A COMPREHENSIVE STUDY IN THE APPLICABILITY OF THE SOLID-SOLID
STRUCTURAL PHASE TRANSITION OF VANADIUM DIOXIDE TO
MICRO-ELECTRO-MECHANICAL DEVICES
by
Rafmag Cabrera

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Electrical Engineering - Doctor of Philosophy
2014

ABSTRACT
A COMPREHENSIVE STUDY IN THE APPLICABILITY OF THE SOLID-SOLID
STRUCTURAL PHASE TRANSITION OF VANADIUM DIOXIDE TO
MICRO-ELECTRO-MECHANICAL DEVICES
by
Rafmag Cabrera
This dissertation presents a series of studies which explores the use of the structural
phase transition of V O2 in MEMS actuators. MEMS actuators enable the interaction of
devices with their surroundings, instead of simply sensing it. Actuation mechanisms, which
are very common in the macro scale, such as pneumatics, induction motors, and combustion
engines, are very difficult and inefficient to implement in the micro scale. Therefore, new
actuation techniques must be developed and optimized. Current actuation mechanisms such
as electrostatic or thermal expansion have limited performance in terms of total displacement,
applied force, operating voltage, and power requirements. This has resulted in a push to
incorporate new multifunctional smart materials into standard MEMS devices to improve
performance. V O2 has proven to be one of the most promising materials for the creation of
new sensors and actuators. Although most of its electrical and optical properties have been
thoroughly studied, there is very little work related to its SPT and its application to MEMS
actuators. Throughout this dissertation the SPT of V O2 is extensively studied, methods
are developed for the optimal design of V O2 -based actuators. The memory effect caused
by the hysteresis in the SPT is exploited to create very robust programmable multiple state
actuators. And fabrication processes are developed which combine the use of V O2 with
standard MEMS fabrication processes.

To my family and friends.

iii

ACKNOWLEDGMENTS

I would like to thank my advisor Dr. Nelson Sepulveda for all the help, guidance and patience
he has provided during the years I have known him. He has been a great influence in every
step of my professional development. I can confidently say that I would not be where I am,
without his help. I would also like to thank all the members of the Applied Materials Group,
for their friendship and research collaborations. Im also deeply grateful for the support of
Dr. Armando Rua and Dr. Felix Fernandez, who provided me with support through the
beginning of my graduate studies.
Im also grateful for the support from the members of my graduate committee; Dr. Tim
Hogan, Dr. Shantanu Chakrabartty, and Dr. David Grummon.
I would also like to thank the National Science Foundation for funding my Ph.D through
NSF ECCS Award ECCS-1306311 and NSF Award ECCS 1139773 monitored by Anupama
Kaul.
Finally I would like to thank my family, without whom none of this would have been
possible.

iv

TABLE OF CONTENTS

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii
CHAPTER 1 INTRODUCTION . . . . . . . . . . . . .
1.1 Problem Description and Motivation . . . . . . . . . .
1.2 Thesis Statement . . . . . . . . . . . . . . . . . . . . .
1.3 Anticipated Research Contributions . . . . . . . . . . .
1.4 Dissertation Outline . . . . . . . . . . . . . . . . . . .

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

.
.
.
.
.

. . .
. . .
. . .
. . .
. . .

1
2
3
3
4

CHAPTER 2 BACKGROUND . . . . . . . . . . . . . .
2.1 Thermal Actuation in MEMS . . . . . . . . . . . . . .
2.1.1 Geometrically constrained devices . . . . . . . .
2.1.2 Pseudo bimorphs . . . . . . . . . . . . . . . . .
2.1.3 Bimorph Cantilevers . . . . . . . . . . . . . . .
2.2 Phase Change materials . . . . . . . . . . . . . . . . .
2.2.1 GeSbTe . . . . . . . . . . . . . . . . . . . . . .
2.2.2 Shape Memory Alloys . . . . . . . . . . . . . .
2.3 The Structural Phase Transition of Vanadium Dioxide .
2.4 Strain Energy Density . . . . . . . . . . . . . . . . . .
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .
. . .

5
5
6
7
8
10
10
11
12
14
16

CHAPTER 3 THE MEMORY EFFECT IN VANADIUM DIOXIDE
3.1 Electrical Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Optical Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Mechanical Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . 17
. . . 17
. . . 21
. . . 24
. . . 28

CHAPTER 4 EMBEDDED ELECTRIC MICRO-HEATERS
4.1 Micro-Cantilever Design and Fabrication . . . . . . . . . . . .
4.2 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . .
4.3 Heater Design . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.4.1 Static Analysis . . . . . . . . . . . . . . . . . . . . . .
4.4.2 Frequency Response . . . . . . . . . . . . . . . . . . .
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

. . . 30
. . . 30
. . . 33
. . . 35
. . . 36
. . . 37
. . . 40
. . . 42

CHAPTER 5 CONTROL OF THE MEMORY EFFECT
5.1 Design of the memory device . . . . . . . . . . . . . . . .
5.2 Static and transient response . . . . . . . . . . . . . . . .
5.3 Open-loop versus closed-loop behavior . . . . . . . . . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

. . . 44
. . . 44
. . . 46
. . . 49

v

.
.
.
.

.
.
.
.

.
.
.
.

5.4
5.5

Performance of the Vanadium Dioxide-based MEMS memory . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

52
54

CHAPTER 6 MEMS COMPATIBLE FABRICATION PROCESS FOR
VANADIUM DIOXIDE . . . . . . . . . . . . . . . . . . . . . . 55
6.1 Vanadium Dioxide-based Voltage Controlled Resistor . . . . . . . . . . . . . 55
6.2 Device Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.3 Mask Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.4 Fabrication Process Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
6.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
CHAPTER 7 SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.1 Summary of Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.2 List of Problems Solved in this Thesis . . . . . . . . . . . . . . . . . . . . . . 68
BIBLIOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

vi

LIST OF TABLES

Table 4.1

Material properties used in the FEM model.

. . . . . . . . . . . . . . . .

34

Table 4.2

Estimated Power Consumption . . . . . . . . . . . . . . . . . . . . . . . .

40

vii

LIST OF FIGURES

Figure 1.1

Characteristic behavior of the IMT in V O2

. . . . . . . . . . . . . . . .

1

Figure 2.1

Example of an array of bent-beam actuators [1]. . . . . . . . . . . . . . .

6

Figure 2.2

Example of a cascade configuration which quadruples the total deflection.

7

Figure 2.3

An example of a pseudo-bimorph amplification of thermal expansion. . .

8

Figure 2.4

An example of a bimorph cantilever. . . . . . . . . . . . . . . . . . . . .

9

Figure 2.5

bimaterial cantilever bending upon incident heat, side view of the polymersilicon micro-cantilever bending as temperature increases from 20 to 40
o C, [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9

0.9 µm3 working composite shape memory actuator. On thermocycling
the actuator exhibits reversible free end deflection of 230 nm, [3]. . . . .

12

Crystal structure of the monoclinic, low-temperature phase (left) and
the rutile, high-temperature phase (right). Vanadium atoms are depicted in red and oxygen atoms in a lighter color. Because of a symmetrybreaking pairing of vanadium atoms (violet lines), the axis definitions
change [4]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

Multiple exposure picture of V O2 -coated silicon cantilever during heating [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

Strain energy densities of different micro-actuators including the V O2 based micro-actuator studied in [6] (green). . . . . . . . . . . . . . . . .

15

Series of hysteresis curves for V O2 -coated microbridge sample. Interior
curves are obtained by cooling before the end of the transition region is
reached [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18

Transient of the resistance to a 2 s light pulse. The inset shows the
response in the value of resistance when a 2 s red (λ = 672 nm) light
pulse illuminates the sample. The bolder trace in one of the hysteresis
curves suggests the thermal evolution of the sample [7]. . . . . . . . . . .

19

Resistance as a function of time as a series of four light pulses with
increasing duration illuminate the sample [7]. . . . . . . . . . . . . . . .

20

Figure 2.6

Figure 2.7

Figure 2.8

Figure 2.9

Figure 3.1

Figure 3.2

Figure 3.3

viii

Figure 3.4

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Hysteresis curve for transmittance at λ = 1.31µm for V O2 -coated flat
sample [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

Transmittance measured at λ = 1.31µm for the V O2 -coated flat sample
after it was illuminated with a series of red (λ =672 nm) light pulses
with equal irradiance and increasing duration, starting with a 0.6 s pulse
and increasing by 0.6 s each time [7]. . . . . . . . . . . . . . . . . . . . .

22

Sample transmittance at =1.31 µm for the V O2 -coated flat sample in
response to a series of 80 V electric pulses with increasing duration lower
curve and to a series of increasing voltages with equal 1 s duration upper
curve [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

SEM image of the commercially available silicon micro cantilever chips
(MikroMasch). The center cantilever is 350 µm long. . . . . . . . . . . .

24

Measurement set up. The entire system was in ambient pressure. As
shown in the insets, the measuring laser (L1) was focused at the tip of
the cantilever, while the heating or programming laser (L2) was focused
close to the center of the cantilever beam [8]. . . . . . . . . . . . . . . . .

25

Deflection of the cantilever as a function of temperature. A relatively
wide hysteresis width for V O2 thin films of 30 o C can be noticed. The
left y-axis represents the voltage reading form the PSD, and the right
y-axis is the cantilever deflection [8]. . . . . . . . . . . . . . . . . . . . .

26

Figure 3.10 Non-uniform mechanical states found from linearly increasing light pulses
(blue trace) and for five consecutive light pulses of 4 mW (red trace) [8].
Figure 4.1

27

General fabrication process flow. (A) Bottom electrode deposition and
patterning by lift-off, (B) Sacrificial layer (amorphous silicon) deposition
and patterning by RIE, (C) First SiO2 deposition and patterning, (D)
Top electrode deposition (Heater), (E) Second SiO2 deposition, (F)
Patterning of the second SiO2 layer, (G) Release of the MEMS actuator,
and (H) V O2 -deposition by pulsed-lased deposition using in-situ shadow
mask [9]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

Resistance as a function of temperature for the V O2 film used. The
V O2 film was a circular pattern of 6 mm in diameter, and the probes
were 2 mm apart [9]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32

Figure 4.3

SEM image of cantilevers with dimensions (top view) [9]. . . . . . . . . .

32

Figure 4.4

Temperature dependent thermal expansion coefficient of V O2 (plot of
α(T ) in Equation (4.1)) [9]. . . . . . . . . . . . . . . . . . . . . . . . . .

35

FEM simulation results for heater optimization [9]. . . . . . . . . . . . .

36

Figure 4.2

Figure 4.5

ix

FEM simulation results of the temperature (o C) distribution for the 300
µm cantilever in air and vacuum. The anchor is at the left [9]. . . . . . .

37

SEM images of cantilevers. The top and bottom images show the actuator before and after actuation, respectively [9]. . . . . . . . . . . . . . .

38

Figure 4.8

Static response and FEM simulation results in vacuum and in air [9]. . .

39

Figure 4.9

Schematic of system used for frequency measurements [9]. . . . . . . . .

41

Figure 4.10 Frequency response curves for the cantilevers in vacuum and in air [9]. .

42

Figure 4.6

Figure 4.7

Figure 5.1

Top view false color image of the memory cell. The Pt/Ti heater is
shown in red [10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

45

Measured resistance of the heater (and connecting traces) as the temperature of the entire chip is varied [10]. . . . . . . . . . . . . . . . . . .

45

Measured deflection of the device as the temperature of the entire chip
is varied [10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

46

Deflection of the cantilever as the current is cycled, the deflection is
plotted as a function of the device resistance [10]. . . . . . . . . . . . . .

47

Step response of the measured resistance of the device. Scattered points
are individual data measurements, and solid curve represents the average of the previous 10 resistance measurements [10]. . . . . . . . . . . . .

48

Circuit schematic showing how the device is connected, where the voltages are read, and a block diagram of the FPGA controller [10]. . . . . .

50

Figure 5.7

Close-loop response of the device at 30 and 40 o C [10]. . . . . . . . . . .

50

Figure 5.8

Open-loop response of the device at 30 and 40 o C [10]. . . . . . . . . . .

52

Figure 5.9

Repeatability test of the memory device in close loop [10]. . . . . . . . .

53

Figure 5.10 Demonstration of multiple stable states in closed loop [10]. . . . . . . . .

53

Figure 6.1

Schematic symbol of a VCR. . . . . . . . . . . . . . . . . . . . . . . . . .

55

Figure 6.2

Characteristics of ideal VCR . . . . . . . . . . . . . . . . . . . . . . . . .

56

Figure 6.3

Schematic of the VCR device connected in feedback for linearization.
blue (VCR), red (“feedback resistor”), and green (resistive heater). . . .

57

Resistivity of a 187 nm thick V O2 film. . . . . . . . . . . . . . . . . . . .

58

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

Figure 5.6

Figure 6.4

x

Figure 6.5

Top view of the interdigitated resistive array design used. . . . . . . . . .

59

Figure 6.6

Mask design for one of the VCR designs. . . . . . . . . . . . . . . . . . .

60

Figure 6.7

Full wafer Mask used in the VCR process. . . . . . . . . . . . . . . . . .

61

Figure 6.8

Fabrication Results.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

62

Figure 6.9

Resistance as a function of temperature for the V O2 film right after
deposition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

63

Figure 6.10 Open loop test on the 200mw 200k VCR. . . . . . . . . . . . . . . . . . .

64

Figure 6.11 Exampe of the metal pealing off the V O2 film. . . . . . . . . . . . . . . .

64

Figure 6.12 Open loop test on the 100mw 500k VCR, showing large differences in
R1 and R2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

65

Figure 6.13 Transient response of the resistance of the VCR, due to a step input. . .

66

Figure 6.14 Transient response of the resistance of the 500k 100mw VCR, due to a
step input for the heating and cooling cycle. . . . . . . . . . . . . . . . .

66

Figure 6.15 Close Loop Test on the 200 mW, 200k VCR. . . . . . . . . . . . . . . . .

