University of Southern California Dissertations and Theses

Acoustic ejector employing lens with air-reflectors and piezoelectrically actuated tunable capacitor

(USC Thesis Other)

Acoustic ejector employing lens with air-reflectors and piezoelectrically actuated tunable capacitor

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

ACOUSTIC EJECTOR EMPLOYING LENS WITH AIR-REFLECTORS AND
PIEZOELECTRICALLY ACTUATED TUNABLE CAPACITOR

by
Chuang-Yuan Lee

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ENGINEERING)

August 2007

Copyright 2007 Chuang-Yuan Lee

ii
Acknowledgements
First, I would like to thank my advisor, Dr. Eun Sok Kim, for offering me the
opportunity to explore the exciting and fantastic field of MEMS and giving me
constant guidance and encouragement far beyond research, without which it would
have been impossible to carry out this work.
I am indebted to all the members in the USCMEMS group for their
camaraderie and support. My deepest gratitude goes to Dr. Wei Pang and Dr.
Hongyu Yu, who taught me almost everything about micromachining when I first
joined the group and gave me tremendous helps in the past four years.
My thanks also go to Qiang Zou and Hao Zhang for their valuable advice and
friendship. I would also like to thank Merrill Roragen, Derrick Chi, Joe Lo, Shih-Jui
Chen, Ta-Shun Chu, Huisui Zhang, and Sanat Kamal-Bahl for their friendship and
assistance in research.
Finally, nothing that I have done would be possible without the love,
encouragement, and support of my parents. The same appreciation is extended to my
brother. My special thanks go to Chen-Yin Ou, who has been by my side on my
every good and bad day and has turned my Ph.D. years sweet and fruitful.

iii
Table of Contents
Acknowledgements......................................................................................................ii
List of Tables ..............................................................................................................vi
List of Figures ............................................................................................................vii
Abstract ......................................................................................................................xv
Chapter 1
Introduction..................................................................................................................1
1.1 Review of Droplet Ejection and Acoustic Focusing Methods.............1
1.2 Review of Micromachined Tunable Capacitors...................................3
1.3 Overview of the Chapters ....................................................................5
Chapter 1 References ...................................................................................................7

Chapter 2
Approach to Acoustic Ejector with LWAR ...............................................................10
2.1 Design and Simulation.......................................................................10
2.1.1 Working Principle`.....................................................................10
2.1.2 Liquid Delivery for Continuous Droplet Ejection .....................13
2.1.3 Directional Acoustic Ejection ....................................................14
2.2 Device Fabrication.............................................................................16
2.3 Experimental Testing Setup ...............................................................21
Chapter 2 Summary....................................................................................................22
Chapter 2 References .................................................................................................24

Chapter 3
Results and Applications of Acoustic Ejector with LWAR.......................................25
3.1 Ejector Performance Characterization ...............................................25
3.1.1 Droplet Formation Sequence......................................................25
3.1.2 Ejections at High Ejection Rates................................................26
3.1.3 Stable and Continuous Ejection .................................................27
3.1.4 Comparisons with Ejector without LWAR................................28
3.1.5 Directional Ejections..................................................................28
3.2 Nanoliter Droplet Coalescence in Air................................................30
3.2.1 Coalescence of Two Droplets ....................................................31
3.2.2 Application to Micromixing.......................................................35
3.2.3 Coalescence of Four Droplets ....................................................37
3.3 On-Demand DNA Synthesis by Four Directional Ejectors
on a Chip ............................................................................................39

iv
3.3.1 Review of Microarray Fabrication Methods..............................39
3.3.2 Synthesis by Cyanoethyl Phosphoramidite Chemistry ..............42
3.3.3 A 15-mer Oligonucleotide Synthesized by Four
Directional Ejectors on a Chip ...................................................46
3.4 Droplet-Based Microreactions with Oil Encapsulation .....................50
3.4.1 Review of Contemporary Microfluidic Networks .....................50
3.4.2 Microreaction Platform Based on Directional
Acoustic Ejectors.......................................................................53
3.4.3 Experimental Steps for Microreactions with Oil
Encapsulation .............................................................................55
3.4.3.1 Construction of Oil Microreaction Chambers....................55
3.4.3.2 Dispensation of Reagent Droplets into Oil
Microchambers ..................................................................57
3.4.3.3 Characterization of Water Evaporation Rate .....................61
3.4.4 Microreaction Applications by the New Microreaction
Platform......................................................................................63
3.4.4.1 Physical Mixing Reaction..................................................63
3.4.4.2 Chemical Presipitation Reaction........................................63
3.4.4.3 Long-term Iodine Clock Reaction .....................................65
3.5 Harmonic Operation for Ejection of Micron-Sized Droplets.............68
3.5.1 Introduction................................................................................68
3.5.2 Design and Fabrication..............................................................69
3.5.3 Results and Discussion...............................................................71
Chapter 3 Summary....................................................................................................76
Chapter 3 References .................................................................................................77

Chapter 4
Piezoelectrically Actuated Tunable Capacitor...........................................................82
4.1 Design and Simulation.......................................................................82
4.1.1 Cantilever Structure...................................................................82
4.1.2 Bulk-Micromachined Tunable Capacitor with Mass
Structure.....................................................................................83
4.1.3 Bridge-Type Surface-Micromachined Tunable Capacitor.........84
4.1.4 Deflection Analysis....................................................................86
4.1.5 Structure Optimization...............................................................88
4.1.6 Theoretical Capacitance Variation.............................................91
4.2 Device Fabrication.............................................................................93
4.2.1 Surface-Micromachined Parylene-Supported Tunable
Capacitor ....................................................................................93
4.2.2 Bulk-Micromachined Silicon-Supported Tunable
Capacitor ....................................................................................96
4.2.3 Bridge-Type Surface-Micromachined Tunable Capacitor.........99

v
4.3 Experimental Results and Discussion ..............................................103
4.3.1 Surface-Micromachined Parylene-Supported Tunable
Capacitor ..................................................................................103
4.3.2 Surface-Micromachined Cantilevers Utilizing Other
Supporting Materials................................................................109
4.3.3 Bulk-Micromachined Silicon-Supported Tunable
Capacitor ..................................................................................111
4.3.4 Bridge-Type Surface-Micromachined Tunable Capacitor.......114
4.3.5 Discussion................................................................................122
Chapter 4 Summary..................................................................................................123
Chapter 4 References ...............................................................................................125

Chapter 5
Conclusion and Future Directions............................................................................126

Bibliography ..........................................................................................................130

vi
List of Tables
Table 2.1 Particle displacements on the liquid surface (integrated over a
400 μm by 400 μm area) for different electrode patterns. ....................16
Table 4.1 Parameters for the unimorph cantilever design and simulation ............89
Table 4.2 Optimization of cantilever structure for largest deflection based
on constant voltage and constant field conditions.................................89
Table 4.3 Comparison of displacement amplitude for different structures.........118

vii
List of Figures
Figure 2.1 Cross-sectional view of the ejector employing the acoustic lens
with air-reflectors..................................................................................11
Figure 2.2 Vertical particle displacements on the liquid surface for
ejectors with and without lens...............................................................13
Figure 2.3 Integrated ejector for automatic liquid delivery. (a) Cross-
sectional view. (b) SEM photo of the buffer and the chamber
before being integrated with the ejector (upside-down view................14
Figure 2.4 Schematic diagram of directional acoustic ejector array. Each
ejector is integrated with its own reservoir for continuous
liquid supply for ejections.....................................................................15
Figure 2.5 Simulation of particle displacements at liquid surface for 90
o

sector electrode......................................................................................16
Figure 2.6 Brief fabrication steps for acoustic ejector employing lens with
air-reflectors ..........................................................................................17
Figure 2.7 Algor simulation showing negligible displacement as one
atmosphere static pressure is exerted on the parylene lens
structure.................................................................................................18
Figure 2.8 SEM photos of the fabricated ejector with LWAR: (a) top view
and (b) side view. ..................................................................................19
Figure 2.9 SEM photos of the release holes: (a) before and (b) after being
filled with parylene. ..............................................................................20
Figure 2.10 Photos of (a) ejector array and (b) chamber array.................................20
Figure 2.11 Fabricated device in a DIP package connected to transmission
lines and SMA connectors. ...................................................................20
Figure 2.12 Schematic diagram of testing setup for droplet ejections. ....................22
Figure 2.13 Photo of the experimental testing setup... .............................................22
Figure 3.1 Photos of droplet formation sequences. ................................................26

viii
Figure 3.2 Photos of droplet ejections at (a) 1 kHz, (b) 2 kHz, (c) 4 kHz,
and (d) 10 kHz.......................................................................................27
Figure 3.3 Photos of continuous ejections at different times. ................................28
Figure 3.4 Comparisons between the ejectors with and without the lens...............28
Figure 3.5 Stable and continuous directional ejection photos taken (a) at
the beginning of the ejections and (b) after 1800 droplet
ejections.................................................................................................29
Figure 3.6 Photos of the directional droplet ejections at the ejection rates
of (a) 1 kHz and (b) 2 kHz. ...................................................................30
Figure 3.7 (a) Optical micrographs with strobe to show the digital
droplets during the ejection process (Ejectors 1 and 2
activated). (b) Optical micrographs without strobe to illustrate
the droplets traveling trajectories. The shutter-opening time of
the CCD camera was set to 2 ms to record the images. ........................31
Figure 3.8 Optical micrographs of the droplets ejection process (Ejectors
2 and 4 activated). (a) Overview pictures of the ejection
process. (b) Detailed coalescence sequence of the droplets
around 1100 μs. (c) Rotations of the traveling coalesced
droplet with a periodicity of 400 μs ......................................................34
Figure 3.9 (a) Optical micrographs of the water, ink, and mixed droplets
ejected onto a glass slide by directional ejections. Water was
used as the medium for Ejector 1, while red ink was used for
Ejector 2. A glass slide was placed above the ejector to collect
the droplets. For comparison, one water droplet (~0.27 nL) was
first ejected by Ejector 1 and collected (image 1) and then one
ink droplet (~0.27 nL) was ejected by Ejector 2 (image 2).
Finally, one water and one ink droplet were simultaneously
ejected and coalesced in air (image 3). The coalesced droplet
continued traveling until it was collected on the glass slide
(image 4). (b) Schematic diagram showing that the trajectories
of any two different liquid droplets intersect in air...............................37
Figure 3.10 Optical micrographs of the ejection process when all the four
ejectors were activated. .........................................................................38

ix
Figure 3.11 Schematic diagram illustrating a spot inked by four directional
ejectors carrying four DNA bases.........................................................40
Figure 3.12 Illustration of standard phosphoramidite DNA synthesis
chemistry...............................................................................................44
Figure 3.13 Photos showing how a glass slide is rotated between the
synthesis and wash areas for DNA synthesis. The left photo
shows the motorized rotation stage along with the wash and
synthesis areas, while the right photo shows how the glass
slide is placed over the ejector array.....................................................45
Figure 3.14 Synthesized 15-mer 5’-CGCCAAGCAGTTCGT-3’ DNA
sequence on a poly-l-lysine glass substrate. (a) Schematic
illustration of the oligonucleotide synthesized by
phosphoramidite chemistry. (b) Micrograph of the hybridized
DNA sequence at the spot where the DNA bases were ejected
by the four ejectors. (c) Fluorescence image at another spot on
the same glass which did not have the synthesized DNA
sequence. ...............................................................................................47
Figure 3.15 Schematic representation of droplet-based microreactions with
oil encapsulation. (a) Diagram of microreaction platform based
on four directional acoustic ejectors (two in the front and the
other two in the back) along with four inlet ports. (b) Sequence
of microreactions using nanoliter droplets – (1) Formation of
oil microreaction chamber. (2) Dispensation of reagent A. (3)
Microreaction inside oil with concurrent or sequential
dispensing of reagents B and C from the two ejectors in the
back. ......................................................................................................54
Figure 3.16 Construction of oil microreaction chambers. (a) Stable and
continuous directional ejections of oil droplets without any
frequency tuning. (b) Successive oil droplets ejected onto the
same position on glass and accumulated into oil microchamber.
(c) Array of oil microchambers constructed on glass substrate. ...........56

x
Figure 3.17 Dispensation of red ink droplet into the oil microreaction
chamber. (a) Sequential steps to form oil microreaction
chamber and to dispense red ink droplet – (1) Oil droplets
ejected by directional ejector and collected on glass. (2)
Microreaction chamber constructed with 20 oil droplets. (3)
Ejection of red ink droplets. (b) Photo showing red ink droplet
encapsulated inside oil. (c) Schematic showing relative
position of encapsulated aqueous droplet inside oil..............................58
Figure 3.18 Detailed sequence of encapsulation process. Water was used as
the liquid medium. One water droplet was ejected onto oil to
create a water crater, which was then sealed radially from
outside by a thin film of oil...................................................................59
Figure 3.19 Alternative scheme for droplet-based microreactions with oil
encapsulation. (a) Schematic representation of the alternative
scheme. Aqueous droplets were dispensed to penetrate through
an oil film for oil encapsulation. (b) Experimental results
showing a liquid droplet before and after it passed through the
oil film – (1) Water droplet ejected toward the oil film
constructed by a metal frame. (2) Water droplet encapsulated
in oil after penetration through the oil film...........................................60
Figure 3.20 Characterization of water evaporation rate: (a) Photos showing
slow evaporation of one water droplet (~0.27 nL) inside 10-oil-
droplet microchamber. It took more than 10 hours for the water
droplet to dry out. (b) Measured decrease in evaporation rate as
a function of time. Also plotted is the surface area, over which
diffusion is taking place, as a function of time. ....................................62
Figure 3.21 Physical mixing microreaction using nanoliter droplets with oil
encapsulation. (a) Micrographs showing red ink droplets
ejected and encapsulated inside oil microchambers with
different sizes. (b) Micrographs showing various combinations
of red and blue ink droplets encapsulated and mixed inside oil
microchamber........................................................................................64

xi
Figure 3.22 Chemical precipitation microreaction using nanoliter droplets
with oil encapsulation. (a) One droplet of 0.1 M silver nitrate
solution ejected and encapsulated inside oil. (b) No observable
evaporation after one hour. (c) Dispensation of one droplet of
0.1 M sodium carbonate solution into oil. (d) Microreaction of
silver nitrate and sodium carbonate with oil encapsulation. (e)
Formation of silver carbonate precipitate in 1 sec. ...............................65
Figure 3.23 Long-term iodine clock reaction. (a) Macroreaction: Pipetting
1 mL of solution A and 1 mL of solution B into test-tube for
reaction. Blue color appeared after 126 sec. (b) Microreaction:
Ejecting one droplet (0.3 nL) of solution A and one droplet
(0.3 nL) of solution B into oil microchamber for reaction. Blue
color appeared after 110 sec..................................................................66
Figure 3.24 Photos of fabricated acoustic lenses for harmonic operations. .............71
Figure 3.25 Photos of stable ejections for the 3rd harmonic....................................71
Figure 3.26 Ejections for the 3rd harmonic at high ejection rates............................72
Figure 3.27 Photos of stable ejections of 10- μm-diameter droplets with the
9th harmonic operation. ........................................................................72
Figure 3.28 Ejections at ejection rate of 3 kHz for different harmonics. .................73
Figure 3.29 Droplet size reduction by harmonic operations. ...................................73
Figure 3.30 Measured quality factor for harmonics. ................................................74
Figure 3.31 Droplet size as a function of frequency.................................................75
Figure 3.32 Droplet separation time versus droplet size. .........................................75
Figure 4.1 Schematic representation of the cantilever-type surface-
micromachined piezoelectric tunable capacitor....................................83
Figure 4.2 Schematic representation of the bulk-micromachined tunable
capacitor with mass structure................................................................84
Figure 4.3 Schematic representation of the bridge-type surface-
micromachined tunable capacitor. ........................................................86

xii
Figure 4.4 (a) Theoretical deflection as a function of SiN thickness for
constant field condition. (b) Neutral plane position simulation............90
Figure 4.5 Theoretical capacitance variation as a function of applied
voltage for cantilever-type surface-micromachined tunable
capacitor. ...............................................................................................92
Figure 4.6 Fabrication process flow for the cantilever-type surface-
micromachined tunable capacitor. ........................................................95
Figure 4.7 SEM photos of the fabricated cantilever-type surface-
micromachined tunable capacitors (a) with the top capacitor
electrode connected to the pad through a lead and (b) with the
top capacitor electrode floating.............................................................96
Figure 4.8 Fabrication process flow for the bulk-micromachined tunable
capacitor ((a)-(c) are viewed from AB cross section, and (d)-(f)
are viewed from CD cross section). ......................................................97
Figure 4.9 Illustration of the method used for the bonding process: (1)
place the parylene slip, (2) apply the external force, and (3)
apply the epoxy, in that order................................................................99
Figure 4.10 SEM photos of the fabricated bulk-micromachined tunable
capacitor: (a) top view and (b) side view..............................................99
Figure 4.11 Fabrication process flow for the bridge-type surface-
micromachined tunable capacitor. ......................................................102
Figure 4.12 SEM photos of the fabricated bridge-type surface-
micromachined tunable capacitor. ......................................................103
Figure 4.13 Measured piezoelectric displacement characteristics of the
parylene-supported microcantilever for unbiased (a) sinusoidal,
(b) square, and (c) triangular voltage waveforms at 500 Hz...............105
Figure 4.14 Measured piezoelectric displacement characteristics of the
microcantilever for (a) positive and (b) negative applied
voltage with a duty cycle of 30%........................................................106
Figure 4.15 Measured piezoelectric displacement versus actuation voltage
of the cantilever-type tunable capacitor. .............................................107

xiii
Figure 4.16 Measured mechanical frequency response of the fabricated
cantilever-type surface-micromachined tunable capacitor..................108
Figure 4.17 Comparison of measured piezoelectric displacement for the
parylene-, SU-8-, and SiN-supported cantilevers. ..............................110
Figure 4.18 SEM photos of the SU-8-supported tunable capacitors with the
SU-8 thickness of (a) 1.6 μm, (b) 2.7 μm, and (c) 6.5 μm..................110
Figure 4.19 Measured piezoelectric displacement of the bulk-
micromachined tunable capacitor at the capacitor region versus
actuation voltage. ................................................................................112
Figure 4.20 Capacitance-versus-voltage curves for bulk-micromachined
tunable capacitor with mass structure mass. (a) Measured
capacitance and tuning ratio versus actuation voltage. (b)
Comparisons of experimental and theoretical results. ........................112
Figure 4.21 S
11
Smith charts for bulk-micromachined tunable capacitor
with mass structure at the actuation voltage (V
act
) of (a) −35 V,
(b) −25 V, (c) 0 V, (d) 18 V, and (e) 25 V, from 2 GHz to 6
GHz. ....................................................................................................114
Figure 4.22 Measured piezoelectric displacement characteristics of the
basic cantilever (LPCVD-Si
x
N
y
-supported) for unbiased (a)
sinusoidal, (b) triangular, and (c) square voltage waveforms at
1 kHz. ..................................................................................................115
Figure 4.23 Measured piezoelectric displacement versus actuation voltage
for (a) basic cantilever, (b) cantilever for simply-supported
bridge, and (c) simply-supported bridge. ............................................117
Figure 4.24 (a) Schematic showing how the electrical signals are applied to
the electrodes for bridge actuation. (b) Electrical signal
waveforms applied to bottom actuation electrodes of the two
cantilevers. ..........................................................................................118
Figure 4.25 Measured mechanical frequency response for the fabricated (a)
basic cantilever (LPCVD-Si
x
N
y
-supported), (b) cantilever for
simply-supported bridge, and (c) simply-supported bridge. ...............120

xiv
Figure 4.26 Capacitance-versus-voltage curves for bridge-type surface-
micromachined tunable capacitor. (a) Measured capacitance
versus actuation voltage. (b) Comparisons of experimental and
theoretical results. ...............................................................................121
Figure 4.27 S
11
Smith charts for bridge-type surface-micromachined
tunable capacitor at the actuation voltage (V
act
) of (a) −30 V, (b)
0 V, (c) +25 V, and (d) +40 V, from 2 GHz to 5 GHz........................121
Figure 5.1 Schematic representation of an exemplary array of 10 ejectors
for droplet-based microfluidic system. (a) Diagram showing 10
ejectors (along with 10 peripheral inlet ports) targeting the
same spot in the center with 10 different liquids. (b) Detailed
perspective drawing of a composed unit of the array. (c)
Demonstration of dynamic control of ejection angle through
combinatory actuation of any sector electrode – (1) Actuation
of one sector for most oblique ejection angle. (2) Actuation of
three sectors for less oblique ejection angle. (3) Actuation of
five sectors for least oblique ejection angle. (Electrodes of one
ejector are divided into 12 equal sectors in this example.) .................127
Figure 5.2 Top and perspective views of the piezoelectrically actuated
switch composed of two-air-gap capacitors........................................129

xv
Abstract
This thesis presents microfluidic and radio-frequency (RF)
microelectromechanical systems (MEMS) based on acoustic ejectors employing lens
with air-reflectors (LWAR) and piezoelectrically actuated tunable capacitors,
respectively.
For microfluidic MEMS, a new acoustic ejector with LWAR has been
developed. This ejector requires no nozzle, nor heat for actuation, but uses LWAR to
effectively focus acoustic waves for droplet ejection. LWAR utilizes the innate
acoustic impedance mismatch between solid and gas, and has wide tolerance for its
lens geometry. Ejections by the LWAR ejectors are demonstrated to be stable and
repeatable. The acoustic ejectors are further integrated with reservoirs and channels
for automatic liquid delivery, and continuous ejections have been attained without
any frequency tuning.
Based on the LWAR ejectors, ejections of micron-sized droplets are attained
through harmonic operations, and ejections in directional angles are realized through
electrode pattern designs. Both harmonic operations and directional ejections are
demonstrated to be stable and reproducible. Also the LWAR ejector is shown to
dispense oil with large viscosity without any difficulty. With the mentioned abilities,
this droplet-on-demand acoustic ejector has been successfully utilized for various
biochemical applications, including DNA synthesis, droplet coalescence in air, and
microreactions with oil encapsulation.

xvi
For RF MEMS, the first MEMS piezoelectric tunable capacitors employing
ZnO actuation have been developed. Three different types of ZnO-actuated tunable
capacitors are fabricated with surface and bulk micromachining techniques. The
piezoelectric deflection is demonstrated to be linear and bi-directional, and the
frequency response is fast and immune to environmental vibrations.
With the design of a bulk-micromachined mass structure driven by a ZnO
unimorph, a 2,100% continuous capacitance tuning range (from 0.46 pF to 10.02 pF)
is achieved with applied voltages from −35 V to +25 V. This is the highest tuning
ratio ever reported for parallel-plate tunable capacitors. Another bridge-type surface-
micromachined tunable capacitor with a much smaller dimension is fabricated on a
single chip without any bonding. Through the implementation of a simply-supported
bridge structure driven by two ZnO-actuated cantilevers, 1400% continuous tuning
from 0.13 pF to 1.82 pF is demonstrated without large stress developed in the bridge.

1
Chapter 1
Introduction

This thesis presents microfluidic and radio-frequency (RF)
microelectromechanical systems (MEMS) based on acoustic ejectors employing lens
with air-reflectors (chapters 2 & 3) and piezoelectrically actuated tunable capacitors
(chapter 4), respectively. General reviews of the two areas are covered in this chapter.

1.1 Review of Droplet Ejection and Acoustic Focusing Methods
The interest in creating and controlling microdroplets is growing rapidly,
driven by emerging applications in biomedicine, chemistry, and pharmaceutics [1-2].
Various droplet dispensing mechanisms have been developed and employed to
realize high resolution, fast, and reliable biochemical assays at low cost. Existing
droplet generation methods fall into two general categories ― contact pin dispensing
and non-contact inkjet dispensing. Since the introduction of monitoring gene
expressions by systematically printing deoxyribonucleic acid (DNA) microdroplets
[1], pin-based dispensing has gained extensive attention for printing DNA and
protein microarrays [3-4]. However, the pin is subject to clogging by contaminants,
and the chip coating is prone to damage, because the liquid-loaded metal pin
frequently contacts the chip surface [5].
Alternatively, inkjet dispensing is a non-contact method to shoot out bio-
chemical microdroplets onto a chip surface. Thermal and piezoelectric actuations are

2
the two most widespread mechanisms, and have been broadly reported for use in
mass spectrometer analysis [6], metals deposition [7], tissue engineering [8], and
printing polymer light-emitting diodes [9]. In the thermal method, liquid droplets are
ejected from a nozzle through the growth and collapse of vapor bubbles generated by
a heater. In the piezoelectric method, a hydrostatic pressure produced by the
mechanical bending of a piezoelectric unimorph forms a droplet at the nozzle. Both
methods eject droplets through droplet-defining nozzles, which are difficult to
construct with good uniformity, and suffer from the inherent tendency of nozzle
clogging. In addition, droplets can be ejected only in a direction perpendicular to the
nozzle surface.
Nozzleless dispensing is possible with acoustic actuations [10]. An acoustic
beam focused on the liquid surface can overcome the restraining surface tension and
expel liquid droplets from an open space without any nozzle. A variety of acoustic
focusing mechanisms have been reported. A spherical lens is an obvious and
straightforward choice [10], but its fabrication by mechanical grinding and polishing
is difficult and expensive. Fresnel lens has the advantage of being planar, but the
lens thicknesses are tightly restrained to acquire the desired ° 180 phase difference
between the waves traveling through the liquid and the lens [11]. Surface acoustic
waves (SAW) have also been utilized for droplet ejections [12]. The SAWs are
generated by the interdigital transducers (IDT) and leak into the liquid to form a
focused beam. However, the F-number of this lens, given by 0.5 tan Φ , is not a
design parameter and is typically smaller than one due to the small Raleigh angle Φ.

3
A flextensional ultrasound transducer is demonstrated to eject droplets by exciting
the axisymmetric resonance modes in a clamped circular plate [13]. However, it
requires nozzles to define the droplet size and inherits the issues from the nozzle-
based ejection. Another technique called self-focusing acoustic transducer (SFAT)
uses no physical lens, but focuses acoustic waves through near-field wave
interference by patterning the electrodes into annular rings [14]. The fabrication of
SFAT is very simple, but due to the patterned annular electrodes, the undesirable
heat generation or effects of lateral electric field could be an issue.
In this work, an effective focusing scheme called “lens with air-reflectors
(LWAR)” is developed by utilizing the innate acoustic impedance mismatch between
solid and gas. This new type of lens has wide tolerance for its lens geometry, and has
been shown to be very effective in focusing acoustic waves for droplet ejection.
Based on this idea, a series of biochemical applications are reported.

