Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Development of optical devices for applications in photonic integrated circuit and sensing
(USC Thesis Other)
Development of optical devices for applications in photonic integrated circuit and sensing
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
DEVELOPMENT OF OPTICAL DEVICES FOR APPLICATIONS IN
PHOTONIC INTEGRATED CIRCUIT AND SENSING
by
Xiaomin Zhang
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MATERIALS SCIENCE)
December 2013
Copyright 2013 Xiaomin Zhang
ii
Acknowledgments
First of all, I would like to thank my academic advisor, Prof. Andrea Armani, for
her support and guidance throughout my PhD study at USC. I am so happy to say that the
five years I have spent in her research group is truly fruitful and enjoyable. I feel very
fortunate to be in her research group and also being one of her very first PhD students. I
had the chance to witness her transformation of a new lab to now a well-established, big
and promising research group and I have learned so much from her.
I am always extremely impressed by her expertise in research and positive attitude.
She gives very interesting and challenging projects to start and also encourages me to
think of and pursue my own ideas. She offers invaluable advice on every aspect of
research from experimental skills, project management to presentations. Her suggestion
about how to properly write a lab notebook has benefited me greatly and I have become
so much more organized in research. Her positive attitude in research has influenced me
greatly. One time I came to meeting with her with data not working and felt extremely
frustrated. But after discussions, she said to me “Excellent! At least we know this
approach doesn’t work!” I am also extremely impressed by her patience and her
dedication to help her students succeed. She tirelessly answers the tons of questions and
helps with all kinds of problems I have encountered during my experiment. Whenever I
have questions or concerns, she is never too busy to help despite her very busy schedule.
Her communication skills have also impressed and impacted me a lot. No matter it’s a
meeting or simply an email conversation, she is always clear and insightful. With all that
said, Prof. Armani is not only an academic advisor but also my role model in many
aspects.
iii
I would like to thank all of the group members for creating a professional yet fun
working environment and I have enjoyed working with all of you. I would particularly
like to thank the “waveguide people”: Ashley Maker, Mark Harrison, Soheil Soltani,
Audrey Harker, Kelvin Kuo and Dr. Rasheeda Hawk for their inspirational discussions
and collaborations. Thank you to Ce Shi, Simin Mehrabani and Maria Chistiakova for all
the laughs. Special thanks to Dr. Jason Gamba for his positive comments about my
resume which helped my initial acceptation to Prof. Armani’s research group.
I would like to thank my friends and colleagues I met at USC. Thank you to Lin
Zhang, Yang Yue, and Hari Mahalingam for their helpful discussions. I would also like
to thank Tingwei-Yeh, Youngki Choe, Qian Zhang and Changgeng Liu for their advice
about fabrication in cleanroom. Thank you to Jing Ma for the lunches and outings we
have enjoyed together.
Last but not least, I would like to thank my beloved family. I would like to express
my deepest gratitude to my parents for their love, support and sacrifice over the years. I
feel so blessed to be in a family full of love and care. Thank you to my grandparents, my
Uncle Huadong Wei and Aunt Maoju Wei for their love and encouragement. Thank you
to Yanjie Wang for your love and accompany. You are my harbor away from home.
iv
Table of Contents
Acknowledgments............................................................................................................... ii
Table of Contents ............................................................................................................... iv
List of Tables .................................................................................................................... vii
List of Figures .................................................................................................................. viii
Abstract ............................................................................................................................ xvi
Chapter 1 Introduction ..................................................................................................... 1
1.1 Motivation .......................................................................................................... 1
1.2 Chapter overview .............................................................................................. 3
Chapter 2 Overview of optical waveguides and microtoroidal resonators ...................... 5
2.1 Introduction to optical waveguides ................................................................. 5
2.2 Background of optical waveguides .................................................................. 7
2.2.1 Total internal reflection................................................................................... 7
2.2.2 Numerical aperture.......................................................................................... 9
2.2.3 Propagation loss ............................................................................................ 10
2.2.4 Bending loss .................................................................................................. 12
2.3 Introduction to whispering gallery mode optical resonators ...................... 13
2.4 Background of optical resonators ................................................................. 17
2.4.1 Quality factor ................................................................................................ 17
2.4.2 Free spectral range ........................................................................................ 19
2.5 Testing procedure for waveguides and integrated microtoroid ................. 20
2.5.1 Testing procedure for waveguides ................................................................ 20
2.5.2 Testing procedure for standing alone microtoroid ........................................ 21
2.5.3 Testing procedure for integrated microtoroid-waveguide system ................ 24
2.6 Finite element method simulations ................................................................ 25
2.6.1 Optical modes ............................................................................................... 26
2.6.2 Waveguide simulations ................................................................................. 27
2.6.3 Microtoroid simulations ................................................................................ 29
Chapter 2 References .................................................................................................. 31
Chapter 3 Suspended waveguides and splitters ............................................................. 32
3.1 Introduction ..................................................................................................... 32
3.2 Trapezoidal waveguide ................................................................................... 33
3.2.1 Fabrication of trapezoidal waveguide ........................................................... 33
3.2.2 FEM modeling and finite differential time domain simulations ................... 39
3.2.3 Experimental setup........................................................................................ 43
v
3.2.4 Experimental results and discussion ............................................................. 44
3.3 Silica Beam Splitter......................................................................................... 50
3.3.1 Fabrication of the splitter .............................................................................. 50
3.3.2 Experimental results and discussion ............................................................. 56
3.4 Conclusions ...................................................................................................... 60
Chapter 3 References .................................................................................................. 62
Chapter 4 Ultimate quality factor of silica microtoroid resonant cavities ..................... 64
4.1 Introduction ..................................................................................................... 64
4.2 Theoretical calculations .................................................................................. 64
4.3 Experimental verification ............................................................................... 67
4.3.1 Fabrication of microtoroids........................................................................... 67
4.3.2 Refractive index measurements .................................................................... 70
4.3.3 Quality factor testing and analysis ................................................................ 71
4.4 Conclusions ...................................................................................................... 73
Chapter 4 References .................................................................................................. 75
Chapter 5 Integrated microtoroid-waveguide system .................................................... 77
5.1 Introduction ..................................................................................................... 77
5.2 The design concept .......................................................................................... 78
5.3 Initial fabrication tries .................................................................................... 81
5.3.1 Fabrication with 4 µm thick thermal oxide on silicon .................................. 83
5.3.2 Fabrication with PECVD silica ..................................................................... 86
5.3.3 Fabrication with 12µm thick thermal oxide .................................................. 87
5.4 One of the fabrication routes which enables microtoroid-waveguide system
with sub-micron gap ................................................................................................... 91
5.4.1 First Cr deposition......................................................................................... 94
5.4.2 First photolithography and Cr wet etching ................................................... 94
5.4.3 First AOE etching ......................................................................................... 96
5.4.4 Second Cr deposition .................................................................................... 98
5.4.5 Second photolithography and Cr etching .................................................... 100
