Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigation for improvement and application of MEMS-based micro-electro-discharge machining (M3EDM) Wang, Ningyuan 2011

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

Item Metadata

Download

Media
24-ubc_2011_spring_wang_ningyuan.pdf [ 4.18MB ]
Metadata
JSON: 24-1.0071595.json
JSON-LD: 24-1.0071595-ld.json
RDF/XML (Pretty): 24-1.0071595-rdf.xml
RDF/JSON: 24-1.0071595-rdf.json
Turtle: 24-1.0071595-turtle.txt
N-Triples: 24-1.0071595-rdf-ntriples.txt
Original Record: 24-1.0071595-source.json
Full Text
24-1.0071595-fulltext.txt
Citation
24-1.0071595.ris

Full Text

Investigation for Improvement and Application of MEMS-Based MicroElectro-Discharge Machining (M3EDM) by  Ningyuan Wang  B.Sc., Beihang University, 2008  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in The Faculty of Graduate Studies (Electrical and Computer Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) January 2011 © Ningyuan Wang, 2011  Abstract MEMS-based micro-electro-discharge machining (M3EDM) is a batch microfabrication technique utilizing planar electrode actuators to machine conductive materials. The low contrast pattern transfer issue in electroplating mold fabrication is firstly analyzed and improved by eliminating the contact gap and adding a rehydration step. The new method gives a better structure profile with near vertical sidewalls. The causes and mechanisms of spin coating nonuniformity and bonding voids are discussed, as well. The deformation on the foil electrode area where discharges occur is explained by abnormal heat shock, tool wear, material softening and discharge-brought reactive force. A feedback control circuit with pulse discrimination is developed to detect the harmful short pulse and prevent thermal shock. Nickel is proposed and tested as the new material for actuators owing to its higher mechanical and thermal resistance. The optimized nickel based electrodes together with the affiliation circuit are applied to cantilever MEMS contact switch fabrication. The photoresist melting in the photoresist sandwich structure is observed. A new reverse fabrication process is proposed and processed in order to minimize the photoresist melting. The method partially addresses the issue. The further directions for improvements and the potential application of the reverse process to reusable M3EDM devices are discussed.  ii  Table of Contents Abstract .......................................................................................................................................... ii Table of Contents ......................................................................................................................... iii List of Tables ................................................................................................................................. v List of Figures ............................................................................................................................... vi Acknowledgements ...................................................................................................................... ix Dedication ...................................................................................................................................... x Introduction ................................................................................................................................... 1 1.1 Micro-Electro-Discharge Machining (μEDM) ................................................................... 3 1.1.1 Principle of μEDM ....................................................................................................... 3 1.1.2 Setup of μEDM ............................................................................................................ 5 1.1.3 Application of μEDM................................................................................................... 9 1.2 Batch Mode μEDM........................................................................................................... 10 1.3 MEMS-Based Micro-Elctro-Discharge Machining (M3EDM) ........................................ 12 1.3.1 Machining Principle ................................................................................................... 13 1.3.2 Devices and Machining Results ................................................................................. 14 1.4 Thesis Outline ................................................................................................................... 16 2 Analysis and Improvement of MEMS-Based Micro-Elctro-Discharge Machining (M3EDM) ..................................................................................................................................... 17 2.1 Introduction ...................................................................................................................... 17 2.2 Electrode Actuator Formation .......................................................................................... 18 2.2.1 Electroplating Mold.................................................................................................... 18 2.2.2 Structure Uniformity .................................................................................................. 22 2.3 Improvement of Process Control: Feedback System ........................................................ 25 2.3.1 Circuit Design ............................................................................................................ 26 2.3.2 Test Results and Discussion ....................................................................................... 30 2.4 Other Material Options and Development of Nickel-Based Electrode Actuator ............. 32 iii  2.4.1 Motivation .................................................................................................................. 32 2.4.2 Design for Nickel-Platform Devices .......................................................................... 38 2.4.3 Nickel Wet Etching .................................................................................................... 42 2.4.4 Devices Fabrication .................................................................................................... 47 2.4.5 Results ........................................................................................................................ 49 3 Application of M3EDM-Preliminary Study ........................................................................... 52 3.1 Introduction ...................................................................................................................... 52 3.2 Design ............................................................................................................................... 54 3.3 Fabrication and Experimental Results .............................................................................. 56 3.4 Discussion ......................................................................................................................... 61 3.5 Conclusion ........................................................................................................................ 63 4 Conclusion and Future Work ................................................................................................. 64  iv  List of Tables Table 2.1: Copper and Candidate Metals and Their Characteristics............................................ 37  v  List of Figures Figure 1.1: The principle and process of μEDM ........................................................................... 4 Figure 1.2: The setup of sinking EDM .......................................................................................... 5 Figure 1.3: (a) The RC type pulse generator (b) the transistor type generator .............................. 7 Figure 1.4: Three types of sacrificial electrode machining ........................................................... 8 Figure 1.5: Stainless steel 3-D structure machined by 5-axis μEDM system................................ 9 Figure 1.6: The setup of batch-mode μEDM ............................................................................... 11 Figure 1.7: The fabrication process of tool electrode for batch-mode μEDM............................. 11 Figure 1.8: Batch mode μEDM pulse generation (a) single pulse generator setup (b) parallel pulse generator setup..................................................................................................................... 12 Figure 1.9: The machining principle of M3EDM ........................................................................ 13 Figure 1.10: Schematic views of (a) single layer device and (b) double layer device ............... 15 Figure 1.11: (a) SEM image of a single layer device (b) optical image of a double layer device flipped over and a detailed image taken by SEM ......................................................................... 15 Figure 1.12: (a) SEM image of the surface machined by a single layer device (b) optical measurement of pattern “UBC” machined by a double layer device (c) SEM image of cavity array machined by a double layer device ...................................................................................... 16 Figure 2.1: (a) SEM image of electroplated patterns on the actuator (b) detailed view of a single electroplated structure under SEM, noting that the sidewalls of the pattern are not vertical ....... 18 Figure 2.2: The old setup of electroplating photoresist mold fabricaition................................... 19 Figure 2.3: The new setup of electroplating photoresist mold fabrication .................................. 21 Figure 2.4: SEM image of a mold fabricated by the new process ............................................... 21 Figure 2.5: (a) SEM image of an electrode array defined by the improved mold (b) detailed view of a single structure in (a) ............................................................................................................. 22 Figure 2.6: (a) Height measurement of dielectric anchors vs. the distance from the center of 3” wafer (b) measured depths of stainless steel holes machined using double layer device with 50 50μm electrode arryas ........................................................................................................... 23 vi  Figure 2.7: Deformation of the copper electrode structure .......................................................... 25 Figure 2.8: (a) Measured current pulses of discharge (b) close up of a single pulse ................... 26 Figure 2.9: The schematic voltage and current patterns of different types of pulses .................. 27 Figure 2.10: The schematic of the feedback control circuit ........................................................ 29 Figure 2.11: The final setup of feedback control circuit.............................................................. 30 Figure 2.12: (a) Wave pattern in long term (b) wave pattern of one working cycle when short pulse happens (c) wave pattern when stable normal pulses happen. ............................................ 31 Figure 2.13: The schematic of a single discharge in μEDM........................................................ 33 Figure 2.14: Measured reaction force of a single pulse (30A) .................................................... 36 Figure 2.15: Sample layout of electrode (a) cantilever design for single layer device (b) crab leg design for double layer device ...................................................................................................... 39 Figure 2.16: (a) The photoresist pattern (30 30μm holes with 200μm spacing) (b) nickel etching sample with poor profile (c) optimized nickel etching result with vigorous manual agitation. ... 44 Figure 2.17: The setup of air blow agitation etching. .................................................................. 45 Figure 2.18: (a) The etching result of air blow etching (5min, 40oC, 60 60μm holes) (b) the etching result of manual agitaion etching (12min, 40oC, 60 60μm holes, ~150rpm) (c) SEM image of nickel etching hole (d) the front view of copper etching under mild manual agiataion (20min room temperture, 30 30μm holes, ~60rpm) (e) the rear view of the same hole in (d) ... 46 Figure 2.19: (a) Fabrication of single layer device (b) fabrication of double layer device ......... 48 Figure 2.20: Experiment setup of M3EDM test ........................................................................... 49 Figure 2.21: SEM image of a nickel single layer device after machining ................................... 50 Figure 2.22: (a) The top view of a crab leg electrode under SEM (b) the flipped over view of a double layer device (c) the surface machined by the electrode shown in (b) ............................... 51 Figure 3.1: (a) Image of a toggle switch/ohmic contact (b) image of a shunt airbridge/capacitive coupling switch ............................................................................................................................. 53 Figure 3.2: One example of M3EDM designed for contact switch fabrication ........................... 56 Figure 3.3: The fabrication process of contact switch from photoresist sandwich structure....... 57  vii  Figure 3.4: Different views of the same location (shown in the rectangle framework) (a) the front side view of the foil after machining (b) the backside view of the foil after machining (c) the photoresist under the foil after machining (part of photoresist was damaged during stripping) (d) the stylus scanning at the arrow direction shown in the (c) .................................................... 58 Figure 3.5: The reverse fabrication process for contact switch ................................................... 