A MEMS Valve for the MIT Microengine
by
Xue'en Yang
B.S. Mechanical EngineeringUniversity of California, Berkeley (1999)
Submitted to the Department of Mechanical Engineeringin partial fulfillment of the requirements for the degree of
Master of Science in Mechanical Engineering
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
May2001 V"© 2001 Massachusetts Institute of Technology
All Rights Reserved
Author ..........................Depirth t of Mechanical Engineering
May 22, 2001
C ertified by ................................................. . .. . ........ .................... ..... ....Martin A. Schmidt
Professor of Electrical Engineering and Computer ScienceThesis Supervisor
A ccepted by ................................ ....... . .........................................Ain Sonin
Professor, Department of Mechanical EngineeringChairman, Department Committee on Graduate Studies BARKER
MASSAHUSETS TTUTEOF'TEGHN OLOGY
JUL 1 6 001
LIBRARIES
2
A MEMS Valve for the MIT Microengine
by
Xue'en Yang
Submitted to the Department of Mechanical Engineeringon May 22, 2000, in partial fulfillment of the requirements forthe Degree of Master of Science in Mechanical Engineering
Abstract
A microfabricated, electro-statically actuated, on/off gas valve made of silicon materialhas been designed, fabricated and tested. The valve will be a fuel control component in amicro-scale gas turbine engine. Room-temperature testing results using nitrogen havedemonstrated repeatable valve functions and choked flow characteristics.
MIT has initiated a project to build a micro-scale gas turbine generator for high powerdensity output in applications such as portable power source or micro air vehicles. Forclosed-loop operation, a valve is required to be able to withstand 10 atm upstream pressureunder high-temperature operating environment (700K), and result in a maximum flow rateof 600 sccm while has very low gas leakage rate. These system requirements can not bemet by previously reported MEMS valve, many of which are designed for low tempera-ture or low pressure applications.
The microengine prototype valve comprises of three fusion-bonded SOI wafers. Electro-static-actuation is used to lift the silicon boss actuator supported on four L-shaped tethersand open against high pressure. Polysilicon is chosen as the seat material for high-temper-ature operating environment. The flow path of the valve is designed to be choked andbecause of the micro-scale nature, both viscous and compressible effects are taken intoconsideration in flow analysis with axis-symmetric geometric.
It is demonstrated that at operating pressure of 10 atmosphere, the valve can be opened atless than 150 V with power consumption that is less than 0.04 mW. The gas leakage at thesame pressure is estimated to be less than 0.03 sccm Helium, while the open flow rate is43 sccm (3 g/hr) nitrogen. Commercial fluid analysis package CFD FLUET is used tomodel the flow and very good agreement with experimental data is obtained.
In the future, an array of 20 on/off valves (to obtain 5% accuracy in flow rate) will be usedto accomplish the fuel control scheme of the microengine.
Thesis Supervisor: Martin SchmidtTitle: Professor of Electrical Engineering and Computer Science
4
Table of Contents
Table of Contents .............................................................................................................................. 5List of Figures ................................................................................................................................... 7List of Tables ................................................................................................................................... I IAcknowledgments ........................................................................................................................... 13Nomenclature .................................................................................................................................. 151. Introduction ................................................................................................................................ 19
1. 1 Background .................................................................................................................... 191.2 The Valve Team and Facility ......................................................................................... 211.3 Thesis Organization ........................................................................................................ 22
2. Design Process ........................................................................................................................... 232.1 System Requirem ents ..................................................................................................... 232.2 Design of Prototype Valve ............................................................................................. 26
2.2.1 Design Schematic .............................................................................................. 262.2.2 Design History ................................................................................................... 292.2.3 Design Parameters ............................................................................................. 29
2.3 Summ ary ........................................................................................................................ 413. M icrofabrication ......................................................................................................................... 43
3.1 Fabrication Process ........................................................................................................ 443.2 Fabrication Considerations ............................................................................................. 543.3 W afer Bonding and Diesawing ...................................................................................... 553.4 Summ ary ........................................................................................................................ 58
4. Test Package and Testing Setup ................................................................................................. 594.1 Packaging ....................................................................................................................... 594.2 Testing Setup .................................................................................................................. 61
4.2.1 Electrode Characterization ................................................................................ 614.2.2 System Characterization .................................................................................... 624.2.3 Flow Characterization ....................................................................................... 64
4.3 Summ ary ........................................................................................................................ 665. M odeling and Testing ................................................................................................................ 67
5.1 Electrode Characterization ............................................................................................. 685.2 System Characterization ................................................................................................. 71
5.2.1 Quasi-Static M ode ............................................................................................. 725.2.2 Dynam ic m ode ................................................................................................... 75
5.3 Flow Characterization .................................................................................................... 785.3.1 Valve Function .................................................................................................. 855.3.2 Gas Leakage ...................................................................................................... 88
5.4 Summ ary ........................................................................................................................ 896. Conclusions and Future W ork .................................................................................................... 91
6.1 Conclusions .................................................................................................................... 916.2 Future W ork ................................................................................................................... 92
Appendix A M ask Drawings .......................................................................................................... 97Appendix B Valve Process Flow ................................................................................................. 117
B.1 Top W afer .................................................................................................................... 117B.2 Boss W afer .................................................................................................................. 119B.3 Seal W afer ................................................................................................................... 123
Appendix C M ask Drawings ........................................................................................................ 127
5
6
List of Figures
Figure 1.1. Schematic of the control system in the microengine. (Drawing by Diana Park.) ........ 20
Figure 1.2. Cross section of the microengine. (Drawing by Diana Park.).................................21
Figure 2.1. 3D schematic of the three structural layers that comprise the design the first generation
of prototype valve.........................................................................................................27
Figure 2.2. Cross-sectional view of the three structural layers..................................................28
Figure 2.3. Top view of boss supported by tethers. A) Straight L-shaped. B) Revised L-shaped teth-
ers w ith rounded corners............................................................................................ 30
Figure 2.4. Force balance when a voltage is applied to open the valve in its closed position........34
Figure 2.5. Breakdown voltage of nitrogen as function of pd from the generalized Townsend theory
(original data points from Meek & Craggs [13]).....................................................36
Figure 2.6. Cross-sectional view of the valve seat geometry.....................................................37
Figure 3.1. Fabrication flow for top wafer................................................................................. 46
Figure 3.2. SEM im age of top w afer...............................................................................................47
Figure 3.3. Fabrication flow of boss wafer ................................................................................. 48
Figure 3.4. SEM im age of boss w afer.............................................................................................49
Figure 3.5. Fabrication flow of bottom wafer.............................................................................51
Figure 3.6. SEM images of the seal wafer showing two different magnifications............52
Figure 3.7. AFM photos of polysilicon and silicon surfaces scanning on a 5 by 5 area. The grains
and stripes shown on the silicon photo are the scan line artifacts. .......................... 53
Figure 3.8. Profiles of oxide undercut A) without using step-oxide etch and B) after using step-oxide
etch ................................................................................................................................ 54
Figure 3.9. Results of wafer bonding of the three-wafer stack after annealing. The size of the fringe
is a measure of the local gap between the surfaces caused by particles..................56
Figure 3.10. Valve schem atic as bonded..................................................................................... 57
Figure 3.11. Pictures of the valve die showing the top view and contact pads for the various elec-
trodes in the bottom view . ........................................................................................ 57
Figure 4.1. Assembly of the valve chip package that attains both flow and electrical connections for
testing purpose. Drawing by Alexander Hoelke.......................................................60
Figure 4.2. Test package on an air-floating table........................................................................61
Figure 4.3. The circuit used to actuate the valve using a voltage source and obtain voltage and cur-
rent m easurem ents. ....................................................................................................... 62
7
Figure 4.4. 2D and 3D images of the tethers taken by Wyko as they are deflected to the upmost po-
sition. The tether deflection can be read from the 2D profile...................................63
Figure 4.5. A chart representation of the flow test system showing the nitrogen flow path...........65
Figure 5.1. Cross-sectional schematic of the second generation valve to show the probes and the
four contact pads....................................................................................................... 68
Figure 5.2. I-V curves of the four contact pads measured using HP semiconductor analyzer by
sweeping -100V to 100 V across the same contact pad............................................69
Figure 5.3. I-V characteristics between the two parallel plate electrodes...................................70
Figure 5.4. Lumped model of the electrostatic actuator.............................................................71
Figure 5.5. Plot of equilibrium position of boss as function of voltage using measurement data from
T able 5.1. ............................................................................................................ 73
Figure 5.6. Plots of boss deflection measured using Wyko vs. voltage applied between the top ac-
tuator and the boss for two different dies. Also in the plot is the theoretical curve using
m easured dim ensions................................................................................................ 74
Figure 5.7. Tether deflection as voltage is applied across boss and bottom electrode. .............. 75
Figure 5.8. Step response of the boss with a step voltage of 41 V............................................ 76
Figure 5.9. Undamped natural frequency of the system as a function of the voltage input............77
Figure 5.10. Valve open flow rate measured at different absolute pressures of gas inlet for two dies
w ith different seat geom etry. ................................................................................... 78
Figure 5.11. Simplified flow geometry showing the flow direction......................................... 80
Figure 5.12. Open flow rate as function of absolute pressure at low pressure range for Die I. The
model matches the experimental data well at pressure lower than about 1.5 atm........81
Figure 5.13. Open flow rate as function of pressure in high pressure region. The model neglects vi-
sous effect. .................................................................................................................... 82
Figure 5.14. Velocity contours in flow region for 1.2 atm upstream pressure. .......................... 83
Figure 5.15. Flow profile in A) the throat and B) the channel showing subsonic flow that is fully
developed......................................................................................................................83
Figure 5.16. Mach number and pressure contours for 10 atm upstream pressure. .................... 84
Figure 5.17. At 10 atm upstream pressure, A) shows choked flow in the throat and B) shows pres-
sure drops on the boss along the valve seat............................................................ 84
Figure 5.18. Voltage required to open the valve against applied upstream differential pressure...85
Figure 5.19. Leakage current between the boss and top electrode as voltage is applied for Die I. 86
Figure 5.20. Flow rate at certain pressure as voltage is gradually increased to open the valve......87
8
Figure 5.21. Helium leakage rate of two dies with different seat areas....................88
Figure 6.1. Valves distributed on microengine chip. .................................................................. 93
Figure A.1. Mask: ALIGN, wafer level, with streets..................................................................98
Figure A.2. Mask: TOPELEC, die level, with streets...............................................................99
Figure A.3. Mask: Top_.ELEC, device level.................................................................................100
Figure A.4. Mask: TOPOX_2, die level, with streets ................................................................. 101
Figure A.5. Mask: TOPTHROUGH, die level, with streets ....................................................... 102
Figure A.6. Mask: TOPTHROUGH, device level ...................................................................... 103
Figure A.7. Mask: BOSSOX1, die level, with streets.................................................................104
Figure A.8: Mask: BOSSFEET, device level..............................................................................105
Figure A.9. Mask: BOSSOX-2, die level...................................................................................106
Figure A. 10. Mask: BOSSTETHER, die level, with streets ....................................................... 107
Figure A. 11. Mask: BOSSTETHER, device level......................................................................108
Figure A.12. Mask: BOSSDEEP, die level, with streets ............................................................ 109
Figure A.13. Mask: STREETS, wafer level..................................................................................110
Figure A. 14. Mask: SEALOXIDE, die level, with streets .......................................................... 111
Figure A.15. Mask: SEALOX, device level ............................................................................... 112
Figure A.16. Mask: SEALBACK, die level, with streets ........................................................... 113
Figure A. 17. Mask: SEALSEAT, die level, with streets ............................................................ 114
Figure A.18. Mask: SEALCHANNEL, die level, with streets ................................................... 115
Figure C. 1. AutoCAD layout of the valve package: window plate .............................................. 128
Figure C.2. AutoCAD layout of the valve package: top plate ...................................................... 129
Figure C.3. AutoCAD layout of the valve package: spacer plate ................................................. 130
Figure C.4. AutoCAD layout of the valve package: bottom plate ................................................ 131
Figure C.5. AutoCAD layout of the valve package: pin holder....................................................132
Figure C.6. . AutoCAD layout of the valve package: pin holder..................................................133
9
10
List of Tables
Table 2.1. Microengine system requirements for fuel valve and test valve................................24
Table 2.2. Design parameters of valve geometry........................................................................ 40
Table 3.1. Wafer layout and the status of dies as fabricated.......................................................45
Table 3.2. Types of SOI wafers used for the three structural layers..........................................45
Table 5.1. Planar dimensions of fabricated valve as well as constants calculated from these dimen-
sion m easurem ents. ......................................................................................................... 67
Table 5.2. Pull-in Voltages of different dies for both the top electrode and the bottom electrode.
*Die VI has a tether that is buckled and is not considered in statistics.......................74
Table 5.3. Valve performance for four dies at 10 atmosphere upstream pressure......................87
11
12
Acknowledgments
I would like to thank foremost my advisor Professor Martin Schmidt, who has givenme the opportunity for this invaluable research, who has mentored me throughout theproject, and who has always supported his students in many ways. This project could nothave been accomplished without Dr. Alexander Hoelke, who initiated the design, andtaught me through everything about valves, MEMS, graduate life, etc. Vielen Dank, Alex!
I also owe much gratitude to Professor Alan Epstein, who has always guided methrough the project, and Professor Jeffrey Lang, Dr. Auturo Ayon, Dr. Stuart Jacobsenand Dr. Stephen Umans for their advice on the different aspects of the valve.
It has been a great experience to work with the Schmidt group, namely, Samara, Joel,Ole, Becky, Christine, Sam and Zony. I would like to thank them sincerely for their friend-ship and the help they never hesitate to provide. There is another group that I owe manythanks; they are the people I have spent the most time with for a long while. Tom, Ravi,Dennis, Yoav et al., thanks for sharing the many experiences and providing many helps inthe cleanroom! I owe many thanks to the people in the microengine group, whom I learneverything about microengine from. I need to thank Dr. Xin Zhang specifically for herexceptional help on wafer bonding and her bountiful advice.
I am also grateful to Dr. Vicky Diadiuk and the MTL staff, especially Kurt Broderik,Paul Tierney and Bernard Alamariu for their training and caring. I would like to thank Dr.Carol Livermor for her patience with my questions and what I did to the Microvision sys-tem, Paulo Lozano for his prompt help with helium leak detector, Yifang Guo for his gen-erous assistance in using CFD FLUENT and many others who have helped and shared theexperiences in various ways.
The most special thank goes to Simon, who has added different colors to my life andhas tried to change my perspectives on many things for the past year. Le printemps est ici,Simon!
I owe the most to my parents. I thank them deeply from my heart for giving me theguidance and strength: wuyan ganji.