5.4.6 Second AOE etching ................................................................................... 104
5.4.7 XeF
2
etching................................................................................................ 105
5.4.8 CO
2
laser reflow .......................................................................................... 109
5.5 The other fabrication routes which enables microtoroid-waveguide system
with sub-micron gap ................................................................................................. 114
5.5.1 Detailed fabrication process ........................................................................ 116
vi
5.5.2 Fabrication result ........................................................................................ 117
5.6 Characterization of the microtorid-waveguide system ............................. 118
5.7 Analytical calculation of free spectral range .............................................. 120
5.8 Conclusions .................................................................................................... 121
Chapter 5 References ................................................................................................ 122
Chapter 6 Integrated microtoroid-waveguide system sensors ..................................... 123
6.1 Introduction ................................................................................................... 123
6.2 Power dependence ......................................................................................... 124
6.3 Temperature and UV sensing demonstrations ........................................... 125
6.3.1 Detection mechanism and testing set-up..................................................... 125
6.3.2 Temperature sensing ................................................................................... 128
6.3.3 UV sensing .................................................................................................. 129
6.4 Conclusions .................................................................................................... 130
Chapter 6 References ................................................................................................ 132
Appendix A: Hybrid Silica-Polymer Ultra-High-Q Microresonators ............................ 133
A.1: Introduction....................................................................................................... 133
A.2 Theory ................................................................................................................. 133
A.2.1 Loss mechanism ............................................................................................ 133
A.2.2 FEM simulations ........................................................................................... 134
A.3 Fabrication of the hybrid device ....................................................................... 137
A.4 Experimental result and discussions ................................................................ 137
A-5 Conclusions ......................................................................................................... 139
Appendix A References ............................................................................................ 141
vii
List of Tables
Table 3-1 Comparison of the effective refractive indices of TE and TM modes 40
Table 4-1 The refractive indices of different thermal oxide with different doping
concentrations in the substrate at three wavelengths and the corresponding measured
quality factors of the microtoroids fabricated on these different wafers. 69
Table A-1 Summary of Model and Experimental Fit Parameters. 139
viii
List of Figures
Figure 2-1(a) SEM image of a silicon strip waveguide (Green, W.M., et al.) and (b)
demonstration of red light (633nm) guided in a polymer optical waveguide
fabricated on a printed circuit board (Leng, Y., et al.) 6
Figure 2-2 Illustrations of various types of waveguide geometries: (a) ridge (b) embedded;
(c) immersed. Grey color indicates materials with higher refractive index and where
light propagates in. 7
Figure 2-3 (a) the ray is incident from the medium of smaller refractive index n
1
< n
2,
(b)
the ray is incident from the medium of greater refractive index (n
1
> n
2
). (c) For n
1
>
n
2
, if the angle of incidence is greater than critical angle, it will undergo total
internal reflection. 8
Figure 2-4 A dielectric waveguide which is composed of a thin layer of high-index
material (core index n
1
) sandwiched between two layers of low index (cladding
index n
2
) material. 10
Figure 2-5 Propagation loss measurement: the transmission loss (dB) from the
waveguides over a series of different 11
Figure 2-6 When light rays reach a bend, rays with incidence angle smaller than the
critical angle become unguided and lost. 12
Figure 2-7(a) Whispering gallery in St. Paul’s Cathedral, the acoustic wave (red arrow)
travels along the gallery (b) Snapshot of an acoustic whispering-gallery mode
calculated at a frequency of 69 Hz in a cylinder filled with air with the same
diameter as the whispering gallery in St Paul’s Cathedral. Red and blue spots
represent higher and lower whisper intensity, respectively. As indicated from the
location of the acoustic wave pattern, the acoustic wave propagates along the
periphery. 14
Figure 2-8 (a) A microtoroid resonator coupled with an optical fiber taper, the light
(shown in red) travels along the periphery of the resonator (b) optical field
distribution of a whispering gallery mode inside the microtoroid cavity. 15
Figure 2-9 (a) the wavelength of the light fits an integer number of times of the resonator,
the resonator is on resonance at that wavelength (b) the resonator is off resonance 15
Figure 2-10 Different types of whispering gallery mode optical resonators 17
ix
Figure 2-11 Microscopic image showing (a)Microscopic image showing a lensed fiber
coupled to a waveguide (b) two output of a splitter, an aspheric lens is to focus the
output into a beam profiler instead of a lensed fiber 21
Figure 2-12 All the components of testing setup for waveguide and splitter 21
Figure 2-13 (a) Testing setup for standing alone microtoroid (b) microscopic image of a
microtoroid coupled to an optical fiber taper 22
Figure 2-14 Components of the testing setup for a standing alone microtoroid 23
Figure 2-15 Microscopic image showing (a) lensed coupled to the input of the waveguide
(b) the microtoroid-waveguide system (c) lensed coupled to the output of the
waveguide 24
Figure 2-16 Components of the testing setup for the integrated microtoroid-waveguide
system 25
Figure 2-17 Meshing created to find a solution to a electromagnetic problem in FEM
software 26
Figure 2-18 Electric field distribution of the TE fundamental mode of a trapezoidal shape
silica waveguide at 1550nm 28
Figure 2-19 Magnetic field distribution of the TE fundamental mode of a trapezoidal
shape silica waveguide at 1550nm 29
Figure 2-20 Transverse component of electric field squared intensity distribution of a TE
fundamental mode of a microtoroid with major and minor diameter for 37.5 µm and
5µm respectively (a) Surface plot of the mode (b) electric field plot along y=0 axis
30
Figure 3-1 Fabrication process of the trapezoidal waveguide (a) 80µm wide silica
trapezoids are defined using photolithography and buffered HF etching. (b) A
second photolithography and buffered HF etching are used to create two 8µm
trapezoids at each side of the 80-µm trapezoids, and (c) XeF
2
etching is used to
isotropically undercut the silica. (d) Cross section schematic of the waveguide
indicating the critical dimensions. In the present experiments, W=80µm, w=64µm
and h=2µm. 34
Figure 3-2 Scanning electron micrograph (SEM) images of the trapezoidal waveguide (a)
SEM side view image of a single waveguide SEM, (b) SEM side view image of a
pair of waveguides (c) top view image of a pair of waveguides. 36
x
Figure 3-3 Scanning electron micrograph (SEM) images of the trapezoidal waveguides (a)
when the membrane thickness is much less than 900nm (b) zoomed out view (c)
when the amount of undercut is much more than 10µm 37
Figure 3-4 Scanning electron micrograph (SEM) images of the curved trapezoidal
waveguide (a) The S-curve waveguide geometry depicting the bending radius (R),
which was varied between 125 and 400µ m. (b) SEM of the bending section and (c)
cleaved end of a curved waveguide device. 38
Figure 3-5 Finite element method simulation of the electric field distribution (TE mode)
of (a) trapezoidal, (b) rectangular waveguide and (c) circular waveguide at 1550nm.
(d) Comparison of the intensity of the mode profile along the x (horizontal) direction,
normalized to highest intensity in trapezoidal device. 40
Figure 3-6 Finite-difference time-domain simulation of the electric field intensity bent
trapezoidal waveguide with an inner radius of 75µm at 1550nm. The optical field
clearly leaks into the silica membrane in-between the two waveguide arms. This
leakage is a significant source of loss in the serpentine devices. 43
Figure 3-7 Measured propagation loss of the trapezoidal silica waveguide at three
wavelengths: 658, 980 and 1550nm 45
Figure 3-8 Transmission at four different polarization states 47
Figure 3-9 Measured bending loss of the trapezoidal silica waveguide at 658, 980 and
1550nm and the FDTD results 48
Figure 3-10 The output power of the trapezoidal waveguide shows a linear dependence
on the input power up to 200mW. 50
Figure 3-11 Overview of the fabrication process of the splitter (a, d) Rendering and
cartoon of the dual photolithography and buffered HF etching proceedure to achieve
the rectangular shaped silica waveguides with a taperd region, (b,e) Rendering and
cartoon of the XeF
2
etching to isotropically undercut the silica structure, (c)
Rendering of the CO
2
laser reflow to form the circular silica waveguide channels
shown in operation and (f) Cartoon of CO2 laser reflow to form the circular silica
waveguide channels. 51
Figure 3-12 Composite optical microscope image of the splitter 52
Figure 3-13 SEM image of the splitter which has two input and output silica circular
waveguide channels with a suspended, tapered region 53
xi
Figure 3-14 SEM image of the end of the splitter showing two circular silica waveguides
53
Figure 3-15 Microscopic images showing fabrication results of directional couplers (a)
the coupler before reflow with very thin (less than 1µm) silicon pillar (b) reflow
result which shows wavy waveguides as a result of the thickness ununiformity 55
Figure 3-16 Optical characterization of the 2x2 splitter The a) splitting ratio versus
wavelength and b) excess loss versus wavelength were measured from 1520 to
1630nm. Both metrics are quite smooth and flat over the wavelength range. 56
Figure 3-17 Wavelength dependence of splitting ratio showing oscillatory variations
when the alignment is imperfect 57