59 Figure 3.6: (a) Backside of machined area under front illumination (b) detailed image under front illumination (c) backside of machined area under background illumination (d) detailed image under background illumination .......................................................................................... 61 Figure 3.7: The draft of reusable M3EDM device ....................................................................... 62  viii  Acknowledgements I would like to thank my family for the endless support. I am certain I would have not achieved anything in my life if it were not for their guidance, encouragement, and support. I would like to acknowledge the help and guidance of my supervisor, Dr. Kenichi Takahata. I would like to thank him for all the suggestions and advice. His thoughts, reviews, and feedback were the base of my work. I would, also, like to thank Dr. Boris Stoeber and Dr. Edmond Cretu for their suggestions and ideas. Their insights and suggestions encouraged me to approach problems from a number of different perspectives. As for my colleagues and friends in lab, they were my second family throughout the past two years. For their willingness to discuss the research and share their cumulative experience, I would like to thank Greg Wong, Mohamed Sultan Mohamed Ali, Prasad Bharath, Anas Bsoul, Babak Assadashangabi, Masoud Dahamardeh and Dan Brox. I would like to offer them my sincere and full gratitude. My friends Hussein, Mrigank, Hadi, Eile and Simon, thank you for your friendship, support and the time together.  ix  Dedication  To My Family  x  Chapter 1 Introduction Microelectromechnical systems (MEMS) have attracted much interest recently. MEMS products are now commercially available. Microfabrication plays a crucial part in realization of MEMS devices designs and commercialization of prototypes. The matured silicon based micromachining technologies used in integrated circuit (IC) manufacturing have greatly stimulated the development of microfabricaiton. These techniques often incorporate parallel processes which facilitate high throughput production and cost effectiveness for volume production. A limitation of these technologies is that they are generally designed for thin layer structures or two-dimensional structures, which have limited mechanical performance. This limits MEMS fabrication when high aspect ratio structures or three-dimensional structures are desired to achieve better performance. Efforts have been made to enhance the structure aspect ratio based on current micromachining principles. Deep etching technologies such as DRIE (deep reactive ion etching) and micromolding technologies such as LIGA (German acronym for x-ray lithography, electrodepositing and molding) have successfully been developed to achieve structures etching and growing on the substrate with high aspect ratio, respectively. However, there is another restraint in IC related fabrication techniques. Most of technologies transferred from IC manufacturing are limited to standard microelectronic materials, e.g. silicon and silicon oxide. Materials that are suitable for the mechanical design, e.g. hard metals and robust alloys, are difficult to pattern using these concept technologies.  1  Miniaturizing mechanical machining technologies is another way to approach micromachining. The idea is to create the micron-scale or nanometer-scale counterparts of technologies utilized in the mechanical machining industry. The miniaturized mechanical machining technologies have the ability to machine most materials, including engineering alloys. Materials can be removed by mechanically cutting and milling, thermal evaporation or electrochemical reaction. Micro-electro-discharge machining (μEDM) is one of the techniques in this approach. However, the application of the technique is still limited by several factors, e.g. cost and size. In order to fulfill the micron level accuracy machining requirement, ultra-high precise numerical control (NC) system and stepper motors, usually expensive, have to be used. Also, it is difficult to miniaturize the control stage and the relevant servo system accordingly. Moreover, the throughputs of these technologies are low. A new micromachining approach has been presented to overcome these constraints. Instead of using complex macro-scale systems, the micromachining is processed and controlled by MEMES actuators. These actuators are much cheaper than the NC system and stepper motors. Besides, since the total area needed for unit machining device is in magnitude of mm2 or less, it will be easy to integrate devices in a small area. Development of this compact layout design may lead to the realization of the so-called „micro-factory‟. Chaitanya et al. have successfully demonstrated μEDM based on this new concept. Two versions of MEMS actuators have been fabricated directly on stainless steel wafers for μEDM in a dielectric oil environment. The actuator or moving electrode is actuated by electrostatic force created by a DC voltage which also serves to charge the electrode via a resistance-capacitance (RC) circuit for discharge ignition. This thesis will investigate and evaluate improvements of the formerly mentioned approach.  2  Possible application of this new machining technology in MEMS contact switch fabrication is also undertaken. In section 1.1, the principle, basic setup and applications of μEDM will be briefly introduced. Section 1.2 is about the efforts to achieve batch μEDM on conventional platform. Section 1.3 will introduce previous research on the new concept μEDM. In the last section 1.4, an outline of the thesis will be presented.  1.1  Micro-Electro-Discharge Machining (μEDM) Electro-discharge machining (EDM) is a thermoelectric machining process. The machining  is based on the thermal electrical energy generated between a tool electrode and a workpiece submerged in a dielectric fluid. The heat created by the pulsed discharge between the two parts will remove the workpiece material by melting and evaporation. The process does not apply any contact force to the workpiece. Traditionally, EDM has been applied to machine hard and brittle metals in complex designs. Recently, the development of high precision control and spark generation systems allowed for μEDM to be developed and utilized as means to create microfeatures with high aspect ratio and good surface quality [1, 2]. This attractive method has enabled high aspect ratio precise machining on nontraditional MEMS materials from hard metals such as WC-Co [3] to soft materials such as carbon nanotube forests [4]. MEMS devices fabricated by μEDM, e.g. shape memory alloy stents [5] and capacitive pressure sensors [6], have been presented, as well. 1.1.1 Principle of μEDM An electrode and a conductive workpiece submerged in a dielectric fluid are the basic components involved in μEDM. The principle and process of μEDM are illustrated in figure 1.1. 3  Initially, the electrode and the workpiece are separated by a small gap called spark gap, which is filled with a dielectric liquid. The voltage difference applied to the tool electrode (cathode) and the workpiece (anode) will ignite the discharging process once it exceeds the dielectric breakdown threshold. Upon discharge, a plasma channel immediately builds up, allowing conduction of electrons and ionized molecules. Typically, the discharging energy of a mini-arc pulse occurred in μEDM process is 0.01-10μJ with a duration about 10-100ns. The temperature rises dramatically in magnitude of several 1000oC locally at the erosion point on the workpiece due to the high discharge current pulse, melting and evaporating the surrounding areas. Meanwhile, the extreme change of temperature also evaporates the dielectric fluid, which generates air bubbles and a shockwave. The melting area of the workpiece resolidifies into metal particles that are flushed away by the dielectric fluid shockwave. A crater is left on the workpiece at the machining point. These steps are repeated until the desired level of machining is reached [7, 8].  Figure 1.1: The principle and process of μEDM [7] 4  1.1.2 Setup of μEDM The setup of sinking EDM is shown in figure 1.2 [9]. This is a common configuration used in commercial μEDM machines. The conductive workpiece is fixed to the bottom of a tank which is mounted on an X-Y table and the tool electrode is held by a Z-axis stage. Both workpiece and tool electrode are submerged in a dielectric liquid. A pumping system, which is designed for recycling and renewing the dielectric liquid, is connected to the tank. The  Figure 1.2: The setup of sinking EDM [9]  discharge condition between the two parts is monitored and controlled by a gap condition analyzer and a pulse generator, respectively. To achieve better machining rates and quality, a servo feedback system is usually added to control the discharge gap. All the movements and pulse generation are controlled by an NC system using the gap condition analyzer as feedback. There are generally two types of pulse generation system, an RC pulse generator and a transistor type pulse generator (figure 1.3). The RC pulse generator utilizes the charge and discharge of the capacitor via a resistor to control the pulse timing. This simple structure design is to be capable  5  of providing the short pulse-on time for high quality μEDM. However, the discharging frequency of the RC generator is low due to the charging time of the capacitor. Moreover, it is difficult for the RC generator to achieve uniform pulse lengths and discharge intervals. However, the transistor type generator easily overcomes these disadvantages. The voltage pulse is created by switching of a transistor allowing the discharge to be triggered at a higher frequency. The machining uniformity and discharge interval is easily maintained by a switching control signal. The drawback to the transistor generator is that it is difficult to provide 10-8 second scale short pulses needed for μEDM, owing to the response delay of the electronics [10, 11]. The tool electrode can be fabricated in several methods. The electrode material used is usually a hard metal or alloy such as tungsten in conventional μEDM [3]. Patterning the tool electrode by μEDM with another sacrificial electrode is a common method of fabrication. The sacrificial electrode could be a stationary metal block or a rotary thin metal disc or a guided running wire (figure 1.4). The latter is commonly known as wire electrode-discharge grinding (WEDG), which is the favored method for tool fabrication. WEDG has a very accurate machining dimension control, however, produces lower surface smoothness than the metal block and the disc machining under the same condition. The process of the tool machining is outlined below. The tool electrode, usually a thin wire, is mounted onto an accurately controlled spindle. The spindle spins the tool electrode as it approaches the sacrificial electrode so that discharge could occur. Sparks between the two electrodes machine the tool electrode, meanwhile the spindle is moved forward or backward in direction of the rotary axis according to the electrodes contact condition. When the tool electrode completes the machining for the set point with desired machined length, it will be moved closer to the sacrificial electrode and the process will be repeated until the machine depth reaches the given value [12].  6  Figure 1.3: (a) The RC type pulse generator (b) the transistor type generator [10]  To further optimize the performance of μEDM, ultrasonic vibration has been introduced to μEDM process. It is known that the machining speed of μEDM may drop dramatically when it reaches a certain depth in high aspect ratio micro-hole drilling due to the debris accumulation at the machining surface. The debris is typically the resolidified materials vaporized in the machining process. Once concentration of the debris between the machining gap reaches a certain threshold, the harmful discharge occurs. With a higher current level, the harmful discharge brings abnormal thermal damage to both tool electrode and workpiece. Ultrasonic  7  vibration has been demonstrated to successfully remove debris during microstructure fabrication. The aspect ratio and the machining quality have both been improved [13]. The tool electrode vibration is not desirable in high accuracy or ultra-tall structure machining processes as tool electrodes are more sensitive and easier to damage by vibration in these cases. To overcome this, the vibration unit may be attached to the workpiece instead [14].  Figure 1.4: Three types of sacrificial electrode machining [12]  In previous efforts, the dielectric liquid has also been investigated. In conventional EDM, kerosene or EDM oil is commonly used as the dielectric liquid. Different types of nanopowders or additive powders have been added into the liquid for a better machining process [15]. Applications of other dielectric liquids, e.g. deionized (DI) water, have been reported, as well. Although DI water has larger machining gap than kerosene, higher machining rate and lower tool wear have been demonstrated in DI water μEDM. Unlike large craters on the machined surfaces in kerosene, the machined surfaces in DI water are covered with small pits resulting from 8  electrochemical effects [16]. The application of DI water as the dielectric liquid could also eliminate the production of carbon byproducts that exist in kerosene case [17]. 1.1.3 Application of μEDM μEDM has the capability to cut a wide range of electrical conductive materials, e.g. metals, alloys, cemented carbides, conductive ceramics and highly doped silicon. With proper setup and design, μEDM could fabricate 3D structures with high accuracy. The example shown in figure 1.5 is a 3-D metal structure machined by a 5-axis μEDM system with two rotational axes. The structure is a combined design of several mechanical components [18]. However, since the process is serial, the process has a low throughout and is very time consuming.  