Citation
Precision Molding of Metallic Microcomponents

Material Information

Title:
Precision Molding of Metallic Microcomponents
Creator:
BARDT, JEFFREY A. ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Aspect ratio ( jstor )
Crystallization ( jstor )
Etching ( jstor )
Geometry ( jstor )
Glass transition temperature ( jstor )
Heating ( jstor )
Metallic glasses ( jstor )
Molding ( jstor )
Platens ( jstor )
Silicon ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Jeffrey A. Bardt. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
8/31/2006
Resource Identifier:
495636961 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ETIWHEJK0_VXXIGQ INGEST_TIME 2017-08-09T19:14:47Z PACKAGE UFE0011822_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

PRECISION MOLDING OF METALLIC MICROCOMPONENTS By JEFFREY A. BARDT A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Jeffrey A Bardt

PAGE 3

ACKNOWLEDGMENTS I would like to thank my parents, David and Nancy Bardt, for all their support over my entire life. They always believed in me and made me believe in myself. I would like to thank my love, Jessica Reyes, for being there for me when I needed her the most. Special thanks go to my advisor, Dr. Greg Sawyer, for the advice and guidance throughout my graduate studies. Also, thanks go to the other members of my committee, Dr. John Ziegert and Dr. Tony Schmitz, for their support and involvement with this project. Special thanks go to Gerald Bourne for his materials knowledge and for his help with the Major Analytical Instrumentation Center equipment. Thanks go to Nathan Mauntler for his help throughout this project. I thank all the members of the Tribology Lab for their advice and support. I thank Dr. Mark Sheplak along with two of his students, Yawei Li and Stephen Horowitz, for their help with the development of the silicon wafer molds. iii

PAGE 4

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iii LIST OF TABLES .............................................................................................................vi LIST OF FIGURES ..........................................................................................................vii ABSTRACT .........................................................................................................................x CHAPTER 1 INTRODUCTION........................................................................................................1 2 BACKGROUND..........................................................................................................5 Bulk Metallic Glass......................................................................................................5 Applications..................................................................................................................7 Micro-Electro-Mechanical-Systems......................................................................7 Textured Surfaces..................................................................................................8 Fuel Cell Interconnects..........................................................................................9 3 LITERATURE REVIEW...........................................................................................10 Fabrication Techniques...............................................................................................10 Bulk Metallic Glass and Processes.............................................................................11 4 EQUIPMENT DESIGN/CONSTRUCTION..............................................................15 Molding Components.................................................................................................15 Heating Process..........................................................................................................18 Cooling Process..........................................................................................................19 Final Design................................................................................................................21 5 UNCERTAINTY ANALYSIS...................................................................................24 Temperature................................................................................................................24 Force...........................................................................................................................27 6 SILICON WAFER MASTER MOLD........................................................................29 iv

PAGE 5

Design.........................................................................................................................29 Deep Reactive Ion Etching.........................................................................................31 7 EXPERIMENTAL PROCESS...................................................................................37 Procedure....................................................................................................................37 Conditions...................................................................................................................41 8 RESULTS AND DISCUSSION.................................................................................44 Aspect Ratio, Geometry, and Proximity.....................................................................44 Temperature, Applied Pressure, and Molding Time..................................................47 Additional Flow Characteristics.................................................................................53 9 CONCLUSIONS........................................................................................................60 APPENDIX A THEORETICAL CALCULATIONS FOR PLATEN DESIGN................................62 B ADDITIONAL AGP IMAGE RESULTS..................................................................64 C TWO DIMENSIONAL CHANNEL GEOMETRIES................................................68 D GRID PATTERN FOR TRIBOLOGICAL APPLICATIONS...................................72 LIST OF REFERENCES...................................................................................................77 BIOGRAPHICAL SKETCH.............................................................................................80 v

PAGE 6

LIST OF TABLES Table page 1-1 Fabrication techniques..................................................................................................1 2-1 Material properties of Vitreloy-1 at room temperature................................................6 5-1 Temperature components with uncertainty in measurement......................................25 5-2 Uncertainties for the four molding temperatures........................................................26 5-3 Uncertainty values for force measurement.................................................................28 5-4 Uncertainty values for the four applied forces...........................................................28 7-1 Experimental test matrix for AGP pattern silicon wafer master................................43 vi

PAGE 7

LIST OF FIGURES Figure page 2-1 Atomic structure of crystalline material versus amorphous material...........................5 4-1 Components of molding apparatus.............................................................................15 4-2 Molding components secured to MTS machine.........................................................17 4-3 Experimental heating profile for stainless steel platen...............................................19 4-4 Cooling water path......................................................................................................20 4-5 Experimental cooling time response of stainless steel platen.....................................21 4-6 Final schematic of molding setup stack......................................................................22 4-7 Experimental time response for complete heating and cooling process.....................22 4-8 Equipped MTS machine.............................................................................................23 5-1 Temperature versus molding time of mold sleeve......................................................26 6-1 Individual and combinatorial parameters for AGP silicon wafer mold......................30 6-2 AGP 5 mm x 5 mm grid layout..................................................................................31 6-3 4” silicon wafer master layout....................................................................................31 6-4 Optical microscope picture of AGP silicon wafer master..........................................35 6-5 SEM micrographs of four sections of AGP pattern silicon wafer master..................36 7-1 Centorr arc-melter furnace used for remelting bulk metallic glass material..............37 7-2 Recasted bulk metallic glass in copper square pin mold............................................38 7-3 8” South-Bay Technologies polishing wheel and South-Bay diamond saw..............38 7-4 Sonicator used for acetone and methanol baths..........................................................39 7-5 VEECO white light interferometer.............................................................................40 vii

PAGE 8

7-6 FEI focused ion beam with scanning electron microscope capabilities.....................41 8-1 White light interferometer scan of a circle and square feature. Both had heights near 11 m..................................................................................................................45 8-2 Two-dimensional height values for 10 mm diameter features with increasing spatial distance away from the center.........................................................................46 8-4 Viscosity versus temperature for Vitreloy-1 with a heating rate of 20 K/s................47 8-5 10 m features molded at 115 MPa for 15 seconds...................................................48 8-6 Aspect ratio plotted against molding time for the four pressures...............................50 8-7 Aspect ratio versus maximum temperature for 400 C and 425 C............................51 8-8 X-ray diffraction patterns for the four samples seen in Figure 8-5............................52 8-9 TEM micrograph of part crystalline, part amorphous section of 50 m feature........53 8-10 Illustration of scallops produced from deep reactive ion etching.............................54 8-11 10 m feature cross section in silicon wafer mold...................................................54 8-12 Reproduced scallops or cusps along the sidewall of the molded feature.................55 8-13 Reproduced scallops along entire sidewall of channel geometry.............................56 8-14 Footing created on one side of silicon wafer during deep reactive ion etching process......................................................................................................................56 8-15 Illustration of side etching which creates footing in the deep reactive ion etching process......................................................................................................................57 8-16 Filled footing in molded bulk metallic glass............................................................58 8-17 Additional SEM micrographs of AGP results..........................................................59 A-1 Time versus temperature of various materials under consideration..........................62 A-2 Time versus temperature profiles. A) in air, B) in aluminum block.........................63 B-1 SEM micrographs of molded bulk metallic glass with little bulk flow.....................65 B-2 SEM micrographs of molded bulk metallic glass......................................................67 C-1 Eight channel geometry designs................................................................................68 C-2 SEM micrographs of four channel geometries..........................................................70 viii

PAGE 9

C-3 Angled SEM micrographs of channel geometries.....................................................71 C-4 Scalloped wall on molded channel geometry............................................................71 D-1 Grid pattern design for solid lubricants.....................................................................72 D-2 SEM micrographs of molded bulk metallic glass grid pattern..................................73 D-3 Optical microscope images of solid lubricants in molded metallic glass grid..........74 D-4 Tribometer used for pin-on-disk composite tests......................................................75 D-5 Friction coefficient plot over sliding distance...........................................................75 D-6 Optical microscope image of wear track in PTFE-metallic glass composite............76 ix

PAGE 10

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science PRECISION MOLDING OF METALLIC MICROCOMPONENTS By Jeffrey A. Bardt August 2005 Chair: W. Gregory Sawyer Major Department: Mechanical and Aerospace Engineering The steady trend towards the miniaturization of mechanical devices and systems has been restricted by the limitations of the manufacturing technology and available materials. Alternate fabrication technologies suitable to these complex component geometries and materials must be developed. Bulk metallic glasses, also known as amorphous metals, enables the advances of micro-devices. Bulk amorphous metals are high strength and high precision which make them ideal for forming micro-components. Also, since no phase change occurs, there is relatively little shrinkage, attributed only to thermal expansion and contraction, as the material cools below its working temperature. This allows for exceptional tolerance control of molded features. Bulk amorphous metal alloys develop good surface finishes because of the lack of crystallinity, which is important because secondary-finishing operations are limited at the micron-level. The micro-molding of bulk amorphous metals to create surface features from submicrometer to fractions of a millimeter was investigated. The goal was to x

PAGE 11

demonstrate that it is possible to produce such features in a metallic material from a master. The bulk metallic glass material was embossed at several different temperatures in the “supercooled liquid” region, which is the temperature regime between the glass transition temperature and crystallization temperature. In addition, several different pressures and molding times were selected for experimentation. A tension/compression MTS load frame was outfitted with custom platens to serve as the molding apparatus. Cartridge heaters in both platens allow for the necessary critical heating rates to avoid crystallization. A pump flowing water through channels in the platens rapidly cools the sample below the glass transition temperature. Silicon wafers patterned by deep reactive ion etching were used as the masters. The patterns were designed to test the effects and interactions of aspect ratios, geometry, and spatial proximity. The sensitivity of the moldability of the amorphous alloy to these important process variables was examined. Results showed excellent moldability of the bulk metallic glass. Geometry and proximity did not show strong trends, suggesting that they do not play major roles in the micromolding process. Over the ranges of temperature, pressure, and molding times, temperature was the most dominant process variable as the viscosity is so dependent on temperature and changes largely with small fluctuations in temperature. The aspect ratio which defines how well the material flowed showed inconsistent variations over the applied pressures and molding times. Expected results of higher aspect ratios for higher pressures and longer times were not always observed suggesting that localized temperature was the main concern. Aspect ratios were observed as high as 13.5 to 1. xi