Figure 3-18 Output intensity profile from the 2x2 silica splitter showing a 50/50 splitting
ratio with low crosstalk and high uniformity. 58
Figure 3-19 The splitting ratio (red circles, black diamonds) and output power (blue
squares) versus input power. The splitting ratio is constant over the entire range and
the output power changes linearly with the input power. Both indicate that the
device’s behavior is not being affected by any non-linear effects over this range of
input power 59
Figure 3-20 Polarization dependence of the splitter. The splitting ratio doesn’t change at
each polarization state at each wavelength 60
Figure 4-1 The calculated dependence of the quality factor on the refractive index and the
resonant wavelength. The Q factor decreases as the refractive index decreases and
the wavelength increases. 67
Figure 4-2 Renderings of the fabrication process: (a) photolithography and buffered HF
etching are used to form circular silica pads on a silicon substrate, (b) XeF2
undercuts the oxide, forming silica microdisks sitting on silicon pillars, (c) a CO
2
laser reflows the silica, forming silica microtoroids on silicon pillar 68
Figure 4-3 Microscopic images of the device at each step: (a) silica pad, the wedged
shape is a result of isotropic buffered HF etching, (b) silica microdisk, the bright
circle in the middle is the silicon pillar underneath (c) silica microtoroid 68
Figure 4-4 A scanning electron micrograph of the fabricated silica microtoroid resonator
68
xii
Figure 4-5 The acquired ellipsometric parameters and fit to a Si-SiO
2
material
system. The solid (open) squares, triangles and circles represent ( ) recorded at
64, 69 and 74 degrees. The solid lines are fit data. 70
Figure 4-6 A fine scan (the forward scan direction) of the fundamental transverse mode
of the microtoroid fabricated from the film with boron concentration of 1.63× 10
14
cm
-3
in silicon at 848.8nm with a dual-Lorentz fit (red). The resonance shows
splitting and the quality factor of the left and right peak is 3.39× 10
8
and 2.67× 10
8
separately. 72
Figure 4-7 Measured quality factor as a function of refractive index or boron dopant
concentration at three wavelengths: 630, 850 and 980nm. The Q factor decreases as
the refractive index decreases and the wavelength increases. 73
Figure 5-1 schematics of the cross section of a microdisk (a) and a microtoroid (b), the
reflow process transforms the microdisk to a microtoroid with smaller manor
diameter 79
Figure 5-2 schematics of the proposed design concept, (a) cross section of a microring
with a waveguide (b) microtoroid and reflowed waveguide with a sub-micron gap as
a result of reflow and volume conservation 80
Figure 5-3 Mask design schematic: the blue layer is designed for the membrane and the
grey layer is for the microring and waveguide; two different gaps of 3µm and 5µm
respectively is designed to optimize the gap 81
Figure 5-4 schematics of fabrication process: (a) shape after first AOE etching (b) shape
after second AOE etching (c) XeF
2
undercut (d) CO
2
laser reflow 83
Figure 5-5 optical microscopic images of the device after two step AOE etching (a), XeF
2
etching(b), and CO
2
laser reflow(c) and SEM image of the device 84
Figure 5-6 improved fabrication result with 4 µm thick thermal oxide on silicon: (a)
microscopic image of the device after XeF
2
etching, (b) after CO
2
laser reflow 85
Figure 5-7 Quality factor tested with an optical fiber taper of the microtoroids fabricated
using AOE etching 86
Figure 5-8 microtoroid and waveguide system fabricated using 10µm thick PECVD silica
(a) microscopic image of the device before reflow, (b) after reflow 87
Figure 5-9 microscopic images of the device fabricated using 12µm thick thermal oxide
(a) before reflow (b) after reflow 88
xiii
Figure 5-10 SEM images showing the side view of the reflowed device (a) and the gap
between the microtoroid and the waveguide (b) 88
Figure 5-11 microscopic image of the device right after second AOE etching (a) and after
additional cleaning (b) 89
Figure 5-12 SEM images showing the top view of the reflowed device (a) and the
reflowed microtoroid and waveguide system (b) 90
Figure 5-13 SEM images showing the side view of the reflowed device (a) and the gap
between the microtoroid and the waveguide (b) 91
Figure 5-14 the schematics of the new mask 92
Figure 5-15 schematic of the fabrication process for the two time AOE etching (a) after
first time AOE etching (b) after second time AOE etching 93
Figure 5-16 Microscopic image of the device after first photolithography and Cr wet
etching 96
Figure 5-17 Microscopic image of the device after the first AOE etching and Matrix O
2
plasma stripping 98
Figure 5-18 (a) microscopic image of a waveguide with photoresist residue on top and
caused (b) the subsequent Cr deposition peeling off 99
Figure 5-19 (a) SEM image of the side view of the waveguide deposited with 550nm Cr,
(b) zoomed in view of the sidewall 100
Figure 5-20 SEM image of the second photolithography with AZ4620 101
Figure 5-21 SEM images of the photoresist at different hard baking temperatures and Cr
wet etching profile at that temperature: (a) 130 degree for 5min (b) 120 degree for
5min(c) 115 degree for 5min 103
Figure 5-22 SEM side view of the cross section of the waveguide right after the second
AOE etching 105
Figure 5-23 Oxide not etched away in the gap region as indicated by the red arrows 105
Figure 5-24 Miroscopic images showing XeF2 etching result of (a) not well cleaned
device (b) well cleaned device 106
Figure 5-25 SEM image showing the microring and the waveguide bridged together 107
xiv
Figure 5-26 Optical microscopic image showing the remaining silicon near the gap region
(indicated by red arrows) 107
Figure 5-27 SEM images of the device after XeF
2
etching (a) top view (b) angled view (c)
side view 108
Figure 5-28 SEM image showing the amount of proper undercut, red arrows indicate
where the inner sidewall and the silicon pillar are 109
Figure 5-29 microscopic image of the device (a) before reflow (b) after reflow with a
submicron gap 111
Figure 5-30 SEM image of a microtoroid-waveguide system with submicron gap 112
Figure 5-31 Microscopic images of a reflowed device (a) with submicron gap, (b) after
reflow with a higher power again, the gap increases beyond submicron. 113
Figure 5-32 Microscopic images of a device (a) reflowed at certain power which didn’t
reflow thoroughly (b) after reflow with a higher power again, the microtorod and the
waveguide are reflowed together. 114
Figure 5-33 schematics of the device after (a) first AOE etching (b) second AOE etching
115
Figure 5-34 Microscopic images of the device (a) before reflow (b) after reflow 118
Figure 5-35 The optical characterization of the integrated ultra-high-Q microtoroid
resonant cavity. a) Transmission spectra of the 70 µm diameter integrated
microtoroid with a bent waveguide with a radius of curvature of 100µm. The free
spectral range of this device is 5.53nm. b) An example resonance spectra with the
lorentzian fit. The full width at half maximum (dl) from the fit is 3.1 × 10
-4
, yielding
a loaded Q value of 4.3× 10
6
. 119
Figure 5-36 a) Calculated effective refractive index for a range of major and minor
diameter combinations. b) Calculated free spectral range for a range of major and
minor diameter combinations. 121
Figure 6-1 As the input power is increased, the resonant wavelength slightly shifts.
However, because of the low thermo-optic coefficient of silica, several hundred
microwatts of input power, which corresponds to several watts of circulating power,
are needed to induce a multi-linewidth shift. 125
xv
Figure 6-2 The testing set-ups for the detection experiments. (a) The cylindrical heater is
integrated directly under the integrated resonator, and the thermocouple is adjacent
to the sample. (b) The UV lamp is position directly above (13mm gap) the resonant
cavity. 127
Figure 6-3 Temperature sensing experiments. (a) Sensor response when the temperature
is increased. Inset: The histogram from the noise measurement with a Gaussian fit.
(b) The results from part (a) are re-plotted to highlight the relationship between the
resonance shift and the temperature change. The solid line is the linear fit. 129
Figure 6-4 UV sensing results. (a) Sensor response with several different exposure
intensity cycles, increasing from 54mW/ cm
2
to 100mW/ cm
2
and then decreasing to
54mW/ cm
2
(b) The characteristic UV sensing curve showing both the forward and
reverse response at 100mW. The resonance undergoes a large, rapid wavelength
shift once the UV is turned on. When the UV is turned off, the resonant wavelength
returns to its original value. Inset: The histogram from the noise measurement with
a Gaussian fit. 130
Figure A-0-1 (a) Scanning electron micrograph of a toroidal microresonator. Finite
element method simulation results for the optical field intensity distribution: (b)
silica resonant cavity, (c) hybrid resonant cavity with a 100nm thick PS film, and (d)
200nm thick PS film. Note that the optical field shifts from the silica towards the
polymer film as the thickness of the polymer film increases. The toroid major
(minor) diameter is 40(8)m and λ=850nm. 135
Figure A-0-2 Finite element method results. (a) Percentage of the optical field in the
polymer layer as a function of the minor diameter as the major diameter increases
from 40 to 100μm at λ=980nm. The PMMA film thickness is fixed at 500nm. (b)
Percentage of the optical field as a function of the polymer film thickness for
PMMA and PS at λ=850, 980nm. 136
Figure A-0-3 Quality factor (Q) and the percentage of the optical field for the hybrid
polymer resonant cavity devices as a function of coating thickness. PMMA at (a)
850nm and (b) 980nm. PS at (c) 850nm and (d) at 980nm. The results were fit to an
equation of the form y=ax
b
, which is included in the figure as a solid (dashed) red
line for the theoretical (experimental) Q results. The parameters (a,b) are
summarized in Table 1. The black dotted line indicates the highest Q demonstrated
with a silica toroidal resonant cavity to date, setting an upper bound on Q [17]. 139
xvi
Abstract
High qualify factor (Q) integrated micro-cavities are not only one of the key
elements in integrated photonics but also have numerous other applications such as
fundamental physics studies and biosensing. Since the invention of silica ultra-high
quality factor micro-disks and micro-toroids, they have proven to be very efficient in
these applications. However, currently, optical fiber tapers are used to couple light into
the micro-disks and toroids. In order to use them in practical applications, they have to be
fabricated with a waveguide on the same substrate. The focus of this dissertation is the
design of a new waveguide which enables the development of a fully integrated
waveguide-resonator system.
