abstract  There has been a significant rise in interest in carbon nanotube (CNT) structures to be implemented in micro electro mechanical systems (MEMS) devices owing to the appealing characteristics possessed by CNTs. Electrical discharges may be used to modify and machine CNT forests generating a myriad of applications from the treated structures. High current density arc discharge may be used to machine the CNT forest and generate micro structures while low current density glow discharge may be used to treat the CNT surface and change its characteristics for relevant applications. Techniques for fabricating CNT based MEMS are being developed for some time owing to the escalating demand, and have seen substantial progress over the years. The goal is to develop a technique that can provide precise machining with high throughput, and being economically feasible at the same time. This process employs an array of Cu electrodes microfabricated through an advanced UV-LIGA process enabled with a new photoresist system in combination with electroplating, providing a low-cost path to constructing high-density arrays of ?EDM electrodes for high-throughput parallel processing. The fabricated arrays of 85-?m-tall electrodes are utilized to demonstrate and characterize planar dry ?EDM for post-growth patterning of CNT forests in air. Die sinking and scanning processes are tested to show pattern transfers with a 4 ?m tolerance and an average surface roughness of 230 nm. An elemental analysis suggests that contamination of the electrode material on the produced patterns is minimal. Key characteristics in the use of planar electrodes for batch processing of CNT forests are revealed through experimental analysis and discussed in detail. The results suggest that the investigated process is a promising approach toward offering a cost-effective manufacturing technology for future products functionalized with custom-designed microstrucutres of CNT forests. In addition to selectively manipulating the height of the CNT   iii  forest to generate desired structures for MEMS devices, research is conducted to modify the diameter of the nanotubes locally by means of glow discharge to attain further characteristic enhancements derived from such modified CNTs. Literature exists where CNTs have been treated by plasma (mostly for functionalization), but local treatment is a novel approach. The outcome of the research has given rise to results similar to literature where the diameter is decreased by rupturing of outer walls of MWCNT. In addition, extension of the research has resulted in coalesced structures that are reported for the first time. Further extension of the research to radio frequency discharge has been proposed and designed. It is postulated that such local configuration of characteristic parameters of CNT forest surface can have a number of applications in MEMS, electronics, and other possible fields.                 iv  Preface   The major part of fabrication and characterization of the first part of the thesis was conducted by me, while the forests were grown by Masoud Dahmardeh in Dr. Nojeh?s research facility, for the thesis. The second part of the thesis was aided by Dr. Zhimming Xiao and Dr. Tanveer Saleh, who helped in the initial stages of setting up the glow discharge experiments, while the rest of the research was conducted by me.                 v  Table of contents   ABSTRACT ..................................................................................................................................................................... II PREFACE ...................................................................................................................................................................... IV TABLE OF CONTENTS ....................................................................................................................................................... V LIST OF TABLES ............................................................................................................................................................. VII LIST OF FIGURES .......................................................................................................................................................... VIII ACKNOWLEDGEMENTS .................................................................................................................................................... X DEDICATION ................................................................................................................................................................. XI 1. CHAPTER 1 ..................................................................................................................................................... 1 INTRODUCTION .............................................................................................................................................................. 1 1.1 CARBON NANOTUBES OVERVIEW .......................................................................................................................... 4 1.2.1 SYNTHESIS OF CARBON NANOTUBES: ..................................................................................................................... 7 1.2.2 CARBON NANOTUBE FORESTS: ............................................................................................................................. 8 1.2.3 FUNDAMENTALS OF DISCHARGES AND PROCESSING CNTS WITH THEM ......................................................................... 9 1.3.1 MICRO ELECTRO DISCHARGE MACHINING OVERVIEW .............................................................................................. 12 1.3.2 TYPES OF ?EDM ............................................................................................................................................. 13 1.3.3 COMPARATIVE ADVANTAGES OF ?EDM ............................................................................................................... 14 1.3.4 COMPARATIVE COMPATIBILITY OF ?EDM ............................................................................................................ 14 1.3.5 DEVELOPMENT OF ?EDM ................................................................................................................................. 15 1.3.6 PRINCIPLE OF OPERATION ................................................................................................................................. 16 1.3.7 ?EDM COMPONENTS ...................................................................................................................................... 18 1.4 OUTLINE OF THESIS  ......................................................................................................................................... 24 2. CHAPTER 2 ................................................................................................................................................... 26 BATCH MODE ?EDM .................................................................................................................................................... 26 2.1 PROCESS FLOW ............................................................................................................................................... 26 2.2.1 CHOICE OF PHOTORESIST .................................................................................................................................. 28 2.2.2 SUBSTRATE PREPARATION ................................................................................................................................. 29 2.2.3 DEPOSITION OF CU SEED LAYER .......................................................................................................................... 30 2.2.4 PHOTOLITHOGRAPHY PROCESS  .......................................................................................................................... 31 2.2.5 ELECTROPLATING ............................................................................................................................................ 32 2.2.6 REMOVAL OF PHOTORESIST ............................................................................................................................... 33 2.2.7 DIE FORMATION .............................................................................................................................................. 34 2.2.8 GROWTH OF CNT FOREST ................................................................................................................................. 35 2.3 EXPERIMENTAL SETUP ...................................................................................................................................... 36 2.4 CHARACTERIZATION OF THE PROCESS .................................................................................................................. 37 3. CHAPTER 3 ................................................................................................................................................... 47 LOCALIZED GLOW DISCHARGE ........................................................................................................................................ 47 3.1 INTRODUCTION ............................................................................................................................................... 48 3.2 EXPERIMENTAL SETUP FOR LOCAL  DC GLOW DISCHARGE ........................................................................................ 50 3.3 EXPERIMENTAL RESULTS  ................................................................................................................................... 52 3.3.1 EFFECT OF AMBIENCE ON DIAMETER .................................................................................................................... 52 3.3.2 EFFECT OF PROLONGED EXPOSURE TO GLOW DISCHARGE ......................................................................................... 55   vi  3.3.3 ELEMENTAL ANALYSIS OF TREATED CNT SURFACE .................................................................................................. 59 3.4 EXPERIMENTAL SETUP FOR RF GLOW DISCHARGE AND RESULTS ............................................................................... 68 4. CONCLUSIONS AND FUTURE WORK ............................................................................................................. 72 5. REFERENCES ................................................................................................................................................ 75                       vii  List of tables  Table 1.1: Comparison of mechanical properties of CNT with other materials [29] .................................... 6 Table 1.2: Overview of uEDM processes ................................................................................................... 13 Table 1.3: Overview of compatible substrates for different machining techniques .................................... 15 Table 2.1: Spin speeds of HMDS and KMPR 1050.................................................................................... 32 Table 2.2: Path resistance with and without deposited layer ...................................................................... 35 Table 2.3: Results from EDX elemental analysis of patterned CNT forest ................................................ 43                viii  List of figures   Figure 1.1: Conceptual diagram of (A) single walled carbon nanotube and (B) multiwalled carbon nanotube [3] .................................................................................................................................................. 5 Figure 1.2: Vertically aligned CNT forest [46] ............................................................................................. 8 Figure 1.3: Principle of operation [54] ........................................................................................................ 17 Figure 1.4: Schematic of the ?EDM system [56] ....................................................................................... 19 Figure 1.5: Resistor-capacitor pulse generator [57] .................................................................................... 20 Figure 1.6: Transistor type pulse generator [58] ......................................................................................... 22 Figure 2.1: Fabrication process flow diagram ............................................................................................ 28 Figure 2.2: (a) Array of electrodes on a chip (b) SEM images of array...................................................... 34 Figure 2.3: (a) Schematic of experimental setup (b) image of experimental setup..................................... 37 Figure 2.4: Patterns of CNT forest created in the sink mode using (a) 40 V and 10 pF and (b) 60 V and 200 pF ......................................................................................................................................................... 40 Figure 2.5: Patterns of CNT forest created in the sink and scan mode using (a) 40 V and 10 pF and (b) 60 V and 200 pF ............................................................................................................................................... 41 Figure 2.6: AFM images of patterned regions uisng (a) 40 V and 20 pF and (b) 60 V and 200 pF. The measurement was performed with Easyscan 2 (Nanosurf AG, Liestal, Switzerland) ................................ 42 Figure 2.7: A sample waveform of discharge pulses captured in the batch-mode ?EDM of CNT forest .. 44 Figure 2.8: The position of electrode array on the Z axis tracked in real time during machining .............. 46 Figure 3.1: Schematic of DC glow discharge experimental setup .............................................................. 51 Figure 3.2: DC glow on CNT with a 100um diameter electrode ................................................................ 51 Figure 3.3: SEM Images of CNT forest surface before and after glow discharge under varying ambient conditions (a) untreated CNT forest surface (b) glow discharge in Ar = 100%, O2 = 0%, (c) glow discharge in Ar = 97.1%, O2 = 2.9%, (d) glow discharge in Ar = 95.8%, O2 = 4.2%,............................... 53 Figure 3.4: Analysis of CNT diameter from SEM images .......................................................................... 54 Figure 3.5: Exposure to different intensity of glow discharge .................................................................... 56 Figure 3.6: Effect of discharge on CNT at 100?A, varying time (a) untreated surface (b) CNT diameter thinning out in addition to CNTs fusing together (c) CNTs fused together forming thicker bundles and granular structures (d) another case of CNTs fused together ...................................................................... 57 Figure 3.7: Effect of varying current on CNT morphology (a) I = 100 ?A (b) I = 176 ?A (c) I = 240 ?A (a) I = 320 ?A ............................................................................................................................................. 57 Figure 3.8: Outlines that suggest a fusion process ...................................................................................... 58 Figure 3.9: Effect of current ........................................................................................................................ 59 Figure 3.10: EDX summary of DC glow processed CNT forest surface .................................................... 62 Figure 3.11: EDX results with the CNT forest tilted at 45o ........................................................................ 63 Figure 3.12: Raman Spectroscopy results (a) reference CNT surface without background correction (b) reference CNT surface with background correction (c) treated CNT surface without background correction (d) treated CNT surface with background correction) ............................................................... 65 Figure 3.13: Raman spectrum showing gradual degradation of characteristic peaks ................................. 67 Figure 3.14: Schematic of RF glow experimental setup ............................................................................. 69   ix  Figure 3.15: RF glow on stainless steel ...................................................................................................... 