PAGE 12

CHAPTER 1 INTRODUCTION There has been a steady trend towards the miniaturization of mechanical devices and systems, however, it has been severely restricted by two key aspects. First, there exist limitations on the manufacturing technology to produce these miniaturized systems. Second, the available materials for these devices are extremely limited due to the material properties necessary for forming micro-devices. The types of components as well as the range of geometries that can be produced are exceptionally narrow. Alternate fabrication technologies capable of manufacturing these complex micro-scale component geometries and materials must be developed in order to realize these systems. Succeeding in the development of this fabrication technology will enable new classes of miniaturized mechanical devices and systems. A list of some current fabrication techniques along with their capabilities for creating micrometer sized geometries can be seen in Table 1-1. Table 1-1. Fabrication techniques Technology/ Feature Geometry Minimum feature size/Feature tolerance Feature positional tolerance Material removal rate Materials Focused Ion Beam/ 2D & 3D 200 nanometers/ 20 nanometers 100 nanometers 0.5 cubic microns/sec Any Micro-milling or micro-turning/2D or 3D 25 microns/ 2 microns 3 microns 10,400 cubic microns/sec PMMA, Aluminum, Brass, Mild Steel 1

PAGE 13

2 Table 1-1. Continued Technology/ Feature Geometry Minimum feature size/Feature tolerance Feature positional tolerance Material removal rate Materials Excimer laser/2D or 3D 6 microns/ submicron Submicron 40,000 cubic microns/sec Polymers, ceramics, and metals to a lesser degree Femto-second laser/ 2D or 3D 1 micron/ Submicron Submicron 13,000 cubic microns/sec Any Micro EDM (sinker or wire)/2D or 3D 25 microns/ 3 microns 3 microns 25 million cubic microns/sec Conductive materials LIGA / 2D Submicron/0.02 mm ~ 0.5 mm ~0.3 mm nom. across 3” N/A Electroformable: copper, nickel, permalloy [1] Note: LIGA can also be used to fabricate parts in polymers, pressed powders, ceramics, and rare-earth magnets with little degradation in machining performance specifications. Advantages and disadvantages exist for each of the fabrication techniques. As can be seen in the table, minimum feature sizes can range from 200 nanometers with a Focused Ion Beam to 25 micrometers with a Micro-machining or Micro-EDM process. Another important thing to note is the material removal rate which can vary orders of magnitude. It ranges from as low as 0.5 cubic microns per second up to 25 million cubic microns per second. The material that one could use also depends on which fabrication technique is chosen based on what the final application is. One class of materials that enables further advancement of micro-devices is amorphous metals, or bulk metallic glasses. The unique combination of properties make bulk amorphous metals ideal for forming high-strength, high precision micro

PAGE 14

3 components. Several of these alloys have strength properties which equal or exceed current high-strength crystalline alloys. The alloy commercially known as Vitreloy-1, for example, has a room-temperature yield strength of 1.90 GPa and a Young’s modulus of 96 GPa [2]. Unlike crystalline metals, however, metallic glass can be molded to produce micron sized or smaller features at temperatures around and above 400 C. This low glass transition temperature allows molds to be made of conventional materials such as tool steel, or even aluminum or copper depending on molding pressure. It also suggests that conventional thermoplastic molding equipment may be able to be modified for this application, resulting in low capital costs for manufacturers. The unique atomic structure leads to a set of characteristic properties for some amorphous metals which include: very high yield strength, high hardness, superior strength/weight ratio, superior elastic limit, and high wear resistance. Since no phase change occurs, there is relatively little shrinkage as the material cools below the glass transition temperature. This will allow for exceptional tolerance control of cast and molded features. Also due to lack of crystallinity, bulk amorphous metal alloys tend to develop extremely good surface finish upon vitrification, which is important because options for secondary-finishing operations are extremely limited. Control of the viscosity of the material via temperature in its “supercooled liquid” state will allow additional flexibility in determining optimal processing parameters when molding fine features with high aspect ratios, where the high surface area to volume ratios lead to high heat transfer rates. The components that result will have exceptional mechanical strength properties, very high flexibility, good fracture toughness and fatigue strength, and will be electrically conductive.

PAGE 15

4 Applications for this technology can easily be seen in micro-electro-mechanical-systems (MEMS). But in addition to this area, one could imagine much more. There exist possibilities in fields such as Tribology, which studies interacting surfaces in relative motion as well as associated matters such as friction, wear, and lubrication. Textured surfaces can act as storage for solid lubricant delivery to help reduce wear or friction between two surfaces. Another application deals with miniaturizing fuel cells to utilize their energy potential in areas besides automobiles and other forms of transportation. The ability to stack fuel cells and increase their energy output and efficiency is a key research area. Bulk metallic glasses can improve fuel cells by molding the interconnect plates which is responsible for delivering the hydrogen and oxygen components of the reaction with micrometer sized channels. The major goals of this micro-molding process are 1) rapidly produce surfaces with engineering features; 2) create feature sizes on the micron-scale, possibly even sub-micron; 3) form features with aspect ratios of at least 5:1; aspect ratio is defined as the height divided by the diameter; 4) remain a relatively inexpensive and low capital cost process; 5) mold with a high strength, high precision material.

PAGE 16

CHAPTER 2 BACKGROUND Bulk Metallic Glass Bulk metallic glasses have a set of unique properties that make them a compelling option for many process or applications. Historically, metallic glasses could only be developed by techniques such as melt spinning or splat cooling due to the extremely high cooling rates needed to maintain the amorphous structure of the material. Researchers in the early 1990s worked on alloying elements to create bulk metallic glasses that had deep eutectic points on their phase diagrams. The suppressed melting temperature along with the kinetics involved in atomic rearrangement impedes crystallization as the alloy solidifies. This allows for more manageable cooling rates of the material so the amorphous structure can be utilized. One of the major defining characteristics of these materials is that they have an amorphous structure. This lack of a crystallized structure leads to many ideal characteristics for micro-molding. An example of an amorphous structure compared to a crystallized structure can be seen in Figure 2-1. Figure 2-1. Atomic structure of crystalline material versus amorphous material 5

PAGE 17

6 There exists many classes of bulk metallic glasses; the specific metallic glass used for these micro-molding experiments is Zr 41.2 Ti 13.8 Cu 12.5 Ni 10 Be 22.5 , also known commercially as Vitreloy-1. This material has a high-yield strength, around 1.9 GPa, which is higher than many steels, and a high hardness which makes it a good fit for molding complex micro-components. This material also exhibits good elastic limits and a good strength to weight ratio. Some key properties of Vitreloy-1 can be seen in Table 2-1. The elastic modulus is reported around 96 GPa, similar to aluminum. The stress-strain curve starts comparable to aluminum but exceeds it in orders of magnitude until failure. This allows for large stresses without much plastic deformation in the material. When the material fails, it fails catastrophically, however, if stresses are maintained below the yield strength, the high strength and low modulus mechanical properties can be utilized in secondary operations or conditions to the material. Table 2-1. Material properties of Vitreloy-1 at room temperature Property Value Density 6,000 kg/m 3 Young’s Modulus 96 GPa Poisson’s Ratio 0.36 Elastic Strain Limit 0.02 Tensile Yield Strength 1.90 GPa Vickers Hardness 534 kg/mm 2 [2] The current commercial uses of bulk metallic glasses vary over a wide range of uses. They can primarily be seen commercially in sporting goods equipment, including golf clubs, tennis racquets, and skis. But they also have many other applications that include casings for electronic devices such as cell phones. There also exist medical

PAGE 18

7 applications for bulk metallic glass such as surgical blades. Defense applications can include composite armor and night vision casings. Applications Micro-Electro-Mechanical-Systems Currently, many MEMS devices are limited with the materials and manufacturing technology that can be used to create them. These processes are primarily limited to silicon in combination with sputtered and etched thin metallic coatings. This layered nature imposes severe limits on the types and range of component geometries, and thus the mechanical motion which can be realized. Compared to macroscopic fabrication processes, the lack of ability to achieve tighter relative tolerances also limits the applicability of current MEMS fabrication technology. By extending the thermoplastic molding process, there is a possibility to economically fabricate miniature components with complex, three-dimensional geometries from high strength materials. Strength properties which can exceed current high-strength alloys, but which can be molded to produce sub-micron sized features at temperatures around 400 C. One could envision fabricating such items as micro-resonators, micro-flexures, micro-surgical tools and devices, micro-motors, micro transmission components, micro-fluidic arrays, non-planar reflective micro-optics, and high frequency microwave components such as waveguides, connectors, and enclosures. The only potentially feasible technology for fabrication of high strength metallic micro-components is LIGA when considering all the current manufacturing technologies for micro-components. If LIGA produces metal dies, then it is possible to think of higher production rates of low temperature material such as PMMA using replication techniques. One might also consider low melting temperature metallic alloys such as

PAGE 19

8 solders, but the tolerances might be rather large due to the solidification shrinkage that occurs. Also, most low melting alloys are weak and tend to creep at room temperature. It is a possibility though to make these components out of higher strength bulk metallic glass. Textured Surfaces Tribological benefits exist from texturing the surface(s) of rotating or sliding bodies. Many current techniques involve using lasers to pattern the surface with microdimples which can act as reservoirs for lubrication. This can in return lower the coefficient of friction or the overall wear of the materials in contact. Longer life cycles may be realized for applications that see highly loaded contacts. In addition to acting as lubrication reservoirs, textured surfaces can help with debris and other contaminants that may come off the contacting surfaces. These potentially harmful fragments of material will not remain at the interface between the materials but rather be buried in the pockets of the textured surface. This will ensure a cleaner surface and thus a lower coefficient of friction. Other patterns besides creating dimples are being researched. Other shapes for texturing include raised features as well. It may be possible to mold such features out of a bulk metallic glass to create a textured surface for tribological applications. It is also not limited to specific shapes or sizes like many rapid laser techniques. Imagine an intricate design of features varying in shape, size, depth, and spatial distances. A truly customized textured surface could be created based on the desired application. Another possible surface, other than a pattern of features, is to create reservoir channels throughout the surface. A series of intersecting channels filled flush to the top of the interface can serve a similar purpose for tribological applications.