13
14
Nomenclature
Greek
interfacial tension from air to liquid interface
S permittivity of air
dynamic viscosity
V specific heat ratio
0 water contact angle
Or water contact angle on roughened surface
T time constant
COC cutoff frequency
(On natural frequency
Roman
A area where choked flow occurs in the flow path (the throat)
Aactuator area of actuator
b damping constant
C1 constant used in Townsend's theory
C2 constant used in Townsend's theory
d gap used in Paschen's law
D boss diameter
ds gap distance when spark breakdown occurs
E young's modulus
Es electrical field strength
F capillary force
Ftank pressure force acting on the boss from the fuel tank
Ftether tether force
15
g gap between the two parallel plates in the capacitor
go the original gap between the two parallel plates
go gap between the two parallel plates at an operational point
gmnin minimum gap between the two parallel plate electrodes
h distance of gap between boss and valve seat when valve is fully open
I flow region in the seat area
I-XII die numbers
II flow region in the channel area
k spring constant
Kbend minor loss coefficient in the bend
Kinlet inlet minor loss coefficient
1 tether length
m mass of boss
m mass flow rate
p pressure used in Paschen's law
P fluid pressure
PO stagnation pressure
Po fluid pressure at an operational point
P1 pressure at seat inlet
P2 pressure at seat outlet
P3 pressure at channel inlet
Ptank pressure of tank
Q charge on capacitor
Q1 volume flow rate in flow region I
Q2 volume flow rate in flow region II
R resistance
R universal gas constant
r; radius of water droplet
r, radial dimension in flow region I
r2 radial dimension in flow region II
r; radius of flow channel / inner radius of valve seat
16
ro outer radius of valve seat
t tether thickness
TO stagnation temperature
V voltage
VB voltage applied to boss
VBT voltage applied to boss, with top electrode grounded
VBS voltage applied to boss, with bottom electrode grounded
VTB voltage applied to top electrode, with boss grounded
VSB voltage applied to bottom electrode, with boss grounded
VS voltage applied to bottom electrode
VL voltage applied to landing pad
VT voltage applied to top electrode
VP; pull-in voltage
Vr velocity in radial direction
VS spark voltage used in Paschen's law
w tether width
x1 state of charge on capacitor
x10 charge on capacitor at an operational point
X2 state of boss displacement
x2o boss displacement at an operational point
X3 state of boss velocity
z displacement of boss from the original position
zI vertical axis from seat used in flow region I
Z2 vertical axis from seat used in flow region II
Acronyms
AFM atomic force microscope
BOE buffered oxide etch (HF)
17
BOX buried oxide
CFD computational fluid dynamics
DRIE deep reactive ion etch
DSP double side polished
FEM finite element analysis
FIB focused ion beam
LPCVD low pressure chemical vapor deposition
MEMS micro-electro-mechanical system
SOI silicon on insulator
STS deep reactive ion etcher from Silicon Technology Limit
VLSI very large scale integration
18
Chapter
1Introduction
1.1 Background
This thesis examines the design, analysis, fabrication, packaging and testing of a MEMS
(Micro-Electro-Mechanical Systems) fuel valve for the application of a micro gas turbine
engine.
MIT has initiated a research project on micro power systems, with an aim to build
a micro-scale gas turbine generator to produce high density power. This heat engine is
designed to produce tens of watts of electrical power per cubic centimeter, which is about
ten times the energy density of batteries. The microengine technology can be used to
power micro-air vehicles, micro-fluidic control, miniature cooling systems and micro-
rocket engines. The microengine will be built using semiconductor fabrication techniques
(microfabrication) developed in the microelectronic industry. The structural material is sil-
icon and silicon carbide, which possess good mechanical properties such as high strength
and toughness.
For self-contained applications, the design of a microengine includes a built-in control
system for fuel metering as shown in Figure 1. This control scheme includes a start valve
for initial engine stabilization and a fuel valve for fuel level control in response to the sig-
19
20 Chapter 1: Introduction
nals of pressure, temperature and engine spinning speed. Therefore, the valve must be able
to modulate the flow according to the input control signal.
PressurizedFuel Tank
Fuel Valve Fuel LineStart
User ValveInterface
Start Control ControlSignal
Pressure
Tomarature,
RPM
E Igniter
Micro Gas Turbine
Figure 1.1. Schematic of the control system in the microengine. (Drawing by DianaPark.)
Figure 2 is the cross-sectional view of the microengine, showing the compressor, com-
bustor, turbine supported by air bearings and integrated electric generator. The fuel system
consists of the fuel manifold and the fuel injectors, which supply fuel to the combustor
from the fuel tank (not shown in the picture). A valve is placed between the fuel tank and
the plenum to modulate the fuel flow requested by the controller. Microfabrication tech-
nology constrains the design of the microengine to be a 2D-extruded structure. Therefore,
the valve will be designed to stack on top of the engine.
21
Starter/Generator
Flame Fuel Fuel CompressorHolders Manifold Injectors Diffuser Rotor
\Vanes Bades I
CombustionChamber 7I~i
TurbineNozzleVanes
Figure 1.2. Cross
Turbine ExhaustRotor NozzleBlades Centerline Rotor
of Rotation
section of the microengine. (Drawing by Diana Park.)
1.2 The Valve Team and Facility
The valve team consists of post doctoral associate, Alexander Hoelke, me and an advisory
committee including Professors Alan Epstein, Martin Schmidt, Jeffrey Lang, Dr. Arturo
Ayon, Dr. Stuart Jacobsen, and Dr. Stephen Umans.
Dr. Hoelke initiated the preliminary design of the fuel valve in the summer of 1999.
After I joined him in the Fall, we fabricated the first generation of the prototype valve and
tested its functions. Based on the testing results, I revised the design, fabricated, and test-
ing the second generation.
The valve was microfabricated in Microsystems Technology Laboratories (MTL) at
MIT. MTL possesses sufficient facilities for the valve fabrication, including photolithog-
raphy, DRIE (Deep Reactive Ion Etch), plasma etch, thermal oxidation, nitride deposition
and fusion wafer bonding.
IEMM
Section 1.2: The Valve Team and Facility
Gasnlet Path
22 Chanter 1: Introduction
1.3 Thesis Organization
This thesis will emphasize the design, fabrication, packaging, testing and modeling of the
second generation of the prototype valve.
Chapter 2 presents the design of the second valve based on the preliminary design and
the testing results of the first generation.
Chapter 3 explains the detailed fabrication process and presents the results of fabrica-
tion.
In Chapter 4, the packaging design and testing apparatus needed for different testing
purposes are described.
In Chapter 5, the testing results of the second valve are presented, together with model
analysis for comparison.
The final chapter concludes the work and lays out the future work for the valve.
22 Chapter 1: Introduction
Chapter
2Design Process
This chapter introduces the design of a prototype valve based on the system requirements
of the microengine. Two generations of the valve were fabricated and tested. This chapter
will emphasize the design of the second generation, which has improved function com-
pared to the first one.
2.1 System Requirements
The goal of the microengine is to produce 20 Watts mechanical power, while consuming
about 45 grams per hour of propane fuel. To achieve this goal, the fuel valve must operate
under high temperature and high pressure. Such requirements rule out many currently
available designs of microvalves developed in both industry and academe.
A valve can be categorized mainly by its actuation type and sealing material. Conven-
tional valves used in engines typically employ solenoids for magnetic actuation. On the
micro scale, however, induced magnetic forces are usually too weak to act against high
pressure flows [2]. Successfully commercialized microvalves that use bimetallic and ther-
mopneumatic driving techniques have been reported [3]. These valves often operate under
relatively low temperatures (between 0*C to 60*C in the case of thermally actuated
valves) because of the materials used. Piezoelectric and electrostatic actuation have been
23
widely used in design because of their low power consumption [4][5]. Such drivers
require high voltage input and small deflections in order to produce large actuation forces.
Other actuation techniques include electrolysis-bubble and shape memory alloys [6][7].
Conventional valve sealing materials can be either hard or soft. To achieve extremely low
leakage rate, many researchers have employed soft materials as contact surfaces such as
polyimide or silicone for their high flexibility and fatigue resistance [8]. However, such
materials are not apt for high temperature applications. Other valves conveniently use sili-
con as hard contact surfaces, typically in the form of cantilever and diaphragm [9]. Vari-
ous studies have shown that sealing properties are not only dependent on materials used,
but also on the fabrication process.
Table 2.1. Microengine system requirements for fuel valve and test valve.
Application Microengine Prototype Valve
Throttle
Fluid Propane N2, Propane
Mount Engine Test Package
Temperature 700K 300K
Flow Rate (g/h) 45 2.25 (5% of 45)
A Pvalve, max (atm) 6 9
Ptank (atm) 10 10
Modulation 0.5-1 On/Off
Precision 5% 5%
Time Response ms ms
Shock Resistance 1Og log
Size (cm 2) 2.1 0.25
Mass (mg) 1200 80
24 Chapter 2: Design Process
Table 2.1 lists the system requirements of the microengine fuel valve as well as a pro-
totype valve. The goal of the valve project is to build a prototype valve that satisfies the
design criteria of the microengine and to evaluate the function of the valve in a test pack-
age before integrating it with the engine. In other words, the prototype valve is used to val-
idate the design concept. For simplicity, testing conditions are set at room temperature
while nitrogen is used as the testing fluid. The items in Table 2.1 will be explained in the
following paragraphs.
For a portable engine design, it is desirable to integrate the valve with the engine by
wafer bonding. Doing so also reduces packaging complexity and improves power density.
Integration with the engine, however, will cause rapid heat transfer from the combustion
walls to the valve and as a result, the valve will be heated. For a combustion temperature
at 1600 K, it is estimated that the valve will be operating at about 700 K. The operating
temperature requirement forbids the use of polyimide or elastomer as a sealing material.
Furthermore, as will be explained in the fabrication chapter, the 1100*C annealling tem-
perature of the valve precludes the use of metal for actuation.
In order for the valve to control fuel level, two possible design schemes have been pro-
posed. One is a proportional valve that adjusts flow according to an input actuation signal,
another is an array of on/off valves that modulate the flow by turning on an appropriate
number of valves in response to the input signal. The second scheme is often easier to
design and implement. However, an array of valves requires complex wiring path. In addi-
tion, flow accuracy is limited by the number of elements in the array. For a maximum
engine flow rate of 45 g/h, in order to obtain an accuracy of 5%, 20 on/off valves are
needed, each of which will supply a fuel flow of 2.25 g/h when turned on. The number of
valves in the array will also be limited by the planar dimension of the microengine, which
is 2.1 cm 2 .
Section21l: System Reqient 25
26 Chanter 2: Design Process
The pressure of the fuel tank will initially be at 10 atmospheres. The tank walls form
the external shell of the microengine package and they enclose about 800 cm 3 of fuel.
This package is designed to supply tens of hours of electrical power between refuelling.
The maximum pressure drop across the valve should be about 6 atm. The prototype valve,
however, will be tested at an outlet pressure of one atmasphere, for design simplicity.
It is desired that the valve have a response time in milliseconds and a shock resistance
of 100 g. The size of the final valve is limited by that of the microengine. For the engine to
be efficient, the valve should weight no more than 1.2 g.
2.2 Design of Prototype Valve
Based on the requirements stated above, a first design of the prototype valve has been cre-
ated. The valve uses electrostatic force as the actuation method. This is chosen because the
microengine is integrated with an electric generator and is designed to provide 300 V of
electrical signal. The actuation mechanism can be described as a parallel plate capacitor
acting on a mass-spring-damper mechanical model. We have chosen silicon material for
the valve seat because soft materials are eliminated in our design by high temperature con-
straints.
2.2.1 Design Schematic
The valve is comprised of three 4" wafers fusion bonded together at room temperature and
subsequently annealed at 1100*C. Figure 2.1 shows the 3D cross-sectional view of the
three layers.
The top wafer contains the valve inlet, view-port, main electrode and landing pads.
Gas fuel enters the valve from the tank above via the inlet. A view-port is opened for test-
ing purposes so that a fiber optic sensor can be inserted to detect the motion of the boss.
26
The main electrode is primarily a thin layer of single crystal silicon on top of a 1 gm thick
silicon dioxide layer that acts as insulation. The substrate is a 500 pm thick silicon wafer.
Such a three-layer structure is commonly called an SOI (silicon on insulator) wafer. The
landing pads are holes etched into the silicon dioxide insulation layer. They are used as the
mechanical stop and prevent the boss from crashing onto the main electrode upon pull-in.
The substrate is grounded so as to avoid being electrically floating.
Valve Inlet
Viewport
Main Electrode
Landing Pad
Landing Feet
Boss
Tethers
Valve Seat
Secondary Electrode
Valve Outlet
Figure 2.1. 3D schematic of the three structural layers that comprise the design the first gen-eration of prototype valve.
The middle wafer features a movable boss that is supported by four L-shaped tethers,
which offer the boss sufficient flexibility in the vertical direction. The tethers are made
also from an SOI layer to ensure uniform thickness. We are only interested in the vertical
Section 2.2: Design of Prototype Valve 27
28 Chapter 2: Design Process
motion of the boss and will not consider its in-plane rotation. The boss initially closes
down under 10 atm pressure from the fuel tank. When a voltage is applied between the top
electrode and the boss, electrostatic force will attract the boss to displace upwards and
hence open up the valve. During pull-in, the landing feet on top of boss will touch down
on the landing pads and make electrical contact with the top substrate, which is also
grounded. By doing so, a minimum gap is formed between the two electrodes and a short
circuit is avoided.
The bottom wafer includes three main objects: valve outlet, valve seat, and secondary
electrode. The outlet is a through-hole in the silicon wafer and it leads the gas to the com-
bustor through the engine manifold and the injectors. When the boss moves down, it lands
on the valve seat and hence closes the flow path. The secondary electrode is added for test-
ing purpose. It forms another parallel plate capacitor with the boss and is used to attract
the boss in the downward position. This setup helps to characterize the total boss displace-
ment.
To illustrate the design further, Figure 2.2 shows a schematic cross-sectional view of
the valve with the three wafers bonded together.
Valve Inlet Landing Foot
Tether View Port Main Electrode
Top
Seat
Backup Electrode Valve Outlet Pin Ports
Figure 2.2. Cross-sectional view of the three structural layers.
29
2.2.2 Design History
Two generations of the valve have been successfully fabricated. The first generation used
smooth silicon as the valve seat. This valve allowed us to validate the design and develop
the fabrication techniques needed for the process. Testing results showed that this valve
had fully functioning actuation mechanism and predicted flow characteristics. However,
there were a few drawbacks. First, the actuation force was too weak to open against full
pressure flow. Second, the current leakage between the electrodes worsened after high
voltage usage. Finally, over time, the boss adhered to the valve seat because of stiction.
Based on these problems, a second generation was designed and fabricated. The new
design has improved geometry and fabrication process, and uses polysilicon as the sealing
material. This section will introduce the design concept and the design variables.
2.2.3 Design Parameters
The goal of the preliminary valve is to be able to open against 10 atm pressure with a volt-
age no more than 300 V, and a response time of milliseconds. Therefore, the design
dimensions must be able to satisfy the following conditions:
1. The tethers must have appropriate stiffness to support the boss. Furthermore, the
resonant frequency of the tether-boss structure should be much less than that of the
rotor, which has achieved a speed of 1.2 million RPM.
2. The silicon based capacitor should exert enough attraction force over the boss to
counteract the pressure force when the valve is in the closed position.
3. The dimensions of the gap between the boss and valve seat, as well as the diameter
of the gas outlet, should be chosen such that the gas flow rate is 2.25 g/h.
Section 2.2: Design of Prototype Valve
30 Chapter 2: Design Process
Based on these design considerations, an analysis of the tether structure, parallel
capacitor electrodes and fluid dynamics is carried out to estimate the valve dimensions.
This analysis will then be compared with testing result to check its validity.
Tether Design
A top view of the boss supported by L-shaped tethers are shown in Figure 2.3.A. The
tethers are L-shaped rather than straight because such a design has many advantages. It
allows more linear downward deflection and offers better attenuation over packaging
stresses and thermal stresses [10]. During fabrication, however, we have found that this
boss structure is very fragile and that the fabrication yield is low. Tethers tend to break
along the straight corners due to stress concentration. To resolve this problem, we have
revised the straight corners with round fillets. Also, the width of the tether is increased to
make it stiffer. Such a design is illustrated in Figure 2.3.B).
A) B)
1W
Figure 2.3. Top view of boss supported by tethers. A) Straight L-shaped. B) Revised L-shaped tethers with rounded corners.
30 Chapter 2: Design Process
The tether thickness is determined by that of the SOI layer, which is used in our design
to ensure uniformity throughout the wafer during the etching process. Given the tether
length 1, width w and thickness t, the total force FTether exerted on boss for tether end deflec-
tion z can be expressed as
Ftether Ewt3 Z (2.2)
where E is the Young's modulus of silicon. In our design, we use (100) wafers and the
tethers are oriented at 450 with the primary flat, i.e., in the [001] direction. The Young's
modulus corresponding to this direction is 130 GPa [10].
The tether-boss structure can be modeled as spring-damper-mass mechanical system,
which can be represented by the following system equation,
mz + bz + kz = a (2.3)
where b is the damping coefficient, m is the mass of boss, k is the spring constant of the
tethers, and a is the input signal. The natural frequency of this system is then
O) = -z (2.4)
Time constant t for a step function can be expressed as,
m (2.5)
Squeezed-Film Damping
The damping constant can be evaluated using squeezed-film damping theory, which
applies as the boss moves up and down relative to the stationary, parallel actuator. Consid-
ering the top electrode, we assume the following:
1. The gap between boss and top electrode is much smaller than the boss diameter.
Section 2.2: Design of Prototyp av 31
2. The gas obeys the ideal gas law, is fully developed and isothermal.
3. The boss moves in slow motion such that the gas attains a small Reynold's number,
and viscous effects dominate.