70 Figure 3.16: RF Discharge Experimental Setup ......................................................................................... 71                    x  Acknowledgements  I offer my enduring gratitude to the faculty, staff, and my fellow students at the ECE department at UBC, who have inspired me to continue my work in this field. I owe particular thanks to Dr. Kenichi Takahata for being my supervisor and my mentor in this field. I also thank Dr. Alireza Nojeh for all his suggestions and comments regarding my work to make it better and for allowing me to use his lab facilities for my research. I wish to thank Masoud Dahmardeh for helping me with my research from time to time, and more importantly for growing my carbon nanotube forests for me. I also wish to thank Tanveer Saleh for training me in the equipment in the lab and easing me into the research I would be doing. Finally I wish to thank Zhimming Xiao, who helped in in the initial stages of my glow discharge research.  I wish to extend a very special thanks to my parents and my sister for supporting me through my years of education.               xi  Dedication   To my Grandmother                       1  1. Chapter 1 Introduction  Carbon nanotubes (CNT) discovered by S. Lijima [1] in 1991, has been a topic of substantial research since it possesses a myriad of exceptional mechanical, electrical and thermal properties. CNTs may be classified into two main types; single walled carbon nanotubes (SWNT), and multi walled carbon nanotubes (MWNT) [2]. Vertically aligned carbon nanotubes, so called CNT forests, are being considered as a functional bulk material promising for forthcoming micro-electro-mechanical systems (MEMS) and various other applications owing to their unique electrical, mechanical, thermal, and other properties [3]?[6]. To be employed as base materials for MEMS devices and other such applications it is necessary to be able to manipulate the forests to attain desired changes in characteristics (such as conductivity, elasticity, tensile strength, optical reflectivity) and machine structures (such as cones) out of the forests. Different electrical discharges may be used to accomplish these tasks. Locally selective growth of CNT forests can be performed through a typical CVD process with pre-patterned catalyst. However, these patterned structures are limited to two-dimensional like geometries with uniform height. To facilitate the application of this material to further frontiers, it is essential to establish a technique to create free-form, three-dimensional (3D) structures from post-growth forests. For machining the forest, the technology employed needs to be cost effective, fast, and yet produce good quality structures. Micro electro discharge machining (?EDM) employing arc discharge is one such technology that does not require a large capital cost compared to lasers.   2  Literature has proved its ability to machine fine structures on CNT forests [7]. The process gives rise to desirable features with appreciable surface morphology. One drawback of the process is the serial nature of machining owing to the use of a single electrode. To enhance this aspect of the process and render it closer to a commercially viable technology, a batch mode fabrication electrode is proposed whereby using the array of electrodes several structures may be machined simultaneously enhancing the throughput by a factor of 100X. Parallel ?EDM that uses arrays of high-aspect-ratio electrodes were reported for wet processing of metals [8]?[11]. The arrays of electrodes were fabricated using LIGA technology in many cases [8], [9]. LIGA is the process that uses deep X-ray lithography to generate high-aspect-ratio micro molds and replicate the shapes of the molds into metal structures using an electroplating process [12], [13]. Arrays of these electroplated microstructures were demonstrated to effectively work as ?EDM electrodes, enabling parallel micromachining of bulk metals in dielectric oil with >100X throughput over conventional serial ?EDM. X-ray lithography requires the use of synchrotron sources whose availability is significantly limited and renders the process to be costly thus incongruous for commercial implementation. The first part of this thesis is focused on two main objectives: (1) the development of arrayed ?EDM electrodes of copper fabricated through standard ultraviolet (UV) photolithography enabled with a chemically amplified, thick photoresist system, and (2) characterization of batch-mode dry ?EDM of CNT forests, using the fabricated arrays of copper electrodes and that of patterned CNTs. The parallel processing approach to micropatterning of CNT forests is expected to open a novel path to high-throughput production of different types of advanced devices enabled with 3D-patterned structures of the material. Moreover, the use of optical lithography in combination with electroplating, known as UV-LIGA [14]?[16], will make the fabrication of arrayed ?EDM electrode devices compatible with typical integrated-circuit   3  microfabrication facilities hence significantly raising the applicability and cost effectiveness of the approach. Glow discharge is being exploited by researchers to generate plasma and treat CNT forests to enhance specific characteristics by functionalization. Literature proposes the use of glow discharge in functionalization of the CNTs, by adding useful N2 and H2 bearing functional groups among others, which result in enhanced polymer lubrication, increased conductivity, increasing water solubility etc. and may find useful applications in electronics, composites and material chemistry [17]?[20]. It has also been proposed that functionalization is preceded by degradation of the nanotube due to the high energy charged particles bombarding the nanotubes and in effect incur a form of etching, breaking the C-C bonds and generating active sites for bonding of functional groups present in the plasma [18]. With larger energy, sputtering phenomenon can occur, leading to substantial breaking and destroying of the nanotubes [18]. Although there are a number of proposed mechanisms of this degradation the true process is yet unknown. In all these cases during the treatment, the whole CNT forest was being exposed to the plasma. The second part of the thesis focuses on developing a technology to sustain a glow discharge on a CNT forest surface locally, within a small area of a few 1000 ?m2. Being able to locally treat CNT forests would enable the possible designs of a myriad of new devices. The thesis looks into a specific effect of glow on the CNT surface, i.e. the degradation of the CNTs upon exposure to glow discharge. Ar plasma was used to generate an inert atmosphere, and it was observed that the diameters of the nanotubes are reduced which is consistent with literature. The aim in case of [21] was micromachining of the CNT forest to generate micro-structures and the authors used arc discharge. In the mentioned process, the CNTs in the treated regions resulted in having a reduced diameter seen in SEM images [21]. Different theories for the   4  mechanism of truncating the outer walls of the MWCNTs have been proposed. One proposed mechanism is the removal of the C as an oxide when the ambience contained O2 [22], but experiments in an inert atmosphere are also conducted where the rupturing have been hypothesized to have been due to the possibility that CNTs conduct balistically and the energy to break C bonds originates from highly localized dissipation at defect scattering sites [23]. While the literature suggest only reduction of the diameter, this research has observed that the CNTs bundle together being treated by glow discharge, and form coalesced structures that are denser, and can have interesting characteristic properties in terms of electrical, thermal and mechanical properties. A similar phenomenon where coalescence of CNTs have been observed were brought about by annealing [24]. It is postulated that such treatment will change the characteristic properties, and achieving that locally can lead to a number of applications in designing of MEMS devices or electronics.    1.1 Carbon nanotubes overview  Both SWNT and MWNT have been defined in an earlier section of the thesis. SWNTs are structurally similar to a single graphite sheet wrapped into a cylindrical tube and MWNTs comprise a concentric array of such tubes as shown in Figure 1.1. SWNTs can be either metallic or semiconducting, depending on the direction about which the graphite sheet is rolled to form a nanotube cylinder. MWNTs on the other hand are mostly metallic and are able to carry high current densities [25].      5          In case of SWNT the direction determining the conductivity; in the graphite sheet plane and the nanotube diameter are obtainable from a pair of integers (n, m) that denote the nanotube type [26]. Depending on the appearance of a belt of C bonds around the nanotube diameter, the nanotube is either of the armchair (n = m), zigzag (n = 0 or m = 0), or chiral (any other n and m) variety. All armchair SWNTs are metals; those with n ? m = 3k, where k is a nonzero integer, are semiconductors with a small band gap; and all others are semiconductors with a band gap that inversely depends on the nanotube diameter [26]. The nearly one-dimensional electronic structure leads to the electronic transport in metallic SWNTs and MWNTs being ballistic (i.e., without scattering) over long nanotube lengths, enabling them to carry high currents with essentially no heating [27]. Phonons also propagate easily along the nanotube. The measured room temperature thermal conductivity for an individual MWNT (>3000 W/m.K) is greater than that of natural diamond [28]. Superconductivity has been observed at low temperatures. Mechanical properties of CNT are summarized in Table 1.1.   Figure 1.1: Conceptual diagram of (A) single walled carbon nanotube and (B) multiwalled carbon nanotube [3]   6    Their high conductivities and aspect ratios enable them to generate strong electric fields. This renders them to be an efficient source of electrons, as has been demonstrated in field emission displays [25]. Arrays of MWCNTs have been used as electrodes in rectifiers [29], X-ray sources [30], gas sensors [31] and other electrical devices, as nano-electrodes for liquid crystal based electro-optical devices [32], solar cells to increase efficiency [33], optical antenna arrays [34] and photonic crystals [35]. Their applications range further and wider with the progression of technology.    Table 1.1: Comparison of mechanical properties of CNT with other materials [29] Material Young?s Modulus (GPa) Tensile Strength (GPa) Density (g cm-3) SWNT/MWNT ~1000 ~100-200 ~0.7-1.7 High Tensile Steel 210 1.3 7.8 Toray Carbon fibres 230 3.5 1.75 Kevlar 60 3.6 1.44 Glass Fibres 22 3.4 2.6   7  1.2.1 Synthesis of carbon nanotubes  SWNTs and MWNTs are usually fabricated by carbon-arc discharge, laser ablation of carbon, or chemical vapor deposition. Arc-Discharge Method: This is among the pioneering attempts to produce CNTs. Parts of a graphite anode are evaporated by arc discharge generated between two closely spaced graphite rods with a DC voltage applied between them, in a reactor under inert atmosphere [36]. The parts from the evaporated anode are deposited on the reactor walls and the cathode. MWNT can be obtained from the deposits on the cathode. SWNTs can be produced by incorporating metal catalysts in the anode eg. Co and Ni. The yield and quality are substantial. Impediments associated with this method are the need to separate the CNTs from other C deposits and also the difficulty to produce aligned CNTs [37]. Laser Ablation: A high power laser is used to ablate a graphite target with Ar flowing through the system. A furnace heats a quartz tube containing the graphite target (may contain catalysts), and a cooled collector to trap the evaporated C deposits [37]. The yield and quality is good but similar to arc discharge it requires harvesting the CNT.  Chemical Vapor Deposition: A hydrocarbon gas flowed through a tubular reactor decomposes being heated by a furnace and obtaining C deposits on a substrate inside the reactor with catalysts on which the CNTs are formed. It is low cost, simple, ease of control, high yield and operates in ambient pressure [37].    8  1.2.2 Carbon nanotube forests  Vertically aligned carbon nanotubes, so called CNT forests, are being considered as a functional bulk material promising for forthcoming MEMS and various other applications owing to their unique electrical, mechanical, thermal, and other properties [3]?[6]. The CNT forests are grown in in the chemical vapor deposition (CVD) process, in which the nanotubes (including both types SWNTs and MWNTs) are self-aligned due to crowding at the beginning of their growth [38]. These properties along with their immense internal surface area renders them suitable for a range of applications that may include but are not limited to sensors [39], [40], electrical interconnects [41], heat sinks [42], biomimetic adhesives [43], and scanning probes [44].  Figure 1.2: Vertically aligned CNT forest [46]       9  1.2.3 Fundamentals of discharges and processing CNTs with them  There are different types of discharges that are initiated by the breakdown of a dielectric by an electric field and a subsequent avalanche effect with the generation of further charged entities to sustain the discharge. The two discharges relevant to this study are:  Arc discharge: It is a form of electric discharge with a high current density (as high as 1 MAcm-2). An arc may be initiated by ionization and glow discharge as the current is increased. In case of an arc the terminal voltage is much lower than that of glow and has a characteristic high temperature (can be a few hundreds of degree Centigrade going up to a thousand in extreme cases [45]). Another distinguishing feature of arc from glow is that it has approximately equal effective temperature of both electrons and positive ions, while in glow discharge ions have less thermal energy compared to electrons. The current is sustained by both thermionic and field emission. Arcs have ample energy to sustain thermo-mechanical removal of matter from a sample, which is the basis of EDM.  Glow discharge: Glow discharge is basically plasma, i.e. partially ionized gas consisting of electrons, ions and a number of neutral species. It is generated by the application of a potential difference between electrodes (with typical electric field strength of a few MVm-1) inside a chamber filled with gas (usually Ar, with the pressure being typically 50-700 Pa, although atmospheric pressure glow discharge also has applications. With the application of a potential difference, electrons accelerated towards the anode give rise to ionization and excitation collisions with the gas atoms. The ionization collisions create ions and further electrons having an avalanche effect. Electron excitation of neutral atoms give rise to atoms in excited levels which later decays to a lower energy level by emission of radiation. This emission is the cause of   10  the ?glow? of the discharge. Glow discharge has been extensively used in analytical chemistry in a process called ?sputtering?, where the material being analyzed is bombarded by Ar ions to release atoms of the cathode material which upon arrival into the plasma undergo collisions creating ions that may be measured in a mass spectrometer. The excitation collisions yield excited atoms that release photons characteristic of elements present in the material to be analyzed that may be detected with an optical emission spectrometer. In addition to analytical chemistry, Glow discharge is used extensively in etching surfaces or deposition of thin films in semiconductor industry. Glow discharge in DC mode usually operates at hundreds of volts with currents up to maximum of tens of mA, and pressure as low as 50 ? 700 Pa, in addition to atmospheric pressure environments. Radio-frequency (RF) glow discharge is also quite popular for non-conducting materials. In this case, there will be a buildup of negative charge on the ?cathode? at the initial stage owing to a layer of trapped electrons on the non-conducting ?cathode?. The RF frequency is usually 13.56 MHz, with other operating parameters similar to DC mode Glow discharge.   