1.2 Review of Micromachined Tunable Capacitors
RF MEMS tunable capacitors have offered the opportunities to realize tuning
filters, matching circuits, and voltage-controlled oscillators due to their large tuning
ratio, low insertion loss, and high quality factor [15-21].
Among various actuation mechanisms such as electrostatic [15-19],
piezoelectric [20] and thermal [21] actuations, electrostatic actuation has received
most attention for MEMS tunable capacitors, and been reported in numerous papers.
Different types of capacitors have been investigated, among which the gap-closing

4
capacitor consisting of two parallel plates (one being movable and controlled by an
applied voltage and the other being fixed) is most widely used. However,
electrostatic actuation is a nonlinear function of the applied voltage, and the
controllable region is only one third of the initial plate gap. This pull-in instability
unfortunately limits the theoretical tuning range to 1:1.5. Several structure
modifications have hence been made to overcome the restriction, where a
geometrical separation of actuation and capacitor electrodes has successfully
provided a 1:6 tuning [17]. Recently, lateral [18] and angular vertical [19] comb-
drive actuators are implemented for tunable capacitor application with tuning ratios
of 1:8.4 and 1:31 respectively. However, these comb-drive actuators generally
require large actuation voltage or actuation area.
Compared to these tunable capacitors with electrostatic actuation, piezoelectric
actuation has several inherent advantages such as bi-directional deflection, high
linearity, low driving voltage, wide dynamic range due to no “pull-in” phenomenon,
and no electrostatic charging effect. However, up to now only one piezoelectric
tunable capacitor has been reported [20], where lead zirconate titanate (PZT)
actuators were used. Despite the innate potential for bi-directional actuation and
tuning, only unidirectional characteristics were utilized and a 1:3 capacitance tuning
was achieved in reference [20]. Further, although PZT has a large piezoelectric
constant, deposition of PZT thin films is relatively difficult. Sputter deposition of
PZT requires very tight process control for repeatable quality, and sol-gel PZT
typically has large residual stress. These processing difficulties are possible reasons

5
why piezoelectric actuation has been rarely explored for tunable capacitor
application.
By contrast, piezoelectric zinc oxide (ZnO) film can easily be deposited over
various materials, and has been proven to be reliable, reproducible, and compatible
with IC integration [22]. Through detailed calculation and by proper design, a fairly
large capacitance tuning can be realized by ZnO actuation in spite of its
comparatively weak piezoelectric nature.
This work describes the design, fabrication, and characterization of the first
MEMS piezoelectric tunable capacitors employing ZnO actuation. Relatively simple
design rules for the device-structure optimization for largest deflection are shown
from simulation results based on theoretical equations. Three different types of
tunable capacitors with surface and bulk micromachining techniques are presented.

1.3 Overview of the Chapters
In Chapter 1, reviews of the current methods used for droplet ejections and for
micromachined tunable capacitors along with the motivation of the thesis work are
described as a brief introduction to the thesis.
Chapter 2 presents the design, simulation, and fabrication of the proposed
acoustic ejector with LWAR. The experimental testing setup for droplet ejection is
also described.
Chapter 3 describes the characterization results of the designed acoustic ejector
with LWAR. Also demonstrated is the use of the acoustic ejector for various

6
biochemical applications, including on-demand DNA synthesis, droplets coalescence
in air, microreactions with oil encapsulation, and harmonic operations for smaller
droplets.
Chapter 4 describes the design, fabrication and characterization of
piezoelectrically actuated tunable capacitors. Three different types of ZnO-actuated
tunable capacitors with surface and bulk micromachining techniques are presented.
Finally, Chapter 5 presents conclusions and future research directions.

7
Chapter 1 References
[1] M. Schena, D. Shalon, R. W. Davis, and P. O. Brown, “Quantitative monitoring
of gene expression patterns with a complementary DNA microarray,” Science,
vol. 270, pp. 467-470, 1995.
[2] J. L. DeRisi, V. R. Iyer, and P. O. Brown, “Exploring the metabolic and genetic
control of gene expression on a genomic scale,” Science, vol. 278, pp. 680-686,
1997.
[3] D. J. Duggan, M. Bittner, Y. Chen, P. Meltzer, and J. Trent, “Expression
profiling using cDNA microarrays,” Nat. Genet., vol. 21, pp. 10-14, 1999.
[4] G. MacBeath and S. L. Schreiber, “Printing proteins as microarrays for high-
throughput function determination,” Science, vol. 289, pp. 1760-1763, 2000.
[5] H. B. Hsieh, J. Fitch, D. White, F. Torres, J. Roy, R. Matusiak, B. Krivacic, B.
Kowalski, R. Bruce, and S. Elrod, ”Ultra-high-throughput microarray
generation and liquid dispensing using multiple disposable piezoelectric
ejectors,” J. Biomol. Screen., vol. 9, pp. 85-94, 2004.
[6] T. Laurell, J. Nilsson, and G. Marko-Varga, “Silicon microstructures for high-
speed and high-sensitivity protein identifications,” J. Chromatogr. B, vol. 752,
pp. 217-232, 2001.
[7] G. G. Rozenberg, E. Bresler, S. P. Speakman, C. Jeynes, and J. H. G. Steinke,
“Patterned low temperature copper-rich deposits using inkjet printing,” Appl.
Phys. Lett., vol. 81, pp. 5249-5251, 2002.
[8] T. Xu, J. Jin, C. Gregory, J. J. Hickman, and T. Boland, “Inkjet printing of
viable mammalian cells,” Biomaterials, vol. 26, pp. 93-99, 2005.
[9] T. R. Hebner, C. C. Wu, D. Marcy, M. L. Lu, and J. Sturm, “Ink-jet printing of
doped polymers for organic light emitting devices,” Appl. Phys. Lett., vol. 72,
pp. 519-521, 1998.
[10] S. A. Elrod, B. Hadimioglu, B. T. Khuri-Yakub, E. G. Rawson, E. Richley, C. F.
Quate, N. N. Mansour, and T. S. Lundgren, “Nozzleless droplet formation with
focused acoustic beams,” J. Appl. Phys., vol. 65, pp. 3441-3447, 1989.
[11] S. C. Chan, M. Mina, S. S. Udpa, L. Udpa, and W. Lord, “Finite element
analysis of multilevel acoustic Fresnel lenses,” IEEE Trans. Ultrason.,
Ferroelect., Freq. Contr., vol. 43, pp. 670-677, 1996.

8
[12] U. Demirci, “Droplet-based photoresist deposition,” Appl. Phys. Lett., vol. 88,
pp. 144104-1-3, 2006.
[13] G. Percin, T. S. Lundgren, and B. T. Khuri-Yakub, “Controlled ink-jet printing
and deposition of organic polymers and solid particles,” Appl. Phys. Lett., vol.
73, pp. 2375 -2377, 1998.
[14] D. Huang and E. S. Kim, “Micromachined acoustic-wave liquid ejector,” J.
Microelectromech. Syst., vol. 10, pp. 442-449, 2001.
[15] L. E. Larson, R. H. Hackett, M. A. Melendes, and R. F. Lohr, “Micromachined
microwave actuator (MIMAC) technology − a new tuning approach for
microwave integrated circuits,” IEEE Microwave and Millimeter-wave
Monolithic Circuits Symp., pp. 21−30, 1991.
[16] T. G. S. M. Rijks, J. T. M. van Beek, P. G. Steeneken, M. J. E. Ulenaers, J. De
Coster, and R. Puers, “RF MEMS tunable capacitors with large tuning ratio,” in
Proc. IEEE International Micro Electro Mechanical Systems Conference,
Maastricht, The Netherlands, January 25 −29, 2004, pp. 777−780.
[17] Z. Xiao, W. Peng, R. F. Woffenbuttel, and K. R. Farmer, “Micromachined
variable capacitor with wide tuning range,” in Proc. Solid-State Sensor and
Actuator Workshop, Hilton Head Island, SC, June 2–6, 2002, pp. 346–349.
[18] R. Borwick, P. Stupar, J. DeNatale, R. Anderson, C. Tsai, and K. Garrett, “A
high Q, large tuning range, tunable capacitor for RF applications,” in Proc.
IEEE International Micro Electro Mechanical Systems Conference, Las Vegas,
NV, January 20−24, 2002, pp. 669−672.
[19] H. D. Nguyen, D. Hah, P. R. Patterson, R. Chao, W. Piyawattanametha, E.K.
Lau, and M.C. Wu, “Angular vertical comb-driven tunable capacitor with high-
tuning capabilities,” J. Microelectromech. Syst., vol. 13, no. 3, pp.
406−413, June 2004.
[20] J. Y. Park, Y. J. Yee, H. J. Nam, and J. U. Bu, “Micromachined RF MEMS
tunable capacitors using piezoelectric actuators,” IEEE MTT-S Int. Microwave
Symp. Dig., vol. 3, pp. 2111–2114, 2001.
[21] Z. Feng, W. Zhang, B. Su, K. F. Harsh, K. C. Gupta, V. Bright, and Y. C. Lee,
“Design and modeling of RF MEMS tunable capacitors using electro-thermal
actuators,” IEEE MTT-S Int. Microwave Symp. Dig., vol. 4, pp. 1507–1510,
1999.

9
[22] M. J. Vellekoop, C. C. G. Visser, P. M. Sarro, and A. Venema, “Compatibility
of zinc oxide with silicon IC processing,” Sensors Actuators, vol. A23, pp.
1027–1030, 1990.

10
Chapter 2
Approach to Acoustic Ejector with LWAR

This chapter describes the design, simulation, and fabrication of the proposed
acoustic ejector employing lens with air reflectors (LWAR). The experimental
testing setup for droplet ejection is also presented.

2.1 Design and Simulation
The acoustic ejector mainly consists of an acoustic transducer and an acoustic
lens (Fig. 2.1). The detailed working principle of the lens and the ejector are
described as follows.

2.1.1 Working Principle
PZT is used as the piezoelectric material for acoustic transducer due to its large
electromechanical coefficient and capability to produce large acoustic power. The
127-μm-thick PZT sheet (with the fundamental thickness-mode resonance frequency
of 18 MHz) sandwiched between two nickel electrodes serves as the acoustic
transducer. By taking advantage of the fact that air has an acoustic impedance (~400
Rayl) much smaller than that of any solid, LWAR provides a method to focus the
acoustic waves. As acoustic waves propagate at the interface of two different
materials (from material 1 to material 2), the fraction of the incident energy that is

11
reflected is calculated as
2
1 1 2 2 1 1 2 2
)] cos / cos / /( ) cos / cos / [( α α α α Z Z Z Z R + − = ,
where Z
1
and Z
2
are the acoustic impedances of the two materials and α
1
and α
2
are
the incident and transmitted angles. Due to impedance mismatch, the acoustic waves
produced by the PZT are mostly reflected at the transducer-air interface. To ensure
efficient acoustic transmission, the lens structure was built with parylene (that is
biocompatible), because its acoustic impedance (2.8 MRayl) is between that of water
(1.5 MRayl) and the transducer (33 MRayl). As a result, the acoustic waves are
transmitted into liquid through the parylene but reflected back by the air pockets as
illustrated in Fig. 2.1.

Fig. 2.1 Cross-sectional view of the ejector employing the acoustic
lens with air-reflectors.
Coupling
Reflection
Air Air Air Air
127μm PZT
Parylene
2mm
Water
800μm
Top view Nickel
Coupling
Reflection
Air Air Air Air
127μm PZT
Parylene
2mm
Water
800μm
Top view Nickel

12
The parylene lens is patterned into Fresnel half-wave bands (with the kth
radius given by ) 4 / ( λ λ k F k + × [1], where λ and F are the acoustic wavelength and
the lens focal length, respectively) so that the transmitted acoustic waves arrive at
the liquid surface in phase, constructively interfering with each other and
intensifying the acoustic pressure. The intensified acoustic beam ejects liquid
droplets with their size primarily determined by the diameter of the focused acoustic
beam that is directly related to the acoustic wavelength.
It is noted that for a conventional Fresnel lens, in order to get the desired ° 180
phase shift difference between the waves traveling through the liquid and through the
lens, thickness of the lens material, h, should be tightly controlled to be
[ ] ) ( 2 / 1
1 1 − −
− =
s l
V V f h , where f is the frequency, and V
l
and V
s
are the acoustic
velocities in the liquid and lens, respectively [2]. However, unlike the conventional
Fresnel lens, LWAR does not necessitate tight control over the parylene thickness or
the gap distance of the air pocket.
Since the droplet is ejected in the vertical direction and is most related to the
vertical particle displacement, simulations are performed to visualize the improved
focusing effect of the lens as shown in Fig. 2.2. The ejector without lens is also
shown for comparison. The sharp increase of the vertical particle displacement at the
focal point indicates that the acoustic wave is well focused into a narrow beam width
at the center of the top liquid surface. The focused acoustic beam is 5.5 times
stronger in intensity and 4 times narrower in width (full-width-half-maximum
(FWHM)) for the ejector with lens than without lens.

13

2.1.2 Liquid Delivery for Continuous Droplet Ejection
To get continuous ejection, the ejector is integrated with a reservoir, buffer and
microchannel, which are microfabricated with two silicon wafers (Fig. 2.3). A
silicone o-ring is used as a dam for the reservoir. Due to hydrostatic pressure and
surface tension, all water in the reservoir can be delivered to the ejection chamber,
where the liquid level is maintained at the top surface of the silicon chamber, as the
liquid is ejected out by the ejector. With this automatic liquid delivery, continuous
ejections can happen without any liquid refilling or frequency tuning until all the
liquid in the reservoir is consumed.

Fig. 2.2 Vertical particle displacements on the
liquid surface for ejectors with and without lens.
-1000 -500 0 500 1000
0
2
4
6
8
10
12

Relative vertical particle displacement
Distance from focal plane center (μm)
Ejector with air-reflected lens
Ejector without air-reflected lens
with lens employing air-reflectors
without lens employing air-reflectors
-1000 -500 0 500 1000
0
2
4
6
8
10
12

Relative vertical particle displacement
Distance from focal plane center (μm)
Ejector with air-reflected lens
Ejector without air-reflected lens
with lens employing air-reflectors
without lens employing air-reflectors

14

2.1.3 Directional Acoustic Ejection
To generate directional ejections, an asymmetrical electric field is intentionally
created within the piezoelectric layer to produce lopsided acoustic waves (Fig. 2.4),
which are focused on the liquid surface to obliquely eject nanoliter droplets.
When both top and bottom nickel electrodes of the PZT transducer are
patterned into a pie shape (Fig. 2.4), asymmetric acoustic fields are generated from
the “pie-shaped” sector electrodes, in contrast to symmetric acoustic fields generated
from circular electrodes [3]. The electric field applied across the thickness of the
piezoelectric PZT causes the PZT to vibrate, producing acoustic waves. Since, to the
first order, the vibrations happen only at the regions covered with the sector
electrodes, uneven acoustic pressure distributions are produced at the liquid surface.
Fig. 2.3 Integrated ejector for automatic liquid delivery. (a) Cross-sectional
view. (b) SEM photo of the buffer and the chamber before being integrated
with the ejector (upside-down view).
Si wafer #2
180 μm
1 mm opening
800 μm
Si wafer #1
PZT
2 mm lens
reservoir
chamber
250 μm
Si wafer #2
Si wafer #1
chamber
silicone o-ring
buffer
buffer
(a)
(b)
Si wafer #2
180 μm
1 mm opening
800 μm
Si wafer #1
PZT
2 mm lens
reservoir
chamber
250 μm
Si wafer #2
Si wafer #1
chamber
silicone o-ring
buffer
buffer
(a)
(b)

15

The simulation of particle displacements at the liquid surface for 90
o
sector
electrodes is shown in Fig. 2.5. As the sector angle decreases, the vertical particle
displacement becomes less intensified, while the relative lateral displacement
becomes larger (Table 2.1). As the acoustic radiation pressure is unbalanced in the
plane of the liquid surface, the droplet ejection happens in a direction oblique to the
liquid surface. The ejection becomes more tilting as the sector angle gets smaller [3].
However, a smaller sector angle means a smaller actuation area and weaker acoustic-
power generation. Thus it requires a larger minimum pulsewidth for stable ejections.
Though it has been reported that the ejection stability and droplet size are
sensitive to the pulsewidth, the pulsewidth effect on acoustic ejection remains
unclear for lack of theory [4]. In the case of directional ejection, we find that the
Air
PZT Transducer
Parylene
Air
Ejector 1
Ejector 3
Ejector 4
Ejector 2
Reservoir
Ejector
Focal point
Silicon
PZT
90
o
sector
Air
PZT Transducer
Parylene
Air
Ejector 1
Ejector 3
Ejector 4
Ejector 2
Reservoir
Ejector
Focal point
Silicon
PZT
90
o
sector
Fig. 2.4 Schematic diagram of directional acoustic ejector array. Each ejector is
integrated with its own reservoir for continuous liquid supply for ejections.

16
directionality is also a function of pulsewidth and that a larger pulsewidth would
result in a less oblique ejection angle. A 90
o
sector is used in our design. In addition
to a single ejector, four ejectors are coordinately arrayed on a single chip to target
one spot in the center with multiple liquids.

Apex angle of sectored electrode 360
o
270
o
180
o
90
o

Normalized particle displacement 8.1 6.4 4.4 2.0
Ratio of lateral to vertical particle displacement 0 0.16 0.37 0.59

2.2 Device Fabrication
The acoustic ejector was built on a 127-μm-thick PSI-5A4E PZT sheet (Piezo
Systems, Cambridge, MA). The fabrication process steps are shown in Fig. 2.6.
Top view: 90
o
sector
X
-20 0 -10 0 0 10 0 20 0
-10
0
10
20
30
40
50
Focal point
Particle displacement vector at liquid surface
(Y-axis position = 0)
0
0.2
0.4
0.6
0.8
1.0
X-axis position (μm)
Relative value (unit less)
Y
Z
X
Y
Top view: 90
o
sector
X
-20 0 -10 0 0 10 0 20 0
-10
0
10
20
30
40
50
Focal point
Particle displacement vector at liquid surface
(Y-axis position = 0)
0
0.2
0.4
0.6
0.8
1.0
X-axis position (μm)
Relative value (unit less)
Y
Z
X
Y
Fig. 2.5 Simulation of particle displacements at liquid surface for
90
o
sector electrode.
Table 2.1 Particle displacements on the liquid surface (integrated over a 400 μm
by 400 μm area) for different electrode patterns.

17

On both sides of the PZT sheet, nickel electrodes were first patterned, followed
by spinning and patterning of 3-μm-thick photoresist as the sacrificial layer. After
we deposited (in Special Coating Systems, Indianapolis, IN) and patterned (with O
2

reactive ion etching) 3-μm-thick parylene-D as the lens material (with release holes
of 30 μm in diameter), the photoresist was removed with acetone. Another 4-μm-
thick parylene was then deposited to fill the release holes.
The microfluidic components (embedded microchannels, ejection chambers
and reservoirs) were microfabricated with two silicon wafers. Both sides of silicon
wafers were first deposited with 0.8-µm-thick Si
x
N
y
by low-pressure chemical vapor
deposition (LPCVD). The front-side Si
x
N
y
was then patterned, followed by
anisotropic etching of bulk silicon in KOH. After etching silicon for the microfluidic
Fig. 2.6 Brief fabrication steps for acoustic ejector employing lens with
air-reflectors.
(a) Pattern nickel electrodes on
127μm thick PZT sheet
(c) Deposit and pattern 3μm
parylene as lens material
(d) Remove photoresist with
acetone
(e) Deposit 4μm parylene to fill
release holes
(f) Bond PZT sheet and silicon
chamber with epoxy
(b) Spin and pattern 3μm photo-
resist as sacrificial layer
(a) Pattern nickel electrodes on
127μm thick PZT sheet
(c) Deposit and pattern 3μm
parylene as lens material
(c) Deposit and pattern 3μm
parylene as lens material
(d) Remove photoresist with
acetone
(d) Remove photoresist with
acetone
(e) Deposit 4μm parylene to fill
release holes
(e) Deposit 4μm parylene to fill
release holes
(f) Bond PZT sheet and silicon
chamber with epoxy
(f) Bond PZT sheet and silicon
chamber with epoxy
(b) Spin and pattern 3μm photo-
resist as sacrificial layer

18
components, the Si
x
N
y
was removed, and two silicon wafers were bonded together
with epoxy. Finally, the PZT sheet was adhesively bonded to the silicon wafers in
which the 800 μm deep (matching the lens focal length) chambers were
microfabricated.
Since the release-hole sealing was realized with parylene coating, the formed
air pocket was practically sealed in vacuum. Though parylene has a relatively small
Young’s modulus, since the dimensions of the air-reflector are also small, the
ambient pressure was calculated not to cause significant structure deformation. Finite
element analysis (FEA) software (Algor) was used to analyze the design. Simulation
results showed that there was negligible displacement of less than 0.1 μm as one
atmosphere static pressure is exerted on the parylene lens structure (Fig. 2.7).

(μm)
Parylene structure
Parylene sealing
PZT
7 μm
60 μm
1000 μm
1 atm
3 μm
(μm)
Parylene structure
Parylene sealing
PZT
Parylene structure
Parylene sealing
PZT
7 μm
60 μm
1000 μm
1 atm
3 μm
Fig. 2.7 Algor simulation showing negligible displacement as one
atmosphere static pressure is exerted on the parylene lens structure.

19
Figure 2.8 shows the scanning electron microscope (SEM) photos of the
fabricated devices. Also shown in Fig. 2.9 are SEM photos of the release holes
before and after being filled with parylene. It can be seen in these photos that the
release holes are properly sealed and no deformation of parylene structure is
observed.
Figure 2.10 shows the photos of the fabricated array of PZT directional
ejectors and silicon chambers before they were adhesively bonded together. The
fabricated device was packaged in a dual-in-line (DIP) package, and the package was
placed on a mounting station with high frequency microstrips. The photo of the
finished device in a DIP package is shown in Fig. 2.11, where we denote the inlet
ports for liquid refilling and the openings (500 μm × 500 μm) for ejections.

Fig. 2.8 SEM photos of the fabricated ejector with LWAR: (a) top
view and (b) side view.
(a) (b)
parylene with air underneath
parylene directly on PZT
parylene structure layer
PZT sheet
parylene structure layer
PZT
parylene
air gap
(a) (b)
parylene with air underneath
parylene directly on PZT
parylene structure layer
PZT sheet
parylene structure layer
PZT
parylene
air gap
parylene with air underneath
parylene directly on PZT
parylene with air underneath
parylene directly on PZT
parylene with air underneath
parylene directly on PZT
parylene structure layer
PZT sheet
parylene structure layer
PZT
parylene
air gap
parylene structure layer
PZT sheet
parylene structure layer
PZT
parylene
air gap

20

Fig. 2.9 SEM photos of the release holes: (a) before and (b)
after being filled with parylene.
release hole
release hole filled with parylene
(a)
(b)
release hole
release hole filled with parylene
(a)
(b)

Fig. 2.10 Photos of (a) ejector array and (b) chamber array.
2mm
chamber 3
buffer 3
chamber 4
chamber 1
buffer 1
chamber 2
2mm
1.4mm
PZT
lens electrode
ejector 4
ejector 2 ejector 1
ejector 3
(a) (b)
2mm
chamber 3
buffer 3
chamber 4
chamber 1
buffer 1
chamber 2
2mm
1.4mm
PZT
lens electrode
ejector 4
ejector 2 ejector 1
ejector 3
2mm
1.4mm
PZT
lens electrode
ejector 4
ejector 2 ejector 1
ejector 3
(a) (b)

SMA
transmission line
DIP
ejection
opening
liquid
inlet
3mm
ejector 4
ejector 2 ejector 1
ejector 3
SMA
transmission line
DIP
ejection
opening
liquid
inlet
3mm
ejector 4
ejector 2 ejector 1
ejector 3

Fig. 2.11 Fabricated device in a DIP package connected to
transmission lines and SMA connectors.

21
2.3. Experimental Testing Setup
For the device testing, pulses of sinusoidal signals are applied to the ejectors to
eject droplets in a setup shown in Fig. 2.12. First, the sinusoidal signal is modulated
with RF pulses through a high-speed switch. The pulse repetition frequency (PRF)
used is ranging from 1 Hz to 10 kHz, and the pulsewidth is from 7 to 100 μs. The
pulsed signal is then amplified with an RF power amplifier (A041, LCF Enterprises)
and fed into the device. For an array of four ejectors, the amplified pulsed signal is
split into four equal signals through a power splitter and an RF switch array, before
being applied to each of the ejectors. Each ejector is individually actuated by a
computer program. A charge-coupled device (CCD) camera (SONY SSC-DC54A)
with a microscope is placed horizontally to record the ejection process as we
stroboscopically blink a light-emitting diode (LED). Synchronization of the flash
illumination with the sinusoidal pulse input is achieved by turning on the LED with
another pulse source triggered by the pulse generator that pulses the sinusoidal signal.
By varying the delay time between the illumination of LED and the RF signal
applied to the transducer, we observe the ejection process at any moment. The photo
of the actual testing setup is shown in Fig. 2.13.

22

Summary
A new acoustic ejector with LWAR has been proposed, designed, simulated,
and fabricated. This ejector requires no nozzle, nor heat for actuation, but uses a
Fig. 2.13 Photo of the experimental testing setup.
Device under test
Low speed switch
array
Computer control
Power supplies,
pulse generator and
RF signal generator
CCD camera and
microscope lens
Device under test
Low speed switch
array
Computer control
Power supplies,
pulse generator and
RF signal generator
CCD camera and
microscope lens
Fig. 2.12 Schematic diagram of testing setup for droplet ejections.
RF Signal
Generator
High
Speed RF
Switch
RF Power
Amplifier
Power
Splitter
Pulse
Generator
Delayed
Pulses
(strobe)
Microscope Lens
Computer
Control
Program
1/PRF
Switch Array Controlled by
Computer
pulsewidth
LED
CCD
Camera
RF Signal
Generator
High
Speed RF
Switch
RF Power
Amplifier
Power
Splitter
Pulse
Generator
Delayed
Pulses
(strobe)
Microscope Lens
Computer
Control
Program
1/PRF
Switch Array Controlled by
Computer
pulsewidth
LED
CCD
Camera

23
novel lens, which does not require tight thickness control for effective focusing. The
ejector has further been integrated with a reservoir and channel for an automatic
liquid delivery. The design and simulation for directional ejections have also been
described along with the experimental testing setup.

24
Chapter 2 References
[1] D. Huang and E. S. Kim, “Micromachined acoustic-wave liquid ejector,” J.
Microelectromech. Syst., vol. 10, pp. 442-449, 2001.
[2] S. C. Chan, M. Mina, S. S. Udpa, L. Udpa, and W. Lord, “Finite element
analysis of multilevel acoustic Fresnel lenses,” IEEE Trans. Ultrason.,
Ferroelect., Freq. Contr., vol. 43, pp. 670-677, 1996.
[3] J. W. Kwon, Q. Zou, and E. S. Kim, “Directional ejection of liquid droplets
through sectoring half-wave-band sources of self-focusing acoustic transducer,”
Proceedings of the 15th IEEE International Conference on
Microelectromechanical Systems, Las Vegas, NV, 2002, pp. 121–124.
[4] S. A. Elrod, B. Hadimioglu, B. T. Khuri-Yakub, E. G. Rawson, E. Richley, C. F.
Quate, N. N. Mansour, and T. S. Lundgren, “Nozzleless droplet formation with
focused acoustic beams,” J. Appl. Phys., vol. 65, pp. 3441-3447, 1989.

25
Chapter 3
Results and Applications of Acoustic Ejector with LWAR

This chapter describes the characterization results of the designed acoustic
ejectors with LWAR. Based on the LWAR ejectors, a series of biochemical
applications using nanoliter droplets are presented, including on-demand DNA
synthesis, droplets coalescence in air, microreactions with oil encapsulation, and
harmonic operations for ejection of smaller droplets.

3.1 Ejector Performance Characterization
The droplet formation sequence, droplet ejection rate, continuous droplet
ejection, and directional ejection performance are thoroughly characterized as
follows.