Figure 1.5: Stainless steel 3-D structure machined by 5-axis μEDM system [18]  μEDM has been used to machine the high aspect ratio structures components with a high degree of accuracy. Nozzle machining is one of the common applications. The nozzle could be either used in ink-jet [19] or diesel engine [20]. The fabrication of magnet heads for VCR by μEDM has also been reported [21]. μEDM has been applied to the fabrication of alignment microshafts and pins used in the miniature assembly, as well [22]. Micro-mold fabrication is another major application of μEDM. The mold materials are usually robust metals or alloys 9  which are difficult to machine by other conventional machining methods. A mold could be used in further process including batch μEDM [7]. The additional applications of μEDM include microfilters, components of accelerometers, force balanced transducers, micro-gears, microsprings, fiber-optic light detector fixtures [23].  1.2  Batch Mode μEDM Batch mode μEDM is a technology developed to increase the throughput of conventional  μEDM. Instead of using a single micro-needle tool electrode, the new concept setup uses an array of high aspect ratio structures as the tool electrode to achieve a parallel machining process [22]. In order to realize the concept, X-ray related processes such as LIGA have been used to fabricate the array electrode. The set up of batch mode μEDM is illustrated in figure1.6 and the fabrication process is shown in figure 1.7. The use of an array electrode has greatly improved the throughput [24]. The electrode array fabricated by μEDM has been also reported [25]. However, machining an array of repeating structures with high accuracy is a difficult process. The processes discussed above are under a single pulse generator. The discharge ignition is a serial process, which limits the final machining rate of batch mode machining. Parallel pulse generation has been designed to address the problem. Each individual electrode is connected to its own pulse generation circuit (figure1.8), which allows multi-discharges at the same time. The throughput of the batch process using a parallel pulse generator improved throughput more than 100 fold compared with a traditional serial μEDM process [24]. It is worth noting that the expensive NC systems are still used in these cases. These techniques still belong to miniaturizing the macro machining concept.  10  Figure 1.6: The setup of batch-mode μEDM [24]  Figure 1.7: The fabrication process of tool electrode for batch-mode μEDM [26]  11  Figure 1.8: Batch mode μEDM pulse generation (a) single pulse generator setup (b) parallel pulse generator setup [27]  1.3  MEMS-Based Micro-Electro-Discharge Machining (M3EDM) MEMS-based micro-electro-discharging machining (M3EDM) is a new concept for batch  mode μEDM based on MEMS actuators. This approach is proposed to achieve the precise machining at low cost by eliminating the use of expensive NC systems. Accurate machining is achieved by applying MEMS actuators with a pre-patterning tool electrode. The machining pattern or the tool electrode is designed in batch mode for a high throughput. 1.3.1 Machining Principle The machining principle of M3EDM is shown in figure 1.9. The electrode is fabricated directly on the top of the workpiece (e.g. a stainless steel wafer), but is separated by dielectric  12  photoresist anchors with a pre-determined gap size. The RC pulse generation circuit is connected to the electrode and the workpiece. The capacitor is fully charged initially. (step-1) When the switch is turned on, a DC voltage is applied to the electrode and the workpiece. The voltage acts as electrostatic actuation, pulling the electrode towards the workpiece. The DC voltage applied is higher than the so-called pull-in voltage of the electrode. In this case, the mechanical force of the electrode structure is not able to balance the electrostatic force so that the electrode  Figure 1.9: The machining principle of M3EDM [28]  13  will approach to the workpiece until it touches the workpiece surface. Before the physical touch happens, a discharge spark is ignited. (step-2) This discharge releases the energy stored in the capacitor and lowers the voltage between the electrode and the workpiece, thereby lowering the electrostatic force on the electrode and withdrawing electrode back to the original position. The thermal impact of discharge leaves a crater on the workpiece. (step-3) The capacitor is then recharged via a resistor and the voltage is restored. The step-2 and step-3 are repeated until it reaches the limit, i.e. the voltage is unable to pull the electrode to the discharging gap. When the machining is completed, the electrode is lifted off [28]. The machining depth may be increased by applying larger DC voltage. However, since the voltage acts as charging voltage for the discharge pulse ignition, increasing the voltage increases the energy of a single pulse and decreases the machining quality. To overcome this trade-off, external assistance actuations such as a hydrodynamic force have been introduced. The external force can allow for additional displacement which would enable further machining [29]. 1.3.2 Devices and Machining Results There are two types of designs fabricated in the previous efforts: single layer and double layer. The schematic views are shown in figure 1.10. In the case of single layer design, the tool electrode is the same metal foil structure that forms the microactuator. This design has an ability to cut micro-pillars out of the workpiece. The double layer device has electroplated patterns under the foil-formed actuator as the tool electrode, which is suitable to machine cavities. Images of these structures and their machining results taken by scanning electron microscope (SEM) or optical microscope or optical profiler are shown in figure1.11 and figure 1.12, respectively [30].  14  Figure 1.10: Schematic views of (a) single layer device and (b) double layer device  Figure 1.11: (a) SEM image of a single layer device (b) optical image of a double layer device flipped over and a detailed image taken by SEM [30]  15  Figure 1.12: (a) SEM image of the surface machined by a single layer device (b) optical measurement of pattern “UBC” machined by a double layer device (c) SEM image of cavity array machined by a double layer device [30]  1.4  Thesis Outline The chapters of this thesis are set out as follows. In chapter 2, the structure fabrication  profiles and the electrode deformation in previously discussed designs are analyzed, improved and addressed by optimizing the lithographic process, adding feedback control and utilizing new electrode material. In chapter 3, a preliminary investigation of potential M3EDM application in MEMS contact switch fabrication is presented. Chapter 4 concludes the overall efforts.  16  Chapter 2 Analysis and Improvement of MEMSBased Micro-Electro-Discharge Machining (M3EDM) 2.1  Introduction  Chaitanya‟s work has successfully demonstrated the potential of M3EDM as mentioned in chapter 1. However, there are still a number of hurdles that exist [30]. The structural measurement revealed a poor electroplated pattern and spatial non-uniformity in photroresist structures. Deformation of the electrode structure after long term machining was also observed under SEM. To overcome the electrode structure formation problems, analyses of the photolithography process, mechanisms of the spin coating and the polymer bonding process are investigated. Solution for the low-contrast patterns transfer and possible solutions to structure non-uniformity are presented. Thermal shock caused by short pulses, tool wear, material softening and discharge reaction force have been addressed as possible causes of structural deformation of the electrode. A feedback control circuit is designed to prevent thermal shock damage. Possible material change from copper to a metal with higher resistance to the above mechanism is considered. The design and fabrication of actuators based on a new material are also demonstrated. 17  Section 2.2 contains the analyses of the electroplating mold fabrication process, the spin coating process, the polymer bonding process and experimental solution or possible solutions to the structure formation issues. In section 2.3, the design of the feedback control is presented. Section 2.4 discusses the investigation of suitable actuator material and the fabrication of actuators based on the new material.  2.2  Electrode Actuator Formation  2.2.1 Electroplating Mold  Figure 2.1: (a) SEM images of electroplated patterns on the actuator (b) detailed view of a single electroplated structure under SEM, noting that the sidewalls of the pattern are not vertical [30]  Electroplating using a thick photoresist mold is an economical method to achieve structures with relative high aspect ratio. The photoresist could be either negative (e.g. Microchem Su-8), or positive (e.g. Rohm & Haas SPR220-7), or wet (e.g. Diaplate 132) or dry (e.g. Ordyl P-50100) [31]. SPR220-7 is one of the suitable candidates for M3EDM application as it can be easily removed by acetone in sacrificial releasing and has good resistance to the acidic electroplating 18  environment. In the previous work, SPR220-7 has been used successfully as a mold for the tool electrode electroplating. However, the pattern transferred shows a low profile with non-vertical sidewalls as shown in figure 2.1, which limits and downgrades the quality of the machining. A better mold fabrication process is needed.  Figure 2.2: The old setup of electroplating photoresist mold fabrication  Figure 2.2 illustrates the old setup of mold fabrication. An 18μm copper foil is fixed to a glass wafer by taping on the front side. A thick layer photoresist is then spin coated and patterned by hard contact lithography. The tape (~60μm) introduces the largest gap between the mask and sample, which reduces the pattern transfer resolution. Using polymer bonding [32] is one possible solution. But removing the foil after electroplating without damaging the photoresist structure above is difficult due to the soft and fragile nature of the foil. Fixing the foil by taping on the backside of the glass plate is another option, though the fixing strength is low. To increase the strength, a thin layer of water is applied between the foil and the glass plate. The air pressure is sufficient to hold the foil firmly in place on the glass. This method has been reported as temporary bonding after alignment in the glass bonding process [33]. It is worth noting that this water bonding strength decays with time, which allows for easy removal of the foil after the electroplating. 19  A further factor that prevents a good contact between the photoresist and mask is the edge bead (EB) [34, 35]. This is the accumulation of the photoresist at the edge of a wafer during the spin coating process, which creates a thicker photoresist ring around. Addition of short high speed and high acceleration spinning steps in the middle and at the end of the spin coating process can reduce the width of the EB. A further acetone spraying after soft baking can eliminate the bead [34]. This would, however, require a spray gun in the spinner system, which is not available in the AMPEL cleanroom. Instead, a lithography EB removing method presented in Kraft‟s work was applied [35]. The EB was exposed to a high UV light dose (exposure energy per unit area) and was successfully dissolved in the MF-26 developer before the mold patterning. A relaxing time or a rehydration step is important for the thick photoresist process [36]. SPR220-7 is a novolak resin based positive tone photoresist [31]. During the soft baking, the solvent in the photoresist is dried out, which strengthens the structure and promotes the chemical resistivity of the structure. However, the moisture in the polymer is also depleted. The water level plays an important part in the photoactive reaction and development of the novolak resin photoresist. This is of little concern in the thin layer process as the moisture is easily reabsorbed from the atmosphere. The diffusion of vapor, however, is much slower in a thick photoresist structure. This demands an additional step of water level restoration by either soaking in the water or exposing to the atmosphere with sufficient moisture for a long term [36]. In the experiment, both samples with and without rehydration have been tested. After experiencing the same baking and exposure, samples were developed in the MF-26 developer. The developing time for rehydrated samples was ~4 minutes compared with 6-8 minutes for the samples without rehydration. In addition, the exposure time needed for completely pattern transferring under the same baking conditions was decreased from 14 min to ~10min, as well. The improved setup of  20  the process is shown in figure 2.3. The mold fabricated by the improved process is shown in figure 2.4. SEM images of the electroplated pattern defined by the new mold are presented in figure 2.5. Pattern with better profile and nearly vertical sidewalls has been achieved.  