It is important to be able to pattern the CNT forest for it to be useful for the mentioned applications. Such patterns have been formed widely by selective growth by CVD on pre-patterned catalyst on a substrate. However this is limited to two dimensional patterns with uniform height. Micro-patterning of three dimensional, free form structures with high aspect ratios in CNT forests have been demonstrated by use of laser machining and micro electro discharge machining (?EDM). Lasers tend to be substantially more expensive compared to ?EDM including general issues such as depth control, thermal damage, tapered profile of focused beam etc., and with the latter, it is possible to have electrodes as small as 5?m in diameter for machining extremely fine features. In case of EDM, high intensity arc discharges   11  are used to remove matter by thermo-mechanical impacts and generate desired machined structures on the sample. Such micro-patterning using ?EDM have already been demonstrated [7]. The process gives rise to desirable features with appreciable surface morphology. In terms of commercialization of the process, the serial nature of machining acts as an obstacle. A solution to this would be to use an array of electrodes instead of a single one, to machine an array of structures at a time to obtain high throughput. The motivation of developing a cost effective technique to enable high throughput micro machining of CNT forests has been discussed in an earlier section of this thesis. UV LIGA has been incorporated replacing the high cost X-Ray LIGA source in literature that was used to fabricate the high aspect ratio electrode array, following which the machined CNT structures are characterized as a part of this research.  Further to machining complex 3D structures, the second part of this thesis has investigated selective control of the diameter of the CNTs. Modifying the diameter and studying the effects of manipulating the aspect ratio of the MWCNTs (length to diameter ratio) can be fascinating to probe into, and may lead to discovering possible applications that may rise from it, as mentioned in a prior section of this thesis. Literature has treated CNT forests using glow for functionalization, exposing the entire CNT surface to the glow. This thesis looks into using glow discharge generated using an electrode, to treat CNT forests locally. Using direct current (DC) glow discharge, thinning of the CNTs has been observed, and prolonged treatment led to generation of thick coalesced structures that have not been reported before. In addition, RF discharge has been proposed and designed as an extension of this work.    12  1.3.1 Micro electro discharge machining overview  ?EDM is a micro-machining technique applicable to any electrically conducting material including metals, alloys, and even highly doped semiconductors. The setup consists of a machining electrode and a sample bearing stage that possesses multiple degrees of freedom and offers precise maneuverability during machining, conducted by electrical discharges generated between the electrode and the sample being machined. Being a non-contact form of machining, it is possible to machine fragile/soft materials since there are no stresses being applied. Using computer numeric control  (CNC) programs to control the machining stage and electrode with tens of nanometer range accuracy, extremely precise micro structures are attainable. Incorporating ?EDM to the process of fabricating MEMS devices enabled researchers to exploit a broad range of materials incompatible with conventional MEMS fabrication process.  Material removal from sample to be machined (referred to as the workpiece) is conducted by thermo-mechanical impacts created by pulses of miniaturized discharges, generated between the machining electrode and the workpiece immersed in a dielectric. The discharge occurs as a result of the dielectric breakdown owing to the application of an electric field. The discharge heat causes the material to melt and evaporate from the site. The discharge energy is set to be minimal to achieve the smallest mass being removed ensuring a fine machined surface. ?EDM tends to be popular in industries that are associated with applications requiring difficult to generate features e.g. nozzles, micro-mechatronic actuator parts, fabrication of micro-tools etc. [46], [47].       13  1.3.2 Types of ?EDM  Current industrial ?EDM technology can be categorized into five types [48]: I. Die sinking: An electrode with micro-features is employed to produce its mirror image in the workpiece. II. ?EDM milling: Micro electrodes are employed to produce 3D cavities by adopting movement strategies similar to conventional milling. III. ?EDM drilling: Micro electrodes are used to drill micro holes in the workpiece IV. Micro wire EDM: A wire of diameter as small as 20 ?m is used to cut through a conductive workpiece.  V. Micro wire electro discharge grinding: The microwire is used to grind structures in the workpiece using the EDM mechanism.  Table 1.2 below shows a comparison of the specifications of the different machining mechanisms [47].    Table 1.2: Overview of uEDM processes  ?EDM Variant Geometric Complexity Minimum Feature Size Maximum Aspect Ratio Surface Quality Ra (?m) Drilling 2D 5 ?m ~ 25 0.05 ? 0.3 Die-Sinking 3D ~ 20 ?m ~ 15 0.05 ? 0.3 Milling 3D ~ 20 ?m ~ 10 0.5 ? 1 WEDM 2 ? D ~ 30 ?m ~ 100 0.1 ? 0.2 WEDG Axi-sym. 3 ?m 30 0.8   14  1.3.3 Comparative advantages of ?EDM    Compared to traditional Si micromachining technologies, EDM has substantial advantages [49]. I. EDM required low installation cost compared to lithographic techniques II. It requires little job overhead (e.g. designing masks etc.) III. It is flexible and ideal for prototyping or small batches of products IV. It can easily machine complex 3D structures and shapes that prove difficult for etching.  1.3.4 Comparative compatibility of ?EDM   Compatibility of machining technique with substrates is an important aspect of the technology. The wider the compatibility the better the technique and the greater the range of applications it has. In terms of versatility ?EDM it has an advantage over its competitors since it can be used to machine any substrate that is conductive in nature e.g. metals, semiconductors, alloys, and even certain ceramics. Table 1.3 below provides an overview of the compatible substrates for each technique [49].       15           1.3.5 Development of ?EDM   The basis ?EDM is believed to be the discovery of an English chemist Joseph Priestly [50]. In 1770, Joseph Priestly discovered that electrical discharges had erosive effects.  Later in 1943, the Lazerenko EDM system was developed by B.R. Lazarenko and N.I. Lazarenko at the Moscow University using a resistance-capacitance type power supply to generate these sparks. Their system enabled machining of difficult to machine materials in a controlled manner by vaporization of the material from the workpiece [51]. This resistor-capacitor (RC) relaxation type pulse generator was later on extensively used. Parallel to this, developments in America occurred that were the basis of vacuum tube EDM and electronic-circuit servo system providing automatic proper electrode-to-workpiece spacing for discharges without the electrode coming in Table 1.3: Overview of compatible substrates for different machining techniques Micro-Machining Technology Feasible Materials LIGA Metals, Polymers, Ceramic Materials Etching Metals, Semiconductors Excimer-LASER Metals, Polymers, Ceramic Materials Micro-Milling Metals, Polymers Diamond cutting Non-Ferro Metals, Polymers Micro-Stereolithography Polymers ?EDM Metals, Semiconductors, Ceramics   16  contact with the workpiece [52]. In 1980?s introduction of Computer Numeric Control (CNC) brought about major improvements in the machining process. It enabled the machining process, beginning from placing the workpiece, till the complete machining of the final structure, to be completely automated [53]. Over the years the state of the art of the process has progressed to solid state power supplies and modern six axis numeric controlled machining. Such growing merits of ?EDM rendered it to be a highly sought after manufacturing process in the industry, with its base of applications growing substantially and protruded into micro fabrication technology.  1.3.6 Principle of operation   The removal process is essentially converting electrical energy into thermal energy via discrete electrical discharges between the electrode and the workpiece immersed in a dielectric (usually a liquid). The discharges occur at frequencies up to 10s of Mhz (i.e. a period of 100 ns).     Temperatures at the site of machining can reach up to 1000s of K [54]. Pulse energy can be of the order of 0.01 ?J ~ 10 ?J. The high temperature melts and evaporates the workpiece, and residues are flushed away by a pressure difference created during the intervals of the discharges. A detailed explanation of the process is illustrated in Figure 1.3. Upon the application of a growing electric field between the electrode (the cathode) and the workpiece (the anode), the dielectric breaks down and electrons are accelerated towards the workpiece. An arc discharge is sustained by electron emission from the cathode by thermionic and field emissions, and secondary emission. Since both temperature and electrical field are strong, the emission process is strongly dependent on both variables. Thus the plasma is highly ionized resulting in high current densities with comparatively low discharge voltage.    17  Secondary emissions take place when the fast moving electrons hit the dielectric molecules causing ionization and generate further electrons resulting in rapid expansion of the bubble. An avalanche process takes place with a surge of electrons bombarding the workpiece, causing a thermal effect that leads to melting of the workpiece and evaporation. The evaporated atoms and molecules are solidified and condensed to form debris particles or gases generated form bubbles. Also, the discharge causes a pressure wave that removes any residue left at the point of discharge. There is also wear at the electrode surface caused by the ions striking the electrode, but the wear is minimal compared to the material removal at the workpiece since the electron propagation is much faster than the ions. Once the discharge is complete, the plasma channel breaks down and dielectric comes rushing in, flushing out any remaining fragments at the site. The process repeats to obtain further machining [54].  Figure 1.3: Principle of operation [54]   18    Literature suggests there are three distinct phases of the discharge process [55], preparation for ignition, discharge, and interval between discharges. In the preparation phase the electric field is created with the application of gap voltage. When the strength of the electrical field becomes high enough to break down the insulating properties of the dielectric fluid, the site is prepared for a discharge. In the discharge phase, when the resistivity of the fluid is lowest, a spark propagates through the ionized flux tube and strikes the workpiece. At this point there is a drop in voltage as a current is generated. The electrical energy is converted to thermal energy and the resulting spark vaporizes the surface of the workpiece, and the dielectric fluid, encases the spark in a sheath of gasses. At the interval phase between discharges, when the current is switched off, the heat source is eliminated and the sheath of vapor around the spark implodes. Its collapse creates a void or vacuum and draws in fresh dielectric fluid to flush away debris and cool the area.   1.3.7 ?EDM components   The components of a ?EDM system consists of a Computer Numeric Control system to control the X-Y-Z axes with a feedback mechanism, a pulse generator, a tank with the dielectric, the electrode and a X-Y stage standing the workpiece.  The components are shown in Figure 1.4 [56] and discussed in detail in the following section.   19   Figure 1.4: Schematic of the ?EDM system [56]  a) Computer numeric control system: This is the intelligence that controls the machining process. It has the ability to precisely control the maneuvering of the electrode and the X-Y stage with a resolution of around 100s of nm hence machining excellent micro structures. The system can be programmed to fully automate the machining process beginning with placing the workpiece at the stage, till the end structures are created. The NC also controls the pulse generation, and a feedback from the system enables the NC to maintain an optimal discharge gap to prevent short circuits and damage the workpiece by coming in contact with it. If the system detects a short, the NC immediately pulls the electrode in the positive z direction till it reaches a safe distance, and then resumes a slow decent to continue machining.  b) Pulse generator: There are two major types: I) Resistor-capacitor relaxation type: This was developed in early years and is illustrated in Figure 1.5 below [56].   20   Figure 1.5: Resistor-capacitor pulse generator [57]   The capacitor charges up, and as the electrode approaches the workpiece, upon reaching a suitable electric field (depending on the dielectric) to break down the dielectric, a spark passes through the dielectric discharging the capacitor. The cycle continues for successive discharges and subsequent machining of the workpiece.  To minimize the energy of the discharge for fine machining the capacitance needs to be small. The energy of discharge is        , where C is the capacitance, and V is the source voltage. In a typical ?EDM setup the system capacitance is accompanied by parasitic capacitance Cp, hence the true energy of discharge is             . The parasitic capacitance exists in (i) the circuit between the electric feeders, (ii) between the electrode holder and work table, (iii) between the electrode and workpiece [57]: With careful design of the equipment (incorporating measures such as shorter feed cables, etc) the parasitic capacitance can be reduced to 10 ? 12 pF [58]. In certain cases machining is conducted setting the system capacitance to zero and using solely the parasitic capacitance.    21  Another way of minimizing the discharge energy is by lowering the applied voltage (typically voltages of 60 V ? 100 V are used). This approach risks the incidence of frequent short circuits and in turn possible damage to the work piece. This is because if the voltage is lower, the gap distance requires to be reduced to maintain the necessary electric field for the discharge to occur. With shorter gap distance the probability of the electrode coming in contact with the surface of the workpiece increases leading to the projected increased number of shorts. In addition to damaging the workpiece it also slows down the process since every time a short occurs, the electrode is retraced rapidly, and resumes decent after attaining a safe height. In spite of being an early process, the relaxation type pulse generators are still used because of its ability to generate short but large current pulses, and low parasitic capacitance compared to the later pulse generators.  In relaxation type generators there is always a presence of leak current [57] and it hinders the charging of the capacitor resulting in interruption of pulse discharge [57].  The leak current generates joule heat at discharging area that may be an obstacle to the recovery of dielectric strength of the gap resulting in machining instability [57]. However, the relaxation type pulse generators are used in finishing and micro-machining because it is difficult to obtain significantly short pulse duration with constant pulse energy using the transistor type pulse generator.  II) Transistor type: With the development of transistors with high current carrying capability, transistor type pulse generators started to gain popularity. The setup is shown in Figure 1.6 below [58]. Discharge pulses are generated by switching the transistors on and off periodically with a duration of 1000 ns and a period of 100 ns. However it is   22  difficult to make sure that electrical breakdown occurs at that exact instance and that the discharge delay is shorter than the pulse duration. This leads to a lower discharge frequency compared to the RC generator and poor uniformity.   