4. There is no pressure gradient in the vertical direction.
5. The no-slip boundary condition can be applied because the ratio of the mean free
path of the gas molecules to the gap is small.
Applying the Navier-Stokes equations and combining with the ideal gas law, the
squeezed-film damping phenomenon can be described by the Reynolds equation [11]:
3
(Pg) = &V2p 2 (2.6)6g.
where P is the pressure of the film that is a function of radius and time, g is the gap
between the two plates, and R is the dynamic viscosity of the fluid film. This partial dif-
ferential equation is nonlinear, and in order to obtain an analytical solution, linearization is
performed near an operating point Po and go. We can then find the pressure response to a
velocity impulse. The total force acting on the plate can be calculated by integrating the
pressure over the plate. A first order approximation of this force in Laplace transform is
[11]
F(s) = b sz(s) (2.7)+ _
where b is the damping constant
b 96gr4 (2.8)g go
oC is the cutoff frequency defined by
32 Chapter 2: Desigzn Process
33
2^2'it g0 P0O =C 2O (2.9)
C 2R 21211r
and r is the radius of the actuator.
This solution demonstrates that b is a function of geometry only. The existence of oC
suggests that the gas behaves also like a spring due to the compressibility effect. At low
frequency, the compressibility effect can be ignored, and b can be viewed as a constant.
However, at high frequency, we have to take into consideration the spring effect. If o is
much larger than the frequency that we are interested in, we can use b as a constant.
There is also a damping effect from the bottom electrode. As the radius of the seat is
much smaller than the radius of the boss, we will ignore this damping effect and consider
only the top electrode.
Force Analysis
Actuation force for the boss comes from a parallel plate capacitor formed by the top wafer
and the boss. When a voltage V is applied between these two electrodes, the electrostatic
force exerted on the boss for a gap distance g is
F uactuator ctuatorV2 (2.10)2g2
where,
e = permittivity of air, which is 8.85x10- 2 F/m, and
Aactuator = Actuation area defined by the enclosed area of top wafer electrode and boss
Under the 10 atm pressure of the fuel tank, the valve is normally closed. The boss
experiences a net pressure force due to the pressure difference in the valve seat area. For
design purposes, the worst case of this pressure force is evaluated in order to estimate the
largest actuation force needed. This case corresponds to the largest pressure drop AP
Section 2.2: Design of Prototype Valve
34Chpe2:DsgPrcs
across the valve channel (9 atm) and the vacuum condition on the valve seat area. This
external pressure force can be expressed as
Ftank = A~r+P tankC(r -r?) (2.11)
where
ri = inner diameter of the valve orifice
r0 =outer diameter of the valve seat
To open the valve from its close position, force balance requires that
Factuator + Ftether > Fank (2.12)
The force diagram of the boss is demonstrated in Figure 2.4.
+jriro
V
Figure 2.4. Force balance when a voltage is applied to open the valve in its closed position.
Minimum GapThe landing feet create a minimum air gap between the top electrode and the boss when
the boss is in its upmost position. Without the landing feet, the boss would crash onto the
top electrode upon pull-in, causing the two surfaces to stick. A proper choice of the mini-
mum gap prevents breakdown between the two electrodes.
The dielectric strength of air is usually reported as 3x10 6 V/m at atmospheric pres-
sure. This value, however, does not hold when the gap between the two plane electrodes is
Chapter 2: Design Process34
reduced to micron dimensions. It has been observed from experiments that the breakdown
voltage depends on the product of the gas pressure and the gap separation, as stated in Pas-
chen's law [13]
V, = f(pd) (2.13)
where p is the gas pressure and d is the gap separation. This law can be interpreted as that
the breakdown voltage at small gaps can be predicted by using data taken at small pres-
sures. Experiments carried out in vacuum, i.e., very low (pd) values, have shown that
many gases exhibit a minimum breakdown voltage. At lower (pd) values, the breakdown
strength of gases will increase. This behavior is captured in Townsend's breakdown the-
ory, which explains the breakdown phenomenon as a number of collision processes that
ionize the gas [13]. At very low pressure or gas separation distances, particle collisions are
less likely to occur, therefore making sparking breakdown more difficult. The criterion for
breakdown is given as
c,%e -( 1 = 1 (2.14)
where c, and c2 are constants that can be obtained by measurements of pre-breakdown
ionization current, c1 is the coefficient representing ionization by electrons, c2 represent-
ing ionization of gas by positive ions, and d, is the gap distance when spark breakdown
occurs. The breakdown voltage then relates to the gap distance by Vs = Esd,, where Es is
the strength of the uniform electrical field. Using this criterion, the breakdown characteris-
tics of nitrogen can be described by the curve shown in Figure 2.5.
Section 2.2: Design of Prototyve Valve 35
36 Chapter 2: Design Process
550-
500-
450-
400-
C: 350-
0
( 300-
250-
200
0 10 20 30 40 50 60
pd (atm-um)
Figure 2.5. Breakdown voltage of nitrogen as function of pd from the generalized Townsendtheory (original data points from Meek & Craggs [13]).
The curve suggests that the minimum breakdown voltage for nitrogen is about 300 V.
At 10 atmospheres, the gap separation at this breakdown voltage is about 1 gm. Experi-
ments at very low gap separation, however, have shown that this breakdown voltage is not
always achievable. Surface contamination or surface roughness could cause the actual
breakdown voltage to be much smaller. In such cases, the electric field can be locally con-
centrated, enhancing gas ionization and resulting in a lower average breakdown field.
Therefore, a larger gap separation will be advantageous. In the design of the valve, a min-
imum gap of 2 gm is chosen.
Flow Analysis
In order to control the mass flow rate, choked flow is designed in the gas flow path. A
magnified cross-sectional view of the valve seat geometry is shown in Figure 2.6.
36 Chapter 2: Design Process
37
Po, To h
Patm
Figure 2.6. Cross-sectional view of the valve seat geometry.
The geometry is axisymmetric so that only the 2D case needs to be considered. For
invisid flow in a duct, the choked condition determines the maximum mass flow rate to
area ratio to be
S1/2(l V -(V + 1)
0a + Y-2 ((1 - 1) (2.15)
where
m = mass flow rate
A = 2nrih, is the area of the flow path where flow is choked
h= distance of gap between boss and valve seat when valve is fully open
PO = stagnation pressure of flowing fluid, same as the pressure of the fuel tank
To = stagnation temperature of flowing fluid
v = specific heat ratio of flowing fluid; 1.13 for propane; 1.14 for nitrogen
R = gas constant; 189 J/kg-K for propane; 287 J/kg-K for nitrogen
Section 2.2: Design of Prototype Valve
38 Chapter 2: Design Process
This equation might be used to estimate the flow rate at high pressure, where the Rey-
nold's number is relatively high. But in the real situation, there will be pressure drops
across the seat and within the channel because of viscous effects. Also, various minor
losses in the flow inlet and the bend have to be considered. For design simplicity, this
equation will be used to estimate the size of the flow path.
StictionIn Figure 2.5, we have shown three forces acting on the boss. However, a stiction force is
omitted from the picture. This section is devoted to describe this force, which is hard to
quantify as it depends on surface roughness, humidity and other factors. Nonetheless, dur-
ing testing of the first generation of the valve, we often observed that the boss easily
adhered to the top or the bottom surface, and could not be released. The second generation
aims to prevent these problems by using rougher material as the valve seat and by reduc-
ing the valve contact area.
Stiction is more well known in the microscopic world and often proves to be detrimen-
tal to MEMS devices. The causes of stiction include capillary forces, Van Der Waals
attraction and electrostatic forces. For a water droplet between two parallel plates, the
attracting capillary force is given as [15]
F = 2ycos0 r2 (2.16)h 1
where y is the interfacial tension from air to liquid interface, 0 is the water contact angle,
h is the gap between two surfaces and r, is radius of the water droplet. This equation says
that the capillary force is proportional to the plate area and the cosine of the water contact
angle. At a contact angle of 90*, the force is zero. To reduce the capillary force, possible
solutions are then to decrease the area of the plate and to increase the water contact angle.
Section 2.2: Design of Prototype Valve
A completely hydrogen terminated silicon surface is hydrophobic with a contact angle
around 90'. When the surface is exposed to air or water, a native oxide forms on the sili-
con surface, causing it to be hydrophilic with a contact angle less than 60'. This contact
angle can be increased by roughening the silicon surface. It is derived from quasithermo-
dynamic model that the water contact angle on roughened surface can be expressed as [16]
cosor = rcos0 (2.17)
where Or is the contact angle of the roughening surface and r is the ratio of the actual area
of roughened surface to the projected area. Therefore, increasing the roughness reduces
the value of cos0 and hence the capillary force. In experiments, however, such an appar-
ent relationship is not always achievable. The advantage of roughening the surface, fur-
thermore, is that it reduces the actual contact area and as a result, the Van der Waals force
and the electrostatic force will also be reduced [17].
Conventional ways to modify silicon surfaces include using focused ion beam (FIB) to
create dimples [18], chemical etching by NH4F [19], or silicon anodization [20]. For a
valve, however, there is a trade-off between the roughness of the valve seat surface and
the leakage rate. Rougher surface produce less flow resistance in closed position and result
in larger leak. Polysilicon is chosen as the seat material because of its rough surface and
the fabrication advantage. Deposition of polysilicon is compatible with VLSI, and the thin
film structure is uniform and stable. Furthermore, roughness of polysilicon can be con-
trolled through deposition conditions.
Chosen Dimensions
Summarizing the structural and flow analysis from above, we chose the dimensions of
various geometry and some resulting constants as listed in Table 2.2. Comparison of the
39
4040Chte2:DinPrcs
dimensions used in the 1st generation and the 2nd generation are also made in the table.
Table 2.2. Design parameters of valve geometry.
Items Units 1st Generation 2nd Generation
Tether:
Tether Thickness t jim 17 (SOI) 17 (SOI)
Tether Width w pim 20 60
Tether Length 1 jim 680 800
Total Spring Constant K N/m 161 300
Boss:
Boss Diameter D gm 670 1080
Boss Mass m Kg 4.1x10 7 10.5x10~
Natural Frequency oo KHz 2.76 2.68
Original Gap g, pm 6.2 4.9
Actuation Gap g gm 7.2 5.4
Minimum Gap gmin Im 2.86 2
Valve Seat:
Inner Radius r; jm 15 18
Outer Radius ro jm 100/200 34/42
Flow Gap h jm 4 3.2
Oscilliary Feature:
Landing Feet Diameter jm 50 30
Landing Pad Diameter jm 100 60
View Port Diameter pm 100 290
The second generation aims not only to increase fabrication yield but also to allow the
valve to open against full pressure. In order to do so, the tether stiffness is increased by tri-
pling the tether width. Furthermore, the actuation area is enlarged while the actuation gap
was decreased, thus resulting in an increased actuation force. The valve seat radius is also
decreased by three times hence reducing the pressure force acting on the boss from the
Chapter 2: Design Process
Section 2.3: Summary 41
fuel tank by about 17 times. Given these dimensions, we would predict that for the second
generation, the valve would be able to open against 10 atm (1.013x106 Pa) at an actuation
voltage of 148 V. Because the flow gap is decreased as a result of the smaller actuation
gap, the flow channel radius is increased from 15 ptm to 18 gm in order to obtain a simi-
lar flow rate.
2.3 Summary
A prototype valve was designed to meet the system requirements of the MIT microengine.
Some design issues are discussed. Dimensions of the valve are assigned based on the pre-
liminary structural, electrical and fluidic analysis.
An on/off valve is proposed as it is more practical to build by microfabrication. Fuel
control can be accomplished by using an array of such valves. Electrostatic actuation is
chosen as the actuation method as voltage can be supplied from the engine generator. A
boss supported by tethers forms the actuation mechanism, which can be modeled as a par-
allel plate capacitor with mass, damper and spring. The flow is designed to be choked at
the valve seat at high pressure. For high temperature application, silicon is used for the
valve seat.
Design of the valve involves several interesting phenomena. The first one is the
squeezed-film damping, which occurs when two parallel plates have relative motion.
Damping turns out to be caused both by viscous (dashpot) and compressibility (spring)
effects. At frequencies much lower than the cutoff frequency, spring effects can be
ignored. The second one is the electrical breakdown between two parallel plates separated
by submicron distance. A minimum voltage appears as the distance decreases further, in
which case, the breakdown voltage is limited by the surface condition. The third one is
41Section 2.3: Summary
42 Chapter 2: Design Process
stiction, which is a function of contact area and water contact angle. Effective ways to
reduce stiction include reducing the area and roughening the contact surface.
This chapter introduces the design concept of the prototype valve. The second genera-
tion of the design differs to the first one by: 1) a revised geometry to increase the net valve
opening force; 2) using polysilicon as the valve sealing surface instead of smooth silicon
to reduce stiction; and 3) improving the fabrication process to reduce current leakage
between the electrodes. The fabrication process will be introduced in the next chapter.
Chapter
3Microfabrication
Fabrication of the prototype valve was carried out in the MIT Microsystem Technology
Laboratories (MTL). The techniques of microfabrication were similar to VLSI, i.e., using
photolithography for mask patterning, and various wet and dry etching methods for cut-
ting exposed geometries. In particular to MEMS fabrication, surface micromachining
refers to the process of making free-standing thin-film structures by use of sacrificial lay-
ers and bulk micromaching refers to the process of etching deep into the substrate [11].
Deep etches are usually achieved by using thick photoresist as a masking material or by
using a high selectivity material such as an oxide thin-film. MEMS microfabrication often
involves bonding two or more wafers together to achieve various geometries. In this case,
the surface of the wafer must be flat and smooth in order for wafer bonding to be success-
ful.
It is worth noting that there are a couple of current technologies which have made the
current valve design feasible:
1. SOI (silicon on insulator) wafers are made by thermally growing oxide on a silicon
substrate and subsequently bonding to another silicon substrate, which would then
be thinned down and chemical-mechanically polished. We used these wafers as the
starting material for two purposes: 1) They possess excellent uniformity as well as
43
surface smoothness for wafer bonding. For this purpose, the SOI layer was used for
tether structures; and 2) the buried oxide could be used as an electrical insulation
from the substrate to the electrode.
2. The high aspect ratio silicon structures in the design would not have been accom-
plished without access to deep reactive ion etching (DRIE). In particular, we used
the time-multiplexed deep etching technique developed by Robert Bosch. This tech-
nique cycles an etching phase (using SF6) and a sidewall passivating phase (using
C4F8to prevent etching of the sidewalls) [12]. With this technique, etching of silicon
as deep as 300 gm with thin walls is possible.
The following section will explain the valve fabrication process.
3.1 Fabrication Process
The valve requires a total of fifteen masks to be patterned on three wafers, four sides, and
approximately twelve shallow plasma etches, four deep plasma etches, three thermal oxi-
dations and two thin film depositions. Fabrication of the second generation differs from
the first one mainly because of the use of polysilicon in the bottom wafer. In this section,
we will discuss only the process flow for the fabrication of the second generation. The
wafer layout and all the masks used in fabrication are shown in Appendix A. Fabrication
details of each wafer are described in Appendix B.
Each wafer layout has twelve evenly spaced devices containing four different designs
of valve seat geometry (the numbering of dies on the wafer is shown in Figure A.1 in
Appendix A). The performance of each die after fabrication is listed in Table 3.1. Note
that because of fabrication constraints, all dies with grooves were not successfully fabri-
cated and as a result, only 6 dies contained functioning valves.
Chapter 3: Microfabrication44
Section 3.1: Fabrication Process 45
Table 3.1. Wafer layout and the status of dies as fabricated.