3.1.1 Droplet Formation Sequence
The fabricated ejector is driven with pulses of 18 MHz sinusoidal signals of
±60V
peak-to-peak
(the peak electrical field across the PZT substrate being around 0.47
MV/m, which is smaller than the PZT polarization field of around 2 MV/m). The
time evolution of the droplet formation by an RF pulse having width of 7 μs and
energy of 63 μJ is shown in Fig. 3.1. The droplet separation time and the liquid-
surface relaxation time are estimated to be 100 μs and 130 μs, respectively. The

26
droplet size of the acoustic ejector depends mostly on the wavelength of the acoustic
wave (which is determined by the RF resonant frequency of the PZT substrate). For
the same ejector driven with the same electrical condition, there is no observable
variation in droplet size. For a set of 20 ejectors, the droplet size ranges from 70 μm
to 80 μm possibly due to the resonant frequency variations from the PZT substrate.
The droplet size can be reduced by using a thinner PZT substrate (or film) or through
harmonic operations of the transducer, which will be described in a later paragraph.

3.1.2 Ejections at High Ejection Rates
Figure 3.2 shows the ejections at different ejection rates. The ejection is one
droplet per pulse and free of satellite droplets. The images in Fig. 3.2(a), (b), and (c)
are a superposition of 16, 32, and 64 successive droplets, respectively, and the
sharpness of the images demonstrates that the ejection speed is consistent (1.5 m/s);
the droplet size is invariable; and the formation process is consistent for every
200 μm
0 μs
20 μs 50 μs
100 μs 120 μs 130 μs
200 μm
0 μs
20 μs 20 μs 50 μs 50 μs
100 μs 100 μs 120 μs 120 μs 130 μs 130 μs

Fig. 3.1 Photos of droplet formation sequences.

27
droplet ejection. Moreover, though the liquid-relaxation time is around 130 μs,
ejection is observed at a rate up to 10 kHz, i.e. with an interval of 100 μs between
two consecutive shootings, as shown in Fig. 3.2(d). This indicates the feasibility of
ejection without the liquid-surface’s returning to its flatness, and a higher ejection
rate is achievable.

3.1.3 Stable and Continuous Ejection
When driven with pulses of 7 μs pulsewidth at 60 Hz PRF, the ejector
produces continuous ejection for more than 75 seconds, during which the RF
frequency is fixed at 18 MHz without any fine tuning. Fig. 3.3 shows the continuous
ejection photos taken every 15 seconds from the beginning. Frames at different time
are almost identical, again exhibiting the uniformity of droplet sizes and the stability
of ejection process.

(a) 1 kHz
70 μm
2.0 mm
70 μm
(b) 2 kHz
70 μm
(d) 10 kHz
70 μm
(c) 4 kHz
(a) 1 kHz
70 μm
2.0 mm
70 μm
(b) 2 kHz
70 μm
(d) 10 kHz
70 μm
(c) 4 kHz
Fig. 3.2 Photos of droplet ejections at (a) 1 kHz, (b) 2 kHz,
(c) 4 kHz, and (d) 10 kHz.

28

3.1.4 Comparisons with Ejector without LWAR
An ejector without the lens has also been fabricated and tested for comparison.
As expected, this ejector requires higher power (minimum pulsewidth of 20 μs) for
ejection and the produced droplet size (165 μm) is larger due to lack of focusing
effect (Fig. 3.4). It is also found that the highest ejection rate for this ejector is only
500 Hz due to its larger droplet size and greater liquid mounting-up before ejection.

3.1.5 Directional Ejections
The directional ejections of the ejectors with quadrant electrodes are
characterized. The ejection process is as stable as the vertical ejection. Fig. 3.5
165 μm
2 mm 2 mm
70 μm
ejection without lens ejection with lens
165 μm
2 mm 2 mm
70 μm
ejection without lens ejection with lens

Fig. 3.4. Comparisons between the ejectors with and without the lens.
after 15sec after 30sec
after 45sec after 60sec after 75sec
at the beginning
1mm opening
Si cover
70 μm
after 15sec after 15sec after 30sec after 30sec
after 45sec after 45sec after 60sec after 60sec after 75sec after 75sec
at the beginning
1mm opening
Si cover
70 μm

Fig. 3.3 Photos of continuous ejections at different times.

29
shows the continuous ejections driven with ±60-V
peak-to-peak
pulses of 18-MHz
sinusoidal signals without any fine frequency tuning. The pulsewidth and the PRF
are 7 μs and 60 Hz, respectively. The ejection is one droplet per pulse and free of
satellite droplets. The frame taken right at the beginning of the ejections is almost
identical to the one after 30 seconds of ejections (1800 droplet shootings).
Fig. 3.6 shows the ejections at the ejection rates of 1 kHz and 2 kHz. For the
same ejector driven with the same electrical condition, there is no observable
variation in droplet directionality. For ejectors fabricated within a same PZT
substrate, the directional angle varies within 5 degree. This is possibly due to the fact
that the PZT ejectors were manually bonded to the silicon chambers and the
misalignment of the ejector to the ejection opening would result in a different
boundary condition for the ejection. In addition, the relatively large undercut of the
nickel electrodes on the PZT may also cause the electrode pattern variation among
the devices and influences the directional angle.

100 μm
(a) (b)
100 μm
(a) (b)

Fig. 3.5 Stable and continuous directional ejection photos taken (a) at
the beginning of the ejections and (b) after 1800 droplet ejections.

30
In conclusion, we have demonstrated an acoustic ejector that requires no
nozzle, nor heat for actuation, but uses a novel lens. The LWAR focuses the acoustic
beams at the liquid surface, and facilitates uniform and consistent droplet ejections.
This efficient droplet-on-demand ejector proves to be an alternative to current
nozzle-based ink-jet printers, particularly in the applications where nozzle clogging
is an issue.

3.2 Nanoliter Droplet Coalescence in Air
We present here a controlled coalescence of nanoliter liquid droplets in air by
acoustic directional ejections. Four directional ejectors, integrated with their own
800- μm-deep reservoirs, are arrayed on a single chip to target one spot in the center.
Up to four obliquely-ejected droplets coalesce in air into a single droplet, which then
continues to travel, rotating at 16,000 rad/s and producing effective micromixing in
air.
500 μm
(a) (b)
500 μm
(a) (b)

Fig. 3.6 Photos of the directional droplet ejections at
the ejection rates of (a) 1 kHz and (b) 2 kHz.

31
3.2.1 Coalescence of Two Droplets
We first actuated Ejectors 1 and 2 using water as the liquid medium (Fig.
3.7(a)). The ejectors were driven with ±80-V
peak-to-peak
pulses. The pulsewidth and the
PRF were 16 μs and 60 Hz, respectively. The droplets had a diameter of around 80
μm, very close to the acoustic wavelength of 82 μm. Two spherical droplets broke
off simultaneously from their own bulk liquids. They then approached each other
with an ejection angle of ° 60 at a speed of 2.3 m/s. At around 1100 μs, they arrived
at the same place in air at the same time and coalesced into one larger ellipsoid. The
coalesced droplet kept on traveling in air with no velocity along the x ˆ direction. The
droplets traveling trajectories are illustrated in the micrographs without strobe (Fig.
3.7(b)). The droplet size (80 μm) is much smaller than the distance between the two
ejectors (2 mm), yet the two droplets still meet in a three-dimensional space owing to
the same ejection/traveling speed and precise directionality.

Fig. 3.7 (a) Optical micrographs with strobe to show the digital droplets during
the ejection process (Ejectors 1 and 2 activated). (b) Optical micrographs
without strobe to illustrate the droplets traveling trajectories. The shutter-
opening time of the CCD camera was set to 2 ms to record the images.
1100 μs 600 μs 1500 μs 200 μs
500 μm
x
y
z
(a)
(b)
500 μm
1100 μs 600 μs 1500 μs 200 μs
500 μm
x
y
z
(a)
(b)
500 μm

32
As two droplets collide, bounce will occur if the droplets’ surfaces undergo a
flattening deformation without making contact due to a thin air layer between the
droplets. The colliding droplets will coalesce when the air layer thickness reaches a
critical value, typically of the order of 100 Å [1]. In coalescence, a post-collision
droplet with its mass equal to the sum of the masses of the pre-collision droplets is
formed. The droplets may coalesce permanently or temporarily, depending on the
collisional kinetic energy (CKE) and impact parameter, b, which is defined as the
distance from the center of one droplet to the relative velocity vector placed on the
center of the other droplet as they collide [1]. The CKE of the droplet pair composing
of the same liquid is given by
2
33 3 3
12 1 2 1 2
( /12) /( ) CKE D D D D v v ρπ=+−
rr
, where ρ is
the liquid density; D
1
and D
2
are the diameters; and
2 1
v v
r r
− is the relative velocity
of the droplets [2]. Since the post-collision droplet has a smaller surface energy,
C
S ,
than the pre-collision droplets,
T
S , the decrease in surface energy is calculated as
2
22 3 3
3
12 1 2
() ( )
TC
SS S D D D D
σ
πσ πσ Δ= − = + − + , where σ is the surface tension of the
liquid [2]. The total collisional energy E
TC
, given by
TC
ECKE S
σ
= +Δ , is dissipated
through the deformations and rotations of the droplets. The deformed droplets
dissipate the energy off to the air and through the viscous action of the liquid. The
rotational energy of the coalesced droplet is calculated as
2
(1/ 2)
r
EI ω = , where I
and ω are the mass moment of inertia and angular frequency, respectively. As the
rotational energy is transferred from collisional energy, the proceeding velocity,
c
v
r
,

33
and translational kinetic energy, E
t
, of the post-collision droplet are given by
11 2 2 1 2
()/( )
c
vmv mv m m =+ +
rr r
and
2
12
(1/ 2)( )
tc
Emmv =+ , where m
1
and m
2
are the
masses of the droplet pair. The total kinetic energy of the coalesced droplet, E
k
, is the
sum of
t
E and
r
E . In temporary coalescence,
TC
E is so large that the coalesced
droplet undergoes catastrophic fragmentation into numerous small droplets [2].
According to experimental data, whether droplets bounce, coalescence or fragment is
typically predictable in the (b, We) plane, where the Weber number is defined as
2
12 1 2
() /2 We D D v v ρ σ =+ −
rr
.
Here we used D
1
=D
2
=80 μm,
2 1
v v
r r
− =1.7 m/s, b=50 μm, σ =0.073 N/m,
ρ=1000 kg/m
3
and ω=16,000 rad/s to calculate We, CKE, S
T
, S
C
, ΔS
σ
, E
TC
, E
r
, E
t
and
v
c
to be 3.2, 0.19 nJ, 2.9 nJ, 2.3 nJ, 0.6 nJ, 0.79 nJ, 0.07 nJ, 1.2 nJ and 2.1 m/s,
respectively. With b=0.6D and We=3.2, the collision falls into the stable coalescence
region [1]. The droplet size, relative velocity and angular frequency are in good
quantitative agreement with other droplet collision studies on atmospheric raindrop
formation and spray combustion [1, 3]. In Adam’s work, as two 60- μm water
droplets collide with a relative velocity of 2 m/s, the coalesced droplet rotates at
15300 rad/s. However, that particular experiment required a very large apparatus
with two ejectors that were manually mounted and aligned.
The ejection sequences of Ejectors 2 and 4 show the coalescence from another
angle (Fig. 3.8). After coalescence, the coalesced prolate spheroid (cigar-shaped
ellipsoid) droplet rotated with a periodicity of 400 μs (corresponding to an angular

34
frequency of 16,000 rad/s), as its airborne travel continued (Fig. 3.8(c)). At 1100 μs
(the start of one cycle), the semi-major axis is along y ˆ direction. At 1200 μs, the
ellipsoid has rotated 90
o
and the semi-major axis is perpendicular to y ˆ direction (on
the xz plane). At 1300 μs, the ellipsoid has rotated 180
o
and the semi-major axis is
along y ˆ direction again (but with 180
o
phase difference). At 1400 μs, the ellipsoid
has rotated 270
o
and the semi-major axis lies on the xz plane. At 1500 μs, the
ellipsoid has rotated 360
o
, completing one cycle with a periodicity of 400 μs.

Fig. 3.8 Optical micrographs of the droplets ejection process (Ejectors 2 and 4
activated). (a) Overview pictures of the ejection process. (b) Detailed
coalescence sequence of the droplets around 1100 μs. (c) Rotations of the
traveling coalesced droplet with a periodicity of 400 μs.
1500 μs 1100 μs 600 μs 200 μs
500 μm
1100 μs 1060 μs 1080 μs
100 μm
1040 μs
200 μm
1100 μs 1200 μs 1300 μs 1400 μs 1500 μs
v
c
(a)
(c)
(b)
x
y
z
1500 μs 1100 μs 600 μs 200 μs
500 μm
1100 μs 1060 μs 1080 μs
100 μm
1040 μs
200 μm
1100 μs 1200 μs 1300 μs 1400 μs 1500 μs
v
c
(a)
(c)
(b)
x
y
z

35
3.2.2 Application to Micromixing
Rapid and homogenous mixing of two or more fluidic species is essential for
micro total analysis systems ( μTAS). The rotating characteristics of the coalesced
droplet can be applied to micromixing. In a conventional microfluidic system,
laminar flow occurs due to a low Reynolds number. Diffusion, dominant at the
microscale, takes a relatively long time and decreases the overall system efficiency.
The coalescence by directional ejections offers the following benefits for
micromixing. First, by ejecting and mixing reagents in air without long channels on a
substrate, which are typically required for lab-on-a-chip [4], the chip real estate is
greatly reduced. Second, through rotations from an off-axis collision or oscillations
from a head-on collision [5], mixing can be achieved more efficiently than by
passive diffusion.
To illustrate the mixing effect, one ink and one water droplet were ejected for
coalescence in air. Performing mixing analysis for airborne droplets from the side-
view micrographs was relatively difficult, and we examined the droplets collected on
a glass slide placed 2-cm above the device (Fig. 3.9(a)). The number of revolutions
N
r
(at a traveling distance S) is calculated as /2
rc
NS v ω π = . With ω=16,000 rad/s
and v
c
= 2.1m/s, the coalesced droplet was mingled swiftly by rotating about 25
revolutions before it reached the glass slide. The droplet landing position could be
controlled within 80 μm. The droplet homogeneity exhibited the excellent mixing by
the precise directional ejections and the continuous flight and rotations after the
coalescence.

36
To further quantify the mixing performance, the mixing index is calculated as
2
1
1
()
N
k
k
cc
Mixing index
Nc
=
−
=
∑
……………………………………………......(3.1)
where c
k
and c are the color index at pixel k and the average of N pixels over an
image area [6], respectively. The smaller the index is, the better the mixing is.
We compared the time required to mix the same liquid volume by rotation in
air and by diffusion. To carry out the diffusion experiment in small volume without
evaporation, one water and one ink droplets were ejected inside the oil and mixed by
diffusion. It was observed that diffusion took 6 sec to reach the mixing index of 0.13
while rotation in air took less than 0.05 sec to reach 0.11, a 120-times reduction in
time. Thus, as the mixing was performed by rotation in air, the passive diffusion
effect was much smaller than the mixing effect by an active rotation. In addition, the
trajectories of any two different liquid droplets will lie on the same plane and
intersect in air as long as the two ejectors are placed symmetrically along a line as
shown in Fig. 3.9(b). Thus even liquids with different viscosity and density can meet
at a same spot by controlling the delay time between the actuation pulses to the
ejectors.

37

3.2.3 Coalescence of Four Droplets
To demonstrate that more than two droplets can be merged in air, we
simultaneously actuated four ejectors on a chip. Driven with ±80-V
peak-to-peak
pulses
of 30- μs pulsewidth and 60-Hz PRF, the four ejected droplets moved toward the
center with an ejection angle of ° 75 and joined together in air after 1800 μs (Fig.
3.10). It is noted that when more ejectors were simultaneously actuated, a larger
pulsewidth was required and droplets would meet at a higher meeting point due to a
less oblique ejection angle. For coalescence of two droplets, 80% of the devices can
Fig. 3.9 (a) Optical micrographs of the water, ink, and mixed droplets ejected
onto a glass slide by directional ejections. Water was used as the medium for
Ejector 1, while red ink was used for Ejector 2. A glass slide was placed
above the ejector to collect the droplets. For comparison, one water droplet
(~0.27 nL) was first ejected by Ejector 1 and collected (image 1) and then
one ink droplet (~0.27 nL) was ejected by Ejector 2 (image 2). Finally, one
water and one ink droplet were simultaneously ejected and coalesced in air
(image 3). The coalesced droplet continued traveling until it was collected on
the glass slide (image 4). (b) Schematic diagram showing that the trajectories
of any two different liquid droplets intersect in air.
2 mm
3
100 μm 100 μm
1
100 μm
2 4
Ejector 1 Ejector 2
v
1
v
2
θ
1
θ
2
Top view
Side view
Meeting point
(a) (b)
2 mm
3
100 μm 100 μm
1
100 μm
2 4
Ejector 1 Ejector 2
v
1
v
2
θ
1
θ
2
Top view
Side view
Meeting point
v
1
v
2
θ
1
θ
2
Top view
Side view
Meeting point
(a) (b)

38
eject droplets for coalescence in air. In addition, all the working devices can 100%
repeat the coalescence process. It becomes harder to coalescence three or four
droplets in air due to the resonant frequency variations among the ejectors. However,
for the working ejectors, the coalescence process is 100% repeatable.
When a single ejector is actuated, there is no satellite droplet at all. When two
or more ejectors are simultaneously actuated, satellite droplets have been observed.
This is because that the PZT substrate that we used has a 5% thickness variation,
resulting in resonant frequency difference among ejectors. When only one frequency
was used, not all ejectors were operated at their optimum condition and a larger
pulsewidth was required to compensate for the lack of resonance enhancement.
Hence satellite droplets were produced. One can apply separate frequency signals to
different ejectors or use a more uniform piezoelectric substrate to reduce the satellite
droplets upon simultaneous ejections.

500 μm
200 μs600 μs 1000 μs 1700 μs 1800 μs
500 μm
200 μs600 μs 1000 μs 1700 μs 1800 μs
Fig. 3.10 Optical micrographs of the ejection process when all the
four ejectors were activated.

39
In conclusion, we have demonstrated a microdroplet dispensing method
capable of targeting a same spot in air, with combinations of different liquids.
Because of its high throughput, positional accuracy, reproducibility, and low
susceptibility to clogging, this digital fluid manipulation by directional droplet
ejections will be useful not only for inkjet printing but also for various biochemical
applications.

3.3 On-Demand DNA Synthesis by Four Directional Ejectors on a Chip
A synthesis technique is presented for any random DNA sequences on solid
surfaces. An array of 2 × 2 directional ejectors having 90
o
sector electrodes is
integrated with microfluidic components on a single chip to synthesize
oligonucleotides. Since a spot on a substrate can be inked by four ejectors (carrying
four DNA bases) without any mechanical movement, the directional ejection
removes the need for aligning the substrate, and minimizes the automation and
control circuitry (Fig. 3.11).

3.3.1 Review of Microarray Fabrication Methods
DNA microarray allows combinatory, parallel genetic analysis [7-8]. There
exist various commercial techniques to produce DNA probes of pre-made
oligonucleotides, which are transferred to thousands of spots on a chip through inkjet
printing or contact dispensing. However, they lack the flexibility for geneticists to

40
explore various DNA sequences as they carry out genotyping [9]. Alternatively,
oligonucleotides can be synthesized in situ to suit individual needs [10].

The Affymetrix GeneChip is one of the major microarray platforms in which
DNA probes are made with a set of 4n photomasks for n-mers with large processing
systems in a clean room environment [11]. As each oligonucleotide set requires a
new mask set, the technique is costly and requires a long turnaround time for custom
designs of DNA probe array. Though maskless technique that uses a digital
micromirror array to form virtual masks has been offered by Xeotron and Nimblegen
[12], the method is still limited by the photolabile-protecting group chemistry [13].
Alternatively, Combinmatrix has proposed to use an electrochemical synthesis [14]
where each microelectrode on the array is selectively addressed to generate acid for
controlling the detritylation reaction during phosphoramidite synthesis. However, the
Directionally ejected bases
Activated target area
G
T
A
C
Poly-l-lysine glass substrate
Directionally ejected bases
Activated target area
G
T
A
C
Poly-l-lysine glass substrate

Fig. 3.11 Schematic diagram illustrating a spot inked
by four directional ejectors carrying four DNA bases.

41
challenges include avoidance of the side reactions on the electrode surface as well as
ensuring the effectiveness of electrochemically-generated acid [13].
Inkjet-based technology developed by Agilent and Rosetta [15] is a more
flexible method that allows faster construction of DNA microarrays. By dispensing
DNA monomers from multi-channel inkjet print-heads, any desired sequences can be
chemically synthesized in parallel. However, the commercial inkjet print-heads shoot
out liquid droplets through nozzles, and hence the smallest droplets depend on the
nozzle size [16]. The phosphoramidite precipitates for DNA synthesis can
accumulate on the print head and clog the nozzles, especially when volatile solvents
are used [17]. In addition, it is challenging to produce directional ejections, and there
should be many mechanical movements when nozzle-based ejectors are used for
DNA synthesis.
The fabrication method of DNA microarrays reported here utilizes an
ultrasonic ejector that requires no nozzle to eject droplets of DNA bases on demand
for any DNA sequences. The nozzleless ejector is capable of ejecting droplets in any
direction, so that a spot on a chip can be inked by any of four such ejectors without
mechanical movement and alignment. The synthesis method is new, and yet faster
and more flexible than existing DNA synthesis method based on photolithography.
For example, for a new 25-mer microarray, up to 100 photolithographic process
steps along with 100 new photomasks are required for the existing method, even
when only one oligonucleotide is changed in the array. The method described here,
however, is flexible and can synthesize a new microarray without any additional

42
costs, i.e., fabrication of a new microarray consumes the same amount of time and
money as reconstruction of a pre-designed microarray. Moreover, the coupling using
standard phosphoramidite chemistry for the new method is more efficient than that
using photolabile chemistry for the existing method, and allows longer
oligonucleotides to be synthesized, hence decreasing the required multiple features
for unambiguous identification of each gene [17]. Since washing and drying are the
rate-limiting steps for the phosphoramidite synthesis and the synthesis time mainly
depends on the sequence length, it will only take hours to fabricate a new microarray.
Compared with the inkjet method, where 4 movements are required to complete
adding any one DNA base of the four bases to the DNA sequence, the new method
not only decreases the mechanical movements by 4 folds, but also minimizes the
automation and control circuitry.

3.3.2 Synthesis by Cyanoethyl Phosphoramidite Chemistry
We used standard cyanoethyl phosphoramidite chemistry [18] to synthesize
any oligonucleotide sequence on a poly-l-lysine coated glass slide. Since
phosphoramidite chemistry can be used for DNA synthesis also on silicon [19] and
plastic [20] surfaces that are modified with poly-l-lysine, this technique is also
suitable for DNA synthesis on plastic or silicon substrates. Four activated monomers
(Alpha DNA, Quebec, Canada) initially contained in four on-chip reservoirs were
brought to the ejection chambers through the embedded microchannels and ejected
by the directional ejectors. A pulse of 18-MHz sinusoidal wave was applied to eject

43
droplets of DNA bases onto a glass slide placed above the device. Each ejector was
individually actuated by a computer program. The 3’-phosphorous of the ejected
monomer was linked by trivalent phosphite bond to the free 5’-hydroxyl on the poly-
l-lysine-coated glass slide. The glass slide, attached to a motorized rotation stage
(Newmark Systems, Mission Viejo, CA), was then rotated by 180
o
to a wash/dry
area to be treated with capping solution, which blocked out the nucleotides that had
failed to couple. An oxidation step was then performed to stabilize the coupling,
which was followed by deblocking process to remove dimethoxytrityl (DMT)
protecting group at 5’-end for incoming oligonucleotides (Fig. 3.12). Next, the glass
slide was rotated back without any alignment with the ejector array to the synthesis
area at the starting step of next iterating cycle.
We have carried out the DNA synthesis under a fume hood (Micro Air
Systems, Santa Clara, CA) with air filters to prevent potential contamination. Fig.
3.13 shows the synthesis and wash areas under a fume hood for DNA synthesis and a
close-up photo of the poly-l-lysine glass slide placed above the ejector array. This
oligonucleotide synthesis is a cyclic sequence of reactions, adding one nucleotide to
the growing oligonucleotide chain and proceeds in four main steps during each cycle:
activation/coupling, capping, oxidation, and deblocking [21]. The glass slide was
rotated between the synthesis and wash areas repeatedly.

44

Fig. 3.12 Illustration of standard phosphoramidite DNA synthesis chemistry.
HO
O
T
DMT
O
T
Deblocking
O
O
P NCCH
2
CH
2
O
⊕
N(IpR)
2
G
ibu
DMT
H
O
O
NCCH
2
CH
2
O
N(IpR)
2
G
ibu
DMT
P
Activation
O
O
T
O
O
NCCH
2
CH
2
O
G
ibu
DMT
P
+
Coupling
O
O
T
H
3
C
O
Capping
O
O
T
O
O
P
NCCH
2
CH
2
O
G
ibu
DMT
O
+ Iodine/H
2
O/Pyridine
Oxidation
HO
O
T
HO
O
T
DMT
O
T
DMT
O
T
Deblocking
O
O
P NCCH
2
CH
2
O
⊕
N(IpR)
2
G
ibu
DMT
H
O
O
NCCH
2
CH
2
O
N(IpR)
2
G
ibu
DMT
P
Activation
O
O
T
O
O
NCCH
2
CH
2
O
G
ibu
DMT
P
+
Coupling
O
O
T
H
3
C
O
O
O
T
H
3
C
O
Capping
O
O
T
O
O
P
NCCH
2
CH
2
O
G
ibu
DMT
O
+ Iodine/H
2
O/Pyridine
Oxidation

45

The phosphoramidites used were dA-CE, dC-CE, dG-CE, and dT-CE
phosphoramidite (Glen Research, Sterling, VA). Since the phosphoramidites are
sensitive to moisture and oxygen, for preparation of the amidite solutions, dry
oxygen-free acetonitrile (“DNA synthesis grade”) was used, and the procedures
suggested by Glen Research were followed to minimize exposure of the acetonitrile
to air. 0.45 M 1-H-Tetrazole in anhydrous acetonitrile was used as the activator.
Phosphoramides and activator were premixed as 30% w/v solutions immediately
prior to synthesis. The ancillary reagents consisted of Cap A (tetrahydrofuran
(THF)/pyridine/acetic anhydride), Cap B (10% v/v solution of N-methylimidazole in
THF), oxidizer (0.1 M solution of iodine in THF/pyridine/water), and deblocker (3%
w/v solution of trichloroacetic acid in methylene chloride). Repeating the ejection
and solidification steps n times produces n-mer DNA probes.
Fig. 3.13 Photos showing how a glass slide is rotated between the synthesis
and wash areas for DNA synthesis. The left photo shows the motorized
rotation stage along with the wash and synthesis areas, while the right
photo shows how the glass slide is placed over the ejector array.
Directional ejector array
below glass slide
Wash area Synthesis area
Mounting station
with DIP package
Poly-l-lysine glass slide
Motorized
rotation stage
Directional ejector array
below glass slide
Wash area Synthesis area
Mounting station
with DIP package
Poly-l-lysine glass slide
Motorized
rotation stage

46
3.3.3 A 15-mer Oligonucleotide Synthesized by Four Directional Ejectors on a
Chip
For automated DNA synthesis, a computer program was used to control the
power delivery to each ejector and to decide the synthesis sequence. We synthesized
a 15-mer 5’-CGCCAAGCAGTTCGT-3’ sequence as shown in Fig 3.14(a), using a
pulsewidth of 30 µs and 10-Hz PRF for 3 seconds for each DNA base. The substrate
was placed at the height of 3 mm from the array, predetermined by the
characterization tests.
To test whether the DNA sequence was properly synthesized, hybridization
was performed using complementary DNA probes with FITC fluorescence tagging
(Alpha DNA, Quebec, Canada) and the SlideHyb kit from Ambion Inc., Austin, TX.
The hybridization procedures were performed as suggested by Ambion. SlideHyb
Buffer was preheated to 68°C for 15 minutes and then added to the labeled cDNA in
EDTA buffer solution. The glass slide was treated with the SlideHyb/cDNA mixture
and hybridized in a humidity chamber at 40°C for 16 hours. Before using
fluorescence imaging to determine the validity of the hybridization, we washed the
glass slide in low stringency wash buffer 2X SSC (0.9% sodium citrate, 1.8%
sodium chloride, pH 7.0) for 30 minutes and then in high stringency wash buffer
0.5X SSC (0.225% sodium citrate, 0.45% sodium chloride, pH 7.0) for 30 minutes.
Fluorescence microscopy was carried out with a Zeiss Axiovert 200
epifluorescence microscope at an excitation wavelength of 495 nm. Fluorescence
images were captured and analyzed with PicoStar HR12 camera (LaVision,

47
Goettingen, Germany). Fig. 3.14(b) shows the fluorescence image at the spot on
which the DNA bases were repeatedly ejected, while Fig. 3.14(c) shows the
fluorescence image at another spot (~3 mm away from the ejected droplets) on the
same glass which did not have the synthesized DNA sequence, and acted as a control
to verify that the fluorescence was not caused by dusts, particles or any
contamination. The observation of fluorescence only at the spot where the droplets
had been ejected confirms that we had the correct sequence synthesized.