PAGE 20

9 Fuel Cell Interconnects Increased power and efficiency can be gained by improving the flow fields and current collection capabilities of Proton Exchange, or Polymer Electrolyte, Membrane (PEM) fuel cells which will lead to performance improvements. The potential to miniaturize fuel cells hinges, in part, on the ability to miniaturize the collection plates is uniquely positioned to support future efforts in design and fabrication of next generation, low weight fuel cells for aviation and space applications. State-of-the-art PEM fuel cell performance can be extended through a combination of rapid plate production with varying channel geometry and system modeling and control, such as computational fluid dynamics models, impedance spectroscopy characterization, and/or the PEM thermal management system. The interconnects compose the outer layer of the PEM fuel cell. One could imagine rapidly producing these channel geometries with a micro-molding technique using bulk metallic glasses. There would be freedom in producing a truly custom pattern of channels over the entire plate. With capabilities on the micron scale, the miniaturization for next generation applications can be realized.

PAGE 21

CHAPTER 3 LITERATURE REVIEW Fabrication Techniques Fabrication methods such as LIGA can be used to produce high aspect ratio microstructures [3-5]. However, LIGA is limited to a fairly narrow set of materials such as polymers or electroformable metals and is a line-of-sight technique. There are different forms of LIGA such as ion beam, x-ray, and ultra-violet lithography. There are issues involved with using lithography techniques; i.e., Cheng and Chen studied key issues when fabricating high aspect ratio microstructures using deep x-ray lithography such as adhesion and film cracking [6]. The LIGA technology is particularly well suited for producing polymer components with high aspect ratios, smooth surfaces and submicron accuracies according to Malek’s and Saile’s LIGA review [7]. Micro-EDM has the highest material removal rate of the fabrication techniques seen in Table 1-1. It is definitely one of the enticing aspects of this technique if the minimum feature size and tolerance isn’t on the submicron size. Pham et al. [8] studied recent developments and research issues involved with Micro-EDM. They focused on the planning of the EDM process and electrode wear problems. A main conclusion from this research suggests that reliable algorithms and strategies with repeatable results should be used for Micro-EDM processes to remain competitive as a micro-machining technology. There are new strategies to replace complex calculations with simpler ones that may make this process more attractive to industries in the future. 10

PAGE 22

11 There are many forms of micromachining currently in progress. Studies are ongoing using a broad range of machining techniques from waterjets [9] to the typical end mills made of tungsten carbide with sputtered cobalt steel. Adams et al. [10] combines two of the fabrication techniques in Table 1-1, focused ion beam and micromachining. Using a focused ion beam, they sputtered cobalt M42 high-speed steel and C2 micrograin tungsten carbide tool blanks to make end mills with diameters of ~25 m. Using these end mills, 1 mm long trenches were created in a variety of materials such as aluminum, brass, and steel. There has also been other research done in the area of focused ion beams for micro or nanofabrication. Both machining with the FIB [11-13] as well as creating tools with the FIB to then use in machining other materials [14] has been studied. Although, the focused ion beam technique provides for nano-sized features and accuracies, one of the main concerns with a focused ion beam is the material removal rate when trying to machine a finished part or even a tool to use for a secondary application. Powder injection molding of metallic microstructures has also been studied [15-17]. Added strength is achieved using alloys of metals and ceramics. Many fabrication techniques are available for features on the micrometer scale. Advantages and disadvantages exist for all of them and the appropriate method is determined based on considerations of application, time, cost, and accuracy. Bulk Metallic Glass and Processes There has been a lot of research into understanding the properties of many different classes of bulk metallic glasses. Great effort has gone into characterizing properties such

PAGE 23

12 as strain rates, viscosities, crystallization kinetics, time-temperature-transformation diagrams, even deformation behaviors and shear flows. The crystallization of the metallic glass is to be avoided if at all possible. The amorphous structure can be utilized for several reasons: 1) better reproduction of the mold, less shrinkage; 2) good mechanical properties. Work has been done by several researchers [18-22] varying critical heating and cooling rates and oxidizing environments. Important results from these papers include critical cooling rates shown to be as low as 1 K/s. Also, it is very difficult to heat the material quick enough, around 200 K/s, to avoid crystallization, however, the faster the heating rate is the higher the crystallization temperature becomes. For instance, for a heating rate of about 2 K/s, the crystallization temperature is about 740 K or 466C compared to a heating rate of 10 K/s which has a crystallization temperature of about 810 K or 536 C. A time-temperature-transformation diagram reflecting the onset of crystallization in the undercooled liquid state which is from the melt temperature down to the glass transition temperature was developed [23]. A “C” shaped curve is exhibited as expected. An interesting observation is below 800 K, two crystallization events occur. The primary crystallization can be attributed to the formation of a nanocrystalline Be-poor and Ti-rich face centered cubic phase that is embedded in a noncrystalline Be-rich matrix. The second crystallization event occurs from the remaining Be-rich supercooled liquid crystallizing [24]. Viscosity has also been studied [25] which is extremely temperature dependent. It is shown to vary orders of magnitude decreasing through the supercooled liquid region until it reaches crystallization where the viscosity begins to increase. However it is

PAGE 24

13 important to note that the relaxation of the viscosity into its equilibrium state is directly related to its glass transition temperature, which in and of itself is directly related to the heating rate of the material. Therefore the viscosity throughout the supercooled liquid regime changes based on the heating rate of the material. Deformation of fully amorphous and partially crystallized bulk metallic glass was studied [26] showing Newtonian flow for the fully amorphous state under low strain rates and high temperatures. For a given temperature, a transition to non-Newtonian flow could be detected when strain rates were increased. The strain rates corresponding to this transition decreases with decreasing temperature. For partially crystallized bulk metallic glass, an increase in viscosity occurs along with a reduction of the strain rate which causes the shift from Newtonian to non-Newtonian flow. Stress-strain relationships were developed over a variety of temperatures from room temperature to the supercooled liquid regime [2]. Important conclusions from this study are that the material exhibits high strain rate sensitivity in the supercooled liquid region. Also, there are three modes of deformation, Newtonian, non-Newtonian, and shear localization. Shear localization should be avoided and is reported to occur at stresses in the range of 1100-1700 MPa. Efforts have been made for forming bulk metallic glasses in a variety of ways including superplastic forging and extrusion. Die forging was studied under low pressures in the supercooled liquid regime [27]. At temperatures of 643 K or 370 C and 653 K or 380 C, pressures around 5 MPa were applied to fill microparts of 200 m in width and 500 m in height. Also, a finite element model of the steady thermal flow of Zr 41.25 Ti 13.75 Cu 12.5 Ni 10 Be 22.5 during continuous extrusion was simulated [28]. A 5 mm

PAGE 25

14 channel within a 100 mm thick copper mold was used in the simulation with the bulk metallic glass melt injected at 1 mm/s. The material was above the melting temperature of ~1000 K and the walls were actively cooled at 373 K. It concludes that the melt is quenched fast enough to avoid crystallization and consequently a 5 mm purely amorphous plate is extruded. There has not been a lot of research in the field of micro-molding bulk metallic glasses. There have been studies forming features with a wet etch process, which consists of heating the metallic glass to the melt temperature and then filling a master mold. Kndig et al. [29-30] performed wet etches of bulk metallic glasses on the micrometer scale with high aspect ratios. They heated the material to a temperature of 1250-1350 K and then spread over a silicon wafer master and allowed to fill the mold. The sample is then quenched with a copper block that applies a small load while cooling. Structures consisting of holes and trenches on the order of 10-100 m are created with a constant depth of 20 m. This project aims to study the micromolding fabrication technique. Higher aspect ratios are desired than that seen in the other wetting techniques done by Kndig et al. Also, molding will occur at a lower temperature at just above the glass transition temperature of the material. The key items to be studied are aspect ratio, proximity, and geometry.

PAGE 26

CHAPTER 4 EQUIPMENT DESIGN/CONSTRUCTION In order to micro-mold bulk metallic glass, multiple components must be designed and incorporated together. These components consist of the MTS machine, top and bottom platens, mold pocket sleeve, key stock punch, along with the bulk amorphous metal and the master mold. The design concepts for these individual components can be seen below in Figure 4-1. Figure 4-1. Components of molding apparatus In addition to these components, a heating and cooling system is required to successfully mold bulk amorphous metals by increasing the material temperature above the glass transition temperature, T g , molding the material, and decreasing it back below T g quickly enough to avoid crystallization. Molding Components An important aspect of the molding process was the platens which house the heating and cooling systems as well as provide a pocket area to perform the molding of 15

PAGE 27

16 the amorphous metal. An approximately 30 mm diameter, 8 mm deep pocket in the bottom platen provided the work area during molding experiments. Additional heat loss due to convection is avoided by performing the molding in a recessed pocket as opposed to on the surface of the bottom platen. Calculations were performed with different material possibilities to observe which material had the best heating capabilities when given a certain heat input. A temperature versus time graph was constructed based on initial calculations of the heating capabilities and used to compare the different materials. This graph can be seen in Appendix A. Aluminum was initially chosen as the material for the first set of platens because it performed well and it is more easily available than the other materials considered. However, as later observed, higher stresses were needed than initially thought, which required the transition to a harder material at temperatures exceeding 400 C. The second set of platens was made from a harder stainless steel. In addition to a block platen for molding, a thin plate was considered as the molding surface. A molding platen would be heated to the desired temperature and then brought into contact with the thin plate mold and the amorphous metal. Theoretically a quick time response could be accomplished this way, thus ensuring a minimal time for crystallization to occur. Both ideas accomplish the heating requirements necessary during the molding process. The first idea of block platens as the molding apparatus was chosen for its simplicity over the multipart thin plate idea. Within the recessed pocket of the bottom platen was where the molding of bulk metallic glass occurred. It consisted of an assembly of six parts. First, a stainless steel disc was inserted as a protector to the platen. If anything went wrong during an