Figure 1.6: Transistor type pulse generator [58]  c) X-Y stage: This is a platform that positions the tank bearing the workpiece immersed in the dielectric fluid. The stage is driven by computer numeric programs. Moving the stage, moves the workpiece relative to the electrode which has only one degree of freedom (in the Z-axis), and achieve the desired machining of the structure.   d) Z-stage: The Z-stage control is most crucial. The servo feed control of the electrode in the z-axis keeps the gap at a constant optimum. If the gap becomes too small, the gap voltage reduces and the decent slows down, or in an extreme case the electrode may be retracted. This prevents the electrode from coming in contact with the workpiece and incurring possible damage. This is conducted by the feedback provided by the gap condition analyzer and the intelligence in the numeric control centre.  e) Dielectric: Literature [59] reported that most of the metal removal occurs due to boiling of the superheated molten mass in the crater at the end of discharge. Results [56] indicate   23  liquid dielectric is not necessary for metal removal and higher material removal rate in O2 have been reported [60]. In air, however, most of debris particles were reattached to the workpiece surface. This is because the melted debris particles move in air and do not solidify until they hit the electrode surface. In the case of liquid, in contrast, the debris particles proceed straight through the bubble and penetrate the bubble wall, and as a result decelerate. They then solidify into a spherical shape under the influence of surface tension. This confirms that the dielectric liquid is important for the cooling and flushing of debris particles but not for material removal. It is shown in [56] that the material removal rate with a dielectric liquid poured into the gap in air is higher than that with the gap submerged in a tub of the dielectric liquid because the ability of flushing debris in former case is higher. Dielectric strength of liquid (>10 MV/m) is higher than gas (<4 MV/m) which leads to a higher electric field necessary for the dielectric breakdown to occur, resulting in a smaller gap distance. As a consequence the stray capacitance increases which limits the minimum energy of the discharges. Also, thermal conductivity and heat capacity affects solidification and cooling of electrode and workpiece, and they have proved to be higher in liquids.  Combination of gas and liquid, have also been studied with water mists in addition to air, N2 and Ar gas used as dielectric medium [56]. The liquid phase dispersed in the gas changes the electric field for discharges, enabling larger gap distances. This results in stable machining under lower energy. The debris reattachments can also be avoided as the liquid in the dielectric fluid flushes them away.     24  1.4 Outline of thesis   This thesis is divided into two parts. The first part aims to develop a batch mode fabrication technology to machine CNT forests and create free form 3D structures using ?EDM. The fabrication process of a batch mode electrode array is developed, that can replace the single electrode being used, to enhance the throughput yet retaining the quality of structures machined. UV LIGA was employed to render the process economically feasible and compatible with current electronics fabrication facilities compared to X-Ray LIGA used to develop similar arrays of electrodes in literature.. The machining process and the machined structures are then characterized. The process parameters have proved to be consistent with the single electrode case and the morphology of the machined structures has also been observed to have an acceptable quality.   The second part of the thesis looks into sustaining a local glow discharge on the CNT forest surface using the ?EDM electrode, in an inert ambience passing Ar in the chamber. DC glow was observed to have degrading effects on the CNT forest surface localized to an area of a few 1000 ?m2. The diameters of the CNTs seem to be decreased in SEM images, which is consistent with literature [21]. With prolonged exposure and higher energy discharge, coalescence of the CNTs occurred; bundles and later amorphous looking spherical structures began to manifest. Such treatment may be able to change characteristics of CNTs such as conductivity, tensile strength etc. locally, allowing the possible development of a wide range of novel MEMS devices. In addition to Ar plasma, this work may be extended to be used to functionalize CNT surfaces locally with N2 and H2 bearing functional groups among others. Localized RF glow discharge is also designed and proposed. RF glow discharge is normally used in cases when the cathode is non-conducting, and a layer of electrons get trapped on it to form   25  the virtual cathode. The RF process has been observed on stainless steel and has not yet been optimized for CNTs, but it is postulated to have interesting outcomes resulting from treating the CNT surface with it.                      26  2. Chapter 2 Batch mode ?EDM   This part of the thesis aims to developing a parallel process that is potentially compatible with the current electronics fabrication facilities, of fabricating a batch mode ?EDM electrode array that can enhance throughput of machining CNT forests with appreciable machined surface quality, followed by characterization of both the process and resulting structures. This chapter discusses a microfabrication process developed employing UV LIGA, using facilities available in any typical cleanroom. The electrodes created are high aspect ratio and the machined structures bear a surface morphology of acceptable quality.       2.1 Process Flow  The fabrication process is based on conventional UV photolithography using a chemically amplified negative photoresist called KMPR 1050. The substrate preparation involves cleaning thoroughly using piranha etch, followed by oxide removal by HF dip. The clean substrate is then dried and a seed layer of Cu is deposited by electron beam physical vapor deposition. Once prepared, the photoresist is spun on the substrate, and then heated. The substrate is then exposed using a conventional UV source under a mylar mask possessing the required patterns. The exposed substrate is then soft baked and then developed using SU8 developer. It is further treated by plasma etching to remove any exposed remnants that may hinder the electroplating process. The substrate is then electroplated, and the photoresist is    27  removed using Remover PG. The structures attained a height of 85 ?m with acceptable aspect ratios. The electrode chips are obtained by cutting the substrate using a diamond disc saw. Once the chips are obtained, a thin layer of Cu is deposited on the back and sides of the chip to create a better conducting path from the bottom of the chip to electrodes on top using the electron beam physical vapor deposition method. A number of different patterns of molds have been fabricated, and then electroplated using Cu to obtain the electrode structures. Cu has been chosen due to its thermal and electrical properties, resistance to wear, and ease of electroplating. A fabrication flow diagram in Figure 2.1 summarizes the process.  The machining is conducted by attaching the die to an electrode holder by System 3R, set on a bracket to be connected to the discharge circuit. The CNT Forest is set on the stage and machining is conducted with air as dielectric. Both sink in and scanning processes are characterized. The process parameters that work suitably are found by trial and error. The current pulses are measured using a current probe. The machined structures are analysed by atomic force microscopy (AFM) and scanning electron microscopy (SEM) imaging, and the topography of the machined surface is observed to be rather fine. An energy dispersive X-Ray spectroscopy (EDX) analysis shows minimal wear of the electrodes upon machining.        28   Figure 2.1: Fabrication process flow diagram  2.2.1 Choice of photoresist   One key factor involved in the development of the arrayed electrode device is the photoresist suitable for the targeted microfabricaiton process. An epoxy-based photoresist called SU-8 (MicroChem Corp., MA, USA) has been widely used in UV LIGA, as it can be spin coated to form a very thick layer that allows for high-aspect-ratio photo-patterning owing to its mechanical properties and low optical absorption in the near UV range leading to a uniform exposure condition across the thickness [61]?[63]; patterning thickness of over 1000 ?m   29  (established by repeated spin coatings) with an aspect ratio of 40 was reported [61]. Regardless of these appealing features, the use of SU-8 poses an inherent difficulty in the removal of the cross-linked epoxy material, a requirement for the resist to be used as the mold for electroplating [14], [64] to provide free-standing electroplated structures on the substrate. Poly(methyl methacrylate), or PMMA, is another photoresist traditionally used for thick patterning but requires UV exposure dose 10 times that of SU-8; this option is more suited for synchrotron X-ray lithography as used in typical LIGA processes. KMPR is a new chemically amplified photoresist jointly developed by MicroChem Corp. (MA, USA) and Nippon Kayaku Ltd. (Tokyo, Japan). It is a negative tone, i-line (365 nm) photoresist based on epoxy that is high contrast and can be coated for >100 ?m thickness by a single spin coating, enabling thick, high-aspect-ratio photo-patterning. A UV-exposed layer of this material can be developed in a conventional aqueous alkaline developer (such as tetramethylammonium hydroxide), leaving the cross-linked material that provides high chemical resistance suitable for the subsequent electroplating process. One of the significant features of this photoresist is that the cross-linked material can be readily stripped using a commercial resist stripper (e.g., Remover PG). All these characteristics of KMRP make it ideal for the UV-LIGA process, including the microfabrication of ?EDM electrode array.   2.2.2 Substrate preparation   The fabricated chip needs to be highly conductive for proper discharges to occur from the electrodes for optimal machining of the CNT Forests. Under this consideration the Si wafer substrate was chosen to be a highly doped (100) n-type wafer with ? = 0.001-0.005 ?-cm. It is   30  extremely critical to make sure that the wafer is clean before initiating any process. To ensure cleanliness the substrate (wafer), is cleaned using piranha etch, followed by a HF dip to remove the native oxide. Piranha solution is a mixture of concentrated sulphuric acid (H2SO4) and hydrogen-peroxide (H2O2), and used for removing organic residues on the semiconductor wafer. Extreme caution must be taken when preparing, handling and disposing it. Desired amount of concentrated H2O2 is poured in a clean glass beaker and then 5 times that volume of H2SO4 is added very slowly. Since the mixing is an exothermic reaction, a large amount heat will be released. The solution is stirred using a glass rod. It is then put on a hotplate at 100oC. The substrate is put into the solution when it starts to bubble, and is left for 15 minutes. The substrate is then pulled out using Teflon tweezers and rinsed with DI water. Dilute hydrofluoric acid (HF) is then used to remove native Si dioxide from wafers. Since it acts quickly, one needs to only expose the wafer for a short time (?dip?). The substrate is soaked for 1-2 minutes in 2% HF solution. The substrate is then removed and rinsed in running DI water. Finally the substrate is heated on a hotplate at 100 oC for 1-2 mins, to ensure that it is completely dry.   2.2.3 Deposition of Cu seed layer   A seed layer of Cu is to be deposited on the substrate to enable growth of structures by electroplating. A thin layer is deposited by the electron beam physical vapour deposition process. The process involves the insertion of the substrate facing a crucible containing the metal to be deposited inside a vacuum chamber. The metal is heated up by a beam of electrons. Upon evaporation, the metal reaches the surface of the substrate, comes in contact, and gets deposited. A slow rate of deposition ensures the formation of a uniform layer. 20 nm of Cr is deposited to   31  enhance the adhesion of Cu, followed by the deposition of 200 nm of Cu. The seed layer is illustrated in Figure 2.1 (a).  2.2.4 Photolithography process   The photoresist KMPR 1050 is spin coated on the substrate at speeds deemed optimal for the process. This process is preceded by spinning a layer of HMDS for better adhesion of the photoresist on the substrate. Details of the spin parameters are shown in Table 2.1. A thickness of 90 ?m is achieved from the mentioned spin speeds. The substrate is then soft baked on a hot plate at 100 oC, for 23 minutes. The soft bake step is necessary to: (i) drive away the solvent from the spun-on resist, (ii) improve the adhesion of the resist to the wafer and (iii) anneal the shear stresses introduced during the spin-coating. Immediately after soft bake, the surface of the photoresist layer may be in a semi-liquid state that may adhere to the mask. Hence upon removal from the hot plate, the substrate is allowed to cool for a couple of minutes for the baked photoresist to set before being exposed. The substrate with the PR applied is shown in Figure 2.1 (b). The photoresist is exposed using Canon PLA-501F double-side 100 mm mask aligner. It is powered by a 250 W Hg lamp. The exposure dose is calibrated by trial and error and adequate results are obtained for an exposure time of 150 s. The Cu seed layer dictates the exposure time to be longer than in the case of the conventional Si substrate. A mylar mask, set on a glass plate was used for exposure. The mask was printed at high resolution to ensure fine features. Exposure was conducted in contact mode to prevent diffraction, which may lead to undesired exposure under the pattern.    32  The post exposure bake (PEB) step involves baking the exposed photoresist on a hot plate at 100 oC for 4 mins. This step is necessary for completion of the cross-linking process. The wafer is then developed in SU-8 developer solution. It is hand agitated for 4 mins, with subsequent use of a sonic bath for 1 min for the developer solution to propagate properly into the deep structures and conduct proper dissolution of the photoresist. The developed photoresist is then rinsed with IPA and blow dried with N2. To ensure appropriate removal of photoresist and other possible contaminants, Reactive Ion Etching using O2 plasma is conducted using Trion PECVD device.       It is imperative to ensure complete removal of the developed photoresist since remnant portions of the photoresist would substantially hinder the electroplating process.  The developed pattern is shown in Figure 2.1 (c).  2.2.5 Electroplating   The substrate with the patterned mold is then prepared for electroplating. It is cleaned with RONACLEAN? PC 960 (Dow Chemical Company MI, USA) to get rid of any oxides that Spin Speeds of HMDS and KMPR 1050 Species Step Acceleration(rpm/s) Speed(rpm) Time (s) HMDS I 330 3000 20 KMPR I 110 500 15 KMPR II 330 1000 35 Table 2.1: Spin speeds of HMDS and KMPR 1050   33  may have been formed, followed by a wash with 10% H2SO4. It is then connected to a constant current supply and submerged in a bath of CuSO4 solution. A pump pushes the solution into the electroplating chamber, and the solution then flows back into the reservoir. A mechanical arm is used to agitate the solution constantly, enhancing the flow of ions well into the depths of the high aspect ratio molds. The current is ramped at a rate of 3 mA/cm2 every 30 s until the final rate of 32 mA/cm2 is reached. It is then held constant for 90 minutes assuming a rate of deposition of   0.75 um/min. A slow and steady ramp process will lead to the formation of a dense and uniform structure necessary for a sturdy tall electrode. Upon completion the electroplated substrate is rinsed with DI water, and dried using compressed air. The electroplated substrate is shown in Figure 2.1 (d).  