67

xi

CHAPTER 1
INTRODUCTION

The insulator to metal transition (IMT) in vanadium dioxide (V O2 ) has been studied since
1959 when it was first reported by Morin [11]. V O2 has proven to be very useful in the
creation of new devices due to the fact that many of the materials properties change drastically during the transition (i.e. the material is considered to be a multifunctional smart
material)[12, 13, 14]. But the characteristic that gives V O2 the edge over other phase change
materials in practical applications, is the fact that the transition temperature is at around
68 o C and exhibits temperature hysteresis of several degrees, which can be tuned through
doping or choice of substrate. [15]
V O2 can be deposited in thin films through various methods. All V O2 films in this dissertation where deposited using pulsed laser deposition (PLD). Figure 1.1 shows the electrical
and optical characteristic behavior of some V O2 thin films (deposited by our research group)
during the IMT.[16]

1 M

T r a n s m is s iv ity ( % )

R e s i s t a n c e ( Ω)

5 0

1 0 0 k

1 0 k

H e a tin g
C o o lin g
1 k

4 0
3 0
2 0
1 0

H e a tin g
C o o lin g
0

3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

3 0

T e m p e ra tu re (oC )

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

T e m p e ra tu re (oC )

(a) Resistance curve for a V O2 thin film over
SiO2 .

(b) Transmissivity of V O2 films during the
transition.

Figure 1.1 Characteristic behavior of the IMT in V O2
The crystal structure of V O2 is tetragonal in its metallic phase. On cooling through

1

the IMT, this changes to a monoclinic phase by slight reordering of the vanadium ions,
which causes a doubling of the unit cell. This change in the structure of the unit cell,
hereinafter referred to as the structural phase transition (SPT), can be exploited in the
creation of actuators or sensors which, in previous actuator technologies, depended only on
the difference in the thermal expansion coefficient between two materials. Because of the very
large contraction of the V O2 ’s crystal plane (during the monoclinic-to-tetragonal change) in
the direction parallel to the surface it was deposited on, and the fact that the phase transition
of V O2 occurs over a small temperature variation, the performance of micrometer-sized
transducers based on this SPT could provide theoretical limits well beyond the traditionally
used mechanisms. More details about this are provided in the second chapter of this thesis.

1.1

Problem Description and Motivation

When a V O2 film is deposited over a released structure such as a micro-cantilever beam, the
resulting changes in film stress during the SPT cause very large deformations in the released
structure. These deformations can be exploited to create novel sensors and actuators. In
2010 our research group was the first to report the large deformations achievable in V O2 based bimorph microcantilevers. Nevertheless, several issues needed to be resolved before
V O2 -based actuation technologies became practical. Some of these problems included:
• The need for the characterization of the effects of deposition conditions on the properties of the V O2 film across its phase change.
• The deformation of released structures with V O2 films as well as the dynamics of these
deformations must be characterized and modeled.
• Different actuation methods need to be studied in order to determine their influence
and advantages for different applications.
• Fabrication processes that incorporate V O2 films must be developed.
2

This thesis not only addresses this issues, but also explores the difficulties and limitations
that come with the integration of V O2 thin films in MEMS and explores the memory effect
in the SPT of V O2 .

1.2

Thesis Statement

This dissertation consists of a series of studies that allow for the development of a fully
monolithic MEMS device capable of providing unprecedented capabilities in terms of total
actuator displacement. The main contributions of the present work include: the integration
of a non-standard MEMS material (V O2 ) in well established fabrication technologies; full
characterization of V O2 -based MEMS devices by using different actuation mechanisms; development of a fully monolithically integrated device that can exploit the properties of V O2
to achieve better performance than other available technologies; and the design of a control
system that increases the stability and usability of the developed MEMS device in practical
applications.
Thesis Statement: By characterizing and modeling the effects of the structural phase
transition of vanadium dioxide when applied to micro-electro-mechanical devices, we intend
to significantly expand the capabilities of transducers through the integration of phase-change
materials and the development of the necessary tools to cleverly design structures and optimize their performance.

1.3

Anticipated Research Contributions

This dissertation will directly address each of the research problems listed in Section 1.1.
A discussion of the deposition conditions and its effects on the SPT of V O2 will provide a
method for optimizing the properties of the V O2 film for different applications. By characterizing and modeling the deformation of released structures with V O2 films, it will be

3

possible to predict the behavior of SPT, thus allowing for the design of V O2 -based devices
and simulation of their expected response. By developing and optimizing actuation and detection methods that are compatible with a MEMS fabrication process that includes V O2 ,
a general architecture that can be used to design future V O2 -based MEMS transducers is
enabled. Finally, by incorporating the deposition and patterning of V O2 films into standard
fabrication processes, the design of more complicated MEMS devices is made.

1.4

Dissertation Outline

The remainder of this dissertation is organized as follows; In Chapter 2 the background
information on different MEMS actuating mechanisms, as well as the use of V O2 and the
phase transition, with particular emphasis on the SPT and its effects on the bending caused
V O2 films when deposited over micro-cantilever beams, is discussed in more detail. Chapter
3 discusses the general memory effect seen in the different properties of the material, and
how it can be exploited for different applications.
Up to Chapter 3 the discussion has only involved simple structures with no built-in
actuation method. Chapter 4 includes a discussion on the design and characterization
of a V O2 -coated cantilever beam with built-in micro electric heaters, and their potential
use as actuators. The focus of Chapter 5 is on the application of the memory effect in
the SPT, where a close-loop system that compensates for external temperature variations
and permits precise control of the programmed states is described. Chapter 6 presents
exploratory studies on possible ways to incorporate V O2 films into standard fabrication
processes. And finally, Chapter 7, presents conclusions, and summarizes the contributions
of this dissertation.

4

CHAPTER 2
BACKGROUND

In MEMS, actuators enable the interaction of devices with their surroundings, instead of
simply sensing it. Actuation mechanisms, which are very common at the macro scale (i.e.
with dimensions in the orders of meters), such as pneumatics, induction motors, and combustion engines are very difficult and inefficient to implement in the micro scale (i.e. with
dimensions in the order of 10−15 meters). Yet, actuation techniques such as thermal or
electrostatic actuation, which are mostly impractical at the macro scale, are ideal at the
microscopic level. Micrometer-sized electrostatic actuators use the attractive (or repelling)
force between two charged plates or surfaces. Electrostatic actuation has the benefit of a
very fast response and low power, but requires very large voltages (100V), and can produce
very little force (µN ). If lower voltages and higher forces are required, the best approach is
to use a thermal-based actuation mechanisms.

2.1

Thermal Actuation in MEMS

When a material is heated its volume changes, this is known as thermal expansion. In a
linear dimension, this can be expressed as:
∆L
= α∆T,
L0

(2.1)

where L0 is the original length, ∆L is the change in length, ∆T is the change in temperature, and α is known as the coefficient of thermal expansion (CTE). For most materials
α is positive. α can also vary with temperature, although it is mostly considered constant in
the temperature range of interest in MEMS. But most importantly, α is small. For silicon, a
change in temperature of 100 o C only produces a 0.03% change in size. This means that for

5

this mechanism to be useful, the deflection must be amplified in some way. There are three
main ways this is accomplished in MEMS:
1. The geometry of the structure and its constraints are chosen so that a small expansion
in one direction translates to a large deflection in another. (Section 2.1.1)
2. The device is designed in such a way so that the temperature of one part of the structure
is different from the other. (Section 2.1.2)
3. Materials with different CTEs are employed. (Section 2.1.3)

2.1.1

Geometrically constrained devices

One of the most common actuator types in this category is known as the bent-beam actuator
[17, 1]. This actuator consists of a clamped-clamped bridge, shaped as a “V” made from a
single material. An example of this type of actuator is shown in Figure 2.1.

Figure 2.1 Example of an array of bent-beam actuators [1].
The beam is usually heated uniformly by passing a current through it. This will cause
joule heating and thermal expansion, which moves the “apex” of the beam. The total amount
of force is dependent on the dimension of the beams, the angle of the “V”, and the material’s
mechanical properties. As an example, a 400 µm long, 4 µm wide, and 4 µm thick polysilicon beam, with an incline gradient of 0.1, has a total displacement of 3 µm when heated
600 o C. That actuator has a maximum blocked force of 800 µN . That force is about 100
times larger than that of electrostatic comb-drives of similar dimensions [17].
By connecting these beams in parallel, as shown in Figure 2.1 it is possible to increase
the total force; and by connecting them in cascade as shown in Figure 2.2, it is possible to
6

amplify the total deflection. The structure shown in Figure 2.2 has a 4X increase in total
deflection.
Primary Unit
(With Current)

Secondary Unit
(No Current)

L2
Secondary
Tip Motion
Primary Unit
(With Current)

Figure 2.2 Example of a cascade configuration which quadruples the total deflection.

However it is important to note that these results are obtained by heating the devices by
600 o C, near the maximum temperature poly-silicon can sustain before self-annealing. This
translates to large power consumption (221 mW in the device discussed in [17]). And even
in the cascaded configuration the total deflection is less than 20 µm for a total device size
in the mm range.

2.1.2

Pseudo bimorphs

If more deflection per size is required with less power consumption, another solution is to
use pseudo bimorph structures. In theses structures the thermal and electrical properties
of the device are altered in different locations in order to induce a temperature difference
between different parts of the structure. This temperature difference will cause one part of
the structure to expand more than another, causing bending. One very common geometry
is shown in Figure 2.3.
In this asymmetrical structure, the width of one “arm” is larger, reducing the total
resistance of that arm with respect to the other, and increasing the surface area (which
translates into an increase of the heat dissipation in that arm). The end result is that when
7

+

"Hot" arm

Contact
pads

Direction
of motion

"Cold" arm

Figure 2.3 An example of a pseudo-bimorph amplification of thermal expansion.
a current is passed through the structure, one arm will heat up (and consequently expand)
more than the other. This will cause bending in the direction shown.
This type of device is discussed and tested in [16]. In this work, devices with lengths of
150, 200, 250, 300 µm were tested. For different conditions of “arm” widths, the maximum
deflection obtained was 22 µm for the 300 µm device, while dissipating 42 mW of power.
This type of thermal device is clearly capable of more deflection per power and size, but the
tradeoff is force. The largest blocked force reported for these actuators was 19 µN . Similar
to the bent-beam actuators, it is possible to connect these actuators in parallel to increase
the total amount force. Because of the small value of the CTE in most materials, it is still
necessary to heat these actuators to large temperatures, which still dissipates large amounts
of power.

2.1.3

Bimorph Cantilevers

If even more deflection is desired, one approach is to increase the difference in thermal
expansion (i.e. increase ∆α), by combining materials with very different CTEs. For example
the classical bimorph cantilever (also known as a bi-metalic strip) as seen in Figure 2.4.
Here a cantilever beam, composed of two materials with very different CTEs, is heated.
One of the layers will expand at a different rate from the other, creating a stress gradient
that causes the cantilever beam to bend. It is possible to achieve very large deflections if the
proper materials are chosen. For example in [2], a 300 µm long silicon cantilever was coated
with a 200 nm thick layer of plasma-polymerized polystyrene (PS). When heated from 20 to
40 o C the cantilever beam deflected by almost 50 µm (see Figure 2.5)
8

Direction
of motion

Material with higher coefficient
of thermal expansion

Figure 2.4 An example of a bimorph cantilever.

Figure 2.5 bimaterial cantilever bending upon incident heat, side view of the polymer-silicon
micro-cantilever bending as temperature increases from 20 to 40 o C, [2].
This device, is a significant improvement to the other to methods presented in terms
of total deflection, since it can produce very large deflections for very small temperature
increases. It shows a thermal sensitivity in the order of 1 nm/mK which is 100 times larger
than other metal/ceramic bimorph cantilevers and 1000 times larger than the geometrically
constrained device discussed in Section 2.1.1. However, these types of actuators are made
of polymers, which usually have a very low elastic modulus, and therefore their work capacity
is limited.
It is possible to create bimorph structures with materials that rely on other physical
effects besides thermal expansion. One example is piezoelectric ceramics. In [18], PZT/AgPd
bimorph disks with a diameter of 25 mm could produce 75 µm of deflection under an applied
voltage of 500 V. However in general piezoelectric materials produce very small deflections

9

per actuator size and require very large voltages (100s of volts). Therefore their performance
can not really be compared to the thermal bimorph structures we are discussing here.

2.2

Phase Change materials

Phase change materials make use of changes in the size of their crystalline structures during
a phase transition. The type of materials that are of interest in the use of actuators are
usually materials with solid-solid or martensite type transformations. When these materials
undergo their phase transition their crystalline structure changes, and with these changes
most of the materials properties change as well. Two popular examples of phase change
materials are GeSbT e and shape memory alloys (SMA).

2.2.1

GeSbTe

GeSbT e is a phase change material commonly used in re-writable compact discs (CD-RW
or DVD-RW). Its crystalline structure can be reversibly changed between amorphous and
crystalline states by sequences of heating and rapid cooling. These two phases produce
big differences in optical reflectivity [19, 20] and electrical resistivity [21, 22]. But more
importanly for MEMS actuator applications, wafer curvature measurements have shown
large stresses in the GPa range during the transition [23]. These generated stresses can be
used to create a bimorph type actuator. For example in [24], Amorphous Ge2 Sb2 T e5 films
were deposited on the SiN cantilevers measuring 50 µm in length. A laser was used to heat
the cantilever, this caused a crystallization of the film which caused a change in volume and
the cantilever deflected by 8.4 µm. However even though the transition could theoretically
be reversed, this requires raising the temperature to the melting point of the film, followed by
a rapid cooling usually in the nanosecond range. This rapid cooling is not feasible in micro
actuators. This also makes the transition discontinuous, in other words it becomes very
difficult to control the deflection of the cantilever at an intermediate value of the transition.

10

2.2.2

Shape Memory Alloys

Another popular phase change material is SMA (Shape Memory Alloy). SMAs can also
undergo a thermally induced crystalline transformation between a low temperature martensitic phase and a high temperature austenitic phase. During the low temperature phase the
material is highly ductile and can be easily deformed. When heated the material strength
increases and it recovers its initial shape. Ni-Ti based SMA have been shown to recover
up to 8% strain or produce up to 500 MPa stress when constrained [25]. Since the SMA
transition is ”one-way”, in other words the material can only return to the high temperature
state, some bias element is required to deform the material to the cold temperature state
[26, 27]. Currently SMA have the largest strain energy density as will be shown in Section
2.4. It was also shown the SMA transition also occurs at the microscopic level [28]. Figure
2.6 shows a 3 µm long T i5 N iCu/PT bimorph cantilever when heated it exhibited a 230 nm
deflection [3].
In [3] the Pt film served as the bias element (a form of spring) to restore the cantilever
deflection to its cold state. Heating was performed using photo-thermal actuation.
It is important to note that each of these materials brings their own complications when
used in MEMS devices. For example, piezoelectric ceramics require extremely large voltages,
and although they have the potential to create large forces, they create very small displacements –millimeter size structures are needed to produce 10s of µm deflections. SMAs can
produce displacements in the required scale, but their integration in MEMS involves very
complicated fabrication processes. Furthermore, the phase transition in SMAs is not fully
reversible and starts fatiguing (dependent on the magnitude of the applied stress) after a
few hundred cycles [29].