PAGE 28

17 experiment it is much easier to replace this disc than to replace the platen. Second, a mold pocket sleeve made from stainless steel was placed on top of the platen protector disc. This sleeve would ensure that the items placed inside stayed vertical and did not become misaligned and consequentially fall over during a molding experiment. The sleeve was approximately 5 mm in thickness with a 5 mm square hole broached through the center. Next, the silicon wafer mold was inserted into the mold sleeve followed by a thin slice of bulk metallic glass. A section of 5 mm square key stock would serve as the punch for this molding process and would be placed on top of the metallic glass. The final piece was another platen protector placed above the key stock punch for the top platen. The top platen was then brought into position with a light load to hold the components in place before testing began. The last task for the design of the platens was to fasten them to the MTS machine. In order to do this the platens were bolted to a ” thick piece of aluminum. This piece was then sequentially bolted to a cylindrical aluminum piece designed to fit the hardware of the MTS machine. The cylindrical piece was inserted into a metal sleeve already fastened to the MTS machine and a locking pin was inserted through it. Figure 4-2. Molding components secured to MTS machine

PAGE 29

18 Heating Process Several heating options were investigated to accommodate the high heating rate needed for the bulk metallic glass to remain amorphous. Calculations were performed to determine how much power would be required to heat the platen based on initial dimensional goals of 40mm mm mm to 400 C in a reasonable amount of time, on the order of 30-60 seconds. Cartridge heaters were chosen as the heat source because the molding platens are small and cartridge heaters have high power densities compared to other heater options. Tests were also performed to quantify the time response of cartridge heaters both in air and in metal. An aluminum block roughly 2 inches cubed was used with a 150 W, 0.5” diameter cartridge heater inserted in the middle of the block. Temperature data was collected using a thermocouple probe at the surface of the aluminum block. The temperature versus time graphs for these heating tests can be seen in Appendix A. Using the experiments and theoretical calculations based on the initial dimensions of the platen, 1800 Watts of heating power would have a quick enough time response to heat the platen during the molding process. Cartridge heaters that were 200 W, 0.25” diameter, and roughly 1.5” long were selected to ensure the required 1800 W. Nine cartridge heaters are used in each platen for the heating application. The set of stainless steel platens for higher force molding required more heating power as it trailed its aluminum counterpart in both thermal conductivity and thermal diffusivity. For the stainless steel, 5000 W of heating power was used. Each cartridge heater had 250 W and there were 11 in the bottom platen and 9 in the top platen.

PAGE 30

19 The final time response of the cartridge heaters in the stainless steel platen can be seen in Figure 4-3. Figure 4-3. Experimental heating profile for stainless steel platen The platen reaches 400 C in under 1 minute. One option during the molding process was to control the temperature to maintain 350 C until it was time to heat above the glass transition temperature. From this graph it can be seen that it takes roughly 10-15 seconds to heat up to 400 C. However this option turned out to be not necessary. Cooling Process The cooling process is needed to decrease the temperature of the bulk amorphous metal as quickly as possible back below the T g once the molding was completed. Cooling is achieved by piping a low temperature fluid through the molding platen. Different fluids, including high temperature oils, were considered in mass flow calculations performed to determine relative temperature decreases of theoretical platens with the same pumping rate. Of all the fluids considered, water was chosen because it

PAGE 31

20 performed sufficiently and is an easy resource to acquire. Calculations were performed to determine the temperature rise of the water that would be encountered and the nucleate boiling point. Based on this information an appropriate centrifugal pump was selected and a path through the platens was mapped out. A B Figure 4-4. Cooling water path. A) through aluminum platen, B) through stainless steel platen Once the cooling fluid and pump were determined, the plumbing system was designed. It consists of a water reservoir and stainless steel hoses. Flexible stainless steel hose was chosen to accommodate the movement of the MTS machine during the

PAGE 32

21 molding process as well as the high temperatures that could be encountered from the heated water upon exit from the molding platen. Water exiting the pump divides into two channels towards each platen inlet. A rotary valve is connected to each channel to control the water flow to each platen independently in case water cooling is desired in one platen only. Water exiting the platens flows independently back into the water reservoir. Tests of the cooling process show excellent cooling capabilities with a time response as seen in Figure 4-5. Figure 4-5. Experimental cooling time response of stainless steel platen Final Design The final dimensions of the molding platen are 40.64mm .64mm .55 mm and details of the stainless steel components are shown in Figure 4-6.

PAGE 33

22 Figure 4-6. Final schematic of molding setup stack The final temperature versus time response for the complete process can be seen below in Figure 4-7. Figure 4-7. Experimental time response for complete heating and cooling process

PAGE 34

23 The heating equipment was designed to perform at a higher level than what is necessary for this specific molding process. The heating system has the capability of heating the platens to higher than 600 C, a temperature far greater than the molding window of this type of bulk metallic glass. The MTS machine equipped with all installed equipment can be seen in Figure 4-8. Figure 4-8. Equipped MTS machine

PAGE 35

CHAPTER 5 UNCERTAINTY ANALYSIS With any process comes uncertainty. This process relied on measuring two key variables, temperature and force. Temperature was important because bulk metallic glass is a temperature dependent material. The viscosity changes greatly with small changes in temperature. Tests were performed at many different temperatures and it is essential to understand any involved errors. In addition, force was extremely important as it determined at what pressure the experiment was conducted. Temperature To understand the uncertainty of the temperature, it is crucial to understand the form in which the temperature was being measured. Many components make up the measurement of temperature, all which have an error associated with them. Temperature was measured by an Omega Type-K Thermocouple embedded into each of the molding platens. This thermocouple was wired into an Omega CNi3252 Temperature Controller and Monitor. After the thermocouple reading was converted using an analog to digital converter, the temperature was displayed on the front face of the temperature controller. The value was then converted back via a digital to analog converter and transmitted in the form of a voltage to a National Instruments PCI-6014 data acquisition card connected to a computer. This voltage was then recorded and scaled appropriately to determine the temperature in C. The temperature measuring components can be written together into a single equation which produces the measured temperature. This equation can be seen in 24

PAGE 36

25 Equation 5-1. It simply states that the temperature value is a voltage received to the data acquisition card multiplied by a scaling constant which converts it into a temperature. Equation 5-1: Using the law of propagation of uncertainty, an equation can be written to quantify the uncertainty of the measured temperature value. The uncertainty is temperature dependent, so a reported value of uncertainty is necessary for each of the molding temperatures. The uncertainty equation can be seen in Equation 5-2. Taylor series expansion is conducted for the two variables in the temperature equation. In addition to these sources of uncertainty, there exists uncertainty in the thermocouple reading and is added to the overall equation. Equation 5-2: Table 5-1. Temperature components with uncertainty in measurement Measuring Component Units Uncertainty Type-K Thermocouple Temperature [C] 0.4% of Temperature Calibration Constant [C/V] 0.9999 Data Acquisition Card Voltage [V] 0.0003 Assuming a rectangular distribution, the uncertainty is defined as the manufacturer’s error range divided by 3. Inserting the known values into the uncertainty equation, the associated uncertainties can be seen in Table 5-2 for temperature.

PAGE 37

26 Table 5-2. Uncertainties for the four molding temperatures Temperature Uncertainty 400 C 2.4 C 425 C 2.6 C 450 C 2.7 C 475 C 2.9 C This uncertainty analysis focuses on the measurements taken inside the platens. However, this is not the area where the actual molding took place. It wasn’t possible to place the thermocouples closer to the molding sample because it was too far from the heating cartridges and made for unstable temperature control. So, measurements were taken with an external handheld thermocouple placed inside the mold pocket sleeve close to the interface where the actual molding took place. These temperature measurements were compared to the measurements from the platens. Figure 5-1. Temperature versus molding time of mold sleeve

PAGE 38

27 The tests were conducted with a prescribed temperature of 450 C. It can be seen that it takes about 30 seconds after the loading pressure was applied before the mold sleeve temperature enters between the range of platen temperatures. This would suggest that for smaller molding times of 15 seconds, the actual temperature was lower than expected, however still close to the desired molding temperature. Force Force control was conducted completely inside a closed loop within the MTS controller and frame. All load cells including the one used for the molding experiments were calibrated by a MTS technician upon arrival to the site. Calibration sheets were provided upon the completion of the technician’s tests and are used in the uncertainty analysis for the force measurement. To record the force for later processing and evaluation, an analog output from the MTS controller to a data acquisition card was set up. The MTS controller acted independently from any recording of force data using LabVIEW. The MTS controller accounted for calibration constants associated with the load cell internally and therefore the only necessary steps to receive that force value was to first scale it into a voltage, receive it with the data acquisition card, and then re-scale it back to the original force value. The uncertainty can be broken up into two areas, the uncertainty of the MTS frame and controller to accurately produce a force value and the data acquisition card’s resolution to receive the voltage via the analog output. The uncertainty value of the MTS frame and controller is quantified from the calibration tests performed by the MTS technician. The data acquisition card’s uncertainty value is the same as shown in the

PAGE 39

28 temperature uncertainty analysis as the resolution of the card. The uncertainty values are listed in Table 5-3. Table 5-3. Uncertainty values for force measurement Component Uncertainty MTS frame and controller 0.46%*Force Data acquisition card 0.0003 V Again, assuming a rectangular distribution of potential values, these terms can be applied to Equation 5-3. The associated values of uncertainty for the four applied force values can be seen in Table 5-4. Equation 5-3: able 5-4. Uncertainty values for the four applied forces T Force Uncertainty 2500 N 11.5 N 2875 N 13.2 N 3250 N 14.9 N 3750 N 17.2 N he uncertainty in the force measurement is dominated by the uncertainty of the calibrated MTS machine value as the resolution of a 16-bit data acquisition card provides relatively low uncertainty. T

PAGE 40

CHAPTER 6 SILICON WAFER MASTER MOLD Design Important parameters studied to help understand the flow characteristics of bulk metallic glass include aspect ratio, geometry, and proximity. With knowledge of these three parameters one could get a decent understanding of the types of components that could be molded from bulk metallic glass. Without a real understanding of how this material behaves in a molding process, a combinatorial approach was utilized. There were nine possible combinations including the three individual parameters already mentioned. With all possible combinations accounted for, one could better understand how this material flows and what the important parameters include. The six combinations include: Aspect Ratio and Geometry (AG) Aspect Ratio and Proximity (AP) Geometry and Proximity (GP) Aspect Ratio and Geometry and Proximity(AGP-a) Aspect Ratio and Geometry and Proximity(AGP-b) Aspect Ratio and Geometry and Proximity(AGP-g) Alpha, beta, and gamma represent three different spatial distances, respectively 5 microns, 10 microns and 100 microns. The design of the 9 individual and combinatorial parameters can be seen in Figure 6-1. The beta spatial distance is represented from the Aspect Ratio-Geometry section since beta represents the standard spatial distance. 29

PAGE 41

30 Figure 6-1. Individual and combinatorial parameters for AGP silicon wafer mold The mold sample was designed just less than 5mm x 5mm to fit into the mold pocket sleeve. Each 5 mm x 5 mm silicon sample was divided into a grid of 1mm squares giving 25 square areas for the three parameters and their combinations to be tested. An important assumption was that over each 1mm square section the pressure was equal, this allows all features within these 1mm square areas to be compared without compromising results. Each parameter and combination was given squares in the grid and ensured to have a section close to the middle as well as one towards the outside. This placement was randomly completed. The final layout can be seen in Figure 6-2.