Figure 2.3: The new setup of electroplating photoresist mold fabrication  Figure 2.4: SEM image of a mold fabricated by the new process  21  Figure 2.5: (a) SEM image of an electrode array defined by the improved mold (b) detailed view of a single structure in (a)  2.2.2 Structure Uniformity Structure uniformity is a determining factor in M3EDM process. Non-uniform structures will lead to incompletely M3EDM pattern transferring and variance in machining depth. In order to evaluate the uniformity of the fabrication, dielectric anchor heights of single layer structures and the depths of holes which are machined by a double layer device with 5x5 electrode array have been characterized using a stylus profiler. The results are shown in figure 2.6. Spatial variation in anchor heights and non-uniformity in machining depths were observed. The plot in figure 2.6a shows a random pattern with an average height of 68μm and a trend of height increasing from center to edge of the wafer for anchors heights. The machining depths of holes in batch machining have an average value of 20.7μm with ±17.4 % margin (figure 2.6b). This is due to the structure unevenness in the form of dielectric anchor heights differences and electroplating structures non-uniformity. Both single layer and double layer structures are formed  22  by spin-coated photoresist structures with polymer bonding. These results indicated the need to further investigate the spin coating and polymer bonding processes in order to find the causes to identify possible solutions for the structure non-uniformity.  Figure 2.6: (a) Height measurement of dielectric anchors for single-layer electrode device vs. the distance from the center of 3” wafer (b) measured depths of stainless-steel holes machined using a double layer device with 50 × 50-μm2 electrode arrays  The photoresist spin coating process has been modeled and discussed in previous studies [37, 38]. There are two main mechanisms in the spin coating process, the fluidic outflow and the solvent evaporation mechanisms. In the experiments using thin positive photoresist OFPR-800, the behavior of the photoresist can be separated into 3 stages. In the first stage, the photoresist is spread over the wafer by the initial acceleration. The photoresist weight loss in this stage is mainly due to the outflow rather than the evaporation. In the following stage, the outflow  23  decreases dramatically while the evaporation increases. After reaches a maximum level, the evaporation speed begins to decrease. The photoresist behaves as a classic Newtonian fluid in this stage. Both thickness and variation of the photoresist are reduced. Through continuous spinning, the outflow, the evaporation and the thickness of the photoresist decrease, however, the uniformity of the photoresist thickness begins to decline because of the non-Newtonian effect and the temperature variation on the wafer. In order to achieve a uniform coating, identification of an optimal spin time is important [37]. Other factors, e.g. air turbulence [37], spin rate [38], environment temperature [38, 39], air humidity level [39], soft baking surface leveling [40] and relaxing time between spinning and soft baking (especially for the thick photoresist process) [41], are all reported. Current experiment conditions prohibit the regulation and optimization of the majority of these factors, which necessitates much experimental work and accurate measurements for each specific application. Other possible improvements including constantvolume-injection method may enhance the structure uniformity [40]. A thin layer of S1813 photoresist polymer is spin-coated on top of the photoresist sacrificial layer and serves as a bonding layer between the laminated structure and the sacrificial layer. The bonded sample is then cured by baking. Although the manual bonding can successfully build up a structure, bubbles or voids in the bonding layer were observed. The main causes of the voids are air trapped during the lamination process and solvent evaporation during the curing. Bonding in a vacuum environment and a prebaking before bonding easily eliminate air trapped and reduce the solvent evaporation as reported [32]. It is worth noting that the photoresist structures do not have high resistance to liquid S1813. Leaving liquid S1813 on the photoresist structure for an extended period of time, e.g. 1 minute, may soften the structure and introduce additional  24  structure variance. Hard baking of the photoresist sacrificial layer before bonding could strengthen the structure and partially address the problem.  2.3  Improvement of Process Control: Feedback System  Ideally, M3EDM process under constant voltage will automatically stop once it reaches the device limit due to the increase of the gap between the electrode and the workpiece surface and the decrease of the electrostatic force. In this case, the electrode should be separate from the workpiece. However, short circuits between the electrode and the workpiece were often observed. The SEM image (figure 2.7) confirms physical contact and shows the copper foil electrode structure deformation of the area where discharge occurs. The deformation is one of the causes for non-uniform machining. Several mechanisms may relate to the deformation. Abnormal discharging is one of the possible causes. It is reported that the harmful discharge pulse creates huge thermal shock and damages both the electrode and the workpiece [42].  Figure 2.7: Deformation of the copper electrode structure  25  The M3EDM process was conducted under a simple RC circuit in the previous demonstration. Figure 2.8 shows the measured current wave pattern during the machining process. Although there was an in situ current probe monitor, the process control was based on manual observation and switching. In the following section, a feedback control circuit for the thermal shock prevention is presented.  Figure2.8: (a) Measured current pulses of discharge (b) close up of a single pulse [30]  2.3.1 Circuit Design Research into macro EDM process identified that the discharge pulses between the electrode and the workpiece can be generally separated into 5 types, which have different current and voltage wave patterns. Figure 2.9 shows the schematic voltage and current patterns of these pulses. As its name suggests, the open pulse means there is no discharge between the gap. In this case, the energy is continuously charged, and a high voltage level with low current readout is observed. A discharge happens in both the spark pulse and the arc pulse. These two types are  26  distinguished by a short ignition delay time (open pulse) before gap current peak. The spark pulse has the delay before the discharge. The arc pulse that has no delay produces a higher removing rate but rougher surface, for this reason the spark pulse is considered as the „normal pulse‟ in many cases. The off pulse has no current and voltage readout, which indicates that the machining is at rest. The short pulse is the most damaging pulse. It has a high current output with a negligible small voltage readout between the gap, which shows a metal bridge is built up between the workpiece and the electrode, i.e. short circuit. The electrode must be removed immediately from the workpiece in order to avoid further damage. A servo feed system in the setup continuously controls the electrode position according to the pulse type in order to get good machining quality [42].  Figure 2.9: The schematic voltage and current patterns of different types of pulses [42]  27  According to the measurements shown in figure 2.8, the peak current for the ordinary pulse is ~ 2A with a pulse-on time of ~50ns. The pulse frequency is in magnitude of MHz or higher. Current is estimated using Ohm law (I = U/R). Substituting the values of the circuit (U=100V150V, R=20KΩ), the short circuit current is at I= 0.005-0.0075A. In the design assumption, the peak current of normal pulse is very high with its spectrum mainly focuses in the MHz or higher waveband, i.e. the normal pulse produces lower amplitude output than the short pulse in the KHz or lower waveband. In order to get the readout for the low frequency waveband, an accurate current sensing resistor is used to convert the current into a voltage signal. The voltage signal is then filtered and amplified by an active low pass filter and detected by an oscilloscope. Verification experiments were processed using single layer devices with the formerly mentioned setup. The wave pattern of low pass filtered signal had a DC pattern with minor ripples. Its output level fits the model assumption. Unlike the commercially available μEDM machine, the tool electrode in M3EDM is controlled by the electrostatic force created by the DC voltage. In order to separate the tool electrode from the workpiece, the electrostatic force has to be removed by shutting down the DC voltage. This can be achieved by utilizing the same type of switch transistor which is used in the transistor pulse generator [11]. After the DC voltage is cut off, the electrode will be pulled up to the starting position by the mechanical force of the structure. As mentioned in Chapter 1, the pulse generation will also be off since its power is provided by the actuation DC voltage. It has been observed that the tool electrode in the short pulse state needs a cooling time to guarantee the separation after the DC voltage is off. To achieve this, a delay signal created by a timer circuit is applied.  28  The detailed schematic and the setup of designed feedback control circuit are shown in figure 2.10 and figure 2.11, respectively. The circuit first reads out the converted in situ current signal via the current sensing resistor. The signal then passes through the low pass active filter with amplifying based two operational amplifiers (OP). Once the output level of the filtered signal is higher than the pre-set threshold tested from an actual short circuit, the OP comparator will create a dropping edge immediately, which is detected by the following timer. The timer creates a delay signal with a hardware-set duration. This delay signal is then converted into a control signal. The switch transistor receives the control signal and shuts off the DC voltage for a period of time. The machining process stops while the output of the filtered signal drops. After which, the feedback control is ready for a new detection cycle.  Figure 2.10: The schematic of the feedback control circuit  29  Figure 2.11: The final setup of feedback control circuit  2.3.2 Test Results and Discussion The designed feedback control circuit was linked to a M3EDM device. Low-pass filtered current signal, in situ current signal and discharge created bubbles and arcs in the process were recorded. The typical working signal of feedback control is shown in figure 2.12. Figure 2.12b shows a detection cycle of the feedback control. Once the filtered signal reaches the short pulse threshold, the voltage is shut off. The discharge will be stopped, i.e. no bubbles and arcs will be observed. Meanwhile, the output of the filtered signal will return to the off circuit level. The circuit continuously shuts down the voltage when the short pulses occur so extending the machining time (figure 2.12a). Figure 2.12c shows wave pattern of the low pass filtered signal when stable normal pulses occur. In this case, the signal level is lower than the threshold, i.e. no  30  short pulse occurs. The pulse generation circuit is always on and continuous bubble release and arc discharges can be observed. It is worth noting that the in situ current signal detected by the current probe in figure 2.12 suffered from a severe data loss due to the low sampling rate. Further test have shown acceptable results for short pulse detection and thermal shock prevention.  Figure 2.12: (a) Wave pattern in long term (b) wave pattern of one working cycle when short pulse happens (c) wave pattern when stable normal pulses happen  Voltage detection is not considered in the design, however, in the previous research, pulse discrimination based on voltage is well documented [42, 43]. The combination of voltage and current detection using a computing unit may provide greater accuracy in pulse discrimination than the current setup. In the experiment, machining with a higher low pass filtered signal output 31  (still lower than the short pulse threshold) tended to have larger machining depth and area. The machining progress is optically invisible because the machined area is covered by the actuator. It is possible to build a critical relationship between the output of low pass filtered signal and the material removing rate. This relationship may potentially be applied to machining progress estimation. The circuit successfully detected the short pulse and prevented thermal shock. However, thermal effects of the normal pulse may also cause the deformation of the copper foil structure. The following section will investigate the possible materials that may reduce the deformation caused by normal pulse effects.  2.4  Other Material Options and Development of Nickel-Based Electrode Actuator  2.4.1 Motivation Although the feedback control has negated the harmful impact of short pulse discharges, the normal pulse discharge may also damage the electrode copper foil structure in various mechanisms. The so-called tool wear is one of the mechanisms. Tool wear in a single discharge process in μEDM was modeled as shown in figure 2.13. A plasma channel will build up once the discharge occurs. The radius of the channel at the cathode (the tool electrode) is smaller compared with the anode (the workpiece) as the cathode emits electrons and the anode emits positive ions. Both anode and cathode are bombarded and molten by the emitted charges from the opposite polarities in the machining process. Because the mass of an ion (e.g. 1.67x10-27kg) is much larger than the mass of an electron, 9.11x10-31kg, the anode (the workpiece) has a larger molten area compared with the cathode (the tool electrode) [8]. Consequently, the tool electrode 32  and the workpiece are both machined in the discharge process, though the machining speed for the tool electrode is much slower than the machining for the workpiece. The electrode and the supporting actuator are the same structure in M3EDM. Therefore, tool wear also weakens the actuator structure for the single layer device.  