11

Figure 2.6 0.9 µm3 working composite shape memory actuator. On thermocycling the actuator exhibits reversible free end deflection of 230 nm, [3].

2.3

The Structural Phase Transition of Vanadium Dioxide

Another phase-change material that has proven to be very useful is V O2 . As was briefly
described in Chapter 1, V O2 is a multifunctional material with a transition temperature at
around 68 o C. The crystal structure of V O2 is monoclinic (designated M1 ), as its temperature rises above 68 o C the structure changes to its tetragonal phase,(rutile type, designated
R), by slight reordering of the vanadium ions. This is known as a Structural Phase Transition (SPT) Figure 2.7 [4]. This SPT is strongly coupled to the IMT in the material. It is
important to note that vanadium has 4 stable adjacent oxidation states; V O, V2 O3 , V O2 ,
and V2 O5 . There are also various other distinct phases of vanadium oxides, for example
V6 O13 . Some of these oxides also exhibit solid phase transitions, but they either occur at
very low temperatures (-123 o C for V6 O13 [30]), or very high temperatures (280 o C for V2 O5
[31]). The fact that the transition temperature of V O2 is so close to room temperature, is

12

what makes the material so useful for transducer applications.

Figure 2.7 Crystal structure of the monoclinic, low-temperature phase (left) and the rutile,
high-temperature phase (right). Vanadium atoms are depicted in red and oxygen atoms in
a lighter color. Because of a symmetry-breaking pairing of vanadium atoms (violet lines),
the axis definitions change [4].
Using the lattice parameters for both phases of V O2 (and noting that the unit cell in the
monoclinic phase corresponds to two unit cells in the tetragonal phase), it can be calculated
that the volume of the V O2 unit cell increases during this SPT. Although the overall volume
increases during the phase change (that is, during the heating cycle), the areas of some planes
are reduced [32]. When polycrystalline V O2 films are deposited by pulsed-laser-deposition
(PLD) over amorphous SiO2 or single crystal silicon (SCS), the material tends to orient
itself with its (011) and (110) planes parallel to the substrate. The area of the plane for the
tetragonal phase is smaller than that of the monoclinic phase. Thus, when SiO2 or SCS
microcantilever structures are coated with V O2 films, the structure bends concave toward the
V O2 film coating during the phase change. ”It can be readily calculated that the area of the
plane parallel to the substrate shrinks by 1.7% which results in a theoretical maximum strain
of a perfectly oriented and unrestricted film of approximately -0.8%. However, since the film
is polycrystalline, and the orientation of every grain is not exactly in the same direction
and the film is partially restricted by the substrate, which is precisely what generates the
bending, the measured strain was closer to -0.3% [5].
In a preliminary observation [5], the V O2 /Si bimorph cantilevers exhibited very large
reversible changes in curvature through the SPT. The superimposed pictures in Figure 2.8,
for a 350 µm long cantilever, were taken through a low-power microscope as the temperature,
13

measured with a thermocouple attached directly to the chip, was raised. Cantilever bending
toward the side with the deposited film evidences increasing film tension. As the temperature
was then reduced, the reverse response was observed, with some noticeable hysteresis.

Figure 2.8 Multiple exposure picture of V O2 -coated silicon cantilever during heating [5].
If the amount of deflection per actuator size is compared, V O2 -based actuators can produce twice as much deflection per actuator size compared to the best case polymer/ceramic
bimorph actuator. In terms of sensitivity to temperature, V O2 /Si bimorph cantilevers
demonstrate an extremely high rate of change in curvature of up to 485 m−1 K −1 near
the transition temperature, which is almost an order of magnitude higher than for the best
cases reported for polymer-coated or SMA-coated cantilevers [5].

2.4

Strain Energy Density

Strain energy density is a value that can be used as a figure of merit to compare microactuators based on different materials or physical phenomena. It is a value that describes
the capability of the actuator to do work, in terms of the actuator size; or, in other words,
it normalizes the work that can be generated by the micro device per unit size. in [25], the

14

Strain Energy Density for different materials was studied. These results are summarized in

6

1 0

S o
E n e r g y D e n s ity ( J m
lid
- li N
q
V a u id i- T i
S
n
T h a d tra n M A
iu
e
s
T h r m o m d itio n
e r - p io x
ne
m
id
E l ale um e
ec xp a
tr o a tic
-m n s i
E l ag on
ec ne
P i tr o tic
ez st
o e a ti
le c
ct
M
r
ic r M u ic
s
ob
u b c le
bl
es

-3

)

Figure 2.9, together with a V O2 -based microactuator studied in [6].

1 0

5

1 0

4

1 0

3

1 0

2

Figure 2.9 Strain energy densities of different micro-actuators including the V O2 -based
micro-actuator studied in [6] (green).
Even though V O2 -based actuators do not exhibit the highest energy density, it does
posses some advantages over its closest competition. In terms of strain energy density,
electroactive polymers have demonstrated strain energy densities of 5 × 104 J/m3 , and
thermal expansion mechanisms have shown values in the range of 105 J/m3 . Shape memory
alloys (SMAs) have shown the largest strain energy densities (3 × 106 J/m3 ) [25], while
V O2 -based actuators follow with reported values of 8.1 × 105 J/m3 [6] (6.3 × 105 J/m3 for
initially curved actuators [33]). SMAs are made of ductile metals (typically titanium, nickel,
and copper) and are thus more susceptible to plastic deformation, creep, oxidation, and
fatigue. In contrast, the bonding in monoclinic V O2 is ionic, and the compound has little or
no ductility. Also, typical SMAs need a wider temperature variation for inducing the phase
change, lose their memory capability, and are currently limited to relatively low operating
frequencies [33, 25]. Long term studies on the efficiency and fatigue of V O2 -based devices
have not been performed, but the actuators discussed in Chapter 4 of this dissertation have
been operated for thousands of cycles at full deflection, without any noticeable evidence of
15

degradation.

2.5

Summary

This chapter discussed the different actuation mechanisms used in MEMS. Thermal based
actuation has proven to be the most practical for large deflection and force. Compared
to other forms of actuation, such as electrostatic and piezoelectric, it operates with much
lower voltages that are compatible with most electronic devices. The different techniques
used to amplify deflection where discussed, of which bimorph structures produced the most
deflection per actuator size. Finally, the effects of choosing the right materials in actuators
was shown to be critical in the design of efficient bimorph actuators.
Of the many materials that can be used in bimorph structures, V O2 offers many advantages over current state-of-the-art micro-actuation technologies, as the results in this thesis
show. Some of those advantages include high sensitivity, relatively large strain energy densities, and lower power consumption than typical thermal actuators. During the rest of this
dissertation the properties and performance of V O2 -based devices will be discussed as well
as the possibility of incorporating V O2 films in MEMS fabrication processes.

16

CHAPTER 3
THE MEMORY EFFECT IN VANADIUM DIOXIDE

The changes in the properties in V O2 during its phase transition show hysteretic behavior.
This hysteresis allows for the creation of systems with multiple stable outputs for one input.
This endows the material with memory capability. Because a similar hysteretic behavior is
seen in all of the material properties that change during the transition, different types of
memories (e.g. optical, electrical, mechanical) can be created. Furthermore, the intrinsic
multi-functionality of this material allows the use of different ’writing mechanisms’, the
most commonly used being thermal. In this chapter, the memory behavior in the optical,
electrical, and mechanical properties of V O2 is demonstrated.
The experiments described were performed with three different samples coated with V O2
thin films: 1. a SiO2 microbridge (used for experiments in which only resistance was measured) 2. a flat SiO2 glass substrate, used for experiments in which IR optical transmittance
was measured, and 3. a single crystal silicon (SCS) microcantilever used for measuring the
deflection. The SiO2 bridge was fabricated by standard lithographic techniques as described
in [34] and had length, width, and thickness of 200 µm, 35 µm, and 3.3 µm, respectively.
The work related to this chapter was published in [7, 8].

3.1

Electrical Memory

The experiments described in this section were performed on the SiO2 microbridges described
before. The microbridges were kept under vacuum ( 10 mTorr or less) in order to reduce
convective heat losses. The back side of the substrates (opposite to the side with the V O2
thin film coating) was cemented with silver paste to a heater integrated in a sample holder.
Thermocouples were directly attached to the samples, and read by temperature controllers
feeding the heaters. To allow film resistance readings, gold wires leading to the vacuum

17

chamber electrical feedthroughs were attached to each of the two bridge anchors, allowing
current flow along the bridge.
In order to characterize the hysteresis of the sample, The resistance was measured while
employing computer-controlled heating-cooling cycles, were the breadth of the cycles were
progressively narrowed. This resulted in the family of curves shown in Figure 3.1. These
curves can be used to predict the ‘’path” the resistance will follow given a known temperature
variation.

1 M
o

R e s i s t a n c e ( Ω)

6 4

C

1 0 0 k

1 0 k

1 k
3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

o

T e m p e ra tu re ( C )

Figure 3.1 Series of hysteresis curves for V O2 -coated microbridge sample. Interior curves
are obtained by cooling before the end of the transition region is reached [7].

The memory effect only exists within the hysteresis region of the transition. Therefore,
in order to characterize the memory of the film, the sample needs to be kept at a background
temperature of 64 o C. This was done using the temperature controller. This temperature,
about half way through the transition will be reffered to as the “operating temperature”. It
is important to note that this is the temperature of the sample, which is measured far from
the microbridge. Therefore, if the microbridge is locally heated (its localized temperature
increases), the rest of the sample will remain at 64 o C, and once that temporary source
of heat is removed the localized temperature of the microbridge will return to 64 o C. The

18

doted line in Figure 3.1 shows that the measured resistance at 64 o C can have any value
from 95 kΩ to 4 kΩ depending on the ’history’ of the sample temperature. In other words
the same heat input could result in two different values of resistance; depending on what
the initial resistance value was (i.e. depending on what was the resistance state prior to the
pulse). In this experiment the temporary heat source (actuation method) is the increase in
temperature of the sample due to absorbed photons (i.e. photothermal actuation).
In this work, photothermal actuation is achieved through laser pulses. A current-controlled
continuous diode laser with 672 nm wavelength and a maximum power output of 300 mW
is used to illuminate the samples. The output power was calibrated with an optical power
meter (Thorlabs PM100D); and an electronic shutter (Uniblitz T-132) was used to control
the duration of the pulses. The laser was focused to illuminate an area of the sample on
which the resistance measurements were performed. Figure 3.2 shows the change in resistance due to a light pulse irradiating the sample for 2 s, after it was heated from 30 o C to
64 o C.

1 M
o

C

B e fo r e P u ls e

1 0 0 k

A fte r P u ls e
1 0 k

R e s i s t a n c e ( Ω)

R e s i s t a n c e ( Ω)

6 4

9 0 .0 k
6 0 .0 k
3 0 .0 k
0 .0

1 0

2 0

3 6

4 2

1 k
2 4

3 0

3 0
T im e ( s )

4 8

4 0

5 0

5 4

6 0

6 6

7 2

7 8

T e m p e ra tu re (oC )

Figure 3.2 Transient of the resistance to a 2 s light pulse. The inset shows the response in
the value of resistance when a 2 s red (λ = 672 nm) light pulse illuminates the sample. The
bolder trace in one of the hysteresis curves suggests the thermal evolution of the sample [7].

The resistance drops rapidly while the pulse lasts, then partially recovers more slowly,
19

but it does not follow the same path during the duration of the pulse. It stabilizes at a value
an order of magnitude lower than the initial resistance. The dynamic is suggested by the
highlighted hysteresis trace in Figure 3.2, which was chosen to connect the initial, lowest,
and final resistance values measured before, during, and after the laser pulse, respectively.
The behavior shown in Figure 3.2 suggests that the final value of the resistance is only
dependent on the maximum temperature the sample temporarily reaches. This can be
confirmed by adjusting the duration of the light pulses. In Figure 3.3, all light pulses had
the same intensity as before, but durations were chosen to produce roughly similar resistance
drops.

1 0 0 .0 k

R e s i s t a n c e ( Ω)

8 0 .0 k
6 0 .0 k

3 m s
2 s

2 0 0 m s

4 0 .0 k
2 0 .0 k

0

6 0 0 m s
2 0

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

T im e ( s )

Figure 3.3 Resistance as a function of time as a series of four light pulses with increasing
duration illuminate the sample [7].
Starting from the initial condition, as shown in Figure 3.3, the magnitude of the drop in
resistance increases with the energy of the laser pulse. In this case, the energy of the pulse is
increased by increasing its duration. It is noted that the single 2 s pulse (inset Figure 3.2)
causes nearly the same resistance drop as that due to the cumulative effect of all the pulses
recorded in Figure 3.3, of which the last pulse is also 2 s long. This suggests that the final
resistance value, after the pulse, is not a function of the current value, but only a function

20

of the largest temperature achieved during the temperature excursion.