Die Number Seat Outer Diameter Seat Pattern Fabrication Status
I, VIII, X 34 flat good
II, VII, IX 42 flat good
III, V, XII 34 grooved, 2 rings bad
IV, VI, XI 42 grooved, 2 rings bad
Table 3.2 lists the three wafers used for the three structural layers. The fabrication pro-
cess for each wafer will be explained in detail in the following sections.
Table 3.2. Types of SOI wafers used for the three structural layers.
Structural Layer SOI Thickness Buried Oxide Thickness Silicon Substrate Thickness
gm gm
Top Wafer (SOI) 0.34 1 500
Boss Wafer (SOI) 17 0.4 380
Seat Wafer (DSP) N/A N/A 450
Top Wafer
The top wafer contains the top electrode, landing pads and the view port. The S01 layer is
used as an electrode and the buried oxide as an insulator. Figure 3.1 shows the fabrication
flow.
The major fabrication steps are:
Section 3. 1: Fabrication Process 45
46 Chapter 3: Microfabrication
Silicon Oxide Polysilicon Nitride
Substrate
1. Etch alignment marks Buried Oxide
Sol
2. Plasma shallow etchtop electrode
3. BOE etch oxide step *
Stepped oxide etch--
4. DRIE through etch forview port and gas inlet a1
Landing 7View PortLandingContact Pad to
Pad Top Electrode Landing Pad
Figure 3.1. Fabrication flow for top wafer.
1. Prepare the SOI wafer and etch alignment marks on both sides (mask: ALIGN). The
top side alignment is to be used for aligning to the other two wafers during wafer-
bonding.
2. Shallow plasma etch the SOI layer to define the top electrode area as well as the
landing pad areas (mask: TOPELEC). The top electrode has a diameter of 1100
gm and the diameter of the landing pad is 80 gm. The valve is enclosed in a square
of 1.6 mm2
3. Etch the buried oxide layer in buffered HF (BOE) (mask: TOPOX_2). Instead of
etching through the oxide using the same mask as in step 2, this oxide etch creates a
projected oxide layer (section 3.2 provides further explanation).
4. DRIE to etch through the wafer to open the flow inlet as well as a through hole for
Chapter 3: Microfabrication46
Section 3. 1: Fabrication Process 4
viewing purposes under microscopes (TOPTHROUGH). Diameter of the hole is
290 pm.
Figure 3.2 shows an image of the fabricated top wafer using a Scanning Electron
Microscope (SEM). The white band shown in the picture is the "electrical wire" from the
electric contact located near the edge of the die. The wire is insulated from the rest of the
wafer surface by the oxide thin film below it. We could also see the step-oxide etch by not-
ing the color contrast at the edge of the cylinder.
Figure 3.2. SEM image of top wafer.
Boss Wafer
The boss wafer forms another half of the parallel plate capacitor. It contains the mov-
able boss that is supported by four tethers. The fabrication process is shown in Figure 3.3.
and the process flow can be summarized into the following steps.
47
48 ChaDter 3: Microfabrication
Silicon Oxide Polysilicon Nitride
1. Thermal oxidation, 1 pm;Nitride deposition; Etchalignment marks
2. Wet etch oxide anddefine device area
3. Plasma shallow etchlanding feet, 3.4 pm
4. BOE etch oxide fromback side
5. DRIE etch tether fromback side, using BOXas etch stop
6. DRIE etch boss fromfront side using BOXas etch stop
7. Wet etch oxide to releaseboss; Remove nitride inphosphoric acid hot bath
NitrideThermal OxideSubstrateBuried Oxide
Landing Feet
Bos
TetherContact Pad
to BossContact Through Hole
to Top Electrode
Figure 3.3. Fabrication flow of boss wafer
1. Thermal oxidation of SOI wafer under wet conditions at 11000 C for 1.5 gm.
LPCVD nitride deposition of 0.1 gm. This nitride layer is used as the etch mask for
releasing the tether structure in BOE in the last step. Etch alignment marks on both
48 Chap~ter 3: Microfabrication
Figure 3.4. SEM image of boss wafer.
6. DRIE to etch the boss structure (mask: BOSSDEEP). The buried oxide again is
used as the etch stop.
7. Etch the buried oxide in BOE and release the boss structure using the nitride film as
49
sides of wafer (mask: ALIGN).
2. Plasma etch nitride and then use BOE to etch top oxide layer to define the device
area for landing feet etching (mask: BOSSOX_1).
3. Plasma shallow etch of silicon to create the four landing feet, each with diameter of
30 gm. Etch depth is 3.4 pm (mask: BOSS_FEET_2).
4. Plasma etch nitride and use BOE to etch backside oxide layer (mask:
BOSSOX_2).
5. DRIE to etch the tether from the back side for 17 gm (mask: BOSSTETHER).
The buried oxide is used as the etch stop.
50 Chanter 3: Microfabrication
the etch mask. Remove nitride layers in hot phosphoric acid.
An SEM photo of the as fabricated boss wafer is shown in Figure 3.4. The rough walls
seen in this picture are the result of DRIE process, which etches the side walls slightly.
Bottom WaferThe bottom wafer has the valve seat and flow orifice. The process flow is shown in
Figure 3.5.
1. Thermal oxidation of double sided polished wafer under wet conditions at 1 100 0C
for 0.7 gm after etching of alignment marks on both sides (mask: ALIGN in the
front and mask: STREETS at the back). This oxide thin film is used for two pur-
poses. First, it is the insulation from the bottom electrode and the sealing material,
which is polysilicon. Second, it is used as an etch mask (in replacement of photore-
sist) to etch the bottom electrode.
2. LPCVD polysilicon deposition at 625 C, for 0.95 gm. This polysilicon layer is
used as the seat material.
3. Thermal oxidation on top of polysilicon under wet conditions at 1 100*C for 1 jm.
This process consumes about 0.5 gm of polysilicon and hence the final thickness of
polysilicon is 0.45 gm.
4. Three etches in a row to define the valve seat area and the bottom electrodes (mask:
SEALOX). First use BOE to etch the top oxide, then plasma-etch the polysilicon
thin film, and finally use BOE again to etch the bottom oxide thin film.
50
Section 3.1: Fabrication Process 5
Silicon Oxide Polysilicon Nitride
1. Thermal oxidation,0.7 gm
2. Undoped polysilicodeposition, 0.95 pt
3. Thermal oxidation,1 jm
4. DRIE etch flow outfrom back side, 22
5. Etch top three layeand define seat anbottom electrode
6. Etch top two layersexcept the seat are
7. Etch flow channel,100 jm
8. Etch bottom electrcusing oxide as mas100 sim
9. BOE remove oxide
4 Thermal Oxide4 Substrate
4 Polysilicon
4- Thermal Oxide
nn
let0 gm
rsd
aOR
de
Bottom Electrode A- Valve Seat
Through Hole for Flow Outlet Through Hole forBoss Contact Landing Pad Contact
Figure 3.5. Fabrication flow of bottom wafer.
5. DRIE to etch the flow outlet from the bottom side (mask: SEALBACK). The etch
depth is 240 gm. The reason for this etch is to reduce the etch depth of the small
channel from the front. It is difficult to use DRIE to produce straight walls for wafer-
51
deep features, and especially in this case, where the cylindrical wall of the valve seat
has a thickness of only 16 jm. Furthermore, because of thinning of photoresist at
the corners during etching, the hole diameter tends to expand. Therefore, longer etch
times will result in less accurate geometric dimensions.
6. Etch the top two thin films for nested mask (mask: SEALSEAT). The oxide left on
the top of the valve seat and the bottom electrode will be used as the etch mask for
the DRIE etch of these features later.
7. DRIE to etch the flow channel from the top side for 100 jm (mask:
SEALCHANNEL). After this flow channel is etched, it would be very difficult to
spin on photoresist and expose the next mask evenly. This is the reason why the
oxide is used as the etch mask.
8. DRIE to etch down the bottom electrode as well as to etch through the flow channel
using only oxide as mask. The etch depth here for the bottom electrode is 100 jm.
9. Remove oxide layer using BOE.
Figure 3.6. SEM images of the seal wafer showing two different magnifications.
Chapter 3: Microfabrication52
Two SEM photos of the fabricated seal wafer are shown in Figure 3.7. The one on the
right is an enlarged view of the valve seat area with the bottom electrode. With an etch
depth of 100 pim, straight walls of the channel are obtained using DRIE.
We will now examine the roughness of polysilicon deposited in this process. Figure
3.7 shows the topography of polysilicon surface using tapping mode Atomic Force Micro-
scope (AFM). The average roughness is measured to be 21 nm, and the difference between
the actual surface area and the projected area is 5% as provided by AFM analysis. A com-
parison with smooth silicon is also made in the picture (note that the z scales in the two
images are different). The roughness of silicon shown here is 0.4 nm, and the area differ-
ence is 0.08%. As have been described in Equation (2.17), the increase of actual surface
area in polysilicon results in an increase of water contact angle. Quantitative comparison
in the actual stiction force exerted on silicon and polysilicon surfaces, however, is not able
to be made in this experiment.
NanOsanpe Tapping AFN
Scs e 5.000 ON
scan rate 0.4984 LaNumbser of Sam
X1 000 pa/div2 150.000 nw/Aiv X~-" 1.000 Pm/dIV
Z 20.000 nA4/div
Polysilicon Silicon
Figure 3.7. AFM photos of polysilicon and silicon surfaces scanning on a 5 by 5 Rm2 area.The grains and stripes shown on the silicon photo are the scan line artifacts.
53
Chapter 3: Microfabrication
3.2 Fabrication Considerations
This section will discuss a number of techniques employed in fabrication that are essential
to the valve function, and why some dies failed to be fabricated as expected.
Firstly, we have used the step-oxide etch method to eliminate possible leakage current
between two electrodes. Because etch of oxide in HF is isotropic, undercut of oxide after
etch is unavoidable (Figure 3.8.A). The hydrophilic oxide attracts water molecules and
creates a "shelter" for moisture and debris, which are the two possible sources for shorting
the SOI electrode with the silicon substrate. By using a different mask for the oxide etch,
we can etch the oxide far from the SOI edge and hence reduce the chance of current leak-
age path (Figure 3.8.B). Testing results have proved this technique to be effective.
Oxide Oxidie
A) Oxide undercut B) step-oxide etch
Figure 3.8. Profiles of oxide undercut A) without using step-oxide etch and B) after usingstep-oxide etch.
Secondly, it is very important to protect the surfaces of the wafer for bonding during
fabrication processes. For DRIE etch, the wafer is mounted on a quartz or silicon handle
wafer for through etch. Thick photoresist (10 gm) or oxide thin films are used as etch
masks to protect the front side of the wafer. The back side of the wafer, however, can be
damaged during etching by footing or pitting, which occurs when the ions reflected off the
handle wafer attack the back side. This problem can be avoided by coating the back side
with thin resist. To protect the wafer further, oxide or nitride thin films prove to be useful.
54
55
Finally, the grooved seal geometries were not fabricated successfully. This problem is
caused by the nested-mask process, which requires two oxide thin films to be grown.
Etching of each oxide film results in different degrees of undercut depending on the time
etched. In order for the polysilicon layer to survive the undercut, it can be found that at
least 2.2 Jtm of margin must be used in mask design. Because of over-etch during the pro-
cess, the grooves which had widths of 5 gm and 6 gm failed to survive the etches and
hence resulted in seat-less dies.
3.3 Wafer Bonding and Diesawing
After the three wafers were fabricated, they were bonded together using direct fusion
bonding. When two flat and smooth silicon or thermally oxidized silicon surfaces are
pressed to each other to atomic distances, the interfacial forces such as the Van der Waals
forces and electrostatic forces will attract the two surfaces and form good adhesion. Upon
annealing at high temperature, atoms at the two wafer surfaces migrate and reorient them-
selves to reach a state of minimum free energy, and in doing so, filling up the macroscopic
voids and result in bonding strength at the interface as strong as silicon. This mass trans-
port model is analogous to the sintering mechanism in metal surface [21].
To prepare for bonding, firstly, the wafers must be flat. By our experience, wafer bows
greater than 20 pm usually cause the wafers to fail to adhere to each other. Secondly, the
surface of the wafer must to be cleaned using NH 4 C1 and HCl solutions to remove metals
or any organic contaminants. Direct bonding is most vulnerable to surface contaminants.
Any particles on the surface will create gaps and depending on the height of the particles,
the bond may fail locally or globally on the whole wafer. During cleaning, the wafer sur-
face is also hydrated. This step is important as hydrophilic surface will result in greater
bond energy. The presence of hydroxyl group will attract the water molecules in the envi-
Section 3.3: Wafer Bonding and Diesawing
Chapter 3: Microfabrication
ronment and enhance the hydrogen bonds. Thirdly, the wafers will be contacted by press-
ing to each other. After the alignment contact, the wafer stack is continued to be
compressed under 4 atmospheric pressure for as long as more than 10 hours. This is to
ensure good adhesion. Finally, the wafer stack is annealed at 1 100'C for 1 hour. Bonding
is then complete.
Figure 3.9 shows an optical interference photo of the bonding stack after annealing.
Bonding is shown to be successful in most areas except a few local gaps seen as fringes on
the picture. Since these gaps are located outside of the valve area, they are "harmless" to
the structures.
Figure 3.9. Results of wafer bonding of the three-wafer stack after annealing. The size of the
fringe is a measure of the local gap between the surfaces caused by particles.
Figure 3.10 shows a schematic view of the cross-section of the bonded three-wafer
stack. This schematic shows two contact holes for making electric contact from outside
electronics to the boss and the landing pads. The contact hole for the top electrode is omit-
ted in this picture. It is worth noting that the contact made to the boss is through the tether
layer, and that there is a thin oxide film between the tether and the boss. This means that
56
57
when a voltage is applied to the contact pad, the actual voltage of the boss is floating. This
is the weakness of the current design that is difficult to overcome.
Landing
LandingFeet
Contact Pto Boss
Pad View Port Flow Path
Top Electrode
S V f Bottom Electrode
Tether alve Seat
Flow Outlet Conto L
tact Padanding Pads
Figure 3.10. Valve schematic as bonded.
We diced the wafer stack using a stainless steel blade impregnated with diamond. It is
important to place tapes on both sides of the wafer stack to ensure no water and slurry get
into the valve structure during the operation. Photos of the valve die taken from the top
and bottom sides are shown in Figure 3.11. The bottom view shows the contact pads that
make electrical contacts by using Pogo pins. How this is done will be explained in detail in
Chapter 4.
Figure 3.11. Pictures of the valve die showing the top view and contact pads for the variouselectrodes in the bottom view.
Section 3.3: Wafer Bonding and Diesawing
ad
3.4 Summary
This chapter has introduced necessary details of how the valve is microfabricated. The
fabrication process for each wafer is illustrated and explained. Because of a mishap
between design and fabrication, only six dies out of the twelve are rendered useful valves.
Three fabricated wafers were bonded together successfully, resulting in 100% bonding
yield.
It is also mentioned that the use of silicon-on-insulator (SOI) wafers and deep reactive
ion etch (DRIE) methods have enabled the current design of the valve. Several other tech-
niques employed in fabrication are also essential to the function of the valve. They are
namely: 1) creating an oxide step between SOI and substrate to avoid current leakage
paths; 2) protecting the back side of wafer during DRIE etch with a thin resist, oxide, or
nitride; and 3) using a nested-mask process to etch through the flow channel as well as 100
gm deep bottom electrode. In conclusion, it is found from the valve fabrication experi-
ence that microfabrication in an important sense constraints the design, and therefore a
successful MEMS design must take into consideration the details of fabrication issues.
58 Chapter 3: Microfabrication
Chapter
4Test Package and Testing Setup
For the prototype valve, a package was designed specifically for testing purposes. This
package need to contain both electrical and flow connections, be able to withstand 10
atmosphere pressure and deliver a voltage of 300 V. Different testing setups are needed in
order to characterize the electrodes, system dynamics, gas flow rate, and leakage rate.
4.1 Packaging
In MEMS devices, packaging is known to be costly and as important as the device itself.
Hence, design of MEMS is a process inseparable from the consideration of packaging
issues. For testing purposes, the valve chip has incorporated contact pads for making elec-
trical contacts as explained in Chapter 3.