DMT
C
P NCCH
2
CH
2
O
O
O
O
G
O
P
NCCH
2
CH
2
O
O
O
O
T
O
Poly-l-lysine glass
Poly-l-lysine
Synthesized DNA
sequence
Glass substrate
b)
c)
a)
O
DMT
C
P NCCH
2
CH
2
O
O
O
O
G
O
P
NCCH
2
CH
2
O
O
O
O
T
O
Poly-l-lysine glass
Poly-l-lysine
Synthesized DNA
sequence
Glass substrate
b)
c)
a)
O

Fig. 3.14 Synthesized 15-mer 5’-CGCCAAGCAGTTCGT-3’ DNA
sequence on a poly-l-lysine glass substrate. (a) Schematic illustration of the
oligonucleotide synthesized by phosphoramidite chemistry. (b) Micrograph
of the hybridized DNA sequence at the spot where the DNA bases were
ejected by the four ejectors. (c) Fluorescence image at another spot on the
same glass which did not have the synthesized DNA sequence.

48
There are several approaches to obtain better-defined and more uniform
fluorescence distribution. First, though short oligonucleotides should theoretically
provide better discrimination between related sequences, it has been experimentally
shown that oligonucleotides shorter than 25 bases often suffer from poor
hybridization specificity and that the use of spacer molecules with a length of at least
40 atoms is recommended [22]. We used standard cyanoethyl phosphoramidite
chemistry without adding spacers to move oligonucleotides away from the surface.
For enhanced hybridization performance, spacers can be added by modifying the
surface chemistry. Second, for better inkjet-based array fabrication efficiency, the
distance between the print-head and the array substrate should be as close as possible
to increase the droplet contact speed and to minimize the droplet deflection from the
side wind [23]. In the case of acoustic directional ejections, the distance is governed
by the ejector geometry and by the directional angle. To minimize the distance from
ejectors to the substrate, one can tilt the directional ejection angle further through
electrode design or use higher acoustic frequency to reduce the ejector size. Third,
because the PZT substrate we used had a 5% thickness variation, the resonant
frequency, which is proportional to the substrate thickness, varied among the ejectors.
As a result, when only one frequency was used to drive the four different ejectors,
satellite droplets were produced due to a higher power needed to make all the four
ejectors shoot out droplets. These satellite droplets of DNA bases were ejected along
with the main droplet onto the glass slide, resulting in poorly defined and
nonuniform fluorescence distribution. One can use multiple frequencies to match the

49
resonant frequencies for the different ejectors or use a more uniform piezoelectric
layer to resolve the satellites issue. Fourth, the repeatability decreased when the
synthesis process was carried out under the fume hood due to the fact that trajectory
of the ejected droplet was inevitably influenced by the high-speed air flow (120 fpm).
To ensure enough overlapping of DNA bases, we ejected 30 droplets of DNA bases
at each coupling step, broadening the spot area. An air shield can be used to
minimize the air-flow effect so that the number of the droplets and consequent spot
size can be reduced. Fifth, in a control test using water as the medium, the ejector
ejected a water droplet of 80 μm in diameter to form a round spot size of less than
150 μm in diameter on a glass slide. However, the spot shape has been reported to be
varying from round to ragged (or sometimes even vanishingly small) when surface
tension is low (as in the case of solvents used for phosphoramidite chemistry) [17].
Several schemes can be employed to help maintain spot integrity, such as ejecting
phosphoramidites and then tetrazole for activation or using a more hydrophobic
synthesis surface [10].
In conclusion, we present here an automated DNA microarray fabrication
technique by an array of acoustic directional ejectors. The ability of the ejectors to
directionally eject droplets of DNA bases removes the control circuitry and
mechanism that would otherwise have been needed to align a glass substrate with the
array. For proof of concept, a sequence of 5’-CGCCAAGCAGTTCGT-3’ has been
synthesized on a glass substrate by four directional ejectors. The synthesis
demonstrates the feasibility of a portable DNA synthesis system based on directional

50
ejector array and integrated microfluidic components. This portable DNA synthesis
system will provide significant benefit to geneticists by allowing them to sequence
DNAs at an affordable cost in their own labs.

3.4 Droplet-Based Microreactions with Oil Encapsulation
We describe here a microreaction technology for biochemical assay using
nanoliter droplets encapsulated inside oil droplets. Microreaction chambers are
constructed on a glass substrate by accumulating oil droplets that are dispensed by a
directional droplet ejector. Droplets of different aqueous reagents are then
directionally ejected (by other directional droplet ejector adjacent to the oil droplet
ejector) into the oil microchambers for parallel and combinatory analysis. Because
the reagents are encapsulated in oil, the evaporation rate is reduced by several orders
of magnitude, and only small amount of reagents are required for each assay. The
microchamber size and the reagent amount are digitally controlled by the number of
ejected oil and reagent droplets, respectively.

3.4.1 Review of Contemporary Microfluidic Networks
Microfluidic networks are useful for experiments that require minimal use of
reagents, and have been used in fields ranging from analytical chemistry to
biomedicine [24–26]. In principle, a microfluidic system can offer high throughput at
a reduced cost by densely integrating complex biochemical assays on a chip.
Although the first microfluidic chips were fabricated by etching a substrate such as

51
silicon [27] and glass [28] with conventional photolithography, more and more
devices are made by molding elastomers such as silicone [29] and
polydimethylsiloxane [30] (PDMS) with soft lithography due to their low production
cost.
In most of the reported microfluidic networks, liquids travel through sealed
interconnect channels, which are embedded into a solid matrix. The mass transfer of
liquid mediums in these systems is dominated by diffusion due to the small Reynolds
number [31]. The liquid flow can be manipulated by pressure difference in the
conduits. However, the passive pressure-driven networks typically have a broadened
velocity profile, and suffer from reagents diffusion through junctions and channels
over time [32]. Another drawback is that the pressure-driven flows do not scale well
with miniaturization and often require complicated off-chip plumbing [33].
Alternatively, electro-osmotic flow can be used for liquid transport with more
homogeneous velocity distribution and better scalability with miniaturization due to
its lack of any moving parts [34]. However, with the incorporation of the active
electrical or pneumatic pumping mechanisms into the microfluidic networks, the
interface complexity will also increase. Moreover, a practical issue for all these
systems with complex microchannel networks is the priming problem of filling the
device with fluids. In addition to the potential contamination of the samples, the
initial introduction of fluids can also generate undesirable air bubbles [35].
Liquids can also be confined and guided on open surfaces without a closed
conduit, and several approaches have been developed to pattern the wettable areas on

52
a surface, including electrowetting using an array of electrodes [36], wedge-wetting
by patterning the surface topology [37], and chemical wettability engineering [38].
The main disadvantage of these open microfluidic systems is the inevitable
contamination and drying of the liquid samples.
Another candidate for the miniaturization of microfluidic systems is inkjet
technology. The inkjet printing is able to deliver droplets on demand and can be
potentially applied to the nanoliter-scale assays [39]. However, it offers no control
over the fluid once the droplet has left the confinement of the ejection nozzle,
especially in regards to fast evaporation rate of nanoliter droplets [40]. Therefore, the
major challenge is to carry out the experiments utilizing small droplets and yet to
overcome the problematic liquid evaporation as well as contamination that may
happen without any closed channel networks. One prospective solution is to
encapsulate the aqueous droplets with oil by taking advantage of the immiscibility
between water and oil [41-42].
Here we present a new microfluidic platform for biochemical assays using
directional acoustic ejectors without complex channel networks. Acoustic actuation
is employed because droplets of both aqueous and non-aqueous fluids (even oil
droplets with a viscosity of 55 cSt) can be easily dispensed by an acoustic ejector
[43]. In addition to the ability of oil dispensation, the acoustic ejector is also capable
of ejecting droplets in any direction and hence suitable for parallel and combinatory
assays. In this new microfluidic platform, we first construct an array of oil
microreaction chambers on a glass substrate by acoustically ejecting and

53
accumulating oil droplets, and then eject droplets of different aqueous reagents into
the oil microchambers for various biochemical reactions.

3.4.2 Microreaction Platform Based on Directional Acoustic Ejectors
The schematic diagram of the proposed microreaction platform based on
directional acoustic ejectors is shown in Fig. 3.15(a). In general, several reagents are
needed for one biochemical assay, and the two-dimensional array of 2 × 2 ejectors
having sector electrodes are integrated on a single chip for efficient analysis. One of
the ejectors is used to eject oil droplets, and the other three ejectors are for different
targeted reagents. The microreactions using nanoliter droplets with oil encapsulation
are performed according to the following sequence (Fig. 3.15(b)). First, we
directionally eject and collect oil droplets on the glass substrate to form oil
microreaction chambers, where reaction will take place inside. Because the acoustic
ejector can eject one liquid droplet per one electrical pulse, the microchamber size
(V
MC
) can be digitally controlled by the number of ejected oil droplets (N) (i.e.,
D MC
V N V × = , where V
D
is the volume of one ejected droplet). After the oil
microchambers are constructed, droplets of reagents are then directionally ejected
into the microchambers for reaction. A combinatory assay not only entails utilization
of numerous reagents, it also involves different trials using various dosage (or
concentration) of individual reagent. The dispensation of reagents can be carried out
either sequentially or concurrently, and the reagent dosage is also digitally
controllable by the number of ejected reagent droplets.

54

Since only one reagent is to be run through a single ejector, which is integrated
with its own microchannel and reservoir, the individual ejectors dispense discrete
droplets without cross contamination. In addition, because the microreaction is
carried out inside the oil, the contamination and evaporation of the reagents are
greatly reduced and thus the assay volume is miniaturized. With each microchamber
representing one individual assay, this microreaction platform allows complex but
economical high-throughput combinatory analysis, easily through a computer
program.
Fig. 3.15 Schematic representation of droplet-based microreactions
with oil encapsulation. (a) Diagram of microreaction platform based on
four directional acoustic ejectors (two in the front and the other two in
the back) along with four inlet ports. (b) Sequence of microreactions
using nanoliter droplets – (1) Formation of oil microreaction chamber.
(2) Dispensation of reagent A. (3) Microreaction inside oil with
concurrent or sequential dispensing of reagents B and C from the two
ejectors in the back.

55
3.4.3 Experimental Steps for Microreactions with Oil Encapsulation
We describe here the experimental steps used for microreactions with oil
encapsulation and the measurement results of the reduction in water evaporation
with the new microreaction method.

3.4.3.1 Construction of Oil Microreaction Chambers
The first step is to construct the oil microchambers. We examined the ejection
of oil droplets by an LWAR ejector from the side-view micrographs. The oil used
was Inland-19 Ultra with a viscosity of 55 cSt. Figure 3.16(a) shows the continuous
ejections driven with ±60-V
peak-to-peak
pulses of 18-MHz sinusoidal signals without
any fine frequency tuning. The pulsewidth and the PRF were 100 μs and 60 Hz,
respectively. The ejection was one oil droplet per pulse and free of satellite droplets.
There is no measurable variation in droplet size for a given pulsewidth after
successive droplet ejections. The frame taken at the beginning of the ejections is
almost identical to the one after 30 seconds of ejections (1800 droplet shootings),
exhibiting the uniformity of droplet sizes (100 μm in diameter; 0.5 nL in volume)
and the stability of the ejection process.

56

To construct the oil microreaction chambers, we next tested the oil ejection
with a glass substrate placed 1 cm above the ejector and examined the ejection
process from the top view. In order to clearly see the oil droplets accumulate through
time, a low PRF was intentionally used for the testing. The oil droplets were
At beginning
(first shooting)
After 15 sec
(900 shootings)
After 30 sec
(1800 shootings)
300 μm
First droplet 10 droplets 5 droplets
Collected on glass
100 μm
1 oil droplet (0.5nL)
5 oil droplets (2.5nL)
10 oil droplets (5nL)
500 μm
Glass
(a)
(b)
(c)
At beginning
(first shooting)
After 15 sec
(900 shootings)
After 30 sec
(1800 shootings)
300 μm
First droplet 10 droplets 5 droplets
Collected on glass
100 μm
1 oil droplet (0.5nL)
5 oil droplets (2.5nL)
10 oil droplets (5nL)
500 μm
Glass
(a)
(b)
(c)
Fig. 3.16 Construction of oil microreaction chambers. (a) Stable and
continuous directional ejections of oil droplets without any frequency tuning.
(b) Successive oil droplets ejected onto the same position on glass and
accumulated into oil microchamber. (c) Array of oil microchambers
constructed on glass substrate.

57
directionally ejected and collected as the microchambers on the glass. The droplet
position was so consistent that sequential droplets hit at the same spot and the oil
microchamber grew through time, demonstrating the precise directionality for every
oil droplet ejection (Fig. 3.16(b)).
After the ejection characterization, oil droplets were then systematically
ejected to form microreaction chambers on a glass substrate. Because an oil
microchamber can hold limited amount of reagents before its breakdown, a larger
microchamber is generally needed for a greater amount of reagents. With this drop-
on-demand capability, the microchamber size was digitally controlled by the number
of ejected oil droplets, and an array of oil microchambers with different sizes was
formed (Fig. 3.16(c)).

3.4.3.2 Dispensation of Reagent Droplets into Oil Microchambers
After the construction of oil microchambers, the next step for microreaction is
to eject the reagent droplets into the oil microchambers. For easy visualization of the
dispensation process, the diluted red ink was first used as the liquid medium. We
formed the microchamber by accumulating 20 oil droplets on the glass, and then
directionally ejected the red ink droplets (80 μm in diameter) into the oil by another
ejector with ±60-V
peak-to-peak
pulses of 10- μs pulsewidth (Fig. 3.17(a)). The droplet
landing position could easily be controlled within 80 μm from ejection to ejection.
Since the first two successive droplets always overlap each other, the landing
position accuracy is conservatively estimated to be within one droplet size. The ink

58
droplet penetrated through the oil-air interface and was wrapped by a thin film of oil
(Fig. 3.17(b)). The droplet stopped as its kinetic energy was balanced by the viscous
loss from the drag force that resisted the motion. The ink droplet was placed very
close to the oil-air interface and isolated from air ambient by a thin oil film as
illustrated in Fig. 3.17(c). By flipping over the glass substrate and yet attaining the
same relative position of the encapsulated aqueous droplet inside oil, we confirmed
that the aqueous droplet’s proximity to the oil-air interface (rather than the oil-glass
interface) was not caused by the gravity effect.

Oil
500 μm
Glass
Oil
Red ink
100 μm
Oil thin film
Glass
Oil
Aqueous droplet
(a)
(b)
(c)
12 3
Oil Oil
500 μm
Glass
Oil
Red ink
100 μm 100 μm
Oil thin film
Glass
Oil
Aqueous droplet
Oil thin film
Glass
Oil
Aqueous droplet
(a)
(b)
(c)
12 3

Fig. 3.17 Dispensation of red ink droplet into the oil microreaction chamber. (a)
Sequential steps to form oil microreaction chamber and to dispense red ink
droplet – (1) Oil droplets ejected by directional ejector and collected on glass.
(2) Microreaction chamber constructed with 20 oil droplets. (3) Ejection of red
ink droplets. (b) Photo showing red ink droplet encapsulated inside oil. (c)
Schematic showing relative position of encapsulated aqueous droplet inside oil.

59
Figure 3.18 demonstrates the detailed encapsulation sequence from another
point of view. One water droplet was ejected into the oil microchamber. As the water
droplet first impacted the oil, it created a crater. The water crater was then sealed by
a thin film of oil radially from outside and completely encapsulated inside oil in less
than 0.7 seconds.

Also, we have proposed and performed another scheme for the droplet-based
microreactions with oil encapsulation. Figure 3.19 shows the schematic
representation of the alternative scheme. Instead of forming the oil microchamber
first by ejecting oil droplets onto the glass substrate, we dispensed the reagent
droplets directly through an oil film for oil encapsulation. We formed the oil film by
surface tension. A 160- μm-thick copper wire was shaped into a hydrophilic metal
frame, around which the oil film was suspended. The aqueous reagent droplets were
then ejected to penetrate through the oil film and then wrapped with oil after
penetration. However, there exist several drawbacks with this method. First, since
some portion of the oil is taken away with the flying reagent droplet for oil
encapsulation, a more complicated microfluidic system will be needed to
dynamically supply and reconstruct the oil film upon oil consumption. Second,
0 s 0.03 s 0.30 s 0.40 s 0.63 s
100 μm
Oil
Water
0 s 0.03 s 0.30 s 0.40 s 0.63 s
100 μm
Oil
Water
Fig. 3.18 Detailed sequence of encapsulation process. Water was used as the
liquid medium. One water droplet was ejected onto oil to create a water crater,
which was then sealed radially from outside by a thin film of oil.

60
during the penetration process some reagent residues are undesirably left in the oil,
resulting in cross-contamination issues for subsequent reactions. Third, since many
reagents are typically required for one assay, reagent droplets have to first
coalescence in air and then go through oil film for encapsulation. It becomes
extremely challenging especially when many droplets are needed for digital control
of the reagent amount. Thus, although this alternative method also can be employed
for oil encapsulation, it will not be too practical until the mentioned issues are
resolved.

Fig. 3.19 Alternative scheme for droplet-based microreactions with oil
encapsulation. (a) Schematic representation of the alternative scheme.
Aqueous droplets were dispensed to penetrate through an oil film for oil
encapsulation. (b) Experimental results showing a liquid droplet before and
after it passed through the oil film – (1) Water droplet ejected toward the oil
film constructed by a metal frame. (2) Water droplet encapsulated in oil after
penetration through the oil film.

61
3.4.3.3 Characterization of Water Evaporation Rate
Evaporation rates of droplets with and without oil encapsulation were
characterized. When one water droplet was ejected into the microchamber formed by
10 oil droplets, the evaporation rate with the oil encapsulation was so slow that it
took more than 10 hours for the droplet of only around 0.27 nL to evaporate (Fig.
3.20(a)). In a control test, a water droplet of the same volume was observed to spread
and dry out within 3 seconds after being ejected onto a glass surface without any oil
encapsulation. This kind of fast evaporation is the main challenge in carrying out any
biochemical experiment with a small reagent amount. Thus, with the oil
encapsulation, miniaturized assays using reagent volumes of only nanoliters are
achievable.
Though it was difficult to measure the thickness of the oil film between the
water and air either from the side-view (Fig. 3.17) or top-view micrograph (Fig.
3.18), it can be estimated by the water evaporation rate. According to the Fick’s law
of diffusion [44], the water encapsulated in oil evaporates at a rate given by
lRT
P P AD
t
Q
a sat
) ( −
=
∂
∂
………………………………………………………...…….(3.2)
where A is the surface area over which diffusion is taking place, D is the diffusion
coefficient of water in oil, P
sat
is the saturation water vapor pressure in the air, P
a
is
the actual water vapor pressure in the air, l is the oil film thickness, R is the universal
gas constant, and T is the air temperature. The measured decrease in evaporation rate
as a function of time is plotted in Fig. 3.20(b). The evaporation rate is observed to be
directly proportional to the droplet surface area, indicating that the oil thin film

62
thickness remains about the same throughout the 10 hours. With the experimental
evaporation results, the oil film was calculated to be around 0.5 μm. Though the oil
film was thin, it has been proven to be very efficient in reducing the evaporation rate.

After 2 hours After 4 hours
After 6 hours After 8 hours After 10 hours
At beginning
100 μm
Oil
Water
02 46 8 10 12
0.00
0.01
0.02
0.03
0.04
0.05

Evaporation Rate
Normalized Surface Area
Time (hour)
Evaporation Rate ( ΔnL/hour)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
Normalized Surface Area
(a)
(b)
After 2 hours After 4 hours
After 6 hours After 8 hours After 10 hours
At beginning
100 μm
Oil
Water
After 2 hours After 4 hours
After 6 hours After 8 hours After 10 hours
At beginning
100 μm
Oil
Water
02 46 8 10 12
0.00
0.01
0.02
0.03
0.04
0.05

Evaporation Rate
Normalized Surface Area
Time (hour)
Evaporation Rate ( ΔnL/hour)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
Normalized Surface Area
(a)
(b)

Fig. 3.20 Characterization of water evaporation rate: (a) Photos showing slow
evaporation of one water droplet (~0.27 nL) inside 10-oil-droplet
microchamber. It took more than 10 hours for the water droplet to dry out. (b)
Measured decrease in evaporation rate as a function of time. Also plotted is
the surface area, over which diffusion is taking place, as a function of time.

63
3.4.4 Microreaction Applications by the New Microreaction Platform
The new microreaction platform has been utilized for a series of microreaction
applications, including physical mixing reaction, chemical precipitation reaction, and
long-term iodine clock reaction.

3.4.4.1 Physical Mixing Reaction
First, we conducted a physical mixing experiment. We formed the
microchambers by ejecting and accumulating oil droplets on the glass slide, and then
dispensed diluted ink droplets into the oil. It has been demonstrated that the ink
droplets can be dispensed and encapsulated into the oil microchambers of different
sizes (Fig. 3.21(a)). In addition, various combinations of red and blue ink droplets
were ejected and mixed by diffusion inside the microchamber, and showed expected
color mixing even with the assay volume in the range of only nanoliters. Figure
3.21(b) shows the mixing results 5 seconds after ink droplets were ejected and
encapsulated inside the oil microchamber.

3.4.4.2 Chemical Precipitation Reaction
Not only the physical reaction can be carried out inside oil, the chemical
reaction can also be completed with this method. For visual demonstration, we have
used the microreaction platform to perform the precipitation reaction. As silver
nitrate reacts with sodium carbonate, they exchange partners to form silver carbonate,
which cannot be dissolved in water [45].

64
te) (precipita CO Ag 2NaNO CO Na 2AgNO
3 2 3 3 2 3
+ ⎯→ ⎯ + …………...………...(3.3)
We prepared two colorless solutions – 0.1 M silver nitrate and 0.1 M sodium
carbonate. After we ejected and collected 10 oil droplets to form the microreaction
chamber, one droplet (~0.3 nL) of the silver nitrate solution was then dispensed into
the oil (Fig. 3.22(a)). After intentionally waiting one hour to confirm negligible
water evaporation (Fig. 3.22(b)), we then ejected one droplet (~0.3 nL) of the
sodium carbonate solution into the microchamber for reaction (Fig. 3.22(c-d)). The
silver carbonate precipitate was attained in about one second, consuming only one
droplet of silver nitrate and one droplet of sodium carbonate (Fig. 3.22(e)).
The physical mixing and chemical precipitation reactions are the assays that
happen instantaneously or in seconds. Since the evaporation rate with oil
encapsulation is so slow that this microreaction platform can even be applied for the
experiments that require longer time to be completed, as shown in the next paragraph.

2 red + 1 blue 1 red + 2 blue 3 red + 5 blue
1 blue
100 μm
1 red in 1 oil 1 red in 5 oil 1 red in 10 oil 5 red in 10 oil
100 μm
Oil
Ink
(a)
(b)
2 red + 1 blue 1 red + 2 blue 3 red + 5 blue
1 blue
100 μm
1 red in 1 oil 1 red in 5 oil 1 red in 10 oil 5 red in 10 oil
100 μm
Oil
Ink
(a)
(b)

Fig. 3.21 Physical mixing microreaction using nanoliter droplets with oil
encapsulation. (a) Micrographs showing red ink droplets ejected and
encapsulated inside oil microchambers with different sizes. (b)
Micrographs showing various combinations of red and blue ink droplets
encapsulated and mixed inside oil microchamber.

65

3.4.4.3 Long-Term Iodine Clock Reaction
We have further demonstrated the use of the method for long-time reactions by
conducting the iodine clock reaction [46], in which two colorless solutions are mixed,
and at first there is no visible reaction, but the mixed solution suddenly turns blue
after some time delay. There are three reactions occurring in the mixed solution. In
the first reaction, iodide ions are oxidized by hydrogen peroxide to form triiodide
ions in an acid solution.
O 2H (aq) I (aq) 2H (aq) 3I O H
2
-
3
-
2 2
+ ⎯→ ⎯ + +
+
…………………………...…….(3.4)
In the second reaction, the triiodide ions are reconverted back to iodide ions by
thiosulfate.
(aq) O S (aq) 3I (aq) O 2S (aq) I
2
6 4
- 2
3 2
-
3
− −
+ ⎯→ ⎯ + ………………………………...(3.5)
Since the second reaction is much faster than the first reaction, the triiodide ions are
consumed right after they are formed, preventing any observable response in the
following third reaction. After all the thiosulfate ions are used up, the triiodide ions
Fig. 3.22 Chemical precipitation microreaction using nanoliter droplets with
oil encapsulation. (a) One droplet of 0.1 M silver nitrate solution ejected and
encapsulated inside oil. (b) No observable evaporation after one hour. (c)
Dispensation of one droplet of 0.1 M sodium carbonate solution into oil. (d)
Microreaction of silver nitrate and sodium carbonate with oil encapsulation.
(e) Formation of silver carbonate precipitate in 1 sec.
Oil
AgNO
3
Ag
2
CO
3
0 s 0.03 s 0.27 s 1.0 s
100 μm
Na
2
CO
3
(a) (b) (c) (d) (e)
Oil
AgNO
3
Ag
2
CO
3
0 s 0.03 s 0.27 s 1.0 s
100 μm
Na
2
CO
3
(a) (b) (c) (d) (e)

66
react with starch, and produce blue starch-pentaiodide complex, at which point blue
color appears.
(aq) I complex I - starch starch (aq) 2I
-
5
-
3
−
+ ⎯→ ⎯ + …………………………...….(3.6)
We prepared two colorless solutions, Solution A (0.3% w/v starch, 0.003 M
sodium thiosulfate, and 1.6% w/v potassium iodide in the acetate buffer, pH 4.5) and
Solution B (0.1% hydrogen peroxide). For demonstration, both the microreaction
using the new platform and a conventional macroreaction using test tubes and
pipettes have been carried out.
For the macroreaction, one milliliter of solution A and one milliliter of
solution B were separately pipetted into the test tube for reaction. There was no
visible color change in the beginning (Fig. 3.23(a)), and the blue color was observed
after 126 seconds.