3.2

Optical Memory

This experiment was also performed under vacuum (10 mTorr or less) in order to reduce
convective heat losses. The sample was cemented first to a thin copper plate with a small
(4 mm diameter) circular aperture. This copper plate was cemented in turn to the heater.
The light source for the transmittance measurements was a continuous IR laser diode (Mistubishi ML725B8F, 1.31 µm wavelength, 10 mW rated power) with collimating lens, mounted
in a thermoelectric cooler base and driven at constant temperature and power by a laser
diode driver (Melles-Griot DLD203). A light detector with high IR sensitivity (Thorlabs
PDA10CS) was positioned directly opposite. For the transmittance measurements the setup
was calibrated with a similar uncoated substrate in place of the sample. The actuation laser
system was the same used in Section 3.1. Because the area illuminated by the heating laser
did not fill the area sampled by the IR laser, this procedure is adequate for measurements of
changes in average transmittance in a fixed setup, which is sufficient for the purpose of this
work, but would be inadequate for absolute measurements.
The hysteresis curve for sample transmittance at λ = 1.31µm is shown in Figure 3.4.
For this particular sample the hysteresis loop is nearly double that of the sample used in
Section 3.1. Because of the change in the shape of the hysteresis curve a temperature of
70 o C was chosen as the “operating temperature”.
Figure 3.5 shows the effect of a sequence of pulses with equal irradiance and increasing
duration, starting with a 0.6 s pulse and increasing by 0.6 s for each successive pulse. By the
end of this sequence, even though the pulse energy continues to increase, the steps become
smaller as the value of transmittance approaches the lower limit measured before. As in
Section 3.1, in which light pulses were used for heating the sample, it was found that
approximately the same transmittance reduction can be achived with a succession of many
pulses of increasing duration ,that with a single pulse with the same duration as the last
21

(% )
T r a n s m i t t a n c e a t λ= 1 . 3 1 µm

4 0

7 0

5 0

o

C

3 0
2 0
1 0
0
3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

T e m p e ra tu re (oC )

Figure 3.4 Hysteresis curve for transmittance at λ = 1.31µm for V O2 -coated flat sample [7].
in the sequence, as long as the device is allowed to ’settle’ between pulses and the lower

T r a n s m i t t a n c e a t λ= 1 . 3 1 µm

(% )

transmittance limit has not been reached.

4 0
3 5

0 .6 s
1 .2 s

3 0
2 5
2 0

0 .6 s In c re a s e
1 5
1 0
0

6 0

1 2 0

1 8 0

2 4 0

3 0 0

3 6 0

4 2 0

4 8 0

T im e ( s )

Figure 3.5 Transmittance measured at λ = 1.31µm for the V O2 -coated flat sample after
it was illuminated with a series of red (λ =672 nm) light pulses with equal irradiance and
increasing duration, starting with a 0.6 s pulse and increasing by 0.6 s each time [7].

In order to verify the stability of the systems described, the setup for this last case was
used to measure the IR transmittance continuously for 8 h, after the sample was brought up
22

to 70 o C, subjected to a single light pulse during which transmittance was lowered from 38.5%
to 25.0% and maintained at the same temperature. During this time the transmittance value
slowly decreased to 24.8% by the time the experiment was ended. This slight reduction was
attributed to instrumental drift in the temperature control or IR emission-detection system.
A second experiment was performed with this sample in order to demonstrate the capability to combine the electrical and optical properties of the V O2 film. Voltage pulses
were applied through wires directly connected to the V O2 film as the transmittance was
measured. As shown in Figure 3.6, a sequence of pulses with increasing duration but equal
voltage (80 V in this case), produces successive drops in sample transmittance. As in the
previous case with the resistance measurements, there is some overshoot of the transmittance value just as the pulse ends, associated with thermal excursion in the region excited.
The transmittance recovers rapidly, however, and remains at the,new level when the local
temperature stabilizes. The second trace in Figure 3.6 was obtained by applying pulses of
equal duration (1 s) but increasing voltage.

Figure 3.6 Sample transmittance at =1.31 µm for the V O2 -coated flat sample in response
to a series of 80 V electric pulses with increasing duration lower curve and to a series of
increasing voltages with equal 1 s duration upper curve [7].

23

3.3

Mechanical Memory

The samples in this section consist of a V O2 film deposited over commercially available silicon
micro cantilever chips (MikroMasch). As specified by the manufacturer, the cantilever used
for the detailed measurements had length, width, and thickness of, respectively, 350, 35, and
1 µm, and a spring constant of 0.6 N/m. Figure 3.7 shows an SEM image of the cantilever
chip.

Figure 3.7 SEM image of the commercially available silicon micro cantilever chips (MikroMasch). The center cantilever is 350 µm long.
Figure 3.8 shows a schematic of the experimental set-up used for this experiment. The
entire setup is constructed on an optics air-suspended table in order to reduce vibrations. In
this setup, a red laser (L1; λ = 672 nm, intensity = 500 µW , spot size 35 µm), was focused
and aimed at the tip of the V O2 -coated Si cantilever. This laser was used to measure the
deflection of the cantilever beams, and remained on during the experiment. The reflected
beam is then aimed at a position sensitive detector (PSD) (Hamamatsu S3270). As the
cantilever moves the PSD will produce a voltage proportional to the displacement, using the
initial position of the cantilever as a reference. The second laser (L2; λ = 672 nm, spot size
100 µm) was centered at 100 µm from the cantilevers anchor. This laser was used to heat
the cantilever by computer controlled optical pulses of varying intensity, but of an equal
duration of 1 ms. These pulses programmed the different mechanical states. The intensity
24

of the lasers was measured with an optical power meter (Thorlabs model: PM100D). L1 and
L2 are arranged orthogonal to each other so that the light from L2 does not affect the PSD
reading, which relates only to the beam reflected from L1. The voltage output from the PSD
was amplified by a DC coupled amplifier and filtered for high-frequency interference. All the
experiments were done at ambient pressure.

Figure 3.8 Measurement set up. The entire system was in ambient pressure. As shown in the
insets, the measuring laser (L1) was focused at the tip of the cantilever, while the heating
or programming laser (L2) was focused close to the center of the cantilever beam [8].
Similarly to Section 3.1 and 3.2 the first step was to characterize the transition, in
order to determine the ’operating temperature’. The cantilever deflection was monitored
while a PID-controlled Peltier heater (Custom Thermoelectric 03111-9L31-04CG) was used
as the heat source - L2 was not used. The temperature was increased in intervals of 0.5 o C,
and the measurements were taken after the set temperature was reached and stabilized. The
result shown in Figure 3.9.
The V O2 -coated cantilever was bent downwards at room temperature. As the sample
was heated through the SPT, the area of the V O2 plane parallel to the Si surface was

25

reduced, causing the cantilever to bend upwards (See Chapter 2). The heating/cooling
curves showed a rather broad hysteresis for V O2 . However, this was found beneficial for this
particular application, since broad hysteresis allow for larger number of memory states. The
PSD voltage at 30 and 100 o C correspond to the minimum and maximum deflection of the
tip of the cantilever, respectively, taking a perfectly flat cantilever as a reference (0 µm).
With this information the relationship between PSD voltage output (left y-axis in Figure
3.9) and the cantilever tip deflection (right y-axis Figure 3.9) was determined.

3 .0

4

6 0

2 .5

o

C
0

-4

1 .5
1 .0

-8

0 .5

-1 2

0 .0
-1 6

-0 .5
-1 .0

-2 0

H e a tin g
C o o lin g

-1 .5
-2 .0
3 0

4 0

5 0

6 0

7 0

8 0

9 0

D e f l e c t i o n ( µm )

D e fle c tio n V o lta g e ( V )

2 .0

-2 4
1 0 0

T e m p e ra tu re (oC )

Figure 3.9 Deflection of the cantilever as a function of temperature. A relatively wide
hysteresis width for V O2 thin films of 30 o C can be noticed. The left y-axis represents the
voltage reading form the PSD, and the right y-axis is the cantilever deflection [8].
In order to obtain more mechanical states, laser pulses with the minimum power that
caused a detectable deflection were used in this experiment. Since the minimum intensity
from L2 (3 mW) was already beyond this minimum intensity value, optical attenuators had
to be placed between L2 and the sample. The minimum power of a 1 ms pulse that caused a
detectable deflection on the cantilevers was found to be 1 mW. The cantilever was exposed
to a sequence of pulses with powers ranging from 1 to 10 mW (the full range of L2 after the
optical attenuators) with increments of 1 mW in order to measure the different mechanical
states. The resulting deflection is shown as the blue trace in Figure 3.10.
26

1 .0

V a r y in g P o w e r
S in g le P o w e r
In c r e a s in g b y 1 m W

-6
-8
-1 0

0 .5

4 m W
3 m W

0 .0

-1 2

4 m W

-1 4

2 m W
-1 6

1 m W
-0 .5

D e f l e c t i o n ( µm )

D e fle c tio n V o lta g e ( V )

1 .5

0

-1 8
2 0

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

T im e ( s )

Figure 3.10 Non-uniform mechanical states found from linearly increasing light pulses (blue
trace) and for five consecutive light pulses of 4 mW (red trace) [8].
As can be seen, the deflection changed with every pulse, and the new value (i.e. new
mechanical state) was sustained until the next pulse started. Of course, this occurred because
the “operating temperature” was sustained. It can be noticed in Figure 3.10 that the initial
deflection (18.5 µm) is very close to the deflection of the major heating curve value at the
pre-heating temperature (see Figure 3.9). However, the deflection of the final mechanical
state (-4 µm) is far from the major cooling curve value at the pre-heating temperature.
Furthermore, the change in displacement saturates as -7 µm is approached. As it is explained
below, this should not be interpreted as the maximum programmable mechanical memory
state being -7 µm.
The bending profile of the two actuation mechanisms (Peltier heating and laser pulse
heating) cannot be compared. In the Peltier heating method, the entire sample was in contact
with a uniform heat source. The cantilever bending profile in this case had the form of an arc
with a radius of curvature that should be similar for V O2 coated cantilevers with different
lengths but otherwise identical. On the other hand, in the laser pulse heating method, the
illuminated area of the cantilever was heated much more than the rest. Consequently, the

27

area of the V O2 film that transitions from the monoclinic phase to the rutile phase is much
smaller for the laser heating method. In this case, the shape of the bent cantilever cannot be
best described by an arc. Instead, the bending profile of the laser actuated cantilever had the
form of two straight lines joined by an arc. It was also found in a separate experiment that the
location of the L2 spot along the cantilever’s length defined the cantilever’s inflection point.
The difference in bending profile between the two heating methods, and the dependence of
the total cantilever deflection with the location of the heating laser spot are reasons why the
largest deflection obtained with the Peltier heating method (-1 µm) should not be expected
when the laser pulse heating method is used. For the programming experiment, the location
of L2 (100 µm form the cantilever’s anchor), was found empirically to be the location that
resulted in the largest deflection.
After these measurements, the cantilever was ’reset’ for the next experiment, which consisted on exposure to five pulses of the same amount of power of 4 mW (Figure 3.10 red
trace). As can be noticed, this resulted in a displacement (or mechanical state) close to the
value obtained from the previous experiment after a sequence of 1, 2, 3, and 4 mW laser
pulses. This observation suggests that the final value of deflection will be mainly determined
by the energy of the pulse. In other words, a specific amount of energy delivered to the
cantilever represents a specific mechanical state, which can be stored by traversing a single
set or different sets of minor heatingcooling loops.

3.4

Summary

The memory behavior in the optical, electrical, and mechanical properties of V O2 under
photothermal actuation, was studied. Through all three experiments, it becomes evident
that in order for the memory behavior to be useful in practical devices, it is necessary to
very precisely control the temperature of the samples. Due to the abrupt transition in
V O2 , small variations in temperature will cause the ’state’ of the material to change. It
is also important to notice that the changes in states are very sensitive to the input pulse
28

energy. Without very precise control of these input pulses it becomes impossible to reliably
and repeatably program the ’state’ of the V O2 film. Finally, in all three experiments the
material’s memory is erased (i.e. reset) by cooling the entire sample to room temperature. In
order to program new states, the sample needs to be taken to the “operating temperature”
first. Using an external heater for reaching this ’pre-heating’ temperature makes the process
extremely slow and energy inefficient. The next chapters will discuss methods that will
eventually fix (or atleast allevaite) all of these deficiencies by incorporating a micro heater
into the MEMS device, and developing a control system that will improve the programing
of a V O2 -based memory.

29

CHAPTER 4
EMBEDDED ELECTRIC MICRO-HEATERS

The previous chapters have presented a detailed discussion of the V O2 properties, with
particular emphasis on the effects of its SPT in MEMS. However, the actuation mechanisms
discussed so far include photo-thermal and heat conduction mechanisms that are not very
practical for many applications. This chapter presents the development of a V O2 -based
MEMS device (a micro-actuator) driven by electric signals through a micro-electric heater
(i.e. driven by joule heating). The dynamics of this V O2 -based MEMS device and its design
parameters are discussed. Also, a Finite Element Method (FEM) model that integrates the
SPT effects into a lumped model is discussed and validated. This model will facilitate the
design of future devices by enabling a method to simulate the effects of the SPT. The work
related to this chapter was published in [9].

4.1

Micro-Cantilever Design and Fabrication

The developed V O2 -based MEMS actuators are composed of SiO2 with integrated titanium/platinum (Ti/Pt) metal layers, where the Ti layer is only used for adhesion purposes.
The first layer of Pt forms the contact pads and traces necessary for either wire bonding or
probe electrical connections (wire bonding was used in this work). It also forms a bottom
electrode, which could be used for capacitive detection. The second Pt layer forms the integrated heater and provided a reflecting surface, necessary for optical detection (the SiO2
cantilever is transparent to the wavelength used for optical detection (λ=632 nm)). This
second layer of Pt is buried inside the SiO2 cantilever, electrically isolating the heater from
the V O2 film. This isolation is necessary to avoid a short circuit across the resistive heater
when the V O2 becomes conductive in the metallic phase.

30

Figure 4.1 General fabrication process flow. (A) Bottom electrode deposition and patterning
by lift-off, (B) Sacrificial layer (amorphous silicon) deposition and patterning by RIE, (C)
First SiO2 deposition and patterning, (D) Top electrode deposition (Heater), (E) Second
SiO2 deposition, (F) Patterning of the second SiO2 layer, (G) Release of the MEMS actuator,
and (H) V O2 -deposition by pulsed-lased deposition using in-situ shadow mask [9].
Figure 4.1 shows a general overview of the fabrication process. The starting substrate
consisted on a silicon wafer with a 1.2 µm thick layer of silicon dioxide (SiO2 ) deposited by
PECVD. A Ti (50 nm)/Pt(150 nm) layer was deposited by evaporation and patterned by liftoff. The sacrificial layer (500 nm thick a-Si film) was deposited by LPCVD and patterned by
RIE. A second thin layer of SiO2 (500 nm) was deposited by LPCVD (in order to minimize
voids in the cantilever’s structural material) and patterned using plasma etching. A second
metallization Ti(40 nm)/Pt(150 nm) followed, this time by sputtering (to increase the film’s
conformity). This second metal layer was also patterned by using lift-off, and it formed
the integrated resistive heater. The final layer of SiO2 (500 nm) was deposited again by
LPCVD and patterned using plasma etching. The wafer was then diced and the structures
were released using XeF2 , which is a dry (gas, not plasma) etching method.
The final step in the device fabrication is the deposition of the V O2 thin film (200 nm
thick), which was done using a shadow mask to guarantee that V O2 was only deposited
on the area of the chip that contains the suspended cantilever beams and away from the

31

R e s i s t a n c e ( Ω)

1 0 M

1 M

1 0 0 k

T tr= 6 8

H e a tin g
C o o lin g

1 0 k
3 0

3 5

4 0

4 5

5 0

5 5

6 0

o

C

6 5

7 0

7 5

8 0

T e m p e ra tu re (oC )

Figure 4.2 Resistance as a function of temperature for the V O2 film used. The V O2 film
was a circular pattern of 6 mm in diameter, and the probes were 2 mm apart [9].
contact pads. Film quality was verified by measuring the change in resistance of the V O2
film across the IMT, which showed the typical hysteretic behavior of (see Figure 4.2). The
finalized device was wire bonded to an IC package for testing. Figure 4.3 shows a scanning
electron microscope (SEM) image of the two tested cantilevers with their dimensions. The
final thickness of the cantilever is approximately 1 µm.