The housing of the package is made of Plexiglas, as it is simple to machine. The pack-
age assembly is shown in Figure 4.1, and the AutoCAD layout of each plate is included in
Appendix C. The valve chip fits in the middle of a spacer plate, which is made of alumi-
num and lies between the top and bottom plates that have through holes of 2 mm outer
diameter on the side for flow connections. Stainless steel tubes made by Scanivalve will be
fixed into the side holes and sealed by epoxy and make flow connections with the test sys-
tem through Teflon tubing. The window plate of 5 mm thick closes the flow compartment.
59
60Chpe4:TsPakganTetnSeu
O-rings are used for sealing between the chip and the plates. The bottom plate has pin-
holes with a diameter of 1 mm. Pogo pins made of nickel and silver alloy with gold-lined
interior will fit into these holes using instant glue1 . These special probes have a spring
mechanism to ensure ohmic contact with the valve chip. They are installed in receptacles
that are attached to wires and thus enable connection to the outside electronics. The design
of the package provides all electrical access to the valve from the bottom side of the chip.
This setup has been taken into consideration in the design of the fabrication process.
Window Plate
Top Plate
O-ring Groove
Valve
Spacer
Bottom Plate
Pin Holder
Figure 4.1. Assembly of the valve chip package that attains both flow and electrical connec-tions for testing purpose. Drawing by Alexander Hoelke.
Figure 4.2 shows a picture of the assembled testing package with flow and electri-
cal connections. This package was used in all the flow tests. For electrode characterization
where no flow is involved, the top plate was replaced by a thin window plate that had a
hole opened in the middle to allow microscope access to the valve.
1. These Pogo pins are products of IDI, Interconnect Devices, INC.
Chapter 4: Test Package and Testing Setup60
Section 4.2: Testing SetuD 6
Figure 4.2. Test package on an air-floating table.
4.2 Testing Setup
This section will explain the methodology and instrumentation used in each testing setup.
4.2.1 Electrode Characterization
To ensure the electrodes have proper functions, we first characterize the contact resistance
as well as the current leakage between two electrodes.
To obtain I-V curves, an automated Hewlett-Packard semiconductor analyzer is used,
which can be programed to provide a ramp voltage signal from -100 V to 100 V. The data
is graphically displayed and can also be converted to text format by using a special soft-
ware provided by the vendor. This test can be done using a probe station without the use of
the testing package.
61
4.2.2 System Characterization
The purpose of this test is to examine the actuator and how it responds to a voltage input.
We are interested in both the quasi-static response of the system to input voltage as well as
the dynamic response. This can be used to verify the spring constant of the tether, the time
response and the natural frequency of the system.
In tests that involves using the voltage source to actuate the valve, the circuit shown in
Figure 4.3 is used. When the circuit is switched on, a voltmeter is used to measure the
actual voltage across the valve, and a Keithley Picoammeter is used to measure the current
leakage through the valve while the voltmeter is disconnected (to eliminate the current
drawn by the voltmeter). When the circuit is switched off, an RC circuit is used to dis-
charge the voltage across the valve. The voltage and current data are used to estimate the
power consumption of the valve.
O On
Off
Sourge - Valve V Voltmeter
Picoammeter
Figure 4.3. The circuit used to actuate the valve using a voltage source and obtain voltageand current measurements.
To test the quasi-static response, voltage is applied between the top/bottom electrode
and the boss. Boss displacement is measured by using a Wyko surface profilometer, which
is an optical profilometer that renders 3 dimensional images of surfaces. Deflection of
62 Chapter 4: Test Package and Testin2 Setup
Section 4 2: Testing Setun 6
boss can be obtained by measuring the tether deflection. A Wyko measurement displayed
in 2D and 3D image form is shown in Figure 4.4. This image is taken with the boss in its
upmost position.
1.9 mmI
/000-1.2 00
-200-
0.4 2 00-0'100 200 300 400 _ 000 0 '0 700 800
'm 46 22 um 069 um0.0 TI IIR_ 7M.21_um -2.07 um
0.0 0.5 1.0 1.5 2.0 2.5 D: 50 =m -276 m
2D Front View Tether Profile
-2 9 2.5
0.0 .mm
3D Perspective View
Figure 4.4. 2D and 3D images of the tethers taken by Wyko as they are deflected to theupmost position. The tether deflection can be read from the 2D profile.
To test the dynamic response, the Computer Microvision system developed by Profes-
sor Dennis Freeman at MIT has been suggested [22]. This system is designed to visualize
the in-situ motion of MEMS structures to an accuracy of one nanometer by combining the
techniques of light microscopy, video imaging and machine vision. It does so by using a
light microscope to magnify the image and then project it to a camera. To detect the Z
motion, stroscopic illumination is used to take sequence of images at multiple planes of
focus. And at each plane of focus, images are taken at multiple stimulus phases that are in
synchronization with the source signal generated by the computer. This process can be
63
repeated at different frequencies and the images will be analyzed by Computer Microvi-
sion algorithm, which output the motion in three axes as a function of frequency. Using
this algorithm, bode plots can be generated and the dynamics of the system will then be
revealed. The Computer Microvision system has been used in many applications, includ-
ing measuring 3D motion of fatigue structures, mirror alignment in optical system, linear
and nonlinear behavior of a gyroscope and in-plane motion of an MIT tethered-motor.
The source signal of the Computer Microvision system, however, is limited to 10 V
DC. For the valve, this results in displacement that is comparable to the noise. Therefore,
in order to use this system, an amplifying circuit will be needed.
The Computer Microvision system will be a very useful tool to provide information
about the dynamics of the system. Currently, experiments using this system have not been
adequate to make any conclusions. This method, however, will continue to be of interests
for future valve testing.
4.2.3 Flow Characterization
We are interested mainly in three types of flow tests. 1. Flow rate of the valve when it is
fully open at different tank pressures. 2. Voltage needed to open the valve at different tank
pressures. 3. Flow leakage rate when valve is completely closed. A flow chart of the test-
ing setup for the flow tests is shown in Figure 4.5. The setup is part of the fluid control/
measurement system C. C. Lin has used for his microturbine air bearing rig [23].
A Mass-Flo Controller by MKS Instruments is used to measure the upstream flow rate
of the valve. A Honeywell pressure transducer is used to measure the upstream pressure
that is set by the pressure regulator. Data acquisition is done via a National Instruments
GPIB board and the LabVIEW program is used for the PC interface to output the values of
pressure and flow rate. The program is explained in detail in C. C. Lin's Doctoral thesis.
Chapter 4: Test Package and TestinR Setup64
The Mass-Flo Controller has a full scale range of 200 sccm and an accuracy of 1%.
This accuracy, however, will not be enough to detect the leakage rate of the valve. Hence a
Mass-Flo Meter is used to measure the flow rate downstream with a full scale of 50 sccm.
The measurement can be improved by using a more accurate flow meter; however, MKS
mass flow meters have a range limit of 10 sccm and therefore the best accuracy is 0.1
sccm. For very low leakage rate, a helium leak detector will be better.
PCItrae TasueLabview Pressure Power Supply
Pressure Regulator In Flow AtmosphereMKS Mass-Flo MKS Mass-Flo
FlwPah Controller Meter
Nitrogen Valve Package
Figure 4.5. A chart representation of the flow test system showing the nitrogen flow path.
Helium leak detectors are commonly used in vacuum system. Using this tool, helium
leak rates as low as 1 x10- 10 atm-cc/sec can be measured. However, some modification is
needed in order to adapt the hose of the leak detector to the Teflon tubes from the valve
package; this can be done by using Teflon plates. The flow setup is similar to the flow
measurement setup, but instead of nitrogen, a helium source is used to supply the flow,
and the upstream pressure is directly read from the dial. The leak rate is displayed in the
detector, which has an upper limit of l0x 10- atm-cc/sec1 .
1. The helium detector used in this test is 959 Portable Leak Detector manufactured by Varian Vac-uum Inc.
65
66 Chapter 4: Test Package and Testing Setup
4.3 Summary
This chapter introduces the package design for both electrode and flow testing of the
valve. This package is designed to withstand 9 atm pressure drop and high voltage input.
Testing methodology and instrumentation are also described. We are mainly interested in
three types of testing: 1) Obtaining I-V curves and access current leakage in the actuator;
2) Verifying function of the electrode in both static and dynamic modes, and 3) Evaluating
flow characteristics, including flow rate, opening voltage and gas leakage of the valve at
different pressure drops.
Chapter
5
Modeling and Testing
This chapter presents the results for the testing of the actuation mechanism and flow char-
acteristics. These results are then compared with those obtained from the lumped element
and flow models.
Before the tests are carried out, the valve dimensions are measured using Wyko and
Electronic Vision tools. Table 5.1 lists the average measurement values across the wafer
Table 5.1. Planar dimensions of fabricated valve as well as constants calculatedfrom these dimension measurements.
Items Units Designed Measured
Tether:
Tether Height t Rm 17 (SOI) 17.5±0.1
Tether Width w gm 60 55.6±0.4
Tether Length 1 gm 800 800± 1
Total Spring Constant K N/m 300 303±6
Boss:
Landing Feet Height pm 3.4 3.6±0.1
Boss Diameter D pim 1080 1080±2
Boss Height jm 378 378±0.1
Boss Mass m Kg 10.5x10-7 7.97 ± 0.08x10 7
Natural Frequency oo KHz 2.68 3.1 ± 0.1
Valve Seat:
Inner Radius r; jim 20 22.0±0.6
Outer Radius ro jm 34/42 31.1/39.1±0.6
67
68 Chapter 5: Modeling and Testing
as well as the uncertainties, which are the standard deviations of the measurements made.
5.1 Electrode Characterization
The purpose of these tests is to characterize the contact resistance of the metal-semicon-
ductor contacts and the leakage current of the parallel capacitors. I-V curves within each
contact pad are obtained. It is found that for n-doped wafers, the wafer-metal contacts
exhibit diode behavior. The leakage current between two electrodes is also measured.
While the leakage current is very low between the boss and the top electrode, this is not
the case between the boss and the bottom electrode. Ideally, the leakage current should be
extremely low and hence the resistance in the circuit does not matter. However, if the leak-
age current is high, the actuation behavior will be affected.
A schematic of the valve cross-sectional view is shown in Figure 5.1. The four contact
pads are illustrated as those for the boss (which is actually the tether), the landing pad, the
top electrode and the bottom electrode.
Landing Pad: VL= VB
Boss: VB BottomElectrode: Vs
Top Electrode: VT
Figure 5.1. Cross-sectional schematic of the second generation valve to show the probes and
the four contact pads.
Section 5.1: Electrode Characterization 69
Note that the boss and the landing pad are shorted so that the landing pad is not electri-
cally floating. The voltages applied to the boss, the landing pad, the top electrode and the
bottom electrode are VB, VL, VT and VS respectively. In the following sections, VBT
denotes a voltage applied between the boss and top electrode, with the top electrode
grounded, and VTB means the same except that the boss is grounded. (That is, the second
subscript denotes the grounded electrode.) Other cases such as VBS, VSB are similar. When
voltage is applied between two electrodes, the third electrode is usually electrically float-
ing.
To evaluate the contact resistance, I-V curves are obtained by putting two probes to the
same contact pad. Figure 5.2 shows the measurement curves. All wafer substrates used are
lightly doped, and the n-type boss wafer and bottom wafer exhibit the typical Schottky
diode behavior. This diode behavior may affect the actuation voltage of the electrodes.
x1 0-3
e~ -1C.)
Top Electrode- .- Landing Pad
- Boss0 9 Bottom Electrode
Voltage (V)
Figure 5.2. I-V curves of the four contact pads measured using HP semiconductor analyzerby sweeping -100V to 100 V across the same contact pad.
4-
3-
2-
50 -100 50 100
-2-
-3-
-4-
70 Chapter 5: Modeling and Testing
When voltage is applied across the top electrode and the boss, it is found that the elec-
trodes behave differently depending on the voltage polarity. The boss behaves normally
when VBT is applied. If the boss is grounded, i.e., when VTB is applied, it oscillates when it
touches the landing pad. This phenomenon can be described by the I-V characteristics
shown in Figure 5.3.A. In this test, the boss is grounded. When voltage is negative, there is
little current measured. When voltage is positive, there is a high current leakage upon
breakdown, causing the actual voltage between the two capacitors to drop significantly.
The boss therefore releases, and when the voltage drops, the capacitor becomes charged
again, forcing the boss to move up. These experiments suggest that the top electrode
should have a lower potential than the boss in order for the actuator to work.
X 10- x 10-3
1- - I
50 -100 -5 0 50 100 1
- - - - 0.5 ------------ -- - - - 7I-1 . -
TAI
Votlage (V) Votlage (V)
A) I-V curve between boss and top electrode B) I-V curve between boss and bottom electrode
Figure 5.3. I-V characteristics between the two parallel plate electrodes.
The current leakage between the boss and bottom electrode is much more substantial
in both polarities, as shown in Figure 5.3.B. The step-etch approach is not applied to the
bottom electrode, and there might be residuals from fabrication process that act as a cur-
-8 ----------------
----- - ----- 6--------- - -----
--------- - - -
o 76- 1
U)
0
-1
71
rent path between these two electrodes. Because of the diode behavior as illustrated in Fig-
ure 5.2, current leakage is not obvious until voltage breakdown takes place. This could
explain the sudden rise of current at about 60 V in the plot.
5.2 System Characterization
We are ultimately interested in determining the pull-in voltage, the resonant frequency and
the time constant for the actuation system. To do so, we model the system by lumped ele-
ment method using a parallel plate capacitor, and a mass-spring-damper mechanism. In
this section, we will compare the model and the experimental results in quasi-static mode,
and then predict the time response and resonant frequency using the dynamic model.
RFixed Electrode
Z m\ MechanicalVin Stops
T- k b
FixedSurrounding
Electrical Domain Mechanical Domain
Figure 5.4. Lumped model of the electrostatic actuator.
The lumped model of the electrostatic actuator is shown in Figure 5.4. Here, m is the
mass of the boss, k is the spring constant of the tethers, and b is the squeezed-film damp-
ing constant. The two mechanical stops are the landing pads in the top wafer and the valve
seat in the bottom wafer; they restrict the motion of the boss to a certain stroke. R is the
Section 5.2: System Characterization
lumped resistance including the resistance of the power source, the contact pads and the
wafers.
5.2.1 Quasi-Static Mode
In quasi-static mode, voltage is applied across the electrode and the equilibrium positions
of the boss are measured. In the case where there is no current leakage between the elec-
trodes, V = Vn, and hence the value of R is not important. From this measurement, the
pull-in voltage of the electrostatic actuator and the spring constant are determined.
At each equilibrium position, we can equate the electrostatic force with the spring
force, such that
E AV22 = kz (5.1)
2(g 0 - z)
where go is the original gap space between the two electrodes as the tethers are unde-
flected. Using this formula, we can express the voltage as a function of boss displacement
z and by using the design dimensions listed in Table 2.2, a plot of the displacement versus
voltage across top electrode and boss is determined (Figure 5.5). As voltage increases,
there are two solutions for the displacement, but one of the rootshis not stable, and hence is
not the real solution. There exists a maximum voltage at which the displacement con-
verges to a single root in the plot; this point is where pull-in occurs. By setting the deriva-
tive of V with respect to z to be zero, we can find that this maximum voltage happens at
1Zg = 3go. The pull-in voltage is therefore
8kgo (5.2)27eAactuator
This is the minimum voltage required to actuate the valve.
72 Chapter 5: Modelin2 and Testin2
73
6
5
4
0 3
2
0.00 10.00 20.00 30.00 40.00 50.00 60.00
Voltage (V)
Figure 5.5. Plot of equilibrium position of boss as function of voltage using measurementdata from Table 5.1.