The silica integrated waveguides which are directly fabricated on a silicon substrate
are demonstrated in the first part of the thesis. Silica splitters based on the silica
waveguides are demonstrated as another essential element in photonic circuitry. The loss
mechanisms which impact the quality factor of micro-toroids are investigated both
experimentally and theoretically. Then, a novel approach for fabricating fully-integrated
silica ultra-high Q toroidal resonators integrated with on-chip waveguides is
demonstrated. Temperature sensing, UV sensing and power dependence experiments are
performed using this fully integrated ultra-high Q toroidal resonator system.
1
Chapter 1 Introduction
1.1 Motivation
Integrated optical devices such as waveguides and micro-cavities can find
applications including integrated photonics, fundamental physics studies and sensing.
Among the materials that are available to fabricate these optical devices such as
semiconductors, dielectrics, polymers and metals, silica is an ideal material for most
photonics and sensing applications, as it inherently has ultralow loss from the visible
through the near-IR. For example, the ultralow intrinsic loss of silica and the extremely
smooth surface of the silica toroidal micro-cavities have already led to an ultra-high
qualify factor in excess of 10
8
.
Integrated waveguides are one of the fundamental elements of photonic circuits. The
two key factors of a high performance waveguide are low propagation loss and small
bending radius. As mentioned, silica has very low intrinsic loss. As a result, the
propagation loss of a silica waveguide is limited by its extrinsic loss which could be
improved by careful design and fabrication and parasitic to the presence of higher order
modes. The bending radius is determined by the refractive index contrast between the
core and the cladding and the creation of additional modes. Typically the contrast in
silica-based waveguides which have been demonstrated is small (less than 1.5 percent).
This restricts the bending radius to several mm, and thus limits achieving smaller
footprints. By increasing the refractive index contrast, more compact devices are possible,
enabling denser photonic circuit using silica devices. However, the current methods of
increasing the index contrast involve complex and expensive fabrication processes, such
as depositing additional and often dissimilar materials. Therefore, achieving large index
2
contrast silica waveguides with simple fabrication methods is a laudable goal. However,
despite recent advances in the field of integrated silica photonics which leverage ultra-
low loss oxides, the majority of work to date is unable to direct and confine light in
arbitrary directions. In this thesis, a low loss silica waveguide with a bending radius of
less than 375 m is demonstrated. The waveguide is directly fabricated on a silicon
substrate with a simple two step photo-lithography and buffered HF etching. Both
straight and serpentine shape waveguides with various bending radius are fabricated to
characterize the propagation and bending loss. Using this suspended device, a silica 2× 2
optical splitter is also demonstrated. A CO
2
laser reflow process is employed to reduce
the transmission loss, particularly in the coupling region. The splitter is able to split the
input signal evenly with a flat splitting ratio over a wide wavelength range.
A silica toroidal resonator which is fabricated on a silicon chip exhibits an ultra-
high-Q factor which is in excess of 10
8
. The small mode volume and long photon lifetime
have made it extremely suitable for nonlinear optical studies such as cavity Quantum
Electrodynamics and low threshold, narrow linewidth lasers. Its high sensitivity and real
time sensing ability have made it a promising candidate for biosensing. Currently, optical
fiber tapers are used to couple light into the micro-toroid; so all of the applications are
performed in a laboratory. Therefore, to move a micro-toroid out of the lab and find
practical applications, it has to be a fully integrated system, which is a micro-toroid
integrated with an on chip waveguide on the same substrate. In this thesis, a novel
approach for fabricating fully-integrated silica ultra-high Q toroidal resonators integrated
with on-chip waveguides is demonstrated. To achieve this performance, several new
fabrication techniques which are based on a combination of top down and bottom up
3
fabrication methods are developed. Additional sensing experiments are carried out using
this system.
1.2 Chapter overview
Chapter 2 gives an introduction to optical waveguides and micro-cavities.
Characteristics of optical waveguides such as propagation loss and critical bending radius
and characteristics of optical micro-cavities such as optical field distribution, mode
volume and quality factor are discussed. The testing setup built for the characterization of
silica waveguides and integrated toroidal micro-cavities is explained. How finite element
methods (FEM) simulations are employed to simulate the optical modes of the
waveguides and resonators are discussed.
Chapter 3 details the demonstration of the trapezoidal silica low loss waveguides and
splitters on silicon substrate. The suspended trapezoidal silica waveguide is directly
fabricated on a silicon substrate with a simple two step photo-lithography and buffered
HF etching. FEM simulations show that the trapezoidal shape gives smaller mode volume
and less loss compared to circular and rectangular shaped waveguides. Finite-difference
time-domain (FDTD) predicts the critical bending radius of the structure. Both straight
waveguides and serpentine waveguides with a series of bending radius are fabricated to
characterize the propagation and bending loss. The fabrication and characterization of
splitters based on the trapezoidal waveguides are also detailed.
Chapter 4 investigated the reason for the quality factor disparity between silica
microspheres and microtoroids. The reason for this performance disparity is directly
related to type of silicon substrate used in the fabrication process. The theoretical Q of
4
planar toroidal silica resonant cavities is calculated and compared to experimental results
from a series of devices fabricated from oxide on doped silicon wafers. As predicted, the
Q depends on the substrate dopant concentration.
Chapter 5 presents how a microtoroid is integrated with a waveguide on the same
substrate to form a fully integrated system. The design concept and different fabrication
approaches are discussed in detail. The mode structure and quality factor are
characterized using a narrow linewidth tunable laser.
Chapter 6 presents some sensing experiments carried out using the fully integrated
microtoroid system. Specifically, a series of temperature sensing experiments are
performed to explore the stability of the resonance to changes in the environment. For
UV sensing, a clear UV response curve is observed. The resonance undergoes a big
wavelength shift once the UV is turned on and then a slower shift as the UV is left on.
Once the lamp is turned off, the resonance returns to the initial position.
Appendix A presents a project performed in collaboration with Dr. Hong-Seok Choi
which demonstrated hybrid devices composed of ultra-high-Q planar toroidal cavities
with ultra-thin polymer films fabricated on a silicon wafer. Incorporating active elements
into these ultra-high-Q cavities to create dynamic devices would extend their
applicability. Using FEM simulations, the optical field overlap between the cavity and the
polymer film is modeled and experimentally verified using two polymers: poly(methyl
methacrylate) and polystyrene. These hybrid devices have demonstrated material-limited
Q factors above 10
7
.
5
Chapter 2 Overview of optical waveguides and microtoroidal
resonators
2.1 Introduction to optical waveguides
An optical waveguide is a light conduit which is able to confine and transmit light. It
consists of a slab, strip or cylinder of dielectric material surrounded by another dielectric
material with lower refractive index. Light undergoes total internal reflection at the
boundaries of dielectrics, gets confined in and guided through the high refractive index
material. The high refractive index part is usually called the core and the surrounding
lower refractive part is usually referred to as the cladding. One of most commonly seen
waveguides is the optical fiber which consists of two concentric cylinders of low loss
dielectric materials such as glass with slightly different refractive indices.
Integrated waveguides are optical waveguides fabricated on a chip. Figure 2-2 shows
examples of a silicon waveguide used for modulator and a polymer waveguide fabricated
on a printed circuit board for chip to chip interconnects.[1, 2] The primary application of
integrated waveguides is in photonic integrated optics, and now is also a promising
platform in other fields such as biosensing. In integrated optics, integrated waveguides
serve as a transmission medium and route light signals between different elements of the
photonic circuits. There are a variety of materials available to fabricate integrated
waveguides such as semiconductor (GaAs, InP etc), electro-optic material (LiNbO
3
),
glass (SiO
2
, SiN
x
etc), silicon on insulator, metals and polymers. Each of these types of
waveguides has its own advantages and limitations regarding the function to be integrated.
For example, semiconductor waveguides allow the direct integration of semiconductor
lasers, modulators and detectors; silica waveguides is advantageous in passive photonic
6
circuits such as arrayed waveguide gratings (AWGs) due to comparably low loss and low
thermal sensitivity.