Figure 4.3 SEM image of cantilevers with dimensions (top view) [9].

32

4.2

Finite Element Model

Although the simulation of the structural changes in a V O2 crystal could be addressed by
modeling of the lattice changes of V O2 single crystals during the transition, the case for
polycrystalline V O2 is more complicated. Polycrystalline V O2 consists of multiple layers of
V O2 crystals of different sizes, which do not experience perfectly synchronized structural
changes with temperature, and which are not perfectly oriented the induced strain during
the phase change is not perfectly uniaxial. Additionally, the modeling of the shift in the
transition temperature (TT r ) of V O2 with stress is no longer possible by using the uniaxial
stress obtained from the Clausius–Clapeyron equation as it was demonstrated for singlecrystal V O2 [35].
Therefore, experimental data was used in this work to generate a comprehensive lumped
model for V O2 -based MEMS actuators, by fitting a temperature dependent thermal expansion coefficient for V O2 that emulated the strain induced in the micro-actuator during the
phase change. The FEM model was also used to simulate the heat losses during actuation
pulses, the design of the resistive heater, and the power dissipated in the heater.
COMSOL Multiphysics was used to create such FEM model. The model includes two
coupled physics: solid mechanics with thermal expansion, and heat transfer with joule heating. The dimensions of the simulated structures are those of the tested devices. There are
two main heat loss mechanisms in these cantilevers. The first one is the heat loss due to
thermal conduction to the substrate through the anchor, and the other through convective
heat transfer between the structure and its surrounding. To simulate the anchor losses, the
face of the cantilever that is in contact with the anchor was given a constant temperature of
30 o C. The natural convection boundary condition was approximated by estimating a heat
transfer coefficient. This is determined by empirical handbook correlations [36], which the
software calculates automatically. Table 4.1 shows the properties of Pt and V O2 used in
the model.
The FEM model simulated the contraction of V O2 during the transition and was used to
33

Material
Young’s Modulus (GPa)
Poisson’s ratio
Density (kg/m3 )
Heat Capacity (J/kg ∗ K)
Thermal Conductivity (W/(m ∗ K))
Reference Temperature (o C)
Thermal expansion coefficient (1/K)
Electrical Conductivity (S/m)

V O2
Pt
140
169
0.33
0.38
4670
21450
700
133
4
71.6
470
470
α(T ) 8.8×10−6
N/A
σ(T )

Table 4.1 Material properties used in the FEM model.
calculate the electric power dissipated at the Pt heater as a function of temperature. Given
that the thermal expansion coefficient of Pt varies less than 2% in the temperature range that
includes the phase change in V O2 [15], this value was assumed to be constant in the analysis.
The contraction of the V O2 thin film, and the power dissipated at the Pt heater as a function
of temperature were modeled by a temperature dependent thermal expansion coefficient of
V O2 (α(T )) and conductivity of Pt (σ(T )), respectively. The following expressions were
used:

α(T ) =

6 × 10−6
1 + e−0.3(T −341.15)

σ(T ) =

− 3 × 10−8 (T − 273.15) + 6.9 × 10−6

8.9 × 106
1 + 0.003729(T − 293.15)

(4.1)

(4.2)

where T is temperature in Kelvin. Equation (4.1) was obtained from a curve fit to experimental measurements, while the numerator of Equation (4.2) is the conductivity for Pt at
room temperature, and the expression in itself is commonly used for modeling changes in
conductivity as a function of temperature. A plot of Equation (4.1) is shown in Figure
4.4 . This model was used to estimate the operating current, and the dynamic response of
the system under various conditions.

34

1 1
1 0

α ( 10−6/o C )

9
8
7
6
5
3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

9 0

T e m p e ra tu re (oC )

Figure 4.4 Temperature dependent thermal expansion coefficient of V O2 (plot of α(T ) in
Equation (4.1)) [9].

4.3

Heater Design

The heater was designed to localize the power dissipation as much as possible along the
cantilever’s length and to maintain a uniform temperature across its width, to avoid twisting
when heating. A single thin Pt line was chosen as the heater element, due to its simplicity.
The width of this line was 2 µm, which also was the minimum feature size of the fabrication
process. The heater was connected to larger electrode pads through wider traces (8.5 µm)
in order to concentrate most of the power dissipation at the heating element region (i.e. at
the 2 µm wide Pt resistor). Another design parameter was the location of the heater that
would maximize the deflection of the cantilever while minimizing power consumption.
As the heater location approaches the anchor, the conductive heat loss through the anchor
increases, which increases the power necessary to reach a specific temperature. However, the
closer to the anchor the V O2 film contracts, the greater the deflection of the cantilever. This
indicates that the location of the heater involves a tradeoff between power dissipation and
cantilever deflection. In this design the location of the heater was optimized. The thickness
of the Pt heater also affects the electrical resistance of the heater, and consequently the

35

power dissipated in the heating element. For the particular device presented here, it was
found that a Ti/Pt thickness of 200 nm was large enough to avoid film discontinuities and
thin enough to maintain the operating currents within the mA range. The final design was

N o r m a liz e d T ip D is p la c e m e n t ( A .U .)

obtained with the help of the FEM simulations discussed earlier.

1 .0 5
1 .0 0
0 .9 5
0 .9 0
0 .8 5
0 .8 0
0 .7 5
0 .7 0
0

1 0

2 0

3 0

4 0

5 0

H e a te r D is ta n c e fr o m

(a) FEM simulation of the power dissipation
in the heater, scale is in W/m3 .

6 0

7 0

8 0

A n c h o r (% )

(b) Simulation of the deflection of the cantilever as the position of the heater is varied.
Optimum position is near 24%.

Figure 4.5 FEM simulation results for heater optimization [9].
Figure 4.5 shows the final heater design as well as the FEM simulations results for
power dissipation and optimum location of the heater. The optimal location of the heater
was determined to be at around 24% of the cantilever length from the anchor. This result
also agrees with previous work, where the optimum position of a heating laser spot was
determined, for maximum amplitude [37]. The resistance of the heater was found to vary
between 16.5 Ω and 20.5 Ω as the current is increased from 0 to 9 mA.

4.4

Results

The measurements are divided into two main categories: static and frequency response. In
the static analysis the relationship between heater current and tip displacement is studied,
as well as the difference in power consumption when the system is in vacuum or open to

36

atmosphere pressure. The frequency analysis studies the bandwidth of the system and the
effects of the design parameters.

4.4.1

Static Analysis

Figure 4.6 FEM simulation results of the temperature (o C) distribution for the 300 µm
cantilever in air and vacuum. The anchor is at the left [9].
Figure 4.6 shows the simulated temperature profiles for the 300 µm long cantilever under
vacuum and air. A more uniform temperature over the surface of the cantilever is obtained
for vacuum environment. This means that the difference in temperature between the heater
and the rest of the cantilever, in vacuum is smaller than in air, and therefore the heater only
needs to reach about half the temperature needed in air to obtain the same deflection. Also,
the resistance of the heater is smaller at lower temperatures, which translates to less power
consumption in vacuum. This was confirmed experimentally as shown below. Figure 4.7
shows a SEM picture of both cantilevers before and after actuation under vacuum. There is
an initial deflection at room temperature, before actuation, due to the residual stress from

37

the deposition of V O2 which occurs at nearly 470 o C. The total deflection for the 300 µm
and 200 µm cantilevers upon actuation was 68.7 µm and 28.5 µm, respectively.

(a) The 200 µm long cantilever (highlighted
in green).

(b) The 300 µm long cantilever (highlighted
in red).

Figure 4.7 SEM images of cantilevers. The top and bottom images show the actuator before
and after actuation, respectively [9].
Figure 4.8 shows the deflection as a function of input current for both cantilevers. The
deflection measurements were taken from SEM images as the current to the resistive heater
was increased in increments of 0.4 mA. The electrical signals for this experiment were provided by a voltage-to-current converter connected to the chip inside the SEM by electrical
feedthroughs.
It can be seen that the simulation correctly predicts the deflection and general shape of
the response, but it is shifted towards larger current values, which represents an overestimate
of the actual heat loss in the system. The experiment in air is closer to the simulation results,
which shows that the overestimate is more likely due to the modeling of the anchor losses than
to the modeling of the heat convection. The FEM model assumes perfect anchor geometry,
which does not exactly represent the real system, since the metal and SiO2 depositions are
not perfectly conformal. This non-conformity results in a cantilever structure with a smaller
38

V a c u u m ( E x p e r im e n ta l)
V a c u u m ( S im u la tio n )

A ir ( E x p e r im e n ta l)
A ir ( S im u la tio n )

7 2

1 5 0

6 8

1 4 0

T i p D i s p l a c e m e n t ( µm )

T i p D i s p l a c e m e n t ( µm )

A ir ( E x p e r im e n ta l)
A ir ( S im u la tio n )

6 4
6 0
5 6
5 2
4 8
4 4

V a c u u m ( E x p e r im e n ta l)
V a c u u m ( S im u la tio n )

1 3 0
1 2 0
1 1 0
1 0 0
9 0
8 0
7 0

4 0
0 .0

2 .0 m

4 .0 m

6 .0 m

8 .0 m

1 0 .0 m

0 .0

C u rre n t (A )

2 .0 m

4 .0 m

6 .0 m

8 .0 m

1 0 .0 m

C u rre n t (A )

(a) 200 µm long cantilever.

(b) 300 µm long cantilever.

Figure 4.8 Static response and FEM simulation results in vacuum and in air [9].
cross-section at the anchor, making the actual cantilever more thermally insulated from the
substrate than the FEM model. It can also be noticed that, in air, the difference in operating
current between the simulation and the real system is less than 10% of the full range for the
300 µm-long cantilever; while this difference is about 20% for the 200 µm-long cantilever
(see Figure 4.8). This indicates that the error in FEM model predictions becomes larger as
the heat losses through the anchor become more dominant (the 200 µm-long cantilever has a
lower surface-to-anchor area ratio). It is also noted that the displacement curves across the
phase-change are steeper for the device operated in vacuum media. This is because a given
increase in current will produce a larger increase in temperature for the device in vacuum,
since convection losses are almost non-existent.
Another observation can be made in terms of power consumption. Taking the average of
the simulated change in resistance for the Pt resistive heater mentioned earlier (18.5 Ω), and
the experimental measurements shown in Figure 4.8, an estimate of the required power to
achieve full-actuation can obtained. These results are summarized in Table 4.2. It is noted
that the necessary current to achieve full-actuation in air was approximately 3 times larger
than in vacuum, which means that full actuation in air requires approximately 16 times more
power than in vacuum.

39

Environment
Vacuum
Air

Estimated Required Power
for Full Actuation (mW)
300 µm Cantilever 200 µm Cantilever
0.07
0.09
1.15
1.18

Table 4.2 Estimated Power Consumption
4.4.2

Frequency Response

The schematic of the set-up used for the frequency response measurements is shown in
Figure 4.9. The set-up is based on laser deflection (or light scattering) techniques, similar to those used in the past [38]. A network analyzer (Agilent 3589A), was connected to
an adjustable voltage to current converter. A HeNe laser (CW 632 nm) was focused on
the cantilever inside a vacuum chamber. The reflected beam was detected using a silicon
photodiode (thorlabs PDA10A) and a DC-300 MHz amplifier (Stanford Research Systesm
model SR445). The vibration of the cantilever generates oscillatory movements of the reflected laser beam, which the photodiode converts to an oscillatory voltage signal detected
by the network analyzer for determining the frequency. A CCD camera was used for facilitating laser alignment prior to the measurements. The cantilevers were operated using a
positive sinusoidal current with adjusted amplitude and offset values to guarantee that the
cantilevers remained inside the SPT. This was monitored through an oscilloscope (Tektronix
TDS2004C) connected directly to the photodiode.
During the frequency response measurements, it was found that the He-Ne laser used for
detection heated the cantilever structure due to optical absorption. Since the cantilevers are
thermally isolated from the substrate, this power absorption added additional heat flow into
the cantilever besides the heat created by the resistive heater. Therefore, care was taken
to ensure that the laser spot was always placed close to the resistive heater, and a neutral
density filter was placed between the He-Ne laser and the first beam splitter, so that the
minimum amount of intensity necessary that allowed vibration detection from the reflected

40

HeNegLaser
VacuumgChamber
Sample

Filter
WhitegLight

BeamgSplitter

To
Computer

Lens
Detector

Camera

Oscilloscope
Amplifier

V-I
Converter
Network
Analyzer

Figure 4.9 Schematic of system used for frequency measurements [9].
laser beam was used. It is reasonable to assume that the delay between the mechanical
response and the thermal response is negligible since the air damping effects don’t become
significant until the cantilever is actuated at higher frequencies and the SPT of V O2 is much
faster [39]. Therefore, the most important effect to consider in electro-thermally driven V O2
actuators is the dynamic response of the heat transfer in the cantilever. The model that was
used to analyze the frequency response of the cantilevers includes heat transfer and joule
heating mechanisms.
Figure 4.10 shows the frequency response of the cantilevers. These results show the
cut-off frequency as defined by a 3 dB drop in amplitude (half power) with respect to the
maximum amplitude at low frequencies. The frequency range was chosen to include the first
resonant mode of the cantilever, which is shown as a peak in the response close to 10 kHz.
A lower quality factor is observed for the measurements done in air, which suggests that air
damping dominates the energy losses, as expected.
The cut-off frequency increases when the cantilever is exposed to air, which suggests that

41

-1 2

4 2 3 H z
(-3 d B )

3 2 H z
(-3 d B )

-2 4

D e ffle c tio n A m p litu d e ( d B )

D e ffle c tio n A m p litu d e ( d B )

1 2

A ir
V a c u u m
0

-3 6
-4 8
-6 0

A ir
V a c u u m
0

-1 2

2 9 H z
(-3 d B )

-2 4
-3 6

5 4 1 H z
(-3 d B )

-4 8
-6 0
-7 2

-7 2
-8 4
1 0

1 0 0

1 0 0 0

1 0 0 0 0

1 0

F re q u e n c y (H z )

1 0 0

1 0 0 0

1 0 0 0 0

F re q u e n c y (H z )

(a) The 200 µm long cantilever.