2.2.6 Removal of photoresist  The KMPR mold is removed in Remover PG (Microchem Corp, MA, USA). Remover PG is a NMP based solvent stripper. It may also be used as a lift-off solvent. The operating temperature of the stripper is 80 oC. Mechanical/ultrasonic agitation enhances the physical transport of swollen resist away from the substrate. The wafer is immersed for a couple of minutes depending on its size. The KMPR mold would expand, lose adhesion to the substrate and fall off. The transformation may be visualized as the mold turns opaque from transparent. The removed KMPR mold may be in a big piece or chunks and would tend to be very soft. It is essentially the ease of removal of the cross-linked KMPR that rendered it to be a more suited option for this application as discussed in previous sections. The substrate is then rinsed with   34  IPA followed by DI water. It is then blow dried using a N2 gun. Images showing the final electrodes are shown in Figure 2.2.                          Figure 2.2:  (a) Array of electrodes on a chip (b) SEM images of array   The electrodes possess vertical side walls and flat top surfaces that are necessary for proper discharge to occur and conduct decent machining. The final electrodes in the process flow diagram are shown in Figure 2.1 (e).  2.2.7 Die formation   The substrate is then cut into dies using a diamond disc saw. The dies are then coated with a thin layer of Cu (200 nm), by the electron beam physical vapor deposition method, by setting the die at an angle and depositing Cu on the backside and the sides of the die. Although   35  the substrate is of high conductivity, this step ensures the least path resistance incurred by the current. The results are summarized in Table 2.2 which shows that deposition of the Cu layer brings about a substantial reduction of the path resistance. This enables low voltage operation of the process.         2.2.8 Growth of CNT forest  The CNT forests used in this thesis are synthesized using an atmospheric pressure thermal CVD reactor tube built in-house by Dr. Alireza Nojeh?s group. The reactor tube consists of two heating zones. The first zone is used to preheat the gases that flow within the quartz tube up to 850 oC by a tube oven. The sample is placed on top of a resistive heater (highly doped Si wafer) in the second heating zone. When the preheated gas reaches the heated sample the growth process commences. The oven is turned on after loading the sample with Ar flowing inside at 338 ccm until the temperature reaches 850 oC. A current is then passed through the resistive heater to heat it up to 800 oC for 2 minutes. The flow rate of Ar is then reduced to 137 ccm and H2 is flowed at 260 ccm for 5 minutes. Then, Ar flow rate is reduced again to 100 ccm, H2 is Table 2.2: Path resistance with and without deposited layer Resistance of path connecting tip of electrode to base Method Resistance (?) Without post deposited Cu  100 With post deposited Cu  16   36  kept at 260 ccm, and Ethylene (C2H4) is flowed at 110 ccm for the growth period. It takes around an hour to grow a few hundreds of micron height of vertically aligned multi-walled CNTs. Once the growth process is completed the flow of C2H4 and H2 is turned off, along with the current and the oven, but maintaining the Ar flow until the system cools down.   2.3 Experimental setup   The schematic in Figure 2.3 (a) illustrates the setting used for micropatterning experiments. The set-up is based on a commercial ?EDM system (EM203, Smaltec International Inc., IL, USA) that employs a relaxation-type (resistor-capacitor, or RC) pulse generator powered by a variable DC voltage source. In this type of pulse generation circuit, the discharge energy of a single pulse that determines a unit removal in the process can be expressed as       , where C is the total capacitance that exists in the RC pulse generation circuit, and V is the applied voltage. The numerical ranges of theses circuit parameters used in the experiments are also shown in the figure. The system has servo-controlled 3-axis stages with a 100 nm positioning resolution, comprised of the XY stage with the metallic holder on which the workpiece, a CNT forest on the substrate, is held and the Z stage to position the lithographically fabricated electrode array vertically. The Z stage is configured to have a planar sample holder made of ceramics, on which the electrode substrate is fixed and electrically coupled with the discharge circuit using conductive adhesive tape as depicted in Figure 2.3 (a). The image in Figure 2.3 (b) shows the actual set-up of the system. The current pulses generated are monitored using an inductive current probe (CT-1, Tektronix Inc., OR, USA) coupled with the discharge   37  circuit and connected to an oscilloscope; readouts of the current signals are obtained via a GPIB interface connected to a computer (Figure 2.3 (a)).  Dry ?EDM processes with this setting are performed in air, in the die-sinking mode by feeding the electrode array using the Z stage into the CNT forest, partly with horizontal scanning of the forest using the XY stage with respect to the array for lateral removal of the material using the array. The electrode arrays are fabricated on the Si substrate via Cu electroplating to attain a height of up to 85 ?m so that deep patterning for depths of 70 ?m or greater can be tested. These electrodes are defined as the cathode whereas the CNT forest is connected as the anode in the circuit (as shown in Figure 2.3 (a)).   Figure 2.3: (a) Schematic of experimental setup (b) image of experimental setup  2.4 Characterization of the process   The feasibility and characteristics of planar ?EDM of CNT forests was evaluated using the fabricated electrode arrays with different discharge conditions in air. The patterning test was   38  first implemented in the die-sinking mode, where the forest sample was patterned by feeding the array into it vertically using the Z stage. In this test, a 10X10 array of the electrodes shown in  Figure 2.2 (a) and (b) was used to machine the forest for a depth of 70 ?m. The machining tolerance and surface quality of the patterned structures are two key characteristics of a ?EDM process, and the conditions that maximize these characteristics are of interest in the targeted planar process. In light of this, the machining test was initiated from the electrical condition (20 V, 10 pF) that defines the least discharge energy necessary to perform conventional ?EDM of CNTs [65]; however, this condition resulted in the difficulty in establishing proper discharge pulses and thus in the removal of the CNTs. This outcome could be affected by two possible factors: One is relatively high overall parasitic resistance (which includes the contact resistances at the forest sample and at the electrode device) present in the discharge circuit. The other is the detection of short circuits due to local arcing that could be exacerbated by the use of planar, non-rotating electrode (in contrast, the conventional case of a rotating cylindrical electrode easily breaks up an arcing condition). The voltage was then increased gradually while maintaining the capacitance value. This procedure revealed that 40 V was the lowest voltage level that led to stable generation of discharge pulses between the Cu array and the CNT forest, enabling parallel patterning of the forest. This electrical condition defines the theoretical discharge energy to be 8 nJ. A sample result from this patterning process is displayed in Figure 2.4 (a). A comparison between the dimensions of the patterned structures shown in the SEM images and known size of the electrode structures suggests that the tolerance of the process (i.e., a gap clearance between the sidewall surface of the patterned structure and that of the electrode) is approximately 4 ?m. This value is less than half the level reported in [66] for the same voltage and capacitance; the   39  tighter tolerance with this batch-mode implementation might be brought by the fact that the electrodes are stationary (except for Z-direction feeding) with no rotating motion. (Note that the conventional setting of single-tip electrodes with rotation may suffer an enlargement of the gap due to possible wobbling motions.) The discharge energy was further increased to observe its impact on the patterning quality and to compare the result with the lowest energy case. Figure 2.4 (b) shows the structures patterned with a 45X larger discharge energy (with V and C of 60 V and 200 pF, respectively) in the same CNT-forest sample using the same electrode array. The machining tolerance under this condition was measured to be 6-10 ?m. Using the two electrical settings mentioned above, another parallel patterning that combined lateral scanning motions with die sinking was also tested using an array of cylindrical Cu electrodes with 140 ?m diameter; examples of the results are shown in Figure 2.5 (a) and (b). In all cases, the process was able to transfer the pattern of the electrodes (and that of scanning motion when it was combined) to the CNT forest. As can be seen from a comparison between the results in Figure 2.4 (a) and (b), however, it is evident that the former case resulted in sharper edges and a finer surface morphology on the patterned structures compared with the latter case. This tendency is consistent with prior results reported [66], i.e., the use of smaller possible energies (or voltages) contributes to suppressing the amount of discharge gap as well as that of unit removal of CNTs by a single pulse, leading to higher precision, as indicated in the difference in the machining tolerances noted above, and smoother surfaces. Comparing the SEM images, the latter case with the large energy apparently caused bundling of the CNTs on the patterned surfaces, leading to rougher surfaces. These visual observations were verified using AFM. Figure 2.6 shows the AFM images for the bottom surfaces of the patterns created in the forest using the two energy conditions and compares two 3D surface roughness parameters, average (Sa) and peak-valley   40  height (Sy), obtained from the acquired AFM data. These comparisons indicate a degradation of surface smoothness by the use of the higher energy condition, which led to increases in Sa and Sy by 75% and 49%, respectively.   Figure 2.4: Patterns of CNT forest created in the sink mode using (a) 40 V and 10 pF and (b) 60 V and 200 pF  The elements on the ?EDMed surfaces were analyzed using EDX at 20 keV beam voltage. Table 2.3 summarizes the detected levels of the major elements involved in the patterning process and the materials. The result with an original (non-patterned) surface of the same forest sample is also shown in the table for comparison. It is observed from both the K ratio and concentration values that although Cu on the patterned area was detected, its amount is considerably small, suggesting that the wear of the electrode due to its use for the process and the   41  contamination on the patterned surfaces with the electrode material are negligible. This preferable condition is similar to the case of W electrodes [66], [67]. The results also indicate that both the amount of O and its change due to the process are insignificant. The signals of Si are presumed to originate in the Si substrate, by detecting X-ray from it caused by impinging electrons that penetrated through the CNT forest (the detection of Si was also reported in [66]). Certain variations in the levels of C and Si between the original and patterned surfaces can also be seen in Table 2.3   Figure 2.5: Patterns of CNT forest created in the sink and scan mode using (a) 40 V and 10 pF and (b) 60 V and 200 pF    42   Figure 2.6: AFM images of patterned regions uisng (a) 40 V and 20 pF and (b) 60 V and 200 pF. The measurement was performed with Easyscan 2 (Nanosurf AG, Liestal, Switzerland)  In particular, it shows that the level of C decreased whereas that of Si increased after the process. These behaviors seem to be logical, as more the amount of C, the less the signal reflected from the Si substrate. One cause of these variations may be related to the thickness of the forest layer rather than the patterning process itself. The following factors should be noted in terms of the thickness aspect. One is that there is relatively large non-uniformity in the original height of the CNTs on a CVD-grown forest sample. Furthermore, the thickness of the forest was reduced on the patterned area due to its removal process. These conditions could have led to a difference in the forest thickness and resultant signal level of C (and thus that of Si) between the two analyzed locations. Another cause may be associated with scattering of X-ray signals from CNTs, which could occur more on the patterned area due to the surface roughness of the area (Figure 2.6 (a)) induced by the patterning process.      43  Table 2.3: Results from EDX elemental analysis of patterned CNT forest  Element Patterned area Original surface K ratio Conc.(%) K ratio Conc.(%) C 0.586 88.88 0.658 91.74 O 0.008 8.21 0.006 6.08 Si 0.021 2.63 0.017 2.17 Cu 0.002 0.27 0 0   The average pulse current in the low energy case (40 V and 10 pF) was measured to be ~30 mA; sample current pulses captured in the process are shown in Figure 2.7. This current level is comparable to those reported in [7], [66], a few to several 10?s mA, for machining with similar electrical conditions (30 V, 10 pF) with the same polarity setting in air ambient. The pulse durations were recorded to be approximately 200 ns. This level of time is 5-6X larger than those shown in the above reports. Although the exact cause of this result is not clear, it is worth considering the following conditions. The lumped capacitance and parasitic inductance of the discharge circuit are the two parameters that affect the duration [68]. Both the present experiment and those discussed in [7], [66] used the same capacitance (of 10 pF), and comparing between the machining settings used in them, the levels of parasitic inductance are presumed to be similar. However, the planar electrode device used in the present study, which faces the top surface of the CNT forest in close proximity during the process (not the case in the past   44  implementations with single-tip electrodes), can cause a parasitic capacitance much larger than that involved in the single-tip electrode case. This additional capacitance associated with the planar electrode is coupled with the external capacitor of the RC circuit in parallel, potentially serving as a contributory factor in the longer duration of the pulses.  Figure 2.7: A sample waveform of discharge pulses captured in the batch-mode ?EDM of CNT forest  Other possible effects particularly relevant to the use of planar electrodes for dry ?EDM of CNT forests should be noted. In the conventional setting with conventional single-tip cylindrical electrodes, the machining area was exposed to fresh ambient air, which supplied a plenty of oxidation species necessary to drive local plasma etching of the nanotubes at the discharge spot as predicted in [7]. The circumstance is different in the batch-process setting, in which a planar electrode device maintains a micron-scale air gap with the forest sample; therefore, the access of ambient gas to the machining areas is physically limited, especially when the electrodes are fed deeper into a forest hence the gap spacing between the electrode substrate and the forest becomes narrower. Nevertheless, the CNT forest is essentially a porous bulk material, whose 95% of the volume is occupied by air [69], and this internal air is available for dry etching of the CNTs. These conditions related to the supply of etch species might affect the   45  process. Figure 2.8 shows the measured Z position of the electrode array tracked during the ?EDM process performed with a feed rate setting of 30 ?m/min. The ripples seen in the plot were caused by short-circuit events, i.e., when the system detects a short while feeding the electrodes, it automatically retracts the Z state until the short condition is cleared and resumes feeding again after that point. Consequently, the occurrence of shorting slows the process; the particular process shown in Figure 2.8 completed in 240 s, representing a real feed rate of 17.5 ?m/min. As displayed in the plot, shorting occurred more frequently when the electrode approached closer to the target depth (70 ?m). One possible cause behind this outcome may be associated with the non-uniformity of the forest material; as the electrode device was fed deeper, some large portion of the forest surface may have come closer to the substrate of the device, having a gap distance small enough to start generating discharge with the electrode substrate, leading to frequent short circuits due to a resultant enlargement of removal area. Another possibility could be related to the circumstance of the ambient gas noted above, i.e., the etch species may become depleted at the machining site as electrode feeding limits the supply of ambient air. Given the fact that ?EDM of CNT forest without O2 (or only with N2) is not feasible due to frequent arcing [70], the above situation could be relevant to the observed shorting behavior.   