Figure 5.6 plots the measured displacements as function of voltage VBT. Also shown in
the plot is the theoretical curve using the measured dimensions listed in Table 5.1. An ini-
tial upward displacement from 0.3 to 0.5 ptm is observed on all dies, which can be caused
by the initial residual stress in the SOI film. In the plot, however, this initial displacement
is trimmed and the zero stress state is set at zero displacement. Two measurements are
made for each die, and it can be seen that the experimental curves are repeatable. The pull-
in characteristics of 5 different dies are listed in Table 5.2. The table shows the pull-in
voltages from both the top electrode and the bottom electrode, the maximum boss dis-
placements of each stroke, as well as the spring constant. The spring constant is calculated
by using Equation (5.2) and uses g as the sum of the average maximum upward displace-
ment and the landing feet height. From the table, we can estimate that the spring constant
of tethers as fabricated is 316 ± 32 N/m, which is within the theoretical value of 303 N/m.
-- -- --- - ---- -- - - - - - --- - -NN
Unstable-- - - --- - - - --- --- - -- - - - - --- - - - - - - -
------------- -- --- -- ---Pull-In
Stable
Section 5.2: System Characterization
Table 5.2. Pull-in Voltages of different dies for both the top electrode and the bottom electrode.
*Die VI has a tether that is buckled and is not considered in statistics.
Die Initial Tether Vpi of Top Maximum Spring Vpi of Bottom Maximum
Number Deflection Electrode Upward Constant Electrode Downward
Displacement Displacement
(V) (N/m) (V)
I 0.21 36.2 2.65 280 162 0.34
IV 0.21 39.1 2.79 300 141 0.36
VI* 0.36 45.5 2.89 385 180 0.41
V 0.35 40.5 2.70 340 108 0.39
XI 0.31 36.2 2.31 345 162 0.29
Average 0.29 38 2.7 316 140 0.36
STD 0.07 2 0.2 27 22 0.04
Notice from the table that the pull-in voltage of the bottom electrode is significantly
larger although the gap is much smaller. Figure 5.7 shows a typical tether response using
the bottom electrode. Although the theoretical pull-in voltage is 17 V, the actual applied
3.
S 2 --- - - - - - - - - - - - - - - - - - - - - - - - - - - - - -De#
r 2.5---- L -4---------------- ------- D
00
2 1. ----------------- -------------- ---------rtI I I De A r
O i Ix Die #1
II * Die#40
W 0.5 -------- --- - -- - - - - - - - -
0 I I
0 10 20 30 40 50
Voltage Applied to the Actuator (V)
Figure 5.6. Plots of boss deflection measured using Wyko vs. voltage applied between the
top actuator and the boss for two different dies. Also in the plot is the theoretical
curve using measured dimensions.
Chapter 5: Modefing and Testingz74
75
voltage VBS has to be much larger because of current leakage (as shown in Figure 5.3) that
causes large voltage drop. The pull-in voltage differs significantly among dies, due to the
unknown resistance that puts the two electrodes in electrical contact.
) 2 4P 69 89 QO 1
-V o -- - - - --F------
Voltage (V)
-- Theoretical
x Die XI
Figure 5.7. Tether deflection as voltage is applied across boss and bottom electrode.
5.2.2 Dynamic mode
Taking into consideration of dynamics, we could represent the lumped model of the elec-
trostatic actuator in Figure 5.4 by a set of state-space equations.
The governing differential equation for the electrostatic actuator is
mz+bz + kz = -V2 (5.3)2(go - z )2
V(go - Z)The charge of the parallel capacitor can be expressed as Q= M . If we take the
charge Q, the boss displacement z and velocity v as the three state variables x1, x2 and x3 ,
and take Vin as the input, then we can represent the system in state-space form,
C
0
-0.1-
-0.2
-0.3
-0.4
-0.5
-0.6 -
-0.7
Section 5.2: System Characterization
- Vi X1 (go - X2) 54x1 = - -(5.4)X R EA
X2 = X3 (5.5)
2
X3 = - kx2 - bx 3) (5.6)
We are interested in obtaining the time response to a step voltage input of the system.
If we assume that the electrical domain has much faster time response than the mechanical
domain, which is usually the case, then V=Vin. Matlab Simulink can be used to obtain the
step response of the system. Figure 5.8 shows displacement of the boss when a step volt-
age slightly greater than the pull-in voltage is applied. In this plot, R is estimated to be 13
KQ, and b is evaluated using Equation (2.8) at the position where the gap between boss
and top electrode is minimum. This overestimates b and gives the time constant of the step
response to be 4 ins. It can also be shown that the cutoff frequency in the squeezed-film
damping model is two orders of magnitude greater than the natural frequency of the
spring-mass system. Therefore, we can neglect the spring effect in our damping model.
x 10-631
2.5
2
1.5
1
0.5
00
Figure 5.8. Step
0.005 0.01 0.015 0.02
Time (sec)
response of the boss with a step voltage of 41 V.
Chapter 5: Modeling and Testing76
We can also linearize the above system about an operating point. By using the Jaco-
bian matrix, the linearized system can be written as
(go -x 2 o) x 10 0 - --
8xI E A LA 6x 1
8X2j 0 0 1 8x2 + 1 6V, (5.7)
x6 _k bo 6 X
mEA m m
where at the operating point, x 2o is found by Equation (5.1) and choosing the stable
Vino(g -x 2 o)root, x10 = A , and bo can be calculated by Equation (2.8). Using this model,
we can obtain a plot of the undamped natural frequency of the system versus the input
voltage, as shown in Figure 5.9. It is observed that the resonant frequency shifts as the
input voltage increases. As the voltage is approaching pull-in, the frequency drops rapidly.
Such phenomenon is often called "spring softening." At zero input voltage, the natural fre-
quency is 3.1 KHz, the same as the spring-mass system.
3.5-
1. -- ----- - ---- ------ - ----- -------
U.
Z - - - - - - - - - - - - - - - -
00 10 20 30 40 50
Voltage Input (V)
Figure 5.9. Undamped natural frequency of the system as a function of the voltage input.
Section 5.2: Systemn Characterization 77
5.3 Flow Characterization
At the fully opened position, the flow rate of the valve is measured at different upstream
pressures and compared with analytical and finite element results. At low pressures, the
flow can be considered as incompressible as the Reynold's number is low. At high pres-
sures, however, viscous effect can not be neglected. As analytical result is difficult to
obtain, flow analysis is done by commercial FEM package CFD FLUENT A simplified
geometry is able to predict the flow with reasonable accuracy. Where choked flow occurs,
a series of normal shock waves are observed.
When the valve is fully opened, its flow rate is observed to be rather linear with the
differential pressure. Figure 5.10 plots the testing results of two dies with different outer
seat geometry. Die IV has a seat radius 8 gm bigger than Die I. This results in more flow
resistance and hence slightly less flow (by about 2 sccm from the plot). However, such a
difference is usually hard to distinguish from that caused by uncertainty in dimensions.
Also plotted in the figure is the flow rate obtained using CFD FLUENT, which shows very
60
50 ----- -L - - j ----- ---- IL- --- IL ----
I40-- --- --- -- ------ - ------I II
1 1 Die 1, Test 1
50 I A
30 ---- --- - -- - --- -- A- --- -- - ---- - 9 Die 1, Test 2
cc I
Die IV
0 20 --- --- ---- - - ---- ------ 7- -- -- CFD Model
00 2 4 6 8 10 12
Pressure (x 105 Pa
Figure 5. 10. Valve open flow rate measured at different absolute pressures of gas inlet fortwo dies with different seat geometry.
Chapter 5: Modeling and Testing78
79
good agreement with the experimental data. At 9 atm differential pressure, the volume
flow rate for the smaller seat geometry is about 43 sccm, corresponding to 3 g/hr. This
flow rate is larger than desired. The difference is caused by the isentropic model used to
choose the design dimensions, as well as the fabrication error. The testing and modeling
results will be further discussed. All modeling results are obtained by using the measure-
ments values listed in Table 5.1, and for the seat, the smaller diameter is used.
Low Pressure Region
Flow at very low pressure is examined first as solutions can be obtained analytically.
When the Mach number is less than 0.3, the compressibility effect can be usually
neglected. A simplified flow geometry can be used for this analysis as drawn in Figure
5.11. Assuming steady state, fully developed flow, and neglecting gravity effects, we can
write the Navier-Stoke's equation in region I as
2
0 = + r (5.8)r az 2
with boundary conditions:
Vr(zj = 0) = 0
Vr(zi = h) = 0
Similarly, in region II,
P = [ r Vz (5.9)aZ2 r2 r2 2
with boundary conditions:
JVz2(r2 =) = 0 (5.10)3r2
Vz2(z2 = 0) = 0 (5.11)
Section 5.3: Flow Characterization
80 Chapter 5: Modeling and Testing
Po,To h z I 2P3
*z2
II
Seat Profile
atm
Figure 5.11. Simplified flow geometry showing the flow direction.
Solving the above equations, we can obtain the volume flow rates in the two regions
as:
region I:
(P 1 -P2)th'Q, = (5.12)
6gln -or.
region II:
(P3 - P at,,,) iQ2 = (5.13)8gl
We also need to take into consideration of the minor loss at the entrance and the bend.
For sharp-edged entrance, we choose Kent = 0.5, and for a sharp bend, Kbend = 1.1 [24]. A
Matlab program is set up to do the calculations. It is found that at pressures lower than
about 1.5 atm, the flow rate calculated complies quite well with the experimental data, as
shown in Figure 5.12. This value corresponds to a Mach number of about 0.5.
Chapter 5: Modelingy and Testingy80
Section 5.3: Flow Characterization 81
16-
14 --- ------------ ----- --- -
12 - --- -- ------ ------ - - ----
0 10------------------------ --- ------------ Experimental
2 8-------------------- ------ ------------ Model
6 -- - - -- ------- ------- - - -- --- -
0U: 4 ----------- ------------- IL---------
2 ---------------------- L
0 I0 1 2 3 4
Pressure (x 10 5 Pa)
Figure 5.12. Open flow rate as function of absolute pressure at low pressure range for Die I.
The model matches the experimental data well at pressure lower than about 1.5
atm.
High Pressure Region
At 10 atmospheric pressure, the Reynold's number reaches 1500. Although viscous effect
may not be negligible, it would be interested to see how the isentropic model compares to
experimental results.
Assuming isentropic, choked flow, Equation (2.15) can be used to calculate the flow
rate as a function of stagnation pressure. Figure 5.13 compares the calculated values with
the experimental data. It is worth noting that using the isentropic model, we could predict
the flow rate with a simple formula within 8 sccm. Therefore, it is a quick way to access
the design dimensions.
82 Chapter 5: Modeling and Testing
70-
60 ------- -------- - -------
E 40 -------50-- ------
,u- 4 0 -- - - - -- - - - -- - - - --- Experimental
30- -------------------------------------- Model.220 -----------LL
10 -- ------------------- -- ------
06 8 10 12 14
Pressure (x 105 Pa)
Figure 5.13. Open flow rate as function of pressure in high pressure region. The modelneglects visous effect.
CFD FLUENT Model
It is been shown in Figure 5.10 that the CFD FLUENT results are in good agreement with
the experimental data. Furthermore, at low pressure, laminar and fully developed flow is
observed as predicted, and at high pressure, normal shock waves are present.
At 1.2 atm upstream pressure, the velocity distribution in the flow region near the seat
area is shown in Figure 5.14. Note that the geometry drawn is rotated 900, and therefore
the flow inlet is from top and outlet to the right. The contours show rather smooth transi-
tion of velocity. At the throat of the flow path, the Mach number only reaches 0.26, as
revealed in Figure 5.15.A. Downstream in the channel, the flow is seen to be steady and
fully developed. Figure 5.15.B shows half of the parabolic velocity profile of a typical
Poiseuille flow.
83Section 5.3: Flow Characterization
9.31e+01
8,38e+01
745e+01
4.66e+O1
3.72e+01 c o i wro r.2.tmustea.resue
2,79e+01
1t860+01
9.31e+00
O.0O+00
Figure 5.14. Velocity contours in flow region for 1.2 atm upstream pressure.
EzJU
0.3
0.25
0.2
0.15
0.1
0.05
n
0
50-
40-
30-
20 -
10-
n
0 1 2 3 4Position from Seat (0) to Boss (3.2 gm)
A) Mach number profile in throat
Figure 5.15. Flow profile in A) the throat and B)fully developed.
0 5 10 15 20
Distance from Center Line (gm)
B) Velocity profile downstream in channel
the channel showing subsonic flow that is
For a 10 atm upstream pressure, the flow patterns become more interesting. Figure
5.16 shows both the Mach number and the static pressure contours of the flow region. The
flow is choked at the outlet of the throat as expected. The Mach number profile plotted in
- - - - - - - - -
----------------
--- -- -------- - -- -- - -- -- - --
-- -- - -- ----
* I I
I v
84 Chapter 5: Modeling and Testing
Figure 5.17.A further demonstrates this. As flow suddenly expands in the channel, flow
separation is observed near the wall region. Not far from the throat, a normal shock clearly
dominates the flow pattern, and result in a large pressure gradient. Downstream, the
boundary layer thickens and the shocks following are not as obvious. The flow gradually
diverges to subsonic before reaching the outlet of the channel. Such flow pattern resem-
bles the case of a supersonic nozzle.
3060+00 1010+06
2.76e+00 .130+05
2AS.e+O0 8.13e+05
2 1 4e-+O 713e+05
1.84+00 &13e+05
115*+W 5.13e+05
1.2f*004.13o.-0S
a, 13".01 3.13e+05
6,13e-1 m o m2.13e+05
3,06"01 1.13e+05
1,760-6 1.32e+04
A) Mach number contour B) Pressure (Pa) contour
Figure 5.16. Mach number and pressure contours for 10 atm upstream pressure.
1.2 12
0
E
0. -- - - - - -- - - - - -4
040.0 1.0 2.0 3.0 4.0 0 40 80 120 160 200 240
Position from Seat (0) to Boss (3.2 gm) Position from Center Line (gm)
A) Mach number profile in throat B) Pressure distribution on boss
Figure 5.17. At 10 atm upstream pressure, A) shows choked flow in the throat and B) showspressure drops on the boss along the valve seat.
85
Figure 5.17.B shows the pressure distribution on the boss. It is seen that pressure drop
along the seat area can be approximated by linear profile, and the pressure drop in the
channel area has a rather small variation around 6 atm.
5.3.1 Valve Function
In this section, we will examine the voltage and power required to open the valve against
an applied pressure. This ultimately demonstrates how the valve functions.
The current valve closes at a very small pressure drops. In this test, a step voltage is
applied, and the lowest voltage that fully opens the valve is recorded. Figure 5.18 plots the
testing results of Die I. The opening voltage that is predicted using the worst case scenario
as depicted by Equation (2.11) is also plotted. As expected, the actual opening voltage is
less than the designed value. At 10 atmospheres (132 psig), the opening voltage is 136 V.
180160 ---- - -------- --- --- |140 -------------- ------ -- ------
120-------------------------- ----- L
4) 100 --------+------ ---- ----- - - -- ------
0 2 4 6 8 10 12
Pressure (x 1O5 Pa)
Figure 5.18. Voltage required to open the valve against applied upstream differential pres-sure.
Section 5.3: Flow Characterization
86Chpe5:MdlnanTetn
To access the power consumption, the leakage current between the boss and the top
electrode is measured during flow operation and displayed in Figure 5.19. The curve
shows the characteristic shape of ionized current between two parallel plate electrodes. As
mentioned before, current leakage has been a major problem in the 1st generation of the
valve design. We have solved this problem by creating the step-oxide etch as discussed in
Chapter 3.
x 10-90.6
0.3 ----- --------- - - - - -- -- -- ---
I I
0.5 ----------- I--- -
0.1 ------
00 20 40 60 80 100 120 140 160
Voltage Across Boss and Top Electrode(V)
Figure 5.19. Leakage current between the boss and top electrode as voltage is applied forDie I.
It is shown that at 136V, the leakage current is 0.5 nA, resulting power consumption of
68 nW. The opening function has been very repeatable at different times over more than
one hundred cycles. Table 5.3 lists the performance of other tested dies. All dies have been
tested under voltage as high as 300 V and no voltage breakdown is observed. Die VI is
discarded because its current leakage is too high (0.8 mA at 300 V) and hence not able to
open the valve. The table reveals that opening voltage differs more than 10 V for different
dies and current leakage in particular varies significantly. Nonetheless, we could conclude
Chapter 5: Modeling and Testing86
that the valve can be opened with voltage less than 150 V, and it consumes power less than
0.04 mW.