(b) The 300 µm long cantilever.

Figure 4.10 Frequency response curves for the cantilevers in vacuum and in air [9].
the limiting factor when it comes to half-power frequency response is dominated by heat
convection losses. For the two cantilevers in vacuum, the difference in heat dissipation by
conduction through the cantilever’s anchor is minimal, since the anchor area is the same
for both cantilevers. This is shown by very similar cut-off frequencies in vacuum where the
difference is only 9%. In air, however, there is a larger difference in cutoff frequency (about
25%), which is also supportive of the suggestion that heat convection losses dominate the
cut-off frequency, since both cantilevers have different surface areas. This indicates that
when operating in vacuum or very thermally isolated conditions, the size of the cantilever
affects very little the cut-off frequency of the cantilevers as long as you are operating the
cantilever far below the resonant frequency.

4.5

Summary

This chapter discussed the development of a V O2 -based actuator with an embedded micro
electric heater used for actuation. The performance of this actuator (static and dynamic
behavior) was characterized. FEM Models where developed and validated that will facilitate
future V O2 -based MEMS transducer designs. The necessary current for maximum deflection
was found to be very dependent on the ambient conditions particularly external pressure

42

and temperature. In order to be able to precisely control the displacement of these types
of actuators, feedback control must be used to compensate for external ambient conditions.
This control issue is the subject of the next chapter.

43

CHAPTER 5
CONTROL OF THE MEMORY EFFECT

In Chapter 4 it was shown that of the behavior of V O2 -based transducers is very sensitive to
variations in external ambient conditions. This is especially true when operating V O2 -based
transducers near the hysteretic region, where even a small temporary disturbance will change
the state of the V O2 film. This chapter presents a closed loop system which can be used
to compensate for variations in external ambient conditions without further complicating
the MEMS device fabrication or design. This technique will be combined with the results
from Chapter 3 to create a V O2 -based MEMS memory cell with much better performance
and controllability than previously reported devices. The work related to this chapter was
published in [10].

5.1

Design of the memory device

The memory cell described in this chapter consists of a V O2 /SiO2 bimorph cantilever with an
embedded titanium-platinum heater, where the titanium is only used for adhesion purposes.
The thickness of the heater and its “on-chip” electrical connections (hereby referred to as
“traces”) to the metal contact pads is 200 nm. A top view of the device can be seen in
Figure 5.1. The thickness of the cantilever was 2µm and the thickness of the V O2 film
was approximately 200 nm. The chip was cemented to an IC package using silver paint and
gold wires were bonded from the Ti/Pt metal pads on the chip to the connections of the IC
package, through which all the actuation and feedback electrical signals are transmitted.
Because of the large linear sensitivity of the resistivity of Pt to variations in temperature,
it is possible to use the resistance of the heater as a feedback measurement to stabilize the
temperature of the cantilever. Figure 5.2 shows how the resistance of the device varies as
the temperature is raised from 30 to 85 o C. This change in resistance with temperature was

44

Figure 5.1 Top view false color image of the memory cell. The Pt/Ti heater is shown in red
[10].
used on later experiments to monitor the temperature of the microcantilever. The entire
device can be actuated and controlled through this single terminal, simplifying the complete
system.

3 6 0

R e s i s t a n c e ( Ω)

3 5 0
3 4 0
3 3 0
3 2 0
3 1 0
3 0 0

R e s is ta n c e
F it

2 9 0
3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

T e m p e ra tu re (oC )

Figure 5.2 Measured resistance of the heater (and connecting traces) as the temperature of
the entire chip is varied [10].
After deposition of the V O2 film, the deflection of the cantilever was characterized as
the substrate temperature was varied across the SPT window. The results are shown in
Figure 5.3. The transition temperature in this work is defined as the points in the heating
and cooling curves where the rate of change in deflection is the largest. The transition
temperature for the heating and cooling curves are marked in green, and the corresponding
device resistance is also marked in Figure 5.2. This curve shows the classical curve for the
SPT found in previous work, showing that we have a highly oriented film of V O2 . Since the

45

temperature sensor used to measure the temperature of the IC package was placed on the
IC package instead of directly on the die, conductive losses from the device to the sensor
resulted in a lower temperature reading. On a separate experiment, the resistance of the V O2
film was measured as its temperature was varied across the phase transition, and the curve
was found to be shifted approximately 10 o C higher, which matches the typical transition
temperature for V O2 . Because of residual stress during fabrication, there is an initial out-ofplane deflection of the cantilever at room temperature. All deflection measurements obtained
were relative to that initial deflection. The static and transient response analysis discussed
next is done using electro-thermal actuation (i.e., Joule heating).

1 5 0

D e f l e c t i o n ( µm )

1 2 5
1 0 0
7 5
5 0

H e a tin g
F it
C o o lin g
F it

2 5
0
-2 5
3 0

3 5

4 0

4 5

5 0

5 5

6 0

6 5

7 0

7 5

8 0

8 5

T e m p e ra tu re (oC )

Figure 5.3 Measured deflection of the device as the temperature of the entire chip is varied
[10].

5.2

Static and transient response

The deflection of the cantilever was measured as the current through the device was cycled
from 0 to 7 mA. Figure 5.4 shows the resulting deflection curve. It can be readily noticed
that the shape of the deflection curve in Figure 5.3 and 5.4 are different. This is due a change
in temperature distribution when heating with the integrated heater in the cantilever, vs.
46

heating the whole chip with the Peltier heater. In conductive actuation, the Peltier heater
covers the entire bottom surface of the chip and therefore, the temperature is very uniform
across the entire chip. However, in the case of Joule heating, only the suspended structure
is being heated and the cantilever’s anchor acts as a heat sink.

1 0 0

D e f l e c t i o n ( µm )

8 0
6 0
4 0
2 0

H e a tin g
C o o lin g

3 1 7 Ω

0
2 9 0

3 0 0

3 1 0

3 2 0

3 3 0

3 4 0

3 5 0

3 6 0

R e s i s t a n c e ( Ω)

Figure 5.4 Deflection of the cantilever as the current is cycled, the deflection is plotted as a
function of the device resistance [10].
The total measured resistance of the heating electrode can be considered as two electrical
resistances connected in series: (i) the resistance of the heating element, and (ii) the resistance of the traces. It is necessary to estimate the values of these two resistance components,
since they can not be physically measured directly (note the size of the cantilever and the
dimensions of the heater and traces in Figure 5.1). The data from previous curves and the
dimensions of the heater were used for this calculation, and heater resistance was found to be
216 Ω whereas the trace resistance was found to be 84 Ω (both values at room temperature).
The thermal coefficient of resistance (TCR) of the Ti/Pt metallization was measured to be
3.17×10−3 K −1 .
As discussed in Chapter 3, the presented V O2 -based MEMS memory device is required
to operate within the hysteretic region. In order to maximize programmable states, it is best

47

if the device is preheated to a value at which the distance between the outermost hysteretic
loops (heating and cooling) is largest [8, 40]. In Figure 5.4 the outermost loops are the red
and blue lines, and the optimum operating resistance is shown by the green line, 317 Ω in
this case. The vertical separation of the loops at this resistance is the theoretical maximum
difference between the memory “states”, 44 µm in this case. The current needed to maintain
this operating temperature when the device is at 30 o C is 3.6 mA and the resistance of the
heater is 233 Ω which gives us the average power needed to operate the device at room
temperature, 3.02 mW.
In order to maintain the operating resistance of the device under fluctuating ambient
conditions, a closed-loop controller for the system was designed and implemented. For the
design of the controller, it was necessary to know the time response of the system. All the
control design parameters (e.g., sampling time, duration of programming pulses, etc.) will
depend on the speed of the system, which is limited by how fast can the temperature of the
structure be changed (see Section 4.4.2). A full scale open-loop current step response is
shown in Figure 5.5.

3 4 5
3 4 0
3 3 5

R e s is ta n c e

3 3 0
3 2 5

τ=4 .4 m s

3 2 0
3 1 5
3 1 0

R e s is ta n c e
R u n n in g A v e r a g e ( 1 0 p ts )

3 0 5
3 0 0
3 0 .0 m

4 0 .0 m

5 0 .0 m

6 0 .0 m

7 0 .0 m

T im e

Figure 5.5 Step response of the measured resistance of the device. Scattered points are
individual data measurements, and solid curve represents the average of the previous 10
resistance measurements [10].

48

The measured time constant was calculated from the average curve to be 4.4 ms. In order
to guarantee that the structure achieves a steady-state temperature after a step input, the
duration of the programming pulses were 40 ms which is several times larger than the times
constant. To have a more robust system that can compensate for temperature fluctuations,
a feedback system was designed, where the device’s temperature is constantly monitored
and used to control the current supplied to the heater. The closed-loop system should be
independent on the programming pulses; and thus, it was designed to be 10 times slower
than the programming pulse (400 ms). The control system was implemented using a field
programmable gate array (FPGA).

5.3

Open-loop versus closed-loop behavior

Figure 5.6 shows the schematic of the circuit used and the block diagram of the control
system. The programming pulses are added to the output of the controller (which is limited
to voltage amplitudes between 1 and 10 V). The device (represented by RD in the schematic)
was connected in series with a known resistance (R1 ) of 1.2 kΩ. The value of R1 was chosen
to protect the resistive heater in the actuator by limiting the current due to the maximum
applied voltage of 10 V. Using the measured value for the voltage across R1 and simple
circuit analysis, the current through the device and the value for RD was calculated ADD
R=V/I. The controller setpoint RSet is set to the previously determined operating resistance
317 Ω.
The first memory experiment was designed to test the robustness of the system. The
temperature of the chip was maintained at 30 o C using the Peltier, while the closed-loop
control system maintained the device resistance at 317 Ω. Then, programming current pulses
that maximized the difference in states were applied. In order to maximize the difference in
the programmed mechanical states, the pulse had to be large enough to transition the phase
change region completely during heating and cooling. The amplitude of the pulses required
were +1.52V (for heating) and -3V (for cooling). A train of pulses that cycled between the
49

V1

R1

Programing Pulses

V2

1V to 10V

1
R D RSet

PI

1

FPGA Controller

V1

F(u)

2

RD(V1,V2)

V2

Figure 5.6 Circuit schematic showing how the device is connected, where the voltages are
read, and a block diagram of the FPGA controller [10].
two states was applied, and the result is shown in Figure 5.7. This experiment was repeated
when the chip is at a temperature of 40 o C (result also shown in Figure 5.7).

D e f l e c t i o n ( µm )

1 0 0
o

3 0 C
o
4 0 C

O n -S ta te

8 0
6 0
4 0
2 0

O ff-S ta te
0
8

V o lta g e ( V )

6
4
2
0
0 .0

5 0 .0 m

1 0 0 .0 m

1 5 0 .0 m

2 0 0 .0 m

2 5 0 .0 m

3 0 0 .0 m

T im e ( s )

Figure 5.7 Close-loop response of the device at 30 and 40 o C [10].
In memory devices it is important to be able to differentiate between the different states,
for example, by using a comparator with a threshold somewhere between the states. This
means that at all times the states of a memory device must remain inside certain “windows”
that must never overlap. Figure 5.7 shows that as the temperature is raised the states shift
down by a small amount, but they remain inside windows that are separated by a sufficiently
large amount that permits reading of the states independently from temperature.
50

An important detail to notice is that for the higher background temperature of 40 o C,
the steady-state deflection is lower. However, from Figure 5.3 and 5.4, the steady-state
deflection should be larger for a higher temperature. This can be explained by considering
that the resistance of the device, RD , is actually composed of two parts as explained before,

RD (T, V ) = Rtrace (T ) + Rheater (T, V )

(5.1)

Where RD is the measured resistance of the device, T is the ambient temperature in o C,
V is the magnitude of the applied voltage pulse, Rheater is the resistance of the heater in
the device, and Rtrace is the resistance of the Ti/Pt traces, given by:

Rtrace (T ) = 84Ω[1 + T CR(T − 30o C)]

(5.2)

Notice that Rtrace has been assumed to be only a function of the ambient (or background)
temperature (i.e., the traces are assumed to be perfect heat sinks). From these two equations,
it can be noticed that when the ambient temperature rises by +10 o C, Rtrace increases by 2.6
Ω and since the controller forces RD to remain at 317 Ω, the value of Rheater has to decrease
by the same amount of 2.6 Ω. From the data plotted in Figure 5.4 it can be calculated that,
a drop of 2.6 Ω from the operating resistance of 317 Ω corresponds to a drop in deflection
of approximately 14 µm; which matches the difference in deflection between the steadystate deflection values for the two different background temperatures shown in Figure 5.7.
Therefore, the difference between memory states at different background temperatures can
be reduced by reducing the ratio of Rtrace /Rheater .
It is important to demonstrate the advantages of the designed closed-loop system over
the simpler open-loop version. To this end, the same tests were performed on the device in
open loop, with a constant average operating voltage that only keep the device resistance at
317 Ω in room temperature. This voltage was 5.65 V. The device was exposed to the same
sequence of pulses. The results are shown in Figure 5.8.

51

1 0 0
9 0

D e f l e c t i o n ( µm )

8 0
7 0
6 0
5 0
4 0
3 0
2 0
1 0
0 .0

5 0 .0 m

1 0 0 .0 m

1 5 0 .0 m

3 0

o

C

4 0

o

C

2 0 0 .0 m

2 5 0 .0 m

T im e ( s )

Figure 5.8 Open-loop response of the device at 30 and 40 o C [10].
At 30 o C the resulting memory states were very similar to the closed loop response, but
when the temperature was raised by +10 o C the deflection curve shifted up by more than
40 µm. The separation between states for the 40 o C background temperature is half of
that for the 30 o C. Furthermore, with these new curves it is impossible to set a threshold
to differentiate between the two states of the device. Therefore, due to the abruptness
of the changes in the properties of V O2 with temperature, some method for temperature
compensation (like the closed-loop system used here) is necessary for practical V O2 -based
memory applications.