PAGE 42

31 Figure 6-2. AGP 5 mm x 5 mm grid layout Deep Reactive Ion Etching The mold was made from deep reactive ion etching a 4” silicon wafer. The 5 mm x 5mm square molds were arranged on the 4” silicon wafer using AutoCAD. It consisted primarily of the AGP featured pattern silicon wafer molds with a pattern designed to further study the pressure profile placed sporadically throughout the 4” disk. The final layout of the 4” wafer can be seen in Figure 6-3. The layout allowed for 167 of the featured pattern and 16 of the pressure pattern. This allowed extra molds during the process setup of the molding apparatus. Figure 6-3. 4” silicon wafer master layout

PAGE 43

32 The AutoCAD file once completed was sent to Photo Sciences Inc. to have the photomask made for the deep reactive ion etching process. The photomask’s purpose is to transfer the desired etching pattern onto the silicon wafer. The photomask is made from a soda-lime glass with a thin layer coating of chrome or iron oxide. Photoresist is then applied onto the coating of chrome and a pattern is exposed onto the photoresist. The pattern is then etched through the chrome coating. The final result is a photomask with the desired layout for deep reactive ion etching. The silicon wafer used for the etching of the molds was a silicon on insulator or an SOI wafer. This type of wafer was chosen because it was important to etch completely through the wafer. Not knowing how easily the molded bulk metallic glass and the silicon mold would release from each other, the molded samples could be turned over to the backside to make any measurements necessary with the white light interferometer. A SOI wafer simply consists of a device layer made of silicon, which is the top layer, roughly 100 micrometers in thickness, followed by a silicon dioxide layer roughly 2 micrometers in thickness, and finally a handle layer made of silicon roughly 300 micrometers in thickness. For this etching process, the silicon dioxide layer acts as a “hard stop.” The etch rate of silicon dioxide is much slower than that of silicon. Since the feature size affects the etch rate, larger sized features etch faster than smaller features, using the silicon dioxide as the stopping point allows the completion of the smaller features without worrying about varying size depths of the etch. Once the SOI wafer is etched, it can simply be given an acid bath to remove the two silicon layers leaving only the top layer with the desired pattern etched completely through.

PAGE 44

33 The SOI wafer is completely cleaned before any etching can occur. In order to clean it, acetone followed by methanol is applied to the wafer. It is dried before and after each bath using nitrogen gas. All work during cleaning and etching was completed in a clean room located in the Interdisciplinary Microsystems Group Lab in Benton Hall. The first step to etching the 4” silicon wafer was to apply a thin coating of photo-resist about 10 micrometers thick. In order to accomplish this, a specific amount of photo-resist was measured out and applied to the center of the silicon wafer. The wafer was then spun for a precise length of time leaving the 10 micrometer thick coating. The photo-resist is then baked in an oven at 90 C for 30 minutes allowing the coating to dry completely. The next step in the etching process was to transfer the desired pattern to the wafer with the photo-resist coating. The wafer was placed in a machine with the soda lime photomask placed above it. UV rays were exposed through the photomask onto the photo-resist on the wafer essentially etching the pattern through the photo-resist. The desired pattern was now transferred to the wafer and was ready for the deep reactive ion etching machine. The 4” silicon wafer was placed inside the STS deep reactive ion etching machine and transported into the etching chamber by the machine. The deep reactive ion etching process consisted of bombarding ions at the wafer. Wherever the photoresist was gone from the previous step, the ions etched into the silicon wafer. The etching rate varied between 1-2 micrometers per step based on feature size. A total of 110 cycles were performed to ensure that the device layer of the SOI wafer was etched completely through to the silicon dioxide layer. Each step took about 1 minute to complete. The

PAGE 45

34 etch rate was not an exact science, it can vary depending on the current conditions, so there was some uncertainty involved when determining whether the device layer was etched completely through, especially since hole sizes varied from 5 micrometers to 100 micrometers. Once 110 cycles were performed, the device layer was separated from the handle layer. In order to separate the two layers the middle silicon dioxide layer was etched away. An aggressive hydrofluoric acid was chosen over a buffered etch because of the distance between some of the features in the pattern. The middle 1 mm square of the grid was not patterned leaving at least 1 mm of silicon dioxide to etch away. The etch rate of the hydrofluoric acid was about 1.5 micrometers per minute. The hydrofluoric acid used to etch the silicon dioxide layer was 49% by weight. The SOI wafer was placed in the bath for about 10 hours under constant agitation to allow for the silicon dioxide debris to be removed from the area that still needed to be etched. Once 10 hours passed, the wafer was removed from the bath and cleaned thoroughly. The tabbed 250 micrometer channels etched around each 5 mm x 5mm section held the device layer together but these sections were easily broken at the tab afterwards. Once all the mold sections were removed the etching process was complete and the only final preparation for the silicon wafer molds was to scrape away part of the tabs to ensure it would fit into the mold pocket and clean them. An optical microscope picture was taken for the AGP silicon wafer master and can be seen in Figure 6-4.

PAGE 46

35 Figure 6-4. Optical microscope picture of AGP silicon wafer master Scanning Electron Microscope micrographs were taken on portions of the silicon wafer master. These micrographs can be seen in Figure 6-5. They show sections from aspect ratio, geometry, and aspect ratio-geometry-proximity-. A B

PAGE 47

36 C D Figure 6-5. SEM micrographs of four sections of AGP pattern silicon wafer master. A) aspect ratio, B) zoomed-in aspect ratio, C) geometry, D) aspect ratio-geometry-proximity

PAGE 48

CHAPTER 7 EXPERIMENTAL PROCESS Procedure As-received plates and pins of bulk metallic glass were first remelted with a Centorr model 5-BJF arc-melter furnace under an argon atmosphere and cast into a 5.25mm x 5.25mm x 25 mm long block copper mold. The Centorr arc-melter furnace can be seen on Figure 7-1. Figure 7-1. Centorr arc-melter furnace used for remelting bulk metallic glass material. It is necessary to recast because of the porosity of the as-received samples. By recasting these samples we are essentially eliminating the pores in the material that when molding would leave gaps of the replicated mold in the silicon wafer master. An example of a recasted pin in the copper drop mold can be seen in Figure 7-2. 37

PAGE 49

38 Figure 7-2. Recasted bulk metallic glass in copper square pin mold The metallic glass was then cut into approximately 0.7 mm slices using a South Bay Technology low-speed diamond saw. The castings were checked for crystallization with a Phillips APD 3720 diffractometer with a Cu K x-ray source. Assuming no crystallization was found, the samples were then polished using a South Bay Technology 8” polishing wheel. The polishing wheel and low-speed diamond saw can be seen in Figure 7-3. Figure 7-3. 8” South-Bay Technologies polishing wheel and South-Bay diamond saw

PAGE 50

39 The silicon wafer mold was then cleaned in an acetone, methanol sonicator bath, dried with compressed air, and finally inserted into the pocket of the bottom platen. The sonicator used can be seen in Figure 7-4. Care was taken to ensure that the silicon wafer mold was inserted in the same orientation for each test. The bulk metallic glass was added on top of the silicon mold inside the pocket. Finally, the punch was aligned a few millimeters into the pocket and the experiment was ready to begin. An initial light pre-load was applied to facilitate heating of the bulk amorphous metal. When the temperature exceeded the glass transition temperature, a larger molding force was applied for a prescribed time. The sample was then rapidly cooled by the centrifugal pump which forced water through a manifold pattern in both platens. The metallic glass and silicon wafer were then extracted together from the mold pocket. Figure 7-4. Sonicator used for acetone and methanol baths The bulk amorphous metal was then checked at random conditions using the Phillips APD 3720 diffractometer for any signs of crystallographic peaks. These peaks can be a sign of recrystallization, oxidation, or other surface reactions. The molded sample was also turned over and a VEECO white light interferometer was used to measure the flow of the metallic glass into the channels and holes. Due to flow between holes and channels that may block the white light interferometer from measuring

PAGE 51

40 individual features, the silicon wafer mold must be removed from the molded bulk metallic glass. Once this is complete the white light interferometer or other measurement means can be used. The VEECO white light interferometer can be seen in Figure 7-5. Figure 7-5. VEECO white light interferometer In addition, a two step wet etch process is employed to completely remove the molded bulk metallic glass from the silicon wafer mold. First, the sample and silicon are placed in a 20% by volume acetic acid – 80% by volume ethyl alcohol solution. This solution attacks any oxide layer that may have formed on the surface of the silicon wafer during the molding process. Second, a 20% potassium hydroxide – water solution can be applied to the bulk metallic glass and silicon wafer mold to essentially etch away the silicon. This step is performed at 70 to 80 C which increases the etch rate of the silicon.