Figure 2-1(a) SEM image of a silicon strip waveguide (Green, W.M., et al.) and (b) demonstration of red light (633nm)
guided in a polymer optical waveguide fabricated on a printed circuit board (Leng, Y., et al.)
Depending on the waveguide material, integrated waveguides can be fabricated in
various geometries. Figure 2-2 shows examples of useful geometries for integrated
waveguides such as ridge, embedded and immersed. Different fabrication techniques are
used to fabricate these geometries. For example, an embedded strip lithium niobate
waveguide is fabricated by diffusing titanium into a lithium niobate substrate in the
region of the strip to raise its refractive index. GaAs strip waveguides are fabricated by
epitaxial growth of multiple layers on a base substrate. Glass waveguides are usually
made by depositing multiple layers of doped glass followed by reactive ion etching.
7
Figure 2-2 Illustrations of various types of waveguide geometries: (a) ridge (b) embedded; (c) immersed. Grey color
indicates materials with higher refractive index and where light propagates in.
2.2 Background of optical waveguides
2.2.1 Total internal reflection
At the boundary of two dielectric media with refractive index n
1
and n
2
, light gets
reflected and refracted. The angle of reflection
r
is same as the incidence angle
1
.
The
angle of refraction and incidence,
2
and
1
,
is governed by Snell’s law:
1 1 2 2
sin sin nn (2.1)
8
When n
1
is smaller than n
2
(n
1
< n
2,
Figure 2-3a), if the ray is incident from the
medium of smaller refractive index,
21
, the refracted ray bends away from the
boundary.
When n
1
is greater than n
2
(n
1
> n
2,
Figure 2-3b), if the ray is incident from the
medium of greater refractive index,
21
, the refracted ray bends toward the boundary.
In this case, as
1
increases,
2
reaches 90 first. The angle of
1
at which
2
reaches
90 is called the critical angle
c
.
1 2
1
sin
c
n
n
(2.2)
When
1 c
, refraction does not occur, and the incident ray is totally reflected (Figure
2-3c). This phenomenon is called total internal reflection (TIR).
Figure 2-3 (a) the ray is incident from the medium of smaller refractive index n
1
< n
2,
(b) the ray is incident from the
medium of greater refractive index (n
1
> n
2
). (c) For n
1
> n
2
, if the angle of incidence is greater than critical angle, it
will undergo total internal reflection.
9
2.2.2 Numerical aperture
One thing to note is that not all the light coupled into the waveguide satisfies the total
internal reflection condition inside the waveguide. As shown in Figure 2-4, some light is
unguided and vanishes. There is a maximum angle of incidence at which light is guided
and is confined. Let’s consider a dielectric waveguide which is composed of a thin layer
of high-index (core index n
1
) material sandwiched between lower index (cladding index
n
2
) material. As shown in Figure 2-4, a light ray enters the dielectric waveguide at an
angle of
0
such that
0 0 1 1
sin sin nn . Total internal reflection occurs if
1
21
sin ( / )
c
nn
, where
1
/2 . The maximum angle of incidence at which
the light can be guided satisfies
max
0 0 1
sin sin( / 2 )
c
nn , which is often referred to
as the numerical aperture of a waveguide:
max 2 2
0 0 1 1 2
sin sin( / 2 )
c
NA n n n n (2.3)
Where n
1
and n
2
are the refractive index of core and cladding respectively.
So the larger the core and cladding index difference, the larger the numerical
aperture. Larger numerical aperture usually leads to smaller coupling loss as more light is
guided.
10
Figure 2-4 A dielectric waveguide which is composed of a thin layer of high-index material (core index n
1
) sandwiched
between two layers of low index (cladding index n
2
) material.
2.2.3 Propagation loss
One of the key factors of a high performance waveguide is low propagation loss.
There are several major loss mechanisms such as coupling loss (power lost when couple
light into and out of a waveguide,
coupling
), material loss (absorption from the waveguide
material,
material
), scattering loss (mainly due to sidewall surface roughness,
scattering
)
and leakage into the substrate (
leakage
), if the optical field is not sufficiently guided. So
for a straight waveguide with a certain length, the total loss
total
is:
total coupling material scattering leakage
(2.4)
The last three terms are dependent on the length of the waveguide while the first
term is dependent on the two endfaces of the waveguide. Therefore, when measuring the
loss of a waveguide vs the length of the waveguide, the coupling loss is a constant offset
to the overall loss of the device. Quantitatively, the transmission loss of a waveguide in
dB with a certain length is defined as:
11
( ) 10log
out
in
P
Loss dB
P
(2.5)
Where
in
P is the input power and
out
P is the output power from the waveguide.
There are many ways of measuring the propagation loss, but one technique which
improves the accuracy of the measurement by removing the device to device variation
caused by fabrication variation is the cut-back method. This technique assumes constant
coupling loss, constant surface roughness and leakage. As shown in Figure 2-5, the
transmission loss (dB) from the waveguides over a series of different lengths is plotted
and then fitted to linear to get the propagation loss (dB/cm). The intercept is the coupling
loss.
Figure 2-5 Propagation loss measurement: the transmission loss (dB) from the waveguides over a series of different
lengths is plotted and then fitted to linear to get the propagation loss (dB/cm). The intercept is the coupling loss
12
2.2.4 Bending loss
In integrated optics, waveguides don’t always propagate in straight; they often need
to change directions to route signal arbitrarily which result in bending of waveguide.
When light rays reach a bend, some of the rays no longer satisfy the total internal
reflection condition and is lost (Figure 2-6). This contributes to the bending loss of a
waveguide. The estimated bending loss was found to be strongly dependent on the
refractive index difference between the core and cladding.[3] The bigger the difference,
the more confined the light is and the less loss after a bend.
Figure 2-6 When light rays reach a bend, rays with incidence angle smaller than the critical angle become unguided and
lost.
13
2.3 Introduction to whispering gallery mode optical resonators
Optical resonators also confine light by total internal reflection. But in contrast to
an optical waveguide in which the light only goes forward, in optical resonators, light is
bounced back and forth inside the resonator. At certain wavelengths, light waves
traveling back and forth have phase matching, then constructive interference will occur
and the intensity of the light waves will be added up. These wavelengths are called the
resonance wavelengths of the resonator.
Whispering gallery mode optical resonators are a type of resonator which typically
have a circular geometry. The name came from the whispering gallery in St. Paul’s
Cathedral in London where Lord Rayleigh studied the acoustic waves propagating along
the gallery (Figure 2-7a). When one person whispers along the periphery of the gallery,
the whisper can be heard at different location. Figure 2-7b is the calculation results of an
acoustic whispering-gallery mode calculated at a frequency of 69 Hz in a cylinder filled
with air with the same diameter as the whispering gallery in St Paul’s Cathedral. As
indicated from the location of the acoustic wave pattern, the acoustic wave travels along
the periphery of the gallery.
In whispering gallery mode optical resonators, light also travels along the
periphery of the resonator. Figure 2-8a is a microtoroid optical resonator alongside an
optical fiber taper. The optical fiber taper which carries light wave is used to inject light
into the resonator and excite the optical whispering gallery mode. One example of the
optical field of a whispering gallery mode distribution in this device is shown in Figure 2-
8b. [4]
14
Since the light wave propagate along the periphery of the resonator so that they
approximately traverse a distance of about πR in a round trip, where R is the resonator
diameter. If the wavelength of the light fits an integer number of times of the resonator
circumference, phase matching condition is satisfied, the resonator will be on resonance
at that wavelength (Figure 2-9a).
RN (2.6)
Where is the wavelength in the medium and N is an integer. Otherwise, the resonator
will be off resonance (Figure 2-9b).
Figure 2-7(a) Whispering gallery in St. Paul’s Cathedral, the acoustic wave (red arrow) travels along the gallery (b)
Snapshot of an acoustic whispering-gallery mode calculated at a frequency of 69 Hz in a cylinder filled with air with
the same diameter as the whispering gallery in St Paul’s Cathedral. Red and blue spots represent higher and lower
whisper intensity, respectively. As indicated from the location of the acoustic wave pattern, the acoustic wave
propagates along the periphery.
15
Figure 2-8 (a) A microtoroid resonator coupled with an optical fiber taper, the light (shown in red) travels along the
periphery of the resonator (b) optical field distribution of a whispering gallery mode inside the microtoroid cavity.
Figure 2-9 (a) the wavelength of the light fits an integer number of times of the resonator, the resonator is on resonance
at that wavelength (b) the resonator is off resonance
Whispering gallery mode optical resonators can be made from various materials
and have numerous geometries, but usually have circular features. Figure 2-10 listed
typical whispering gallery mode optical resonators invented up to date. One metric
which can be used to compare different types of cavities is the device quality factor or Q.