5.4

Performance of the Vanadium Dioxide-based MEMS memory

The next set of experiments were designed to characterize the reliability of the device, and
its multiple state capability. The device was continuously cycled between its two states for
10 min (approximately 12 000 cycles). The values of the two states before and after the
cycling were measured and are presented in Figure 5.9. After cycling, no noticeable change
in the deflection for each state, nor in the shape of the devices response to the programming
pulses were noticed.
52

D e f l e c t i o n ( µm )

1 0 m in o f C y c lin g
7 5

7 5

6 0

6 0

4 5

4 5

3 0

3 0

1 5

1 5

0
0
0 .0

0 .1

0 .0

0 .2

0 .1

0 .2

T im e ( s ) + 1 0 m in

T im e ( s )

Figure 5.9 Repeatability test of the memory device in close loop [10].
Finally, multiple memory states were demonstrated by controlling the amplitude of the
programing pulse. A sequence of positive and negative pulses were used, and the results are
shown in Figure 5.10.

7 5

1 2

6 0

1 0

4 5

8

3 0

6

1 5
0 .0

1 0 0 .0 m

2 0 0 .0 m

3 0 0 .0 m

4 0 0 .0 m

V o lta g e ( V )

D e f l e c t i o n ( µm )

V o lta g e
D e fle c tio n

4
5 0 0 .0 m

T im e ( s )

Figure 5.10 Demonstration of multiple stable states in closed loop [10].

53

5.5

Summary

This chapter presented a V O2 -based micro-electro-mechanical memory cell with a closedloop system, which used the resistance of the integrated heater as a feedback variable, used
to compensate for variations in external ambient conditions, without further complicating
the MEMS device design or fabrication. This feedback system also expanded the usable
temperature window of the device. The various system parameters that affect the design
and tuning of the control loop were studied by determining the static and transient response
of the device, and the necessity to operate the device in close loop was demonstrated. A
large difference between the programed deection states was shown (30 µm) as well as the
capability to demonstrate multiple programmable states. The final memory cell showed high
reliability by showing no degradation after 12 000 cycles.

54

CHAPTER 6
MEMS COMPATIBLE FABRICATION PROCESS FOR VANADIUM DIOXIDE

All the experiments in previous chapters consisted of a continuous V O2 thin film deposited
over the entire device or substrate as a last step in the fabrication process. This is due to the
incompatibility of V O2 with most CMOS of even MEMS fabrication processes. Although
many different and useful devices can be fabricated this way, it limits the development of
more complicated systems that could benefit from the inclusion of V O2 films. The high
temperature deposition of V O2 as well as its ability to oxidize or reduce to other stable
oxides of vanadium makes it difficult to include it in standard fabrication processes. However,
by carefully choosing the right materials and defining a specific process flow, it is possible
to deposit and pattern V O2 films, as well as including them in more complicated MEMS
fabrication processes. This chapter presents a case study where a more elaborated process
flow is used which includes V O2 film deposition and patterning as part of an intermediary
step. The discussion presents the difficulties of using V O2 in MEMS processes as well a
possible ways to overcome them.

6.1

Vanadium Dioxide-based Voltage Controlled Resistor
2

1
3
Figure 6.1 Schematic symbol of a VCR.
The device presented in this case study is known as a Voltage Controlled Resistor (VCR).
55

A VCR is an electronic device which allows the control of a resistance between two terminals
by the control of a voltage on a third terminal. Its schematic symbol can be seen in Figure
6.1. Ideally the controlled resistance would be independent of the current going through it,
and should work for positive and negative voltages, limited only by power dissipation and
dielectric breakdown. The controlled resistance should also have a linear relationship with
the control voltage. This behavior is summarized in Figure 6.2.
IR

R
Rmax

Vmin
Vmax

VR

Vc

(a) I vs V curve for controlled
resistance.

(b) Resistance vs.
voltage.

control

Figure 6.2 Characteristics of ideal VCR
It is important to note that this device does not exist and, as it will be shown, even
the presented device still falls short of this description. However, there are many benefits
to implementing this device using V O2 . This device has many applications which include;
Variable Gain Amplifiers, Voltage Controlled Oscillators, Automatic Gain Controllers, Low
Distortion Tunable Filters, Audio Dynamic Range Compression, and Electronically Controlled Phase Advance and Retard Circuits.

6.2

Device Design

Figure 6.3 shows a general schematic view of the device. The fabricated device consists of the
elements presented in (blue), (red), and (green).The circuit connected to the device (shown
in black) is used to linearize the response of the VCR. The device consists of two identical
V O2 resistors that are electrically isolated from each other but in thermal contact. One
56

of them is used to monitor temperature (hereinafter referred to as the “feedback resistor”),
while the other is the actual VCR. The two resistors are also in thermal contact with a
resistive heater. By controlling the power delivered to the heater, the temperature of the
V O2 can be varied and therefore the resistance can be controlled. Because of the thermal
contact between both resistors, we can assume they both have the same resistance. It is
also important to consider that when a current flows through the resistor being used in the
application the temperature of that resistor will rise due to Joule heating. However, since
both resistors are in thermal contact, the temperature of the feedback resistor will also rise,
and be detected and corrected by the feedback circuit.

Rdiv

+
_

Vcc

Comp

Vset
Figure 6.3 Schematic of the VCR device connected in feedback for linearization. blue (VCR),
red (“feedback resistor”), and green (resistive heater).
An initial design that was tested involved using two V O2 films deposited one over the
other, separated by a SiO2 layer. Due to the sensitivity of the oxidation level of the V O2
films to the annealing time and deposition temperature, the quality of the first V O2 film was
degraded during the deposition of the second film; most probably due to over-oxidation. In
order for the VCR to function properly, both V O2 resistors would need to be identical. So
even if this degradation could be reduced, it would be very difficult to achieve two identical
V O2 films that were not deposited simultaneously. Therefore the final process flow must
only include one deposition of V O2 .

57

There are two main components to the VCR: the electric micro-heater, and the V O2
resistors. The design of the V O2 resistors requires the characterization of the conductivity
of the V O2 films. This was accomplished by depositing a rectangular patch of V O2 over
SiO2 using a shadow mask. Aluminum electrodes where then deposited on the sides of
the rectangular patch, and the resistance was measured as a function of temperature. The
dimensions of the patch as well as the thickness of the V O2 film was used to calculate the
resistivity of the film. The result is shown in Figure 6.4.

R e s i s t i v i t y ( Ω* m )

1 0

-1

1 0

-2

1 0

-3

1 0

-4

1 0

-5

1 0

-6

0 . 0 1 9 1 Ω∗m
6 . 0 2 8 * 1 0 - 6 Ω∗m

M a x :
M in :

3 0

4 0

5 0

6 0

7 0

8 0

9 0

1 0 0

o

T e m p e ra tu re ( C )

Figure 6.4 Resistivity of a 187 nm thick V O2 film.
With this curve, a V O2 resistor can be designed by choosing the appropriate (L/W ) ratio.
The amount of power needed to raise the temperature of the resistor would depend on its
heat dissipation, which is dominated by the surface area (L × W ). However the relationship
between the surface area of the resistor and the power needed to raise the temperature by a
given amount is difficult to determine analytically, especially for a resistor made of V O2 . So
Finite Element Models (FEM) where used to simulate the thermal response of the device.
This was combined with experimental measurements shown in Figure 6.4 to determine
empirical design equations that could be used to calculate the dimensions of the resistors.
The dimensions of the resistors were calculated so that for a chosen amount of power they

58

would reach 100 o C.
Any material used in the fabrication process before the deposition of the V O2 film needs
to be able to withstand the high temperatures attained during the deposition of V O2 . For
this reason, platinum was chosen for the heater. The design of the heater was much simpler
than the design of the resistors. A serpentine pattern was chosen, which covered the entire
area of the resistor. The total resistance of the heater was chosen based on the chosen
power and chosen operating voltage, and the total number of turns was chosen to match
that resistance. The performance of the heaters were also verified through FEM simulations.
All FEM simulations were done using COMSOL Multiphysics.
Because only one layer of V O2 can be deposited on the device, the resistors had to be on
the same plane. In order to maximize the temperature uniformity over both resistors, the
design shown in Figure 6.5 was used. The V O2 resistors, shown in purple, are broken apart
into thin strips which are alternated and connected in parallel. the metal interconnects,
shown in yellow would be separated by a thin oxide layer, shown in light blue. The parallel
combination of these strips would produce the desired resistance.

Figure 6.5 Top view of the interdigitated resistive array design used.

59

6.3

Mask Design

Various designs were implemented. Three maximum resistance values where chosen (200K,
500K, 1M) as well as three maximum power ratings (50mW, 100mW, 200mW). Mask files
where designed in L-Edit. Figure 6.6 shows an example of one of the devices that where
designed.

Figure 6.6 Mask design for one of the VCR designs.
The maximum wafer size compatible with the PLD chamber used to deposit the V O2
film was 2 in, so the masks where designed for 2 in diameter wafers. the final masks also
included other devices that where compatible with the process. The final full wafer mask is
shown in Figure 6.7.

6.4

Fabrication Process Flow

The process starts with a 2 in diameter silicon wafer, 250 µm thickness. A 2 µm layer of SiO2
is deposited by PECVD. This will isolate the metal layer from the silicon substrate. The
first metal layer of 150 nm platinum with an adhesion layer of Titanium (50 nm) is deposited
using liftoff (Figure 6.8a). A second layer of SiO2 (only 200 nm this time)is deposited using
PECVD. A layer of V O2 is then deposited using PLD at 550 o C in an oxygen atmosphere.
The V O2 layer is then patterned by plasma etching (Figure 6.8b). Vias to the first metal

60

Figure 6.7 Full wafer Mask used in the VCR process.
layer are then etched using wet etching in BOE[1:5] (Figure 6.8c). The second metal layer
is then deposited by lift-off using the same conditions as the first metal layer (Figure 6.8d).
Another 200nm oxide layer is deposited using PECVD. Openings to the contact pads are
etched using BOE[5:1]. Finally a layer of Au:Ti 150nm:50nm is deposited over the contact
pads to facilitate wire bonding to the IC package (Figure 6.8e). The wafer is then diced
(Figure 6.8g).

61

(a) After 1st metal layer.

(b) After V O2 patterning.

(c) After Etching vias.

(d) After 2nd metal layer.

(e) After gold deposition.

(f) Alignment marks.

(g) Final diced wafer.

(h) DIE in IC package.

Figure 6.8 Fabrication Results.
62

6.5

Results

As soon as the V O2 film was deposited, its resistance was measured as a function of temperature. the result is shown in Figure 6.9. These measurements were taken 1 in from the
center of the substrate. These results show a good quality V O2 film over the PECVD SiO2
film.

R e s i s t a n c e ( Ω)

1 M

1 0 0 k

1 0 k

1 k

1 0 0
3 0

4 0

5 0

6 0

7 0

8 0

9 0

o

T e m p e ra tu re ( C )

Figure 6.9 Resistance as a function of temperature for the V O2 film right after deposition.
After device completion, several of the VCR devices were tested without the feedback
circuit (in open loop). The heater voltage was slowly ramped up and down. Both V O2
resistors where simultaneously measured, as well as the power to the heater. The results for
one of these devices (200k, 200mV) are shown in Figure 6.10.
This shows that the minimum resistance was obtained at around the expected power
of 200mW. However the total resistance drop was less that the resistance drop observed in
Figure 6.9. There can be many causes for this drop and further experimentation is required.
One possibility is that the deposition temperature of the PCVD Oxide (300 o C) is sufficient
for causing a small change in the oxidation of the V O2 film. There was also a difference
in the resistance values of the V O2 resistors. This difference was found to be attributed to
bad adhesion between the metal contacts and the individual V O2 strips. In some cases the
63

2 5 0 k

1

R
2

1 8 0 m
1 6 0 m

P o w e r

1 4 0 m
1 5 0 k

1 2 0 m
1 0 0 m

1 0 0 k

8 0 m

P o w e r (W )

R e s i s t a n c e ( Ω)

2 0 0 k

2 0 0 m
R

6 0 m
5 0 k

4 0 m
2 0 m

0
0
0 .0

1 .3

2 .6

3 .9

5 .2

6 .5

7 .8

9 .1

V o lta g e ( V )

Figure 6.10 Open loop test on the 200mw 200k VCR.
metal would peal off these strips, essentially disconnecting that strip from the total resistor,
changing its equivalent resistance. An example of this problem can be seen in Figure 6.11.

Platinum Electrodes

Peal-off Contact
VO2 Strips
Figure 6.11 Exampe of the metal pealing off the V O2 film.
Small resistance devices contained many of these strips forming the V O2 resistors. So
when one or two got disconnected, the total resistance was not affected too much. However
larger resistance devices were composed of very few of these strips. So a few failed strips
would change the resistance significantly. As an example, Figure 6.12 shows the open
64

loop response of a (500k,100mW) device, which is much smaller. In this case the difference
between R1 and R2 is to large to efficiently use feedback compensation. This problem
however does not affect the thermal properties of the device, or the transient response. Future
implementations could compensate for this problem by making lower resistance devices or
by maybe using adhesion promoters.

8 8 0 k
R
1

R
2

R e s i s t a n c e ( Ω)

6 6 0 k

4 4 0 k

2 2 0 k

0
0 .0

1 .1

2 .2

3 .3

4 .4

5 .5

6 .6

7 .7

V o lta g e ( V )

Figure 6.12 Open loop test on the 100mw 500k VCR, showing large differences in R1 and
R2.
Another important parameter of the final VCR device is the response time. This is
defined as how fast the value of the resistance can be changed. In order to measure this,
both resistors were connected as voltage dividers and the output voltages were measured as
a step input voltage was given to the heater. The results for the 200k device are shown in
Figure 6.13. Due to the large thermal conduction between the resistor and the substrate,
the time constant of the device is much faster than those reported in the previous chapters.
As expected there is a difference between the heating and cooling time constants. Since
the heating transient is limited by how “fast” heat can be delivered to the device and is
therefore limited by the input power, the heating transient can be adjusted by changing the
heater power input. However the cooling transient is limited by power dissipation through

65

the substrate which is fixed by the geometry of the device. Figure 6.14 Shows the difference
in heating and cooling time constant for a (500k, 100mW) device.