PAGE 52

41 This process will not attack the bulk amorphous metal, leaving it behind. Once this is complete, a sample with possible aspect ratios as high as 20 to 1 is left for further analyzing. A FEI Strata DB 235 focused ion beam (FIB) machine with SEM capabilities was utilized to analyze the molded features. Using an on-board measuring device of the FIB, heights could be characterized for comparison. The FIB can be seen in Figure 7-6. Figure 7-6. FEI focused ion beam with scanning electron microscope capabilities Conditions A test matrix of 64 tests was completed to understand the flow at various temperatures and pressures for certain amounts of molding times. This information gives a broad understanding of the mold capabilities throughout the temperature window above the glass transition temperature and below the crystallization temperature for this particular bulk metallic glass. Glass transition temperatures for Zr-based bulk metallic glasses can vary depending on the source. They can even vary for this specific bulk metallic glass. The minimum temperature that experiments were performed was well

PAGE 53

42 above the glass transition temperature in the literature. The molding can operate at this higher temperature due to the response time of the heating system. In fact, molding a higher minimum temperature costs only about 10-15 seconds, which is a decent tradeoff to ensure that we are molding above the glass transition temperature. On the other side, consideration should be given to the temperature in which crystallization could be seen. This also varies depending on the literature, but the maximum temperature that experiments were performed is less than most reports of crystallization temperature. This still provides a fairly large temperature window of 90 C for molding to occur. With this knowledge, temperatures were selected at 25 C intervals for a total of 4 different molding temperatures to compare. According to J. Lu et al. [2], there exists a stress strain curve which can act as a guide to initial approximations in which bulk shear flow would exist. It is important to achieve this bulk deformation flow to ensure complete filling of the silicon wafer mold. After initial tests, our data followed closely to the information reported in this paper. With this knowledge an initial minimum pressure was established for the molding experiments. The maximum pressure was determined from the maximum force that could be accomplished by our load frame. The maximum load cell for the MTS machine is 5000 N, which correlates to 200 MPa using a 25 mm 2 sample. From this, we had the range of our pressure values and we selected 2 other flow pressures in between the minimum and maximum. It was later decided to lower the maximum pressure to get more likelihood of flow variations in the material. Time was more arbitrary than the temperature and pressure. At these temperatures and pressures experiments were completed to get an initial time range which may be

PAGE 54

43 applicable to study. The range of times allow for very little flow to complete flow through the mold especially at the lower temperatures. With the three parameters in place a final test matrix was completed. It resulted in 64 tests. The test matrix can be seen in Table 7-1. Table 7-1. Experimental test matrix for AGP pattern silicon wafer master Temperature [C] Applied Pressure [MPa] Molding Time [sec] 400 100 15 425 115 30 450 130 45 475 150 60 Each test was performed under laboratory air and conditions. The tests were not randomized due to temperature controller tuning at each of the set point temperatures. Heating and cooling rates were not varied over the range of experiments. A small stabilization time allowed the molding temperature to reach a steady state before the molding pressure was applied.

PAGE 55

CHAPTER 8 RESULTS AND DISCUSSION Results showed excellent moldability and reproducibility of the features. Aspect ratios as high as 13.5 to 1 have been observed. The silicon wafer thickness was characterized using a SEM micrograph. Nominally it is 100 m thick, however after further investigation it was measured to be around 96 m thick. Since the deep reactive ion etching process is limited to 20 to 1 aspect ratios, the 5 m diameter features had difficulty etching completely through the silicon master as it approached the limit of the process. The 10 m feature was the smallest feature that consistently etched completely through the wafer, therefore this was the feature used for aspect ratio comparison over the experimental test matrix ranges. Over all the prescribed temperatures, bulk deformation flow of the metallic glass occurred. The best flow occurred at 450C. At 475C, for the two largest molding times of 45 and 60 seconds, a darker gold tone was observed on the surface of the bulk metallic glass. This could occur from surface oxides or may be an effect of crystallization. Since this material is temperature, time, and oxygen dependent, sustaining elevated temperatures for extended times can lead to crystallization. Also, crystallization is more likely at higher temperatures because it takes longer to cool below the glass transition temperature causing the cooling rate to be a crucial factor. Aspect Ratio, Geometry, and Proximity Characterization of the 64 tests was done by several methods including optical microscope, white light interferometry, and focused ion beam with SEM capabilities. It 44

PAGE 56

45 was observed that geometry and proximity had little effect on the molding of the bulk metallic glass. Comparing circle and square geometries showed relatively close molded heights. The square feature was slightly higher which could be a function of a larger cross sectional area. A molded circle next to a square can be seen in Figure 8-1 to have similar molded heights. Figure 8-1. White light interferometer scan of a circle and square feature. Both had heights near 11 m The designed proximity section did not etch completely through the wafer on most of the molds with the deep reactive ion etching process. However, data at lower aspect ratios show the proximity varied minimally between holes. Any variations in height are most likely due to localized conditions of temperature and oxygen presence. An example of a two-dimensional cross section plot of proximity can be seen in Figure 8-2. At higher temperatures with greater metallic glass flow, proximity also can be seen to have little effect as features with small spatial distances flowed as much as features

PAGE 57

46 with larger distances. Although a proximity section is not directly available, observations throughout the rest of the sample suggest spatial distances play a minimal role if at all. Figure 8-2. Two-dimensional height values for 10 m diameter features with increasing spatial distance away from the center For the last individual parameter studied, aspect ratio, diameters were varied to produce maximum achievable aspect ratios with the given silicon wafer thickness. It was very easy for the larger sized holes of 50 m and 100 m to be completely filled under lower temperatures, pressures, and times. Early experimentation suggested that the aspect ratio was the same for all given diameters. This was only true when the viscosity levels were still high, once the temperature of the material got to 400 C, aspect ratio in terms of feature diameters was less predictable. With many of the larger diameter holes filled completely to its maximum aspect ratio at low operating conditions, it is difficult to complete an accurate analysis comparing diameters. One would expect the larger diameter features would allow for higher aspect ratios unfortunately the etching process limits these aspect ratios.

PAGE 58

47 Temperature, Applied Pressure, and Molding Time It was observed that the greatest factor affecting the molding of bulk metallic glass was the temperature. The temperature dependence trumped both the applied pressure and molding time conditions. The viscosity of bulk metallic glass in the “supercooled” liquid regime is very dependent on the temperature. In Figure 8-4, the relationship between the viscosity of the material and the temperature can be seen. Figure 8-4. Viscosity versus temperature for Vitreloy-1 with a heating rate of 20 K/s [2] Slight variations in temperature can cause large shifts in viscosity. This is what was observed when characterizing the molded bulk metallic glass samples. In Figure 8-5, 4 pictures can be seen of the same features under an identical pressure and molding time of 115 MPa and 15 seconds respectively. The 4 samples represent the 4 different temperatures in the experimental matrix.

PAGE 59

48 A B C D Figure 8-5. 10 m features molded at 115 MPa for 15 seconds. A) 400 C, B) 425 C, C) 450 C, D) 475 C The average aspect ratio for the features in Figure 8-5A, molded at 400 C, was 0.92. The aspect ratio of the features in Figure 8-5B, molded at 425 C, was 4.45. The features molded at 450 C and 475 C had an average aspect ratio greater than 10, approximately 12.5. Aspect ratio results were accomplished by measuring the heights of the same set of features on all the samples. These features were the two 10 m diameter circles approximately 10 m apart as seen in Figure 8-5. The silicon master was inserted in the

PAGE 60

49 mold pocket sleeve with the same orientation for every test and the features were located towards the middle of the silicon wafer master. Measurements showed that most of the 450 C samples and all of the 475 C samples had 10 m features molded completely through the silicon mold. For these samples aspect ratios are limited by the deep reactive ion etching process. Results however were conducted for each sample over the 400 C and 425 C samples. A plot of these measurements can be seen in Figure 8-6. A

PAGE 61

50 B Figure 8-6. Aspect ratio plotted against molding time for the four pressures. A) 400 C, B) 425 C Results for these measurements show that applied pressure and molding times have little effect on observed aspect ratios. It shows overlap between these process parameters and suggests that temperature is a more dominant parameter. Variations of aspect ratios are due to localized conditions and variations during the molding process. Figure 8-7 shows the aspect ratio measurements for both temperatures plotted together. Aspect ratios were plotted against maximum temperature as opposed to a nominal desired temperature or average temperature because of the dominant effect of temperature and the fact that viscosity can change dramatically over small changes in temperature. It was determined that the maximum temperature the material encountered would characterize the aspect ratios the best.

PAGE 62

51 Figure 8-7. Aspect ratio versus maximum temperature for 400 C and 425 C It can be seen in Figure 8-7 aspect ratio increases with increasing maximum temperature which is what would be expected. The limiting mold aspect ratio is around 10 which suggest some samples were molded further than the thickness of the silicon wafer. It is unclear whether the trend is linear or follows another trend such as a parabolic form. More temperature data in the range between 400 C to 425 C might clear up the shape of the trend. Also at lower temperatures where the viscosity is higher, applied pressure and molding times might factor into aspect ratio more. X-ray diffraction was used to determine any oxidation or crystallization effects on the bulk metallic glass at several random test conditions. A diffraction pattern reports the x-ray intensity in counts as a function of 2 angle. The range of 2 was 30 to 80.