16
This parameter describes the photon lifetime within the cavity. A longer photon lifetime
results in a higher Q factor. Additional details about the limitations on Q factor are
discussed in the following section.
The silica microsphere which was invented in the 1970’s was the very first
whispering gallery mode optical resonators and possesses the highest quality factor which
is typically in the range of 10
9
to 10
10
.[5, 6] It has been intensively studied in various
fields such as lasing, sensing and fundamental physics. This device which is made from a
fiber tip is free standing and thus limits its further practical applications. In 2002, the
silica microtoroid was invented; it is fabricated on silicon chip with standard
semiconductor processing technique.[7] It has comparable quality factors which are on
the order of 10
8
. The invention of the silica microtoroid has advanced the research greatly
and moved applications onto a single chip which is one big step forward to practical
applications. In 2012, wedged disks with size in mm range were demonstrated with
slightly higher quality factor in the order of near 10
9
.[8] Other devices such as wedged
silica microdisks and silicon nitride flat microdisk are also fabricated on a chip but have
significantly lower quality factors and thus limits their applications.[9, 10] CaF
2
toroidal
resonators also have ultra high quality factors but they are significantly larger (mm-scales
vs m-scale in diameter).[11] Therefore, the silica microtoroid provides the most
promising platform for further practical applications.
17
Figure 2-10 Different types of whispering gallery mode optical resonators
One important thing to note is that while lower Q optical resonators such as the
silicon nitride microdisk have been integrated with waveguides, forming a single
platform, a fully integrated, higher Q device has not been demonstrated. This has limited
the application of these devices in the lab. In this thesis, a novel method to integrate a
waveguide to a silica microtoroid is presented.
2.4 Background of optical resonators
2.4.1 Quality factor
When the light signal coupled into the resonator is in resonance with a resonator
mode, intensity builds up in the resonator due to constructive interference. If the light
signal is turned off at some point, due to absorption, scattering and other loss mechanisms,
the signal inside the resonator will decay and after a period of time, it dies out. The time
18
took to decay to the 1/e of its intensity is called the photon lifetime ( ) of the resonator.
Quality factor is defined as frequency times the photon lifetime:
Q (2.7)
So quality factor is a measure of the resonator’s temporal confinement ability. The higher
the quality factor, the better the temporal confinement. An equivalent definition of the
quality factor is:
/ Q (2.8)
Where is the resonance wavelength and is the linewidth of the resonance
There’s a series of loss mechanisms which affect the photon lifetime or the
quality factor of the resonator. In dielectric cavities, these include material loss
(absorption from the resonator material,
mat
Q ), surface scattering loss (scattering from
surface roughness,
ss
Q ), radiation loss (loss from the curvature,
rad
Q ), contamination loss
(from contaminations of the device,
cont
Q ) and coupling loss (loss induced during
coupling into and out of the resonator,
coupl
Q ). Material loss, surface scattering loss,
radiation loss and contamination loss are intrinsic to the device while coupling loss is
extrinsic to the device as it is dependent on the coupling method being used.
1 1 1 1 1 1 1 1
int total mat ss rad cont coupl ext
Q Q Q Q Q Q Q Q
(2.9)
So in order to get a higher quality factor, it is essential to minimize each of these
types of losses. Contamination loss can be reduced by keeping the device clean.
Radiation loss can be reduced to nearly zero if the radius of curvature is big enough.
19
Coupling loss can be minimized by choosing a coupling method such as the optical fiber
taper which has nearly zero coupling loss. Therefore, the dominant loss mechanisms are
material loss and surface scattering loss. In the case of the silica microtoroid, the low
absorption coefficient of silica and ultra-smooth surface by CO
2
laser reflow have led to a
ultra-high quality factor.
2.4.2 Free spectral range
Free spectral range is the wavelength separation between two successive
resonances. One wavelength is on resonance, this satisfies:
1
nR N (2.10)
The wavelength which is next to it satisfies:
21
( 1) ( 1)( ) nR N N (2.11)
Where n is the refractive index of the resonator material, and N is an integer.
From these two equations, we can get:
2
1
1
2nR
(2.12)
In the case of a microtoroid resonator, the wavelength (around a few hundreds of
nanometers to a few microns) is much smaller than the diameter of a microtoroid (tens of
microns), so this equation is reduced to:
20
2
1
2nR
(2.13)
So, the free spectral range is inversely proportional to the diameter. Therefore, the
smaller the device, the larger the free spectral range is.
2.5 Testing procedure for waveguides and integrated microtoroid
2.5.1 Testing procedure for waveguides
In this thesis, an on chip waveguide is first demonstrated followed by the
demonstration of a splitter based on the waveguide. To characterize the on chip
waveguide and splitter, a lensed fiber with a spot size of around 2 m is used to couple
light into the waveguide (Figure 2-11a). Figure 2-11b shows the output from the splitter
with visible wavelength at 653nm.
Figure 2-12 shows the components of the entire testing setup. The light signal
from the laser source is transmitted through the lensed fiber. The alignment process
between the lensed fiber and waveguide is monitored using top and side view machine
vision systems. For precise alignment, a nanometer resolution motorized stage and a
visible, fixed wavelength laser at 633nm are used to align the lensed fiber with the input
ends. The output light is directly focused into an optical beam profiler which is part of an
M2 Beam Quality Analysis system (Thor Labs), instead of using an additional lensed
fiber. This approach results in lower coupling loss and fewer alignment errors, enabling a
more accurate measurement of the behavior of the device. The beam profiler can get both
a power reading and the mode profile of the output.
21
Figure 2-11 Microscopic image showing (a)Microscopic image showing a lensed fiber coupled to a waveguide (b) two
output of a splitter, an aspheric lens is to focus the output into a beam profiler instead of a lensed fiber
Figure 2-12 All the components of testing setup for waveguide and splitter
2.5.2 Testing procedure for standing alone microtoroid
The testing setup for testing a standing alone microtoroid is shown in Figure 2-
13a. Because there isn’t an on chip waveguide alongside it, an optical fiber taper is
placed on a holder and used to couple light into the microtoroid. The tapered fibers are
fabricated from single mode optical fiber using a hydrogen torch. The sample sitting on
22
the sample holder is brought close to the optical fiber taper with a motorized stage. The
microscope is used to visualize the alignment. Figure 2-13b shows the microscope image
of a microtoroid aligned with a fiber taper.
Figure 2-13 (a) Testing setup for standing alone microtoroid (b) microscopic image of a microtoroid coupled to an
optical fiber taper
Figure 2-14 shows the components of the testing setup. A single mode tapered
optical fiber is used to couple light from a single mode, tunable narrow linewidth
(300kHz) CW laser into the microtoroid. To align the taper to the microtoroid, a nm-
resolution motorized stage is used (Optosigma). The output signal is detected with a high
speed photodetector and then digitized by a high-speed NI digitizer.
23
Figure 2-14 Components of the testing setup for a standing alone microtoroid
There are two commonly used scanning methods in characterizing a microtoroid.
One of them is broad scan, when the laser scans across a wavelength range which is a few
times of the free spectral range of the device with relatively faster scanning speed. This
way, we can easily tell the free spectral range of the device and the position of the
resonant wavelengths.
The other scanning method is called fine scan. From broad scan, the approximate
resonant wavelengths are determined. In fine scan, a triangle wave function with 100Hz
frequency is sent from the function generator to the laser to tune the laser with very small
wavelength steps. This enables scanning around the approximate wavelengths to get the
exact resonant wavelength and the full width half maximum (FWHM) of the resonance
by Lorentzian fitting. The quality factor can then be calculated using / Q .
24
2.5.3 Testing procedure for integrated microtoroid-waveguide system
In this thesis, a fully integrated microtoroid-waveguide system is fabricated on the
same silicon substrate which eliminated the need for an external waveguide such as the
optical fiber taper mentioned above. The testing setup of the integrated microtoroid-
waveguide system will be different from the one used for testing standing alone
microtoroid.
To characterize the microtoroid-waveguide system, a lensed fiber is used to shine
light into the waveguide and another lensed fiber is used to couple light out of the system
(Figure 2-15). The output from the lensed fiber is fed into a high speed photo-detector.
Lensed fiber has to be used for the output signal as the photo-detector has a fiber coupled
input.