1 .9

9 .2

τ =1 9 µs

8 .5

1 .5

7 .8

1 .3

7 .1

1 .1

6 .4

0 .9
O u tp u t 1
O u tp u t 2
In p u t V o lta g e

0 .8
0 .0

5 0 .0 µ

1 0 0 .0 µ

1 5 0 .0 µ

2 0 0 .0 µ

5 .7

In p u t V o lta g e ( V )

O u tp u t V o lta g e ( V )

1 .7

5 .0

2 5 0 .0 µ

T im e ( s )

Figure 6.13 Transient response of the resistance of the VCR, due to a step input.

9 .0
O u tp u t 1
O u tp u t 2
In p u t V o lta g e

7 .5

1 .8 0

6 .0

1 .4 4

4 .5

1 .0 8

H e a tin g

C o o lin g

0 .7 2

τ =8 . 7 µs

τ =3 2 µs

0 .3 6

3 .0
1 .5

In p u t V o lta g e ( V )

O u tp u t V o lta g e ( V )

2 .1 6

0 .0

0 .0 0
0 .0

2 0 .0 µ

4 0 .0 µ

3 6 0 .0 0 µ

4 8 0 .0 0 µ

T im e ( s )

Figure 6.14 Transient response of the resistance of the 500k 100mw VCR, due to a step input
for the heating and cooling cycle.
Finally, in order to test the closed loop performance the feedback circuit shown in Figure
6.3 was implemented using an FPGA in order to quickly tune the feedback gains. The results

66

shown in Figure 6.15 show the controlled resistance as the control voltage is varied. These
results show a fairly linear response as desired.

R e s i s t a n c e ( Ω)

2 5 0 .0 k

2 0 0 .0 k

1 5 0 .0 k

1 0 0 .0 k
1 .0

1 .2

1 .4

1 .6

1 .8

2 .0

2 .2

2 .4

V o lta g e ( V )

Figure 6.15 Close Loop Test on the 200 mW, 200k VCR.

6.6

Summary

This chapter presented a case study on the design of a V O2 -based device which included
deposition and patterning of the V O2 film and incorporation of the V O2 film into the rest
of the micro-fabrication process. The resulting device was measured and its open-loop and
close-loop response was characterized. The complications of including V O2 films into microfabrication processes were explored, and a successful V O2 compatible fabrication process
was developed. These results show that with proper workarounds, it is possible to include
V O2 in more complicated designs, greatly expanding its potential applications.

67

CHAPTER 7
SUMMARY

7.1

Summary of Contributions

This dissertation presented a series of studies which explored the use of the structural phase
transition of V O2 in MEMS actuators. MEMS actuators enable the interaction of devices
with their surroundings, instead of simply sensing it. Actuation mechanisms, which are very
common in the macro size domain, such as pneumatics, induction motors, and combustion
engines, are very difficult and inefficient to implement in the micro domain. Therefore, new
actuation techniques must be developed and optimized. Current actuation mechanisms such
as electrostatic or thermal expansion have limited performance in terms of total displacement,
applied force, operating voltage, and power requirements. This has resulted in a push to
incorporate new multifunctional smart materials into standard MEMS devices to improve
performance. V O2 has proven to be one of the most promising materials for the creation of
new sensors and actuators. Although most of its electrical and optical properties have been
thoroughly studied, there is very little work related to its SPT and its application to MEMS
actuators. Throughout this dissertation the SPT of V O2 is extensively studied, methods
are developed for the optimal design of V O2 -based actuators. The memory effect caused
by the hysteresis in the SPT is exploited to create very robust programmable multiple state
actuators. And fabrication processes are developed which combine the use of V O2 with
standard MEMS fabrication processes.

7.2

List of Problems Solved in this Thesis

The following are the key points addressed in this dissertation:
1. Characterization of the temperature dependent bending caused by the SPT of V O2 in
68

bimorph cantilevers.
2. Characterization of the memory type behavior due to the hysteresis of the SPT in
V O2 -based actuators.
3. Optimization of the design of V O2 -based bimorph actuators with embedded electric
micro-heaters in terms of maximum deflection and power consumption.
4. Development of mathematical models which facilitate the future design and simulation
of new V O2 -based devices.
5. Characterization of the dynamic response of V O2 -based actuators.
6. Development of a robust feedback control system which accurately controls the programed states of V O2 -actuators employing the memory effect of the SPT.
7. Development a fabrication process which permits the inclusion of V O2 thin films in
the fabrication of new MEMS devices.

69

BIBLIOGRAPHY

70

BIBLIOGRAPHY

[1] M. S. Baker and A. D. Corwin, “Design and reliability of a MEMS thermal rotary
actuator.,” Sandias National Lab SAND2007-5901C, Sept. 2007.
[2] M. C. LeMieux, M. E. McConney, Y.-H. Lin, S. Singamaneni, H. Jiang, Bunning, and
V. V. Tsukruk, “Polymeric Nanolayers as Actuators for Ultrasensitive Thermal Bimorphs,” Nano Letters, vol. 6, pp. 730–734, Mar. 2006.
[3] D. Zakharov, G. Lebedev, A. Irzhak, V. Afonina, A. Mashirov, V. Kalashnikov, V. Koledov, A. Shelyakov, D. Podgorny, N. Tabachkova, and V. Shavrov, “Submicron-sized actuators based on enhanced shape memory composite material fabricated by FIB-CVD,”
Smart Materials and Structures, vol. 21, p. 052001, May 2012.
[4] P. Baum, D. S. Yang, and A. H. Zewail, “4D Visualization of Transitional Structures in
Phase Transformations by Electron Diffraction,” Science, vol. 318, pp. 788–792, Nov.
2007.
[5] A. R´
ua, F. E. Fern´andez, and N. Sep´
ulveda, “Bending in VO2 -coated microcantilevers suitable for thermally activated actuators,” Journal of Applied Physics, vol. 107,
p. 074506, Jan. 2010.
[6] E. Merced, X. Tan, and N. Seplveda, “Strain energy density of vo2-based microactuators,” Sensors and Actuators A: Physical, vol. 196, no. 0, pp. 30 – 37, 2013.
[7] H. Coy, R. Cabrera, N. Sep´
ulveda, and F. E. Fern´andez, “Optoelectronic and all-optical
multiple memory states in vanadium dioxide,” Journal of Applied Physics, vol. 108,
p. 113115, Jan. 2010.
[8] R. Cabrera, E. Merced, N. D´avila, F. E. Fern´andez, and N. Sep´
ulveda, “A multiplestate micro-mechanical programmable memory,” Microelectronic Engineering, vol. 88,
pp. 3231–3234, Nov. 2011.
[9] R. Cabrera, E. Merced, and N. Sepulveda, “Performance of Electro-Thermally Driven
V O2 -Based MEMS Actuators,” Journal of Microelectromechanical Systems, pp. 1–1,
Jan. 2013.
[10] R. Cabrera, E. Merced, and N. Seplveda, “A micro-electro-mechanical memory based
on the structural phase transition of vo2,” physica status solidi (a), vol. 210, no. 9,
pp. 1704–1711, 2013.
[11] F. Morin, “Oxides Which Show a Metal-to-Insulator Transition at the Neel Temperature,” Physical Review Letters, vol. 3, pp. 34–36, July 1959.
[12] F. C. Case, “Improved V O2 thin films for infrared switching,” Applied optics, vol. 30,
p. 4119, Jan. 1991.

71

[13] N. Sep´
ulveda, A. R´
ua, R. Cabrera, and F. Fern´andez, “Young’s modulus of V O2 thin
films as a function of temperature including insulator-to-metal transition regime,” Applied Physics Letters, vol. 92, p. 191913, Jan. 2008.
[14] G. Stefanovich, A. Pergament, and D. Stefanovich, “Electrical switching and Mott transition in VO2,” Journal of Physics: Condensed Matter, vol. 12, pp. 8837–8845, Oct.
2000.
[15] J. W. Arblaster, “Crystallographic Properties of Platinum,” Platinum Metals Review,
vol. 50, pp. 118–119, July 2006.
[16] J. H. Comtois, M. A. Michalicek, and C. C. Barron, “Electrothermal actuators fabricated in four-level planarized surface micromachined polycrystalline silicon,” Sensors
and Actuators A: Physical, vol. 70, pp. 23–31, Oct. 1998.
[17] L. Que and J. Park, “Bent-beam electro-thermal actuators for high force applications,”
Journal of Microelectromechanical Systems, pp. 1–6, Jan. 1999.
[18] J. Juuti, K. Kord´as, R. Lonnakko, P. V. Moilanen, and S. Lepp¨avuori, “Mechanically
amplified large displacement piezoelectric actuators,” Sensors and Actuators A: Physical, vol. 120, pp. 225–231, Apr. 2005.
[19] N. Yamada, “Erasable phase-change optical materials,” MRS Bulletin, vol. 21, pp. 48–
50, 9 1996.
[20] C. A. Volkert and M. Wuttig, “Modeling of laser pulsed heating and quenching in optical
data storage media,” Journal of Applied Physics, vol. 86, no. 4, pp. 1808–1816, 1999.
[21] D. Wamwangi, W. Njoroge, and M. Wuttig, “Crystallization kinetics of ge4 sb1 te5 films,”
Thin Solid Films, vol. 408, no. 12, pp. 310 – 315, 2002.
[22] I. Friedrich, V. Weidenhof, W. Njoroge, P. Franz, and M. Wuttig, “Structural transformations of ge2 sb2 te5 films studied by electrical resistance measurements,” Journal of
Applied Physics, vol. 87, no. 9, pp. 4130–4134, 2000.
[23] T. P. L. Pedersen, J. Kalb, W. K. Njoroge, D. Wamwangi, M. Wuttig, and F. Spaepen,
“Mechanical stresses upon crystallization in phase change materials,” Applied Physics
Letters, vol. 79, no. 22, pp. 3597–3599, 2001.
[24] J. Kalb, Q. Guo, X. Zhang, Y. Li, and C. Sow, “Phase-change materials in optically
triggered microactuators,” Journal of Microelectromechanical Systems, pp. 1–10, Jan.
2008.
[25] P. Krulevitch, A. Lee, P. Ramsey, J. Trevino, J. Hamilton, and M. Northrup, “Thin
film shape memory alloy microactuators,” Microelectromechanical Systems, Journal of,
vol. 5, no. 4, pp. 270–282, 1996.
[26] Y.-T. Cheng and D. Grummon, “Indentation in shape memory alloys,” in Micro and
Nano Mechanical Testing of Materials and Devices (F. Yang and J. C. Li, eds.), pp. 69–
84, Springer US, 2008.
72

[27] D. Grummon, R. Gotthardt, and T. LaGrange, “Planar extrinsic biasing of thin film
shape-memory mems actuators,” Materials Research Society Symposium - Proceedings,
vol. 741, pp. 287–292, 2002.
[28] W. Ni, Y.-T. Cheng, and D. S. Grummon, “Microscopic shape memory and superelastic effects under complex loading conditions,” Surface and Coatings Technology,
vol. 177178, no. 0, pp. 512 – 517, 2004. Proceedings of the 30th International Conference on Metallurgical Coatings and Thin Films.
[29] P. Krulevitch, A. Lee, and P. Ramsey, “Thin film shape memory alloy microactuators,”
Journal of Microelectromechanical Systems, pp. 1–13, Jan. 1996.
[30] P. Dernier, “Structural investigation of the metal-insulator transition in v6 o13 ,” Materials Research Bulletin, vol. 9, no. 7, pp. 955 – 963, 1974.
[31] J. Feinleib and W. Paul, “Semiconductor-To-Metal Transition in V2 O3 ,” Physical Review, vol. 155, pp. 841–850, Mar. 1967.
[32] V. Eyert, “The metal-insulator transitions of VO2 : A band theoretical approach,” Annalen der Physik, vol. 11, no. 9, pp. 650–704.
[33] K. Liu, C. Cheng, Z. Cheng, K. Wang, R. Ramesh, and J. Wu, “Giant-amplitude,
high-work density microactuators with phase transition activated nanolayer bimorphs,”
Nano Letters, vol. 12, no. 12, pp. 6302–6308, 2012.
[34] A. R´
ua, F. E. Fern´andez, R. Cabrera, and N. Sep´
ulveda, “Young’s modulus of pulsedlaser deposited V6 O13 thin films,” Journal of Applied Physics, vol. 105, p. 113504, Jan.
2009.
[35] J. Cao, E. Ertekin, V. Srinivasan, W. Fan, S. Huang, H. Zheng, J. W. Yim, D. R.
Khanal, D. F. Ogletree, J. C. Grossman, and J. Wu, “Strain engineering and onedimensional organization of metal-insulator domains in single-crystal vanadium dioxide
beams,” Nature Nanotechnology, vol. 4, pp. 732–737, Sept. 2009.
[36] F. P. Incropera, A. S. Lavine, and D. P. DeWitt, Fundamentals of Heat and Mass
Transfer. John Wiley & Sons Incorporated, 2011.
[37] D. Ramos, N. C. Microelectronics, J. Tamayo, J. Mertens, and M. Calleja, “Photothermal excitation of microcantilevers in liquids,” Journal of Applied Physics, vol. 99,
pp. 124904–124904–8, June 2006.
[38] E. Merced, R. Cabrera, H. Coy, F. E. Fernandez, and N. Sepulveda, “Frequency Tuning
of V O2 -Coated Buckled Microbridges,” Journal of Microelectromechanical Systems,
vol. 20, pp. 558–560, June 2011.
[39] R. Cabrera, P. R. Mayag¨
uez, E. Merced, N. Sep´
ulveda, and F. E. Fern´andez, “Dynamics
of photothermally driven V O2 -coated microcantilevers,” Journal of Applied Physics,
vol. 110, pp. 094510–094510–8, Nov. 2011.

73

[40] E. Merced, R. Cabrera, N. D´avila, F. E. Fern´andez, and N. Sep´
ulveda, “A micromechanical resonator with programmable frequency capability,” Smart Materials and
Structures, vol. 21, p. 035007, Mar. 2012.

74