PAGE 63

52 Diffraction tests indicate that most of the material has remained amorphous. An example of a x-ray diffraction graph can be seen in Figure 8-8 for the four examples of Figure 8-5. A B C D Figure 8-8. X-ray diffraction patterns for the four samples seen in Figure 8-5. A) 8-5A, 400 C, B) 8-5B, 425 C, C) 8-5C, 450 C, D) 8-5D, 475 C Broad diffuse peaks suggest that the material has remained amorphous. Sharp peaks would suggest crystallization in the material because evidence of atomic structure would increase the intensity. As can be seen in Figure 8-8, the broad diffuse peaks can be seen over the entire angle range. But there exists areas that may have localized crystalline structures as seen by the sharp peaks in the x-ray diffraction plots. A transmission electron microscope was used to examine a section of bulk metallic glass molded at 450 C for the maximum time of 60 seconds. One 50 m square feature which had molded completely through the silicon wafer mold was checked for crystalline

PAGE 64

53 structure. It showed partial crystallization with the rest of the material amorphous. More characterization would need to be completed to understand the cause of the crystallization. It could be from too slow of a cooling rate, extended time at an elevated temperature, or from the presence of larger quantities of oxygen as it was the leading edge during the molding. An image of the TEM micrograph showing part crystalline, part amorphous structure can be seen in Figure 8-9. Figure 8-9. TEM micrograph of part crystalline, part amorphous section of 50 m feature Additional Flow Characteristics Excellent reproduction of the molds can be seen in the bulk metallic glass. For example, the deep reactive ion etching process produces scallops or cusps along the sidewalls of the silicon mold. These scallops are created because it is a two step process of etching and passivation in cycles of 1 m to 2 m. A schematic of the deep reactive ion etching process scallops can be seen in Figure 8-10.

PAGE 65

54 Figure 8-10. Illustration of scallops produced from deep reactive ion etching The cycles provide for nearly 90 degree sidewalls which allow for high aspect ratio features. A SEM micrograph of cross sections of 10 m holes in the silicon wafer can be seen in Figure 8-11. Figure 8-11. 10 m feature cross section in silicon wafer mold

PAGE 66

55 It can be seen that the scallops are more distinct at one end of the hole and fade to what resembles stretch marks. These vertical “stretch marks” were believed to form from the hydrofluoric acid bath given after the deep reactive ion etching. A scanning electron micrograph of the replicated scallops in the molded bulk metallic glass can be seen in Figure 8-12. A B Figure 8-12. Reproduced scallops or cusps along the sidewall of the molded feature. A) full feature, height>100 m, B) zoom in on scallop area. As noted in the Experimental Procedure section, in order to separate the handle layer from the device layer, HF acid is applied which etches the silicon dioxide layer, however, it also etches silicon although much slower. This is what is believed to have nullified the scallops along most of the sidewalls on the features. To further justify this as the case, one could look at a channel geometry that was not given a hydrofluoric acid bath. It can be seen in Figure 8-13, scallops do indeed develop along the entire 60 m length of the sidewall. The scallops are close to 1.5 m in height.

PAGE 67

56 Figure 8-13. Reproduced scallops along entire sidewall of channel geometry One of the main fabrication techniques for producing high aspect ratio features is forms of lithography as mentioned earlier. A key characteristic of the lithography is that it is a line-of-sight technique. Non-line-of-sight flow has been observed in the molding process of bulk metallic glass, suggesting that more complex three-dimensional geometries could be filled. During the deep reactive ion etching process, a “footing” is created on one end of the silicon wafer mold, see Figure 8-14. Footing refers to a small gap between the end of the etched holes and the top surface of the silicon wafer. Figure 8-14. Footing created on one side of silicon wafer during deep reactive ion etching process

PAGE 68

57 The footing is created because holes of various sizes are being etched. Larger holes etch quicker than smaller holes. The silicon dioxide layer of the SOI wafer was used as a hard stop. Since a range of holes from 100 m to 5 m were being etched, the largest hole reached the silicon dioxide layer first. However, it was necessary to continue to etch in order for the smaller holes to be completely etched through. When it reached the silicon dioxide layer, it began to etch outwards at the bottom of the holes as illustrated in Figure 8-15. If holes are within close proximity it forms a gap as seen in Figure 8-14. Figure 8-15. Illustration of side etching which creates footing in the deep reactive ion etching process This small void has been filled in Figure 8-16. Flow around corners is specific to molding processes and isn’t characteristic of many other fabrication techniques. One could imagine creating a series of silicon wafer molds that when assembled creates a complex shape or complex channel geometries. Applications for complex three-dimensional features could be different micro-devices such as cantilevers, resonators, and pumps to name a few. It could also be series of fluidic channels for biomedical applications.

PAGE 69

58 Figure 8-16. Filled footing in molded bulk metallic glass Some additional images of the AGP results can be seen below in Figure 8-17. Additional images from the AGP experimentation can be seen in Appendix B. Two-dimensional channel geometry results and images can be seen in Appendix C. Tribological studies of molded bulk metallic glass with the addition of solid lubricants can be seen in Appendix D. A B

PAGE 70

59 C D Figure 8-17. Additional SEM micrographs of AGP results. A) aspect ratio, B) aspect ratio – proximity – geometry, C) geometry, D) aspect ratio

PAGE 71

CHAPTER 9 CONCLUSIONS A micro-molding platform has been designed, constructed, and tested. It successfully heats and cools quickly enough for all or most of the bulk metallic glass to remain amorphous depending on the conditions. The platens can obtain 400 C in about 1 minute. The bottom platen was designed with a recessed pocket for better heat transfer to the sample. Experiments were carried out to demonstrate micro-molding of micrometer scale features in bulk metallic glass. A silicon wafer master mold was designed to test the effects of aspect ratio, proximity, and geometry as well as all the possible combinations of these three parameters. A range of temperature, pressure, and molding time combinations were tested and successful reproduction of a variety of amorphous, high aspect ratio features from silicon masters was observed. Aspect ratios as high as 13.5 to 1 were observed in the molded bulk metallic glass. Comparison between different diameter holes in terms of aspect ratio was difficult as the larger sized features filled the silicon mold completely. Proximity showed little effect during the molding process. Closely spaced features filled as much as features alone. Geometry also showed little effect. Squares filled slightly more so than circle, most likely do to the cross sectional area of the squares being larger. Applied pressure and molding times were dominated by temperature. Variations in pressure and time were overshadowed by the smaller fluctuations seen in the temperature profile. 60

PAGE 72

61 Excellent replication of the silicon molds was observed from the scallops reproduced on the sidewalls of the material. The scallops produced during the deep reactive ion etching process approximately 1.5 m in height can easily be seen in the metallic glass. Additionally, non-line-of-sight flow was seen, which suggests the potential for the production of three-dimensional geometries by this method.

PAGE 73

APPENDIX A THEORETICAL CALCULATIONS FOR PLATEN DESIGN Theoretical time versus temperature calculations performed on eleven different materials. Aluminum, the first platen choice, appears near the top of these curves. Stainless steel, although not as thermally good as aluminum, was needed to account for higher stresses at elevated temperatures. Figure A-1. Time versus temperature of various materials under consideration. Equipment tests were performed with available cartridge heaters in laboratory air conditions as well as aluminum blocks. The higher powered smaller diameter cartridge heater performs much better than its counterpart as expected. 62

PAGE 74

63 Figure A-2. Time versus temperature profiles. A) in air, B) in aluminum block

PAGE 75

APPENDIX B ADDITIONAL AGP IMAGE RESULTS Scanning electron microscope images of molded bulk metallic glass with low bulk flow into the silicon wafer mold. A B C D 64

PAGE 76

65 E F G Figure B-1. SEM micrographs of molded bulk metallic glass with little bulk flow Scanning electron microscope images of other molded bulk metallic glass with larger bulk flow or complete filling of silicon wafer mold.

PAGE 77

66 A B C D

PAGE 78

67 E F G H Figure B-2. SEM micrographs of molded bulk metallic glass

PAGE 79

APPENDIX C TWO DIMENSIONAL CHANNEL GEOMETRIES A silicon wafer was also designed to create two-dimensional channel geometries. The purpose of these channels was to show that it is possible to rapidly fabricate these micrometer scale geometries with potential application to fuel cell interconnect plates. Straight sidewalls were molded creating channels between them. Channel widths varied from 10 m to 200 m. In addition to straight channels, more complex sidewalls that are difficult to create with other techniques were produced. Oscillating sidewalls were created with variations in frequency, amplitude, and combined frequency and amplitude. Also, channel widths were varied for the oscillating sidewall geometries. Other geometries that were created include a serpentine pattern and a section of miscellaneous sidewalls. The channel layouts can be seen in Figure C-1. Figure C-1. Eight channel geometry designs 68

PAGE 80

69 The sidewalls were molded to a depth of 50 m to 60 m. All eight variations of channels were molded on every sample with room for repeats on the 5 mm x 5 mm samples. All channel variations reproduced well in the bulk metallic glass. Since the silicon wafer mold for the two-dimensional channel geometries was created with a timed cycle etch and didn’t etch completely through the wafer, a much better surface finish can be seen compared to the AGP features. For the AGP features, the tops of the features were molded against the stainless steel platen protector disc rather than against silicon. Scanning Electron Microscopy (SEM) micrographs of several different channel geometries are shown in Figure C-2. A B

PAGE 81

70 C D Figure C-2. SEM micrographs of four channel geometries. A) Channel Width, B) Amplitude, C) Frequency and Amplitude SEM images rotated to 45 can be seen in Figure C-3. A B

PAGE 82

71 C D Figure C-3. Angled SEM micrographs of channel geometries. A)Channel Width, B) Amplitude, C) Amplitude, D) Serpentine Good reproduction of the channels can be seen for the eight different channel geometries. Scallops were also observed along the entire height of the molded sidewalls. Figure C-4 shows an example of the scalloped wall. Most sidewalls had a height of 50 m. Figure C-4. Scalloped wall on molded channel geometry

PAGE 83

APPENDIX D GRID PATTERN FOR TRIBOLOGICAL APPLICATIONS A grid pattern of varying sized squares was developed essentially creating channels between the molded square posts for the addition of solid lubricants. The posts were molded to a depth of 50 m to 60 m. The 5 mm x 5 mm sample was divided into 1 mm x 1 mm sections. The number of channels in each direction per section was increased from 2 to 6 in the vertical direction and overall channel area was increased from 10% to 50% in the horizontal direction. The layout for the solid lubricant grid can be seen in Figure D-1. Figure D-1. Grid pattern design for solid lubricants The dimensions of the squares vary from 87 m to 226 m creating channels of sizes 7 m to 126 m. Experiments were conducted at 450 C nominally, at a pressure of 115 MPa for 30 seconds. Excellent results were seen throughout the entire 5 mm x 5 mm molded sample. Figure D-2 shows several SEM micrographs of the molded bulk metallic glass. 72

PAGE 84

73 A B Figure D-2. SEM micrographs of molded bulk metallic glass grid pattern. A) Top view of a portion of the sample, B) tilted zoomed view of portion of sample Several different solid lubricants were added to these molded metallic glass samples. The three lubricants were a graphite based composite, polytetraflouroethylene (PTFE), and gold. These lubricants were added to the molded grid pattern using a hydraulic press. The graphitic powder composite was pressed under room temperature conditions for just a few seconds. The PTFE was pressed under minimal pressure and followed a heating curve that ramped the temperature to 360 C in 3 hours, held for 3 hours at temperature, and then cooled back to room temperature in 3 hours. The gold composite was created by pressing a 100 m thick gold foil into the metallic glass with large pressures at a temperature around 100 C. Figure D-3 shows the solid lubricants in the bulk metallic glass grid.