46   Figure 2.8: The position of electrode array on the Z axis tracked in real time during machining  The bundling of CNTs on the surfaces clearly observed with the high-energy processing could be another unique effect related to the use of planar electrodes. The serial ?EDM of CNT forests with rotating electrodes [7], [66], [67] did not appear to cause this type of bundling effect. The rotation of electrodes dissipates heat generated during the discharge process more effectively than the case without it, the case of the batch-mode process. Bundling of CNTs due to thermal effects was reported [24]. One possible measure to circumvent or mitigate the issue may be to combine vertical vibration of the sample with the machining process. The vibratory motion of ?EDM electrode/workpiece is known to enhance debris flushing and smooth processing in traditional wet ?EDM [71]?[73]. In case of dry ?EDM, the combinational use of vibration is expected to promote air flow, which may be effective in fulfilling the two potential needs discussed above, i.e., heat dissipation from and supply of etch species to the machining regions.   47  3. Chapter 3 Localized Glow Discharge   This chapter reports on developing a technique to sustain a stable glow discharge on CNT forest surface, locally, within a small selective area as small as a few 1000 ?m2 using the ?EDM electrode of diameter 100 ?m. In the past, glow discharge has been employed to add functional groups bearing N and H to CNTs to enhance certain characteristics such as polymer lubrication, increased conductivity, increasing water solubility etc. and may find useful applications in electronics, composites and material chemistry [17]?[20], but in literature the functionalization has occurred over the whole CNT surface. This part of the thesis aims to be able to conduct the process locally at desired areas with a high level of selectivity. This may have a number of potential benefits including patterning an electrical circuit based on the difference in electrical properties, on the CNT surface itself. An inert environment using Ar was used in this case. In addition to attaining localized glow discharge, this thesis probes into studying the possible degrading effects that glow discharge may have in terms of reduction of the diameter of the CNTs brought about by the rupturing of the outer walls of the MWCNTs, effects that are similar to cases in literature where current was passed through the CNTs [22] and also when CNTs were exposed to arc discharge [21]. Such thinning of the walls was observed in addition to a coalescence effect similar to one in literature brought about by annealing [24] where the CNTs started bundling up and forming a new type of structure. In the process a new type of thickened structure emerged that has been analyzed in the later sections, and it is postulated that such a treatment may bring about changes in characteristics like conductivity, tensile strength, optical characteristics etc. in selected regions or patterns to aid the development of a myriad of new   48  MEMS devices and electronics.  DC glow has been investigated on steel and CNTs. RF glow has been designed and tested on stainless steel but requires optimization for the CNTs.     3.1 Introduction   The fundamentals of glow discharge have been discussed in chapter 1. Literature shows a substantial amount of work on studying the effects of both DC and RF glow discharge on CNT under different pressure and gas environment. Literature proposes the use of glow discharge in functionalization of the CNTs, by adding useful N and H bearing functional groups. It has also been suggested that CNTs are degraded by low pressure DC and RF glow discharge in O plasma [74], [75]. Functionalization is normally preceded by degradation of the nanotube due to the high energy charged particles bombarding the nanotubes and in effect incur a form of etching breaking the C-C bonds and generating active sites for bonding of functional groups present in the plasma [18]. With larger energy, sputtering phenomenon can occur, leading to substantial breaking and destroying of the nanotubes [18]. Although there are a number of proposed mechanisms of this degradation the true process is yet unknown.   In all these cases in literature mentioned above, studies were conducted by exposing the entire CNT forest or substrate to the glow discharge. The novelty of this research is attaining a localized glow (within an area of 0.008 mm2) on the surface of a CNT forest (with an area of roughly 60 mm2 ~ 80 mm2) and treating only a desired region of the forest surface with the glow discharge. Such local treatment, bringing about changes in characteristic parameters such as electrical conductivity, thermal conductivity, optical characteristics etc. within a desired local   49  space or pattern can have significant number of applications. This research studies the effect of glow using a cylindrical W electrode of diameter 100  ?m as the cathode, and CNT forest as the anode in an inert ambience under atmospheric pressure created by having the reaction chamber pumped with Ar.   One focus of this part of the thesis was to observe changes in the CNT diameter associated with exposure to the glow. Since the ambience is inert, addition of any functional groups were not expected. The degradation aspect of the process is focused on. It is expected that there would be a change in the electrical conductivity and other properties following rupturing of the outer walls of MWCNTs resulting from the glow discharge. A similar previous work has been reported with the exception of having used arc discharge with a high discharge current density through the nanotubes [21]. The aim in that case was micromachining of the CNT forest to generate micro-structures. In the process, the CNTs in the treated regions resulted in having a reduced diameter shown in SEM images. Different theories for the mechanism of truncating the outer walls of the MWCNTs have been proposed. One proposed mechanism is the removal of the C as an oxide when the ambience contained O2 [22], but experiments in an inert atmosphere are also conducted where the rupturing have been hypothesized to have been due to the possibility that CNTs conduct balistically and the energy to break C bonds originates from highly localized dissipation at defect scattering sites [23].  While the literature suggest only reduction of the diameter, this research has observed that the CNTs bundle together being treated by glow discharge, and form coalesced structures that are denser, and can have interesting characteristic properties in terms of electrical, thermal and mechanical properties. A similar phenomenon where coalescence of CNTs have been observed were brought about by annealing [24].    50  It is postulated that such local manipulation of surface characteristics can give rise to a myriad of applications as discussed earlier in this section, and hence the possibility of the emergence of a whole new field of research using CNTs as the base material.   3.2 Experimental setup for local DC glow discharge   The SmalTec ?EDM system is used to generate the glow discharge on the CNT forest. An external DC power source is used in series with an Agilent 34405A Multimeter, used as an ammeter. A schematic of the setup is shown in Figure 3.1.  The CNT forest is positioned on the XY Stage of the ?EDM system, which is in turn connected to the positive of the DC power source. The electrode of the system, with a diameter of 300um or 100um depending on particular experiment, acts as the cathode. The discharge is confined inside a gas chamber to maintain a controlled environment. The ambience is manipulated by controlling the flow rates of Ar and O2 going into the mixing chamber.  Upon reaching the necessary electric field strength (a few MVm-1) for glow discharge to occur, attained by moving the electrode closer to the CNT forest surface with the voltage being applied, a bluish glow is observed.  Current readings range from 100 ?A to 400 ?A, with the application of voltages ranging from 200 V to 400 V,as attained from the ammeter readout. The gap distance between the forest and the electrode is usually a few tens of ?m, and this leads to an electric field in the range of   3 MVm-1 to 10 MVm-1, which is consistent with typical values mentioned in the introduction.   51   Figure 3.1: Schematic of DC glow discharge experimental setup    Figure 3.2: DC glow on CNT forest with a 100um diameter electrode 100 ?m   52  3.3 Experimental results     The effect of the glow discharge is characterized using different techniques and the results are analyzed to be able to draw conclusions regarding the newly formed thickened structures as a result of the discharge. SEM images of the CNTs are used to attain an understanding of the effect of the glow on the diameter of the CNTs, and analysis techniques such as EDX and Raman spectroscopy have been utilized to comment on the composition and the characteristics of the resulting thickened structures.  3.3.1 Effect of ambience on diameter  Experiments have been performed under different ambient conditions by varying the O2 and Ar content inside the gas chamber by controlling their flow rates. SEM images shown in Figure 3.3 are then used to analyze the post treatment effects on the CNT forest surface. The images are used to measure the diameter of the nanotubes and an average is calculated from a large pool of data. The nanotubes are selected manually in a random process and the diameter measured from the image. 30 samples are used in each case to obtain a substantial pool of data to derive conclusions from the exhibited trend. The results are illustrated in Figure 3.4.  .   53   Figure 3.3: SEM Images of CNT forest surface before and after glow discharge under varying ambient conditions (a) untreated CNT forest surface (b) glow discharge in Ar = 100%, O2 = 0%, (c) glow discharge in Ar = 97.1%, O2 = 2.9%, (d) glow discharge in Ar = 95.8%, O2 = 4.2%   It is observable from the chart in Figure 3.4, that there is a reduction in the diameter of the CNTs after being exposed to glow discharge. In this part of the study the current was maintained roughly constant at 100 ?A varying the O2 concentration. A complete Ar ambience brings about the maximum reduction in diameter, while incorporating O2 in the environment shows a lower reduction in diameter. A hypothesis to explain this phenomenon is that it is easier to obtain a stable glow discharge in complete Ar environment, compared to the presence of O2 in the gas chamber, and the difficulty increases with the increase in O2 content. In fact it was observed that introducing O2 into the chamber extinguishes a stable glow in the inert ambience, and necessitates an increase in the applied voltage to regenerate the glow. Hence, a more stable a b c d   54  glow can lead to a greater rapturing of the outer walls of the MWCNT, which in turn causes the observable reduction of diameter. Such a reduction in diameter is consistent with literature [21], [22], where the passage of current through the CNTs bring about the rupturing of the outer walls of the MWCNTs both in the presence of O2 and in an inert ambience. In the presence of O2 a possible mechanism is the removal of the C as an oxide [22], while in an inert atmosphere it is hypothesized to have been due to the possibility that CNTs conduct balistically and the energy to break C bonds originates from highly localized dissipation at defect scattering sites [23].  Figure 3.4: Analysis of CNT diameter from SEM images     00.20.40.60.81regular oxygen4.2%oxygen2.9%oxygen 0%Normalized Diameter   55  3.3.2 Effect of prolonged exposure to glow discharge  Further characterization of the process is performed by studying the effect of varying the current flowing through the nanotubes and sustaining the discharge for longer periods of time upto 60 seconds which gave rise to the coalesced thickened structures. This is brought about by changing the voltage applied ranging from 200V to 400V, resulting in a range of current from 100 ?A to 400 ?A, under a constant Ar rich environment. The discharge is not uniform as seen in Figure 3.2, over the total area under the electrode, resulting in regions of the CNT forest surface being exposed to different intensities of discharge. In Figure 3.5, the inner region is exposed to a higher intensity discharge, forming a granular morphology while the outer region being exposed less, shows the transitional stage, with tubular structures still being evident. It has been observed that for higher currents like 400 ?A, the granular structures begin to appear faster i.e. after 5 s, while being exposed long enough (e.g. 10 s) to a current as low as 100 ?A, it would lead to similar morphology which insinuates it is a function of energy rather than current. Figure 3.6 shows the transition. Compared to the untreated surface, Figure 3.6 (b), shows the CNTs being thinner being exposed to discharge similar to and consistent with literature [21], [22], and furthermore with a prolonged discharge and/or with a high current density discharge, it is suggested that the CNTs start to fuse together forming thicker bundles and granular structures coincident with literature having similar outcomes of annealing [24]        56   Figure 3.5: Exposure to different intensity of glow discharge  Varying the current and exposing the CNT surface to the discharge for 60 s gives rise to a greater granular morphology as illustrated by the SEM images in Figure 3.7. It is seen clearly evident that larger the current, the thicker and granular the structures. It is also apparent that with increasing current, the tubular shape of the forest begins to diminish, and a spherical appearance begins to manifest. The hypothesis of the CNTs being fused together is supported by Figure 3.7  and Figure 3.6 (c). Figure 3.8 shows boundary lines on the spherical structures that suggests smaller structures coming together to form the larger one i.e. a fusion process.   57   Figure 3.6: Effect of discharge on CNT at 100?A, (a) untreated surface (b) CNT diameter thinning out in addition to CNTs fusing together (c) CNTs fused together forming thicker bundles and granular structures (d) another case of CNTs fused together   Figure 3.7: Effect of varying current on CNT morphology (a) I = 100 ?A (b) I = 176 ?A (c) I = 240 ?A (a) I = 320 ?A (a) (b) (c) (d)   58      Figure 3.8: Outlines that suggest a fusion process  To obtain a quantitative understanding of the phenomenon, the diameters of the thickened structures are measured for different currents and the results are shown in Figure 3.9. It can be seen that there is a rise in the CNT diameter with increasing current, and a substantial increase is observed in case of the jump from 100 ?A to 176 ?A. The current density is varied within a range of roughly 10 ? 40 kAm-2 , which is similar to that in literature ranging from 13 kAm-2 ? 26 kAm-2 [76].       59     Figure 3.9: Effect of current  3.3.3 Elemental analysis of treated CNT surface  The treated regions are analysed using EDX. A Hitachi S3000N VP-SEM with EDX was used for the characterization.  EDX is an analytical technique for elemental analysis of a sample. It is based on the principle that each element has a unique atomic structure allowing a unique set of peaks in the X-ray spectrum. A high energy beam is incident on the sample which excites and removes an inner shell electron from the sample with a higher shell electron coming down, emitting energy in the form of an X-Ray photon. The number of specific energy of X-Ray photons provides the analysis of elemental composition of the sample. K Ratio is a good characteristic measure of the amount of an element present in the sample. It is the ratio of the 0.002.004.006.008.0010.0012.00Untreated 100?A 176?A 240?A 320?ANormalized Diameter   60  intensity of the relevant element in the test sample compared to the intensity of the same element in a standard sample.  