Figure 2.13: The schematic of a single discharge in μEDM [8]  Thermal material softening may further decrease the foil integrity. This phenomenon is discussed in the micro-contact switch research. It is reported that the contact hardness, which describes the material response to plastic deformation, is reduced when the contact area reaches the softening temperature of contact material. The softening temperature is much lower than the melting point, e.g. ~ 100oC for gold whose melting point is about 1000oC [43]. The softening temperature of copper is 120oC in a macro-scale switch test [44]. In other material science research, the softening of different commercially available copper foils has been also demonstrated. The softening temperature has a range from 150oC to 250oC [45]. This suggests 33  that the discharge, which could reach magnitude of several 1000oC, creates a larger area with temperature exceeding the softening temperature around the molten zone, which further reduces the foil structure strength. In the machining process, an electrostatic force is applied to the electrode structure for the device operation. The structure weakened by the tool wear and the softening is more susceptible to the deformation. The deformation reduces the gap between the electrode and the workpiece, which increases electrostatic force on the deformed area. This process is self-accelerated. The deformation will continue at an increasing rate until the electrode makes contact with the workpiece, causing a short circuit. The discharge generated forces may increase the load on the deformed area of copper foil actuator structure. Reaction force is one of such forces. The force is created primarily by the expansion and contraction of the bubbles generated by the discharging plasma in the dielectric fluid, which creates the shockwave in the electrical discharging machining process [9]. The reaction force is considered as one of the causes of the wire electrode vibration and deflection in wire EDM (WEDM) [47]. The reaction force can be ignored in needle based μEDM as it has an effect parallel to the gap surface (the shockwave) and a small longitudinal strain effect. However, in WEDM, the strain model is closer to the bending beam model whose spring constant is typically 1000 times smaller than the longitudinal strain model. In the case of M3EDM, the strain caused by the discharge reaction force can be considered as the beam deflection. This additional load could increase the rate of structure deformation. Furthermore, unlike the WEDM whose electrode continuously renews, the electrode used in M3EDM is not renewable. The electrode structure experiences a periodical load of the reaction force during the machining process. This may potentially lead to metal fatigues [48].  34  The magnitude of the reaction force could be simulated or inverse calculated from the experimental measurement [47]. Due to the complexity of the calculation and lack of the data, an estimate of scale from literatures will be presented in the following part. In order to simplify the dynamic bubble creation process, assume a cylinder channel and the reaction force is equal to the pressure multiplied by the top surface area of the cylinder. The discharge energy is estimated according to the energy store in the capacitance (both built-in capacitance and pulse generation capacitance) [49]:  (2.1) The capacitance measured in previous experiments was around 200-300pF [30], the voltage was around 100V-150V in the experiment and the discharge energy was ~1μJ accordingly. The pulse-on time is around ~50ns estimated from the measurement. The peak current is about 2A. After conversion of these parameters into a rectangle wave pattern by approximation, values could be substituted into the following equation [50]: (2.2)  d (t) is the diameter of the ionized channel, t is the pulse duration in second, and ie is the peak value of the current in ampere. t is ~50ns and ie is ~1A after conversion. The diameter of the channel is ~3μm after substituting these values. There are several results reported for plasma pressures. The result of low energy discharge (12 μJ) is chosen in the estimation, which is ~100KPa compared with 10MPa in the high energy case [51]. The final force is in ~μN level. This is much lower than the measurement of marco EDM (maximum 50N). It is worth noting that the reaction force is a dynamic force in single pulse and the effect could be up or down, as shown in figure 2.14 [52]. In the estimation, only maximum value of the reaction force is  35  considered. More critical model and verification experiment are needed to investigate the insight of the reaction force effect.  Figure 2.14: Measured reaction force of a single pulse (30A) [52]  To overcome these effects, materials other than copper have been considered as electrode material. Table 2.1 shows parameters of copper and candidate metals. A material with high boiling and melting point is shown to have a better resistance to tool wear and this tendency is independent of the workpiece material [53]. A material with higher softening temperature and greater hardness than copper is needed for its greater resistance to the thermal softening. The price of material is a key factor for the batch process. Tungsten and rhenium have high softening temperature and high hardness, but are expensive in general. The elastic modulus, which is related to the actuation voltage of actuators, also requires evaluation. Basically, a design fabricated of material with higher elastic modulus needs a higher actuation voltage to reach the discharge distance, which may lead to poorer machining quality in M3EDM. Metals such as chrome are not desirable due to their high elastic modulus. Nickel is a potential candidate under these restraints. Applying nickel as electrode material could partially address the unwanted deformation problem. However, it is worthy to note that nickel is not a good electrode material 36  for EDM process [55]. Combining the copper electrode with the nickel actuator platform may yield a better performance than a single material design. Table2.1: Copper and Candidate Metal and Their Characteristics [54]  Resistivity  Brinell  Elastic  Softening  Melting  Modulus  Temperature  Temperature  Material  (10-8 Ωm)  Hardness  (GPa)  (oC)  (oC)  Copper  1.75  35  120  190  1083  Platinum  11.7  40  154  540  1773  Iridium  5.5  170  530  -  2450  Tungsten  5.5  350  350  1000  3390  Molybdenum  5.8  250  350  900  2620  Nickel  8  70  210  520  1452  Cobalt  9.7  125  210  -  1495  Rhodium  4.5  55  300  1966  1966  Rhenium  9.71  250  470  1400  3170  Iron  10  60  200  500  1540  Chrome  20  90  900  -  1615  Tantalum  14  40  190  800  2996  37  2.4.2 Design for Nickel-Platform Devices Both nickel single layer and double layer devices based on copper structure fabrication process [30] were designed and fabricated. The single layer devices were patterned from a 15μm nickel foil. The actuator could be in cantilever or fixed-fixed or torsion tether configurations. A design of cantilever is shown in figure 2.15a. The dimension of the movable electrode is 860 3160μm. The anchor pad is 2800 2800μm which is constant in all single layer design. There are arrays of 60 60μm through releasing holes with a spacing of 170um in the designs. The diameter of the through hole will be discussed in the following section. The double layer devices were formed by a patterned 15μm thick nickel foil with 25-30μm electroplated copper structures beneath. The actuator layout was similar to the single layer design except for a crab leg platform, shown in figure 2.15b. The 60 60μm releasing holes are located throughout the suspended structure with a spacing of 200μm. The custom electroplated pattern is 100 100μm and the anchor is 2100 2100μm. The crab leg double layer structure (figure 2.15b) is an octagon electrode with four crab leg tethers. The structure is believed to have better actuation stability. The pull-in voltage is a critical design parameter for the electrostatic actuation. It is important to note that two assumptions are preset in the following discussion in order to simplify the calculation. Firstly, the mechanical effects of the releasing holes are assumed to be negligible. Secondly, the pattern transfer in the fabrication is assumed to be perfect, i.e. the fabricated samples have exact dimensions of their mask layout designed. In this case, measurement from mask can be used as input parameters for the estimation. Equations presented in Chaitanya et al. works and design books for parallel plate electrostatic mechanical structures were applied. The pull-in voltage VPI-TS for torsion actuation is described below [30]:  38  (2.3)  Figure 2.15: Sample layout of electrode (a) cantilever design for single layer device (b) crab leg design for double layer device  39  L is the length (perpendicular to the orientation of the torsion beam) of the electrode pad; W is the width of the electrode pad respectively; d is the original distance between the electrode and the workpiece; KTLis the spring constant torsion beams; l is the length of the torsion tether/beam; a is one half of the tether/beam width, b is one half of the tether/beam thickness and G is the torsion modulus of nickel. is the permittivity of the EDM oil. The pull-in voltages for the cantilever or fixed-fixed configurations are shown as follow [30, 57]:  (2.4) A is the area electrode; d is original distance; K is spring constant of each case. The value of K equals to KFF and KCL for the fixed-fixed and the cantilever design, respectively. In both fixedfixed and cantilever designs, lv, w, h and E are the length, width, thickness and Young‟s modulus of suspended structure. P in the fixed-fixed equation is a complex combination force effect because of the intrinsic stress in the suspended and nonlinear effect due to the deformation. Neglected the residual stress of the foil, the P and k could be in the following form [57]:  (2.5) I is the moment of inertia. is the Poisson ratio of the nickel. ymax is the maximum deflection of the beam, it is one third of original distance d in the electrostatic mechanical actuation assumption. After substituting the relevant constant of nickel (E=200GPa, G=75GPa, h=15μm,  ) and EDM oil (  ), the gap to achieve a pull-in voltage in  80V-150V for the torsion beam is ~ 12μm -22μm. The gaps for the fixed-fixed and the cantilever  40  range from 26μm-38μm and 45μm-60μm, respectively. This estimation is also dependent on the design dimensions. The crab leg double layer structure is more complex than single layer device. Here the surface of supporting structure and designed electrode structure are not at the same level. The parallel-plate model applied in the previous discussion is not suitable to analyze the overall electrostatic effect. The critical model derived from system energy analysis [58], or modified parallel plate model can be used. These two models are mathematically equal in this case. The pull-in distance and voltage equation for voltage controlled actuator are presented [58]: (2.6)  (2.7)  x is the displacement between the two plate, xPI is the pull-in distance. Um and C are the mechanical energy stored in the system and the capacitance of the system, respectively. The equations for related parameters of the crab leg structure and the spring constant of crab legs are shown below [56]:  (2.8)  (2.9)  s, c and w are the dimensions of crab leg, which are indicated in figure 2.15b. is the Poisson ratio of the material. A1 and A2 are the electrostatic effective area of the supporting structure and 41  the custom design, respectively. A is the top view area of the electrode. d1 and d2 are the original distance of the supporting structure and the custom design, respectively. h is the height of the custom design, i.e. the height of the electroplated structure. Substituting the above relationship to the equation 2.6, the final form is a complex quadratic equation which is difficult to solve. The pull-in distance should have a value between d2/3 and d1/3. A simplified model is introduced here for further estimation. The model consists of a visual single layer structure representing the overall effects of two layers so that the parallel plate model could be applied. This visual single layer has an area of A= A1+A2. d is the original distance of the visual structure. Let electrostatic force of the virtual structure equals the electrostatic force of real structure at the initial point, i.e. the following relationship:  (2.10)  The estimated value of pull-in distance is d/3. After substituting the pull-in distance into the equation 2.7, the equation gives the pull-in voltage. Assume a fixed 30μm height for electroplating structure so that the gap between the electroplated pattern and the workpiece becomes the only variance to discuss for the design. The gap between the electroplated pattern and the workpiece of the crab leg structure for a pull-in voltage from 80V to 150V is ~ 9μm14μm, which is within the estimated range. 