PAGE 85

74 A B C Figure D-3. Optical microscope images of solid lubricants in molded metallic glass grid. A) Graphite composite, B) PTFE composite, C) gold composite Tribological tests were performed on the PTFE-bulk metallic glass composite with a microtribometer. The microtribometer was outfitted with a rotating stage to run in the pin-on-disk configuration. A 3/32” silicon nitride ball was used as the pin during the tests. A force of 1 N was applied at a sliding speed of 3 mm/s. 10,000 laps were completed for a total sliding distance of 94 m. A picture of the pin-on-disk apparatus used for these tests can be seen in Figure D-4.

PAGE 86

75 Figure D-4. Tribometer used for pin-on-disk composite tests The results show a transient portion of the friction coefficient falling until it reaches a steady-state value. This is due to the transfer film of PTFE forming on the surface of the metallic glass. The plot of friction coefficient can be seen over the course of the sliding distance in Figure D-5. Figure D-5. Friction coefficient plot over sliding distance

PAGE 87

76 An optical microscope picture was taken of the wear track on the PTFE-metallic glass composite. This image can be seen in Figure D-6. Figure D-6. Optical microscope image of wear track in PTFE-metallic glass composite Results from the friction and wear tests show friction coefficients less than 0.2, which is typical for a PTFE composite. There was less than 4 m of wear over the entire sliding distance of 94 m under 1 N of load.

PAGE 88

LIST OF REFERENCES 1. G.L. Benavides, Meso-Machining Capabilities. Sandia Report-SAND2001-1708, Sandia National Laboratories, Albuquerque, NM, 2001. 2. J. Lu, G. Ravichandran, and W.L. Johnson, Deformation behavior of the Zr 41.2 Ti 13.8 Cu 12.5 Ni 10 Be 22.5 bulk metallic glass over a wide range of strain-rates and temperatures. Acta Materialia, 2003. 51: p. 3429-3443. 3. C. Marques, Y.M. Desta, J. Rogers, M.C. Murphy, K. Kelly, Fabrication of high-aspect-ratio microstructures on planar and nonplanar surfaces using a modified LIGA process. Journal of Microelectromechanical Systems, 1997. 6-4: p. 329-336. 4. R. Kondo, S. Takimoto, K. Suzuki, S. Sugiyama, High aspect ratio electrostatic micro actuators using LIGA process. Microsystem Technologies, 2000. 6: p. 218-221. 5. F. Munnik, F. Benninger, S. Mikhailov, A. Bertsch, P. Renaud, H. Lorenz, and M. Gmur, High aspect ratio, 3D structuring of photoresist materials by ion beam LIGA. Microelectronic Engineering, 2003. 67: p. 6-103. 6. C.-M. Cheng, R.-H. Chen, Key issues in fabricating microstructures with high aspect ratios by using deep X-ray lithography. Microelectronic Engineering, 2004. 71: p. 335-342. 7. C.K. Malek, V. Saile, Applications of LIGA technology to precision manufacturing of high-aspect-ratio micro-components and -systems: a review. Microelectronics Journal, 2004. 35: p. 131-143. 8. D.T. Pham, S.S. Dimov, S. Bigot, A. Ivanov, K. Popov, Micro-EDM—recent developments and research issues. Journal of Materials Processing Technology, 2004. 149: p. 50-57. 9. D.S. Miller, Micromachining with abrasive waterjets. Journal of Materials Processing Technology, 2004. 149: p. 37-42. 10. D.P. Adams, M.J. Vasile, G. Benavides, A.N. Campbell, Micromilling of metal alloys with focused ion beam–fabricated tools. Journal of the International Societies for Precision Engineering and Nanotechnology, 2001. 25: p. 107-113. 11. S.T. Davies, B. Khamsehpour, Focused ion beam machining and deposition for nanofabrication. Vacuum, 1996. 47: p. 455-462. 77

PAGE 89

78 12. S. Lipp, L. Frey, G. Franz, E. Demm, S. Petersen, H. Ryssel: Local material removal by focused ion beam milling and etching. Nuclear Instruments and Methods in Physics Research B, 1995. 106: p. 630-635. 13. N.P. Hung, Y.Q. Fu, M.Y. Ali: Focused ion beam machining of silicon. Journal of Materials Processing Technology, 2002. 127: p. 256-260. 14. Y.N. Picard, D.P. Adams, M.J. Vasile, M.B. Ritchey: Focused ion beam-shaped microtools for ultra-precision machining of cylindrical components. Precision Engineering, 2003. 27: p. 59-69. 15. G. Fu, N.H. Loh, S.B. Tor, Y. Murakoshi, R. Maeda, Replication of metal microstructures by micro powder injection molding. Materials and Design, 2004. 25: p. 729-733. 16. L.A. Dobrzaski, G. Matula, A. Vrez, B. Levenfeld, J.M. Torralba, Structure and mechanical properties of HSS HS6-5-2and HS12-1-5-5-type steel produced by modified powder injection molding process. Journal of Materials Processing Technology, 2004. 157-158: p. 658-668. 17. Shinn-Yih Lee, Sintering behavior and mechanical properties of injection-molded zirconia powder. Ceramics International, 2004. 30: p. 579-584. 18. Jan Schroers and William L. Johnson: Crystallization of Zr 41 Ti 14 Cu 12 Ni 10 Be 23 . Materials Transactions, JIM, 2000. 41-11: p. 1530-1537. 19. C.H. Wong and C.H. Shek: Difference in crystallization kinetics of Zr 41 Ti 14 Cu 12.5 Ni 10 Be 22.5 bulk metallic glass under different oxidizing environments. Intermetallics, 2004. 12: p. 1257-1259. 20. Marios D. Demetriou and William L. Johnson: Shear flow characteristics and crystallization kinetics during steady non-isothermal flow of Vitreloy-1. Acta Materialia, 2004. 52: p. 3403-3412. 21. Uwe Koster and Rainer Janlewing: Fragility parameter and noncrystallization of metallic glasses. Materials Science and Engineering A, 2004. 375-377: p. 223-226. 22. Y.J. Kim, R. Busch, W.L. Johnson, A.J. Rulison, and W.K. Rhim: Metallic glass formation in highly undercooled Zr 41.2 Ti 13.8 Cu 12.5 Ni 10 Be 22.5 during containerless electrostatic levitation processing. Applied Physics Letters, 1994. 65-17: p. 2136-2138. 23. Y.J. Kim, R. Busch, W.L. Johnson, A.J. Rulison, and W.K. Rhim: Experimental determination of a time-temperature-transformation diagram of the undercooled Zr 41.2 Ti 13.8 Cu 12.5 Ni 10 Be 22.5 alloy using the containerless electrostatic levitation processing technique. Applied Physics Letters, 1996. 68-8: p. 1057-1059.

PAGE 90

79 24. S. Schneider, P. Thiyagarajan, and W.L. Johnson: Formation of nanocystals based on decomposition in the amorphous Zr 41.2 Ti 13.8 Cu 12.5 Ni 10 Be 22.5 alloy. Applied Physics Letters, 1996. 68-4: p. 493-495. 25. R. Busch, E. Bakke, and W.L. Johnson: Viscosity of the supercooled liquid and relaxation at the glass transition of the Zr 46.75 Ti 8.25 Cu 7.5 Ni 10 Be 27.5 bulk metallic glass forming alloy. Acta Materialia. 1998. 46-13: p. 4725-2732. 26. Q. Wang, J.J. Blandin, M. Suery, B. Van de Moortele, and J.M. Pelletier: High temperature deformation of a fully amorphous and partially crystallized bulk metallic glass. Ann. Chim. Sci. Mat., 2002. 27-5: p. 19-24. 27. Th. Zumkley, S. Suzuki, M. Seidel, S. Mechler, and M.-P. Macht: Superplastic Forging of ZrTiCuNiBe-Bulk Glass for Shaping of Microparts. Mater. Sci. Forum, 2002. 386-388: p. 541-546. 28. Marios D. Demetriou and William L. Johnson: Steady non-Newtonian flow of Vitreloy-1 in continuous extrusion. Materials Science and Engineering A, 2004. 375-377: p. 270-275. 29. A.A. Kndig, M.C., P.J. Uggowitzer, A. Dommann, Preparation of high aspect ratio surface microstructures out of a Zr-based bulk metallic glass. Microelectronic Engineering, 2003. 67-68: p. 405-409. 30. A.A. Kndig, A. Dommann, W.L. Johnson, P.J. Uggowitzer, High aspect ratio micro mechanical structures made of bulk metallic glass. Materials Science and Engineering A, 2004. 375-377: p. 327-331.

PAGE 91

BIOGRAPHICAL SKETCH Jeffrey Bardt was born on March 11, 1981, in Stuart, Florida, to David and Nancy Bardt. He moved to Wellington, Florida, at the age of 5. He moved to Colorado after his sophomore year and graduated from Evergreen High School in May of 1999. Jeff received his Bachelor of Science degree in mechanical engineering from the University of Wisconsin – Madison in May of 2003. He joined the Tribology Lab under the guidance of Dr. Greg Sawyer in August of 2003. Jeff is scheduled to complete his Master of Science degree in August of 2005. 80