The results are summarized in Figure 3.10 below. The first sample indexed as ?0 mA? in the chart is the reference CNT surface that has not been treated. It is observed that C content decreases with increase in the current flowing through the sample due to the discharge, accompanied by an increase in detection of Si. The source of the Si is the substrate on which the CNT forest is grown, and comes up on the detector due to the high energy of the incident beam penetrating through the forest. From the chart, K-Ratio values of C goes down by a roughly consistent 20 % with a 100 % increase in the discharge current, and a consistent increase in Si of 23 % (within a tolerance of 4 %) at the same time. A possible explanation of this result could be the fact that when the CNTs are bundling together, they are creating open spaces and easing the process of the incident beam to reach the substrate and detect more Si. Literature has a similar phenomenon where SWNTs were annealed with electron irradiation and observed coalescence of the nanotubes [24]. It was suggested that two adjacent nanotubes, containing random vacancies and dangling bonds lead to the establishment of a connection between the two tubes where the surface and atom reconstruction were initiated by the electronic irradiation. In the case of this thesis, possible mechanism of the bundling could be the energy of the discharge causing breaking of the bonds between the C atoms, (hence the rupturing of the walls) and then reforming an amorphous structure with the atoms (being close to each other and with simultaneous exposure to the energy from the discharge) which may be sufficient to reform the bonds between them but in different and random orientation. The crystallinity is lost in the process of bundling and reforming into what might be amorphous in structure, and the new bundled structures are expected to have different characteristics from the regular CNT. The   61  absence of W or any contaminants in the EDX results further reinforces the hypothesis that the bundled structures are C possibly in a different form. The EDX result confirms minimal wear of the electrode and as a logical sequence the retention of the elemental composition of the CNT forest to being mostly C and O2 possibly trapped within, or being bonded to the C via the dangling bonds.   Supporting evidence to prove the source of Si is from the substrate is found by conducting EDX on the sample by tilting it at an angle as shown in Figure 3.11. It is observed that the Si detected falls substantially and the C content increases rapidly when the substrate is tilted, where more of the surface is exposed and the substrate is moved further away from the incident beam. From this observation, it may be inferred that the Si is originating from the substrate.  Further characterization of the thickened structures has been conducted using Raman spectroscopy. When a laser beam impinges upon a sample it causes both Rayleigh and Raman scattering. Rayleigh corresponds to the light scattered at the frequency of the incident radiation whereas Raman is the one shifted in frequency. Some of the beam excites the vibration eigen-modes of the molecules involved. These vibrations, in turn, frequency modulate a small proportion of the incident photons resulting in an optical frequency shift. This process can be seen as the inelastic scattering of photons and Raman spectroscopy is based on the detection of these inelastically scattered photons. The Raman spectrum constitutes an intrinsic molecular fingerprint of the investigated sample, revealing detailed information on its chemical composition and structural conformation. Although Raman spectroscopy is a useful technique, it is strongly limited by the strong fluorescence background. Spectrally, this fluorescence occurs at   62  the same wavelength as the Raman signal and is often several orders of magnitude more intense than the weak chemical transitions probed by Raman spectroscopy [77].   Figure 3.10: EDX summary of DC glow processed CNT forest surface    63    Figure 3.11: EDX results with the CNT forest tilted at 45o  For this research Raman scattering was analysed from a treated region with the thickened structures and was compared with the regular untreated CNT forest surface. The results are summarized in fig below. Two peaks at roughly 1342 cm-1 and 1580 cm-1 are observed for the reference CNT which is consistent with literature [7]. The peak at 1580cm-1 (Ig) is due to the in plane C-C bond vibration, while the peak at 1342cm-1 (Id) is due to defects in the system. The ratio if Id/Ig provides information regarding the disorder in the system. The reference region had negligible/non-existent background noise, and hence applying the background correction algorithm resulted in little/no difference. Base-line correction was done by 2 degree polynomial fitting and subtraction from the raw data. The treated region (I = 200 ?A, t = 5 s) however shows massive background noise as seen in Figure 3.12. This is a prevalent issue in Raman   64  spectroscopy faced extensively. In biological analysis this is a major obstacle due to the presence of intrinsic fluorescence emission of the tissue [78]. Similar shifts as reported in this thesis of the baseline and enveloping of the Raman signal have been observed due to organic contaminants [79]. After implementing background correction, similar peaks begin to emerge at the same regions as the reference CNT forest which insinuates the existence of similar inelastic scattering as that of regular CNTs, but enveloped under a substantial background usually generated by the existence of fluorescent material. The ratio Id/Ig in both the cases appears to be approximately 1.62 showing the same level of disorder in the CNT structures.    Fluorescence is a process where an electronically excited state decays to a lower electronic state emitting a photon. The lower state is usually one of many vibrational levels in the electronic ground state. One possible conclusion regarding the thickened structures could be that it is itself the fluorescent entity, or it has contaminants that are fluorescent, but it is surely a transformation of the MWCNTs. The EDX results support the claim that there are nearly no contaminants, which suggests the combined results back the hypothesis that the thickened structures are derivatives of CNTs but lacking the organized graphite like molecular structure that has the ability to vibrate. One possibility could be that there is still a CNT core, surrounded by amorphous coalesced structures, or it could also be possible that the whole structure is coalesced with some retaining fragments of CNT that show up in the Raman after background correction in Figure 3.12 (d).  To further strengthen the claim that the thickened structures are derivatives of CNTs, the CNT forest was treated over different durations of time and the Raman results were analysed. Figure 3.13 shows the gradual decay of the distinct peak in the reference untreated CNT surface to a degraded spectrum for one exposed for 1 sec, to virtually two slight bumps in case of the   65  region exposed for 5 sec. After implementing the background correction algorithm, the characteristic peaks are again visible, which demonstrate the presence of CNTs.   Figure 3.12: Raman Spectroscopy results (a) reference CNT surface without background correction (b) reference CNT surface with background correction (c) treated CNT surface without background correction (d) treated CNT surface with background correction      66  An extensive analysis of this is necessary to be able to derive solid conclusions regarding the composition and structural characteristics of the thickened structures and is proposed to be a part of future work.    a   67    Figure 3.13: Raman spectrum showing gradual degradation of characteristic peaks (a) without background correction (b) with background correction   b   68  3.4 Experimental setup for RF Glow Discharge and results  RF glow is generally used when one of the electrodes is non-conducting. Under these circumstances a layer of negative charge deposits on the electrode which then acts as the cathode. Although CNTs are electrically conducting but the forests that are grown may have both MWNTs and SWNTs, and the latter are usually semiconducting, and so the effects of RF glow discharge on them were deemed to be interesting to investigate. The resultant effects are expected to vary from those in case of DC glow.  In case of RF the experimental setup needs to be modified substantially. The schematic is shown in Figure 3.14. The supply used is a COMDEL CX 600S, which is a 600 W RF source working at 13.56 MHz. In case of RF, the process is more complicated since RF necessitates the output impedance to be matched with the characteristic impedance of the source for proper power transfer into the load. This is accommodated for by the impedance matching network by COMDEL CPMX 1000. Since the output impedance of the glow discharge is not characterized, and fell beyond the limits within which the impedance matching network can automatically tune in to match the impedance, a secondary impedance matching network was designed to bring the output impedance within the working range of the automatic impedance matching network. The design of the secondary output impedance required working backwards from the typical load impedance of the system, and tweaking it to obtain the minimum reflected power. The Agilent 34405A Multimeter, is capable of measuring up to 10 KHz, and hence may not be used for measurements in this experiment. A Rohde & Shwarz Milivoltmeter URV 55 is used to measure the voltage at 13.56 MHz. The combination of the 50 ? resistor with the 120 pF capacitor is well matched with the characteristic impedance of the source, and hence a substantial amount of power flows through with little reflected power. This generates an RF potential difference across   69  this combination which is then used to drive the glow discharge between the electrode and a stainless steel plate, being in series with a 50 ? resistor as shown in Figure 3.14. A URV 55 Milivoltmeter is used with a 40 db attenuator to measure the voltage between the electrodes during the discharge. The glow discharge on stainless steel has been observed as shown in Figure 3.15, where the forward power was 40 W, and reflected being 27 W, hence only 13 W flowing through the output circuit. This can be further improved by attaining better impedance matching. An Ar rich environment was maintained by constantly flowing Ar into the gas chamber.   Figure 3.14: Schematic of RF glow experimental setup   70   Figure 3.16 shows the image of the final setup. The RF supply is connected to the impedance matching network using a coaxial cable. The chassis of the impedance match is connected to the faraday?s cage built to contain the RF field generated by the experiment, to reduce interference with other electronics and also as a safety precaution from the RF radiation. A window is built with very fine mesh to be able to observe the glow. The setup is wrapped with aluminium foil to better prevent radiation of the RF field to the ambience. An image showing the internals; shows the series combination of the 50 ? resistor and 120 pF capacitor to aid in impedance matching and have a power flowing through, the 50 ? resistor in series with the electrode, the gas chamber and the stainless steel plate as the cathode, with the URV 55 Milivoltmeter reading 240 V across the electrodes (which is comparable to the DC case).       Figure 3.15: RF glow on stainless steel                  300?m   71   Figure 3.16: RF Discharge Experimental Setup  The same experimental setup has been used in attempts to obtain a stable RF glow on CNT forests. However it has been observed that the energy of the RF glow appears to be quite high, and resulted in massive uncontrolled eradication of the nanotubes with immense heating effects, rather than a controlled treatment. The process needs to be further modified and optimized by tuning the impedance to attain a stable moderate RF glow to treat the CNT surface locally in a controlled manner as in case of DC.     72  4. Conclusions and future work  This thesis has investigated localized processing of CNT forests via electrical discharges to generate desired geometric structures and treat the CNT surface to change characteristic parameters. A micropatterning method for vertically aligned CNT forests based on lithography-assisted dry ?EDM has been studied. Planar and parallel patterning of microstructures in the nanotube material was demonstrated using the arrays of Cu electrodes that were microfabricated using a UV-LIGA process developed. Batch-mode ?EDM of CNT forests was experimentally shown to be feasible with the lowest discharge energy of 8 nJ in the sinking mode in combination with scan mode using the arrays. The surfaces of patterned microstructures created in the forest exhibited sharp edges with an average 3D surface roughness of ~230 nm. An EDX analysis revealed both negligible contamination of the electrode material and minimal consumption of the electrodes caused by the discharge process. The effects in the use of planar electrodes specific to the batch-mode processing of CNT forests were discussed along with the electrical, surface, and process characteristics observed experimentally. The developed process is potentially scalable to very large-area processing and suitable for mass production of a variety of advanced MEMS and other devices realized by this promising material.  Parallel to applications of arc discharge, treatment of CNT forest surface with localized glow discharge is also studied. Plasma treatment is usually conducted for functionalization of a CNT forest sample, by adding N or H groups. This method may be extended to do that locally in a small region or a pattern. A stable DC glow in inert Ar environment was studied and modulation of CNT diameter is investigated. Reduction of diameter by possible rupturing of outer walls of the MWNTs is observed, and with further exposure to glow discharge coalescence   73  of CNTs occurs. The thickened structures generated are characterized and postulated to be derivatives of the CNTs lacking any general arrangement of the molecules. Generation of the thickened structures at selective regions via local glow discharge may bring about characteristic changes having a myriad of applications in MEMS, and electronics.  Future work encompasses further optimization of the UV-LIGA process for establishing high-aspect-ratio electrodes with larger-scale arrays as well as 3D parallel ?EDM using these electrodes with improved performance in throughput and resolution of the process. In case of glow discharge there is a number of aspects that require further investigations. The characterization of the treated regions needs to be more extensive. EDX may be done at different lower energies to ensure the incident beams only remaining at the surface, and not penetrating down onto the Si substrate. A more thorough Raman spectroscopy needs to be conducted with varying exposure to the glow discharge to see if there is a trend of progressing from the spectra showing two distinct peaks to one showing a lot of fluorescence. This would be a firm evidence of the gradual mutation of the nanotubes into the thickened structure and may help explain the structures better. Electrical and optical characteristic parameters may be measured before and after glow discharge to see the effects. The discharge chamber may be better sealed to ensure complete isolation from the outside environment, and retain a quasi-state of the Ar filled ambience inside rather than having a turbulent flow as in the present case to attain a more stable glow discharge. The RF glow operating parameters will need to be optimized to attain a stable glow on CNTs, and the similar characterization in case of DC, may be performed. A further extension of the project may be local addition of functional groups using this technology. An ambitious development of this project may be that, by locally manipulating the resistance of the CNTs, circuits may be designed, and if the process can be perfected, then CNT based MEMS   74  devices that require a back end electronic circuit would no longer require the electronic part to be a separate Si based device, rather be printed on the CNT itself. Thus this kind of localized treatment or patterned change of characteristics can give rise to an entirely novel research field.                  