Fig. 3.23 Long-term iodine clock reaction. (a) Macroreaction: Pipetting 1 mL of
solution A and 1 mL of solution B into test-tube for reaction. Blue color
appeared after 126 sec. (b) Microreaction: Ejecting one droplet (0.3 nL) of
solution A and one droplet (0.3 nL) of solution B into oil microchamber for
reaction. Blue color appeared after 110 sec.
Mixture of Solution A and B (1 mL of A and 1 mL of B)
After 30 sec After 60 sec After 90 sec After 126 sec
1 cm
Test tube
100 μm
Oil
Mixture of Solution A and B (1 droplet of A and 1 droplet of B)
After 110 sec After 30 sec After 60 sec After 90 sec
(a)
(b)
Mixture of Solution A and B (1 mL of A and 1 mL of B)
After 30 sec After 60 sec After 90 sec After 126 sec After 30 sec After 60 sec After 90 sec After 126 sec
1 cm
Test tube
100 μm
Oil
Mixture of Solution A and B (1 droplet of A and 1 droplet of B)
After 110 sec After 30 sec After 60 sec After 90 sec
(a)
(b)

67
For the microreaction, one droplet (0.3 nL) of Solution A and one droplet (0.3
nL) of Solution B were ejected into the oil microchamber for reaction. Compared
with the macroreaction, this was 3 million times reduction in the reagent volume.
Similarly, there was no observable response in the beginning (Fig. 3.23(b)). The blue
color appeared after 110 seconds, very close to the response time of the
macroreaction with 13% deviation. The difference was possibly caused by droplet
evaporation during its flight in air that would have made the solutions for the
microreaction be a little more concentrated than those for the macroreaction. To
minimize the deviation, the distance between the ejector and the assay substrate
should be kept as close as possible to reduce the droplet evaporation. Since the
distance is governed by the ejector geometry and also by the directional angle, one
can use higher acoustic frequency to reduce the ejector size and/or increase the
ejection angle further through electrode design.
In conclusion, a new biochemical microreaction platform based on directional
acoustic ejectors has been demonstrated. Nanoliter droplets of reagents are utilized
for miniaturized assays with oil encapsulation. This microreaction technology offers
the following advantages for biochemical analysis. First, the acoustic ejectors are
capable of ejecting both aqueous and non-aqueous fluids. Thus ejectors for oil and
reagents can be integrated on a single chip to target one spot on the assay substrate
so that each assay is performed efficiently without any mechanical movement and
alignment. Second, by dispensing the oil and reagents droplet by droplet, we can
digitally control the oil microchamber size and the reagent amount easily through a

68
computer program. Third, with the oil encapsulation, the reagent evaporation rate is
greatly reduced and the required reagent amount can be miniaturized.

3.5 Harmonic Operation for Ejection of Micron-Sized Droplets
This section describes the use of harmonic frequencies for acoustic ejection in
order to eject micron-sized droplets. For an acoustic ejector working at the thickness-
mode resonance, the droplet size is determined by the acoustic wavelength. In the
newly invented design, we do not lap the bulk piezoelectric PZT substrate or use
PZT thin films, but use harmonic frequencies of the bulk form to reduce the
wavelength. The acoustic ejectors utilizing harmonic frequencies have been
fabricated and shown to be very effective up to the 9th harmonic (~180 MHz),
ejecting droplets as small as 10 μm in diameter.

3.5.1 Introduction
For an efficient acoustic ejection, the transducer is typically operated at the
fundamental thickness-mode resonant frequency. Consequently, the ejected droplet
size, primarily determined by the diameter of the focused acoustic beam, is
proportional to the piezoelectric substrate thickness and inversely proportional to the
frequency. Since the ejection of smaller droplets is the key for both better printing
resolution and reduced consumption of dispensed reagents, higher-frequency
operation is desirable for droplet ejections. Additionally, the size of the lens (or

69
electrodes for SFAT) is directly related to the working frequency, and thus higher-
frequency operation is needed when the transducer size needs to be reduced.
For high frequency operation (over 20 MHz), the PZT transducer needs to have
a thickness less than 100 μm. Lapping of the bulk ceramic is a solution to reduce the
thickness. However, this induces a risk of fracture as well as the difficulty for
handling. To avoid these issues, using thin or thick films is an alternative solution,
and several fabrication processes have been developed for high frequency transducer
applications. However, while sputtering and chemical vapor deposition of PZT film
require very tight process control for repeatable quality [47], sol-gel PZT film
typically has large residual stress [48]. In addition to these processing difficulties,
the electromechanical performance obtained with thin PZT film is generally lower
than the same compositions in the bulk form [49]. Moreover, for the air-backed
transducer working in a liquid environment, the continuous large vibration at
resonance can result in membrane breakage, especially for membrane size larger
than 500 μm by 500 μm. This fragility causes a potential reliability issue for thin-
film acoustic ejectors employing piezoelectric films with large residual stress.

3.5.2 Design and Fabrication
In the new design, we do not lap a bulk PZT substrate or use PZT thin films,
but use harmonic frequencies of the bulk substrate.
To effectively generate acoustic waves, the piezoelectric film thickness is
typically chosen to be one half of the acoustic wavelength for an air-backed

70
transducer working at the thickness-mode resonance. For a PZT sheet with a
thickness of 127 μm, the fundamental frequency is around 20 MHz. At this
frequency, the wavelength in water is around 80 μm, which limits the smallest
droplet producible to be about 80 μm in diameter.
In order to work at a higher frequency and still produce large acoustic power,
the transducer is excited with Q-enhancement from resonance at the nth harmonic
frequencies,
l fundamenta l fundamenta nth nth
nf nv v f = = = λ λ / / , where v and λ are the acoustic
velocity and wavelength, respectively. For a symmetric transducer having electrodes
sandwiching a piezoelectric sheet, there exist only odd harmonics, i.e., n=1,3,5…
With the fundamental frequency of 20 MHz, the 3rd, 5th, 7th, and 9th harmonics are
60 MHz, 100 MHz, 140 MHz, and 180 MHz, respectively. Thus, through harmonic
operations, a droplet of 10 μm in diameter can be theoretically obtained at the 9th
harmonic frequency with a 127- μm-thick bulk PZT substrate.
For the nth harmonic operation, the lens is designed with the kth radius of the
Fresnel half-wave band given by ) 4 / (
, nth nth nth k
k F k r λ λ + = . The acoustic ejectors with
the lenses designed for different harmonic operations are fabricated with a similar
process flow for fundamental LWAR ejectors. It is noted that all ejectors were built
on the same 127- μm-thick PSI-5A4E PZT sheet, but had different lens patterns.
Figure 3.24 shows the optical photos of the fabricated lenses.

71

3.5.3 Results and Discussion
The fabricated ejector was driven with pulses of sinusoidal signals. Figure 3.25
shows the continuous ejection with ±70-V
peak-to-peak
pulses of 58-MHz sinusoidal
signals based on the 3rd harmonic operation. The pulsewidth is 8 μs and ejection rate
is 120 Hz. Frames at different time are almost identical, exhibiting the uniformity of
droplet sizes (26 μm in diameter) and the stability of ejection process. It took about
35 μs for the droplet to be formed and separated from the bulk liquid.

Figure 3.26 shows the ejections at higher ejection rates. The ejection is one
droplet per pulse and free of satellite droplets. The images captured by strobing are a
superposition of many successive droplets, and the image sharpness demonstrates
consistent droplet ejections at the ejection rate up to 8 kHz.
Fig. 3.24 Photos of fabricated acoustic lenses for harmonic operations.
300 μm 300 μm
300 μm 300 μm
3rd harmonic 5th harmonic 7th harmonic 9th harmonic
300 μm 300 μm
300 μm 300 μm
3rd harmonic 5th harmonic 7th harmonic 9th harmonic
Fig. 3.25 Photos of stable ejections for the 3rd harmonic.
At beginning
100 μm
After 30 sec
(3600 shootings)
After 60 sec
(7200 shootings)
At beginning
100 μm
After 30 sec
(3600 shootings)
After 60 sec
(7200 shootings)

72

The harmonic utilization for acoustic transducers has been observed to be
applicable for ejection up to the 9th harmonic and eject uniform droplets down to 10
μm in diameter (Fig. 3.27).

Ejections for different harmonics have been thoroughly characterized and
compared. Figure 3.28 shows the stable ejections at the ejection rate of 3 kHz for
different harmonics, and Fig. 3.29 shows the eminent droplet size reduction through
harmonic operations.

Fig. 3.26 Ejections for the 3rd harmonic at high ejection rates.
1 kHz 3 kHz 4 kHz 8 kHz
500 μm
1 kHz 3 kHz 4 kHz 8 kHz
500 μm

Fig. 3.27 Photos of stable ejections of 10- μm-diameter
droplets with the 9th harmonic operation.
At beginning
100 μm
After 10 sec
(1200 shootings)
After 20 sec
(2400 shootings)
At beginning
100 μm
After 10 sec
(1200 shootings)
After 20 sec
(2400 shootings)

73

The quality factor obtained from the measured impedance of the ejector shows
the decreasing trend as the harmonic frequency increases (Fig. 3.30). Though the
threshold energy required for ejection decreases as frequency increases (
3 . 2 −
∝ f E
th

[50]), the acoustic attenuation coefficient increases with frequency (
2
f ∝ α [51]).
Thus, the Q-enhancement from resonance plays an important role in producing
enough power for ejection. We have fabricated the ejectors to work at the 11th and
3rd Harmonic
300 μm 300 μm 300 μm
5th Harmonic 7th Harmonic 3rd Harmonic
300 μm 300 μm 300 μm
5th Harmonic 7th Harmonic

Fig. 3.28 Ejections at ejection rate of 3 kHz for different harmonics.
100 μm 100 μm
3rd harmonic
100 μm 100 μm
Fundamental
100 μm
5th harmonic
7th harmonic 9th harmonic
100 μm 100 μm
3rd harmonic
100 μm 100 μm
Fundamental
100 μm
5th harmonic
7th harmonic 9th harmonic

Fig. 3.29 Droplet size reduction by harmonic operations.

74
13th harmonics, but no ejection has been observed mainly due to their low Q values.
Since the PZT substrate we used has a surface roughness around 8 μm ( ±4 μm),
which is comparable with the wavelength in water at frequencies higher than 100
MHz, more energy is lost at the substrate surface for higher harmonics, resulting in
dramatic Q drops.

The measured droplet size is plotted as a function of the harmonic frequencies
in Fig. 3.31. The theoretical wavelength in water is also plotted to demonstrate the
droplet size’s direct dependence on the wavelength. The measured droplet separation
time (Fig. 3.32) is shorter for a smaller droplet size, and thus higher ejection rates are
possible with smaller droplets.

Fig. 3.30 Measured quality factor for harmonics.

75

In conclusion, the use of harmonic frequencies to reduce the wavelength has
been successfully demonstrated for acoustic droplet ejection. The acoustic transducer
based on this idea is free of residual stress, robust, reliable, but yet efficient in
0 10203040 50607080 90
0
20
40
60
80
100
7th harmonic
Fundamental
5th harmonic
9th harmonic
3rd harmonic

Time ( μs)
Droplet size ( μm)
Exp. droplet separation time
Fit linear[1.25(s/m) x droplet size]

Fig. 3.32 Droplet separation time versus droplet size.
0 50 100 150 200
0
20
40
60
80
100
9th harmonic
7th harmonic
5th harmonic
3rd harmonic
fundamental

( μm)
Frequency (MHz)
Theoretical wavelength in water
Experimental droplet diameter

Fig. 3.31 Droplet size as a function of frequency.

76
producing large acoustic power to eject droplets down to 10 μm in diameter at the
9th harmonic (180 MHz). Consistent droplet ejections for different harmonic
operations have been achieved at an ejection rate up to 8 kHz. The harmonic
operations are excellent in reducing not only the droplet size but also the transducer
size. The later has significant implication in microfluidic systems employing array of
acoustic transducers where the size of an individual transducer matters.

Summary
Ejections by the LWAR ejectors have been demonstrated to be stable and
repeatable. With automatic liquid delivery, continuous ejections have been attained
without any frequency tuning. This efficient droplet-on-demand acoustic ejector has
been successfully utilized for DNA synthesis, micromixing, and microreaction.
Further, harmonic frequencies have been used to produce droplets as small as 10 μm
in diameter.

77
Chapter 3 References
[1] M. Orme, “Experiments on droplet collisions, bounce, coalescence and
disruption,” Prog. Energy Combust. Sci., vol. 23, pp. 65-79, 1997.
[2] T. B. Low and R. List, “Collision, coalescence and breakup of raindrops. Part I:
Experimentally established coalescence efficiencies and fragment size
distributions in breakup,” J. Atmos. Sci., vol. 39, pp. 1591-1606, 1982.
[3] J. R. Adam, N. R. Lindelad, and C. D. Hendricks, “The collision, coalescence,
and disruption of water droplets,” J. Appl. Phys., vol. 39, pp. 5173-5180, 1968.
[4] U. Demirci and M. Toner, “Direct etch method for microfluidic channel and
nanoheight post-fabrication by picoliter droplets,” Appl. Phys. Lett., vol. 88, pp.
053117-1-3, 2006.
[5] F. Mashayek, N. Ashgriz, W. J. Minkowycz, and B. Shotorban, “Coalescence
collision of liquid drops,” Int. J. Heat. Mass. Tran., vol. 46, pp. 77-89, 2003.
[6] L.-H. Lu, K. S. Ryu, and C. Liu, “A magnetic microstirrer and array for
microfluidic mixing,” J. Microelectromech. Syst., vol. 11, pp. 462-469, 2002.
[7] G. Nonglaton, I. O. Benitez, I. Guisle, M. Pipelier, J. Leger, D. Dubreuil, C.
Tellier, D. R. Talham, and B. Bujoli, “New approach to oligonucleotide
microarrays using zirconium phosphonate-modified surfaces,” J. Am. Chem.
Soc., vol. 126, pp. 1497–1502, 2004.
[8] G. H. McGall, A. D. Barone, M. Diggelmann, S. P. A. Fodor, E. Gentalen, and
N. Ngo, “The efficiency of light-directed synthesis of DNA arrays on glass
substrates,” J. Am. Chem. Soc., vol. 119, pp. 5081–5090, 1997.
[9] D. Gershon, “Microarray technology: an array of opportunities,” Nature, vol.
416, pp. 885–891, 2002.
[10] T. R. Hughes, M. Mao, A. R. Jones, J. Burchard, M. J. Marton, K. W. Shannon,
S. M. Lefkowitz, M. Ziman, J. M. Schelter, M. R. Meyer, S. Kobayashi, C.
Davis, H. Dai, Y. D. He, S. B. Stephaniants, G. Cavet, W. L. Walker, A. West,
E. Coffey, D. D. Shoemaker, R. Stoughton, A. P. Blanchard, S. H. Friend, and P.
S. Linsley, “Expression profiling using microarrays fabricated by an ink-jet
oligonucleotide synthesizer,” Nature Biotechnol., vol. 19, pp. 342–347, 2001.
[11] C. Tilstone, “DNA microarrays: vital statistics,” Nature, vol. 424, pp. 610–612,
2003.

78
[12] S. Singh-Gasson, R. D. Green, Y. Yue, C. Nelson, F. Blattner, M. R. Sussman,
and F. Cerrina, “Maskless fabrication of light-directed oligonucleotide
microarrays using a digital micromirror array,” Nature Biotechnol., vol. 17, pp.
74–978, 1999.
[13] X. Gao, E. Gulari, and X. Zhou, “In situ synthesis of oligonucleotide
microarrays,” Biopolymers, vol. 73, pp. 579–596, 2004.
[14] S. V. Chittur, “DNA microarrays: tools for the 21st century,” Comb. Chem.
High T. Scr., vol. 7, pp. 531–537, 2004.
[15] E. E. Schadt, S. A. Monks, T. A. Drake, A. J. Lusis, N. Che, V. Colinayo, T. G.
Ruff, S. B. Milligan, J. R. Lamb, G. Cavet, P. S. Linsley, M. Mao, R. B,
Stoughton, and S. H. Friend, “Genetics of gene expression surveyed in maize,
mouse and man,” Nature, vol. 422, pp. 297–302, 2003.
[16] G. G. Rozenberg, E. Bresler, S. P. Speakman, C. Jeynes, and J. H. G. Steinke,
“Patterned low temperature copper-rich deposits using inkjet printing,” Appl.
Phys. Lett., vol. 81, pp. 5249–5251, 2002.
[17] C. Lausted, T. Dahl, C. Warren, K. King, K. Smith, M. Johnson, R. Saleem, J.
Aitchison, L. Hood, and S. R. Lasky, “POSaM: a fast, flexible, open-source,
inkjet oligonucleotide synthesizer and microarrayer,” Genome Biology, vol. 5,
R58, 2004.
[18] J. H. Butler, M. Cronin, K. M. Anderson, G. M. Biddison, F. Chatelain, M.
Cummer, D. J. Davi, L. Fisher, A. W. Frauendorf, F. W. Frueh, C. Gjerstad, T.
F. Harper, S. D. Kernahan, D. Q. Long, M. Pho, J. A. Walker, and T. M.
Brennan, “In situ synthesis of oligonucleotide arrays by using surface tension,”
J. Am. Chem. Soc., vol. 123, pp. 8887–8894, 2001.
[19] E. S. Y. Li and W.-T. Liu, “DNA microarray technology in microbial ecology
studies–principle, applications and current limitations,” Microbes Environ., vol.
18, pp. 175–187. 2003.
[20] P. Angenendt, J. Glokler, D. Murphy, H. Lehrach, and D. J. Cahill, “Toward
optimized antibody microarrays: a comparison of current microarray support
materials” Anal. Biochem., vol. 309, pp. 253–260, 2002.
[21] S. A. Scaringe, C. Francklyn, and N. Usman, “Chemical synthesis of
biologically active oligoribonucleotides using β-cyanoethyl protected
ribonucleoside phosphoramidites,” Nucleic Acids Res., vol. 18, pp. 5433–5441,
1990.

79
[22] M. S. Shchepinov, S. C. Case-Green, and E. M. Southern, “Steric factors
influencing hybridization of nucleic acids to oligonucleotide arrays,” Nucleic
Acids Res., vol. 25, pp. 1155–1161, 1997.
[23] B.-J. de Gans and U. S. Schubert, “Inkjet printing of polymer micro-arrays and
libraries: instrumentation, requirements, and perspectives,” Macromol. Rapid
Commun., vol. 24, pp. 659–666, 2003.
[24] A. Khademhosseini, R. Langer, J. Borenstein, and J. P. Vacanti, “Microscale
technologies for tissue engineering and biology,” Proc. Natl. Acad. Sci., vol.
103, no. 8, pp. 2480–2487, 2006.
[25] M. Joanicot and A. Ajdari, “Droplet control for microfluidics,” Science, vol.
309, pp. 887–888, 2005.
[26] J. Sager, M. Young, and D. Stefanovic, “Characterization of transverse channel
concentration profiles obtainable with a class of microfluidic networks,”
Langmuir, vol. 22, pp. 4452–4455, 2006.
[27] J. B. Knight, A. Vishwanath, J. P. Brody, and R. H. Austin, “Hydrodynamic
focusing on a silicon chip: mixing nanoliters in microseconds,” Phys. Rev. Lett.,
vol. 80, no. 17, pp. 3863–3866, 1998.
[28] J. Gao, X.-F. Yin, and Z.-L. Fang, “Integration of single cell injection, cell lysis,
separation and detection of intracellular constituents on a microfluidic chip,”
Lab Chip, vol. 4, pp. 47–52, 2004.
[29] M. A. Unger, H.-P. Chou, T. Thorsen, A. Scherer, and S. R. Quake,
“Monolithic microfabricated valves and pumps by multilayer soft lithography,”
Science, vol. 288, pp. 113–116, 2000.
[30] F. Lin and E. C. Butcher, “T cell chemotaxis in a simple microfluidic device,”
Lab Chip, vol. 6, pp. 1462–1469, 2006.
[31] P. J. A. Kenis, R. F. Ismagilov, and G. M. Whitesides, “Microfabrication inside
capillaries using multiphase laminar flow pattering,” Science, vol. 285, pp 83–
85, 1999.
[32] T. Pfohl, F. Mugele, R. Seemann, and S. Herminghaus, “Trends in microfluidics
with complex fluids,” Chemphyschem, vol. 4, pp. 1291–1298, 2003.
[33] M. Z. Bazant and Y. Ben, “Theoretical prediction of fast 3D AC electro-
osmotic pumps,” Lap Chip, vol. 6, pp. 1455–1461, 2006.

80
[34] T. M. Squires and M. Z. Bazant, “Induced-charge electro-osmosis,” J. Fluid
Mech., vol. 509, pp. 217–252, 2004.
[35] C. L. Hansen, E. Skordalakes, J. M. Berger, and S. R. Quake, “A robust and
scalable microfluidic metering method that allows protein crystal growth by
free interface diffusion,” Proc. Natl. Acad. Sci., vol. 99, no. 26, pp. 16531–
16536, 2002.
[36] O. D. Velev and K. H. Bhatt, “On-chip micromanipulation and assembly of
colloidal particles by electric fields,” Soft Matter, vol. 2, pp. 738–750, 2006.
[37] R. Seemann, E. J. Kramer, and F. F. Lange, “Patterning of polymers: precise
channel stamping by optimizing wetting properties,” New J. Phys., vol. 6, 111,
2004.
[38] H. Gau, S. Herminghaus, P. Lenz, and R. Lipowsky, “Liquid morphologies on
structured surfaces: from microchannels to microchips,” Science, vol. 283, pp.
46–49, 1999.
[39] S. Schmatloch, M. A. R. Meier, and U. S. Schubert, “Instrumentation for
combinatorial and high-throughput polymer research: a short overview,”
Macromol. Rapid Commun., vol. 24, pp. 33–46, 2003.
[40] E. Delamarche, D. Juncker, and H. Schmid, “Microfluidics for processing
surfaces and miniaturizing biological assays,” Adv. Mater., vol. 17, pp. 2911–
2933, 2005.
[41] N. E. Chayen, “The role of oil in macromolecular crystallization,” Structure,
vol. 5, no. 10, pp. 1269–1274, 1997.
[42] B. Zheng, L. S. Roach, and R. F. Ismagilov, “Screening of protein
crystallization conditions on a microfluidic chip using nanoliter-size droplets,”
J. Am. Chem. Soc., vol. 125, pp. 11170–11171, 2003.
[43] C.-Y. Lee, H. Yu, and E. S. Kim, “Microreactions using nanoliter droplets with
oil encapsulation”, IEEE International Micro Electro Mechanical Systems
Conference, Kobe, Japan, January 21–25, 2007, pp. 81–84.
[44] H. Zhang and W. Davison, “Performance characteristics of diffusion gradients
in thin films for the in situ measurement of trace metals in aqueous solution,”
Anal. Chem., vol. 67, pp. 3391–3400, 1995.
[45] J. M. Lopez-Cepero, “Silver carbonate staining reveals mitochondrial
heterogeneity,” J. Histochems. Cytochem., vol. 52, pp. 211–216, 2004.

81
[46] C. H. Sorum, F. S. Charlton, J. A. Neptune, and J. O. Edwards, “pH change as
an index to reaction mechanisms,” J. Am. Chem. Soc., vol. 74, no. 1, pp. 219–
221, 1952.
[47] P. Muralt, “Ferroelectric thin films for micro-sensors and actuators: a review,” J.
Micromech. Microeng., vol. 10, pp.136-146, 2000.
[48] L. Lian and N. R. Sottos, “Stress effects in sol-gel derived ferroelectric thin
films,” J. Appl. Phys., vol. 95, pp. 629-634, 2004.
[49] F. Levassort, L. P. Tran Huu Hue, J. Holc, T. Bove, M. Kosec, and M. Lethiecq,
“High Performance piezoceramic films on substrates for high frequency
imaging,” in Proc. IEEE Ultrasonics Symposium, Atlanta, GA, Oct. 7-10, 2001,
pp. 1035-1038.
[50] S. A. Elrod, B. Hadimioglu, B. T. Khuri-Yakub, E. G. Rawson, E. Richley, C. F.
Quate, N. N. Mansour, and T. S. Lundgren, “Nozzleless droplet formation with
focused acoustic beams”, J. Appl. Phys., vol. 65, pp. 3441-3447, 1989.
[51] D. Huang and E. S. Kim, “Micromachined acoustic-wave liquid ejector,” J.
Micro-electromech. Syst., vol. 10, pp. 442-449, 2001.

82
Chapter 4
Piezoelectrically Actuated Tunable Capacitor

This chapter describes the design, fabrication and characterization of
piezoelectrically actuated tunable capacitors. Zinc oxide (ZnO) is employed as the
active material for piezoelectric actuation. Three different types of ZnO-actuated
tunable capacitors with surface and bulk micromachining techniques are presented.

4.1 Design and Simulation
Three different tunable capacitor designs discussed in this work are described
as follows.

4.1.1 Cantilever Structure
The first is a simple cantilever structure, in which the cantilever beam is
released with surface micromaching technique. As illustrated in the schematic
drawing of Fig. 4.1, the parallel-plate tunable capacitor has a variable air gap formed
by a moving top capacitor electrode and a fixed bottom capacitor electrode
integrated on the same silicon substrate. Though the top capacitor electrode is
mechanically connected and driven by a piezoelectric unimorph actuator (100 μm
long) composed of parylene-D/Al/PECVD-SiN/ZnO/Al layers, the capacitor is
electrically separated from the actuator. Parylene-D is chosen as the supporting
material because of its low Young’s modulus and potential for a large deflection. A

83
voltage applied between the two Al electrodes produces lateral contraction (or
expansion) in the ZnO film, giving rise to a bending moment in the beam cross-
section. As the beam moves downward upon the contraction of the ZnO film, the air
gap of the capacitor is reduced, and the capacitance increases consequently.
Oppositely, an upward movement upon ZnO film expansion can lead to a
capacitance decrease.