Table 5.3. Valve performance for four dies at 10 atmosphere upstream pressure.
Die Number Seat outer Opening Flow Rate Current PowerDiameter Voltage Consumption
gm (V) (sccm)/(g/h) (nA) (nW)
I 31.1 136 45.1/3.08 0.50 68
X 31.1 126 41.6/2.84 13.3 1,683
Average 131 43.4/2.96
IV 39.1 143 43.6/2.97 0.61 87
XI 39.1 142 43.1/2.94 270 38,232
Average 142.5 43.4/2.96
It will be interesting to observe the pull-in phenomenon at an upstream pressure. Since
the boss deflection can not be obtained directly during flow tests, the flow rate is instead
measured. Figure 5.20 shows the experimental curves for Die I at four different pressures.
At higher pressures, the valve pulls in much more suddenly and hence a gradual increase
in flow is difficult to record.
0LL
0
E
12
10
8
6
4
2
00 20 40
Voltage (V)
60
-- 4.9 gpsi
_e10.9 gpsi
-*-20 gpsi
--- 29.8 gpsi
80
Figure 5.20. Flow rate at certain pressure as voltage is gradually increased to open the valve.
---- ----- T ---- -------- ----- - ---------
-- -------- - ------- -- - ------ ---------
-- - - - -I - - - - I - - - -
- - - - - - - -
Section 5.3: Flow Characterization 87
88
5.3.2 Gas Leakage
Gas leakage when the valve is closed is of fundamental interest. It can be used to deter-
mines whether the valve is suitable for the desired applications.
The leakage rate measured using a 50 sccm flow meter reads 0.01 sccm, which is the
minimum measurement the meter could detect. In order to measure the small flow more
accurately, a helium leak detector is used. Because of the upper detection limit of 104
cm 3 /s of the particular detector used, the upstream pressure is limited to be less than 2.4
atm (20 psig). Two dies with different seat geometry are tested. The results are plotted in
Figure 5.21.
It can be determined from the trend of the two curves that the larger seat area results in
smaller gas leakage. A leakage model, however, is difficult to obtain, because the surface
roughness of the polysilicon is on the same order of magnitude with the gas mean free
path, and hence the fluid can not be treated as a continuum. If we assume that the flow rate
x 10-3
(Die I
7
----------------- -. 4 - - - - - --- - - - - - - - - Small Seat
cc (Diel1)3 A-- - - - -- - - -L- - - - -- .- -- --
Large Seat2 ------- ------ '------- ------- (Die VI)
00 5 10 15 20
Pressure (psig)
Figure 5.21. Helium leakage rate of two dies with different seat areas.
Chapter 5: Modeling and Testing
is linear at higher pressure, we could estimate that at 10 atmospheres, the valve has leak-
age rates of 0.03 sccm for the small seat and 0.02 sccm for the large seat, which gives per-
centage leak of 0.07% and 0.04%, respectively.
5.4 Summary
Various testing results and corresponding analysis have been presented in this chapter.
They can be summarized as follow.
1. Electrodes
" I-V curves of the contact pads show Schottky diode behavior for n type wafers. This
has caused the actuator to behave abnomally in one polarity. In the future, this can be
avoided by using heavily doped p type wafers.
- Attributing to the step-oxide technique, very little current leakage is observed
between boss and top electrode. Since this technique is not applied, the bottom elec-
trode shows substantial leakage.
2. System
- Equilibrium position of boss at applied voltage is measured using Wyko, and the
experimental value corresponds well with the quasi-static model. Average pull-in
voltage is found to be 38 V and the spring constant is 316 N/m.
* Using lumped element method, the dynamic model of the electro-static actuator is
obtained. The model predicts that the time constant for a step response for this sys-
tem should be less than 4 ms. The system also exhibits "spring softening" effect, i.e.,
its resonant frequency drops as the input voltage approach pull-in voltage.
3. Flow
- In the low pressure region, the low can be considered incompressible and hence an
analytical model is obtained. The model complies well with experimental data.
Section 5.4: Summary 89
- In the high pressure region, viscous effect may not be neglected. Compared to the
isentropic model, the experimental data displays about 15% less flow rate. However,
the simple formula of isentropic chocked model is a quick access in the design of the
dimensions.
- CFD FLUENT is used to analyze flow characteristics. Laminar, fully developed flow
pattern is observed at 1.2 atm upstream pressure as predicted. As pressure is
increased higher, chocked flow occurs, and a set of normal shock waves appear in
the flow path, resembling the case of a supersonic nozzle.
4. Valve Function
- At 10 atm, the valve can be opened at voltage less than 150 V in average, while con-
suming less than 0.04 mW of power.
5. Gas Leakage
" Gas leakage using helium detector is estimated to be less than 0.07% of the full flow.
Larger seat diameter results in less gas leakage.
" Depending on the requirement of the application, gas leakage might have to be
improved. This can be done by using smoother seal surface, or increasing the seat
area. But the trade-off between leakage and stiction forever exists.
90 Chapter 5: Modelin2 and Testin2
Chapter
6Conclusions and Future Work
6.1 Conclusions
The MIT microengine prototype valve has been fabricated, tested, and found to be fully
functional. Two generations of the prototype valve were built and the second generation
was improved based on the testing results of the first one. This thesis emphasizes the fab-
rication and testing of the second generation.
The microengine valve employs electrostatic actuation and uses silicon material as the
valve seat. The actuation mechanism can be described using a parallel plate capacitor and
a lumped spring-mass-damper model. The flow is designed to be choked at the seat. Fabri-
cation of the valve is made possible using SOI wafers and the DRIE process.
The first generation of valve has demonstrated functional electrodes and choked flow
characteristics. However, it fails to function as a valve in three aspects: 1) the actuation
force is too weak to open against pressure force; 2) current leakage is high; and 3) the
valve adheres to a surface, rendering any actuation force useless. Design of the second
generation aims to resolve such problems. First, actuation area is increased and the valve
seat diameter is reduced in order to increase the net opening force. Second, a step-oxide
etch is employed where two electrodes are separated by a thin layer of oxide, preventing
91
undercutting between the two conducting layers. This method has eliminated current leak-
age along the edges. Third, instead of silicon, which has a very smooth surface, polysili-
con is chosen as the sealing material, becuase it has a much rougher surface. The
advantage of using polysilicon, however, is not solely justified. Reduction of the seat area
not only has greatly diminished the pressure force acting on the boss, but also decreased
stiction forces. Nonetheless, use of polysilicon is an alternative for fabrication; deposition
of polysilicon gives more flexibility in choice of thickness compared to using an SOI
wafer.
The new valve is shown to have met most of its specifications. At 10 atmospheres, the
valve can be opened with less than 150 V and consumes less than 0.05 mW of power. The
flow rate at this pressure is 3 g/h. The opening function is very repeatable. Stiction has not
been observed even at 100% ambient humidity. The time response of the valve to a step
voltage is estimated to be in milliseconds. The gas leakage when the valve is fully closed
at 10 atmospheres is estimated to be less than 0.03 sccm.
CFD FLUENT is used to model the flow and has very good agreement with the exper-
imental data. At low pressures, the flow is laminar, fully developed and compressibility
effect can be neglected. At high pressure, flow is choked at the seat and series of shock
waves are predicted downstream as would be observed in supersonic flow in nozzles.
6.2 Future Work
The prototype valve has demonstrated the feasibility of the design of an MIT microengine
fuel valve. It will be of great importance to perform cyclic tests and have a better appreci-
ation of the repeatability of the opening function. To incorporate this valve design into the
microengine will required more experimental work. The prototype valve could also be
used in other application, such as in the microrocket that MIT is developing.
92 Chapter : Conclusions and Future Work
For the cyclic testing, an electrical switch can be used together with the LabVIEW pro-
gram to access the cyclic performance of the valve. The failure mode can be either
mechanical fatigue, or electrical breakdown. Therefore, this test could also be of interests
to study fatigue in silicon or electrical breakdown phenomenon.
The microengine will require an array of valves (designed to be 20) in order to modu-
late the flow and accomplish the control scheme. Distribution of valves on the
microengine valve chip can be planned as shown in Figure 6.1, where the array of valves
is drawn to show the relative position in the engine plenum. An extra valve is set to be the
start valve, which will require a much larger initial flow rate to start the engine. This valve
can be made by using the same design but varying the design dimensions to fit the flow
requirements. Wiring also needs to be carefully designed on the chip in order to effectively
switch on the correct numbers of valves.
L1, 0 N
Figure 6.1. Valves distributed on microengine chip.
Section 6.2: Future Work 93
94 Chapter: Conclusions and Future Work
To expand the applicability of the valve, it will be interesting to carry out testing in liq-
uid fuel. The microrocket engine uses liquid fuels such as ethanol, kerosene or JP-7. How-
ever, liquid is usually has higher conductivity and therefore, current leakage will be large
between the electrodes, resulting in much greater power consumption even if the valve
functions. Nonetheless, there are fluids that possess very good electrical properties in high
strength fields that can be used as the first testing fluids, such as silicon oil. If the valve
can accomodate liquid fuels, it can be expected that its future applications will be much
wider.
References
[1] A. Epstein, et al. Micro Gas Turbine Generators, Third Semi-Annual InterimTechnical Profess Report, January 1997.
[2] D. J. Sadler, K. W. oh, et al. A New Magnetically Actuated Microvalve For LiquidAnd Gas Control Applications, Transducers '99, June 7-10, 1999, Sendai, Japan, pp1812-1815.
[3] R. Zengerle, H. Sandmaier. Microfluidics in Europe, 28th AIAA Fluid DynamicsConference & 4th AIAA Shear Flow Control Conference, June 29 - July 2, 1997,Snowmass Village, CO.
[4] I. Chakraborty, W. C. Tang, D. P. Bame, T. K. Tang. MEMS Micro-Valve For SpaceApplications, Transducers '99, June 7-10, 1999, Sendai, Japan, pp 1820-1823.
[5] Mike L. Philpott, David J. Beebe, et al. Switchable Electrostatic Micro-Valves WithHigh Hold-Off Pressure, Solid-State Sensor and Actuator Workshop, Hilton HeadIsland, South Carolina, June 4-8, 2000, pp 226-229.
[6] A. P. Papavasiliou, D. Liepmann, Al. P. Pisano. Electrolysis-Bubble Actuated GateValve, Solid-State Sensor and Actuator Workshop, Hilton Head Island, SouthCarolia, June 4-8, 2000, pp 48-51.
[7] G. Hahm, H. Kahn, etc. Fully Microfabricated, Silicon Spring Biased, ShapeMemory Actuated Microvalve, Solid-State Sensor and Actuator Workshop, HiltonHead Island, South Carolia, June 4-8, 2000, pp 230-233.
[8] C. Vieider, 0. Ohman, H. Elderstig. A Pneumatically Actuated Micro Valve with ASilicone Rubbber Membrane for Integration with Fluid-Handling Systems,Transducers' 95, Eurosensors IX, Stockholm, Sweden, June 25-29, 1995, pp 284-286.
[9] J. Ulrich, H. Fuller, R. Znegerle. Static And Dynamic Flow Simulation of A KOH-Etched Micro Valve, Transducers'95, Eurosensors IX, Stockholm, Sweden, June 25-29, 1995, pp 17-20.
[10] Mitchell J. Novack. Design and Fabrication of a Thin Film MicromachinedAccelerometer, Master Thesis in the Department of Mechanical Engineering atMassachusetts Institute of Technology, September 1992.
[11] Stephen D. Senturia. Microsystem Design. Kluwer Academic Publishers. Boston2000, pp 332-338.
95
96 References
[12] A. A. Ayon, R. Braff et al. Characterization of a Time Multiplexed InductivelyCoupled Plasma Etcher, Journal of The Electrochemical Society, 146 (1), 1999, pp339-349.
[13] J. M. Meek, J. D. Craggs. Electrical Breakdown of Gases. John Wiley & Sons, Ltd:New York, 1978, pp2 10 -3 18 .
[14] Jo-Ey Wong. Analysis, Design, Fabrication, and Testing of a MEMS Switch forPower Applications. Doctoral Thesis at the Massachusetts Institute of Technology,June 2000, pp2 9 -32 .
[15] C. H. Mastrangelo, C. H. Hsu. Mechanical Stability and Adhesion of Micro Structureunder Capillary Forces, Journal of Microelectromechanical Systems, March 1993,pp33-43.
[16] Arthur W. Adamson. Physical Chemistry of Surfaces. Interscience Publishers, Inc.,New York. 1960, pp2 6 1-2 7 5 .
[17] K. Komvopoulos. Surface Texturing and Chemical Treatment Methods for ReducingHigh Adhesion Forces at Micromachine Interfaces. Part of the SPIE Conference onMaterials and Device Characterization in Micromachining, Santa Clara, California,September 1998. SPIE Vol. 3512, pp 106-122.
[18] Y. Ando, J. Ino et al. Friction and Pull-off Force on Silicon Surface Modified by FIB,Micro Electro Mechanical Systems, 1996, MEMS '96, Proceedings. An Investigationof Micro Structures, Sensors, Actuators, Machines and Systems. IEEE, The NinthAnnual International Workshop, 1996, pp 349 -353.
[19] M. Houston, R. Maboudian, etc. Ammonium Fluoride Anti-Stiction Treatments forPoysilicon Microstructures, The 8th International Conference on Solid-State Sensorsand Actuators, and Eurosensors IX, Stockholm, Sweden, June 25-29, 1995, pp 210-213.
[20] Y. Matsumoto, T. Shimada, etc. Novel Prevention Method of Stiction Using SiliconAnodization for SOI Structure, Sensors and Actuators A 72 (1999), pp 153-159.
[21] M. Horiuchi, S. Aoki. A Mechanism of Silicon Wafer Bonding. Proceedings of theFirst International Symposium of Semiconductor Wafer Bonding: Science,Technology and Applications. January 1992, pp 46-52.
[22] http://umech.mit.edu/MEMS.html
[23] Chuang-Chia Lin. Development of a Microfabricated Turbine-Driven Air BearingRig. Doctoral Thesis at the Massachusetts Institute of Technology, June 1999.
[24] Munson, Yound & Okiishi. Fundamentals of Fluid Mechanics. Third Edition. JognWiley & Sons, Inc. New York: 1998.
96 References
Appendix
AMask Drawings
97
02 pedi:Ms Daig
-i*
4 4#
Figure A. 1. Mask: ALIGN, wafer level, with streets
- - - a Q -1d - --- - _- -1 -; 11 2=
Appendix : Mask Drawins98
Appendix: Mask Drawings
Figure A.2. Mask: TOPELEC, die level, with streets
99
100 Appendix: Mask Drawings
Figure A.3. Mask: TopELEC, device level
0
0 0
0
100 Appendix : Mask Drawingzs
Appendix: Mask Drawings 101
Figure A.4. Mask: TOPOX_2, die level, with streets
.01
101Appendix : Mask Drawin~s
102 Appendix: Mask Drawings
eK2
Figure A.5. Mask: TOPTHROUGH, die level, with streets
Appendix: Mask Drawings 103
( )
Figure A.6. Mask: TOPTHROUGH, device level
103Appendix : Mask Drawings
104 Appendix: Mask Drawings
Figure A.7. Mask: BOSSOX1, die level, with streets
104 Appendix : Mask Drawings
Appendix: Mask Drawings 105
00
Figure A.8: Mask: BOSS_FEET, device level
106 Appendix: Mask Drawings
0
Figure A.9. Mask: BOSSOX_2, die level
106 Appendix : Mask Drawings
ADuendix: Mask Drawines10
0
Figure A.10. Mask: BOSSTETHER, die level, with streets
107
108 Appendix: Mask Drawings
Figure A. 11. Mask: BOSSTETHER, device level
Amcendix : Mask Drawings108
Appendix: Mask Drawings 109
0
Figure A.12. Mask: BOSSDEEP, die level, with streets
109Appendix : Mask Drawings
__________________________________ 1±:
110 Appendix: Mask Drawings
_____________%
Figure A.13. Mask: STREETS, wafer level
Appendix: Mask Drawings 111
0
Figure A.14. Mask: SEAL-OXIDE, die level, with streets
112 A Dendix : Mask Drawings
Figure A.15. Mask: SEAL_OX, device level
112
Appendix : Mask Drawinas13
00
Figure A.16. Mask: SEALBACK, die level, with streets
113
1 ilAnedx Ms1raig
Figure A.17. Mask: SEALSEAT, die level, with streets
Appendix : Mask Drawings
0
114
Appendix: Mask Drawings 115
0 0
0
Figure A.18. Mask: SEALCHANNEL, die level, with streets
115Appendix : Mask Drawings
116 Appendix: Mask Drawings
Appendix
BValve Process Flow
B.1 Top Wafer
In this process, nitrite is used as protection layer.