Figure 2-15 Microscopic image showing (a) lensed coupled to the input of the waveguide (b) the microtoroid-
waveguide system (c) lensed coupled to the output of the waveguide
The procedure to get free spectral range and quality factor for the integrated
microtoroid-waveguide system are same as the standing alone microtoroid. Figure 2-16
shows the components of the testing setup for integrated microtoroid-waveguide system.
25
Figure 2-16 Components of the testing setup for the integrated microtoroid-waveguide system
2.6 Finite element method simulations
In this thesis, finite element method (FEM) simulations are used to simulate the
optical mode profile of the waveguide and microresonator. Many physical phenomena in
engineering and science can be described in terms of partial di fferential equations (PDE).
However, solving the resulting mathematical models is often impossible, especially when
the resulting models are nonlinear partial differential equations. Only very simple
problems of regular geometry such as a rectangular of a circle with the simplest boundary
conditions were tractable. FEM is a numerical approach by which these PDE can be
solved approximately. First, complex PDE equations over a large domain which is
impossible to solve, is divided to many simple element equations over many small
subdomains, named finite elements, then all sets of element equations are systematically
recombined into a global system of equations for the final calculation.
26
Figure 2-17 is the meshing created to find a solution to the electromagnetic
problem using FEM software. The big domains are divided into many little triangles
(subdomains) to get an approximate solution. The triangles are much smaller in the
circular region as that is the domain where the majority of the electromagnetic field is
located.
Figure 2-17 Meshing created to find a solution to a electromagnetic problem in FEM software
2.6.1 Optical modes
The concept of modes, or eigenfunctions, is fundamental for all wave phenomena
in physics like optics, acoustics and quantum mechanics. In optics, the modes are
solutions for the propagation of the light. As mentioned in section 2.2 and 2.3, only light
rays with certain angles can be confined in the waveguide and only light with certain
wavelengths can have resonance in the resonator. These confined light rays can have
27
different optical field distribution profiles which are different optical modes of the
microcavity. The mode which is dealt with most is the fundamental mode. The
fundamental mode is the first solution of the optical system with the simplest mode
profile.
In many cases, the polarization of the mode also matters. In this thesis, we
consider the two linearly polarized cases: TE (Transverse Electric) and TM (Transverse
Magnetic) polarization. TE polarization means that there is no electric field in the
direction of propagation and all the electric field is perpendicular to the direction of
propagation. TM polarization means that there is no magnetic field in the direction of
propagation and all the magnetic field is perpendicular to the direction of propagation.
2.6.2 Waveguide simulations
For a silica trapezoidal shape waveguide which is developed in this thesis, FEM
simulations are used to simulate the optical mode profile, effective refractive index, mode
volume, and polarization dependence. Figure 2-18 and figure 2-19 show the optical field
distribution of the TE and TM polarized fundamental modes at 1550nm respectively. The
color indicates the intensity and the arrows indicate the direction of electric field and
magnetic field. The optical field distribution is very simple and has only one maximum
for fundamental modes.
28
Figure 2-18 Electric field distribution of the TE fundamental mode of a trapezoidal shape silica waveguide at 1550nm
29
Figure 2-19 Magnetic field distribution of the TE fundamental mode of a trapezoidal shape silica waveguide at 1550nm
2.6.3 Microtoroid simulations
The microtoroid simulations are based on Mark Oxborrow’s models with slight
modifications. Figure 2-20 is the electric field distribution of the TE fundamental mode
of a microtoroid with major (D) and minor (d) diameter for 37.5 m and 5 m
respectively. One of the fundamental resonant frequencies is 2.044e14
Hz which equals
1466.106 nm. The optical field is mostly confined close to the interface of silica and air.
30
Figure 2-20 Transverse component of electric field squared intensity distribution of a TE fundamental mode of a
microtoroid with major and minor diameter for 37.5 m and 5 m respectively (a) Surface plot of the mode (b) electric
field plot along y=0 axis
31
Chapter 2 References
1. Leng, Y., et al., Dispensed polymer waveguides and laser-fabricated couplers for optical
interconnects on printed circuit boards. Appl. Opt., 2007. 46(4): p. 602-610.
2. Green, W.M., et al., Ultra-compact, low RF power, 10 Gb/s siliconMach-Zehnder
modulator. Opt. Express, 2007. 15(25): p. 17106-17113.
3. Marcatili, E.A.J., Bends in optical dielectric guides. The Bell System Technical Journal,
1969. 48(7): p. 2103-2132.
4. Zhu, J.G., et al., On-chip single nanoparticle detection and sizing by mode splitting in an
ultrahigh-Q microresonator. Nature Photonics, 2010. 4(1): p. 46-49.
5. Vernooy, D.W., et al., High-Q measurements of fused-silica microspheres in the near
infrared. Opt. Lett., 1998. 23(4): p. 247-249.
6. Gorodetsky, M.L., A.D. Pryamikov, and V.S. Ilchenko, Rayleigh scattering in high-Q
microspheres. J. Opt. Soc. Am. B, 2000. 17(6): p. 1051-1057.
7. Armani, D.K., et al., Ultra-high-Q toroid microcavity on a chip. Nature, 2003. 421(6926): p.
925-928.
8. Hansuek Lee, T.C., Jiang Li, Ki Youl Yang, Seokmin Jeon, Oskar Painter & Kerry J.
Vahala, Chemically etched ultrahigh-Q wedge-resonator on a silicon chip. Nature Photonics,
2012. 6(6): p. 5.
9. Shah Hosseini, E., et al., High quality planar silicon nitride microdisk resonators for
integrated photonics in the visible wavelength range. Optics Express, 2009. 17: p. 14543-
14551.
10. Gondarenko, A., J.S. Levy, and M. Lipson, High confinement micron-scale silicon nitride
high Q ring resonator. Optics Express, 2009. 17: p. 11366-11370.
11. Grudinin, I.S., N. Yu, and L. Maleki, Generation of optical frequency combs with a CaF2
resonator. Opt. Lett., 2009. 34(7): p. 878-880.
32
Chapter 3 Suspended waveguides and splitters
3.1 Introduction
Integrated waveguides are one of the fundamental elements of photonic circuits.[1,
2] The two key factors of a high performance waveguide are low propagation loss and
small bending radius. Previous research has demonstrated waveguides integrated on
silicon substrates based on various materials such as silicon, silicon nitride, polymers and
silica.[3-10] Of these, silica is an ideal material for most photonic applications, as it
inherently has ultralow loss from the visible through the near-IR. In comparison with
semiconductor materials, it has extremely low nonlinear behavior, such as the Kerr Effect,
and thus enables high power transfer. Typically, the refractive index contrast between
the core and the cladding in silica waveguides is small (less than 1.5 percent). This
restricts the bending radius to several mm, and thus limits achieving smaller footprints.
By increasing the refractive index contrast, more compact devices are possible, enabling
denser photonic circuit using silica devices.[6] However, the current methods of
increasing the index contrast involve complex and expensive fabrication processes, such
as depositing additional and often dissimilar materials.[4-6] Therefore, achieving large
index contrast silica waveguides with simple fabrication methods is a laudable goal.
However, despite recent advances in the field of integrated silica photonics which
leverage ultra-low loss oxides, the majority of work to date is unable to direct and confine
light in arbitrary directions.[11]
In this work, a suspended trapezoidal silica waveguide is directly fabricated on a
silicon substrate with a simple two step photo-lithography and buffered HF etching. Both
33
straight waveguides and serpentine waveguides were fabricated and characterized. A
beam splitter based on this waveguide is also demonstrated with 50/50 splitting ratio.
3.2 Trapezoidal waveguide
3.2.1 Fabrication of trapezoidal waveguide
To thoroughly characterize the behavior of the trapezoidal waveguide, both straight
and serpentine devices were fabricated. The fabrication and characterization of the
serpentine devices which contain both inside and outside bends is critical in thoroughly
understanding the behavior of this new waveguide geometry. Previous research with
integrated silica waveguides has focused on devices which only have outside bends,
mimicking resonant cavities.[11] However, optical circuit design cannot be limited to
structures which only curve in one direction. Therefore, to fully explore the potential of
silica waveguides for these applications, it is necessary to understand and be able to
predict its behavior.
The trapezoidal silica waveguides are fabricated from 2 m thick thermal oxide (wet)
on a 300 m silicon wafer (Montco Silicon). The fabrication process for the trapezoidal
silica waveguide is shown in Figure 3-1(a-c). First, 80 m wide silica rectangles are
defined using a combination of photolithography and buffered oxide etching using
buffered hydrofluoric (HF) acid (Figure 3-1a). It is important to note that the trapezoidal
shape (in the z-axis direction) arises as a result of the inherent isotropic nature of the
buffered HF etching process. It is possible to control the various angles in the trapezoid
34
by balancing the exposure time to hexamethyldisilazane (HMDS), which occurs before
the deposition of the photoresist, and the concentration of the buffer used to create the
buffered HF.