4.1.2 Bulk-Micromachined Tunable Capacitor with Mass Structure
In the second design, a novel bulk-micromachined mass structure driven by a
film unimorph is employed (Fig. 4.2). The bulk-micromachined tunable capacitor is
consisted of an air gap formed by a thick silicon mass structure (1.7×3×0.4 mm
3
)
Fig. 4.1 Schematic representation of the cantilever-type surface-
micromachined piezoelectric tunable capacitor.
x-axis
Bottom Al (1
st
)
PECVD SiN (3
rd
)
Top Al (4
th
)
ZnO (2
nd
)
Parylene-D (5
th
)
t
5
t
4
t
3
t
2
t
1
z
5
z
1
z
4
z
2
z
3
z-axis
Actuator
Capacitor
Top actuator
electrode
Bottom actuator electrode Bottom capacitor electrode
Top capacitor electrode
ZnO film
Anchor
Silicon
Parylene supporting layer
x-axis
Bottom Al (1
st
)
PECVD SiN (3
rd
)
Top Al (4
th
)
ZnO (2
nd
)
Parylene-D (5
th
)
t
5
t
4
t
3
t
2
t
1
z
5
z
1
z
4
z
2
z
3
z-axis
Actuator
Capacitor
Top actuator
electrode
Bottom actuator electrode Bottom capacitor electrode
Top capacitor electrode
ZnO film
Anchor
Silicon
Parylene supporting layer

84
with the movable top capacitor electrode and the fixed bottom electrodes on a glass
substrate as shown in Fig. 4.2. The large mass structure is driven by a rectangular
and thin beam (2.1×3×0.025 mm
3
) composed of Al/PECVD-SiN/ZnO/Al/Si/Al
layers. With a similar working principle, when an electric field is applied across the
ZnO film, the unimorph bends, making the mass structure to move vertically, and the
capacitance is accordingly tuned.

4.1.3 Bridge-Type Surface-Micromachined Tunable Capacitor
In the third design (Fig. 4.3), the parallel-plate tunable capacitor has a variable
air gap formed by the movable top capacitor electrode on a simply-supported bridge
structure and the fixed bottom capacitor electrode integrated on the same silicon
substrate. The bridge is completely released and initially floats on two piezoelectric
Fig. 4.2 Schematic representation of the bulk-
micromachined tunable capacitor with mass structure.
Top actuator electrode
Capacitor
Epoxy
Bottom actuator electrode
Bottom capacitor
electrode
Top capacitor electrode
ZnO film
Anchor
Silicon
Glass substrate
Actuator beam
Mass structure
Top actuator electrode
Capacitor
Epoxy
Bottom actuator electrode
Bottom capacitor
electrode
Top capacitor electrode
ZnO film
Anchor
Silicon
Glass substrate
Actuator beam
Mass structure

85
unimorph actuators (100 μm long) composed of Al/PECVD-SiN/ZnO/Al/LPCVD-
Si
x
N
y
/SiO
2
layers. As a DC offset voltage is applied between the top actuator
electrodes of the two actuators, the floating bridge is attracted to (by electrostatic
force) the cantilevers with simply-supported boundary condition at its anchors.
When an electric field is applied across the ZnO film, the two cantilevers bend, and
the air gap of the capacitor is changed, resulting in capacitance variation.
The design using simply-supported bridge structure has several advantages
over conventional cantilever or bridge structure. As a single cantilever is used for
tunable capacitor application, the gap distance between the moving electrode on the
cantilever and the stationary electrode on the substrate is not a constant value across
the whole capacitor area and hence the capacitance tuning ratio is limited by the
inherent incomplete closure of the gap. Alternatively, constant gap distance variation
and complete gap closure can be theoretically attained by a bridge structure.
However, bridge-based MEMS structure usually suffers from the tension that gets
developed as the bridge is deflected. As a result, large resistance from the tension
hinders the deflection and relatively high voltage is required for the bridge deflection.
To minimize the tension developed in the bridge, a new bridge structure has been
designed with simply-supported boundary condition at the anchors. The released
bridge, simply supported by two cantilevers, is free to move without any built-in
stress. The SU-8 confinement posts and confinement bridge are used to enclose the
simply-supported bridge in space as shown in Fig. 4.3. By varying the uniform gap

86
of the parallel-plate tunable capacitor, we can achieve a large capacitance tuning
without the cost of high voltage for large displacement.

4.1.4 Deflection Analysis
For the analysis of piezoelectric cantilever actuators, DeVoe and Pisano [1] use
a multiple-layer model to obtain an optimal thickness for the piezoelectric layer for a
largest deflection. However, with the thicknesses of the other layers undetermined,
the issue of optimization involving all the layers remains to be solved. Weinberg [2]
derives the constituent equations for predicting the static behavior of a multiple-layer
piezoelectric actuator. For a multiple-layer piezoelectric cantilever going through a
bending upon applied electric field, the radius of curvature ( R ) is derived as
Fig. 4.3 Schematic representation of the bridge-type surface-
micromachined tunable capacitor.
Capacitor
Simply-supported
bridge
Fully clamped B.C.
Confinement posts (for
lateral confinement)
Simply supported B.C.
Confinement bridge (for
vertical confinement)
ZnO-actuated cantilever
Capacitor
Simply-supported
bridge
Fully clamped B.C.
Confinement posts (for
lateral confinement)
Simply supported B.C.
Confinement bridge (for
vertical confinement)
ZnO-actuated cantilever

87
∑∑ ∑
∑ ∑ ∑ ∑
− +
+ −
=
ii
i i i i i i i
i
i i
ii
i i i
i
i i i i i i
i
i i i i i
A E z z A I E A E
A E z F d A E A E F d A E z
R
2 2
31 31
) ( ) (
) ( ) (
1
…………...………..(4.1)
and the deflection (δ
act
) as
∑∑ ∑
∑ ∑ ∑ ∑
− +
+ −
× = =
ii
i i i i i i i
i
i i
ii
i i i
i
i i i i i i
i
i i i i i
act
A E z z A I E A E
A E z F d A E A E F d A E z
x
R
x
x
2 2
31 31
2 2
) ( ) (
) ( ) (
2 2
) ( δ ….....(4.2)
where
i is index denoting material layer;
A is cross-section area;
i
z is z-axis position measured from the bottom surface of 1
st
layer to the middle
of i
th
layer;
x is x-axis position measured from the anchor point;
I is area moment of inertia;
31
d

is piezoelectric coefficient coupling z electric field to x strain;
E is effective Young’s modulus; and
F is electric field.
For our structure, since the passive capacitor is separated from the active
actuator, the deflection at the capacitor region (δ
cap
) is further modified as
∑∑ ∑
∑ ∑ ∑ ∑
− +
+ −
×
− +
=
− +
=
ii
i i i i i i i
i
i i
ii
i i i
i
i i i i i i
i
i i i i i
cap
A E z z A I E A E
A E z F d A E A E F d A E z
l x l l
R
l x l l
x
2 2
31 31
1 1
2
1 1 1
2
1
) ( ) (
) ( ) (
2
) ( 2
2
) ( 2
) ( δ
...(4.3)
where
1
l is the length of the actuator beam.

88
4.1.5 Structure Optimization
Based on these equations, simulations are performed to optimize the cantilever
structure in Fig. 4.1 for a largest deflection. In reality, though, there exists a
minimum practicable thickness for each layer. For example, there is a minimum
thickness of Al for electric continuity especially over a step, of ZnO for a well-
oriented columnar structure, of PECVD SiN for a good insulation layer, and of
parylene for sturdy supporting. Thus, the simulations are performed with the
thickness of Al, ZnO, PECVD SiN, and parylene-D larger or equal to 0.15 μm, 0.35
μm, 0.1 μm, and 1 μm, respectively. Other device parameters for simulation are
listed in Table 4.1 with the Young’s modulus values found from the literature [3].
The simulated results for a 100-μm-long actuator are summarized in Table 4.2. For a
constant applied voltage of 1 V, the largest deflection of 0.0454 μm happens with
parylene-D(1 μm)/Al(0.34 μm)/SiN(0.1 μm)/ZnO(0.35 μm)/Al(0.15 μm). For a
constant field of 3 V/μm (~1 V/0.35 μm) across ZnO piezoelectric film, the largest
deflection of 0.0729 μm occurs with parylene-D(1 μm)/Al(0.15 μm)/SiN(0.3
μm)/ZnO(0.35 μm)/Al(0.15 μm). Both cases are important for predicting the
maximum deflection under each of these operating regimes [1]. As the maximum
voltage is constrained by the low-voltage driving electronic circuits, the constant
voltage analysis should be applied. On the other hand, as the maximum electric field
is limited by the breakdown field of ZnO film, the constant field condition is of
concern.

89

Units (μm)
Minimum
thickness
Optimal thickness for
constant voltage (1V)
Optimal thickness for
constant field (3V/μm)
Bottom Al (1
st
) 0.15 0.15 0.15
ZnO (2
nd
) 0.35 0.35 0.35
PECVD SiN (3
rd
) 0.1 0.1 0.3
Top Al (4
th
) 0.15 0.34 0.15
Parylene-D (5
th
) 1.0 1.0 1.0
Deflection - 0.0454 0.0729

As shown in Table 4.2, for the constant field optimization, all layers, except
PECVD SiN, have their minimum thicknesses. For further investigation, we set the
thicknesses of Al, ZnO, and parylene-D to their minimum values, and plot the
simulated displacement of the microcantilever as a function of SiN thickness in Fig.
4.4(a). Figure 4.4(b) plots the positions of the top surface plane of ZnO (0.5 μm
since the bottom Al and ZnO are set to be 0.15 μm and 0.35 μm thick, respectively)
and of the neutral plane
N
z , which is given by ( ) ( )
∑ ∑
=
i
i i
i
i i i N
A E A E z z /.
Layer Width (μm) E (GPa) d
31
(pC/N)
Bottom Al (1
st
) 25 70 0
ZnO (2
nd
) 25 161 5
PECVD SiN (3
rd
) 25 160 0
Top Al (4
th
) 25 70 0
Parylene-D (5
th
) 25 2.8 0
Table 4.1 Parameters for the unimorph cantilever design and simulation
Table 4.2 Optimization of cantilever structure for largest deflection based
on constant voltage and constant field conditions

90

From Fig. 4.4, we see that the largest displacement occurs as the neutral plane
is right on the top surface of the ZnO film, when the SiN thickness is 0.3 μm. Thus,
we conclude that for a largest bending displacement under a constant field, the
cantilever should be designed as thin as possible, while the neutral plane condition is
0.0 0.1 0.2 0.3 0.4 0.5 0.6
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8

Z-axis position (μm)
SiN Thickness (μm)
Neutral plane position
Top surface of ZnO
Satisfaction of neutral
plane condition
0.0 0.1 0.2 0.3 0.4 0.5 0.6
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
Displacement at 100μm from anchor

Displacement (μm)
SiN Thickness (μm)
Deflection simulation for constant field (3V/μm)
Largest
displacement
(a)
(b)
0.0 0.1 0.2 0.3 0.4 0.5 0.6
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8

Z-axis position (μm)
SiN Thickness (μm)
Neutral plane position
Top surface of ZnO
Satisfaction of neutral
plane condition
0.0 0.1 0.2 0.3 0.4 0.5 0.6
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8

Z-axis position (μm)
SiN Thickness (μm)
Neutral plane position
Top surface of ZnO
Satisfaction of neutral
plane condition
0.0 0.1 0.2 0.3 0.4 0.5 0.6
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
Displacement at 100μm from anchor

Displacement (μm)
SiN Thickness (μm)
Deflection simulation for constant field (3V/μm)
Largest
displacement
(a)
(b)

Fig. 4.4 (a) Theoretical deflection as a function of
SiN thickness for constant field condition. (b)
Neutral plane position simulation.

91
satisfied, i.e. the neutral plane coincides with the top surface plane of the
piezoelectric film. This simple design rule is not only accurately deduced from the
simulation results, but theoretically reasonable in that a thinner structure has smaller
stiffness and that as the neutral plane is right on the surface plane of the piezoelectric
film, the bending stress is the largest. However, for the constant voltage condition,
since the voltage (or electric field) across the piezoelectric film depends on the
thickness ratio of the piezoelectric film to the insulation layer, the equations in [1-2]
should be used for optimal design.

4.1.6 Theoretical Capacitance Variation
For the designed cantilever consisting of an actuator (100 μm long (l
1
)) and a
capacitor region (50 μm wide (w
cap
) and 200 μm long (l
2
)) with an initial capacitor
gap (g
0
) of 1 μm, the theoretical capacitance variation as a function of applied
voltage is calculated by
12
1
1
0
21 2 1
0
22
ln
() 2
ll cap
cap
cap cap
cap cap l
l
g
A
ll ll dx
Cw w
ggx g
δ
ε
εε
δδ
+
⎡ ⎤
−
⎢ ⎥
⎡⎤
++
⎢ ⎥ == =
⎢⎥
−
⎢ ⎥ ⎢⎥
⎣⎦
⎢ ⎥
⎣ ⎦
∫
…………...…..(4.4)
where ε , A
cap
, g, δ
cap
are the dielectric constant of air, effective area of the parallel
plate, air-gap distance, and deflection at the tip of the capacitor region, respectively.
As can be seen in Fig. 4.5 that plots Eq. (4.4), a 4 V applied voltage produces a
tuning ratio of 1:4.5 (0.06 pF − 0.27 pF). With a larger voltage of 4.4 V that brings

92
the cantilever tip very close to the bottom electrode, the tuning ratio can be further
increased to 1:12 (0.06 pF − 0.72 pF).

For the optimization for the unimorph beam of the bulk-micromachined
structure in Fig. 4.2 and the cantilever of the bridge-type tunable capacitor in Fig. 4.3,
similar analysis is performed. However, practical thickness of the silicon supporting
beam has to be taken into consideration for the mass structure. The silicon-beam
thickness is critical for optimum actuation and for sturdiness of the structure as well.
In preliminary tests, 5-μm-thick silicon beams (formed with a silicon-on-insulator
(SOI) wafer) were attempted, but all devices were broken during the dicing step.
Hence a relatively large thickness of 25 μm is chosen in our current design for the
robustness of the structure and for fabrication easiness at the cost of a smaller
-5 -4 -3 -2 -1 0 1 2 3 4 5
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.72pF
0.27 pF
0.09 pF
0.06 pF

Capacitance (pF)
Applied Voltage (V)
Theoretical curve
O Theoretical data point

Fig. 4.5 Theoretical capacitance variation as a
function of applied voltage for cantilever-type
surface-micromachined tunable capacitor.

93
bending displacement. According to the simulation, with the elastic silicon beam set
to 25 μm thick, the optimal thickness of the ZnO layer is 3 μm for the constant
voltage condition. The corresponding displacement is 0.17 μm/V at the tip of the
mass structure, and is large enough for our application.

4.2 Device Fabrication
Three distinct approaches have been developed to fabricate the three different
ZnO-actuated tunable capacitors. First, a sacrificial silicon process is used for the
cantilever-type surface-micromachined tunable capacitor with parylene as the
supporting layer. Second, a bulk micromaching process is employed for the silicon-
supported tunable capacitor with a large silicon mass structure. Third, a surface
micromaching process, using both evaporated silicon and bulk silicon as the
sacrificial materials, has been developed to fabricate the bridge-type tunable
capacitor.

4.2.1 Surface-Micromachined Parylene-Supported Tunable Capacitor
For smaller operation voltage, the cantilever-type surface micromachined
tunable capacitor is fabricated according to the optimized structure for constant
voltage given in Table 4.2, and the fabrication process steps are shown in Fig. 4.6.
On both sides of the silicon wafer, 0.6-μm-thick low-stress LPCVD Si
x
N
y
is
first deposited for electrical isolation, followed by deposition and patterning of a
0.15-μm-thick layer of Al for bottom capacitor electrodes (Fig. 4.6(a)). Then 1-μm-

94
thick silicon is evaporated and delineated with HNA (a mixture of hydrofluoric acid,
nitric acid, and acetic acid) as a sacrificial layer (Fig. 4.6(b)). With the bottom
capacitor electrodes protected by silicon, another 0.15-μm-thick layer of Al is then
evaporated and patterned as the bottom actuator electrode (Fig. 4.6(c)). Then, a
0.35-μm-thick piezoelectric ZnO film is sputter-deposited at 300°C, 10 mtorr (with a
mixture of 50%Ar/50%O
2
) and 300 watts, followed by 0.1-μm-thick PECVD SiN
deposition. This PECVD SiN insulation layer is used to minimize the thermal
actuation effect by increasing the resistance between top and bottom electrodes.
After patterning PECVD SiN with CF
4
reactive ion etching (RIE) and ZnO with wet
etchant (CH
3
COOH: H
3
PO
4
: DI water = 1ml: 1ml: 50ml) (Fig. 4.6(d)), a 0.35-μm-
thick layer of Al is evaporated and patterned as the top capacitor and actuator
electrodes through a lift-off process to avoid etching the bottom electrode patterns
undesirably (Fig. 4.6(e)). Then a 1-μm-thick film of parylene-D is deposited and
delineated as the supporting layer to connect the actuator and capacitor regions (Fig.
4.6(f)). Since the thermal stress accounts for most (up to 90%) of the residual stress
of the parylene film [4], in order to minimize the residual stress, a soft bake at only
50°C (without any hard bake at all) is used for the photolithography process of the
parylene-D film. Finally, the cantilever is released with XeF
2
, which hardly attacks
Al, ZnO, and parylene (Fig. 4.6(g)).

95

Figure 4.7 shows the SEM photos of the fabricated devices, where the
capacitor and actuator regions are denoted. Two different capacitor structures are
fabricated. First, as shown in Fig. 4.7(a), the top capacitor electrode is directly
connected to the pad through a lead in the additional arm between the two actuating
arms, and an air-gap capacitor is formed between the top and bottom electrodes.
Another capacitor structure (shown in Fig. 4.7(b)) is designed to insure that the lead
does not obstruct the movement of the cantilever. In the design, the top capacitor
electrode is floating, and the two bottom capacitor electrodes (signal and GND) form
a series air-gap capacitor through the floating top electrode. It is shown that due to
the bending effects resulting from the residual stresses of different layers and the
residual-stress gradient within each layer, the released cantilevers are curved up
along the length direction and warped into a U-shape in the width direction.
Fig. 4.6 Fabrication process flow for the cantilever-type surface-
micromachined tunable capacitor.
(d)
(c)
(b)
(a)
(e)
(f)
(g)
LPCVD Si
x
N
y
PECVD SiN ZnO
Si wafer
Amorphous Si
Al
parylene
(d) (d)
(c) (c)
(b) (b)
(a) (a)
(e)
(f)
(g)
LPCVD Si
x
N
y
LPCVD Si
x
N
y
PECVD SiN ZnO ZnO
Si wafer Si wafer
Amorphous Si Amorphous Si
Al Al
parylene parylene

96

4.2.2 Bulk-Micromachined Silicon-Supported Tunable Capacitor
Figure 4.8 shows the fabrication process for the bulk-micromachined tunable
capacitor. A double-side-polished silicon wafer is used to reduce the surface
roughness effect on the capacitance. As discussed, the silicon-beam thickness is most
critical for optimal actuation and structural robustness, and so a two-step etching is
performed for accurate thickness control. First, 0.3-μm-thick LPCVD Si
x
N
y
is
deposited on both sides of the wafer, and patterned on the front side, followed by
etching of 25-μm-thick bulk silicon (Fig. 4.8(a)). Second, another 0.3-μm-thick
LPCVD Si
x
N
y
is deposited on both sides of the wafer, patterned on the backside (Fig.
4.8(b)), and 25-μm-thick silicon beams are formed with KOH etching from the
backside (Fig. 4.8(c) and Fig. 4.8(d)). Then, a 0.15-μm-thick layer of Al is
Fig. 4.7 SEM photos of the fabricated cantilever-type surface-micromachined
tunable capacitors (a) with the top capacitor electrode connected to the pad
through a lead and (b) with the top capacitor electrode floating.
Actuator
Top capacitor electrode pad
Bottom capacitor electrode
Anchor
Lead connecting top
capacitor electrode to pad
Top capacitor electrode
Type A
Actuator
GND
Anchor
Signal
Floating top
capacitor electrode
Bottom capacitor electrode
Type B
(a) (b)
Actuator
Top capacitor electrode pad
Bottom capacitor electrode
Anchor
Lead connecting top
capacitor electrode to pad
Top capacitor electrode
Type A
Actuator
Top capacitor electrode pad
Bottom capacitor electrode
Anchor
Lead connecting top
capacitor electrode to pad
Top capacitor electrode
Type A
Actuator
GND
Anchor
Signal
Floating top
capacitor electrode
Bottom capacitor electrode
Type B
Actuator
GND
Anchor
Signal
Floating top
capacitor electrode
Bottom capacitor electrode
Type B
(a) (b)

97
evaporated and patterned as the bottom actuator electrode, followed by sequential
depositions of 3-μm-thick ZnO, 0.1-μm PECVD SiN

insulation layer and 0.15-μm
Al top actuator electrode. After SiN and Al are delineated with one mask, ZnO is
patterned to expose the bottom actuator electrode for probe access, and a 0.15-μm
layer of Al is then deposited as the top capacitor electrode on the backside (Fig.
4.8(e)).

Fig. 4.8 Fabrication process flow for the bulk-micromachined
tunable capacitor ((a)-(c) are viewed from AB cross section, and
(d)-(f) are viewed from CD cross section).
(a)
(d)
(b)
(e)
(f)
CD
A
B
(c)
LPCVD Si
x
N
y
PECVD SiN
ZnO
Si glass
Al epoxy
(a)
(d)
(b)
(e)
(f)
CD
A
B
CD
A
B
(c)
LPCVD Si
x
N
y
PECVD SiN
ZnO
Si glass
Al epoxy

98
The wafer is diced, and individual chip is bonded with epoxy to a glass
substrate with patterned bottom capacitor electrodes and coplanar waveguide (CPW)
transmission lines over which a thin PECVD SiN dielectric layer is deposited to
avoid shorting of the electrodes (Fig. 4.8(f)).
To precisely control the initial gap of the tunable capacitor, we use a parylene
film slip as described below. First a 1-μm-thick parylene film is deposited on a
dummy wafer that has been pretreated with soap water, and peeled off from the
wafer. The parylene film slip is then placed over the bottom capacitor electrodes as a
sacrificial layer, as illustrated in Fig. 4.9. After an external force is applied to
sandwich the parylene film between the capacitor electrodes, we apply an epoxy
(Epo-tek 301 from Epoxy Technology Inc., MA) around the edge of the anchor area
and let it wick into the gap between the anchor and glass substrate by capillary force.
After we cure the epoxy at room temperature for 24 hours, we remove the external
force, and then pull out the parylene slip with a tweezer (parylene is mechanically
sturdy enough not to be broken during this pull-out). Thus formed capacitor gap is
mainly determined by the thickness of the parylene film, but is also influenced by the
morphology of the contacting surfaces and flatness of the actuator beam. Typically
an air gap of 1.5 to 2 μm is attained for the tunable capacitors.
SEM photos of the fabricated device are shown in Fig. 4.10, where we denote
the 25-μm-thick silicon beam and the 2-μm-wide air gap (between the silicon mass
structure and the glass substrate) of the capacitor.

99

4.2.3 Bridge-Type Surface-Micromachined Tunable Capacitor
A surface micromaching process, using silicon as the sacrificial layer, has been
developed to fabricate ZnO-actuated tunable capacitor with simply-supported bridge
Fig. 4.9 Illustration of the method used for the bonding process:
(1) place the parylene slip, (2) apply the external force, and (3)
apply the epoxy, in that order.
capacitor
epoxy
anchor
glass substrate
actuator beam Si mass
structure
dummy glass
dummy Si
(external force by gravitation)
dummy glass
parylene slip
c d e capacitor
epoxy
anchor
glass substrate
actuator beam Si mass
structure
dummy glass
dummy Si
(external force by gravitation)
dummy glass
parylene slip
c d e
Fig. 4.10 SEM photos of the fabricated bulk-micromachined tunable
capacitor: (a) top view and (b) side view.
25μm
(a) (b)
2μm
actuator
beam
mass
structure
actuator beam
mass
structure
anchor
anchor
25μm
(a) (b)
2μm
actuator
beam
mass
structure
actuator beam
mass
structure
anchor
anchor

100
structure. The fabrication process steps are shown in Fig. 4.11. In order to suppress
RF losses into the substrate, a high-resistance (~10 kΩcm) silicon wafer is used as
the substrate. On the silicon wafer, 0.03-μm-thick thermal oxide is first grown as the
etching mask for the final XeF
2
release step, followed by deposition of 0.4-μm-thick
low-stress LPCVD Si
x
N
y
as the supporting layer for the cantilever structure. A 0.1-
μm-thick layer of Al is deposited and patterned as the bottom actuator electrode (Fig.
4.11(a)). Then, a 0.4-μm-thick piezoelectric ZnO film is sputter-deposited, followed
by deposition of 0.1-μm-thick PECVD SiN as the insulation layer and 0.1-μm-thick
Al as the top actuator electrode. After patterning top actuator electrode and PECVD
SiN insulation layer with one mask (Fig. 4.11(b)), we pattern ZnO with another mask
(Fig. 4.11(c)) to avoid the potential short problem due to large undercut of ZnO
etching. A layer of PECVD SiO
2
is then deposited and delineated as the XeF
2

etching mask and as the insulation layer between the actuator and capacitor (Fig.
4.11(d)). Next we pattern LPCVD Si
x
N
y
and thermal oxide to form the opening for
the final XeF
2
release step (Fig. 4.11(e)). After the deposition and pattering of 0.2-
μm-thick parylene as the etching mask for the first XeF
2
release step (Fig. 4.11(f)),
0.1-μm-thick Al is deposited and delineated as bottom capacitor electrodes through a
lift-off process to avoid etching the actuator electrode patterns undesirably (Fig.
4.11(g)). Then 0.5-μm-thick silicon is evaporated (Fig. 4.11(h)), followed by
deposition and patterning of 0.1-μm-thick Al as the etching stop for the next silicon
etching step (Fig. 4.11(i)). Another 1-μm-thick silicon is evaporated and delineated

101
with CF
4
RIE as a sacrificial layer (Fig. 4.11(j)). A 0.25-μm-thick layer of Al is
evaporated and patterned as the top capacitor electrode through a lift-off process (Fig.
4.11(k)), followed by spinning and patterning of 4-μm-thick SU-8 to form the
simply-supported bridge structure and the confinement posts (Fig. 4.11(l)). After
another 1.5-μm-thick silicon is evaporated and patterned with CF
4
RIE (Fig.
4.11(m)), 7-μm-thick SU-8 is coated and delineated as the confinement bridge (Fig.
4.11(n)). Finally, we release the structure by first etching all the evaporated silicon to
release the simply-supported bridge with XeF
2
(Fig. 4.11(o)). Since the etching rates
of the evaporated amorphous silicon and the bulk crystalline silicon with XeF
2
are
different, by using parylene as the etching mask for XeF
2
, we divide the release
process into two steps – release of simply-supported bridge and release of cantilevers,
to minimize the undercut during release steps. After the parylene is removed with
oxygen plasma ashing, the cantilevers are released with XeF
2
(Fig. 4.11(p)).
Figure 4.12 shows the SEM photos of the fabricated devices, where the
cantilevers (for actuation) and the simply-supported bridge (for tunable capacitor)
are denoted.

102

Fig. 4.11 Fabrication process flow for the bridge-type surface-
micromachined tunable capacitor.
(a)
(b)
(n)
(m)
(l)
(h)
(p)
(g)
(f)
(e)
(d)
(c)
(o)
(k)
(j)
(i)
ZnO
Al
PECVD SiN
SU-8
Si
SiO
2
LPCVD Si
x
N
y
parylene
(a)
(b)
(n)
(m)
(l)
(h)
(p)
(g)
(f)
(e)
(d)
(c)
(o)
(k)
(j)
(i)
ZnO
Al
PECVD SiN
SU-8
Si
SiO
2
LPCVD Si
x
N
y
parylene
ZnO
Al
PECVD SiN
SU-8
Si
SiO
2
LPCVD Si
x
N
y
parylene

103

4.3 Experimental Results and Discussion
The fabricated ZnO-actuated tunable capacitors are measured for vertical
displacements with a focused-beam laser Doppler vibrometer (Optodyne VS-5010,
with resolution of 2.5 nm), and for capacitance variation with a HP 8753D network
analyzer and Cascade Microtech’s CPW microprobe with a ground-signal-ground
configuration.