1. Nitrite Deposition: all wafersICL, Tube A5Recipe: G460Target: 0.1 umDeposition Time: 36 min
Wafer Number Film Thickness(um) STD
Monitori 0.1012 0.0006Monitor2 0.1001 0.0006
2. Mask 1: ALIGN, both sidesPhotolithography
Coating: Standard thin resist at 3000 rpm
Coat on one sidePrebake 10 min at 90 degCCoat the other sidePrebake 25 min at 90 degC
Exposure: EVI for 2 secBack side alignment
Develop: 10 secExposure: Front side alignmentDevelop: 55 secPostbake: 30 min at 120 degC
117
118
AME Etch: both sidesRecipe: Nitride STD SF6 for nitrideTime: 60 secRecipe: Undoped Poly for SOITime: 60 sec
Step Height Measurement:==> SOI thickness: 0.43 um
BOE 12 min
Double Piranha Strip
3. Mask 2: TOPELECPhotolithography
Coating: Standard thin resist @ 3000 rpmPrebake 30 min at 90 degC
Exposure: EVI for 2 secFront side alignment
Develop: 60 secPostbake: 30 min at 120 degC
AME 5000 EtchRecipe: Nitrite STD SF6Time: 60 secRecipe: Undoped PolyTime: 80 sec (intended for 50 sec)Target: 0.34 um
SOI Thickness Measurement: 0.44 um
4. Mask 3: TOPOX_2Photolithography
Coating: Standard thin resist at 3000 rpmPrebake 30 min at 90 degC
Exposure: EVI for 2 secFront side alignment
Develop: 60 secPostbake: 30 min at 120 degC
AME for NitriteRecipe: Nitrite STD SF6Time: 60 sec
BOE for BOX, 1 umTime: 13 min
Appendix B: Valve Process Flow
Appendix B: Valve Process Flow 119
5. Mask 4: TOPTHROUGHPhotolithography
Coating: NT1-3: Thick resist at 1000 rpm: 10 umNT4,5: Thick resist at 2000 rpm: 8 umCoat on front sidePrebake 60 min at 90 degC
Exposure: EVI for 21 secTop side alignment
Develop: 180 sec
Coating: Thin resist at 3000 rpm back sidePostbake: 30 min at 90 degC
Mounting6" quarze waferThick PR; 2.2 krpmCenter Dot; Middle Ring; Outer RingSoft Bake: 15 min at 90 degC
Through Etch: STS2Recipe: MIT_37Time total: 4 hr 20 min
B.2 Boss Wafer
1. Thermal oxidation, 1.5 umICL, Tube A3Recipe: G22410 min dry oxidation280 min wet oxidation
Wafer Number Film Thickness(um) STD
VB1 1.539 0.005VB2 1.547 0.002VB3 1.549 0.002VB4 1.549 0.002VB5 1.548 0.002VBM1 1.543 0.003VBM2 1.532 0.004
Appendix B: Valve Process Flow 119
120
2. Mask 1: ALIGNPhotolithography
Coating: Standard thin resist at 3000 rpmCoat on one sidePrebake 10 min @ 90 degCCoat the other sidePrebake 20 min @ 90 degC
Exposure: EV1 for 2 secBack side alignment
Develop: 10 secExposure: Front side alignmentDevelop: 55 secPostbake: 30 min @ 120 degC
BOE: 1.54 um oxideHF:H20 7:1 Buffer20 min
Nitrite EtchAME 5000, Chamber ARecipe: Nitrite CF4Time: 60 sec
Alignment Mark Etch: AME 5000All wafers; Both sidesRecipe: POLYSTDTime: 135 secEtch Depth: 1 um
3. Mask 2: BOSSOX1Photolithography
Coating: Standard thin resist @ 3000 rpmCoat on front sidePrebake 30 min @ 90 degCExposure: EV1 for 2 secTop side alignment
Develop: 60 secBack Side Coating: Standard thin resist @ 3000 rpmPostbake: 30 min @ 120 degC
Xueen 12/12/00
AME Nitrite Etch: VB5Recipe: NITRITE STD SF6Time: 60 sec
Appendix B: Valve Process Flow
BOE: 1.54 um oxide
Appendix B: Valve Process Flow 121
19 min, ICL
Double Piranha Strip Photoresist
4. Mask 3: BOSSFEET_2Photolithography
Coating: Standard thin resist @ 3000 rpmCoat on front side
Prebake 10 min @ 90 degCRe-coat PR front sidePrebake 25 min @ 90 degC
Exposure: EVI for 4 secTop side alignment
Develop: 90 secPostbake: 30 min @ 120 degCResist Height: 2.06 um
AME EtchBOE 15 secRecipe: Undoped PolyTime: 470 secTarget: 3.4 um
Feet Etch Step height Measurement:
Wafer Number Feet Height (um)
VB2 3.39 umVB3 3.68 umVB5 3.78 um
5. Mask 4: BOSSOX_2Photolithography
Coating: Standard thin resist @ 3000 rpmCoat on back sidePrebake 60 min @ 90 degC
Exposure: EV1 for 2 secTop side alignment
Develop: 60 secPostbake: 30 min @ 120 degC
AME for Nitrite: VB5Recipe: Nitirte STD SF6Time: 60 sec
Appendix B: Valve Process Flow 121
122 Appendix B: Valve Process Flow
BOE for Oxide, 1.54 umTime: 20 min
6. Mask 5: BOSSTETHERPhotolithography
Coating: Thick resist @ 3000 rpm: 6 umCoat on front side
Prebake 60 min @ 90 degCExposure: EVI for 17 sec
Top side alignment
Develop: 120 sec
Coating: Thin resist @ 3000 rpm
Back side
Postbake: 30 min @ 120 degC
Tether Etch: STS1Recipe: MIT_59Etch Rate: 1.5 um/min ?
BOX Thickness: 0.35 umEtch Time: 10 minTether Height Measurement Using Wyko:
Wafer # Step Height
VB2 17.6VB3 17.6VB5 17.7
Double Piranha Strip
7. Mask 6: BOSSDEEPPhotolithography
Coating: Thick resist @ 2000 rpm: 8 umCoat on front side
Prebake 60 min @ 90 degCExposure: EVI for 20 sec
Top side alignment
Develop: 150 secCoating: Thin resist @ 3000 rpm
Back side
Postbake: 30 min @ 90 degC
Mounting:
4" quarze wafer
122 Avivendix B: Valve Process Flow
Appendix B: Valve Process Flow 123
Thick PR; 2.2 krpmCenter Dot; Middle Ring; Outer Ring
Softbake: 15 min @ 90 degC
Boss Etch: STS1Recipe: MIT_69Etch Rate: 2.04 um/min
Etch Time: 3 hr
B.3 Seal Wafer
1. Double Alignment
Mask 1 & 2: ALIGN & STREETPhotolithography
Coating: Standard thin resist @ 3000 rpm
Coat on front side
Prebake 10 min @ 90 degCCoat on back side
Prebake 25 min @ 90 degCExposure: Top side alignment (ALIGN)
EV1 for 2 seeDevelop: 8 sec
Exposure: Back side alignment (STREET)EV1 for 2 see
Develop: 55 seePostbake: 30 min @ 120 degC
AME, ICLRecipe: UPDOPEDPOLYTime: 68 see
Target: 0.5 um
2. Thermal oxidation, 0.7 um
Piranha strip photo: double piranha, ICLPiranha clean, ICLRCA clean, ICL
ICL, Tube A3Recipe: G1485 min dry oxidation67 min wet oxidation5 min dry oxidationFilm Thickness: 0.713 um
3. Polysilicon deposition, 0.95 um
124
ICL, Tube A6Recipe: G461Deposition rate: 58 A/minDeposition time: 2 hr 30 min
Wafer Number Film Thickness(um) STD
Dummy 1.007* 0.002
* Estimated *
4. Thermal oxidation, 1 umICL, Tube A3Recipe: G224
5 min dry oxidation135 min wet oxidation5 min dry oxidation135 min wet oxidation5 min dry oxidation
5. Mask 2: SEALBACKPhotolithography
Coating: Thick resist @ 2000 rpm: 8 umCoat on back sidePrebake 60 min @ 90 degC
Prebake 60 min @ 90 degCExposure: EVI for 18 sec
Top side alignment
Develop: 150 secCoating: Thin resist @ 3000 rpm
front sidePostbake: 30 min @ 90 degC
BOE: 9 min
Mounting6" silicon waferThick PR; 2.2 krpmCenter Dot; Middle Ring; Outer RingSoft Bake: 15 min @ 90 degC
STS2 EtchRecipe: MIT_37AEtch Time: 80 min
Appendix B: Valve Process Flow
Appendix B: Valve Process Flow 125
Etch Depth: 240 um
Piranha Dismount30 min
Piranha Clean10 min
6. Mask 3: SEALOXPhotolithography
Coating: Standard thin resist @ 3000 rpmCoat on front sidePrebake 60 min @ 90 degC
Exposure: EVI for 2 secTop side alignment
Develop: 60 secPostbake: 30 min @ 120 degC
BOE Etch Top OxideTime: 13 min (intended for 10 min)Target: 1 um
AME Etch PolyRecipe: Updoped PolyTime: 70 secTarget: 0.42 um
BOE Etch Bottom OxideTime: 9 minTarget: 0.7 urn
Piranha clean10 min
7. Mask 4: SEALSEATPhotolithography
Coating: Standard thin resist @ 3000 rpmCoat on front sidePrebake 60 min @ 90 degC
Exposure: EVI for 2 secTop side alignment*Use Arturo's rig for alignment exposure**Make sure use clear area for blokage*
Develop: 60 secPostbake: 30 min @ 120 degC
Appendix B: Valve Process Flow 125
BOE for Top Oxide, TRL15 min
AME for PolysiliconRecipe: Undoped PolyTime: 80 sec
8. Mask 6: SEALCHANNELPhotolithography
Coating: Thick resist @ 2000 rpm: 8 umCoat on front sidePrebake 60 min @ 90 degC
Exposure: EVI for 17 sec
Top side alignmentDevelop: 180 secCoating: Thin resist @ 3000 rpm
Back sidePostbake: 30 min @ 90 degC
BOE UltrasonicTime: 10min*Tencor measurement shows about half micron oxide ontop of silicon. *
Mounting:4" quarze waferThick PR; 2.2 krpmCenter Dot; Middle Ring; Outer RingSoftbake: 15 min @ 90 degC
STS1 EtchMIT 372 hr
Piranha Dismount
9. Wafer BondingRCA Clean
EvAlign-BonderCompressed overnightAnnealing: Tube A2Time: 1 hr
10. Diesaw
126
Appendix
CMask Drawings
127
1.6
0.8
0.65
-- 0.5 --
1.0000±0.0005
Ctearonc
........... ................ ......... . . . .......... 1 1/16__
Clearance Hole for 4-40
Clearance Hole for 4-40 (4 Holes)
RO.03130 0 00 0
Reamed Holes for Dowel Pins-0.0004(2 Holes)
0.50.65
0.8
e Hole for 4-40 (2 Holes)
Material: Polished Plexiglass 1/16 thickTolerances: ± 0.003 unless noted otherwiseAlexander H61ke indow Plate Sheet 109/21/99 of 6
Figure C. 1. AutoCAD layout of the valve package: window plate
00
1.6
0.8
-e - 0.65
--- 0.5
R0.2362"0.0015
[ R6.00mm+0.04m] Clearance Hole for 4-40
Clearance Hole for 4-40 (4 Holes)
R0.03130 00 04 Reamed Holes for Dowel Pins
- - A(2 Holes), tight tolerancesA A on location i 0.001
---
0.50.039[i.00mm - RO.1969[R5.00mm.l6
0.8
4-40 UNC threaded through (2 Holes)0.0394(1.00mm
Section A-A3/64 0.028 ±0.001 [0.7mm]
- ------ - 3/32
1/06 I3/16 0.3 0.024 ±0.001 [0.6mm]
Material: Polished Plexiglass 3/16 thickTolerances: ± 0.003 unless noted otherwiseAlexander Hdlke Top Plate Sheet 209/21/99 of 6
Figure C.2. AutoCAD layout of the valve package: top plate
1.6
0.8
0.65
0.5
0.302 0.001 - - -
4-40 UNC threaded through
Clearance holes for 4-40 (4 holes)
00.06250.0000 Reamed Holes for Dowel Pins<2 HoLes),
0.302±0.001 tight toterances on location t0.001
0.50.65
0.8
Clearance Holes for 4-40 (2 Holes)
NotePerpendicularity of center squareand alignment to the dowel pinsis critical (±0.001)
Material: AluminumTolerances: ± 0.003 unless noted otherwiseAlexander Holke Spacer Plate Sheet 309/21/99 of 6
Figure C.3. AutoCAD layout of the valve package: spacer plate
1.6
1.~)0
- -----
01/8(4 Holes Clearance)
0.0590.001 (1.5mm)
I
1.6
0.8
0.65
0.5 -
Af
0.213
0.039[1.00mm
R0.236+0.0015E R6.00mm+0.04mm0.0000 L0.00MM
Cl
4-th
RO.0 310
0 for- tolerar
--E±iZ + - --- - -00
0 o 0.50 00 )0 000 ~ 0.65++ R.197[R5,00m 1 .
Clearance Holes for0.039[1.00mm
0 0.039 [lm] Through holesfor Pogo Pins (19 Holes)
Tolerances are ±0.001
earance Hole for 4-40
40 UNC threaded 0.030 [0.75mm]
rough 04 Hotes) -- 0.089 [2.25mm]
0.148 [3.75mm30.000 Reamed Holes 0.207 [5.25mm)-0.0004 026[.5mDowet Pins (2 Hotes) - 0.266 [6.75cc)ce for location ±0.001
0.030 10075cc)
0.8 0.089 (225mm)
0.148 [3.75-3 Detail of Center Section0.207 [5.25m ] -
0.266 [6.75mm]
4-40 (2 Holes)
I I I l I I I I I
1/4
1/8
Section A-A
1/016 3/64 1
0.3
Figure C.4. AutoCAD layout of the valve package: bottom plate
0.024 ±0,001 [0.6mm]
Material: Polished Plexiglass 1/4 thickTolerances: ± 0.003 unless noted otherwise
Alexander Hblke Bottom Plate Sheet 409/21/99 of 6
1.60.8
0.65
0.5 --
Clearance Hole for 4-40(4 Holes)
Clearance Hole for 4-40
3 Clearance Hole for 4-40 (4 Holes)
R0.03000$ 4 Reamed Holes for Dowel Pins(2 Holes), toleranceson location are 0.001
I 80.016 #78 Through Holes (19 Holes)for Pogo Pins, same Centers as shown
-- - --- - -for Bottom Plate, Tolerances ±0.001
0.50.65
1.05 .
Clearance Hole for 4-40 (2 Holes)
I III III liii
I ~ [I I II~I ~I I I111111111 II
Material: Polished Plexiglass 1/2 thickTolerances: ± 0.003 unless noted otherwiseAlexander H61ke Pin Holder Sheet 509/22/99 of 6
Figure C.5. AutoCAD layout of the valve package: pin holder
L'.3
.A ALI
a-1 CL00
0 CDI 3 0 u-
CD CCD
Alexander Hlke09/22/99
Assembly Sheet 6of 6
Figure C.6.. AutoCAD layout of the valve package: pin holder
7. -
134