2.4.3 Nickel Wet Etching Before the fabrication process is undertaken, the wet etching of nickel was investigated and verification experiments were performed. Several candidate etchants are available. Electrochemical process may be used in many cases [59]. Commercially available etchant recipes were  42  selected [60], for their confirmed photoresist compatibility. A diluted nitric acid solution or nickel etchant TFB (Transene Co. MA, USA) was first tested. The typical application of this recipe is 150nm thin layer etching with an etching rate of 3nm/s at 25oC. The thickness of the nickel foil is 15μm. The projected etching time is around 2 hours. The surface was patterned successfully by the photoresist mask. However, the etching rate was much slower than expected. The sample was soaked in the solution for 6 hours without etching through. Another etchant based on ferric chloride called nickel etchant type I from the same company is favored since it has relatively high etching rate (3 mil/hr at 40°C or 76μm/hr at 40°C). In the first step large pattern etching verification, the etchant was preheated by a hotplate and successfully etched through the foil in 6-7 minutes at 40°C with similar mild agitation used in the copper etching process (~60rpm). In the second small pattern etching verification, a severe undercut problem was observed after etching for 13min (figure 2.16b). The photoresist mask was an array of 30  30 μm patterns with a 200μm spacing (figure 2.16a). Unlike ideal isotropic etching,  horizontal etching rate was much higher than vertical etching rate. High spatial non-uniformity was also observed. The rate and quality of ferric chloride etching are related to the concentrations of the ions, e.g. the byproducts, ferrous ions and nickel ions, and so-called free acid concentration [61]. Agitation optimizes the etching process by removing the byproducts and maintaining active solution level. To improve the resulting profile, several manual agitation rates were tested. The optimized result is shown in the figure 2.16c, where etching occurred at the fastest agitation (~150rpm). The etching time was ~12 min. The poor etching profile and the non-uniformity issues were partially solved by the high agitation rate. Accordingly, the process was tested for a range of hole diameters, the etching quality was better for large holes than for small holes at the 43  same agitation rate. To reduce the etching difficulty, the etching pattern was enlarged to 60 60μm. The agitation effects were recorded in previous copper etching, though it did not affect the etching quality much.  Figure 2.16: (a) The photoresist pattern (30 30μm holes with 200μm spacing) (b) nickel etching sample with poor profile (c) optimized nickel etching result with vigorous manual agitation  Due to the stronger dependency of nickel etching on the agitation, a stronger and more stable agitation method is needed. The target is a 3” wafer with pattern. Manual agitation is not an ideal approach to obtain consistent and uniform etching results. The magnet stirring bar system is a potential alternative to replace manual agitation, but it is not easy to implement agitation to wafer size sample in a uniform manner. The use of an ultrasonic bath is another alternative. This could yield a strong and uniform source of agitation on wafer sized sample. However, the photoresist bonding can be damaged by the strong wave and vibration. Air agitation seems to hold the most promise of a solution. Here clean air is pumped into the bottom  44  of the liquid filled tank and released from the perforated pipe. Air bubbles will rapidly expand and run into the liquid surface [62]. The setup of air bubble agitation is shown in figure 2.17. The compressed air is pump via a pipe with an inner diameter of ~6mm. There are eleven holes with a spacing of ~5mm in the pipe. Each hole has a diameter of ~2mm. The flow rate is ~4.5 to 6L/min. The bubbles create sufficient turbulence to remove byproducts by mixing them rapidly with the surrounding solution. The etching time of the new set up was reduced dramatically from 12 min to 4-5min with an acceptable or at least compatible etching quality (figure 2.18a) as the fast manual agitation sample (~150rpm) (figure 2.18b). A benefit of the air agitation etching is that it could easily be upgraded to mass production level, which is the ultimate destiny of this design research.  Figure 2.17: The setup of air blow agitation etching  Compared with the copper etching sample, the nickel etching structure has a different pattern. SEM image of a nickel etching pattern achieved by air agitation is shown in figure 2.18c. The opening of the hole has a dimension of ~60μm. However, there is a ring of 40-60μm 45  undercut or minor etched pattern around the hole. The side wall profile is rough and non-vertical. Figure 2.18d and figure 2.18e show the front and rear view of a pattern in the copper etching. The substrate is an 18μm copper foil. The etching was processed by manual agitation (~60 rpm) at room temperature with an array of 30 30μm holes as a mask. The pattern is overetched and becomes circular due to the isotropic etching. But there is only about 5μm diameter difference between the frontside and backside measurement, which suggests there is an only minor undercut for the copper etching. Large undercuts were also observed at the edges of other patterns in the nickel etching. This factor has to be compensated for in the mask design.  Figure 2.18: (a) The etching result of air blow etching (5min, 40oC, 60 60μm holes) (b) the etching result of manual agitation etching (12min, 40oC, 60 60μm holes, ~150rpm) (c) SEM image of nickel etching hole (d) the front view of copper etching under mild manual agitation (20min, room temperature, 30 30μm holes, ~60rpm) (e) the rear view of the same hole in (d) 46  It is worth noting that photoresist adhesion to the metal surface may also play a part in the wet etching undercut. The use of an adhesion promoter, dry film photoresist and hard baking were not effective in resolving this problem. A similar problem has been observed and overcome by a two step soft baking process [33]. Since the current process has fulfilled the basic requirement for the pattern etching, the new method will be tested when more accurate etching is needed. 2.4.4 Devices Fabrication The devices were fabricated using a 3” stainless steel wafer as the machining substrate. The process of nickel device fabrication is similar to the process used for copper structure fabrication [30]. In the single layer fabrication (figure 2.19a), the stainless steel wafer is first cleaned by acetone. After the HMDS adhesion promoter coating, a 35-40μm photoresist (SPR220-7, Rohm Haas Co. PA, USA) is spun on by coating twice and serves as the sacrificial layer. Each coating has a spin rate of 1000rpm with a duration of 40s, which is followed by a 10min soft baking at 90oC. In the next step, the nickel foil is laminated to the thick layer photoresist and bonded by polymer. The polymer applied is S1813. The bonding layer is spin coated with a rate of 7000rpm for 40s followed by a 2min prebaking at 60oC on a hotplate. The bonded sample is further cured at 90oC in an oven for 5min. Then, a 5μm SPR220-7 is coated and patterned by photolithography for using as a nickel etching mask. The etching is processed in the bubble agitation bath discussed above. Eventually, the sacrificial photoresist layer is released in acetone. The fabrication of a double layer device (figure 2.19b) can be separated into two segments, the top electrode and the bottom wafer. An additional 10-15μm sacrificial photoresist layer is spun on the wafer after the HMDS coating. Through-hole alignment marks are pre-patterned on a nickel foil before the top electrode fabrication. The foil is then fixed onto a glass plate by using 47  water and tape. A 35-40μm thick SPR-220-7 photoresist is spun on the foil using the similar twice coating process applied in the single layer device fabrication followed by a 30min final soft baking step. The photoresist is then put in the fume hood for at least 12 hours to restore the moisture level. The photoresist is then exposed and developed as an electroplating mask. After that, a 30μm copper is deposited by electroplating. The top electrode and the bottom wafer are bonded together using the same polymer bonding followed by mask patterning, etching and structure releasing steps.  Figure 2.19: (a) Fabrication of single layer device (b) fabrication of double layer device 48  2.4.5 Results  Figure 2.20: Experiment setup of M3EDM test  The final devices were tested under the experiment setup presented in figure 2.20. The sample is fixed to a probing platform with a thick insulting glass beneath it. The platform is built in the ultrasonic EDM oil (EDM 185TM, Commonwealth Oil Co, ON, Canada) bath for cleaning byproducts. The values of the R and C in the circuit are set to be 20KΩ and 100pF, respectively. The control circuit is linked to the RC circuit to prevent harmful thermal impact caused by the short pulses and the low-pass filtered signal is monitored in situ by an oscilloscope. Meanwhile, the in situ current is also measured by a current probe which is linked to the same oscilloscope. The nickel single layer samples were first tested. Deformation of the electrode in the discharging area was observed to an extent less in nickel when compared to copper (figure 2.21). However, the discharge rate of nickel was lower than copper even with high actuation voltage of 150V. Better results were obtained by the double layer structures that used copper as the discharge 49  electrode material. Sustainable machining was observed at a high discharge rate, especially for the crab leg nickel structure, whose sustainability was up to 100% on one measured wafer. The necessary voltage for discharge to occur is around 100V which is lower than the estimated value (150V). Possible reasons could include mechanical effect of releasing holes on the crab legs, the undercut discussed in the nickel etching process and the estimation inaccuracy. Figure 2.22 shows a group of images of crab leg electrodes. Figure 2.22a is a top view of unused electrode under SEM. One electrode is flipped over after the test in order to see the backside structure (before the clean), shown in figure 2.22b. The pattern machined by this device is presented in figure 2.22c.  Figure 2.21: SEM image of a nickel single layer device after machining  50  Figure2.22: (a) The top view of a crab leg electrode under SEM (b) the flipped over view of a double layer electrode (c) the surface machined by the electrode shown in (b)  51  Chapter 3 Application of M3EDM: Preliminary Study 3.1  Introduction  MEMS switch is a novel device concept that utilizes mechanical movement to achieve ON and OFF states in transmission line. The actuation principle could be electrostatic force, magneto static force, thermal effect or piezoelectric effect [56]. The direction of motion could be either lateral or vertical [63]. Basically, there are two types of MEMS contact designs. One is called the toggle switch or ohmic contact. This design uses the cantilever as part of the line structure and direct metal contact to achieve an ON state (figure 3.1a). The ohmic switching principle gives the device a broad frequency bandwidth. This design is suitable for low frequency application. The second design, the shunt airbridge or capacitive coupling switch, realizes the switching by adjusting the coupling capacitance of the contact. The design contains (figure3.1b) an airbridge structure. The metallic contact is isolated by an airgap and a thin dielectric film. The gap is adjustable via electromechanical actuation in order to achieve a UP/DOWN or ON/OFF states. This design is commonly used in high frequency applications [64]. Compared with the commercial available switch e.g. p-i-n diode or FET switches, the MEMS switches have advantages of near zero power consumption, high isolation, low insertion loss, linear response and potential low cost. While problems such as relatively low switching speed, reliability, high actuation voltage packaging and etc. are still unsolved and hinder the real commercialization of the device [65]. Efforts have been made to investigate in the field of 52  electric contact metallurgies to address the contact wear and decay problem and improve the device reliability. R. Coutu et al. focused on the possible alloy combination selection [66]. Q. Ma et al. have evaluated the performances of the pure metals. Hard metals as Rhodium (Rh), Ruthenium (Ru) and Iridium (Ir) have excellent performances as switches in the presence of adequate contact force. But there is no simple method to pattern these materials [67]. M3EDM has been proved to have great capability to machine electrical conductive metals, e.g. metals, and potential low cost for batch production process [30]. Utilizing M3EDM in the MEMS switch fabrication may enable more material possibilities in batch produce process and potentially improve the device reliability.  Figure3.1: (a) Image of a toggle switch/ohmic contact (b) image of a shunt airbridge/capacitive coupling switch [64]  The following sections are organized as follows. The section 3.2 discusses the design of the movable electrodes for contact switch fabrication. In section 3.3, the fabrication of the contact  53  switch is discussed. Section 3.4 focuses on efforts to address the photoresist melting problem and the test result of the new fabrication method. Section 3.5 concludes the overall efforts.  3.2  Design  The octagon electrode with crab leg supports was chosen as the platform for the machining because it has larger electrode area with more stable and uniform discharge machining compared with other designs in the experiment. M3EDM devices were utilized to cut switch structures out of the foil. The cutting depth is important in this case. A rough estimate for single layer machining depth has been discussed in previous work. In the estimation, the maximum machining depth is at the turning point where the mechanical force exceeds the electrostatic force just before the discharge occurs. The equation is shown below [30]:  (3.1) g is discharging gap, assume a 10μm constant gap in order to simplify the discuss.  is the  projected maximum machining depth. However, the electrostatic force at the discharging point is a function of the machining depth before the supporting actuator surface reaches the discharging gap to the workpiece. If the machining depth is larger than the height of electroplated structure, the relationship is similar to the single layer case. The adjusted equations are shown below:  (3.2)  54  A1, d1, A2 and d2 are the electrostatic effective area and the original distance of the supporting structure and the custom design, respectively. To further simplify, two restricts 40μm (30μm electroplated pattern) and  μ  (15μm gap between the electroplated  pattern and the workpiece) are added. The values of typical double layer structure are substituted for evaluation (A1=~9mm2; A1/A2=4/1; K=~300N/m). The voltage is set to 150V. The solution of equation could be found through graphical solving. The stable solution for function is about 13μm. There is an unstable solution at 24μm for first part of equation. The solution suggests only the electroplated patterns will be transferred during the machining process. With decrease in the electroplated structure height, the difference between the two solutions is reduced. Once the height of electroplated pattern is smaller than a certain value, in this case around 27μm, there will be no stable solution for the first half of equation. The solution will be a value fulfilled the second half of equation, which means both patterns of the electroplated and patterns on the actuator foil structure will be transferred. This effect is not favored in the application, though it gives a higher machining depth. In order to achieve greater machining depth without unwanted actuator patterns transferring, it might be necessary to add a machining depth monitor feedback to stop machining before the actuator pattern begins to transfer. The design of the switch was chosen to be a cantilever-type structure as shown in figure 3.2. The structure of this design is simple and the actuation of the device is easier according to the discussion presented in chapter 2. The structure size is minimized in order to have a higher success rate and a more uniformly machining result. The cut depth is set to be ~10μm. The gap between the foil and supporting wafer is set to be 10μm for easy switching, i.e. pull-in.  55  Figure3.2: One example of M3EDM device designed for contact switch fabrication  3.3  Fabrication and Experimental Results  The fabrication process of cantilever-like contact switch is shown in figure 3.3. Since M3EDM has the machining capability to cut most conductive materials, a trial 10μm copper foil instead of a robust metal foil is utilized in the experiment for the process verification. To further simplify the process, the other side of contact is a stainless steel wafer. First, a 10μm SPR 220-7 photoresist is spun onto a pre-cleaned stainless steel wafer as a sacrificial layer. After coating, the sample is soft baked for 10min at 90oC in an oven and hard baked on a hotplate at 130oC for 20min in order to deplete the solvent in the structure. (step 1). A pre-etched copper foil (~10μm thick) is laminated onto sacrificial layer by S1813 polymer bonding to form the sandwich structure, followed by a 5min soft baking at 90oC in the oven. Then double layer M3EDM devices are fabricated onto the sandwich substrate. The fabrication process is the same as  56  illustrated in figure 2.19b and devices are processed in the setup shown in figure 2.20 for the cutting (step 2). The M3EDM devices will be stripped once contact switch structures are separated from the bulk foil completely (step 3). The final step is stripping the unwanted bulk foil and releasing the sacrificial layer. A critical point drying may be needed as additional step after the water cleaning.  Figure 3.3: The fabrication process of contact switch from photoresist sandwich structure  The M3EDM devices have been successfully fabricated on the proposed sandwich substrate. In the following machining process, however, the complete cutting of designed patterns was found to be difficult. The machining was supposed to continue until the patterns were transferred completely to the foil, i.e. totally cutting through. The photoresist in the sandwich structure was expected to remain stable during the machining process. But after the machining process stopped, 57  the foil was only partially cut through. Careful observation of the samples (figure 3.4) revealed that the electrode patterns were not only transferred to the copper foil but also to the photoresist with a depth of ~6μm. The pattern of electrode could be clearly observed on the back side of foil, as well.  Figure 3.4: Different views of the same location (shown in the rectangle framework) (a) the front side view of the foil after machining (b) the backside view of the foil after machining (c) the photoresist under the foil after machining (part of photoresist was damaged during stripping ) (d) the stylus scanning at the arrow direction shown in the (c)  Melting point of the SPR220-7 is reported at about 150oC [68]. A study conducted by B. Revaz et.al. have demonstrated the heat transfer of needle μEDM on a 50μm steel foil workpiece [51]. The backside temperatures of the foil have been measured locally in both spatial and time 58  domain by an array of thermocouples. The energy of discharge pulse is set to be 12μJ (in our case ~1μJ). The experiment shows up to 270oC peak on the back side foil temperature at the location 35μm from the discharging center in around 0.1ms after discharging. The heat created by discharge may exceed 150oC and melt the photoresist beneath the foil easily. The foil deformation may due to this photoresist melting. In this case, the machining depth becomes the sum of cutting depth and the deformed depth, which prevents the complete cutting through.  Figure 3.5: The reverse fabrication process for contact switch 59  One straightforward solution for the problem is to use material of high thermal stability as the sacrificial layer in the sandwich structure. This material must be an insulator. Polymer materials such as polyimide [69] and epoxy resins such as SU-8 [70] and spin-on-glass (SOG) [71] have high thermal stability, however striping these materials in the final structure releasing is challenging. Minimizing the photoresist deformation is another solution. A new reverse fabrication process is proposed in this concept. The process eliminates the thick photoresist under the foil in the sandwich structure. The details are shown in figure 3.5. The foil is laminated on the thin glass (~100μm) by 1μm S1813 bonding layer (step 1). The double layer devices are then fabricated on this substrate. In the following M3EDM cutting process, the sample is fixed onto the probe station platform to cut the contact switch structures out. (step 2) These structures are then bonded with the photoresist base structure (step 3). The glass substrate is totally etched away by HF wet bulk etching (step 4). The final step is to release the structure as presented previously (step 5). Since it only has 1μm photoresist under the foil and glass has high thermal stability during the discharge machining process, the deform problem is mostly solved. Further, by applying transparent substrate, it is easy to optically check the machining progress from the backside of the substrate without removing the devices above. The result showed an improvement of machining quality. The setup has successfully minimized the influence of the photoresist deformation under the structure during the discharge machining. One result is shown in figure 3.6. The device is stripped after the machining ends. The backside of the machined foil is illuminated from background to demonstrate the cutting through. The detailed images are taken by Nikon measuring microscope. 4 out of 5 designs on one electrode were transferred in the process. However, the individual pattern has not been totally cut out of the foil.  60  Figure 3.6: (a) Backside of machined area under front illumination (b) detailed images under front illumination (c) backside of machined area under background illumination (d) detailed image under background illumination  3.4  Discussion  The result suggests an improvement of the machining quality; however, the non-uniformity from other parts may still hinder the uniformly cutting through of the sample. Possible causes could be photoresist spin coating variance, bonding softening, variation in electroplating thickness and foil thickness non-uniformity. The first problem needs a more controllable and 61  repeatable fabrication environment which is not easily achieved under current conditions. The bonding softening could be reduced easily by eliminating the additional sacrificial layer on the copper foil or by a hard baking. The uniformity of electroplated structures can be improved by an additional lapping. By applying a foil substrate of greater uniformity, the last issue may be resolved. The problem could also be addressed by designing and applying actuators that have greater cutting depths.  Figure 3.7: The draft of reusable M3EDM device  The reverse fabrication concept could also be utilized in the fabrication of reusable M3EDM devices. The M3EDM device was directly fabricated on the substrate and one-time use only at this point. Some devices are actually still usable after the machining. If they could be relocated without damage, these devices might be reused. The reverse fabrication concept may potentially enable the fabrication of reusable M3EDM devices. The draft design is shown in figure 3.7. The metal foil is first laminated on the glass base and patterned by wet etching with a photoresist as mask. A thick photoresist layer is coated as the anchor and the electroplating mold. The glass is then patterned in order to form holes for electrical contact. After patterning the thick photoresist for the first time, the sample is electroplated with copper that forms a custom designed pattern  62  and electrical link for each actuator. Photoresist can be lapped for a more uniform thickness. Then the glass is patterned for the second time to provide the space for the bubbles and byproducts releasing. The last step includes the releasing of the structure by exposing and developing the photoresist. The adhesion of photoresist to the glass and the structural strength of glass may hinder this process. If these problems could be solved properly, the device combined with the parallel pulse generator may be potentially applicable in multiuser standard processes such as PolyMUMPs for custom mold fabrication and enhance commercialization of MEMS.  3.5  Conclusion  Applying M3EDM in MEMS contact switch fabrication could enable more material options for the switch structure. This may address the reliability issue that hinders further applications of MEMS contact switches. A preliminary experimental investigation for potential application of M3EDM to MEMS contact switches has been performed. By introducing sandwich foil substrate, the M3EDM device has been utilized to fabricate contact switch. The problem of photoresist melting and foil deformation has been observed. A new reverse fabrication process based on glass has been proposed as a solution for the problem. The problem was partially solved, however, the cutting through was still not achieved. Further solutions for this issue have been discussed. Refining of the new fabrication concept may lead to a reusable M3EDM device, which will potentially enable a new standard multi user machining service as PolyMUMPs.  63  Chapter 4 Conclusion and Future Work The aim of the research in this thesis is to enhance the machining quality and uniformity of the M3EDM in an effort to make application of the technology feasible. In chapter 2, an optimized photoresist mold fabrication process that yields a better profile was developed. The contact gap existing in the previous process was eliminated by a new fixing method and removal of the edge bead. A rehydration step was added to reduce the exposure and developing time. These improvements produced a structure with near vertical sidewalls. The copper deformation problem was investigated and discussed in conjunction with thermal shock damage, tool wear, material softening, and discharge created reaction force as potential causes of the issue. A feedback control was designed in effort to prevent harmful thermal shock caused by the short pulse. The circuit successfully detected the short pulse by monitoring the low pass filtered current signal and prevented the damage by automatically shutting off the actuation DC voltage. A material that is suitable for batch fabrication with higher thermal and mechanical resistance than copper was investigated. The fabrication process utilizing the new candidate, nickel, was investigated. Both single layer and double layer configuration devices were successfully fabricated with a new bubble etching bath and tested in the probe station with the feedback control circuit. The research achieved a new setup device and enabled possible applications of M3EDM.  64  In chapter 3, the optimized M3EDM device was applied towards the cantilever contact switch fabrication. A fabrication process that combines photoresist sandwich structure with M3EDM devices was proposed and tested. Photoresist melting and foil deformation issues were observed. A reverse fabrication process was proposed to address these problems. The method reduced softening of the photoresist, however, the completely cutting through was not reached potentially due to the structure non-uniformity caused by spin coating variance, bonding softening, electroplating non-uniformity and foil thickness variance. Possible solutions and directions for these issues were discussed. A new reusable M3EDM based on the reverse fabrication was proposed with draft process flow. The future work would involve further improvement of device reliability and machining uniformity. The further investigation of the M3EDM applications and development of the reusable M3EDM device should be included in the future work.  65  References [1] M. D. T. Pham, S. S. Dimov, S. Bigot, A. Ivanov, and K. Popov, “Micro-EDM--recent developments and research issues,” Journal of Materials Processing Technology, vol. 149, no. 1, Jun. 2004, pp. 50-57. [2] M. J. Madou, Fundamentals of microfabrication: the science of miniaturization. CRC Press, 2002, pp. 