4.3.1 Surface-Micromachined Parylene-Supported Tunable Capacitor
First, to discriminate the piezoelectric response from thermal or electrostatic
responses, an unbiased and bipolar signal is applied. Figure 4.13 shows the measured
actuation characteristics at the tip of the actuator for 500 Hz, ±8 V
peak-to-peak
(+8 V ~
−8 V) input signals with different waveforms applied to the actuator electrodes. It is
noted that the displacement at the capacitor region should be much larger, but
difficult to characterize due to the curved-up shape of the cantilever. As shown in
Capacitor
electrodes
Simply-supported
bridge
Confinement
bridge ZnO-actuated cantilever
Capacitor electrodes
Capacitor
electrodes
Simply-supported
bridge
Confinement
bridge ZnO-actuated cantilever
Capacitor electrodes
Fig. 4.12 SEM photos of the fabricated bridge-type surface-micromachined
tunable capacitor.

104
Fig. 4.13, the displacement amplitude is equally around 50 nm peak-to-peak for
different waveforms, and the cantilever deflects exactly at the same frequency (500
Hz) as the applied signal. Since electrostatic and thermal responses are both
independent of the actuation signal polarity, this polarity-dependent behavior shows
that the response is surely piezoelectric. No obvious difference in the bending
responses has been observed for the two structures with different top electrodes
shown in Fig. 4.7.
In order to distinguish the upward and downward motions of the cantilever, a
square wave signal with 30% duty cycle is used, and Fig. 4.14 shows the measured
result directly from the laser vibrometer. As shown in Fig. 4.14(a), an upward
movement is observed as +10 V is applied (a smaller position reading from the laser
vibrometer means that the distance between the incident point on the cantilever and
the laser head is smaller, and corresponds to an upward movement). Oppositely, as
shown in Fig. 4.14(b), a downward movement is attained for −10 V voltage, again
manifesting itself as a piezoelectric response. Moreover, from these measurements,
we can know the sign of coefficient d
31
of ZnO on Al film. In our design, the
piezoelectric ZnO film is located at the lower part of the beam, i.e. the middle plane
of the ZnO film is below the neutral plane of the whole structure. When a positive
electric field (positive field F) is applied, the ZnO film expands (positive strain S) in
the length direction and causes the cantilever to deflect upward. Thus, from S = d
31

× F, it is easily inferred that for our sputter system the coefficient d
31
of ZnO on Al
is positive, i.e. the c-axis is along the +z-axis. The sign of d
31
of ZnO on different

105
materials can be likewise attained through the characterization of the movement
direction with a laser vibrometer. This technique can also be employed for other
piezoelectric materials like AlN and PZT as well as for other actuation applications
where prior knowledge of the c-axis direction is needed for the design.

Fig. 4.13 Measured piezoelectric displacement characteristics of
the parylene-supported microcantilever for unbiased (a) sinusoidal,
(b) square, and (c) triangular voltage waveforms at 500 Hz.
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

sinusoidal
Position (nm)
Time (sec)
(a)
50nm
0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

Position (nm)
Time (sec)
triangular
(c)
50nm
0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

Position (nm)
Time (sec)
square
(b)
50nm
0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

sinusoidal
Position (nm)
Time (sec)
(a)
50nm
0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

sinusoidal
Position (nm)
Time (sec)
(a)
50nm
0.002 sec/cycle 0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

Position (nm)
Time (sec)
triangular
(c)
50nm
0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

Position (nm)
Time (sec)
triangular
(c)
50nm
0.002 sec/cycle 0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

Position (nm)
Time (sec)
square
(b)
50nm
0.002 sec/cycle
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
140
160

Position (nm)
Time (sec)
square
(b)
50nm
0.002 sec/cycle

106

The measured displacement versus applied voltage for sinusoidal inputs is
shown in Fig. 4.15. The cantilever is observed to deflect linearly and bi-directionally.
The upward displacement is around 4 nm/V, while the downward response is 2 nm/V.
This difference is due to the U-shape warping of the beam which hinders the
cantilever from moving down larger.
Fig. 4.14 Measured piezoelectric displacement characteristics
of the microcantilever for (a) positive and (b) negative applied
voltage with a duty cycle of 30%.
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
30% +10V, 70% 0V

Position (nm)
Time (sec)
moving up
30%, +10V
70%, 0V
(a)
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
30% -10V, 70% 0V

Position (nm)
Time (sec)
(b)
30%, -10V
70%, 0V
moving down
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
30% +10V, 70% 0V

Position (nm)
Time (sec)
moving up
30%, +10V
70%, 0V
(a)
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
30% +10V, 70% 0V

Position (nm)
Time (sec)
moving up
30%, +10V
70%, 0V
(a)
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
30% -10V, 70% 0V

Position (nm)
Time (sec)
(b)
30%, -10V
70%, 0V
moving down
0.000 0.002 0.004 0.006 0.008 0.010
0
20
40
60
80
100
120
30% -10V, 70% 0V

Position (nm)
Time (sec)
(b)
30%, -10V
70%, 0V
moving down

107

For the mechanical resonant frequency characterization, a sinusoidal signal of
±6 V
peak-to-peak
is applied, and the measured frequency response of the
microcantilever at the tip of the actuator is shown in Fig. 4.16. The fundamental
resonant frequency is measured to be around 100 kHz (with a quality factor of about
5), which is very close to the value estimated by
eff
A
EI
l
f ) (
2
875 . 1
2
2
ρ π
= ………………………………………………………....(4.5)
where A I E l , , , , ρ are the length, Young’s modulus, area moment of inertia, density,
and cross-section area of the composite cantilever, respectively [5]. With this high
resonant frequency, the actuator and tunable capacitor have fast operating speed and
great immunity to environmental vibrations.

-15 -10 -5 0 5 10 15 20
-40
-20
0
20
40
60

Displacement (nm)
Applied Voltage (V)
downward
upward
-15 -10 -5 0 5 10 15 20
-40
-20
0
20
40
60

Displacement (nm)
Applied Voltage (V)
downward
upward
Fig. 4.15 Measured piezoelectric displacement versus actuation
voltage of the cantilever-type tunable capacitor.

108

Due to the curved-up structure, the gap between the capacitor electrodes is
much larger than the designed 1 μm, as can be seen in Fig. 4.7, and the static
capacitance and capacitance change are too small to be measured. With the initial
tilting angle of about 20°, the static capacitance is estimated to be 0.004 pF. Thus,
the measured displacement of 200 nm at the center of capacitor region (actuated by
15 V) is estimated to produce a capacitance tuning ratio of around 0.6%. The low
tuning ratio is due to the initial curvature of the fabricated cantilever. To realize a
tunable capacitor with a larger tuning range, stress-compensation structure can be
designed for a flatter cantilever.

80k 90k 100k 110k 120k
100
150
200
250

10
1
10
2
10
3
10
4
10
5
10
6
0
50
100
150
200
250

Displacement (nm)
Frequency (Hz)
80k 90k 100k 110k 120k
100
150
200
250

10
1
10
2
10
3
10
4
10
5
10
6
0
50
100
150
200
250

Displacement (nm)
Frequency (Hz)

Fig. 4.16 Measured mechanical frequency response of the fabricated
cantilever-type surface-micromachined tunable capacitor.

109
4.3.2 Surface-Micromachined Cantilevers Utilizing Other Supporting Materials
We choose parylene as the supporting layer because parylene-D has a smaller
Young’s modulus (2.8 GPa) than PECVD SiN (160 GPa), and thus a parylene-
supported cantilever should be more flexible than a SiN-supported one. SU-8, with a
Young’s modulus (5 GPa) a little larger than parylene, has the potential as the
supporting material. However, parylene-D has smaller residual stress than SU-8 and
PECVD SiN, and the stiffness increase from the residual stress is the smallest. For
comparison, similar structures of Al(0.35 μm)/SiN(0.1 μm)/ZnO(0.35 μm)/Al(0.15
μm) with parylene-D (1 μm), PECVD SiN (0.1 μm), and SU-8 (1.6 μm) supporting
layers are fabricated. The measured displacement characteristics of these three
structures are compared in Fig. 4.17. The parylene-supported cantilever has a greater
deflection (4 nm/V) than the SiN-supported one (0.8 nm/V), in agreement with our
design to choose parylene as the optimal supporting material. For the SU-8-
supported structure, due to a large residual stress, the cantilever is more warped than
parylene-supported one, and the deflection is only 0.6 nm/V. We have also
fabricated other SU-8-supported cantilevers with different SU-8 thicknesses and
found that flatter cantilevers can be obtained with thicker SU-8 as shown in Fig. 4.18.
However, as we increase the thickness of the SU-8 film, the neutral plane position
shifts farther away from the top surface plane of the ZnO film, and the deflection
becomes smaller and smaller.

110

Fig. 4.18 SEM photos of the SU-8-supported tunable capacitors
with the SU-8 thickness of (a) 1.6 μm, (b) 2.7 μm, and (c) 6.5 μm.
(c)
(a)
(b)
(c) (c)
(a)
(b) (b)
0 5 10 15 20
0
10
20
30
40
50
60
70

SU8
PECVDSiN
ParyleneD
Displacement (nm)
Applied Voltage (V)
Fig. 4.17 Comparison of measured piezoelectric displacement
for the parylene-, SU-8-, and SiN-supported cantilevers.

111
4.3.3 Bulk-Micromachined Silicon-Supported Tunable Capacitor
The piezoelectric displacements at the capacitor region have been measured for
different applied voltages, and the results are shown in Fig. 4.19. The deflection is
bi-directional and highly linear, with a downward displacement consistently
observed to be around 0.086 μm/V until the 2-μm-wide gap is closed, and with an
equally large upward response from which a 3-μm upward displacement can be
obtained at −35 V. It should be noted that in this design, the middle plane of the ZnO
film is above the neutral plane of the whole structure. Thus, when a positive voltage
is applied, the ZnO film expands, resulting in a downward deflection.
The capacitance measurements have been carried out at frequencies from 2
GHz to 6 GHz. The capacitance variations and tuning ratios at 2 GHz as a function
of the driving voltage are shown in Fig. 4.20(a). For accurate assessment of the
nominal capacitance of the tunable capacitor, the parasitic capacitance of the CPW
(0.35 pF) is de-embedded. With a nominal static capacitance of 1.23 pF, we measure
that the capacitance increases to 10.02 pF at +25 V, and decreases to 0.46 pF at −35
V. Thus, the capacitance tuning ratio is as large as 21.8 to 1. It is worth pointing out
that in addition to the capacitance increase from “gap-closing” phenomenon upon a
positive applied voltage, the capacitance can be decreased by up to 63% from its
static value while the gap is opened wider with a negative applied voltage. This bi-
directional response is inherently absent in electrostatic and thermal actuations.

112

The theoretical capacitance curve due to a 0.086 μm/V gap variation at the
front end of the capacitor is plotted in Fig. 4.20(b). At actuation voltages below +20
upward
downward
-30 -20 -10 0 10 20 30
-2
-1
0
1
2
3

Displacement (μm)
Applied Voltage(V)
upward
downward
-30 -20 -10 0 10 20 30
-2
-1
0
1
2
3

Displacement (μm)
Applied Voltage(V)

Fig. 4.19 Measured piezoelectric displacement of the bulk-micromachined
tunable capacitor at the capacitor region versus actuation voltage.
-30 -20 -10 0 10 20 30
0
5
10
15
20

Capacitance (pF)
Applied Voltage (V)
Experimental Datapoints
Theoretical Plot
Normalized Theoretical Plot
-30 -20 -10 0 10 20 30
0
2
4
6
8
10

Applied Voltage(V)
Capacitance (pF)
0
5
10
15
20

Tuning Ratio
Fig. 4.20 Capacitance-versus-voltage curves for bulk-micromachined tunable
capacitor with mass structure mass. (a) Measured capacitance and tuning ratio
versus actuation voltage. (b) Comparisons of experimental and theoretical results.
(b)
(a)

113
V, the measured capacitances are slightly smaller than the theoretical values possibly
due to the fact that as the device is manually bonded to the glass substrate, the
bottom capacitor electrodes are not perfectly aligned with the edge of the mass
structure, leading to a smaller capacitance value. When the theoretical capacitance is
normalized to the measured one at the static capacitance point (i.e., the theoretical
value is divided by a factor of 1.6), the normalized theoretical plot fits the
experimental data fairly well as shown in Fig. 4.20(b). It is also found that the
measured capacitance deviates from the theoretical value greatly at voltages larger
than +20 V mainly because of the incomplete closure of the gap caused by the
unavoidable particles in the testing environment.
The measured S
11
Smith charts of the tunable capacitor at different actuation
voltages are shown in Fig. 4.21. Like other silicon-based MEMS tunable capacitors,
the finite resistance of the bulk silicon accounts for most energy loss, and the quality
factor of the tunable capacitor at 0.46 pF at 2 GHz is measured to be around 10.
Ultimately having the mass structure fabricated with a high-resistance silicon (or
glass) wafer would be an optimum design for large Q values.

114

4.3.4 Bridge-Type Surface-Micromachined Tunable Capacitor
The piezoelectric displacements have been characterized for the simply-
supported bridge as well as the basic cantilever (LPCVD-Si
x
N
y
-supported) and the
fully-clamped bridge for comparisons. As a signal of ±10 V
peak-to-peak
is applied to a
cantilever (100 μm long), the displacement amplitude is equally around 0.32 μm
peak-to-peak for different waveforms, and the cantilever deflects exactly at the same
frequency (1 kHz) as the applied signal (Fig. 4.22).

Fig. 4.21 S
11
Smith charts for bulk-micromachined tunable capacitor with
mass structure at the actuation voltage (V
act
) of (a) −35 V, (b) −25 V, (c) 0 V,
(d) 18 V, and (e) 25 V, from 2 GHz to 6 GHz.
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(c) V
act
= 0V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(b) V
act
= -25V
(a) V
act
= -35V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(d) V
act
=18V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(e) V
act
=25V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(c) V
act
= 0V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(b) V
act
= -25V
(a) V
act
= -35V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(d) V
act
=18V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0 0
j2
-j2
0
(e) V
act
=25V

115

0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2
1kHz Sinusoidal

Displacement (μm)
Time (sec)
(a)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2

1kHz Triangular
Displacement (μm)
Time (sec)
(b)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2
1kHz Square

Displacement (μm)
Time (sec)
(c)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2
1kHz Sinusoidal

Displacement (μm)
Time (sec)
(a)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2
1kHz Sinusoidal

Displacement (μm)
Time (sec)
(a)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2

1kHz Triangular
Displacement (μm)
Time (sec)
(b)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2

1kHz Triangular
Displacement (μm)
Time (sec)
(b)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2
1kHz Square

Displacement (μm)
Time (sec)
(c)
0.000 0.001 0.002 0.003 0.004 0.005
-0.2
-0.1
0.0
0.1
0.2
1kHz Square

Displacement (μm)
Time (sec)
(c)

Fig. 4.22 Measured piezoelectric displacement characteristics of the
basic cantilever (LPCVD-Si
x
N
y
-supported) for unbiased (a) sinusoidal,
(b) triangular, and (c) square voltage waveforms at 1 kHz.

116
The displacement responses have been measured for different applied voltages
and frequencies. As shown in Fig. 4.23, the deflection is bi-directional (with equally
large amplitude in both directions) and highly linear, with a displacement observed
to be around 0.016 μm/V at the tip of a basic cantilever, 0.012 μm/V at the tip of the
cantilever for a simply-supported bridge, and 0.008 μm/V at the center of the simply-
supported bridge. It is noted that to actuate the simply-supported bridge, certain
electrical signals are applied to the top and bottom actuation electrodes of the two
cantilevers as shown in Fig. 4.24. First, a DC offset voltage is applied between the
top electrodes of the two actuators to attract the floating bridge down to the
cantilevers by electrostatic force. Second, AC electric fields with the same
magnitude and phase are then applied across the ZnO films of the two separate
cantilevers to piezoelectrically actuate the cantilevers and the simply-supported
bridge accordingly. The deflection for the bridge is around 50% of the value for a
basic cantilever, indicating that deflection can be achieved without much tension
developed in the bridge. The bridge deflects downward continuously until the 0.35-
μm-wide gap is closed. It is noted that the gap is smaller than the designed 1-μm-
wide gap because the cantilevers were slightly curved down. For comparison, we
have also fabricated and characterized the bridge with its anchors fully clamped to
the cantilevers, and observed no response at an actuation voltage up to 60 V. The
displacement performance of different structures is summarized in Table 4.3.

117

-60 -40 -20 0 20 40 60
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Basic Cantilever

Displacement (μm)
Applied Voltage (V)
(a)
Upward
Downward
~0.016 μm/V
-60 -40 -20 0 20 40 60
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Cantilever for Simply-Supported Bridge

Displacement (μm)
Voltage (V)
(b)
Upward
Downward
~0.012 μm/V
(c)
Upward
Downward
~0.008 μm/V
-60 -40 -20 0 20 40 60
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
Simply-Supported Bridge

Displacement (μm)
Applied Voltage (V)
Testing point
Testing point
Testing point
-60 -40 -20 0 20 40 60
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Basic Cantilever

Displacement (μm)
Applied Voltage (V)
(a)
Upward
Downward
~0.016 μm/V
-60 -40 -20 0 20 40 60
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Basic Cantilever

Displacement (μm)
Applied Voltage (V)
(a)
Upward
Downward
~0.016 μm/V
-60 -40 -20 0 20 40 60
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Cantilever for Simply-Supported Bridge

Displacement (μm)
Voltage (V)
(b)
Upward
Downward
~0.012 μm/V
-60 -40 -20 0 20 40 60
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
Cantilever for Simply-Supported Bridge

Displacement (μm)
Voltage (V)
(b)
Upward
Downward
~0.012 μm/V
(c)
Upward
Downward
~0.008 μm/V
-60 -40 -20 0 20 40 60
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
Simply-Supported Bridge

Displacement (μm)
Applied Voltage (V)
(c)
Upward
Downward
~0.008 μm/V
-60 -40 -20 0 20 40 60
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
Simply-Supported Bridge

Displacement (μm)
Applied Voltage (V)
Testing point
Testing point
Testing point

Fig. 4.23 Measured piezoelectric displacement versus actuation
voltage for (a) basic cantilever, (b) cantilever for simply-
supported bridge, and (c) simply-supported bridge.

118

Theoretical Experimental
Structure
(with 100-
μm-long
cantilever)
Basic
Cantilever
Basic
Cantilever
Cantilever
for Fully-
Clamped
Bridge
Fully-
Clamped
Bridge
Cantilever
for Simply-
Supported
Bridge
Simply-
Supported
Bridge
Displacement
(per volt)
0.021 μm 0.016 μm0 μm 0 μm 0.012 μm 0.008 μm
V
top, right cantilever
(V
offset
)
V
bottom, right cantilever
(V
offset
±V
act
)
V
bottom, left cantilever
( ±V
act
)
V
top, left cantilever
(GND)
Floating
-V
act
+V
act
0
V
offset
-V
act
V
offset
+V
act
V
offset
V
bottom, left cantilever
V
bottom, right cantilever
(a)
(b)
V
top, right cantilever
(V
offset
)
V
bottom, right cantilever
(V
offset
±V
act
)
V
bottom, left cantilever
( ±V
act
)
V
top, left cantilever
(GND)
Floating
-V
act
+V
act
0
V
offset
-V
act
V
offset
+V
act
V
offset
V
bottom, left cantilever
V
bottom, right cantilever
-V
act
+V
act
0
V
offset
-V
act
V
offset
+V
act
V
offset
V
bottom, left cantilever
V
bottom, right cantilever
-V
act
+V
act
0
V
offset
-V
act
V
offset
+V
act
V
offset
V
bottom, left cantilever
V
bottom, right cantilever
(a)
(b)

Fig. 4.24 (a) Schematic showing how the electrical signals are applied to
the electrodes for bridge actuation. (b) Electrical signal waveforms
applied to bottom actuation electrodes of the two cantilevers.
Table 4.3 Comparison of displacement amplitude for different structures

119
For the mechanical resonant frequency characterization, sinusoidal signals are
applied, and the measured frequency responses of the basic cantilever, the cantilever
for a simply-supported bridge, and the simply-supported bridge are shown in Fig.
4.25. The fundamental resonant frequency is measured to be around 40 kHz (with a
quality factor of about 24) for the basic cantilever. The cantilever for simply-
supported bridge resonates at a lower frequency (~32 kHz) due to the added structure
mass. There is no resonance observed for the bridge due to the squeeze film effect,
and the displacement remains about the same up to 2 kHz.
The capacitance measurements have been carried out at frequencies from 2
GHz to 5 GHz. The measured capacitance at 3 GHz as a function of the actuation
voltage is shown in Fig. 4.26(a). For accurate assessment of the nominal capacitance
of the tunable capacitor, the parasitic capacitance (0.2 pF) of CPW is de-embedded
and the nominal capacitance is shown in Fig. 4.26(b). With a nominal static
capacitance of 0.23 pF, we measure that the capacitance increases to 1.82 pF at +40
V, and decreases to 0.13 pF at −30 V. Thus, the capacitance tuning ratio is as large
as 14 to 1 with an actuation voltage of 40 V. The theoretical capacitance curve due to
a 0.008 μm/V gap variation fits the experimental data fairly well as shown in Fig.
26(b).
The measured S
11
Smith charts of the tunable capacitor at different actuation
voltages are shown in Fig. 4.27. The quality factor of the tunable capacitor at 0.42 pF
at 2 GHz is measured to be around 20.

120

10
0
10
1
10
2
10
3
10
4
10
5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
V
act
= ±10V
Resonant Freq. ~ 32kHz
Q ~ 12
Cantilever for Simply-Supported Bridge

Displacement (μm)
Frequency (Hz)
(b)
(a)
10
0
10
1
10
2
10
3
10
4
10
5
0
1
2
3
4
5
6
7
8
Basic Cantilever
V
act
= ±10V
Resonance Freq. ~ 40kHz
Q~ 24

Displacement (μm)
Frequency (Hz)
(c)
10
0
10
1
10
2
10
3
10
4
10
5
0.00
0.05
0.10
0.15
0.20
V
act
= ±10V
Simply-Supported Bridge

Displacement (μm)
Frequency (Hz)
10
0
10
1
10
2
10
3
10
4
10
5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
V
act
= ±10V
Resonant Freq. ~ 32kHz
Q ~ 12
Cantilever for Simply-Supported Bridge

Displacement (μm)
Frequency (Hz)
(b)
10
0
10
1
10
2
10
3
10
4
10
5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
V
act
= ±10V
Resonant Freq. ~ 32kHz
Q ~ 12
Cantilever for Simply-Supported Bridge

Displacement (μm)
Frequency (Hz)
(b)
(a)
10
0
10
1
10
2
10
3
10
4
10
5
0
1
2
3
4
5
6
7
8
Basic Cantilever
V
act
= ±10V
Resonance Freq. ~ 40kHz
Q~ 24

Displacement (μm)
Frequency (Hz)
(a)
10
0
10
1
10
2
10
3
10
4
10
5
0
1
2
3
4
5
6
7
8
Basic Cantilever
V
act
= ±10V
Resonance Freq. ~ 40kHz
Q~ 24

Displacement (μm)
Frequency (Hz)
(c)
10
0
10
1
10
2
10
3
10
4
10
5
0.00
0.05
0.10
0.15
0.20
V
act
= ±10V
Simply-Supported Bridge

Displacement (μm)
Frequency (Hz)
(c)
10
0
10
1
10
2
10
3
10
4
10
5
0.00
0.05
0.10
0.15
0.20
V
act
= ±10V
Simply-Supported Bridge

Displacement (μm)
Frequency (Hz)

Fig. 4.25 Measured mechanical frequency response for the fabricated
(a) basic cantilever (LPCVD-Si
x
N
y
-supported), (b) cantilever for
simply-supported bridge, and (c) simply-supported bridge.

121

(a) (b)
-60 -40 -20 0 20 40 60
0.0
0.5
1.0
1.5
2.0
2.5
Measured Capacitance w/ Parasitic

Capacitance (pF)
Applied Voltage (V)
-40 -20 0 20 40
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

Capacitance (pF)
Applied Voltage (V)
Experimental Data
Theoretical Plot
(a) (b)
-60 -40 -20 0 20 40 60
0.0
0.5
1.0
1.5
2.0
2.5
Measured Capacitance w/ Parasitic

Capacitance (pF)
Applied Voltage (V)
-40 -20 0 20 40
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0

Capacitance (pF)
Applied Voltage (V)
Experimental Data
Theoretical Plot
Fig. 4.26 Capacitance-versus-voltage curves for bridge-type surface-
micromachined tunable capacitor. (a) Measured capacitance versus actuation
voltage. (b) Comparisons of experimental and theoretical results.
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(b) V
act
=0V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(a) V
act
= −30V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(c) V
act
= +25V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(d) V
act
= +40V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(b) V
act
=0V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(a) V
act
= −30V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(c) V
act
= +25V
0.2 0.5 1 2
j0.2
-j0.2
0
j0.5
-j0.5
0
j1
-j1
0
j2
-j2
0
(d) V
act
= +40V

Fig. 4.27 S
11
Smith charts for bridge-type surface-micromachined
tunable capacitor at the actuation voltage (V
act
) of (a) −30 V, (b) 0 V,
(c) +25 V, and (d) +40 V, from 2 GHz to 5 GHz.

122
4.3.5 Discussion
We compare the measured displacements with the designed values for different
fabricated structures. For the bulk-micromachined silicon-supported tunable
capacitor and the bridge-type surface-micromachined tunable capacitor, the
measured piezoelectric responses are more than half of the theoretical values.
However, for the surface-micromachined parylene-supported tunable capacitor, the
measured response is only one tenth of the theoretical value. Possible reasons are
discussed as follows. First, the underlying surface conditions such as cleanliness and
smoothness play an important role for good ZnO crystalline quality. In the
fabrication process used for the surface-micromachined parylene-supported tunable
capacitor, ZnO was sputter-deposited after the sacrificial silicon layer was
evaporated with the e-beam evaporator. Since the evaporated amorphous silicon has
large grain size and a rough surface, this surface roughness degrades the ZnO quality,
leading to a smaller piezoelectric constant. Second, the residual stress effect has not
been considered for the theoretical simulation that assumes a flat beam in both length
and width directions. However, for the surface-micromachined parylene-supported
microcantilever, the residual stress in each layer and the stress gradient through the
device thickness cause a non-flat structure after the release process. As a result, the
U-shape curvature in the width direction increases the stiffness of the cantilever, and
reduces the deflection amplitude greatly.
It is worth noting that although the bulk-micromachined silicon-supported
tunable capacitor and the bridge-type surface-micromachined tunable capacitor have

123
both attained relatively large piezoelectric displacements and hence a wide
capacitance tuning range, the silicon-supported tunable capacitor with mass structure
is rather bulky and comparatively challenging to be fabricated and integrated on a
single chip with IC technology. On the other hand, the bridge-type surface-
micromachined tunable capacitor has a much smaller dimension and can be
fabricated on a single chip without any bonding. The large tuning by a device with a
small footprint is achieved through the design and implementation of the simply-
supported bridge structure instead of the conventional cantilever or bridge structure.