75  5. References [1] S. Iijima, ?Helical microtubules of graphitic carbon,? , Publ. online 07 Novemb. 1991; {\textbar} doi10.1038/354056a0, vol. 354, no. 6348, pp. 56?58, Nov. 1991. [2] R. H. Baughman, A. A. Zakhidov, and W. A. de Heer, ?Carbon Nanotubes--the Route Toward Applications,? Science (80-. )., vol. 297, no. 5582, pp. 787?792, Aug. 2002. [3] T. W. Odom, J.-L. Huang, P. Kim, and C. M. Lieber, ?Structure and Electronic Properties of Carbon Nanotubes,? J. Phys. Chem. B, vol. 104, no. 13, pp. 2794?2809, Apr. 2000. [4] B. Yakobson and R. Smalley, ?Fullerene nanotubes: C1,000,000 and beyond.,? Am. Sci., vol. v85, no. n4, 1997. [5] M. M. J. Treacy, T. W. Ebbesen, and J. M. Gibson, ?Exceptionally high Young?s modulus observed for individual carbon nanotubes,? Nature, vol. 381, no. 6584, pp. 678?680, Jun. 1996. [6] B. Yakobson, C. Brabec, and J. Bernholc, ?Nanomechanics of Carbon Tubes: Instabilities beyond Linear Response,? Phys. Rev. Lett., vol. 76, no. 14, pp. 2511?2514, Apr. 1996. [7] M. Dahmardeh, A. Nojeh, and K. Takahata, ?Possible mechanism in dry micro-electro-discharge machining of carbon-nanotube forests: A study of the effect of oxygen,? J. Appl. Phys., vol. 109, no. 9, pp. 93304?93308, May 2011. [8] K. Takahata and Y. B. Gianchandani, ?Batch mode micro-electro-discharge machining,? J. Microelectromechanical Syst., vol. 11, no. 2, pp. 102?110, Apr. 2002. [9] M. T. Richardson and Y. B. Gianchandani, ?Achieving precision in high density batch mode micro-electro-discharge machining,? J. Micromechanics Microengineering, vol. 18, no. 1, p. 015002, Jan. 2008. [10] S. M. Yi, M. S. Park, Y. S. Lee, and C. N. Chu, ?Fabrication of a stainless steel shadow mask using batch mode micro-EDM,? Microsyst. Technol., vol. 14, no. 3, pp. 411?417, Nov. 2007. [11] T. Li, Q. Bai, and Y. B. Gianchandani, ?High precision batch mode micro-electro-discharge machining of metal alloys using DRIE silicon as a cutting tool,? J. Micromechanics Microengineering, vol. 23, no. 9, p. 095026, Sep. 2013. [12] H. Guckel, ?High-aspect-ratio micromachining via deep X-ray lithography,? Proc. IEEE, vol. 86, no. 8, pp. 1586?1593, 1998. [13] C. K. Malek and V. Saile, ?Applications of LIGA technology to precision manufacturing of high-aspect-ratio micro-components and -systems: a review,? Microelectronics J., vol. 35, no. 2, pp. 131?143, 2004. [14] H.-K. Chang and Y.-K. Kim, ?UV-LIGA process for high aspect ratio structure using stress barrier and C-shaped etch hole,? Sensors Actuators A Phys., vol. 84, no. 3, pp. 342?350, 2000.   76  [15] H. Lorenz, M. Despont, N. Fahrni, J. Brugger, P. Vettiger, and P. Renaud, ?High-aspect-ratio, ultrathick, negative-tone near-{UV} photoresist and its applications for {MEMS},? Sensors Actuators A Phys., vol. 64, no. 1, pp. 33?39, Jan. 1998. [16] K. Y. Lee, ?Micromachining applications of a high resolution ultrathick photoresist,? J. Vac. Sci. Technol. B Microelectron. Nanom. Struct., vol. 13, no. 6, p. 3012, Nov. 1995. [17] Y. H. Yan, M. B. Chan-Park, Q. Zhou, C. M. Li, and C. Y. Yue, ?Functionalization of carbon nanotubes by argon plasma-assisted ultraviolet grafting,? Appl. Phys. Lett., vol. 87, no. 21, p. 213101, 2005. [18] A. Felten, C. Bittencourt, J. J. Pireaux, G. Van Lier, and J. C. Charlier, ?Radio-frequency plasma functionalization of carbon nanotubes surface O2, {NH3}, and {CF4} treatments,? J. Appl. Phys., vol. 98, no. 7, p. 74308, 2005. [19] B. N. Khare, P. Wilhite, R. C. Quinn, B. Chen, R. H. Schingler, B. Tran, H. Imanaka, C. R. So, C. W. Bauschlicher,, and M. Meyyappan, ?Functionalization of Carbon Nanotubes by Ammonia Glow-Discharge:  Experiments and Modeling,? J. Phys. Chem. B, vol. 108, no. 24, pp. 8166?8172, Jun. 2004. [20] B. Khare, P. Wilhite, B. Tran, E. Teixeira, K. Fresquez, D. N. Mvondo, C. Bauschlicher, and M. Meyyappan, ?Functionalization of Carbon Nanotubes via Nitrogen Glow Discharge,? J. Phys. Chem. B, vol. 109, no. 49, pp. 23466?23472, Dec. 2005. [21] Y. W. Zhu, C.-H. Sow, M.-C. Sim, G. Sharma, and V. Kripesh, ?Scanning localized arc discharge lithography for the fabrication of microstructures made of carbon nanotubes,? Nanotechnology, vol. 18, no. 38, p. 385304, Sep. 2007. [22] P. G. Collins, M. Hersam, M. Arnold, R. Martel, and P. Avouris, ?Current Saturation and Electrical Breakdown in Multiwalled Carbon Nanotubes,? Phys. Rev. Lett., vol. 86, no. 14, pp. 3128?3131, Apr. 2001. [23] J. Cumings, P. G. Collins, and A. Zettl, ?Materials: Peeling and sharpening multiwall nanotubes,? Nature, vol. 406, no. 6796, p. 586, Aug. 2000. [24] M. Terrones, P. M. Ajayan, F. Banhart, X. Blase, D. L. Carroll, J. C. Charlier, R. Czerw, B. Foley, N. Grobert, R. Kamalakaran, P. Kohler-Redlich, M. R?hle, T. Seeger, and H. Terrones, ?N-doping and coalescence of carbon nanotubes: synthesis and electronic properties,? Appl. Phys. A Mater. Sci. Process., vol. 74, no. 3, pp. 355?361, Mar. 2002. [25] D.-S. Chung, S. H. Park, H. W. Lee, J. H. Choi, S. N. Cha, J.-M. W. Kim, J. E. Jang, K. W. Min, S. H. Cho, M. J. Yoon, J. S. Lee, C. K. Lee, J. H. Yoo, J. E. Jung, Y. W. Jin, Y. J. Park, and J. B. You, ?Carbon nanotube electron emitters with a gated structure using backside exposure processes,? Appl. Phys. Lett., vol. 80, no. 21, pp. 4045?4047, May 2002. [26] J. W. G. Wilder, L. C. Venema, A. G. Rinzler, R. E. Smalley, and C. Dekker, ?Electronic structure of atomically resolved carbon nanotubes,? Nature, vol. 391, no. 6662, pp. 59?62, Jan. 1998.   77  [27] W. Liang, M. Bockrath, D. Bozovic, J. H. Hafner, M. Tinkham, and H. Park, ?Fabry - Perot interference in a nanotube electron waveguide,? Nature, vol. 411, no. 6838, pp. 665?669, Jun. 2001. [28] P. Kim, L. Shi, A. Majumdar, and P. L. McEuen, ?Thermal transport measurements of individual multiwalled nanotubes,? arXiv:cond-mat/0106578, Jun. 2001. [29] Y. Chen, C. Liu, and Y. Tzeng, ?Carbon-nanotube cold cathodes as non-contact electrical couplers,? Diam. Relat. Mater., vol. 12, no. 10, pp. 1723?1728, 2003. [30] J. Zhang, G. Yang, Y. Cheng, B. Gao, Q. Qiu, Y. Z. Lee, J. P. Lu, and O. Zhou, ?Stationary scanning x-ray source based on carbon nanotube field emitters,? Appl. Phys. Lett., vol. 86, no. 18, p. 184104, Apr. 2005. [31] J. Suehiro, G. Zhou, and M. Hara, ?Fabrication of a carbon nanotube-based gas sensor using dielectrophoresis and its application for ammonia detection by impedance spectroscopy,? J. Phys. D. Appl. Phys., vol. 36, no. 21, pp. L109?L114, Nov. 2003. [32] T. D. Wilkinson, X. Wang, H. Butt, R. R, and W. I. Milne, ?Sparse multiwall carbon nanotube electrodes arrays for liquid crystal photonic devices,? in Photonic Devices + Applications, 2008, pp. 705011?705011?10. [33] H. Zhou, A. Colli, A. Ahnood, Y. Yang, N. Rupesinghe, T. Butler, I. Haneef, P. Hiralal, A. Nathan, and G. A. J. Amaratunga, ?Arrays of Parallel Connected Coaxial Multiwall-Carbon- Nanotube???Amorphous-Silicon Solar Cells,? Adv. Mater., vol. 21, no. 38???39, pp. 3919?3923, Oct. 2009. [34] Y. Wang, K. Kempa, B. Kimball, J. B. Carlson, G. Benham, W. Z. Li, T. Kempa, J. Rybczynski, A. Herczynski, and Z. F. Ren, ?Receiving and transmitting light-like radio waves: Antenna effect in arrays of aligned carbon nanotubes,? Appl. Phys. Lett., vol. 85, no. 13, p. 2607, Sep. 2004. [35] K. Kempa, B. Kimball, J. Rybczynski, Z. P. Huang, P. F. Wu, D. Steeves, M. Sennett, M. Giersig, D. V. G. L. N. Rao, D. L. Carnahan, D. Z. Wang, J. Y. Lao, W. Z. Li, and Z. F. Ren, ?Photonic Crystals Based on Periodic Arrays of Aligned Carbon Nanotubes,? Nano Lett., vol. 3, no. 1, pp. 13?18, Jan. 2003. [36] C. Journet, W. K. Maser, P. Bernier, A. Loiseau, M. L. de la Chapelle, S. Lefrant, P. Deniard, R. Lee, and J. E. Fischer, ?Large-scale production of single-walled carbon nanotubes by the electric-arc technique,? Nature, vol. 388, no. 6644, pp. 756?758, Aug. 1997. [37] Y. Ando, X. Zhao, T. Sugai, and M. Kumar, ?Growing carbon nanotubes,? Mater. Today, vol. 7, no. 10, pp. 22?29, Oct. 2004. [38] E. Verploegen, A. J. Hart, M. De Volder, S. Tawfick, K.-K. Chia, and R. E. Cohen, ?Non-destructive characterization of structural hierarchy within aligned carbon nanotube assemblies,? J. Appl. Phys., vol. 109, no. 9, pp. 94315?94316, May 2011.   78  [39] J. Li, H. T. Ng, A. Cassell, W. Fan, H. Chen, Q. Ye, J. Koehne, J. Han, and M. Meyyappan, ?Carbon Nanotube Nanoelectrode Array for Ultrasensitive {DNA} Detection,? Nano Lett., vol. 3, no. 5, pp. 597?602, May 2003. [40] K. S. Karimov, M. Saleem, Z. M. Karieva, A. Khan, T. A. Qasuria, and A. Mateen, ?A carbon nanotube-based pressure sensor,? Phys. Scr., vol. 83, no. 6, p. 065703, Jun. 2011. [41] T. Wang, K. Jeppson, L. Ye, and J. Liu, ?Carbon-Nanotube Through-Silicon Via Interconnects for Three-Dimensional Integration,? Small, vol. 7, no. 16, pp. 2313?2317, 2011. [42] Y. Fu, N. Nabiollahi, T. Wang, S. Wang, Z. Hu, B. Carlberg, Y. Zhang, X. Wang, and J. Liu, ?A complete carbon-nanotube-based on-chip cooling solution with very high heat dissipation capacity.,? Nanotechnology, vol. 23, no. 4, p. 045304, Mar. 2012. [43] S. Hu, Z. Xia, and L. Dai, ?Advanced gecko-foot-mimetic dry adhesives based on carbon nanotubes.,? Nanoscale, vol. 5, no. 2, pp. 475?86, Jan. 2013. [44] Z. Xiao, M. Saquib Sarwar, M. Dahmardeh, M. Vahdani Moghaddam, A. Nojeh, and K. Takahata, ?Cone-shaped forest of aligned carbon nanotubes: An alternative probe for scanning microscopy,? Appl. Phys. Lett., vol. 103, no. 17, p. 171603, 2013. [45] S. Tamulevicius, K. Babilius, and A. Matiukas, ?Temperature conditions during arc discharge plasma deposition of titanium nitride,? Surf. Coatings Technol., 1995. [46] K. P. Rajurkar and Z. Y. Yu, ?3D Micro-EDM Using CAD/CAM,? CIRP Ann. - Manuf. Technol., vol. 49, no. 1, pp. 127?130, 2000. [47] K. P. Rajurkar, G. Levy, A. Malshe, M. M. Sundaram, J. McGeough, X. Hu, R. Resnick, and A. DeSilva, ?Micro and Nano Machining by Electro-Physical and Chemical Processes,? CIRP Ann. - Manuf. Technol., vol. 55, no. 2, pp. 643?666, 2006. [48] D. . Pham, S. . Dimov, S. Bigot, A. Ivanov, and K. Popov, ?Micro-EDM?recent developments and research issues,? J. Mater. Process. Technol., vol. 149, no. 1, pp. 50?57, 2004. [49] D. Reynaerts, P.-H. Heeren, and H. Van Brussel, ?Microstructuring of silicon by electro-discharge machining (EDM) ? part I: theory,? Sensors Actuators A Phys., vol. 60, no. 1, pp. 212?218, 1997. [50] S. Kalpakjian and S. Schmid, Manufacturing Processes for Engineering Materials (5th Edition). Prentice Hall, 2007, p. 1040. [51] K. . Ho and S. . Newman, ?State of the art electrical discharge machining (EDM),? Int. J. Mach. Tools Manuf., vol. 43, no. 13, pp. 1287?1300, Oct. 2003. [52] E. C. Jameson, Electrical Discharge Machining. SME, 2001, p. 329. [53] E. C. Jameson, Electrical discharge machining: tooling, methods, and applications. Society of Manufacturing Engineers, Marketing Services Division, 1983, p. 238.   79  [54] K. Takahata, ?Micro-Electro-Discharge Machining Technologies for {MEMS},? in in Micro Electronic and Mechanical Systems, K. Takahata, Ed. InTech, 2009. [55] B. M. Schumacher, ?After 60 years of EDM the discharge process remains still disputed,? J. Mater. Process. Technol., vol. 149, no. 1, pp. 376?381, 2004. [56] M. Kunieda, B. Lauwers, K. P. Rajurkar, and B. M. Schumacher, ?Advancing EDM through Fundamental Insight into the Process,? CIRP Ann. - Manuf. Technol., vol. 54, no. 2, pp. 64?87, 2005. [57] M. Kunieda, ?Challenges to miniaturization in micro EDM,? Proceeding twenty-third Annu. Meet. ?, 2008. [58] F. Han, S. Wachi, and M. Kunieda, ?Improvement of machining characteristics of micro-EDM using transistor type isopulse generator and servo feed control,? Precis. Eng., vol. 28, no. 4, pp. 378?385, 2004. [59] H. Singh, ?Experimental study of distribution of energy during EDM process for utilization in thermal models,? Int. J. Heat Mass Transf., vol. 55, no. 19, pp. 5053?5064, 2012. [60] M. Kunleda, Y. Miyoshi, T. Takaya, N. Nakajima, Y. ZhanBo, and M. Yoshida, ?High Speed 3D Milling by Dry EDM,? CIRP Ann. - Manuf. Technol., vol. 52, no. 1, pp. 147?150, 2003. [61] H. Lorenz, M. Despont, N. Fahrni, J. Brugger, P. Vettiger, and P. Renaud, ?High-aspect-ratio, ultrathick, negative-tone near-UV photoresist and its applications for MEMS,? Sensors Actuators A Phys., vol. 64, no. 1, pp. 33?39, 1998. [62] C.-H. Lin, G.-B. Lee, B.-W. Chang, and G.-L. Chang, ?A new fabrication process for ultra-thick microfluidic microstructures utilizing SU-8 photoresist,? J. Micromechanics Microengineering, vol. 12, no. 5, pp. 590?597, Sep. 2002. [63] H. Lee, K. Lee, B. Ahn, J. Xu, L. Xu, and K. W. Oh, ?A new fabrication process for uniform SU-8 thick photoresist structures by simultaneously removing edge bead and air bubbles,? J. Micromechanics Microengineering, vol. 21, no. 12, p. 125006, Dec. 2011. [64] P. M. Dentinger, W. M. Clift, and S. H. Goods, ?Removal of {SU-8} photoresist for thick film applications,? Microelectron. Eng., vol. 61?62, no. 0, pp. 993?1000, Jul. 2002. [65] W. Khalid, M. S. M. Ali, M. Dahmardeh, Y. Choi, P. Yaghoobi, A. Nojeh, and K. Takahata, ?High-aspect-ratio, free-form patterning of carbon nanotube forests using micro-electro-discharge machining,? Diam. Relat. Mater., vol. 19, no. 11, pp. 1405?1410, Nov. 2010. [66] W. Khalid, M. S. M. Ali, M. Dahmardeh, Y. Choi, P. Yaghoobi, A. Nojeh, and K. Takahata, ?High-aspect-ratio, free-form patterning of carbon nanotube forests using micro-electro-discharge machining,? Diam. Relat. Mater., vol. 19, no. 11, pp. 1405?1410, Nov. 2010. [67] T. Saleh, M. Dahmardeh, A. Bsoul, A. Nojeh, and K. Takahata, ?High-precision dry micro-electro-discharge machining of carbon-nanotube forests with ultralow discharge energy,? in 2012   80  {IEEE} 25th International Conference on Micro Electro Mechanical Systems ({MEMS)}, 2012, pp. 259?262. [68] G. Paul, S. Roy, S. Sarkar, N. Hanumaiah, and S. Mitra, ?Investigations on influence of process variables on crater dimensions in micro-EDM of ?-titanium aluminide alloy in dry and oil dielectric media,? Int. J. Adv. Manuf. Technol., vol. 65, no. 5?8, pp. 1009?1017, May 2012. [69] S. Esconjauregui, R. Xie, M. Fouquet, R. Cartwright, D. Hardeman, J. Yang, and J. Robertson, ?Measurement of area density of vertically aligned carbon nanotube forests by the weight-gain method,? J. Appl. Phys., vol. 113, no. 14, p. 144309, Apr. 2013. [70] T. Saleh, M. Dahmardeh, A. Nojeh, and K. Takahata, ?Dry micro-electro-discharge machining of carbon-nanotube forests using sulphur-hexafluoride,? Carbon N. Y., vol. 52, pp. 288?295, 2013. [71] Q. H. Zhang, J. H. Zhang, S. F. Ren, J. X. Deng, and X. Ai, ?Study on technology of ultrasonic vibration aided electrical discharge machining in gas,? J. Mater. Process. Technol., vol. 149, no. 1, pp. 640?644, 2004. [72] S. H. Yeo and L. K. Tan, ?Effects of ultrasonic vibrations in micro electro-discharge machining of microholes,? J. Micromechanics Microengineering, vol. 9, no. 4, pp. 345?352, Dec. 1999. [73] Y. Zhao, X. Zhang, X. Liu, and K. Yamazaki, ?Geometric modeling of the linear motor driven electrical discharge machining (EDM) die-sinking process,? Int. J. Mach. Tools Manuf., vol. 44, no. 1, pp. 1?9, 2004. [74] L. Vandsburger, S. Coulombe, and J. L. Meunier, ?Degradation of carbon nanotubes in oxygen glow discharges,? Carbon N. Y., vol. 57, pp. 248?258, Jun. 2013. [75] L. Vandsburger, S. Coulombe, and J.-L. Meunier, ?Degradation of carbon nanotubes by electron bombardment in radio-frequency glow discharge afterglows,? J. Phys. D. Appl. Phys., vol. 46, no. 48, p. 485301, Dec. 2013. [76] Y. Liu, L. Liu, P. Liu, L. Sheng, and S. Fan, ?Plasma etching carbon nanotube arrays and the field emission properties,? Diam. Relat. Mater., vol. 13, no. 9, pp. 1609?1613, Sep. 2004. [77] M. Mazilu and A. De Luca, ?Fluorescence background suppression in Raman spectroscopy,? ? Lasers ?, 2010. [78] J. Zhao, H. Lui, D. I. McLean, and H. Zeng, ?Automated autofluorescence background subtraction algorithm for biomedical Raman spectroscopy.,? Appl. Spectrosc., vol. 61, no. 11, pp. 1225?32, Nov. 2007. [79] Z.-M. Zhang, S. Chen, Y.-Z. Liang, Z.-X. Liu, Q.-M. Zhang, L.-X. Ding, F. Ye, and H. Zhou, ?An intelligent background-correction algorithm for highly fluorescent samples in Raman spectroscopy,? J. Raman Spectrosc., vol. 41, no. 6, pp. 659?669, Oct. 2009.    

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0166888/manifest

Comment

Related Items