398-401. [3] K. Takahata, N. Shibaike, and H. Guckel, “High-aspect-ratio WC-Co microstructure produced by the combination of LIGA and micro-EDM,” Microsystem Technologies, vol. 6, no. 5, 2000, pp. 175-178. [4] W. Khalid, M.A. Mohamed Sultan, M. Dahmaredeh, Y.Choi, P. Yaghoobi, A. Nojeh and K. Takahata, “High-aspect-ratio, free-form patterning of carbon nanotube forests using micro-electro-discharge machining,” Diamond and Related Materials, vol. 19, no. 11, Nov. 2010, pp. 1405-1410. [5] K. Takahata, Y. Gianchandani, and K. Wise, “Micromachined Antenna Stents and Cuffs for Monitoring Intraluminal Pressure and Flow,” Microelectromechanical Systems, Journal of, vol. 15, no. 5, 2006, pp. 1289-1298. [6] K. Takahata and Y. B. Gianchandani, “A Micromachined Capacitive Pressure Sensor Using a CavityLess Structure with Bulk-Metal/Elastomer Layers and Its Wireless Telemetry Application,” Sensors, vol. 8, no. 4, 2008, pp. 2317-2330. [7] K. Takahata, Micro Electronic and Mechanical systems, InTech, Dec 2009, pp. 144-147. [8] S. H. Yeo, W. Kurnia, and P. C. Tan, “Electro-thermal modelling of anode and cathode in microEDM,” Journal of Physics D: Applied Physics, vol. 40, no. 8, 2007, pp. 2513-2521. [9] M. Kunieda, B. Lauwers, K. Rajurkar, and B. Schumacher, “Advancing EDM through Fundamental Insight into the Process,” CIRP Annals - Manufacturing Technology, vol. 54, no. 2, 2005, pp. 64-87 66  [10] F. Han, S. Wachi, and M. Kunieda, “Improvement of machining characteristics of micro-EDM using transistor type isopulse generator and servo feed control,” Precision Engineering, vol. 28, no. 4, Oct. 2004, pp. 378-385. [11] F. Han, L. Chen, D. Yu, and X. Zhou, “Basic study on pulse generator for micro-EDM,” The International Journal of Advanced Manufacturing Technology, vol. 33, no. 5, Jun. 2007, pp. 474-479. [12] H. Lim, Y. Wong, M. Rahman, and M. Edwin Lee, “A study on the machining of high-aspect ratio micro-structures using micro-EDM,” Journal of Materials Processing Technology, vol. 140, no. 1, Sep. 2003, pp. 318-325. [13] S. H. Yeo and L. K. Tan, “Effects of ultrasonic vibrations in micro electro-discharge machining of microholes,” Journal of Micromechanics and Microengineering, vol. 9, no. 4, 1999, pp. 345-352. [14] C. Gao and Z. Liu, “A study of ultrasonically aided micro-electrical-discharge machining by the application of workpiece vibration,” Journal of Materials Processing Technology, vol. 139, no. 1, Aug. 2003, pp. 226-228. [15] M. P. Jahan, M. Rahman, and Y. S. Wong, “Modelling and experimental investigation on the effect of nanopowder-mixed dielectric in micro-electrodischarge machining of tungsten carbide,” Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, vol. 224, no. 11, Nov. 2010, pp. 1725-1739. [16] D. K. Chung, B. H. Kim, and C. N. Chu, “Micro electrical discharge milling using deionized water as a dielectric fluid,” Journal of Micromechanics and Microengineering, vol. 17, no. 5, 2007, pp. 867-874. [17] C. Lin, H. Chow, L. Yang, and Y. Chen, “Feasibility study of micro-slit EDM machining using pure water,” The International Journal of Advanced Manufacturing Technology, vol. 34, no. 1, 2007, pp. 104110.  67  [18] K. Takahata, S. Aoki, and T. Sato, “Fine surface finishing method for 3-dimensional micro structures”, JEICE Trans. Electronics, vol. E80-C, no. 2, 1997, pp. 291-296. [19] T. Masuzawa, M. Fujino, K. Kobayashi, and T. Suzuki, “Wire electro-discharge grinding for micromachining”, Annals of the CIRP, vol. 34, 1985, pp. 431-434. [20] G. Cusanelli, M. Minello, F. Torchia, W. Ammann, and P.-E. Grize,“Properties of micro-holes for nozzle by micro-EDM,” in Proc.15th Int. Symp. Electromach. (ISEM XV), Pittsburgh, PA, 2007, pp. 241– 245. [21] K. Takahata and Y. Gianchandani, “Bulk-Metal-Based Mems Fabricated by Micro-ElectroDischarge Machining,” in Electrical and Computer Engineering, 2007. CCECE 2007. Canadian Conference on, 2007, pp. 1-4. [22] H. H. Langen, T. Masuzawa, and M. Fujino, “Modular method for microparts machining and assembly with self-alignment”, Annals of the CIRP, vol. 44, 1995, pp. 173-176. [23] C. Stokes and P. Palmer, “3D Micro-fabrication Processes: A Review,” in MEMS Sensors and Actuators, 2006. The Institution of Engineering and Technology Seminar on, 2006, pp. 289-298. [24] K. Takahata and Y. B. Gianchandani, “Batch mode micro-electro-dischargemachining”, IEEE/ASME Journal of Microelectromechanical Systems, vol. 11, 2002, pp. 102-110. [25] 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,” Microsystem Technologies, vol. 14, no. 3, 2007, pp. 411-417. [26] K. Takahata, N. Shibaike, and H. Guckel, “A novel micro electro-discharge machining method using electrodes fabricated by the LIGA process,” in Micro Electro Mechanical Systems, 1999. MEMS '99. Twelfth IEEE International Conference on, 1999, pp. 238-243.  68  [27] C. Chaitanya, MEMS-Enabled Micro-Electro-Discharging Machining (M3EDM). Master Thesis The University of British Columbia, 2008. [28] C. Alla Chaitanya and K. Takahata, “Micro-electro-discharge machining by MEMS actuators with planar electrodes microfabricated on the work surfaces,” in Micro Electro Mechanical Systems, 2008. MEMS 2008. IEEE 21st International Conference on, 2008, pp. 375-378. [29] C. R. A. Chaitanya and K. Takahata, “M3EDM: MEMS-enabled micro-electro-discharge machining,” Journal of Micromechanics and Microengineering, vol. 18, no. 10, 2008, p. 105009(7pp). [30] C. Chaitanya, N. Wang, and K. Takahata, “MEMS-BASED MICRO-ELECTRO-DISCAHGE MACHINING (M3EDM) BY ELECTROSTATIC ACUTATION OF MACHIINING ELECTRODES ON THE WORKPIECE”, IEEE/ASME Journal of Microelectromechanical Systems, vol. 19, no. 3, 2010, pp. 690-699. [31] E. Koukharenko, M. Kraft, G. Ensell, and N. Hollinshead, “A comparative study of different thick photoresists for MEMS applications,” Journal of Materials Science: Materials in Electronics, vol. 16, no. 11, Nov. 2005, pp. 741-747. [32] C. Pan, H. Yang, S. Shen, M. Chou, and H. Chou, “A low-temperature wafer bonding technique using patternable materials,” Journal of Micromechanics and Microengineering, vol. 12, no. 5, 2002, pp. 611-615. [33] C. Lin, G. Lee, Y. Lin, and G. Chang, “A fast prototyping process for fabrication of microfluidic systems on soda-lime glass,” Journal of Micromechanics and Microengineering, vol. 11, Nov. 2001, pp. 726-732 [34] S. Roth, L. Dellmann, G. Racine, and N. F. D. Rooij, “High aspect ratio UV photolithography for electroplated structures,” Journal of Micromechanics and Microengineering, vol. 9, no. 2, 1999, pp. 105108 69  [35] E. Kukharenka and M. Kraft, “Realization of electroplating molds with thick positive SPR 220-7 photoresist,” Journal of Materials Science: Materials in Electronics, vol. 14, no. 5, May. 2003, pp. 319322 [36] O. P. Lehar, M. A. Spak, S. Meyer, R. R. Dammel, C. J. Brodsky, and C. G. Willson, “Resist rehydration during thick film processing,” in Advances in Resist Technology and Processing XVIII, vol. 4345, 2001, pp. 463-474 [37] T. Yada, T. Maejima, M. Aoki, and M. Umesaki, “Thin-Film Formation by Spin Coating: Characteristics of a Positive Photoresist,” Japanese Journal of Applied Physics, vol. 34, 1995, pp. 62796284 [38] T. Yada, M. Aoki, T. Maejima, and A. Ishizu, “Formation of a Positive Photoresist Thin Film by Spin Coating: Influence of Atmospheric Temperature,” Japanese Journal of Applied Physics, vol. 36, 1997, pp. 372-377. [39] Y. Yang and T. Chang, “Experimental analysis and optimization of a photo resist coating process for photolithography in wafer fabrication,” Microelectronics Journal, vol. 37, no. 8, Aug. 2006, pp. 746-751. [40] C. Lin, G. Lee, B. Chang, and G. Chang, “A new fabrication process for ultra-thick microfluidic microstructures utilizing SU-8 photoresist,” Journal of Micromechanics and Microengineering, vol. 12, no. 5, 2002, pp. 590-597. [41] A. E. Kozlowski, “Process optimization of single-coat positive photoresist for thick film applications,” in Microlithography and Metrology in Micromachining, vol. 2640, 1995, pp. 121-131 [42] Y. S. Tarng, C. M. Tseng, and L. K. Chung, “A fuzzy pulse discriminating system for electrical discharge machining,” International Journal of Machine Tools and Manufacture, vol. 37, no. 4, Apr. 1997, pp. 511-522.  70  [43] S. H. Yeo, E. Aligiri, P. C. Tan, and H. Zarepour, “A New Pulse Discriminating System for MicroEDM,” Materials and Manufacturing Processes, vol. 24, no. 12, 2009, pp. 1297-1305. [44] A. Broue et al., “Validation of bending tests by nanoindentation for micro-contact analysis of MEMS switches,” Journal of Micromechanics and Microengineering, vol. 20, no. 8, 2010, p. 085025(8pp). [45] A. Senouci, H. Zaidi, J. Frene, A. Bouchoucha, and D. Paulmier, “Damage of surfaces in sliding electrical contact copper/steel,” Applied Surface Science, vol. 144, Apr. 1999, pp. 287-291. [46] H. D. Merchant, “Thermal effects in thin copper foil,” Journal of Electronic Materials, vol. 33, no. 1, 2004, pp. 83-88. [47] S. Tomura and M. Kunieda, “Analysis of electromagnetic force in wire-EDM,” Precision Engineering, vol. 33, no. 3, Jul. 2009, pp. 255-262. [48] C. Dai, X. Zhu, and G. Zhang, “Tensile and Fatigue Properties of Free-Standing Cu Foils,” Journal of materials science & technology, vol. 25, no. 06, 2009, pp. 721-726. [49] T. Masaki, K. Kawata and T. Masuzawa, “Micro electro-discharge machining and its applications,” Proc. IEEE Conf on Micro Electro Mechanical Systems, 1990, pp 21-26. [50] F. Han, J. Jiang, and D. Yu, “Influence of discharge current on machined surfaces by thermo-analysis in finish cut of WEDM,” International Journal of Machine Tools and Manufacture, vol. 47, no. 7, Jun. 2007, pp. 1187-1196 [51] B. Revaz, G. Witz, and R. Fl kiger, “Properties of the plasma channel in liquid discharges inferred from cathode local temperature measurements,” Journal of Applied Physics, vol. 98, no. 11, 2005, p. 113305(6pp).  71  [52] T. Mariko, K. Toshinao, and K. Masanori, “Measurement of Process Reaction Force in EDM Using Hopkinson Bar Method”, Journal of the Japan Society for Precision Engineering, vol. 68, no. 6, 2002, pp. 822-826. [53] Y. Tsai and T. Masuzawa, “An index to evaluate the wear resistance of the electrode in micro-EDM,” Journal of Materials Processing Technology, vol. 149, no. 1, Jun. 2004, pp. 304-309 [54] M. Braunović, V. V. Konchits, N. K. Myshkin, and V. V. Konchits, Electrical contacts: fundamentals, applications and technology. CRC Press, 2006, pp. 72. [55] N. Tantra, F. Leao, and I. Pashby, “Evaluating theoretical equations to predict wear in electrodischarge machining,” in Proceedings of the First International Conference and Seventh AUN/SEED-Net Fieldwise Seminar on Manufacturing and Material Processing, 2006, pp. 151-155. [56] G. M. Rebeiz, RF MEMS: theory, design, and technology. John Wiley and Sons, 2003. [57] S. Pamidighantam, R. Puers, K. Baert, and H. A. C. Tilmans, “Pull-in voltage analysis of electrostatically actuated beam structures with fixed–fixed and fixed–free end conditions,” J. Micromech. Microeng., vol. 12, no. 4, Jul. 2002, pp. 458–464. [58] Y. Nemirovsky and O. Bochobza-Degani, “A methodology and model for the pull-in parameters of electrostatic actuators,” Microelectromechanical Systems, Journal of, vol. 10, no. 4, 2001, pp. 601-615. [59] P. Walker and W. H. Tarn, CRC handbook of metal etchants. CRC Press, 1991, pp. 857-861. [60]  Transene  Company,  Inc.,  “Nickel  Etchants  for  Thin  Film  Circuits.”  Internet:  http://www.transene.com/ni_etchant.html, [Nov 22, 2010] [61] D. M. Allen and H. J. A. Almond, “Characterisation of aqueous ferric chloride etchants used in industrial photochemical machining,” Journal of Materials Processing Technology, vol. 149, no. 1, Jun. 2004, pp. 238-245.  72  [62] T. Mooney, “Solution Agitation and Mixing,” Metal Finishing,, vol. 97, no. 1, 1999, pp. 747-754. [63] I. Schiele and B. Hillerich, “Comparison of lateral and vertical switches for application as microrelays,” Journal of Micromechanics and Microengineering, vol. 9, no. 2, 1999, pp. 146-150. [64] B. Schauwecker, K. Strohm, T. Mack, W. Simon, and J. Luy, “Serial combination of ohmic and capacitive RF MEMS switches for large broadband applications,” Electronics Letters, vol. 40, no. 1, 2004, pp. 44-46. [65] G. Rebeiz and J. Muldavin, “RF MEMS switches and switch circuits,” Microwave Magazine, IEEE, vol. 2, no. 4, 2001, pp. 59-71. [66] R. A. Coutu, P. E. Kladitis, K. D. Leedy, and R. L. Crane, “Selecting metal alloy electric contact materials for MEMS switches,” Journal of Micromechanics and Microengineering, vol. 14, no. 8, 2004, pp. 1157-1164. [67] Q. Ma, Q. Tran, T. Chou, J. Heck, H. Bar, R. Kant, and V.Rao, “Metal contact reliability of RF MEMS switches,” in Reliability, Packaging, Testing, and Characterization of MEMS/MOEMS VI, vol. 6463, 2007, pp. 646305.1-646305.13. [68] E. Roy, B. Voisin, J. Gravel, R. Peytavi, D. Boudreau, and T. Veres, “Microlens array fabrication by enhanced thermal reflow process: Towards efficient collection of fluorescence light from microarrays,” Microelectronic Engineering, vol. 86, no. 11, Nov. 2009, pp. 2255-2261. [69] D. N. Khanna and W. H. Mueller, “New high temperature stable positive photoresists based on hydroxy polyimides and polyamides containing the hexafluoroisopropylidene (6-f) linking group,” Polymer Engineering and Science, vol. 29, no. 14, 1989, pp. 954-959. [70] A. D. Campo and C. Greiner, “SU-8: a photoresist for high-aspect-ratio and 3D submicron lithography,” Journal of Micromechanics and Microengineering, vol. 17, no. 6, 2007, pp. R81-R95.  73  [71] A. A. Hamzah, B. Y. Majlis, and I. Ahmad, “Formation of thick spin-on glass (SOG) sacrificial layer for capacitive accelerometer encapsulation,” in Device and Process Technologies for Microelectronics, MEMS, and Photonics IV, vol. 6037, 2005, pp. 227-237.  74  

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:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0071595/manifest

Comment

Related Items