Summary
The first MEMS piezoelectric tunable capacitors employing zinc oxide (ZnO)
actuation have been successfully fabricated. Piezoelectric displacement and
frequency performances of the fabricated devices have been thoroughly
characterized with a laser vibrometer. The deflection is demonstrated to be linear and
bi-directional, and the response is fast and immune to environmental vibrations. For
the first time, parylene is used as the supporting material for the cantilever-type
piezoelectric actuator, and the parylene-supported cantilever is shown to display
larger displacement than those supported by PECVD SiN or SU-8. With the design
of a bulk-micromachined mass structure driven by a ZnO unimorph, a 2,100%
continuous capacitance tuning range (from 0.46 pF to 10.02 pF) is achieved with
applied voltages from −35 V to +25 V. This is the highest tuning ratio ever reported
for parallel-plate tunable capacitors. Another design using a simply-supported bridge

124
structure driven by two ZnO-actuated cantilevers has also been implemented and
shows 1400% continuous tuning from 0.13 pF to 1.82 pF.

125
Chapter 4 References
[1] D. DeVoe and A. P. Pisano, “Modeling and optimal design of piezoelectric
cantilever microactuators,” J. Microelectromech. Syst., vol. 6, pp. 266–270,
Sept. 1997.
[2] M. S. Weinberg, “Working equations for piezoelectric actuators and sensors,” J.
Microelectromech. Syst., vol. 8, no. 4, pp. 529–533, Dec. 1999.
[3] F. J. von Preissig, H. Zeng, and E. S. Kim, "Measurement of piezoelectric
strength of ZnO thin films for MEMS applications," Journal of Smart Materials
and Structures, vol. 7, pp. 396–403, June 1998.
[4] T. A. Harder, T.-J. Yao, Q. He, C.-Y. Shih, and Y.-C. Tai, “Residual stress in
thin-film parylene-c”, in Proc. IEEE International Micro Electro Mechanical
Systems Conference, Las Vegas, NV, January 20 −24, 2002, pp. 435–438.
[5] M. V. Salapaka, H. S. Bergh, J. Lai, A. Majumdar, and E. McFarland, “Multi-
mode noise analysis of cantilevers for scanning probe microscopy,” J. Appl.
Phys., vol. 81, pp. 2480–2487, March 1997.

126
Chapter 5
Conclusion and Future Directions

Acoustic Ejector
An effective focusing scheme called “lens with air-reflectors (LWAR)” is
developed by utilizing the innate acoustic impedance mismatch between solid and
gas. LWAR has wide tolerance for its lens geometry, and has been shown to be very
effective in focusing acoustic waves for droplet ejection. With the LWAR ejectors,
ejections of micron-sized droplets have been attained through harmonic operations,
and ejections in directional angles have been realized through electrode pattern
designs. Both harmonic operations and directional ejections have been demonstrated
to be stable and reproducible. Also the LWAR ejector has been shown to dispense
oil with large viscosity without any difficulty. With the mentioned abilities, a new
microdroplet dispensing method capable of targeting a same spot, either in air or on
a solid surface, with combinations of different reagents is demonstrated.
For combinatory analysis involving more kinds of reagents, an array of more
ejectors can be designed and integrated on a single chip. With an array of N
directional ejectors, a spot can be inked with N species without any mechanical
movement. On the other hand, by dividing the circle electrode of one ejector into M
equal sectors, the ejection can be dynamically controlled into 2 1
M
−
(i.e.,
12 3
... 2 1
MMM M M
M
CCC C +++ + = − ) different angles through combinatory

127
activations of any 1, 2, 3,…, M sectors. A more flexible microfluidic system can thus
be constructed with arrays of multi-directional ejectors. Figure 5.1 demonstrates an
exemplary array of 10 ejectors targeting the same spot in the center. Connections to
cartridges for refill can be achieved peripherally through inlet ports for each
individual ejector.

(a)
1
(b)
2 3
(c)
Inlet ports
(cartridge connection)
Inlet ports
Targeted spot
Reagent
Ejector
(a)
1
(b)
2 3
(c)
Inlet ports
(cartridge connection)
Inlet ports
Targeted spot
Reagent
Ejector
(a)
1
(b)
2 3
(c)
Inlet ports
(cartridge connection)
Inlet ports
Targeted spot
Reagent
Ejector
Fig. 5.1 Schematic representation of an exemplary array of 10 ejectors for
droplet-based microfluidic system. (a) Diagram showing 10 ejectors (along with
10 peripheral inlet ports) targeting the same spot in the center with 10 different
liquids. (b) Detailed perspective drawing of a composed unit of the array. (c)
Demonstration of dynamic control of ejection angle through combinatory
actuation of any sector electrode – (1) Actuation of one sector for most oblique
ejection angle. (2) Actuation of three sectors for less oblique ejection angle. (3)
Actuation of five sectors for least oblique ejection angle. (Electrodes of one
ejector are divided into 12 equal sectors in this example.)

128
By dividing the electrode of one ejector into equal sectors and combinatorially
activating any combination of the sectors (Fig. 5.1(c)), the ejection angle can be
dynamically manipulated to compensate the direction variation due to utilization of
reagents with different viscosities and densities. The method has the potential to
systematically perform economical high-throughput analysis, and is ideal for parallel
and combinatory microreactions on a chip.

Piezoelectrically Actuated Tunable Capacitor
The first MEMS piezoelectric tunable capacitors employing zinc oxide (ZnO)
actuation have been presented. Several inherent advantages of piezoelectric actuation
over electrostatic actuation have been demonstrated, such as linear and bi-directional
deflection, low driving voltage, wide dynamic range due to no “pull-in” phenomenon,
and no electrostatic charging effect. Three different types of tunable capacitors
fabricated with surface and bulk micromachining techniques are presented. Of the
most interest is the bridge-type surface-micromachined tunable capacitor. Through
the implementation of a simply-supported bridge structure, large piezoelectric
displacements and a wide capacitance tuning range have been achieved. In addition,
the bridge-type tunable capacitor has a much smaller dimension than the tunable
capacitor with mass structure, and can be fabricated and integrated on a single chip
without any bonding.
With modifications of the bridge-type tunable capacitor, a contactless RF
MEMS switch with high power handling capability can be developed. The ON-state

129
and OFF-state of the switch are achieved by deflecting the simply-supported bridge
with ZnO actuation to move the electrodes on the bridge closer (or away) to (or from)
the other electrodes of two parallel capacitors that are formed by two air gaps above
and below the bridge (Fig. 5.2). In current MEMS technology, switches are prone to
failures when a high power RF signal passes them, because of the electrostatic force
produced by the RF signal. By contrast, the two-air-gap structure can balance the
electrostatic forces and yields a switch that can handle high RF power and is
insensitive to the RF power passing through the switch. This non-contacting
operation through piezoelectric actuation with the two-air-gap structure would
greatly improve the reliability of RF MEMS switches.

Input
Fixed
bridge
ZnO-actuated simply-
supported bridge
Output
Gap 1
Gap 2
Fig. 5.2 Top and perspective views of the piezoelectrically actuated
switch composed of two-air-gap capacitors.

130
Bibliography
J. R. Adam, N. R. Lindelad, and C. D. Hendricks, “The collision, coalescence, and
disruption of water droplets,” J. Appl. Phys., vol. 39, pp. 5173-5180, 1968.
P. Angenendt, J. Glokler, D. Murphy, H. Lehrach, and D. J. Cahill, “Toward
optimized antibody microarrays: a comparison of current microarray support
materials” Anal. Biochem., vol. 309, pp. 253–260, 2002.
M. Z. Bazant and Y. Ben, “Theoretical prediction of fast 3D AC electro-osmotic
pumps,” Lap Chip, vol. 6, pp. 1455–1461, 2006.
R. Borwick, P. Stupar, J. DeNatale, R. Anderson, C. Tsai, and K. Garrett, “A high Q,
large tuning range, tunable capacitor for RF applications,” in Proc. IEEE
International Micro Electro Mechanical Systems Conference, Las Vegas, NV,
January 20−24, 2002, pp. 669−672.
J. H. Butler, M. Cronin, K. M. Anderson, G. M. Biddison, F. Chatelain, M. Cummer,
D. J. Davi, L. Fisher, A. W. Frauendorf, F. W. Frueh, C. Gjerstad, T. F. Harper, S. D.
Kernahan, D. Q. Long, M. Pho, J. A. Walker, and T. M. Brennan, “In situ synthesis
of oligonucleotide arrays by using surface tension,” J. Am. Chem. Soc., vol. 123, pp.
8887–8894, 2001.
S. C. Chan, M. Mina, S. S. Udpa, L. Udpa, and W. Lord, “Finite element analysis of
multilevel acoustic Fresnel lenses,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr.,
vol. 43, pp. 670-677, 1996.
N. E. Chayen, “The role of oil in macromolecular crystallization,” Structure, vol. 5,
no. 10, pp. 1269–1274, 1997.
S. V. Chittur, “DNA microarrays: tools for the 21st century,” Comb. Chem. High T.
Scr., vol. 7, pp. 531–537, 2004.
E. Delamarche, D. Juncker, and H. Schmid, “Microfluidics for processing surfaces
and miniaturizing biological assays,” Adv. Mater., vol. 17, pp. 2911–2933, 2005.
U. Demirci, “Droplet-based photoresist deposition,” Appl. Phys. Lett., vol. 88, pp.
144104-1-3, 2006.
U. Demirci and M. Toner, “Direct etch method for microfluidic channel and
nanoheight post-fabrication by picoliter droplets,” Appl. Phys. Lett., vol. 88, pp.
053117-1-3, 2006.
J. L. DeRisi, V. R. Iyer, and P. O. Brown, “Exploring the metabolic and genetic
control of gene expression on a genomic scale,” Science, vol. 278, pp. 680-686, 1997.

131
D. DeVoe and A. P. Pisano, “Modeling and optimal design of piezoelectric
cantilever microactuators,” J. Microelectromech. Syst., vol. 6, pp. 266–270, Sept.
1997.
D. J. Duggan, M. Bittner, Y. Chen, P. Meltzer, and J. Trent, “Expression profiling
using cDNA microarrays,” Nat. Genet., vol. 21, pp. 10-14, 1999.
S. A. Elrod, B. Hadimioglu, B. T. Khuri-Yakub, E. G. Rawson, E. Richley, C. F.
Quate, N. N. Mansour, and T. S. Lundgren, “Nozzleless droplet formation with
focused acoustic beams,” J. Appl. Phys., vol. 65, pp. 3441-3447, 1989.
Z. Feng, W. Zhang, B. Su, K. F. Harsh, K. C. Gupta, V. Bright, and Y. C. Lee,
“Design and modeling of RF MEMS tunable capacitors using electro-thermal
actuators,” IEEE MTT-S Int. Microwave Symp. Dig., vol. 4, pp. 1507–1510, 1999.
B.-J. de Gans and U. S. Schubert, “Inkjet printing of polymer micro-arrays and
libraries: instrumentation, requirements, and perspectives,” Macromol. Rapid
Commun., vol. 24, pp. 659–666, 2003.
J. Gao, X.-F. Yin, and Z.-L. Fang, “Integration of single cell injection, cell lysis,
separation and detection of intracellular constituents on a microfluidic chip,” Lab
Chip, vol. 4, pp. 47–52, 2004.
X. Gao, E. Gulari, and X. Zhou, “In situ synthesis of oligonucleotide microarrays,”
Biopolymers, vol. 73, pp. 579–596, 2004.
H. Gau, S. Herminghaus, P. Lenz, and R. Lipowsky, “Liquid morphologies on
structured surfaces: from microchannels to microchips,” Science, vol. 283, pp. 46–49,
1999.
D. Gershon, “Microarray technology: an array of opportunities,” Nature, vol. 416, pp.
885–891, 2002.
C. L. Hansen, E. Skordalakes, J. M. Berger, and S. R. Quake, “A robust and scalable
microfluidic metering method that allows protein crystal growth by free interface
diffusion,” Proc. Natl. Acad. Sci., vol. 99, no. 26, pp. 16531–16536, 2002.
T. A. Harder, T.-J. Yao, Q. He, C.-Y. Shih, and Y.-C. Tai, “Residual stress in thin-
film parylene-c”, in Proc. IEEE International Micro Electro Mechanical Systems
Conference, Las Vegas, NV, January 20−24, 2002, pp. 435–438.
T. R. Hebner, C. C. Wu, D. Marcy, M. L. Lu, and J. Sturm, “Ink-jet printing of
doped polymers for organic light emitting devices,” Appl. Phys. Lett., vol. 72, pp.
519-521, 1998.

132
H. B. Hsieh, J. Fitch, D. White, F. Torres, J. Roy, R. Matusiak, B. Krivacic, B.
Kowalski, R. Bruce, and S. Elrod, ”Ultra-high-throughput microarray generation and
liquid dispensing using multiple disposable piezoelectric ejectors,” J. Biomol.
Screen., vol. 9, pp. 85-94, 2004.
D. Huang and E. S. Kim, “Micromachined acoustic-wave liquid ejector,” J.
Microelectromech. Syst., vol. 10, pp. 442-449, 2001.
T. R. Hughes, M. Mao, A. R. Jones, J. Burchard, M. J. Marton, K. W. Shannon, S. M.
Lefkowitz, M. Ziman, J. M. Schelter, M. R. Meyer, S. Kobayashi, C. Davis, H. Dai,
Y. D. He, S. B. Stephaniants, G. Cavet, W. L. Walker, A. West, E. Coffey, D. D.
Shoemaker, R. Stoughton, A. P. Blanchard, S. H. Friend, and P. S. Linsley,
“Expression profiling using microarrays fabricated by an ink-jet oligonucleotide
synthesizer,” Nature Biotechnol., vol. 19, pp. 342–347, 2001.
M. Joanicot and A. Ajdari, “Droplet control for microfluidics,” Science, vol. 309, pp.
887–888, 2005.
P. J. A. Kenis, R. F. Ismagilov, and G. M. Whitesides, “Microfabrication inside
capillaries using multiphase laminar flow pattering,” Science, vol. 285, pp 83–85,
1999.
A. Khademhosseini, R. Langer, J. Borenstein, and J. P. Vacanti, “Microscale
technologies for tissue engineering and biology,” Proc. Natl. Acad. Sci., vol. 103, no.
8, pp. 2480–2487, 2006.
J. B. Knight, A. Vishwanath, J. P. Brody, and R. H. Austin, “Hydrodynamic focusing
on a silicon chip: mixing nanoliters in microseconds,” Phys. Rev. Lett., vol. 80, no.
17, pp. 3863–3866, 1998.
J. W. Kwon, Q. Zou, and E. S. Kim, “Directional ejection of liquid droplets through
sectoring half-wave-band sources of self-focusing acoustic transducer,” Proceedings
of the 15th IEEE International Conference on Microelectromechanical Systems, Las
Vegas, NV, 2002, pp. 121–124.
L. E. Larson, R. H. Hackett, M. A. Melendes, and R. F. Lohr, “Micromachined
microwave actuator (MIMAC) technology − a new tuning approach for microwave
integrated circuits,” IEEE Microwave and Millimeter-wave Monolithic Circuits
Symp., pp. 21−30, 1991.
T. Laurell, J. Nilsson, and G. Marko-Varga, “Silicon microstructures for high-speed
and high-sensitivity protein identifications,” J. Chromatogr. B, vol. 752, pp. 217-232,
2001.

133
C. Lausted, T. Dahl, C. Warren, K. King, K. Smith, M. Johnson, R. Saleem, J.
Aitchison, L. Hood, and S. R. Lasky, “POSaM: a fast, flexible, open-source, inkjet
oligonucleotide synthesizer and microarrayer,” Genome Biology, vol. 5, R58, 2004.
C.-Y. Lee, H. Yu, and E. S. Kim, “Microreactions using nanoliter droplets with oil
encapsulation”, IEEE International Micro Electro Mechanical Systems Conference,
Kobe, Japan, January 21–25, 2007, pp. 81–84.
F. Levassort, L. P. Tran Huu Hue, J. Holc, T. Bove, M. Kosec, and M. Lethiecq,
“High Performance piezoceramic films on substrates for high frequency imaging,” in
Proc. IEEE Ultrasonics Symposium, Atlanta, GA, Oct. 7-10, 2001, pp. 1035-1038.
E. S. Y. Li and W.-T. Liu, “DNA microarray technology in microbial ecology
studies–principle, applications and current limitations,” Microbes Environ., vol. 18,
pp. 175–187. 2003.
L. Lian and N. R. Sottos, “Stress effects in sol-gel derived ferroelectric thin films,” J.
Appl. Phys., vol. 95, pp. 629-634, 2004.
F. Lin and E. C. Butcher, “T cell chemotaxis in a simple microfluidic device,” Lab
Chip, vol. 6, pp. 1462–1469, 2006.
J. M. Lopez-Cepero, “Silver carbonate staining reveals mitochondrial
heterogeneity,” J. Histochems. Cytochem., vol. 52, pp. 211–216, 2004.
T. B. Low and R. List, “Collision, coalescence and breakup of raindrops. Part I:
Experimentally established coalescence efficiencies and fragment size distributions
in breakup,” J. Atmos. Sci., vol. 39, pp. 1591-1606, 1982.
L.-H. Lu, K. S. Ryu, and C. Liu, “A magnetic microstirrer and array for microfluidic
mixing,” J. Microelectromech. Syst., vol. 11, pp. 462-469, 2002.
G. MacBeath and S. L. Schreiber, “Printing proteins as microarrays for high-
throughput function determination,” Science, vol. 289, pp. 1760-1763, 2000.
F. Mashayek, N. Ashgriz, W. J. Minkowycz, and B. Shotorban, “Coalescence
collision of liquid drops,” Int. J. Heat. Mass. Tran., vol. 46, pp. 77-89, 2003.
G. H. McGall, A. D. Barone, M. Diggelmann, S. P. A. Fodor, E. Gentalen, and N.
Ngo, “The efficiency of light-directed synthesis of DNA arrays on glass substrates,”
J. Am. Chem. Soc., vol. 119, pp. 5081–5090, 1997.
P. Muralt, “Ferroelectric thin films for micro-sensors and actuators: a review,” J.
Micromech. Microeng., vol. 10, pp.136-146, 2000.

134
H. D. Nguyen, D. Hah, P. R. Patterson, R. Chao, W. Piyawattanametha, E.K. Lau,
and M.C. Wu, “Angular vertical comb-driven tunable capacitor with high-tuning
capabilities,” J. Microelectromech. Syst., vol. 13, no. 3, pp. 406 −413, June 2004.
G. Nonglaton, I. O. Benitez, I. Guisle, M. Pipelier, J. Leger, D. Dubreuil, C. Tellier,
D. R. Talham, and B. Bujoli, “New approach to oligonucleotide microarrays using
zirconium phosphonate-modified surfaces,” J. Am. Chem. Soc., vol. 126, pp. 1497–
1502, 2004.
M. Orme, “Experiments on droplet collisions, bounce, coalescence and disruption,”
Prog. Energy Combust. Sci., vol. 23, pp. 65-79, 1997.
J. Y. Park, Y. J. Yee, H. J. Nam, and J. U. Bu, “Micromachined RF MEMS tunable
capacitors using piezoelectric actuators,” IEEE MTT-S Int. Microwave Symp. Dig.,
vol. 3, pp. 2111–2114, 2001.
G. Percin, T. S. Lundgren, and B. T. Khuri-Yakub, “Controlled ink-jet printing and
deposition of organic polymers and solid particles,” Appl. Phys. Lett., vol. 73, pp.
2375 -2377, 1998.
T. Pfohl, F. Mugele, R. Seemann, and S. Herminghaus, “Trends in microfluidics
with complex fluids,” Chemphyschem, vol. 4, pp. 1291–1298, 2003.
F. J. von Preissig, H. Zeng, and E. S. Kim, "Measurement of piezoelectric strength of
ZnO thin films for MEMS applications," Journal of Smart Materials and Structures,
vol. 7, pp. 396–403, June 1998.
T. G. S. M. Rijks, J. T. M. van Beek, P. G. Steeneken, M. J. E. Ulenaers, J. De
Coster, and R. Puers, “RF MEMS tunable capacitors with large tuning ratio,” in Proc.
IEEE International Micro Electro Mechanical Systems Conference, Maastricht, The
Netherlands, January 25 −29, 2004, pp. 777−780.
G. G. Rozenberg, E. Bresler, S. P. Speakman, C. Jeynes, and J. H. G. Steinke,
“Patterned low temperature copper-rich deposits using inkjet printing,” Appl. Phys.
Lett., vol. 81, pp. 5249-5251, 2002.
J. Sager, M. Young, and D. Stefanovic, “Characterization of transverse channel
concentration profiles obtainable with a class of microfluidic networks,” Langmuir,
vol. 22, pp. 4452–4455, 2006.
M. V. Salapaka, H. S. Bergh, J. Lai, A. Majumdar, and E. McFarland, “Multi-mode
noise analysis of cantilevers for scanning probe microscopy,” J. Appl. Phys., vol. 81,
pp. 2480–2487, March 1997.

135
S. A. Scaringe, C. Francklyn, and N. Usman, “Chemical synthesis of biologically
active oligoribonucleotides using β-cyanoethyl protected ribonucleoside
phosphoramidites,” Nucleic Acids Res., vol. 18, pp. 5433–5441, 1990.
E. E. Schadt, S. A. Monks, T. A. Drake, A. J. Lusis, N. Che, V. Colinayo, T. G. Ruff,
S. B. Milligan, J. R. Lamb, G. Cavet, P. S. Linsley, M. Mao, R. B, Stoughton, and S.
H. Friend, “Genetics of gene expression surveyed in maize, mouse and man,” Nature,
vol. 422, pp. 297–302, 2003.
M. Schena, D. Shalon, R. W. Davis, and P. O. Brown, “Quantitative monitoring of
gene expression patterns with a complementary DNA microarray,” Science, vol. 270,
pp. 467-470, 1995.
S. Schmatloch, M. A. R. Meier, and U. S. Schubert, “Instrumentation for
combinatorial and high-throughput polymer research: a short overview,” Macromol.
Rapid Commun., vol. 24, pp. 33–46, 2003.
R. Seemann, E. J. Kramer, and F. F. Lange, “Patterning of polymers: precise channel
stamping by optimizing wetting properties,” New J. Phys., vol. 6, 111, 2004.
M. S. Shchepinov, S. C. Case-Green, and E. M. Southern, “Steric factors influencing
hybridization of nucleic acids to oligonucleotide arrays,” Nucleic Acids Res., vol. 25,
pp. 1155–1161, 1997.
S. Singh-Gasson, R. D. Green, Y. Yue, C. Nelson, F. Blattner, M. R. Sussman, and F.
Cerrina, “Maskless fabrication of light-directed oligonucleotide microarrays using a
digital micromirror array,” Nature Biotechnol., vol. 17, pp. 74–978, 1999.
C. H. Sorum, F. S. Charlton, J. A. Neptune, and J. O. Edwards, “pH change as an
index to reaction mechanisms,” J. Am. Chem. Soc., vol. 74, no. 1, pp. 219–221, 1952.
T. M. Squires and M. Z. Bazant, “Induced-charge electro-osmosis,” J. Fluid Mech.,
vol. 509, pp. 217–252, 2004.
C. Tilstone, “DNA microarrays: vital statistics,” Nature, vol. 424, pp. 610–612, 2003.
M. A. Unger, H.-P. Chou, T. Thorsen, A. Scherer, and S. R. Quake, “Monolithic
microfabricated valves and pumps by multilayer soft lithography,” Science, vol. 288,
pp. 113–116, 2000.
O. D. Velev and K. H. Bhatt, “On-chip micromanipulation and assembly of colloidal
particles by electric fields,” Soft Matter, vol. 2, pp. 738–750, 2006.

136
M. J. Vellekoop, C. C. G. Visser, P. M. Sarro, and A. Venema, “Compatibility of
zinc oxide with silicon IC processing,” Sensors Actuators, vol. A23, pp. 1027–1030,
1990.
M. S. Weinberg, “Working equations for piezoelectric actuators and sensors,” J.
Microelectromech. Syst., vol. 8, no. 4, pp. 529–533, Dec. 1999.
Z. Xiao, W. Peng, R. F. Woffenbuttel, and K. R. Farmer, “Micromachined variable
capacitor with wide tuning range,” in Proc. Solid-State Sensor and Actuator
Workshop, Hilton Head Island, SC, June 2–6, 2002, pp. 346–349.
T. Xu, J. Jin, C. Gregory, J. J. Hickman, and T. Boland, “Inkjet printing of viable
mammalian cells,” Biomaterials, vol. 26, pp. 93-99, 2005.
H. Zhang and W. Davison, “Performance characteristics of diffusion gradients in
thin films for the in situ measurement of trace metals in aqueous solution,” Anal.
Chem., vol. 67, pp. 3391–3400, 1995.
B. Zheng, L. S. Roach, and R. F. Ismagilov, “Screening of protein crystallization
conditions on a microfluidic chip using nanoliter-size droplets,” J. Am. Chem. Soc.,
vol. 125, pp. 11170–11171, 2003.

Abstract (if available)

Abstract This thesis presents microfluidic and radio-frequency (RF) microelectromechanical systems (MEMS) based on acoustic ejectors employing lens with air-reflectors (LWAR) and piezoelectrically actuated tunable capacitors, respectively.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Piezoelectric ultrasonic and acoustic microelectromechanical systems (MEMS) for biomedical, manipulation, and actuation applications

PDF

Piezoelectric MEMS for acoustic sensing, contactless RF switching, and resonant mass sensing

PDF

Audio and ultrasound MEMS built on PZT substrate for liquid and airborne applications

PDF

Integrated wireless piezoelectric ultrasonic transducer system for biomedical applications

PDF

Ultrasonic microelectromechanical system for microfluidics, cancer therapeutics and sensing applications

PDF

Zero-power sensing and processing with piezoelectric resonators

PDF

Design, modeling and analysis of piezoelectric forceps actuator

PDF

MEMS piezoelectric resonant microphone arrays and their applications

PDF

Surface acoustic wave waveguides for signal processing at radio frequencies

PDF

Battery-less detection and recording of tamper activity along with wireless interrogation

PDF

Magnetic spring in electromagnetic vibration energy harvester and applications of focused ultrasonic transducer

PDF

Design and implementation of frequency channelized ultra-wide-band (UWB) transceivers

Asset Metadata

Creator Lee, Chuang-Yuan (author)

Core Title Acoustic ejector employing lens with air-reflectors and piezoelectrically actuated tunable capacitor

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Electrical Engineering

Publication Date 07/23/2007

Defense Date 06/20/2007

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag ejector,OAI-PMH Harvest,tunable capacitor

Language English

Advisor Kim, Eun Sok (committee chair), Katsouleas, Thomas (committee member), Meng, Ellis F. (committee member)

Creator Email chuangyl@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m639

Unique identifier UC1456494

Identifier etd-Lee-20070723 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-518848 (legacy record id),usctheses-m639 (legacy record id)

Legacy Identifier etd-Lee-20070723.pdf

Dmrecord 518848

Document Type Dissertation

Rights Lee, Chuang-Yuan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu