 

ii

3.3!}.

d.L...u<ﬁ.v..
1.x! 23w. J

3
{3.3.3.6
.7.“ u. .

. 1

1. t...

.44.”: .33 5.5 rawzwﬁrw,

O

s .61.. . -.. .
. v a: . :. . ..
43w. .. 44%. g

a...
..

1- .
Kuhn-ﬁt.”
I}! .3‘ .1 ‘
\ .‘\ vi.
1:.

am...
i 1

i k. 1i
:5; “15..
IL) ?

3h“: .
(.115
n... ail;
«t 1‘5.

‘ $9... M} 3.1.

“an.“ .

9.3.1171...»
\v a,.wr!‘.ﬂl:.1n .

 

n
a... . . 3:3“. 5.,

.

 

W "
0 0) ﬁ
0

 

LIBRARY
Michigan State

 

University

This is to certify that the
thesis entitled

THE DEPOSITION OF AIN THIN FILMS, THEIR

CHARACTERIZATION AND THE FABRICATION OF SURFACE

ACOUSTIC WAVE DEVICES

presented by

CHARLEE FANSLER

has been accepted towards fulfillment
of the requirements for the

MS. degree in Electrical Engineering

 

 

 

 

0&va /( 6W1

Major Professor’s Signature
5173/ a7

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DAIEDUE

DAIEDUE

DATEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/CIRC/DateDue.indd-p.1

 

THE DEPOSITION OF AIN THIN FILMS, THEIR CHARACTERIZATION AND THE
FABRICATION OF SURFACE ACOUSTIC WAVE DEVICES

By

Charlee Fansler

A THESIS

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Electrical and Computer Engineering

2007

ABSTRACT

THE DEPOSITION OF AIN THIN FILMS, THEIR CHARACTERIZATION AND THE
FABRICATION OF SURFACE ACOUSTIC WAVE DEVICES

By

Charlee Fansler

This investigation entailed a study of how to fabricate surface acoustic wave
(SAW) devices; speciﬁcally, the fabrication and design of the electrodes, the thicknesses
of the AIN and the UNCD (ultrananocrystalline diamond) layers and the parameters to set
for DC pulsed sputter deposition of the AIN layer. The critical layer in the fabrication
process is the AIN layer.

The resulting layer of AIN is transparent with an optical gap of approximately 6
eV or 200 nm. Preliminary XRD (x-ray diffraction) results show a small peak signal at
36° for AIN (002) and a signiﬁcant AIN (004) peak at ~76°. The (002) plane for AIN is
the desired plane for its piezoelectric properties. The surface roughness of the AIN layer
was low with an Ra of 0.78 nm and an R2 value of 28.77 mm. The deposited layer also
had good adhesion to the substrate.

An AIN layer was successfully deposited on UNCD, speciﬁcally AlN/UNCD/Si
were the order of the layers. Al was also deposited on the AIN layer in order to form the
IDTs for the SAW devices. However, it was not possible to etch and pattern the
aluminum because the 75 mm diameter wafer was bowed after UNCD deposition. SAW
devices were successfully fabricated on a quartz — ST substrate. A signal was obtained
for a SAW ﬁlter, using an HP 8753D network analyzer, as veriﬁcation that the SAW

devices would work.

Copyright by
CHARLEE FAN SLER
2007

For Ashley with lots of love.

iv

ACKNOWLEDGEMENTS

I would like to thank my family for all the support they have given to me over the
years; for always expecting more out of me, which motivated me to do more and be
better at whatever I was doing. If it weren’t for my family I wouldn’t be where I am
today or be the person that I am now.

I would also like to thank my advisor Dr. Donnie Reinhard for all his help and
guidance. I appreciate all that he has done for me and all that he continues to do. I know I
could never show the true extent of my gratitude, but I hope this thank you is a good start.

Finally, I would like to thank Dr. Ed Rothwell, Dr. Tim Grotjohn, Dr. Jes
Asmussen, and Dr. Thomas Schuelke for their guidance and support along the way. I
would like to extend my thanks as well for the support of Michigan State University, the
Graduate School and the Fraunhofer Center for Coatings and Laser Applications. I would
also like to thank Dr. Percy Pierre and Dr. Barbara O’Kelly, without whom I would not

even be in the graduate program.

TABLE OF CONTENTS

LIST OF TABLES ........................................................................................................... viii
LIST OF FIGURES ........................................................................................................... ix
Chapter 1 ............................................................................................................................. 1
Introduction ......................................................................................................................... 1
1.1 Introduction .............................................................................................................. l
1.2 Purpose of Research ................................................................................................. 1
1.3 Chapter Review ........................................................................................................ 3
1.4 Summary .................................................................................................................. 5
Chapter 2 ............................................................................................................................. 6
Background ......................................................................................................................... 6
2.1 Introduction .............................................................................................................. 6
2.2 Piezoelectricity ......................................................................................................... 6
2.3 Surface Acoustic Wave Device ............................................................................... 9
2.4 Diamond Interlayer for SAW Device .................................................................... 11
2.5 Piezoelectric Material for SAW Device ................................................................ 13
2.6 Previous Work on Diamond Based SAW Devices ................................................ 14
Chapter 3 ........................................................................................................................... 17
SAW Fabrication Considerations and Procedure ............................................................. 17
3.1 Introduction ............................................................................................................ 17
3.2 Film Thickness Considerations .............................................................................. 17
3.3 Electrode (IDT) Formation .................................................................................... 20
3.4 Electrode (IDT) Design ......................................................................................... 26
Chapter 4 ........................................................................................................................... 33
Aluminum Nitride Deposition and Characterization ........................................................ 33
4.1 Introduction ............................................................................................................ 33
4.2 Pulse DC Sputtering ............................................................................................... 33
4.3 PVD75 Description ................................................................................................ 38
4.4 PVD75 Operating Procedures ................................................................................ 61
Chapter 5 ........................................................................................................................... 67
Aluminum Nitride Deposition Observations .................................................................... 67
5.1 Introduction ............................................................................................................ 67

vi

5.2 Deposition Observations ........................................................................................ 67

5.3 Deposition Rate ...................................................................................................... 69
5.4 Uniformity ............................................................................................................. 71
5.5 Adhesion ................................................................................................................ 74
5.6 Optical Data: Reﬂection and Transmission ........................................................... 77
5.7 X-Ray Diffraction .................................................................................................. 81
5.8 Scanning Electron Microscopy .............................................................................. 84
Chapter 6 ........................................................................................................................... 87
SAW Device Characterization: Setup and Procedures ..................................................... 87
6.1 Introduction ............................................................................................................ 87
6.2 HP 8753D Description and Operating Procedures ................................................ 87
6.3 HP 8753D Calibration Procedures ......................................................................... 93
6.4 Probes/Probe Station Description .......................................................................... 98
6.5 Measurements/Results ......................................................................................... 102
Chapter 7 ......................................................................................................................... 108
Conclusions ..................................................................................................................... 108
7.1 Introduction .......................................................................................................... 108
7.2 Conclusions ........................................................................................................... 108
APPENDIX ..................................................................................................................... 112
List of Samples ............................................................................................................... 112
REFERENCES ............................................................................................................... 114

vii

LIST OF TABLES

Table 1 Parameters for the Sparc-le V unit in the PVD75 ................................................ 22

Table 2 Parameters for the MDX 1.5 kW power supply to the cathode of the PVD75... 22

Table 3 Parameters for inside the chamber of the PVD75 ................................................ 22
Table 4 Reverse time ranges matched to frequency ranges (from reference [36]) ........... 55
Table 5 Spare-1e V error codes and their meanings, taken from reference [36] ............... 57
Table 6 PVD75 Start-Up Procedure ................................................................................. 62
Table 7 AIN Deposition procedure ................................................................................... 63
Table 8 AIN Parameters for the Spare-1e V unit in the PVD75 ........................................ 64
Table 9 AlN Parameters for the MDX 1.5 kW ................................................................. 64
Table 10 AlN Parameters for inside the chamber of the PVD75 ...................................... 64
Table 11 PVD75 Shut-down procedure ............................................................................ 65
Table 12 Shortened list of AIN deposition parameters .................................................... 68
Table 13 Samples of AIN and some of their parameters .................................................. 68
Table ‘14 Sample deposition rates .................................................................................... 70
Table 15 Deposition parameters for AIN samples .......................................................... 112
Table 16 Layer characteristics for AIN samples ............................................................. 113

viii

LIST OF FIGURES

Figure 1 Time varying voltage applied to electrodes on a piezoelectric substrate ............. 7
Figure 2 Tension and compression due to time varying voltage (side view) ..................... 7
Figure 3 Simple SAW device (ﬁlter) .................................................................................. 9
Figure 4 SAW device layers ............................................................................................. 18
Figure 5 Fabrication steps for forming A1 electrodes ....................................................... 21
Figure 6 SAW Mask with 24 die, each with 5 devices ..................................................... 25
Figure 7 Four die on the SAW Mask ................................................................................ 27
Figure 8 Four die on the Backup SAW Mask ................................................................... 27
Figure 9 Filter with 20 electrode pairs per IDT at 8 mm wide .......................................... 28
Figure 10 Filter with 50 electrode pairs per IDT at 4 um wide ........................................ 28
Figure 11 Filter with a serpentine inductor added to the contact pads ............................. 29
Figure 12 View (50x) of serpentine inductor attached to contact pads of ﬁlter ............... 29
Figure 13 Filter with grated, serpentine inductor attached to contact pads of ﬁlter ......... 29

Figure 14 View (50x) of grated, serpentine inductor attached to contact pads of ﬁlter 30

Figure 15 Oscillator with 100 electrodes 8 pm wide and 36 reﬂectors at each end ......... 30
Figure 16 IDT design of the commercial device used in this investigation ...................... 31
Figure 17 Sinc function output corresponding to electrode pattern in Figure l ............... 32
Figure 18 Band pass output corresponding to electrode design in Figure 16 ................... 32
Figure 19 Cross-section of the PVD75 deposition chamber ............................................. 34
Figure 20 Simpliﬁed cross-sectional view of the magnetron’s magnetic ﬁeld .lines ........ 35
Figure 21 Race track shape erosion on sputtering target .................................................. 36

ix

Figure 22 Pulsed DC voltage supplied to the cathode of the PVD75 ............................... 37

Figure 23 Front of PVD75 system .................................................................................... 39
Figure 24 PVD75 Chamber .............................................................................................. 40
Figure 25 Vacuum screen ................................................................................................. 42
Figure 26 Gas screen ......................................................................................................... 44
Figure 27 Deposition screen ............................................................................................. 47
Figure 28 Maintenance screen .......................................................................................... 48
Figure 29 MDX 1.5 kW magnetron drive ......................................................................... 50
Figure 30 LED status lights on the MDX ......................................................................... 51
Figure 31 Sparc-le V pulsing and are suppression device ................................................ 52
Figure 32 Spare-1e V menus and programming buttons ................................................... 53
Figure 33 M.A.P.S. Heater Power Supply (substrate heater) ........................................... 58
Figure 34 Platen rotation control ...................................................................................... 61
Figure 35 PVD75 Power-box ............................................................................................ 62
Figure 36 Racetrack erosion of used target ....................................................................... 72
Figure 37 Sample CF-019, ~7.5 cm above the target, no rotation .................................... 73
Figure 38 Sample CF-008, uniformly coated sample, with rotation ................................. 73
Figure 39 Sample CF-009, placed off center from plasma, no rotation ........................... 74

Figure 40 0-20 XRD scan for sample CF-O22 (XRD scans performed by Rahul

Rarnamurti) ....................................................................................................................... 76
Figure 41 Reﬂection data collected from sample CF-008 ................................................ 78
Figure 42 Reﬂection data for sample CF-018 for a range of 1300 — 2800 nm ................. 79
Figure 43 Reﬂection data for sample CF-018 for a range of 185 — 305 nm ..................... 80

Figure 44 Transmission data for sample CF-018 .............................................................. 80
Figure 45 Figure 40 0-20 XRD scan for sample CF-008 (XRD scan performed by R.
Ramamurti) ....................................................................................................................... 82
Figure 46 Figure 40 0-20 XRD scan for sample CF-021 (XRD scan performed by R.
Ramamurti) ....................................................................................................................... 84
Figure 47 SEM cross-sectional picture of sample CF-016 without Au coating (SEM
pictures privided by Tran Dzung) ..................................................................................... 84
Figure 48 SEM cross—sectional picture of sample CF-016 with 20 nm Au coating (SEM
pictures privided by Tran Dzung) ..................................................................................... 85
Figure 49 SEM cross-sectional picture of sample CF-016 with 20 nm Au coating (SEM
pictures privided by Tran Dzung) ..................................................................................... 85

Figure 50 SEM cross-sectional picture of sample CF-016 with 20 nm Au coating (SEM

pictures privided by Tran Dzung) ..................................................................................... 86
Figure 51 Hewlett Packard 8753D network analyzer ....................................................... 88
Figure 52 HP 8753D display and other components ........................................................ 89
Figure 53 Parameter control panel .................................................................................... 90

Figure 54 Probe station connected to the HP 8753D in order to characterize the DUT... 92

Figure 55 Metalized patterns on the calibration standard ................................................. 95
Figure 56 High frequency probe ....................................................................................... 99
Figure 57 Probe connections to the probe station ............................................................. 99
Figure 58 Probe Station .................................................................................................. 100
Figure 59 Power box for the probe station lights ............................................................ 101
Figure 60 Stage located in the probe station ................................................................... 102

xi

Figure 61 Mounted commercial SAW device ................................................................ 103
Figure 62 S-parameter, SQ], of the commercial ﬁlter with the top of the packaging
removed ........................................................................................................................... 1.03

Figure 63 S-parameter, 312, of the commercial ﬁlter with the top of the packaging

removed and aluminum foil covering the remaining packaging for the device ............. 104
Figure 64 S-parameter, 82., of the commercial ﬁlter with the packaging in tact ........... 105
Figure 65 S-parameter, 82., of the experimental SAW ﬁlter .......................................... 106

Images in this thesis are presented in color.

xii

Chapter 1

Introduction

1.1 Introduction

This chapter gives an overview of this thesis research in general. It will also give
a short review of the chapter topics for this thesis. Finally, there will be a summary of this
investigation.
1.2 Purpose of Research

A Surface Acoustic Wave (SAW) device is an electronic device which can be
used, for example, as a high frequency band-pass ﬁlter or an oscillator; for this research
ﬁlters and oscillators were fabricated and tested. There are many applications for SAW
devices. For example, a SAW device may act as a transducer for chemical sensing
[l,2,3,4]. SAW devices can also be used in mobile phones, television tuners and optical-
communication systems [2,5,6,7,8]. SAW devices are becoming increasingly important in
wireless communications. Wireless mobile and ﬁxed transceivers require electronic
ﬁlters. SAW devices are a great choice for these applications because of their excellent
performance and small size. They have high signal selectivity, accurate bandwidths and
center frequencies, small transition bandwidths, and low passband ripple [5].

A recent trend in wireless communications is the move toward higher operating
frequencies to transmit more information. Current SAW devices not employing diamond
as an interlayer are reaching their limitations in operating frequency. The frequency at

which a SAW device can operate depends on the material properties of the media being

used to transmit the acoustic wave. For most SAW devices this is the piezoelectric
material, which is used to generate the acoustic wave as well. The operating frequency is
limited by the phase velocity of the material and the wavelength of the acoustic wave.
The wavelength of the acoustic wave is limited by the spacing between the ﬁngers of the
IDTs. In order to get higher frequencies the width and spacing of the electrodes must
become smaller. In today’s fabrication technology the submicron size of the electrodes
required to increase the operating frequencies further compromise device reliability and
yield [5,9]. Another limitation of SAW devices using conventional materials (i.e. quartz
and lithium niobate) is power durability. This becomes more of a problem with the size of
the IDTs, the smaller the electrodes the less durable your structure [7,9,10,11].

As technology improves, devices need to operate at higher frequencies and by
consequence become smaller in size. As this applies to SAW devices, it has led to
research of different materials to avoid reaching the limitations of current lithography
techniques to create smaller widths for the IDT electrodes. Diamond, though not
piezoelectric, is the material that is the most promising for SAW devices because it has
the highest acoustic velocity of any known material. Diamond has a high elastic constant
(in this case Young’s Modulus) and a high phase velocity as well [5,7]. With an increase
in velocity there is an increase in frequency, this will be explained in more detail in
Chapter 2. This means that by simply using diamond as an interlayer, between a
piezoelectric layer and the substrate, and keeping the same dimensions of the IDTs there
is an improvement in performance. Another improvement that could be realized using
diamond would be increasing the quality factor (Q) of a resonator. Diamond also has

other beneﬁts for use in SAW devices; it has a high thermal conductivity which will

allow for higher power capabilities. SAW devices on diamond have a high
electromechanical coupling coefﬁcient, which indicates a high efﬁciency of energy
conversion from the input signal to the SAW [5,7,8,9,10,12,13,l4].

For this research we have chosen to use ultrananocrystalline diamond as an
interlayer for the SAW devices, between piezoelectric aluminum nitride (AIN) and
silicon. Ultrananocrystalline diamond (UNCD) is characterized by a grain size of 2 - 10
nm and is grown by a process developed at Argonne National Laboratories by Gruen and
Krauss [15,16]. UNCD has the beneﬁt of a low surface-roughness which will eliminate or
greatly minimize the need for polishing the diamond which takes a great deal of time in
the fabrication process. This low surface roughness will also minimize propagation loss
of the SAW energy which leads to less insertion loss for the device and lower power
consumption. Due to the small grain boundaries of the UN CD this lowers the expected
acoustic scattering seen at large angle grain boundaries in polycrystalline diamond,
especially if grain dimensions are of the same size/length as the acoustic wavelengths,
device apertures and transducer separation [17,18]. The thermal conductivity of UNCD
however, is low. This could be contributed to the wavelength of the thermal phonons
being equivalent to the grain boundaries. However, despite the lower thermal
conductivity of UNCD it still offers the potential of diamond-based high frequency
performance to SAW devices.

1.3 Chapter Review

Chapter 2 gives an overview of background information needed for understanding

this thesis. In this chapter there will be an explanation of piezoelectricity and why it is

advantageous in surface acoustic wave (SAW) devices to use diamond as an interlayer.

Next, a description of what a SAW device is, how it works, its applications and some of
the performance limitations will be given. Then, a brief description of
ultrananocrystalline diamond (UNCD) is given; what are some of its material properties
and how it may improve the performance of SAW devices. Next, reasoning is given for
the choice of piezoelectric material used for this research. Finally, there is a short review
of previous work on diamond based SAW devices.

Chapter 3 explains the procedures used to fabricate the SAW devices. In this
chapter, reasoning is given for the choice of ﬁlm thickness, i.e. what thickness of
aluminum nitride to use and what thickness of diamond should be used? Next, there will
be an explanation of the design of the electrodes (IDTs) and the process used to fabricate
the electrodes (e. g. the deposition of the aluminum, masking and etching). Finally, there
will be a discussion on the design of the SAW structures.

Chapter 4 explains the PVD (physical vapor deposition) process of pulse dc
sputtering, what is it/how does it work and why is it important, as compared to other
sputtering methods. There will be a description of the sputtering system used, that was
developed by Kurt J. Lesker, and of the operating procedures. Finally, there will be a
discussion of the results of sputtering an aluminum nitride (AIN) layer on a Si substrate;
there will be a comparison of the properties of the sputtered layer vs. the sputtering
parameters.

Chapter 5 will discuss the observations made after the AIN deposition. First, a
discussion of general observations made, such as measurements concerning surface
roughness, substrates used and deposition parameters. Next, characteristics of the ﬁlm

that will be discussed are growth rate, uniformity and adhesion. Optical data,

transmission and reﬂection, was also collected. Then, x-ray diffraction (XRD) was used
to determine the chemical composition and lattice structure of the deposited layer.
Finally, scanning electron microscopy (SEM) was used to determine grain growth.

Chapter 6 will summarize the results for this investigation. In this chapter
conclusions from these results will be made. Finally, suggestions for the possibility of
future work will be made as well.

1.4 Summary

This thesis investigates the fabrication of SAW devices and the deposition of a
piezoelectric thin ﬁlm layer of AIN for the proper function of the SAW device. There is
also a discussion of the beneﬁts of using diamond, speciﬁcally UNCD, as an interlayer
for the SAW device to improve the function of the device. Simply by having the
diamond interlayer in place without making any other changes to the device improves its
functional capabilities due the material properties of diamond.

The background material, such as piezoelectricty, coupling factors, and previous
works with diamond and SAW devices is discussed. Then the design process is discussed
in detail, the thickness of speciﬁc layers that must be used in order to optimize the
function of the SAW device using the diamond interlayer, the design of the electrodes
and their fabrication and the design of the SAW devices themselves. The operation of the
Kurt J. Lesker PVD75 system will be discussed, how it works and procedures for
deposition. Finally, there will be a discussion of deposition observations and

characterization of the AIN thin ﬁlm layer.

Chapter 2

Background

2.1 Introduction
This chapter gives an overview of background information needed for

understanding this thesis. In this chapter there will be an explanation of piezoelectricin
and key related material parameters. Next, a description of what a SAW device is, how it
works, its applications and some of the performance limitations will be given. Then, a
brief description of ultrananocrystalline diamond (UNCD) is given; what are some of its
material properties and how it may improve the performance of SAW devices. Next,
reasoning is given for the choice of piezoelectric material used for this research. Finally,
there is a short review of previous work on diamond based SAW devices.
2.2 Piezoelectricity

Piezoelectric material has the property that when a mechanical stress is put
on the material it creates an electric ﬁeld. Conversely, if an electric ﬁeld is applied to the
material it produces a change in dimensions. In other words, the piezoelectric material
transforms mechanical energy into electrical energy in a reversible process. The
molecules in the crystal are polarized but are symmetrically distributed so that the charge
of the crystal is neutral. When an electric ﬁeld is applied to the material the molecules
align themselves with the ﬁeld causing a change in dimensions. If a mechanical stress (a
change in dimensions) is applied to the material the charges of the molecules are no

longer symmetrically distributed and charges of opposite polarity build up on either side

of the material creating an electric ﬁeld across the material. For example, if electrodes are
placed on a piezoelectric material and a time varying voltage is applied to them (as
shown below in Figures 1 and 2) this creates a time varying electric ﬁeld and this in turn
induces a time varying stress in the piezoelectric material. There is also a time-varying
strain associated with the stress due to the opposite polarity of the electrodes. This time-
varying stress and strain is the excitation of a surface acoustic wave; which means that
when applying electrical energy to a piezoelectric material it is converted to mechanical

energy, as mentioned previously.

 

Figure 1 Time varying voltage applied to electrodes on a piezoelectric substrate
(top view)

 

Figure 2 Tension and compression due to time varying voltage (side view)

Piezoelectricity can be described mathematically using the following piezoelectric

stress equations:

T 2 CS — eE (1)

D = 8E + e5 (2)
where T is stress, S is strain, c is the elastic stiffness constant, e is the piezoelectric stress
constant, B is the electric ﬁeld, a is permittivity and D is the electric ﬂux density. These
equations are simpliﬁed versions of the true representations (for ease of understanding);
the elastic, piezoelectric and permittivity constants are tensors. Using crystal symmetry
these independent elements are limited in number in speciﬁc cases [19]. Equation 1 is the
relation of the strain and the electric ﬁeld to the mechanical stress. Notice that if the
piezoelectric constant, e, is zero (i.e. the material is not piezoelectric) then this equation
reduces to Hooke’s Law, which is the relationship between stress and strain. Equation 2
is the relation of the electric ﬁeld and strain to the electric ﬂux density; also note that if e
is zero then this equation reduces to D = 8B, or if the electric ﬁeld, E, is zero D may still
have some ﬁnite value [19,20,21]. Stress constants can be on the order of 10
coulombs/m2 for piezoelectric materials. As the electric ﬁeld reaches dielectric
breakdown voltages on the order of 107 volts/m the induced stress can be as large as 108
Pa [20].

How efﬁciently the piezoelectric material converts mechanical energy to

electrical energy, and vice-versa, is measured by the coupling constant. The
electromechanical coupling coefﬁcient, K, is deﬁned as follows:

K2 _ MechanicalEnergy

 

. (3)
ElectrzcalEnergy

For bulk materials and bulk acoustic waves, K can be calculated in closed form using
material properties. Due to the surface acoustic waves’ inhomogeneity in the z-direction

(perpendicular to the surface), there is no corresponding explicit electromechanical

coupling factor [19]. The electromechanical coupling efﬁciency, for even the strongest
piezoelectric material, is only about 5%, but this is high enough to be practical and is the
basis for surface acoustic wave devices.
2.3 Surface Acoustic Wave Device

A Surface Acoustic Wave (SAW) device is an electronic device which can be
used, for example, as a high frequency band-pass ﬁlter or an oscillator; for this research
ﬁlters and oscillators were fabricated and tested. A simple SAW device consists of
interdigital transducers (IDT) and a piezoelectric material (i.e. quartz, zinc oxide or

aluminum nitride). An example of a simple SAW ﬁlter is shown below in Figure 3.

Interdigital Transducers Piezoelectric Substrate
Figure 3 Simple SAW device (ﬁlter)
The piezoelectric material used for the SAW devices in this thesis is aluminum nitride
(AIN).
An interdigital transducer (IDT) is a metal structure of interleaved electrodes (also
referred to as ﬁngers) with each set connected to a common contact pad, as shown in

Figure 3. The input IDT is used to initiate surface acoustic wave in a material and the

output IDT is used to detect a voltage. A time varying voltage is applied to the input IDT;
this creates a periodic distribution of an electric ﬁeld. Using the appropriate frequency of
applied voltage, the period is correlated to the spacing between the ﬁngers of the IDT.
This ﬁeld then causes a mechanical stress on the piezoelectric material which creates a
surface acoustic wave. This wave travels through the material to another IDT which
detects a voltage induced from the mechanical stress of the surface wave.

There are many applications for SAW devices. For example, a SAW device may
be coated with organic or inorganic material which allows analytes to be sorbed to the
device surface and the SAW device to act as a transducer for chemical sensing [l,2,3,4].
SAW devices can also be used in mobile phones, television tuners and optical-
communication systems [2,5,6,7,8]. SAW devices are becoming increasingly important in
wireless communications. Wireless mobile and ﬁxed transceivers require electronic
ﬁlters. SAW devices are a great choice for these applications because of their excellent
performance and small size. They have high signal selectivity, accurate bandwidths and
center frequencies, small transition bandwidths, and low passband ripple [5].

A recent trend in wireless communications is the move toward higher operating
frequencies to transmit more information. Current SAW devices not employing diamond
as an interlayer are reaching their limitations in operating frequency. The frequency at
which a SAW device can operate depends on the material properties of the media being
used to transmit the acoustic wave. For most SAW devices this is the piezoelectric
material, which is used to generate the acoustic wave as well. The operating frequency is
limited by the phase velocity of the material and the wavelength of the acoustic wave.

The wavelength of the acoustic wave is limited by the spacing between the ﬁngers of the

10

IDTs. The relationship between ﬁnger width/spacing and wavelength is shown in
Equation 5 on the following page. In order to get higher frequencies the width and
spacing of the electrodes must become smaller. In today’s fabrication technology the
submicron size of the electrodes required to increase the operating frequencies further
compromise device reliability and yield [5,9]. Another limitation of SAW devices using
conventional materials (i.e. quartz and lithium niobate) is power durability. This becomes
more of a problem with the size of the IDTs, the smaller the electrodes the less durable
the structure [7,9, 10,1 1].
2.4 Diamond Interlayer for SAW Device

As technology improves, devices need to operate at higher frequencies and by
consequence become smaller in size. As this applies to SAW devices, it has led to
research of different materials to avoid reaching the limitations of current lithography
techniques to create smaller widths for the IDT electrodes. Diamond, speciﬁcally, is the
material that is the most promising for SAW devices because it has the highest acoustic
velocity of any known material. The acoustic velocity of a bulk material is related to the

elastic constant in the following equation,

C

v = —— (4)
p
where v is the phase velocity, c is the elastic constant of the material and p is the density
of the material . It can be seen that as c increases v increases as well. Diamond has a high
elastic stiffness constant (also called Young’s Modulus) and, based on the relation of c

and v in Equation 4, a high phase velocity as well [5,7]. The phase velocity for diamond

has been calculated to be approximately 18,000 m/s, for the longitudinal velocity, and

11

between 10,500 - 13,000 m/s for the two shear velocities [22]. This is important because
the frequency of the device is related to velocity by the equation,
v

f = Z (5)
where f is the frequency, v is the phase velocity and it is the wavelength
[1,5,7,9,11,17,18,23,24,25,26]. As can be seen from this equation, with an increase in
velocity there is an increase in frequency. This means that by simply using diamond as an
interlayer, in a SAW device, and keeping the same dimensions of the IDTs there is an
improvement in performance. The width and spacing of the IDTs is related to the

wavelength of the acoustic wave by the equation:

IDT... =§ <6)

[5,7,18]. This means by using a larger width for the ﬁngers on a SAW device with
diamond as an interlayer the same frequency can be achieved as with ﬁngers of a smaller
width on SAW devices without diamond as an interlayer. Another improvement that
could be realized using diamond would be increasing the quality factor (Q) of a resonator

which can be approximated by the equation:

 

Q = i (7)

where f0 is the operating (center) frequency and Af is the full width at half maximum

value of the peak [27]. Diamond also has other beneﬁts for use in SAW devices; it has a

high thermal conductivity which will allow for higher power capabilities. SAW devices

on diamond have a high electromechanical coupling coefﬁcient, which indicates a high

efﬁciency of energy conversion from the input signal to the SAW [5,7,8,9,10, 12,13,14].
For this research we have chosen to use ultrananocrystalline diamond as an

interlayer for the SAW devices. Ultrananocrystalline diamond (UNCD) is characterized

12

by a grain size of 2 - 10 nm and is grown by a process developed at Argonne National
Laboratories by Gruen and Krauss [15,16]. UNCD has the beneﬁt of a low surface-
roughness which will eliminate or greatly minimize the need for polishing the diamond
which takes a great deal of time in the fabrication process. This low surface roughness
will also minimize propagation loss of the SAW energy which leads to less insertion loss
for the device and lower power consumption. Due to the small grain boundaries of the
UNCD this lowers the expected acoustic scattering seen at large angle grain boundaries
in polycrystalline diamond, especially if grain dimensions are of the same size/length as
the acoustic wavelengths, device apertures and transducer separation [17,18]. The
thermal conductivity of UNCD however, is low. This could be contributed to the
wavelength of the thermal phonons being equivalent to the grain boundaries. However,
despite the lower thermal conductivity of UNCD it still offers the potential of diamond-
based high frequency performance to SAW devices.
2.5 Piezoelectric Material for SAW Device

Due to the fact that diamond is not piezoelectric, a piezoelectric material must be
used to excite a SAW which may then propagate along and near the surface of the
diamond interlayer. Surface acoustic waves, as the name suggests, travel on or near the
surface of the medium. The atoms in the medium have an elliptical shaped displacement
and this displacement drops to zero within a few wavelenths [5]. Using a thin-ﬁlm
overlayer can cause high dispersion for sound propagation and admits a multiplicity of
allowed modes. However, successful devices using layered structures with diamond as an
interlayer have been reported [18]. AIN is the piezoelectric material used for this

research. It has the ability to be sputtered on a substrate in a reactive environment and no

13

special equipment is required to accomplish this; unlike the case for zinc oxide (ZnO)
which needs a special vacuum pump in order to sputter in the reactive environment with
oxygen. Sputtering aluminum nitride also does not require the removal of hazardous
waste gases. Aluminum nitride with a diamond interlayer when compared with zinc oxide
with a diamond interlayer has a higher acoustic velocity and a higher electrical resistivity
[17].
2.6 Previous Work on Diamond Based SAW Devices

Currently there are two groups who have studied SAW structures using UNCD as
an interlayer to take advantage of its high acoustic velocity [17,18]. Bi, Huang, Asmussen
and Golding [18] studied a layered system of ZnO/UNCD/Si to determine the use of
UNCD as a possible substrate for SAW devices. It was found that several modes exist for
a given value of khZHO and that they are highly dispersive at small values of ktho. For
large normalized thickness (kh, where k is 2m and h is the thickness of the ZnO) of the
ZnO layer the phase velocity of the Rayleigh wave approached that of ZnO and as ktho
approached zero the velocity of the Rayleigh wave approached that of diamond. It was
also shown that the SAW devices on UNCD and on polycrystalline diamond substrates
are indistinguishable, within combined uncertainties [18]. Bénédic, et a1 [17], reported for
the ﬁrst time the successful fabrication of AlN/UNCD/Si layered structure based SAW
devices. It was shown that, related to the wave penetration depth, the acoustic velocity
depends on the IDTs periodicity and on UNCD ﬁlm thickness; this means that by
reducing the IDT ﬁnger size, there is no need for a thick layer of UNCD ﬁlms to

eliminate the silicon substrate inﬂuence. A SAW velocity of 9,500 m/s was realized for

14

an UNCD layer of 19 pm thick and a wavelength of 32 pm. This velocity is higher than
the velocity obtained for the 0th Rayleigh mode ZnO/UNCD/Si structure [17].

Several investigations have studied SAW structures with non-UNCD diamond.
Elmazria, et a1 [23], investigated fabricating high velocity SAW devices using AlN on
CVD diamond. They found that using their technique there was a good combination of
high electromechanical coupling coefﬁcient and temperature coefﬁcient of frequency
(TCF). TCF is the measure of the stability of the frequency with a change in temperature
and it was noted that as khAlN (which is the wave number times the thickness of the A1N
layer) increases the TCF decreases, which means greater operating frequency stability as
temperature changes [23].

Iriarte, et al [28], using parameter extraction, found that for a 4.3 pm thick layer
of AlN ﬁlm deposited on polycrystalline diamond the SAW velocity was about 10,000
m/s and for thinner ﬁlms of about 2 pm thick the velocity increased to about 11,800 m/s.
This means that SAW devices operating in the microwave region are feasible using
standard fabrication techniques. It was also shown that there is a relatively high effective
coupling coefﬁcient of about 1% [28].

Higaki, et a1 [10], investigated the high power durability of a ZnO/diamond/Si
based SAW ﬁlter as compared to a LiTaO3 (lithium tantalate). It was shown that the
ZnO/diamond ﬁlter operating at 2.9 GHz, maintained linearity up to the input power of
36 dBm. The LiTaO3 ﬁlter operating at 822 MHz was nonlinear above 23 dBm and
incurred severe damage at input power of 27.7 dBm. The ZnO/diamond ﬁlter operating at

a frequency 3.5 times higher than the LiTaO3 ﬁlter was able to withstand 8 dBm higher

15

input power than that of the LiTaO3 ﬁlter as well. These results are assumed to be due to
the high thermal conductivity and structural integrity of diamond [10].

Although this is not an exhaustive review of previous diamond based SAW device
work, it does provide a sampling of background work pertinent to this thesis. This thesis
further explores materials and device fabrication issues relative to AIN/UNCD SAW

Sti'UCtUI'C S .

l6

Chapter 3
SAW Fabrication Considerations and Procedure

3.1 Introduction

This chapter will explain the procedures used to fabricate the SAW devices. In
this chapter, reasoning is given for the choice of ﬁlm thickness, i.e. what thickness of
aluminum nitride to use and what thickness of diamond should be used? Next, there will
be an explanation of the design of the electrodes (IDTs) and the process used to fabricate
the electrodes (e. g. the deposition of the aluminum, masking and etching). There will be a
discussion on the design of the SAW structures. This chapter will also give a physical
description of the network analyzer used to characterize the SAW devices. An
explanation will be given of the operating and calibration procedures and a description of
the probe station and probes, which are part of the characterization setup. Finally a
discussion of the results obtained from aluminum electrodes fabricated on quartz.
3.2 Film Thickness Considerations

The ﬁlm thickness of the multiple layers of the SAW device is a critical factor
when designing the device. It is important because, as the name suggests, surface acoustic
waves travel along and near the surface of the material but disperse rapidly to zero within
a few wavelengths from the surface [1,5,19,29,30]. The thickness of the diamond is
important because it needs to be thick enough so that the silicon substrate beneath it does
not affect the velocity of the acoustic wave and so that negligible acoustic energy is lost

to the silicon substrate. The thickness of the aluminum nitride is important because it is

17

related to the electromechanical coupling coefﬁcient and the overall SAW velocity. The
desired thickness of the diamond and the aluminum nitride layers were found based on
theoretical calculations by Benetti, et a1 [30] and Nakahata, et al [14]. The layers for the

SAW device are as shown in Figure 4.

Al — Electrodes

AIN — Piezoelectric Layer
UNCD - Ultrananocrystalline Diamond

  

Silicon - Substrate

Figure 4 SAW device layers

The layer of diamond needs to be thick enough so that the silicon layer has no
effect on the performance of the SAW device; speciﬁcally on the dispersion of the SAW
and the velocity of the SAW. From previous work [14], a minimum normalized diamond
thickness of 4 is appropriate for phase velocity considerations,

khclia 2 4 (8)

where k = 21V). and h is the thickness of the diamond. From Nakahata, et al [14], the
velocity of the surface wave does not appreciably increase for thicker diamond layers.
Utilizing a normalized thickness of 24, in a layered system, yields a near maximum phase
velocity. For this research we chose a wavelength of 16 um; inserting this into the

equation 8 gives a minimum thickness of approximately 10 pm for the diamond layer.

Additionally, as mentioned previously, energy can be lost to the silicon if the diamond is
too thin. Benetti, et al [30], recommend that the diamond be “at least a few acoustic
wavelengths,” thick to avoid energy loss to the silicon, which would correspond, in our
case, to 48 um using a thickness of 37». In summary, the diamond ﬁlm should be at least
10 um thick and preferably thicker.

For this investigation, the desired center frequency, for a device on quartz, is
approximately 500 MHz. Quartz has a phase velocity of 3200 m/s; using Equation 5, the

corresponding wavelength is found (as shown below) to be 16 pm.

 

,t _ 3.2x103m/s
5.0x108 ls
xi =16,um

Now, using Equation 6 the width of the IDT’s was found, as shown below.

16
IDTwidth = J"-

4
I D Twidth = 4W"
These parameters, wavelength and IDT width, were then used to ﬁnd the necessary

thickness for the diamond layer. Using Equation 8 and the relation k = 21:0», the

minimum thickness for the diamond was found, the calculations for this are shown

below.
-2:
16
k z 0.4
4
h . __
‘1‘“ 0.4

19

hdia 2 10m

The thickness of the AIN is important as well, because if the A1N is too thick the
SAW will not reach the diamond and the full beneﬁts of having the diamond layer will
not be realized. The electromechanical coupling coefﬁcient also depends on the AIN
thickness. For analyzing, the thickness of the A1N (aluminum nitride) the same
normalization (kh) is used. Benetti [30] has calculated K2 values versus AIN thickness for
semi-inﬁnite diamond and Nakahata has calculated K2 values versus ZnO (zinc oxide)
thickness for thinner diamond layers. Based on the former, a normalized thickness of 0.2,
for AIN is used; this is where the maximum coupling coefﬁcient is realized (for the
layered structure used in this research)

For a semi-inﬁnite diamond layer and a normalized AIN thickness of 0.2, the
theoretical velocity of approximately 8500 m/s for the Rayleigh mode; higher velocities
and coupling coefﬁcients can be obtained for the Sezawa modes [30]. Using equation 5
(in Chapter 2) and substituting in the above mentioned phase velocity and a wavelength
of 16 pm, the operating frequency of the SAW device should be about 531 MHz. The
maximum coupling coefﬁcient calculated for this layered structure is K2 = 0.0015 or
0.15%.

3.3 Electrode (IDT) Formation
The fabrication of the electrodes is a multi-step process. Figure 5, shown on the

following page, illustrates the step-by-step process of fabricating the electrodes.

20

 

 

 

 

 

 

 

Unexposed PR

   
   

Exposed PR

 

4. Develop the PR and hard bake

 

 

 

 

 

 

 

 

6. Remove the PR with photoresist
stripper

5. Etch (pattern) the Al layer to form
the interdigital transducers that make
up the SAW device

 

 

 

Figure 5 Fabrication steps for forming Al electrodes

21

To begin, the wafer needs to be cleaned using acetone, methanol and isopropyl alcohol. If
the wafer is not clean before the aluminum deposition then both poor adhesion and breaks
in electrode patterns may result. Place the wafer in acetone and put it in an ultrasonic bath
for 15 minutes, rinse with de-ionized water and dry with a nitrogen gun. Then place the
wafer in methanol and put it in an ultrasonic bath for 15 minutes, rinse with de—ionized
water and dry with a nitrogen gun. Use isopropanol (isopropyl alcohol, IPA) to evaporate
the residual water from the wafer surface, dry with a nitrogen gun and put into a 120
degree C oven for 20 minutes to dry.

Next is the deposition process, which is accomplished using DC pulsed
magnetron sputtering on a Kurt J. Lesker system (the PVD75). This system and the
sputtering process used will be discussed in further detail in chapter 4. The PVD75
system was pumped down to a base pressure of less than 4x10'6 Torr, the chamber
pressure during sputtering was set to 3 mTorr and argon was used to sputter the
aluminum. The parameters set on the PVD75 system are shown in Table 1 — 3.

Table 1 Parameters for the Sparc—le V unit in the PVD75

 

 

 

Arc Frequenc Revers Crowba Revers Contro Subs
Handlin y e Time r Delay e 1 Mode trate
g Voltage Rota
tion
Sparc
-le V Self Run 20 kHz 3 us 10 us 10 % Local 5

 

 

 

 

 

 

 

 

 

Table 2 Parameters for the MDX 1.5 kW power supply to the cathode of the PVD75

 

 

Power Ramp Time
(Constant Power Mode)
MDX 1.5 kW 800 W 5 5
(Power Source)

 

 

 

 

 

Table 3 Parameters for inside the chamber of the PVD75

 

 

 

 

 

 

 

Base Pressure Operating Gas Flow
Pressure (Argon)
Chamber < 4* 10" Torr 3 mTorr Mode 2

 

22

An explanation of parameter settings will be given in more detail in Chapter 4 when the
operation of the PVD75 will be explained in detail. Only a thin layer, approximately 200
nm, of aluminum is needed to fabricate the electrodes. Using the above parameters, the
deposition rate for aluminum is about 55 nm/min; in order to deposit 200 nm of
aluminum then the power supplied to the target needs to run at 800 W for about 4
minutes.

Next, to form the IDTs and hence the SAW device, the aluminum needs to be
patterned. To pattern the aluminum the substrate needs to be coated with photoresist (a
light sensitive material), which once exposed to light through a mask (blocks the light in
a speciﬁed pattern) will create the desired pattern when it is developed. Before coating
the wafer it must be cleaned to ensure that the electrodes form properly and don’t end up
with defects such as breaks in the electrodes. The wafer is cleaned using the acetone,
methanol and IPA procedure mentioned previously. The substrate is then coated with a
thin layer of AZ ADH Promoter, for better adhesion of the photoresist to the surface,
using a 5 mL pipette. A programmable spinner is used to spin off the excess promoter
and creates a thin and uniform layer of promoter across the wafer. Using the spinner
ensures that the layer spun on is the desired thickness, determined by the time and rate of
rotation, and ensures that the layer is uniform. The promoter is spun on at 3000 rpm for
30 seconds. Next, the substrate is then coated with 81813 positive photoresist. Positive
photoresist means that when it is exposed to light the photoresist rinses away when it is
developed. The wafer is covered with a thin layer of photoresist using a 5 mL pipette,
spun for 30 seconds at 3000 rpm and the resulting layer of photoresist is approximately

1.23 pm thick. The thickness of this layer is important because the thicker the photoresist

23

is the longer it needs to be exposed, or a higher power of exposure is required to ensure
that the light penetrates the entire layer of photoresist. However, if the layer of
photoresist is too thin it will not resist the acid etch used to pattern the Al. The substrate
is then soft baked in a 70 degree C oven for 20 minutes.

Next, the photoresist is masked, exposed and developed in order to pattern the
photoresist which will eventually pattern the aluminum and form the SAW devices. Two
masks were designed. A SAW Mask was designed with 4 and 6 pm width electrodes and
a Backup SAW Mask was designed with 8 pm width electrodes in the event that the
dimensions on the SAW Mask could not be attained. Each mask has 24 die on it and each
die has 5 different devices. The die is copied 23 times to make up the rest of the 24 die.
The devices and their designs will be explained in the next section. The SAW Mask is
shown on the next page in Figure 6. The Backup SAW Mask has the same general form,

7 rows of die with a total of 24 die.

24

 

 

 

E 'J {— '3 ‘ "" r ‘1,
am ﬁxﬂ gig ~~~~~ _,_.

7‘— '" " ’t “1' ‘ r '5 r r." r '3 ‘- -' r 1
{TI-“'2 "" "" Fl". “ :" Fl") "‘ :" t-IT _ :
m -.__-—__-F"—.__2 ‘-—\. 3,
ca. .1 r- n H.- .—F F .1) x... — r .1) n. _r~ J" 3
(it"s ~ 5‘: 5:11 r:- 5 Fin n— — r11 9 5
=4: ._"‘ _.__.;.._L _::£-
“’53 Err." (:1."7: pqtfr-t
at". n- in $12. :- .—." 5:17: ~ : r. lrzs : :
====§ 22% ____..-~-‘—* $31
n: 3’ d;— 35 n: 3P- f: :5 n: "I c' —i ‘3‘ In It: '5
Fir—3 ': :5" Fin .2: .7.’ 4:1“: :- z"- Flt-s *- :
==i==t #ﬂ _...._..-——‘-“- m1
7: 8" 4;: a F: r.“ 4:: :23 ,3: :37." '5 M f .z‘ :5
631:) a: z" 13 n: :r

g ..
.- JI é; aha/1‘; ‘ 1...“ ”5

Figure 6 SAW Mask with 24 die, each with 5 devices

The wafer is exposed to UV light for 1.3 minutes using a Karl Suss MJB3 mask
aligner. The wafer is then placed in 50 mL of MF-3 l9 developer for approximately 60
seconds. The developer removes the photoresist which has been exposed to light and has
no effect on the photoresist which was not exposed. The substrate is rinsed with DI water,
a nitrogen gun is used to remove the excess water and then it is hard baked in a 110
degree C oven. The hard baking process hardens the photoresist so that it can withstand
the acid etch.

Now the aluminum can be etched in the pattern of the photoresist. An aluminum
etchant (consisting of 80 mL of phosphoric acid, 10mL of water, 5 mL of nitric acid and
5 mL of acetic acid) is heated and the wafer is left in the etchant for 2 minutes or until all

of the exposed (not covered by photoresist) aluminum has been etched away. Extreme

25

care must be taken during this step, so that the aluminum is not over etched. If the
aluminum is over etched then undercutting can cause the ﬁnger width to decrease or may
cause the ﬁngers to be etched away completely. If this happens the SAW device will not
work as expected or may not work at all. Now that the aluminum has been etched, not
only are the IDTs formed but now the SAW devices have been created as well.
3.4 Electrode (IDT) Design

In order to fabricate a working SAW device, there must be a way to excite a
surface acoustic wave in the piezoelectric material. Exciting a surface acoustic wave is
achieved using IDTs which are formed with contact pads, a bus bar and electrodes (also
referred to as ﬁngers). The distance between the IDTs, the number of electrodes, length,
overlap and width are designed to obtain the desired performance of the SAW device.

For this investigation, the parameters listed above are based on previous works
[16,17,28]. Fifty pairs of electrodes were used for each IDT (input and output) for the
ﬁlters and a total of 200 pairs of electrodes were used for the oscillators. The distance
between input and output IDTs is 1 mm. The length of the electrodes is approximately
500 um and the overlap distance of the electrodes is 492 pm. The width (and spacing) of
the electrodes has an effect on the wavelength of the SAW. The width of the IDT is
related to the wavelength by equation 6, as noted in Chapter 2. This means the IDT width
can be calculated for a given wavelength [5,7,18,25]. In this research, an IDT width and
spacing of approximately 4 mm was used, based on a desired wavelength of 16 um. There
are also other parameters pertaining to the electrodes and IDTs which can be optimized

for performance of the SAW which are beyond the scope of this research. The designs

26

used for the ﬁlters and oscillators were taken from a commercially produced SAW
device.

Five different structures were made altogether; close up views of the devices are
shown below in Figure 7 and Figure 8. Both the SAW Mask and the Backup SAW Mask

are shown; Figure 7 is the SAW Mask and Figure 8 is the Backup SAW Mask.

I -._.>{'I"‘J 5‘“ ”“1553 “'“‘ J" " 4D |
“—3.

..._.__L_ ——.—'AL.—A-

 

- El
-_’.~“--:” refs: .1 {—7.

.511 .r-r- '3" III—‘1: ."b :”-r

:=_% __._-—5_‘:—T="___
--
...":_i.g:::5,1_-xg_—.:-5

Figure 7 Four die on the SAW Mask

 

 

Figure 8 Four die on the Backup SAW Mask

The signiﬁcant difference between the SAW Mask and the Backup SAW Mask are the
widths of the ﬁngers and the number of ﬁngers on the IDTs. In Figure 7, A, C, D and E
are ﬁlters that have electrodes 4 pm wide and have 50 pairs of electrodes per IDT; the
ﬁlter labeled F has electrodes 6 pm wide and has 50 pairs of electrodes per IDT; the

oscillator labeled B has 200 pairs of ﬁnger and has 73 reﬂectors (gratings) that are 4 pm

27

wide on each end. In Figure 8, G and I are ﬁlters that have electrodes 8 pm wide and have
20 pairs of electrodes per IDT; the ﬁlters labeled H and I have electrodes 8 pm wide and

have 40 pairs of electrodes per IDT; the oscillator labeled J has 100 pairs of electrodes.

 

Figure 9 Filter with 20 electrode pairs per IDT at 8 pm wide
In Figure 9, above, is a ﬁlter from the Backup SAW Mask. The contact pads for
the devices are designed with dimensions compatible with the high frequency probes
used to probe the devices for characterization. From the top arrow to the bottom arrow
the contact pads are ground, signal, ground. Figure 9 shows the basic design for the ﬁlters
on both masks and the difference between ﬁlters is some variation on this design. These
variations are shown in the following ﬁgures. Figure 10 has a gounded metal bar placed

approximately halfway between the input and the output transducers.

 

Figure 10 Filter with 50 electrode pairs per IDT at 4 mm wide

Figure 1 1, shown on the following page, the ﬁlter has a serpentine inductor attached to
the signal and ground contact pad of the IDT. There is an inductor attached to each of the

IDTs.

Figure 11 Filter with a serpentine inductor added to the contact pads
Figure12, below, is a magniﬁed view of the serpentine shape of the inductors attached to

either IDT.

 

 

Figure 13 shows a ﬁlter with another serpentine type inductor attached to either IDT.

This inductor design was copied from a commercial SAW device.

Figure 13 Filter with grated, serpentine inductor attached to contact pads of ﬁlter

Figure 14 is a magniﬁed view of what the inductor looks like and shows how it differs
from the simple design of the inductor shown in Figure 12. The inductor shown in Figure
14, on the following page, has gratings in the length of the inductor; the inductor in

Figure 12, above, is simply a length of solid metal.

29

 

Figure 14 View (50x) of grated, serpentine inductor attached to contact pads of ﬁlter
Figure 15, shown below, is the general design for the oscillators and the difference
between the oscillators on the two different masks is the width and number of electrodes

and the width and number of reﬂectors at each end.

 

Figure 15 Oscillator with 100 electrodes 8 pm wide and 36 reﬂectors at each end

The design of the IDTs for this research has been simpliﬁed, in one important
regard, from the design used in the commercial device. The electrodes used in this
research are uniform, meaning they are all the same size, shape, length and width. A
simpliﬁed view of the IDT design for this research is shown in Figure 3. The electrodes
for the commercial device are of different lengths forming the shape of a sinc function. A
drawing of the IDT design for the commercial device is shown on the following page in

Figure 16.

 

Figure 16 IDT design of the commercial device used in this investigation

The simpliﬁed design of the IDT was used instead of the commercial design, in order to
simplify the process of designing the mask. The difference in the designs of the
electrodes affects the impulse response (transfer function) of the ﬁlter. The output signal
of the ﬁlter is related to the Fourier transform of the overlap of the electrodes that make
up the IDTs. This means that for the design of electrodes shown in Figure 1 the output

signal will be a sine function as shown in Figure 17.

31

l

I
I
I
I
I
I
l
I
I
I
I
A

v v v v v v v v v

Figure 17 Sinc function output corresponding to electrode pattern in Figure 1
For the electrode design used in the commercial device (shown in Figure 16) the output

signal will be a bandpass signal as seen in Figure 18 [5,19].

 

l
l
!
l
l
l
!

 

 

 

Figure 18 Band pass output corresponding to electrode design in Figure 16

32

Chapter 4

Aluminum Nitride Deposition and Characterization

4.1 Introduction

This chapter will explain the PVD (physical vapor deposition) process of pulse dc
sputtering, what is it/how does it work and why is it important, as compared to other
sputtering methods. There will be a description of the sputtering system used, that was
developed by Kurt J. Lesker, and of the operating procedures. Finally, there will be a
discussion of the results of sputtering an aluminum nitride (AIN) layer on a Si substrate;
there will be a comparison of the properties of the sputtered layer vs. the sputtering
parameters.
4.2 Pulse DC Sputtering

There. are different types of sputtering methods; for example, dc, rf (or ac),
reactive, magnetron and other variations and combinations of these methods. In this
investigation we used pulsed dc magnetron sputtering in a reactive process in order to
produce a thin ﬁlm of AIN on our substrate. For pulse dc magnetron sputtering, there is a
negative dc voltage applied to the planar circular cathode, which is where the target sits.

A cross—section of the chamber setup is shown on the following page in Figure 19.

33

 

 

 

AA: 5
'u

I I‘. ;,' ; TL
.. '8 “tﬂ’ﬁa

- 1 Dark Space

.1,”

Magnets (located
below the target)

 

Figure 19 Cross-section of the PVD75 deposition chamber

Under the target and cathode there are also magnets. One magnet is centered
under the target and a magnet of opposite polarity encircles the perimeter of the target.
This set up creates magnetic ﬁeld lines from the center of the target to the perimeter of

the target. Figure 20, below, shows a simpliﬁed cross-section of the magnetic ﬁeld lines.

34

N Magnetic Field Lines
S

 

Target Magnetron
Figure 20 Simpliﬁed cross-sectional view of the magnetron’s magnetic ﬁeld lines
The magnetic ﬁeld crossed with an electric ﬁeld causes the electrons to travel in a helical
path around the magnetic ﬁeld lines. This lengthens their mean free path before they are
lost to the anode or recombination and hence allows for more collisions [31]. The high
energy electrons collide with the argon atoms and produce ions, and other species within
the plasma, and hence the magnetic ﬁeld helps to increase ionization in the area where
the ﬁeld exists. These ions are then accelerated towards the target and eject atoms from
the target when striking it. The atoms are adsorbed to the surface of the substrate but the
atoms travel in all directions and coat other surfaces of the chamber as well. The
localized ionization in the region of the magnetic ﬁeld lines causes the plasma to form in

a ring shape (also called a race track) on the target as shown below in Figure 21.

35

Impurity
deposition

Figure 21 Race track shape erosion on sputtering target

 

As sputtering occurs some particulates, insulating layers and impurities get
deposited onto the target as well as other surfaces in the chamber, which can be seen in
Figure 21. These impurities cause a charge to build up on the target (and other surfaces).
In order for the charge build-up to be discharged an arc occurs. These arcs can either be
hard arcs or micro-arcs. A micro-arc is an are that is eliminated with one reverse pulse
and a hard arc is one that is eliminated with multiple reverse pulses [32]. The pulsing
frequencies are usually between 10 and 350 kHz and the duty cycles are commonly

between 50-90% [33]. A diagram of the pulsing voltage is shown below in Figure 22.

36

trev

l—i 3;“

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 22 Pulsed DC voltage supplied to the cathode of the PVD75

When the negative dc voltage is applied to the cathode it is referred to as the “on” time

(ton). The “on” time is determined by the reverse time and the frequency set by the user.

The reverse time (trey), as labeled in Figure 22, is the duration of time that the polarity of

the voltage is reversed. The voltage is not 100% reversed, for our machine the options for
the percentage of reverse polarity are 10%, 15%, and 20%. Reversing the voltage during
the reverse time allows the target to be discharged and when the parameters are optimized
for the process, prevents arcing from occurring. Limiting or eliminating arcs altogether is
important because arcing contributes to particulate formation, which if embedded in a
thin ﬁlm will cause uniformity problems, and can damage the power supply [32,33] The
parameters and their settings used in this research will be explained in detail in the
following section.

Pulsed dc magnetron sputtering was chosen for its beneﬁts over other processes.
RF sputtering is usually used when sputtering dielectrics due to the charge build-up and

hence arcing issues. RF sputtering has the drawback of low deposition rates. Pulsed do

37

magnetron sputtering has the beneﬁt of arc suppression and increased deposition rates as
compared to RF sputtering and ion beam deposition which have low deposition rates.
Due to the increased ionization from the magnetron the sputtering rate is increased,
thereby increasing the deposition rate as well. Using a reactive process is also possible
with this sputtering process as opposed to ion beam deposition or e-bearn deposition. In
short, pulsed dc magnetron sputtering is employed in this research due to its increased
deposition rates, are suppression capabilities and the ability to use a reactive process
when sputtering.
4.3 PVD75 Description

The PVD75 is a physical vapor deposition machine that is produced by the Kurt I.
Lesker Company. A picture of the PVD75 used for this investigation is shown on the

following page in Figure 23.

38

   
 
   
  

Touchscreen
Control System

 

1. DC Power
SUPply

2. Spare-1e V
3. Substrate

Heater
Shamber 4. Platen
rewport Rotation

 

 

 

Vacuum
Rough Pump
Access

 

 

Figure 23 Front of PVD75 system

As can be seen, there is a touch screen and a chamber viewing window located in the
chamber door, below this door is the door to access the vacuum roughing pump. On the
right hand side is the door to access such parameter controls as the dc power supply
(MDX 1.5 kW), arc suppression (Spare-1e V; frequency, reverse time, reverse voltage,

etc.), the substrate heater (M.A.P.S Power Supply) and the substrate (platen) rotation.

39

 
  
  
 
 
   

‘ Substrate \-. z: 0'
Heaters

[1!" I“

 

Substrate Holder
- Platen

 

Cathode

 

Figure 24PVD75Ch;rnber
Figure 24 shows the inside of the chamber. As can be seen, the chamber has a “D-
shaped” volume. The heaters can heat the substrate to a maximum of 350° C for a time of
4 — 6 hrs. The thermocouples are located next to the heaters. The platen or substrate
holder can be a distance of 15 cm from the target or 7.5 cm from the target. For this
research, it is a distance of 7.5 cm from the target. The platen rotation speed ranges from

0-10. There are no units for the rotation speed, but as is mentioned later in this chapter at

40

a speed of 5 the rotations per minute is approximately 41, so the assumption would be
that at a speed of 10 the rpms would be about 82, double the speed at 5. The other item
within the chamber is the cathode. The cathode is where the target sits and it is water
cooled in order to keep the heat of the target transferring to the rare earth magnets of the
TORUS source. If the cathode is not properly cooled then it could result in damage to the
magnets.

High purity argon (99.999%) is used as the sputtering gas because it is a heavy,
inert gas, which prevents it from reacting with the sputtered atoms and creating a ﬁlm
with a high amount of impurities. High purity oxygen (99.99%) and nitrogen (99.999%)
are connected to the PVD75 to be used in reactive processes. For this investigation
nitrogen is used to react with the aluminum atoms sputtered from the target to deposit as
a ﬁlm of AIN on the substrate. Nitrogen, not high purity, is also used to operate the
pneumatic valves and shutter and for system venting.

The touch screen has four different screens for the operator to use; vacuum,
deposition, gas and maintenance screen. The vacuum screen is the default screen that
comes on when the machine is initially turned on. The possible operations from this
screen are pump down, vent, abort, changing the color scheme of the screens and
switching to one of the other 3 screens (maintenance, deposition and gas). The vacuum
screen is shown on the next page in Figure 25. Upon initial start-up of the PVD the

vacuum screen is the default screen.

41

      

"till . . . . . . 3 n. in, ,. g mil-‘3‘.
{WW »‘ . ‘1 .;
*i '.- ‘ , ‘9' ., 1‘1 - ‘ ‘ . " IIIIII‘IlIlIt‘ I :2 mi

tilbﬁm .. i 3' ' -. r “ ' ' I ll "‘" “" ‘ ‘ It

in. ii». \
. W
. - 'l . _

Status Message
Bar

 

 

 

 

 

 

 

 

 

Figure 25 Vacuum screen

The items that aren’t labeled already (on the screen), or for which their function/purpose
is unclear to the user I have labeled, in Figure 25, and all icons are explained in detail.

The status message bar displays the individual steps in a sequence as they are
being executed and interlock information when they have not been satisﬁed. When a
sequence has been initiated the message bar displays the current step in that sequence.
When an interlock has not been satisﬁed the message bar displays the interlock
information.

The status light informs the user of the mode the machine is in by the color of the
bar. If the bar is green, this represents standby mode. When the status light is blue the
machine is in clean mode or a sequence is running. When clean mode has been initiated,

(from the maintenance screen) all operations through the touch screen will be disabled for

42

30 seconds enabling the user to clean the touch screen without initiating/deactivating
other modes. This mode cannot be aborted and all interlocks running before initiation of
this mode continue running. When the status light is yellow an interlock has not been
satisﬁed and red means that a sequence has failed, or that the abort mode has been
initiated [34].

Touching the pump down icon initiates the pump down sequence to begin
pumping down the machine. During the sequence no other sequence may be initiated
except the abort sequence. It takes about 10 minutes for the sequence to be completed;
this means that the machine has reached a pressure of less than 5 x 10'5 Torr [34]. It takes
approximately 2 hrs to reach a pressure of ~6 x 1045 Torr, and it takes about 3 - 4 hrs to
reach a pressure of 2 x 10'6 Torr.

Pressing the vent icon will begin the venting procedure and no other sequence
may be activated during this mode of operation. It takes approximately 5 minutes for the
PVD75 to reach atmospheric pressure. After atmospheric pressure has been reached the
chamber door may be opened and closed.

When the WRG pressure display is touched, the pressure will be displayed in bar
on all screens, if the pressure was displayed in Torr prior to being touched. If the pressure
was displayed in bar, prior to being pressed, it will be displayed in Torr. This icon is left
in Torr to avoid confusion caused by switching between the two pressure units.

The abort mode is accessible from every screen. However, the abort mode is
disabled when the screen clean mode is enabled. Pressing this icon on the touch screen
will initiate a series of steps including closing the vent, rough, throttle and gas valves off,

turn off the turbo molecular vacuum pump, set MFC (mass ﬂow controller) mode to 0

43

and the MFC ﬂow to 0%, the pressure set point is set to 0 Torr (or bar) and any sequence
running is deactivated [34]. After this sequence of steps has ﬁnished running the PVD
returns to standby mode.

In the G ON SCH mode the icon colors are green and red. Green represents on or
open and red represents off or closed. Touching this icon will switch the color scheme to
the R ON SCH. In the R ON SCH the icon colors are blue and red. Blue represents off or
closed and red represents on or open.

Provided the pump down and vent sequences are not running and the clean screen
mode has not been activated, any of the other three screens may be accessed. Normally,

the next screen selected is the gas screen.

Pressing the gas icon will switch the user to the gas screen shown in Figure 26.

 

     

  

8. Be+82mT
CRPMFIN

 

 

MFC Control
Box

 

 

 

 

 

 

 

9.2 st 3.9 2
FL SP 2 can

“ nrc:
3.1 zrs a
FLnu none .

 

 

     

 

Figure 26 Gas screen

From the gas screen the user can set the pressure at which to sputter the ﬁlm, the mode to
run the MFC’s and switch to the vacuum or deposition screen. There are two mass ﬂow
controllers (MFC) labeled MFCl and MFC2. For this investigation, MFC2 is the
sputtering gas, high purity Ar (99.999%), and MFCl is connected to high purity nitrogen
(99.999%) and oxygen (99.99%) for use in reactive processing. The high purity oxygen
has not been used (at this time) for processing; only high purity nitrogen is used for this
research to deposit an AIN ﬁlm.

There are 3 modes to run each MFC and they are labeled 0, 1 and 2. The mode
can be changed by touching the number labeled mode. It will become highlighted and can
be changed using the IN C/DEC icons. If the INC/DEC buttons are not touched for 5
seconds the number will be un-highlighted and must be touched (highlighted) again
before it can be changed. Mode 0 and 1 are ﬂow modes (i.e. the system adjusts to
maintain a speciﬁed ﬂow) and mode 2 is a pressure mode (i.e. the system adjusts to
maintain a speciﬁed pressure). MFC 1 and 2 can be in mode 0 at the same time, but both
MFC’s cannot be in mode 1 or 2 at the same time; this is because a circular relationship is
not allowed by the software. Mode 0 allows the user to control the ﬂow of gas, for each
channel, by using a set point. For each channel, the number in the upper left corner of the
MFC control box marked FL SP is the ﬂow set point and can be changed by touching the
number, it then becomes highlighted and can be increased or decreased to the desired
ﬂow using the INC or DEC icons. Valid inputs for the ﬂow are 0 — 100% of full scale.
For the nitrogen processing gas full scale is 100 sccm (standard cubic centimeters per
minute) and for the argon processing gas full scale is 139 sccm. Mode 1 is the “slave”

mode; this means that the MFC in mode 1 is “slaved” to the other MFC (called the

45

“master”), which must be set to mode 0 or mode 2. The MFC set to be in mode 1 will
ﬂow a percentage of the ﬂow rate set for the master MFC and the throttle valve will
adjust the pressure as needed. This percentage is the contribution percent labeled %
CONT and located in the upper right hand comer of the MFC Control Box, in Figure 23.
To set this percentage touch the % CONT number to highlight it and using the INC/DEC
icons Set the contribution to the desired percentage. For example, if MFC] is ﬂowing
40% of full scale and MFC2 is slaved to it with a percent contribution of 50%, then
MFC2 will ﬂow 20% of full scale. Mode 2 is a pressure mode. In mode 2, the desired
deposition pressure is set by the user and values from 0 — 1000 mTorr are valid. Touching
the number labeled SP, in the upper left hand comer of the touch screen, will highlight it
and then set the chamber pressure to the desired pressure. Now the total gas ﬂow adjusts
to maintain a set pressure and the slave MFC will contribute a percentage of the ﬂow of
the master MFC.

When all the desired ﬂows and or pressures are set touch the GAS valve icon to
turn the output on (it will be green in color when it is on) and allow gas to ﬂow to the
chamber, then touch the THROTTLE valve icon to turn its output on (it will be green
when it is on); this valve helps control the pressure in the chamber. After the pressure is
set and the number (labeled SP) is no longer highlighted the software will automatically
adjust the ﬂow rate of the MFC until the pressure reaches the set point. The actual
pressure of the chamber is shown just below the set point with the number labeled
CAPMAN and the actual ﬂow of the MFC is the number in the lower left hand corner of

the MFC control box labeled FLOW.

46

To switch to the deposition screen touch the icon labeled DEP. The deposition

screen is shown below in Figure 27.

   

 

   
  

_ Flow Switch
; g j . ' Controls “

   

8. BIB-+62 mT
“FBI 8. 2Zf’s

   
 

'\

_3THL2<SH784I

 

 

Shutter
Controls Box

 

 

 

 

   

 

Figure 27 Deposition screen

The screen in Figure 27 shows the shutter controls and the ﬂow switch indicators for the
deposition screen. The PVD75 (for this research) is equipped for sputtering only and has
one sputtering gun, therefore the icon labeled SHTRl (shutter l) is the only shutter used.
Touch the icon to open or close the shutter, it will be green when the shutter is open and
red when the shutter is closed. However, for this investigation, because the platen has
been modiﬁed to be closer to the target the shutter has been removed because it would
not open entirely if it were attached to the cathode. The ﬂow switch controls indicate
whether or not the ﬂow set point for the MFC has been satisﬁed, it is green if it is

satisﬁed and red if it is not. The numbers in the lower left hand corner of the shutter

47

controls box are as follows; the number labeled CAP is the actual pressure of the
chamber, MFCl is the actual ﬂow of MFC l and MFC2 is the actual ﬂow of MFC2.

Touching the maintenance icon on the vacuum screen will switch the user to the

maintenance screen, as shown in Figure 28.

 

 

. ' — .0
I. -. PH 1 ?
1-503-245-1636 , , , ‘
....... 16:51:54" WWI PRESSURE CONTROL
HTLC-UI I. .. ..-.mw-mmm-q
1b IHsHOUEE lane
Hastings, East Bussea i 9.18
THTS 4HH U.H. 31
I44! 3 1424 T19161 3F TERM
... ﬁuaq
l
I
1.93
I TERM

 

   

 

Figure 28 Maintenance screen

From this screen Kurt J. Lesker’s contact information is available making it convenient to
reach them while you are still near the PVD75 when there is a problem. Depressing the
icon labeled clean on the touch screen will initiate the clean mode.

The pressure control icons are for manually setting the proportional and integral
terms (PID pressure control parameters). These terms may need to be adjusted for the
pressure control loop to function properly and cannot be adjusted while the pressure
control loop is running. The pressure control loop is the program/software that adjusts the

ﬂow of the gas to attain the pressure set point when the MFC is in mode 2. By touching

48

the term to be changed, it will be highlighted and using the INC/DEC icons to the right of
the screen allows the user to increase or decrease the term to the desired set point.
Acceptable values for these terms are 0 —— 99.99. If the INC/DEC square is not pressed
within 5 second the term chosen will no longer be highlighted. These terms are saved
when changed and will initialize to the changed values when the machine is powered up
again. There are no speciﬁed default values for the proportional and integral controls,
however, the user could simply make note of these values, which for our machine
happened to be 0.1 for the P term and 1.0 for the I term, at the initial start-up (out-of-the-
box start-up) of the machine then the machine can be setback to these values at any time.
Appropriate values, for processing conditions, need to be set experimentally.

The parameter controls on the left hand side of the of the PVD75 machine, as
shown in Figure 23, are the dc power supply (MDX 1.5 kW), the Spare-1e V (arc
suppression controls), the substrate heater (M.A.P.S. power supply) and the platen
rotation. The MDX 1.5 kW magnetron power supply, as shown in Figure 29, supplies
power to the cathode of the sputtering gun. Pushing the POWER button in the lower left

hand comer of the MDX will turn it on.

49

Output Set
Point

.RAMP ADJUST RIGHT DISPLAY

SE‘I’PT ACTUAL

- ADVANCED
A: ENERGY‘

 

The magnetron drive has the option of supplying output in the form of power, current or
voltage regulation. These buttons are located under the label REGULATION in Figure
29. Pressing the buttons marked power or current will regulate the output with these
parameters. If both the power and the current button are pressed at the same time the
output will be regulated by voltage. The regulation mode cannot be changed when the
output has been turned on, the lights will go on or off when pressed but the mode will not
change when the output is on [35]. The power output range is from O — 1500 W, the
voltage output range is from 0 — 750 V, and the current output range is from 0 — 2 A. The
actual output, based on what is chosen, is displayed to the right of the Actual Output

label. For this investigation, power output regulation is used.

50

To set the output value of the MDX press the button marked SETPT under the
label RIGHT DISPLAY and then turning the knob under the LEVEL label until the
desired set point is reached. If the SETPT button is pressed a second time the user can
adjust the ramp time of the output by turning the knob below the RAMP ADJUST label
until the desired ramp time is reached. The time for ramping the output can be set from 0
-10 either in seconds or in minutes. Pressing the ACTUAL button once, next to the
SETPT button, will display the actual output power in watts, pressing it twice will show
the actual output voltage in volts and pressing it a third time will show the actual output
current in amps.

There are four pairs of LEDs just below the output displays (actual and set point),

as shown in Figure 30; these are the LED status messages.

 

ARC SETPOINT INTLK OVERTEMP

 

 

 

I RAMP IOUTPUT liLASMA ‘ [REMOTE I

Figure 30 LED status lights on the MDX

From left to right, the ARC LED lights when an arc has been detected by the MDX and
has triggered the activation of the ARC-OUT circuit. The ARC-OUT circuitry helps to
suppress and quench arcing. The RAMP indicator light ﬂashes quickly when the output is
ramping toward the set point and turns off when ramping is ﬁnished. The SETPOINT
light turns on when the output is within .2% of the programmed set point and ﬂashes

when the output is on and the output is not within .2% of the set point. This light turns off

51

when the output is turned off. The OUTPUT LED is lit when the output has been turned
on. The INTLK indicator light is lit when all interlocks have been satisﬁed and ﬂashes if
all interlocks have not been met. The PLASMA light is lit when the output is on and the
output current is greater than 2% of the maximum current available for the MDX model.
The OVERTEMP lights if the MDX has overheated. The REMOTE LED lights if control
of the MDX is set to remote and is controlled by computer. Once all the parameters are
set the output can be turned on and off by pressing these buttons below the OUTPUT
label.

The Spare-1e V pulses the dc output from the MDX and sends it through to the
cathode in the chamber; it is meant to enhance the arc handling capability of the MDX
power supply. Sparc-le V actually stands for Small Package Arc Repression Circuit-Low

Energy Variable. The Spare-1e is shown in Figure 34 and 35. Figure 34 shows the entire

Sparc-le face and Figure 35 shows only the parameter menus and the programming

Digital
Display

buttons.

l,/
sit/:4 _
AEENERGYE‘ED Sparc-Ie /_//;/ ,\<\:\

[m

 

Figure 31 Sparc—le V pulsing and are suppression device
The digital display window above the parameter settings displays the value of each
setting when it is chosen (the LED in front of the parameter is lit). There are 4 menus,

shown on the following page in Figure 32; the function menu, which is the main menu;

52

two submenus, mode control and are handling, which can be accessed from the function
menu; and a status menu, which gives the current status of the Sparc—le V. In order to
access the submenus (mode control and are handling) from the function menu, use the
Select arrow keys to move the lit LED to desired submenu, push the Program button to
select the submenu, the LED in front of the submenu will blink, the user can now use the
Select arrow keys to choose a setting for that submenu (i.e. select either local or remote
for mode control), once the setting for that submenu has been chosen push the Program
button and the LED in front submenu will stop blinking and this signiﬁes that the user

can move through the function menu items.

F _ ‘. s." I . . Mode Control
unctron , 1., Menu
Menu '5 ﬂ ,

Status

I I ‘ -. J . Control Mode Override
; . . . - Loeel Crowbar

Men! Are COMM Remote Undervoitage
. . - An: Hlﬂdllng Mode Supervisory
‘ u i V Frequency (kHz) Arc Handling Interlock
‘r- ' Reverse Time (us) Self Run

Program (Reset) CWWMV DOIIV fuel Active Arc _
, Reverse Voltage (7.] Passive Arc Handllng
Control Mode Menu

 

Figure 32 Sparc-le V menus and programming buttons

As mentioned above, to scroll through the menus use the up/down arrows under
the Select label on the left hand side of the Spare—1e. In the function menu the ﬁrst three
items are are counts (total, micro and hard), which are counted while the output of the
MDX is on. These can be cleared at any time by pressing the Program (Reset) button

located on the bottom left hand side of the Spare-1e. The rest of the items in the function

53

menu are the arc handling mode, frequency (in kHz), reverse time (in us), crowbar delay
(in us), reverse voltage (in %) and the control mode. With the exception of the reverse
voltage percent, all of these items can be changed by scrolling down to the item; push the
Program (Reset) button to select the item. The light in front of the item will blink and the
user now has access to that parameter or submenu. To set the parameter or the submenu
setting to the desired point use the up/down arrows, when the parameter is set push the
Program (Reset) button again, the LED will stop blinking and stay lit until a select button
is pushed to move to the next item. Before changing the reverse voltage, be SURE that
the MDX output is turned off. The reverse voltage percentage can be changed by
scrolling down to the LED labeled reverse voltage. When the LED is lit, lift up on the T-
handle (located on top of the Spare-1e) and move it to one of percent choices (10%, 15%,
or 20%).

There are three different are handling modes, self-run, active arc and passive
mode. Self-run mode is the mode in which the Spare-1e is always operated, in this study.
This mode pulses the dc power from the MDX at a user deﬁned frequency. This pulsing,
discharges any charge that has built-up in the chamber and slows the build up of an
insulating ﬁlm [36]. Self—run mode allows for the frequency, reverse time, reverse
voltage and the crowbar (hard arc) delay to be adjusted. Information regarding the active
arc and the passive modes are available in reference [36].

The frequency parameter allows the user to set the frequency at which the power
from the MDX is pulsed. This frequency is shown in kilohertz (kHz) and ranges from 1 —

100 kHz. This range goes from 1 — 5 kHz in increments of 1 kHz, the frequency then

54

jumps from 5 kHz directly to 20 kHz. The frequency range then goes from 20 — 75 kHz
in increments of 1 kHz and it goes from 75 — 100 kHz in increments of 5 kHz.

The reverse time, trey, is the amount of time that the voltage to the target is
reversed in polarity (during the ton the voltage is negative, thus during the trey the voltage
is positive). This time is in us (micro-seconds) and the range to choose from differs
depending on the frequency that is selected. Table 4 listed below, shows the break down
of the frequency ranges that match the reverse time ranges [36].

Table 4 Reverse time ranges matched to frequency ranges (from reference [36])

 

 

 

Frequency 1-5, 5 1-55 56-60 61-65 66-70 7 1-75 80-
(kHz) 20-50 100
Reverse 1-10 1-7 1-6 1-5 1-4 1-3 1-2
Time
(Its)

 

 

 

 

 

 

 

 

 

The crowbar delay is the last attempt to quench an arc. This parameter can be set
to 10, 30, or 50 us. At 10 us, the Spare-1e tries to clear the are once before it shorts the
output of the MDX power supply. At 30 us the Spare-1e will attempt to clear the are twice
before it shorts the output of the power supply and at 50 us it will attempt to clear the are
three times. When the output of the power supply is shorted it is shutdown for 7 ms (long
enough to clear the arc) and then the output from the MDX is reinstated and continues its
normal operation.

Reverse Voltage is the percent of the initial voltage that is reversed in polarity.
For example, if the ton voltage (initial voltage) is set to 500 V (as mentioned previously,
this voltage is negative in polarity) and the reverse voltage is set to 10%, then the reverse
voltage will be 50 V (of positive polarity). The reverse voltage can be set to 10, 15, or

20%.

55

There are two different control modes at which to set the Sparc-le V, they are
local and remote. For this investigation the Spare-1e is always set to local mode; this
means that the Sparc-le parameters are adjusted from the front of the Spare-1e (they are
set manually). If the Spare-1e is in remote mode then the parameters are adjusted
remotely from the rear of the machine via the user port or the host port.

The status menu, located on the far right in Figure 32, shows the status of the
Sparc-le V. The status LEDs are labeled as follows, mode override, crowbar, under-
voltage, supervisory and interlock. The mode override LED in the status menu will light
when there are any one of the following faults detected, input over-current, over-
temperature, over input power, over-voltage, line voltage fault, tap select interlock fault,
switch over-current, slave fault, slave error, or processor error. When one of these errors
occurs the Spare-1e V is disabled, but the power to the load is not interrupted. This is an
undesirable situation because there is neither are suppression nor a reverse recovery. If
this situation occurs, system troubleshooting is necessary. After the condition has cleared
the Spare-1e V will turn on again automatically to the same mode prior to being disabled.
The LED will remain lit until the fault has been cleared by pressing the program (reset)
button twice [36]. When a mode override error occurs an error code will display for 1 s
and a normal display resumes. The error code will redisplay by pressing any button on
the Spare-1e V. If multiple errors occur only the last error code will display. For
convenience, the table of error codes, Table 5 shown on the following page is copied

from reference [36].

56

Table 5 Spare-1e V error codes and their meanings, taken from reference [36]
Error Code Description

E01 This is an over-voltage error. The active arc and self—run switches
are disabled for 1 ms but power is not interrupted from the MDX to
the chamber.

 

 

 

 

E02 This is an input over-current error code. The crowbar is initiated to
short the dc input power to the Sparc-le.
E03 This is an over-temperature error code. The Sparc-le is disabled and

the passive mode is enabled until the unit cools down. The error
code display remains until the unit has cooled.

E04 An over input power has occurred, which means the power has
exceeded 12.5 kW. The active arc and self-run switches are disabled
until the fault is cleared.

 

 

 

E05 This is a line voltage fault. When this error occurs the active arc and
self-run modes are disabled for 100 ms.
E06 A tap select interlock fault, this means that the T—handle to adjust the

reverse voltage is up. The active arc and self-run modes are disabled
for 100 ms while the handle is raised. This interlock fault could also
mean that the ARC—OUT enhancement relay drive has
malfunctioned and the Spare-1e requires repair service.

 

 

E07 This indicates a switch over-current, at which time the reverse time
is shortened until the error clears
E08 This is a slave fault and it means there is one of the following errors

in the slave Spare-1e, over-temperature, line voltage input, over
input power, switch over-current, or over-voltage.

(Note: for this research we do not have multiple Spare-le’s set up in
the master/slave orientation)

E09 Slave error means that the slave has lost line voltage or the
master/slave cable is bad.

(Note: for this research we do not have multiple Spare-le’s set up in
the master/slave orientation)

E10 This indicates a processor error (internal) or power has been lost to
the user port and requires a manual startup.

 

 

 

 

 

 

The crowbar status LED will ﬂash for 5 ms each time a crowbar event occurs.
The under-voltage will light when the dc input voltage is lower than 150V. When the
supervisory LED is lit (continuously) this means that the Sparc-le V has an internal
malfunction. The LED sometimes blinks during hard arcing but it only indicates a

problem when it is continuously lit. When the interlock light is lit when the back cover on

57

the output connector is off or the T-handle for the reverse voltage is not seated properly
[36].

The M. A. P. S. heater power supply is the power supply that controls substrate
heating. It also monitors the temperature of the substrate. The substrate heater is shown in

Figure 33.

Hill".

‘ SETIJOINT
50“..

cuanrmr

i‘OVVER

 

. OLI l M II

. INll'Itlfll K
LED Bar Display Mode Temp. Limit
Graph Display Selector Switch Controller Controller

 

Figure 33 M.A.P.S. Heater Power Supply (substrate heater)

To turn on the heater power supply on ﬂip the switch, under the POWER label in the
lower right hand corner of Figure 33, to the up position. To turn off the heater power
supply, ﬂip the switch to the down position. This switch also acts as overall current
protection as well.

The LEDs in the lower left hand corner are the power, output and interlock
indicators. When all interlocks are satisﬁed the LED labeled INTERLOCK will be lit.
When power to the heater power supply has been switched on the LED labeled POWER

will be lit. When output from the heater has been turned on the OUTPUT LED will be lit.

58

To turn on the output from the heater power supply to the quartz lamps in the
chamber, push the ON portion of the switch under the label OUTPUT. Once it is pushed
the output LED will light and stay lit. Pushing the OFF portion of the OUTPUT switch
turns the output LED off and output to the quartz lamps is shut off. When the ON or OFF
switch is pushed it will not stay locked in place, it will return to the neutral position.

The LED bar graph display, labeled in Figure 33, has markings of 0, 50 and 100
% to the right of the bar graph display. The bar graph display has 20 LED bars total and
is linearly scaled so that each bar represents 0.5 VDC of the control signal to the SCR
(silicon controlled rectiﬁer), which controls power delivery and has a 0 — 10 VDC signal
range [37]. It should be noted that the set point is not linear with the applied voltage.

The display mode selector switch selects the function of the bar graph display.
There are three settings which are labeled, SETPOINT, CURRENT and POWER. The
current and power selections are non-functional and are there for future use (i.e.
upgrades). The setpoint setting gives the user a visual of the voltage control signal sent to
the SCR.

The temperature controller, as labeled in Figure 33, allows the user to set the
desired temperature. The temperature controller allows the user to program, heating
processes into the device, which allows for plethora of variables to be set for strict control
over the heating process. For further information on how change speciﬁc variables and/or
program the temperature controller please see reference [38]. Our process is simple and
does not need such explicit control. The temperature is set to the desired set point and
allowed to ramp up to temperature, before the plasma is ignited. The substrate is kept at

this temperature for the entire deposition process. To set the temperature, push the

59

up/down arrow at the bottom right hand side of the temperature controller until the
desired temperature is set. The set point is in small yellow numbers in the second row of
the digital display. The actual temperature (as measured by the thermocouples in the
chamber) is in large red numerals in the ﬁrst row of the display. There are three process
LEDs, labeled 1, 2 and 3 in the upper left hand corner of the temperature controller.
When the temperature is set manually, as just described, the LED labeled 1 will blink
until the set point has been reached.

The limit controller, in the upper right hand comer of Figure 33, is used to prevent
process conditions from going out of control limits and damaging the substrate and or
chamber parts. The limit controller allows for many different parameters to be set by the
user, but our process is a simple one and only the over-temperature is set. For more
information on parameters that can be set and how to program the limit controller please
see reference [39]. To set the temperature limit, use the up/down arrow keys located in
the bottom right hand comer of the limit controller until the desired absolute maximum
process temperature is set. The set point is in small yellow numerals on the far right of
the digital display. The actual temperature, as measured by the thermocouples in the
chamber, is in large red numbers on the left of the display.

The last of the parameter controls for the PVD75 is the platen (substrate) rotation

control, shown in Figure 34.

60

 

Figure 34 Platen rotation control

To turn the platen rotation on, ﬂip the toggle switch on the left hand side of Figure 31 to
the ON position. To turn the platen rotation off, ﬂip the toggle switch to the OFF
position. Before turning the rotation on the rotary knob should always be set to 0. Turn
the rotation on and slowly adjust the knob to the desired speed. The rotation speed setting
can range from 0 — 10. Based on experience, if the speed is set above 5 the rotation can
become unstable and cease to rotate; therefore the rotation speed is never set higher than
5. At a setting of 5 the rotations per minute (rpm) is 41. At the time this thesis was
written, the rotation would cease when the plasma was turned on. The source of this
problem has yet to be ascertained.
4.4 PVD75 Operating Procedures

To begin working with the PVD75, one must go through the start-up procedure to

turn on the machine; the steps are listed on the following page in Table 6.

61

Table 6 PVD75 Start—Up Procedure
Steps Description

1 Turn on the main circuit breakers (2) located behind the machine. The breaker
on the left controls power to the PVD75 except for the roughing pump. The
breaker on the right controls the power to the roughing pump
2 Turn on the nitrogen gas tanks (2), located to the right of the PVD75 in the
center row. There are 3 rows of tanks. One tank controls the valves and
sputtering shield and the other vents the chamber.
3 Turn on the chiller located to the right of the PVD75. Be sure the chiller is set
to 20 °C once it is on. The chiller MUST be on in order to operate the PVD75.
(Note: If it is below 20 °C on hot humid days condensation will build up on the
machine creating an electrical hazard)
4 Turn on the power-box circuit breaker, located inside the left rear panel at the
top of the machine, as shown in Figure 35. The green LED located above this
switch will be lit when the switch is in the ON position
5 Turn on the power switch to the MDX 1.5 and the Spare—1e V (located on the
back of the respective devices) accessible from the rear of the PVD75.

 

 

 

 

 

 

 

 

 

 

Power-box
Circuit Breaker

 

Figure 35 PVD75 Power-box

Now that the PVD75 is on, afﬁx the substrate to the platen and place it back into the
chamber. Wipe down the chamber walls using IPA (isopropyl alcohol, to avoid leaving a
residue when it evaporates, as with acetone, or methanol) and target to remove any loose
and unwanted particulates, a source of arcing. This should be done before or after every
run. Check the target to ensure that the proper target has been installed, for target removal
and installation instructions please refer to reference [40]. Close the chamber door and
begin the pump down procedure as described previously. Wait until the pressure is at

least 4 x 10'6 Torr before beginning the deposition procedure.

62

After the pump down procedure has ﬁnished the parameters can now be set for
the deposition process. The process used to deposit the AIN layer is a reactive process;
for this investigation it means that a target of Al was sputtered in a nitrogen gas enriched
environment and combined with the nitrogen before depositing on the substrate as a layer
of AIN. Listed in Table 7 is a description of the deposition procedure.

Table 7 AIN Deposition procedure

 

 

 

 

 

 

 

 

 

 

 

 

Steps Description

1 Set the substrate temperature and turn on the heater output. Allow
approximately 15 - 30 minutes for the temperature to reach its set point.

2 Turn on the high purity nitrogen and argon gases.

3 Set the deposition chamber pressure from the gas screen on the PVD75.

4 Set the mode for each gas (nitrogen and argon) and the parameters related to
each mode.

5 Turn on the gas and throttle valves.

6 Set the parameters on the Spare-1e V.

7 Set the parameters for the MDX.

8 Turn on the output from the MDX for the speciﬁed time.

9 When the deposition time is reached, turn off the output to the MDX.

10 Turn off the substrate heater and allow to the platen to cool overnight before
removing the substrate from the chamber.

 

 

 

In Tables 8 — 10 are listed the parameter settings used to deposit AlN onto
ultrananocrystalline diamond (UNCD). The frequency of 50 kHz and a reverse time
range of 3 - 5 us were chosen because for the AIN reactive process this was the frequency
and reverse times at which arcing was minimal and made for a more stable process for
the duration of the run time. The crowbar delay is set at 10 us so that the Spare-1e will try
to clear the are once and then will cut-off power from the MDX to quench the arc. Arcing
causes unwanted effects to the deposited layer and at worst it could do damage to the
MDX power supply. The control mode is always set at local so the Spare-1e can be

controlled from the front panel.

63

Table 8 AlN Parameters for the S arc-1e V unit in the PVD75

 

Arc Frequency Reverse Crowbar Reverse Control Substr

Handling Time Delay Voltage Mode ate
Rotati
on

 

Spar
c-le Self Run 50 kHz 3 — 5 us 10 us 10 % Local 0

 

 

 

 

 

 

 

 

 

 

Table 9 shows the parameter settings of the MDX power supply. The power was set to
the maximum power that could be output given the Spare-1e settings.

Table 9 AIN Parameters for the MDX 1.5 kW

 

 

 

 

 

 

Power Ramp Time
(Constant Power Mode)
MDX 1.5 kW 450 - 375 W 7 3
(Power Source)

 

Table 10 shows the chamber pressure settings used for the AIN deposition. As previously
noted, the base pressure (pressure the chamber is initially pumped down to) is pumped
down to at least 4 x 10'6 Torr to have minimal impurities in the chamber at the time of
deposition. The rest of the chamber parameters were set based on previous works
[20,21,23,30,32,41,42].

Table 10 AlN Parameters for inside the chamber of the PVD75

 

 

 

Base Deposition Gas Flow Substrate Substr
Pressure Pressure (Argon ; Nitrogen) Temperature ate
Distan
cc to
Target
Chamber < 4"‘10'6 3 mTorr Mode l ; Mode 350 oC ~ 7.5
Torr 2 cm
33.3% cont. ;

 

 

 

 

 

 

 

Table 11, shown on the following page, explains the steps of the shut-down procedure for

the PVD75.

Table 11 PVD75 Shut-down procedure

 

 

 

 

 

 

 

 

 

 

 

 

Steps Description

1 Turn off the power button on the front of the MDX.

2 Turn off the gas valve and the throttle valve on the gas screen.

3 Shut off gas ﬂow from the high purity nitrogen and argon tanks (on the
tanks).

4 Switch to the vacuum screen, vent the chamber, wait for the chamber to
reach atmospheric pressure and wait for the status message bar be in stand-
by mode.

5 Turn off the power switch to the MDX 1.5 and the Sparc-le V (located on
the back of the respective devices) accessible from the rear of the PVD75.

6 Turn off the power-box circuit breaker, located inside the left rear panel at
the top of the machine, as shown in Figure 35. The green LED located
above this switch will turn off when the switch is in the OFF position.

7 Turn off the the main circuit breakers (2) located behind the machine

8 Turn off the nitrogen gas tanks (2), located in the center row

9 Allow the chiller to remain on for at least 15 minutes to avoid damaging the
magnetrons and then turn the chiller off.

 

Some expectations for a typical run would be that the power would stay within

20% of its set-point. The reason for this large percentage is because as the process runs

the chamber conditions change and the target degrades as well, which causes the actual

power output to ﬂuctuate and decrease as the power output is on. Even from run to run

the power cannot always be set to the initial start value of the previous run due to this

phenomena and it is sometimes difﬁcult to gauge where the power needs to be set upon

start—up in order to achieve optimal results (i.e. high deposition rate, uniformity, etc.). It

would also be expected for a typical run that upon ignition of the plasma there is a high

amount of arc activity, usually this spike occurs within the ﬁrst 30 - 60 seconds of

igniting the plasma (turning on the power output). This spike in arc activity can be

minimized by adjusting chamber parameters, but at the cost of deposition rate. This initial

spike in arcing could be due to build-up of insulative material from the previous run,

redeposition of material on the target, ﬂakes, or particulates on the target. The target can

65

 

be “scrubbed”, a pre-clean of the target with the shutter closed to speciﬁcally rid the
target of most of these issues, prior to deposition for 10 - 15 minutes. However, the target
cannot be “scrubbed” during the A1N deposition due to the platen distance from the
target. The shutter is has been removed because the distance from the platen to the target
would hinder it from opening all the way. If it was left on it would not only damage the
substrate/sample, it would partially block the plasma from getting to the substrate as well.
An average of 150 arcs or less is appropriate upon plasma ignition. Also an average of
250 arcs or less is expected for total arcs for an entire run. This is of course dependent on
the expectations for the layer upon completion, if the uniformity does not matter (e. g. it is
a sacriﬁcial or masking layer) than more arcs than this may acceptable. As for hard arcs,
no matter what the situation an average of 1 or 2 is acceptable for an entire run, but the
absolute maximum would be 5 (and only upon initial start-up, because it is over a short
duration of time, between 30 and 60 seconds, as mentioned above), any more than this
and the system would need to be shut down and the cause would need to be found and
eliminated. The expectations for the completed layer would determine whether or not the
platen rotation would need to be on or off. In this research for a typical AIN run the
rotation is off. For an average run it would also be expected to be an error free run, i.e.

none of the errors described in Table 5 would be experienced.

66

Chapter 5
Aluminum Nitride Deposition Observations

5.1 Introduction

This chapter will discuss the observations made after deposition. First, a
discussion of general observations are made, such as measurements concerning surface
roughness, substrates used and deposition parameters. Next, characteristics of the ﬁlm
that will be discussed are growth rate, uniformity and adhesion. Optical data,
transmission and reﬂection, were also collected. Then, x-ray diffraction (XRD) was used
to determine the chemical composition and lattice structure of the deposited layer.
Finally, scanning electron microscopy (SEM) was used to determine grain growth.
5.2 Deposition Observations

Aluminum nitride was deposited on different surfaces, glass (microscope slides),
silicon and UNCD. A list of all samples for this thesis is included in Appendix A. The
initial (trial) runs were done on glass microscope slides; these are called trial runs
because this was the initial phase of using the PVD75 and involved becoming familiar
with the parameters. These runs were with a non- heated substrate and performed under
arc-less conditions under the assumption that arcing causes undesirable effects to the
deposited ﬁlm layer. However, under these arc-less conditions we were unable to take
advantage of the higher deposition rates by using pulsed dc magnetron sputtering. For

some of the later samples, on silicon substrates, the substrate was heated. As noted

67

previously, the deposition parameters for these samples are listed in Tables 8 — 10. A

shortened list is shown below in Table 12.

Table 12 Shortened list of AlN deposition parameters

 

 

 

 

 

 

 

 

Parameter Value
Pressure 3 mTorr
Frequency 50 kHz
Trev 3 HS
N22 AT 321
Tempsubstrate 350 0C
Power 370 — 450 W

 

 

 

The samples with AIN deposited onto Si were observed to be smooth and

transparent. Measurements with a Dektak were shown to have Ra values in the range of

0.6 to 2 nm and RI values in the range of 10 to 50 nm for scan lengths of 1- 2 mm. A

representative sample, CF—018, had an R3 value of 0.78 nm and an R2 value of 28.77 nm

and a thickness of 800 nm. Some samples (using glass and silicon substrates) and their

parameters are listed below in Table 13.

Table 13 Samples of AlN and some of their parameters

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample Deposition Distance Rotation Thickness Deposition Young’s
Time to Target Rate Modulus
CF-008 120 (min) 15 (cm) Yes 120 (nm) 60 (nm/hr) 150
(GPa)
CF-009 120 7.5 No 300 150 x
CF-016 64 7 .5 No 493 462 280
CF-017 163 7.5 No 1500 551 330
CF-018 120 7 .5 No 800 400 353
CF-019 35 7.5 No 187 322 x

 

The Young’s modulus was also lower on the sample run with a distance to the

target of 15 cm as opposed to the samples that ran at a distance of 7.5 cm to the target.

Optical data was taken for sample CF—018 to conﬁrm the thickness of the A1N layer. A

thickness of approximately 1267 nm was found using this method and the thickness

found using the Dektak proﬁlometer was 800 nm, as listed in Table 13. A possible reason

68

 

for this discrepancy is that the A1N layer was deposited without platen rotation, which
means that due to the non-uniformity of the plasma (characteristic race—track formation)
the deposited layer will be non-uniform as well. If the rotation were turned on then the
deposited layer would be more uniform. Another possible reason for the discrepancy is
that when using the proﬁlometer to assess the thickness of a sample a precise step is
needed not a gradual slope. A method for getting a crisp step was not found and used
until the last sample (CF-022) was run in the PVD75.

5.3 Deposition Rate

The growth rate of the ﬁlm is dependent on the parameters set on the Sparc-le V
and the chamber parameters. These parameters affect the maximum setting, i.e. the
power/voltage/current setting that the MDX can output. For this investigation the power
was used as the setpoint for the MDX output. Once the current of the MDX reaches 2 A
then this is the upper limit of the power output from the MDX.

As mentioned in a previous chapter, the settings on the Spare-1e V are as follows;
are handling mode, frequency (in kHz), reverse time (in Its), crowbar delay (in Its),
reverse voltage (in %) and the control mode. The settings that were adjusted for this
investigation are the frequency and reverse time. It was found experimentally that the
frequency had the most effect on minimizing arcs and that the reverse time had the
largest effect on maximizing the possible output voltage from the MDX power supply.
The effect on the MDX output is important because the more power there is the higher
the deposition rate. This means that any parameter that has an effect on the output of the
power supply will cause the deposition rate to be effected as well. The reverse time

range depends on the frequency setting, the higher the frequency the lower the possible

69

setting of the reverse time. These times and frequency ranges are listed in Table 4,
Chapter 4, but in general the reverse time could possibly be set from 1-10 us. The higher
the setting of the reverse time means that the maximum output of the power supply is
lower and thus causing the deposition time to be lower. For example, the earlier “test”
runs for the PVD75, that were done on glass microscope slides, were run for
approximately one hour with the reverse times set to 10 us (in order to deposit under are-
less conditions) and the layer thickness for some of these runs are listed in Table 14
below. In comparison to some of the later runs (which are listed in Table 13), it can be
seen that changing the reverse time makes a big difference in deposition. For example,
the deposition time for sample CF-018 was 2 hours, which corresponds to a deposition
rate of 400 nm/hr.

Table 14 Sample deposition rates

 

 

 

 

 

 

Sample # Deposition Rate Deposition Reverse Time Thickness Distance
(rim/hr) Time (min) (us) (nm) to Target
(cm)
CF-001 70.2 65 10 76 15
CF-002 57 60 10 57 15
CF-003 62 60 10 62 15
CF—004 40 60 10 40 15

 

 

 

 

 

 

 

The trial samples, on glass, which were run under arc-less conditions and at a
distance of 15 cm from the target had low deposition rates. Those samples (on silicon
substrates) that were run at a distance of approximately 7.5 cm from the target had higher
deposition rates. As can be seen from sample CF-009 above, simply by making the
distance to the target smaller the deposition rate was over triple what sample CF-004 was

and slightly larger than double the deposition rate for sample CF-001.

70

Another parameter that also has an affect on the deposition rate is the wear of the
target over time. As the target is used, the maximum power level that can be attained
decreases; sometimes it can even be seen within the same run, if the run is long enough.
For example, sample CF-019 was run for 2 hrs and the power was set to 420 W. The
setpoint was attained initially, but by the end of the run the power level had decreased to
364 W. No measurements were made for this effect due to the nature of the deposition
process. A systematic study was not performed to determine how this variation in power
levels during deposition affects the deposition rate. It can be said that the deposition rate
decreases but by how much exactly cannot be said, based on the data collected to date.
5.4 Uniformity

For this investigation, achieving a uniform AlN ﬁlm was not a high priority. All
that was necessary for this research was creating a piezoelectric ﬁlm of A1N with a (002)
orientation. However, for future work and for the layer of Al needed to create the
electrodes of the SAW devices a uniform layer is necessary. The parameters that have an
effect on uniformity are, target erosion/wear, distance to target from the substrate,
substrate rotation and the position of the substrate over the target. Each of these
parameters affects the others, in other words they are not independent from each other.

As a target is used it erodes unevenly in a “racetrack” shape around the target, as

shown below in Figure 36.

71

    

Racetrack \

Figure 36 Racetrack erosion of used target

This racetrack erosion is caused by the design of the magnetron. This erosion pattern
leads to a non-uniform plasma and this in turn leads to a non-uniform deposition layer. If,
however, the substrate is rotating this aids in achieving a uniform coating. The other two
parameters that also help in getting a uniform layer are distance to target and the
placement of the substrate over the target (plasma). If the distance to the target is small
then the placement of the substrate is crucial, the entire wafer must be placed as directly
over the plasma as possible. The substrate must also be rotated in order to get a uniform
deposition layer. Figures 37, 38, and 39, show photographs of samples CF—Ol9, CF-008
and CF-009, respectively. Sample CF-019 had no rotation and was a distance of ~7.5 cm
to the target and it was placed directly over the target plasma for deposition. The sample
thickness is phi-symmetric, decreasing radially from the center. Sample CF-008 was a
distance ~15 cm from the target with rotation and was not placed directly over the target.
Although, the entire wafer is not shown in Figure 38 due to sectioning it for XRD and
other analysis, the entire wafer was uniformly coated. Sample CF-009 was a distance of

~7.5 cm from the target and was placed slightly off-center from the target and the

72

deposition was done with no rotation of the substrate. Under these conditions the

thickness is not phi-symmetric.

 

Figure 37 Sample CF-019, ~7.5 cm above the target, no rotation

 

Figure 38 Sample CF-008, uniformly coated sample, with rotation

73

 

Figure 39 Sample CF-009, placed off center from plasma, no rotation
If the distance to the target is large then the sputtered atoms have a greater distance over
which to spread out. This allows for a larger area of placement of the substrate on the
platen and still leaves the possibility of achieving a uniform coating of the substrate.
5.5 Adhesion

The adhesion for all the samples was tested using a tape test. All of the samples
had excellent adhesion to the substrate. Sample CF-022 was found to have good adhesion
to the diamond interlayer; this was found using two different methods. The ﬁrst method
was the tape test and the second method was XRD. XRD is not a method of determining
adhesion, but due to the circumstances we were able to use it to determine that the
adhesion between the UNCD interlayer and the AIN layer was good. The sample of
UNCD that was used in this investigation was not a ﬂat sample; it was convex due to
internal stresses caused during the diamond grth process. The AlN was then deposited
onto the sample this way. During the photolithography procedure the sample was broken;

this caused the AIN layer to break away from the substrate in pieces (ﬂakes). It was at

74

ﬁrst assumed that these layers were freestanding AIN, but when held to the light for
visual inspection they were not transparent as would be expected. X-ray diffraction was
used on one of these ﬂakes to determine why it was not transparent as would be expected
and it was found that there was diamond present on the sample. This information was
used to also determine that there was good adhesion between the two layers because if
there was poor adhesion between the two only AIN would have been present on the XRD
analysis. The XRD scan for this sample is shown on the following page, in Figure 40. All

the signiﬁcant peaks are labeled, they are AIN (002), (004) and diamond (111).

75

32: Such N

 

. com.

com

8».

08

com

08 _.

come

83

com _.

Isuexul

(369) M

Figure 40 0-20 XRD scan for sample CF-022 (XRD scans performed by Rahul

i)

Ramamurt

76

5.6 Optical Data: Reﬂection and Transmission

Reﬂection and transmission optical data was collected for this research. The
reﬂection data, shown in Figure 41, on the following page, is used to calculate the
thickness of the ﬁlm and it also shows that the bandgap of the A1N ﬁlm is approximately
6 eV or ~200nm. The equation used to calculate the thickness from Figure 41 is as
follows

t—‘l =2n( 1 — l )

,1 peak] Apea/<2

 

(9)
where n is the index of refraction and Area“ is the wavelength associated with the ﬁrst
peak and W2 is the wavelength associated with the second peak, from Figure 41.
Plugging in 219 nm for hp”; and 366 nm for M2 and using n=2 for the index of
refraction, the thickness is found to be 136 nm. This thickness is pretty close to the 120
nm measured by the proﬁlometer. The bandgap was also found using data from the
graph. There is a steep “drop-off” in signal at a lambda of approximately 200 nm and this

corresponds to a bandgap of 6 eV.

77

Reﬂection - Sample CF-OOB

 

I e Seriest

Reflection °/o

 

 

 

 

185 235 285 335
Lambda

Figure 41 Reﬂection data collected from sample CF-008

This same methodology was also used to calculate the thickness of sample CF=018. The
data for this is shown on the following page in Figure 42. Again, using Equation 9 and
plugging in 1615 nm and 2413 nm for Area“ and 11mm, respectively and n=2 for the
refractive index. The thickness of the sample is then calculated to be 1.22 pm. To conﬁrm
this calculation the valleys were also used and the calculation should give the same result.
The values of 1393 nm and 1950 nm were used for Laney] and Ivaneyg, respectively and
n=2 for the refractive index. The thickness was then calculated to be 1.22 pm. This
conﬁrms the calculation for the thickness of the AIN ﬁlm. Although, the peak and the
valley calculations of the thickness of the ﬁlm correspond with each other they do not
correspond to the measurements of thickness made by the proﬁlometer, which was 0.8
pm. As mentioned earlier, this could be for a number of reasons. The Dektak needs a
sharp step in order to accurately measure the thickness of a ﬁlm, not a gradual slope
(which is what was used in this case because that is all that was available). It could also

be due to the fact that the deposited AlN ﬁlm is not uniform and is thicker in the center

78

than near the edges of the wafer, which is where the measurements were taken using the

proﬁlometer.

Reﬂection - Sample CF-O18

 

 

 

I e Series1

 

 

 

110447»_-s_7-1_m
o l l t

 

Figure 42 Reﬂection data for sample CF-018 for a range of 1300 - 2800 nm
Looking at this same data for the range of 185 - 305 nm, as shown in Figure 43 below, it

can be seen that the transparency of the ﬁlm goes to zero at ~230 nm which corresponds

to a bandgap of about 5.4 eV.

79

Reﬂection - Sample CF-O18

 

 

 

‘ 80

70 L e i 4
..6° 2+ == L
l 5 504* r x '9” ‘—— e‘, :1 _
I; g 40% —l — --—~ ——-,.——Ov-. ,-;—— eSeriesI}
l:3° _ e s r :z ”37"“
- 20 <4 , 44* LL- *‘gfofgm. «V ’a

10‘ we iw’ V. 9 ‘j 1‘— a

o

 

185 205 225 245 265 285 305
i Lambda I

Figure 43 Reﬂection data for sample CF-018 for a range of 185 — 305 nm
Transmission data was also collected for use in this investigation, as shown in

Figure 44 below.

Transmission - Sample CF-O18

Transmission %

 

-10 i. .L LLL L LLL. L: L L L L L L L L L L L. L ,_
l 0 500 1 000 1 500 2000 2500 3000 3500

l i
: Lambda l
I

Figure 44 Transmission data for sample CF—018
This set of data shows that the deposited AIN ﬁlm is transparent causing the constructive

and destructive interference. It also shows the onset of absorption by the silicon substrate.

80

Using Equation 9 again, we can conﬁrm the calculations for the thickness of the ﬁlm on
sample CF-018. For the values of Mm = 1396 nm, Xpeakz = 1946 nm and n=2 for the
refractive index, the thickness of the ﬁlm is calculated to be 1.23 pm, which is
approximately what is found using the reﬂection data.
5.7 X-Ray Diffraction

The crystal structure of the AIN layer was determined using X-ray diffraction.
There were 3 samples that were analyzed using this method. The ﬁrst sample to be
analyzed was CF-022. This sample consisted of four layers, from the bottom up, a silicon
substrate, a UNCD layer, an aluminum nitride layer and ﬁnally the patterned Al which
formed the electrodes. This Al layer, after patterning was only present where there were
electrodes, it was not across the entire substrate as the other layers were. The XRD scan
for this sample is shown in Figure 40. As discussed earlier this shows that there is AIN
ﬁlm. An interesting point to note is that the (002) is needed for the SAW devices and yet
there is only a small signal on the scan. There is however quite a large signal for the
(004). X-ray diffraction analysis was also completed for two other samples, CF—021 and
CF-008. The 0—20 XRD scans for these samples are shown below and on the following
pages in Figures 45 and 46. As can be seen from both scans there is no signiﬁcant peak at
all other than Si that is picked up by the scan. What has caused this to occur has not yet

been assessed, at the time this thesis was written.

81

tom) IS L, L LL

 

 

80 70 60 50 40 30 20 10

Figure 45 Figure 40 0-20 XRD scan for sample CF-008 (XRD scan performed by R.
Ramamurti)

82

 

(sdo) Mgsuaaul

83

Figure 46 Figure 40 0-20 XRD scan for sample CF—021 (XRD scan performed by R.
Ramamurti)
5.8 Scanning Electron Microscopy

SEM cross-sectional pictures were taken in order to ascertain what the grain
structure of the A1N layer might look like. As shown below in Figure 47 on the following
page, a cross-sectional picture was taken with the SEM of sample CF—016. This sample
was not coated with a conducting layer of gold (Au) before being placed in the SEM for

analysis. Because the AIN and Si layers are not conducting, an image was not obtained,

so that the grain structure could be seen under the microscope.

2 pm

  

Figure 47 SEM cross—sectional picture of sample CF—016 without Au coating (SEM
pictures privided by Tran Dzung)

As can be seen on the following page, in Figures 48, 49, and 50, the grain structure of the

A1N layer can be seen. The sample was coated with 20 nm of gold in order to be able to

84

see the grain structure of AlN layer. It can be seen from these ﬁgures that the grain

structure is columnar in shape.

 

     

.' , .
~71

ss-sectional picture of sampleCF-016 wichO nm Au coating (SEM

   

Fgure 48 SM cro

pictures privided by Tran Dzung)

 

Figure 49 SEM cross-sectional picture of sample CF—016 with 20 nm Au coating (SEM
pictures privided by Tran Dzung)

85

SJJ nn

 

Figure 50 SEM cross-sectional picture of sample CF-016 with 20 nm Au coating (SEM
pictures privided by Tran Dzung)

86

Chapter 6

SAW Device Characterization: Setup and Procedures

6.1 Introduction

This chapter will discuss the characterization techniques, setup and procedures
employed in the characterization of the SAW devices. To begin, there will be a
description of the network analyzer, Hewlett Packard (HP) 8753 D, and its operating
procedures. Next, there is a description of the calibration procedures for the HP 8753D.
Then follows, a description of the high frequency probe tips used in order to probe the
devices and a description of the probe station and its setup. Finally, there is a discussion
of results obtained from a commercial SAW device as well as a SAW device fabricated
on quartz using the procedures described in Chapter 3.
6.2 HP 8753D Description and Operating Procedures

In this investigation, a Hewlett Packard (now under the name Agilent
Technologies) 8753D network analyzer was used to characterize the SAW devices. The
network analyzer allows measurement of the scattering parameters, Sij, of a ‘device—
under-test (DUT) over a broad frequency range. For example, consider a two port SAW
ﬁlter with the input terminal connected to port 1 of the spectrum analyzer and the output

terminal connected to port 2. The Sij are deﬁned as

5.. =_.- (10)

87

where v; is the amplitude of the voltage wave incident on the DUT from port j and v,— is
the amplitude of the voltage wave reﬂected from the DUT to port i. Thus, as described by

Pozar, “Sij is found by driving port j with an incident wave of magnitude v; and

measuring the reﬂected wave amplitude v,‘ coming out of port i” [43]. This is
accomplished by the 8753D. For a SAW ﬁlter, as illustrated, for example, in Figure 3 of
Chapter 2 and Figure 9 of Chapter 3, one wishes to have a high transmission from port 1
to port 2, or in other words, a large value of S12, at resonant frequency. At non-resonant
frequencies, S12 should be low. A picture of the network analyzer is shown below, in

Figure 51.

Display I
Screen

 

Parameter
‘ Controls

This will not be an in-depth description of the network analyzer, but all the parameters

Figure 51 Hewlett Packard 8753D network analyzer

vital to conducting the necessary measurements on the SAW devices will be discussed.

On the left hand side of the HP 8753D is the display screen and on the right hand side of

the machine are the parameter control buttons, as shown above in Figure 51.

88

Soft Keys

 

   

3.5” Disk Drive

Figure 52 HP 8753D display and other components

Shown in Figure 52, on the previous page, are some of the major components associated
with the display screen. The plug below the power button is the grounding tether for the
user to wear when connecting and disconnecting the device under test (DUT). This tether
grounds the user to avoid an accidental static discharge when handling the DUT which
could potentially harm the device. There is also a 3.5” disk drive, located below the
display, so the user can save the measurement results, calibration settings, etc. Directly to
the right of the display screen are soft keys which allow the user to access menus shown

on the right hand side of the display screen, next to the soft keys.

89

 
 
    
   
 
  

  

Active
Channel

 

Response

    
   
  

Instrument
State

R Channel

 

‘ Probe Power
Connectors

 
  
 
   

 

 

Ports
1 & 2

 

Figure 53 Parameter control panel

In Figure 53, shown above, the keys to control the measurement parameters are organized
in sections which are labeled both on the photograph and with side captions. The two
Active Channel buttons are independent display channels and these keys allow you to
choose the active channel. The functions then chosen apply to the channel that is active at
the time [44]. Just below these buttons are the Response keys. The Response keys allow
the user to control the measurement and display parameters. The Stimulus buttons control
parameters such as frequency, power, etc. The Entry block of keys is for entering
numerical data and controlling the markers. In the box marked Instrument State are the
keys which control channel-independent system functions, such as save/recall, tuned
receiver mode and test sequence function. Finally, at the bottom of the ﬁgure are Ports 1

and 2 which give an output signal from the source and receive an input signal from the

90

device under test. Port 1 allows the user to measure Sn and S12 parameters and Port 2
measures S22 and S21. This is a simple overview of the parameter sections and their uses.
For a detailed description of all the buttons and their uses I will refer the reader to
reference [44].

Before operating the HP 8753D, the user must always be grounded. This will
ground the user and avoid damaging the device under test (DUT) with static discharge.
First, the user must slide the ground tether around the wrist and adjust it for a snug ﬁt.

Now the DUT can be connected to the HP 8753D. For this research, this includes
connecting high frequency cables to Ports 1 and 2 with a male connector on the end. For
non-packaged devices, the male connector at the end of each cable of Port 1 and Port 2
needs to be connected to a high frequency probe on either side of the probe station as
shown in Figure 57, on the following page. The probe station is described in more detail
in section 6.4. The Ports must be connected to ONLY those connections pointed to in
Figure57, because these are the only connections for the high frequency probes on the
probe station. For procedural purposes, Port 1 will always be connected to the connector
shown on the left and Port 2 will always be connected to the connector pointed to on the
right (which is beneath a metal shield, but the connector is the same as that shown to the

left in the ﬁgure).

91

 

High frequency
cable connections

 

1‘4: , ———___, -

Figure 54 Probe station connected to the HP 8753D in order to characterize the DUT

Now that the HP 8753D is connected to the probe station, the DUT is placed on a
stage inside the probe station. The stage is located inside the cavity, as shown in Figure
54. The wafer containing the SAW devices is placed on the stage and the DUT (the SAW
device probed by the user) is connected to the probes. The setup/conﬁguration of the
probes and the probing of the DUT will be explained, in detail, in section 6.3.

In order to account for mismatch in impedance and channel cross-talk, the HP
8753D needs to be calibrated, before measurements are taken. However, if there is any
uncertainty in what range the frequency response for the device lies, I would suggest
doing preliminary measurements to ﬁnd the device response, then calibrate the machine
for this range and then take measurements. This would be beneﬁcial and time saving

because once the machine is calibrated for a speciﬁc frequency range and that range is

92

then changed, the machine must then be recalibrated for the new frequency range.
Calibrating the HP is cumbersome and time consuming, so it is highly recommended that
device response range is found ﬁrst and then calibrate the machine for this range. The
calibration procedure will be discussed, in detail, in the following section. The calibration
procedure involves setting the frequency range for which the s-parameter measurements
will be taken and it also includes setting the number of points and averaging, so these
steps do not need to be included in this section.

In order to measure the s-parameters of the SAW device the following steps
should be followed. In the Active Channel box on the HP 8753D, push the CHAN 1
button in order to select channel 1 to display. Next, under the Response area on the HP,
push the SCALE REF button. Using the soft keys, next to the display, push the soft key
with AUTO SCALE next to it on the display screen. This selects the auto-scaling
function for the display. Next, under the Response area, push the button labeled
FORMAT. Using the soft keys, push the soft key with LOG MAG next to it on the
display screen; this displays the magnitude of the DUT on a logarithmic scale. Now press
the button labeled MEAS, in the Response box, on the HP 8753D. Finally, using the soft
keys choose which s-parameter to measure. The S11 parameter is the reﬂection
coefﬁcient of the forward traveling wave, S 12 is the transmission coefﬁcient for the
forward traveling wave, S21 is the transmission coefﬁcient for the reverse traveling wave
and S22 is the reﬂection coefﬁcient for the reverse traveling wave.
6.3 HP 8753D Calibration Procedures

In order to optimize the measurements of the S-parameters of the SAW devices,

the HP 8753D needs to be calibrated by doing a TRL (Through, Reﬂect Line) calibration.

93

For packaged devices to which simple cable connections can be made, one uses the HP
‘calibration kit standard’. However, for devices tested with the probe station, one must
modify the calibration kit standard to a user deﬁned standard as the ﬁrst step of the
process. The second step is to perform the TRL calibration using an impedance standard
for the 3 input high frequency probes. The steps for both of these procedures can also be
found in reference [44], which is the Hewlett Packard 8753D User’s Guide.

The substrate used as an impedance standard was Model 101-190 from Cascade
Microtech, Incorporated [45]. This is a 0.8 x 0.6 x 0.025 inch aluminum oxide substrate,
on which are repeated arrays of metallization patterns corresponding to 50 ohm loads,
‘shorts’ and ‘thru’ lines. When a probe is placed on a 50 ohm load metal pattern, a
precise 50 ohm load is presented between signal and ground for that probe. When a probe
is placed on a short, the signal is shorted to ground. An open circuit is synthesized by
raising the probes in the air at least 250 pm above the surface. When the probe is placed
on a thru line, it may be directly connected to an opposing second probe also placed on
the same through line. In this case, two probes are connected ground-to-ground and
signal-to-signal with a 1 ps delay between them. Also, on the impedance standard are thru
lines with longer delays, up to 40 ps for a length of 5.25 mm. The three different types of
calibration pads are shown generically in Figure 55, on the following page. There are 40

of each type on the chip, as well as several thru lines with longer delays.

94

 

 

 

50 Ohm Load Short (50 x 650 um bars) Thru Line (1 ps, 200 um long)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

Figure 55 Metalized patterns on the calibration standard

These are the steps that need to be followed in order to switch from ‘calibration
kit standard’ to ‘user deﬁned standard’. Push the button labeled CAL, located in the
Response area of the HP 8753D. Now using the soft keys, select CAL KIT, then
MODIFY and then DEFINE STANDARD. Next, to select a short, press 1 and then the x1.
button (the reﬂect standard is a short for our purposes). Next, press SHORT, then
SPECIFY OFFSET and OFFSET DELAY. To set this delay to 0 seconds, press 0, x1,
STD OFFSET DONE and then STD DONE (DEFINED). Next, deﬁne the thru/line
standard. To do this, press DEFINE STANDARD, 4 and x1. Then press DELAY/THRU,
SPECIFY OFFSET, OFFSET DELAY, 1, and XI. The number one is selected here
because the delay/thru line on the standard has a 1 picosecond delay and not 0 as ideally
noted in the HP User’s Guide. Now press STD OFFSET DONE and then STD DONE
(DEFINDED). Next, to deﬁne the line/match standard press DEFINE STANDARD, 6,
X]. To set the delay for the line standard push DELAY/THRU, SPECIFY OFFSET,
OFFSET DELAY, .04 and G/n. For the standard used this sets the delay at 40
picoseconds as speciﬁed by the standard veriﬁcation line. Now press STD OFFSET

DONE.

95

The standards are now deﬁned and the labels for these standards can be changed
for clarity. Press LABEL STD and using the knob in the next to the keypad modify the
name to LINE. When the new name appears press, DONE and STD DONE (DEFINED).

Now we need to assign the standards to TRL classes. In order to do this press,
SPECIFY CLASS, MORE, MORE, TRL REFLECT. Because the number 1 was selected
for the Reﬂect standard earlier press 1 and then XI. The number 6 was used for the
line/match standard press, TRL LINE OR MATCH, 6 and x1. The number 4 was used for
the thru/line standard press, TRL THRU, 4 and XI. The assignment of TRL classes is
now ﬁnished, press SPECIFY CLASS DONE.

Next we need to label the classes we have just assigned. Press LABEL CLASS,
MORE and MORE. Change the label of the TRL REFLECT class to TRL SHORT, then
change the label of TRL LINE OR MATCH class to TRL LINE and lastly change the
label of the TRL THRU class to TRL THRU. Now press LABEL CLASS DONE.

Finally, we need to label the calibration kit. Press LABEL KIT and name the kit
using a maximum of 8 characters and then press DONE. In order to save the kit in non-
volatile memory press KIT DONE (MODIFIED) then SAVE USER KIT. A user deﬁned
kit has now been deﬁned and can now be used to calibrate the HP 8753D.

Before beginning the calibration, there are some parameters that need to be set
ﬁrst. In the box labeled Response on the HP 8753D, push the button labeled AVG. Using
the soft keys push the one labeled AVERAGING and turn the averaging ON. Now push
the button labeled MENU in the Stimulus area on the HP 8753D. Using the soft keys,
push the one labeled NUMBER of POINTS and select the number of sampling points to

be taken, 201 points were used in this research. This is enough points to get an accurate

96

measurement and not too many so that the measurement takes too much time. Finally, the
frequency range needs to be set. Push the button labeled START in the Stimulus box on
the HP and type in the starting frequency, using the number pad or use the knob to dial in
the correct starting frequency and push the appropriate label, e. g. k/m for kHz or M/u for
MHz. Push the STOP button in the Stimulus box on the HP and type in the stopping
frequency and then push the appropriate label button as explained previously.

The HP is now ready to be calibrated to the probe station. There must be a TRL
calibration kit deﬁned and saved in the USER KIT. To begin the calibration, press the
button labeled CAL in the Response section of the HP 8753D. Using the soft keys, push
CAL KIT, SELECT CAL KIT, USER KIT and RETURN. Now push, CALIBRATE
MENU and TRULRM 2-PORT. To measure the TRL THRU, place the probes on either
end of the, 1 ps, thru line on the calibration standard and press THRU TRLTHRU (or
whatever you labeled this action in the user kit deﬁned previously). When the
measurement is completed the name will be underlined. To measure the TRL SHORT,
place each probe on a short bar located on the calibration standard. To measure the
reﬂection from port 1, press S11 REFL: TRLSHORT and to measure port 2, press S22
REFL: TRLSHORT. Now place the probes at either end of the longest thru veriﬁcation
line, the same one you deﬁned with a delay of 40 picoseconds when deﬁning the
calibration user kit. Press LINE/MATCH and DO BOTH FWD + REV. For this research
the Isolation calibration did not need to be performed and for this simply push OMIT
ISOLATION and then DONE TRULRM. The HP then calculates the calibration
coefﬁcients, when it has ﬁnished with the calculations the DUT may be attached and the

s-parameters may be measured. Again, as mentioned, before the same frequency range

97

used for calibration must be used when measuring the s-parameters otherwise the
calibration is null and void and the HP must be recalibrated. An error message will
appear on the display stating that calibration coefﬁcients will be turned off (not used) if
the frequency range is changed and a prompt will ask to conﬁrm that the change in
frequency will turn off the calibration coefﬁcients.

6.4 Probes/Probe Station Description

Due to the small size of the SAW devices and the frequency range in which they
are designed to operate, high frequency probes with micrometer size contacts are required
to probe and characterize the SAW devices. The use of these probes also requires the use
of a probe station. The probe station allows for the substrate to be held motionless and to
be unaffected by vibration while it is being probed.

The high frequency probes are co-planar coaxial probes. The frequency range for
the probes used in this research is DC to 40 GHz, with an insertion loss of less than .8 dB.
The structure for these probes is a ground-signal-ground (GSG) structure; this means that
there are three co-planar contacts. The outside contacts are the ground contacts and the
center contact carries the signal. In Figure 56, on the following page, is a picture of the

high frequency probe used in this research.

98

 

Figure 56 High frequency probe

The probes are connected to the probe station, as shown below, in Figure 57.

Probe connectors

    

Figure 57 Probe connections to the probe station
Extreme care and caution must be taken when handling the probes. Avoid getting
anything near the probe tips. This includes hands and fingers, as they are fragile and are
easily broken. To attach the probes, screw them in to the connectors, as labeled in the
ﬁgure above. It is easiest to attach the probes prior to connecting the cables from the HP

8753D. The probe connectors can be rotated with a wrench to connect the probes, as

99

labeled in Figure 57. If the probes are attached in this way, it avoids the issue of having to
rotate the actual probe to screw them in place, which increases the chances damaging the
probe tips. Always be sure to remove the probes from the probe station and put them
away upon ﬁnishing taking measurements.

The probe station used in this research, was shown earlier in this chapter and is
shown below in Figure 58 with different attributes labeled. A nitrogen tank is connected
to the probe station which supplies a steady ﬂow of gas to the legs of the table on which
the probe station sits. This ﬂow of gas allows the table to “ﬂoat” on a layer of nitrogen,
isolating the table from the ground. Isolating the table (i.e. the probe station) from the

ground keeps vibrations from effecting the measurements.

 

Figure 58 Probe Station

100

The camera is connected to a television monitor. Turn the monitor on and this allows the
user to see the substrate without leaning over probe station, which would risk disturbing
or perhaps damaging the substrate and/or the devices if the probe station is accidentally
bumped. The camera also allows the user to zoom in/out in order to see one device, one
section of a device, etc. For this research, it was used to see the input/output contact pads
on the device to ensure that all three points of the probes were making contact with the
pads. If the probe is not properly making contact then the s-parameters would not be
correct.

The lights can be turned on via a power box near the probe station, shown below

in Figure 59.

 

Figure 59 Power box for the probe station lights

By turning the knob on the right-hand side of the power box the lights can be turned on
and the intensity of the lights can be adjusted as well. The position of the lights can also
be adjusted so the light can be focused anywhere the user chooses. There tends to be a

glare from the light, making it difﬁcult to see, if it is focused directly on the DUT; so

101

adjusting the position of the lights is important to avoid this glare and have a clear view
of the DUT.

The device under test is placed on the stage located inside the probe station as can

be seen in Figure 60, below.

  

Figure 60 Stage located in the probe station
Before placing the substrate on the stage be sure that the probes (if already attached to the
probe station) are moved as far back, away from and above, the stage as they can be
moved; this is done in order to minimize the risk of damaging the probe tips.
6.5 Measurements/Results

S-parameter measurements were taken of several devices. To begin,
measurements were taken of two commercial devices; both were the same type of device
(a SAW ﬁlter), made by the same manufacturer, but one device measurement was taken
with the packaging in tact and the other device was measured with the top portion of the
packaging removed. The samples were mounted, as shown on the following page in
Figure 61, so the devices could be connected to the HP 8753D network analyzer. The

samples were mounted by another graduate student working with Dr. Ed Rothwell.

102

 

Figure 61 Mounted commercial SAW device

The characterization results of the commercial devices varied drastically,
depending on the presence or absence of chip packaging. For the device with the top part
of the packaging removed the s—parameter measurement of the transmission parameter,
821, resulted in a line beginning at -85 dB and decreasing to approximately

-95 dB, behaving pretty much like an open circuit, as shown below in Figure 62.

 

Magnltude (dB)

 

mama:ooaomcnmcnooaocooooooooooomcnoocooocncocomaooo
00000000000OOOOOOOOOOOOOOOOOO
+++++++++++++++++++++++++++++
LLIUJLUIJJUJUJLUUJUJUJUJLULIJUJLUUJLUUJUJUJLULUUJUJUJLUUJUJLU
ONIONONIONONmNQNVNO’NVNQv-VOQv—VOG’
'5"".".°9°‘!"‘3QQQQQQQQQQT‘TT‘TN‘Y‘Y‘YQQQ”
vvvvvvvvvvvvvmmmmmmmmmmmmmmmm
Frequency(Hz)

Figure 62 S-parameter, $2], of the commercial ﬁlter with the top of the packaging
removed

After placing a small piece of aluminum foil over the top of the remaining packaging,

results were obtained which were similar to the results of the device with the packaging

103

still intact. The s—parameter result for the device with the cap off and the aluminum foil

covering the remaining packaging are shown on the following page, in Figure 63.

 

a
3
3
3
E
9
I
0‘65 GPQPPQ’QQ’ﬁbre?QRGPQQ’GPQP@QPQPPP6PSPQPQPQR§@,§@§§
eaeeweeeeeeaeeeyesedaaaeeeewe
hésf’mhgsési’ ell-5959‘ stse'v-éséséh"\ 05.3”.“- Ié‘q’ ‘“ Raul-9&3»? g5” ‘95§L&¢h@h9 I?

quoncv (H1)
Figure 63 S-parameter, S12, of the commercial ﬁlter with the top of the packaging
removed and aluminum foil covering the remaining packaging for the device
The results obtained for the device with the packaging intact are shown, on the following
page, in Figure 64. In comparing the results from Figure 63 to the results from Figure 64,
they are similar. Figure 63 shows the transmission s-parameter (S12) results for the
commercial SAW ﬁlter. The insertion loss is approximately -20 dB and the center
frequency is approximately 480 MHz; these parameters match with those given in the

device data sheet, reference [46]. Figure 64, for the commercial device that was left intact

and measured, shows similar results. The insertion loss is approximately -20 dB and the

center frequency is approximately at 480 MHz, which again matches the manufacturer’s

data sheet speciﬁcations.

-10

 

_._.J

 

Magnitude (dB)
““5
la"
E?

I'I
-70 it ,, ~— w.

 

 

 

.100 -L ___.--L L _. .LL _ ___ LLLLLL_L___,._LL._LL.. L

9939359999SSSSSSSSSSSSSS959599933659939393365 93°65

9

XX XXXX X
9m9s,g§<~\‘°9°xgx Wmsé’993’9999 99<§<"\‘°‘°9‘°9s9'\3’<§°
‘9‘?)‘9‘9‘9 9°39“ gekehvagoaa’xaa 9‘99 9&9‘9‘99

Frequency (NH!)
Figure 64 S-parameter, S2], of the commercial ﬁlter with the packaging in tact

The S2; parameter was investigated for a SAW structure formed by aluminum
electrodes deposited on an ST-cut quartz substrate during the course of this research. The
aluminum was deposited as described in chapter 3 and patterned using the electrode mask
pattern described in chapter 4. The preliminary results obtained for the s-parameter
measurement of the experimental SAW device (SAW ﬁlter) is shown on the following

page in Figure 65.

105

 

 

 

 

 

Magnltude (dB)

 

 

 

 

 

 

 

 

-105L»———~ ,
oowcooomcnoooococncnoocooooooocommaacoaooomuomoocnmmuooom
0000000OOOOOOOOOOOOOOOOOOOOOOOOOO
+++++++++++++++++++++++++++++++++
LummuJLummLoumuJwwwwmmmwwmmwwwwwwmmwmw
oe-Neooow'mcorxcooaw—e—vamorxNQQNe—NMV-vmohm
QOQQQQQQQQQoici‘T'T'Z'T'T‘T‘T‘f‘TTZd‘Y‘Y‘Y‘Y‘Y‘Y‘Y‘YN
mmmmmnnmmmm mommmmmmmm mnmmnmmmm

FMWICV (HZ)
Figure 65 S-parameter, 821, of the experimental SAW ﬁlter

These results are for an uncalibrated frequency scan using the probe station and the
network analyzer as explained in the previous section of this chapter. A band-pass
frequency centered at approximately 315 MHz with a height of about 18 dB is illustrated.
The band-pass is not as well shaped as the packaged commercial device. This may be
partly due to the fact that the probed device was still on the quartz wafer, unpackaged and
also because the analyzer had not been calibrated.

It was realized after-the-fact that the center frequency is higher than would be
expected for the ST—cut quartz surface wave, which is 3080 m/s. Using Equation 5 and a
wavelength of 16 pm, the expected resonance would be at about 190 MHz. One possible
explanation for this is that the IDT structures can also excite bulk acoustic waves to

various extents. The bulk wave velocity would be approximately 5000 m/s which would

106

be in agreement with the center frequency. Alternatively, as previously noted, due to over
etching the electrode widths and spacing were not equal. This can facilitate generation of
higher harmonics. However, these issues were not investigated further during the course

of this research.

107

Chapter 7

Conclusions

7 .1 Introduction

This chapter will discuss conclusions based on the results found in the previous
chapters. It will also cover future work to be done with this project.
7.2 Conclusions

For the fabrication of the electrodes, it is paramount that the spacing and width of
the electrodes are calculated and fabricated properly for the desired operating frequency
of the device. Using Equation 5 one can calculate the wavelength. Then using the
calculated wavelength and Equation 6 the necessary IDT width (and hence the spacing
between the ﬁngers) can be calculated. For this investigation the wavelength was
calculated to be 16 um, using the desired operating frequency of 500 MHz. Using the
calculated wavelength, the necessary IDT width, for achieving the desired center
frequency, was found to be 4 pm. The aluminum layer must be uniform and one must
take care not to over etch when patterning the electrodes, otherwise they will be too thin
and will not work properly. The aluminum layer was successfully patterned to form the
IDTs for the SAW devices. For each IDT 20 pairs of electrodes were formed for some of
the SAW devices. The actual width of the electrodes and the spacing between them were
2.3 pm and 5.7 pm, respectively. In order to get a functioning SAW device, the proper

equipment needs to be used for fabricating these devices. There needs to be a proper

108

cleanroom environment in order to fabricate the electrodes. Due to their small size, 4 pm,
it is easy for particles to wreak havoc on the electrode fabrication process (the
photolithography steps).

There were two different SAW devices that were fabricated, a ﬁlter and an
oscillator. There were also different variations of each of these devices. Some of the
variations include inductors at one end of each IDT, different size electrodes (larger or
smaller width) and hence spacing between the ﬁngers as well and different numbers of
electrodes on some devices.

Using a reactive dc pulsed sputter process for the deposition of aluminum nitride
(AlN), a range of parameters were investigated. Some of these parameters include
frequency, reverse time, power, deposition pressure, distance to target and substrate
temperature. The recommended parameter settings for deposition frequency set to 50
kHz, the reverse time set between 3 - 5 us, power set between 375 and 450 W, set the
deposition pressure to 3 mTorr, the distance to the target should be 7.5 cm, and the
substrate temperature used was 350° C . It is vital to have the right substrate temperature
in order to get the (002) facing for the AlN layer; this allows for the desired piezoelectric
effect in the AlN. At the time of writing this thesis getting the (002) facing for the AlN
has been illusive. The full set of parameter settings are listed in Tables 8 — 10 in chapter
4. The reason for the range on the reverse time and the power is due to chamber
conditions. For each run the chamber conditions are different and as the run is in progress
the chamber conditions are constantly changing. The power and the reverse time are set

based on these conditions. With these parameter settings the AlN layer had a thickness

109

that ranged from 187 nm — 1500 nm, for different deposition times ranging from 35 - 163
minutes. The Young’s Modulus for some of these samples ranged from 280 — 353 GPa.

The AlN layer also does not need to be too thick but getting the appropriate
thickness is important. The normalized thickness needs to be .2 because this corresponds
to the maximum electromechanical coupling coefﬁcient. In this investigation the
thickness of the AlN layer needs to be 500 nm thick. To get good deposition rates an arc-
less process can not be used, it would take hours in order to get the appropriate thickness
of AlN. The deposition rates for an arcless process range from 40 - 70 nm/hr as
compared to a range of 150 — 550 nm/hr for a process that produced some arcing.

The resulting layer of AlN is transparent with an optical gap of approximately 6
eV or 200 nm. These results are consistent with literature values for AlN. Preliminary
XRD (x-ray diffraction) results show a small peak signal at 36° for AlN (002) and a
signiﬁcant AlN (004) peak at ~76°. This scan was performed on an AlN layer deposited
on UNCD. Results for samples on silicon were inconclusive. The scans show no peaks
for AlN at all, not even for Al or N. But based on other data and a simple visual
inspection it can be seen that there is a layer of AlN on the substrate. Further XRD scans
would be beneﬁcial. The surface roughness of the AlN layer, as measured on sample CF-
018 (on a Si substrate), was low with an Ra of 0.78 nm and an R2 value of 28.77 nm.
This means the deposited layers were smooth after deposition. The deposited layer also
had good adhesion to the substrate and has a columnar grain growth structure as would be
expected. The deposited layer is uniform if the substrate is rotated during deposition but

this cuts the deposition time as well. The appropriate heating and power need to be used

110

in order to get the (002) plane to grow, but this has been proven to be difﬁcult on the
PVD 75 system.

The diamond layer needs to be at least 2 — 3 ). thick so that the velocity of the
SAW is not adversely affected by the substrate layer, which has a much lower phase
velocity than the diamond. In our case, it needed to be 10 pm thick. Such a wafer was
provided for this research with UNCD deposited by Tran Dzung and was polished by
Martin Kutchler.

An AlN layer was successfully deposited on UNCD, speciﬁcally AlN/UNCD/Si
were the order of the layers. Al was also deposited on the AlN layer in order to form the
IDTs for the SAW devices. However it was not possible to etch and pattern the aluminum
because the wafer was damaged during the lithography step of processing. Due to internal
stresses in the diamond layer the sample had signiﬁcant bowing effects. The lithography
step includes a contact mask, where the sample must be ﬂush against the mask for the
photoresist to be fully exposed. When the wafer was put in contact with the mask it broke
into pieces too small to work with and the layers of AlN and diamond ﬂaked off the Si
substrate. For future work the internal stress/bowing issue needs to be addressed and/or
working with smaller samples. However, SAW devices were successfully fabricated on a
quartz — ST substrate. A signal was obtained using an HP 8753D network analyzer as
veriﬁcation that the devices would work as designed.

The diamond layer needs to be at least 2 — 3 it thick so that the velocity of the
SAW is not adversely affected by the substrate layer, which has a much lower phase

velocity than the diamond. In our case, it needed to be 10 pm thick. Such a wafer was

111

provided for this research with UNCD deposited by Tran Dzung and was polished by

Martin Kutchler.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

APPENDIX
List of Samples
Table 15 Deposition oarameters for AlN samples
Deposition Substrate Distance
Reverse Pressure Frequency Temperature to Target
_S__ar_nple # Time (us) (mTorr) (kHz) Power (W) (° C) (cm) Rotation
Cli-001 10 7.5 50 187 0 15 Yes
CF-002 10 7.5 50 200 0 1 5 Yes
CF-003 1 0 12 50 186 0 1 5 Yes
CF-004 1 0 7.5 50 201 0 1 5 Yes
_Qf-OOS 1 0 12 50 203 0 1 5 Yes
‘ CF-006 10 7.5 50 221 o 15 Yes
_(_3_E-007 1 0 7.5 50 x 0 1 5 Yes
CF-008 1 0 7.5 50 200 0 1 5 Yes
CF-009 10 7.5 50 205 0 7.5 No
£010 10 7.5 50 205 0 7.5 No
CF-011 10 7.5 50 225 0 7.5 No
CF-016 2 - 5 3 - 10 30 - 70 300 - 500 0 7.5 No
CF-O17 1 - 10 3 20 - 100 223 - 500 0 7.5 No
CF-018 4 3 50 420 0 7.5 No
CF-019 3 3 50 434 - 450 0 7.5 No
CF-020 5 3 50 364 - 420 350 7.5 No
£021 5 3 50 361 - 391 350 7.5 No
CF~022 5 3 50 367 - 375 350 7.5 No

 

Samples CF—012 to CF-015 were not AlN samples and were unrelated to this thesis

research and therefore are not included in Tables 15 and 16.

112

 

Table 16 Layer characteristics for AlN samples

113

 

 

Young's

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Layer Surface
Deposition Deposition Thickness MOd“l°“S* Roughness
Sample # Rate (nm/hr) Time (min) (nm) (Gpa) (nm) Color Uniform

£001 70.2 65 76 x Ra = 1.8 Clear Yes
£002 57 60 57 x Ra = 2.5 Clear Yes
05-003 62 60 62 x x Clear Yes
__CE-004 4O 60 40 x x Clear Yes
CF005 x 60 x x x Clear Yes
CEOOS x 60 x x x Clear Yes
CF-007 x 60 x x x _Clear Yes

Royal
CF-008 60 120 120 150 x Blue Yes

Blue,

Purple,
£009 1 50 120 300 x x Yellow No
£010 x H60 x x x x No

CF01 1 x 600 x x x ﬁx No

Purple,

CF-016 462.2 64 493 280 x geen No

Purple,

Ra = 1.57, Yellow,
CF-017 552.1 163 1500 330 R2 = 45.17 _Green No
Purple,
Ra = .78, Yellow,
CF-018 400 120 800 353 F_Iz = 28.77 G_reen No
Ra = 2.07, Pink,
CF019 320.6 35 187 x R2 = 7.28 Green No
Blue,
Pink,
Green,
Ra = 7.8, Purple,
CF020 255.8 120 512 x R2 = 47.71 Yellow No
BTue,
Pink,
Green,
Purple,
CF021 395.5 130 857 x x YeLlIow No
Blue,
Pink,
Green,
Purple,
CF-022 x 104 x x x Yellow No

 

*Young’s Modulous measurements provided by M. K. Yaran.

REFERENCES

114

 

References

1. U. Wolff, F. L. Dickert, G. K. Fischerauer, W. Greibl, C. C. W. Ruppel, “SAW
Sensors for Harsh Environments,” IEEE Sensors Journal, Vol. 1, No. 1, pp. 4-13, June
2001

2. R. Weigel, D. P. Morgan, J. M. Owens, A. Ballato, K. M. Lakin, K. Hashimoto, C. C.
W. Ruppel, “Microwave Acoustic Materials, Devices and Applications,” IEEE

Transactions on Microwave Theory and Techniques, Vol. 50, No. 3, pp. 738-749, Mar.
2002

3. R. A. McGill, R. Chung, D. B. Chrisey, P. C Dorsey, P. Mathews, A. Pique, T. E.
Misna, J. L. Stepnowski, “ Performance Optimization of Surface Acoustic Wave

Chemical Sensors,” IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency
Control, Vol. 45, No. 5, pp.1370-l380, Sept. 1998

4. L. M. Reindl, A. Pohl, G. Scholl, R. Weigel, “SAW-Based Radio Sensor Systems,”
IEEE Sensors Journal, Vol. 1, No. 1, pp. 69-78, June 2001

5. J. T. Glass, B. A. Fox, D. L. Dreifus, B. R. Stoner, “Diamond for Electronics: Future
Prospects of Diamond SAW Devices,” MRS Bulletin, pp. 49-55, Sept. 1998

6. H. Meier, T. Baier, G. Riha, “Miniaturization and Advanced Functionalities of SAW
Devices,” IEEE Transactions on Microwave Theory and Techniques, Vol. 49, No. 4, pp.
743-748, Apr. 2001

7. H. Nakahata, S. Fujii, K. Higaki, A. Hachigo, H. Kitabayashi, S. Shikata, N. Fujimori,
“Diamond-Based Surface Acoustic Wave Devices,” IoP Electronic Journals,
Semiconductor Science and Technology, Vol. 18, No. 3, 896-8104, Mar. 2003

8. S. Shikata, H. Nakahata, A. Hachigo, N. Fujimori, “High Frequency Bandpass Filter
Using Polycrystalline Diamond,” Diamond and Related Materials, 2, pp. 1197-1202,
1993

9. S. Shikata, H. Nakahata, A. Hachigo, “New Material Systems for Diamond Surface
Acoustic Wave Devices,” New Diamond and Frontier Carbon Technology, Vol. 9, No. 1,
pp. 75-92, 1999

10. K. Higaki, H. Nakahata, H. Kitabayashi, S. Fujii, K. Tanabe, Y. Seki, S. Shikata,
“High Power Durability of Diamond Surface Acoustic Wave Filter”, IEEE Transactions

on Ultrasonics, Ferroelectrics and Frequency Control, Vol. 44, No. 6, pp. 1395-1400,
Nov. 1997

11. M. Ishihara, T. Nakamura, F. Kokai, Y. Koga, “Preparation of Lithium N iobate Thin

Films on Diamond-Coated Silicon Substrate for Surface Acoustic Devices,” Diamond
and Related Materials, 12, pp. 1809-1813, 2003

115

12. V. Mortet, M. Nesladek, J. D’Haen, G. Vanhoyland, O. Elmazria, M. B. Assouar, P.
Alnot, M. D’Olieslaeger, “Deposition of Aluminum Nitride Film By Magnetron

Sputtering for Diamond-Based Surface Acoustic Wave Applications,” Physica Status
Solidi (a), Vol. 193, No. 3, pp. 482-488, 2002

13. P. R. Chalker, T. B. Joyce, C. Johnston, J. A. A. Crossley, J. Huddlestone, M. D.
Whitﬁeld, R. B. J ackman, “Fabrication of Aluminum Nitride/Diamond and Gallium
Nitride/Diamond SAW Devices,” Diamond and Related Materials, 8, pp. 309-313, 1999

14. H. Nakahata, A. Hachigo, K. Higaki, S. Fujii, S. Shikata, N. Fujimori, “Theoretical
Study on SAW Characteristics of Layered Structures Including a Diamond Layer,” IEEE

Transactions on Ultrasonics, Ferroelectrics and Frequency Control, Vol. 42, N o. 2, pp.
362-375, May 1995

15. D. M. Gruen, A. R. Krauss, US Patent Speciﬁcation No. 5.772.760, 1998
16. D. M. Gruen, Annual Review of Materials Science, 29, p. 211, 1999

17. F. Bénédic, M. B. Assouar, F. Mohasseb, O. Elmazria, P. Alnot, A. Gicquel, “Surface
Acoustic Wave Devices Based on Nanocrystalline Diamond and Aluminum Nitride,”
Diamond and Related Materials, 13, pp. 347-353, 2004

18. B. Bi, W. S. Huang, J. Asmussen, B. Golding, “Surface Acoustic Waves on
Nanocrystalline Diamond,” Diamond and Related Materials, 11, pp. 677-680, 2002

19. A. A. Oliner, G. W. Farnell, H. M. Gerard, E. A. Ash, A. J. Sloabodnik, Jr., H. 1.
Smith, Topics in Applied Physics: Acoustic Surface Waves, Vol. 24, Springer-Verlag,
Berlin, Heidelberg, New York 1978

20. B. A. Auld, Acoustic Fields and Waves in Solids, Vol. 1, John Wiley & Sons, New
York 1973

21. L. Solymar, D. Walsh, Lectures on the Electrical Properties of Materials, Oxford
University Press, New York 1988

22. C. M. Flannery, M. D. Whitﬁeld, R. B. J ackman, “Acoustic Wave Properties of CVD
Diamond,” Semiconductor Science and Technology, 18, pp. S86-S95, 2003

23. O. Elmazria, V. Mortet, M. El Hakiki, M. Nesladek, P. Alnot, “High Velocity SAW
Using Aluminum Nitride Film on Unpolished Nucleation Side of Free-Standing CVD

Diamond,” IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, Vol.
50, No. 6, pp. 710-715, Jun. 2003

116

24. T. Lamara, M. Belmahi, O. Elmazria, L. Le Brizoual, J. Bougdira, M. Rémy, P.
Alnot, “Freestanding CVD Diamond Elaborated By Pulsed-Microwave-Plasma for
ZnO/Diamond SAW Devices,” Diamond and Related Materials, 13, pp. 581-584, 2004

25. M. Ishihara, T. Nakamura, F. Kokai, Y. Koga, “Preparation of AlN and LiNbO3 Thin
ﬁlms on Diamond Substrates By Sputtering Method,” Diamond and Related Materials,
11. Pp. 408-412, 2002

26. M. B. Assouar, O. Elmazria, L. Le Brizoual, P. Alnot, “Reactive DC Magnetron
Sputtering of Aluminum Nitride Films for Surface Acoustic Wave Devices,” Diamond
and Related Materials, 11, pp. 413-417, 2002

27. T. Makkonen, V. P. Plessky, W. Steichen, W. Steichen, “Surface-Acoustic-Wave
Devices for the 2.5-5 GHz Frequency Range Based on Longitudinal Leaky Waves,”
Applied Physics Letters, Vol. 82, No. 19, pp. 3351-3353, May 2003

28. G. G. Iriarte, G. Engelmark, I. V. Katardjiev, V. Plessky, V. Yantchev, “SAW COM-
Parameter Extraction in AlN/Diamond Layered Structures,” IEEE Transactions on
Ultrasonics, Ferroelectrics and Frequency Control, Vol. 50, No. 11, pp. 1542-1547, Nov.
2003

29. A Frass, G. Lehmann, A. Lomonosov, P. Hess, “Linear and Nonlinear Elastic Surface
Waves: From Seismic Waves to Material Science,” Analytical Sciences, Vol. 17, pp S9-
S12, Apr. 2001

30. M. Benetti, D. Cannata, F. Di Pietrantonio, E. Verona, “ Growth of AlN Piezoelectric
Film on Diamond for High-Frequency Surface Acoustic Wave Devices,” IEEE
Transactions on Ultrasonics, Ferroelectrics and Frequency Control, Vol. 52, No. 10, pp.
1806-1811, Oct. 2005

31. B. Chapman, Glow Discharge Processes, John Wiley & Sons, New York 1980

32. D. Carter, H. Walde, G. McDonough, G. Roche, “Parameter Optimization in Pulsed
DC Reactive Sputter Deposition of Aluminum Oxide,” Society of Vacuum Coaters, pp.
570-577, 2002

33. A. Belkind, A. Freilich, J. Lopez, Z. Zhao, W. Zhu, K. Becker, “Characterization of
Pulsed DC Magnetron Sputtering Plasmas,” New Journal of Physics, Vol. 7, No. 90, pp.
2-15, Apr. 2005

34. Kurt J. Lesker Operation Manual PVD Series, Version

35. Advanced Energy User Manual MDX 1.5K Magnetron Drive, Apr. 1997

36. Advanced Energy User Manual Spare-1e V 100 kHz, Nov. 2002

117

37. Kurt J. Lesker M.A.P.S. Power Supply Installation and Operation Manual, Version
1.2

38. Watlow Series SD User’s Manual PID Controller and PID Proﬁling Controller, Apr.
2004

39. Watlow Series SD User’s Manual Limit Controller, Dec. 2002

40. Kurt J. Lesker TORUS Magnetron Sputtering Sources Installation and Operation
Manual, Version 2.2

41. M. B. Assouar, O. Elmazria, M. Elhakiki, P. Alnot, “Study of Structural and
Microstructural Properties of AlN Films Deposited On Silicon and Quartz Substrates For

Surface Acoustic Wave Devices,” Journal of Vacuum Science and Technology B, Vol.
22, No. 4, pp. 1717-1722, Jul/Aug. 2004

42. F. Engelmark, G. F. Iriarte, I. V. Katardjiev, M. Ottosson, P. Muralt, S. Berg,
“Structural and Electroacoustic Studies of AlN Thin Films During Low Temperature

Radio Frequency Sputter Deposition,” Journal of Vacuum Science and Technology A,
Vol. 19, No. 5, pp. 2664-2669, Sept/Oct. 2001

43. D. M. Pozar, Microwave Engineering, Addison-Wesley, Reading, MA, 1990, p. 221.

44. Hewlett Packard User’s Manual HP 8753D Network Analyzer, Dec. 1997

45. Cascade Microtech Inc., 2430 NW 206th Avenue, Beaverton, Oregon 97006, USA

46. ECS Inc. International, ECS-D479.5B/D480A Series SAW Filter Data Sheet, 1105 S.
Ridgeview, Olathe, KS 66062, USA

118

   

lllllllllllllllllI