Next, a second photolithography and buffered HF etching step are used to carve out
two 8 m wide trapezoids at each side of the 80 m trapezoids (Figure 3-1b). The
photolithography and buffered HF etching process are optimized to create trapezoids with
45 degree sidewall. The trapezoidal waveguides are then diced using a diamond scribe to
create the input and output ends. In order to isolate the lower refractive index silica
waveguide from the higher refractive index silicon substrate, XeF
2
etching was used to
undercut the silica until the silicon pillar is approximately 10 m away from the
waveguides. A schematic of the cross section is shown in Figure 3-1d. While w, W and h
were held constant at 64 m, 80 m and 2 m respectively, the membrane thickness varied
between 900 nm and 1 m.
Figure 3-1 Fabrication process of the trapezoidal waveguide (a) 80 m wide silica trapezoids are defined using
photolithography and buffered HF etching. (b) A second photolithography and buffered HF etching are used to create
two 8 m trapezoids at each side of the 80- m trapezoids, and (c) XeF
2
etching is used to isotropically undercut the
silica. (d) Cross section schematic of the waveguide indicating the critical dimensions. In the present experiments,
W=80 m, w=64 m and h=2 m.
Given the dimensions defined in Figure 3-1d, this device supports higher-order
modes and operates in a multi-modal fashion. However, by changing the initial masks or
35
adding a cladding layer, it should be possible to optimize this performance, potentially
achieving single mode devices. Additionally, although the present work is focusing on
fabricating pairs of waveguides separated by the pillar, by changing the initial photomask,
it will be possible to fabricate multiple adjacent waveguides or single waveguides.
Figure 3-2 is the SEM images of the trapezoidal waveguide. Figure 3-2a is the
zoomed in image of one trapezoidal waveguide. It is important to note that the ends are
quite smooth, which is verified by measuring the coupling loss. Figure 3-2b is the
sideview/ends of the trapezoidal waveguide which shows a pair of trapezoidal
waveguides supported by the silicon pillar. Figure 3-2c is the top view of the complete
trapezoidal structure.
There are two important things to note here. First, the thickness of the membrane
which connects the two trapezoidal waveguide should be around 900nm to 1 m. If it is
too thin, then the membrane is easy to break and also will make the straight waveguides
wavy (Figure 3-3a, b). If it is too thick, there will be possible leakage to the membrane.
The amount of undercut is also important; a good value is around 10 m. If it is too big,
then it will also cause the waveguides to be wavy (Figure 3-3c), but if it is too close, then
there will also be possible leakage.
Serpentine waveguides are also fabricated using this process. An S-curve geometry
(Figure 3-4) is chosen in order to study the effect of both outside and inside bending on
the optical field. The inside bending radius (R) is varied from 125 to 400µ m at 25µ m
increments, yielding twelve different curved devices. The length of the waveguide arms
is held constant to allow straightforward calculation of the bending loss.
36
Figure 3-2 Scanning electron micrograph (SEM) images of the trapezoidal waveguide (a) SEM side view image of a
single waveguide SEM, (b) SEM side view image of a pair of waveguides (c) top view image of a pair of waveguides.
37
Figure 3-3 Scanning electron micrograph (SEM) images of the trapezoidal waveguides (a) when the membrane
thickness is much less than 900nm (b) zoomed out view (c) when the amount of undercut is much more than 10 m
38
Figure 3-4 Scanning electron micrograph (SEM) images of the curved trapezoidal waveguide (a) The S-curve
waveguide geometry depicting the bending radius (R), which was varied between 125 and 400µ m. (b) SEM of the
bending section and (c) cleaved end of a curved waveguide device.
39
3.2.2 FEM modeling and finite differential time domain simulations
As shown in Figure 3-2 and Figure 3-4, the trapezoidal waveguide is suspended
off of the silicon substrate by a silica membrane. One concern is that the optical field
might leak into the membrane, thereby increasing the overall loss of the device or
distorting the mode profile. Therefore, the optical field profile of the fundamental mode
of the device was modeled at 658, 980 and 1550nm, using FEM analysis. For comparison,
we also modeled silica rectangular and circular waveguides with the same width at the
same wavelength. The results at 1550nm for the TE mode are plotted in Figure 3-5 (a-c)
and used to form Figure 3-5d. The electric field distribution in all three devices is
symmetric, indicating that there is minimal leakage into the membrane. However, the
effective mode area of the trapezoidal waveguide is significantly smaller than the other
geometries.
Using these FEM results, the effective refractive indices for the TE and TM
polarization states were calculated and they are slightly different for the wavelengths
studied, indicating that the device should have polarization dependent operation
observable over long propagation distances. Additionally, based on these results, the
effective refractive index contrast ((n
2
core
-n
2
clad
)/(2n
2
core
)) of the device is 25.6%, 25.3%
and 24.5% at 658, 980, and 1550nm, for the TE modes [12]. The behavior for the TM
mode is similar.
40
Figure 3-5 Finite element method simulation of the electric field distribution (TE mode) of (a) trapezoidal, (b)
rectangular waveguide and (c) circular waveguide at 1550nm. (d) Comparison of the intensity of the mode profile along
the x (horizontal) direction, normalized to highest intensity in trapezoidal device.
Wavelength (nm) TE TM
658 1.4313 1.4305
980 1.4222 1.4199
1550 1.4011 1.3933
Table 3-1 Comparison of the effective refractive indices of TE and TM modes
One of the present hurdles in developing compact integrated circuits is bending
or radiation loss, which results from mode radiation, reflection resulting from phase
mismatch, and increased interaction with sidewall roughness. In the present devices,
optical field leakage into the supporting membrane also increases radiation loss, as the
effective refractive index contrast is slightly asymmetric. One approximation for the
bending or radiation loss (
b
) is:
1/2
2
32
3
2
exp
3
Bg
R
V R
(3.1)
41
Where R is the bend radius,
g
is the propagation constant of the waveguide, V is
proportional to the effective refractive index contrast between the core and the cladding.
Lastly, is proportional to the refractive index of the waveguide, the free space
propagation constant and
g
, and is proportional to the refractive index of the cladding
layer, the free space propagation constant and
g
One approach for solving this equation
is simulating the optical field profile throughout the bend. However, because of the
complex three-dimensional profile of the device, it is necessary to use finite difference
time domain (FDTD) simulations.
Using Lumerical FDTD, curved devices having similar geometrical values to
those fabricated were simulated. Specifically, inner bend radii of 50µ m to 175µ m, at
25µ m increments, as well as one device with an inner bend radius of 250µ m operating at
1550nm were modeled. The light travelled first through a 90-degree inside bend followed
by a 90-degree outside bend. In order to accurately capture behavior of the waveguides
with a 1550nm light signal, we did not use the built in SiO
2
material to draw the structure,
but instead used the user-defined dielectric material with a refractive index set to 1.444,
which is the refractive index of fused silica at 1550nm. [13] Using a mode source object,
we injected the fundamental mode of the straight waveguide into the waveguide arm that
goes through an inner bend first. We used the mode of the straight waveguide in order to
capture the mode-mismatch loss that will be present in the experimental results, as we
coupled light into a straight portion of the waveguide first during experiments.
In order to determine the loss in the simulations, we used two 2D Y-normal
frequency-domain power monitors, one close to the source and one at the very end of the
waveguide. Additionally, we used one 2D Z-normal monitor to capture the field profile in
42
the overview of the entire structure. Both of the 2D Y-normal monitors were placed so
that they contained just the waveguide arm guiding the light signal while including as
little of the silica membrane as possible. The monitor near the source was used to make
sure that the source power was completely coupled into the waveguide arm, and this
monitor always had an output for y-transmission very close to 1 since the source power is
always normalized to 1 in Lumerical. The monitor at the output was used to measure the
power going out of the far waveguide arm, P
out
which was used to easily calculate the
loss in dB. Finally, the loss was divided by the arc-length of the waveguide arms through
which the light travelled in order to normalize the loss at each bending radius to dB/cm.
Because of the way the loss was measured and calculated, the simulation loss results are
directly comparable to the experimental results.
Unfortunately, we were unable to include the silicon pillar in our simulations.
From the simulations, as light travels through an inside bend of a curved waveguide, a
significant amount of the optical field leaks into the membrane (Figure 3-6). When this
occurs, a portion of this field may be lost into the silicon pillar, which has a higher
refractive index than the silica membrane and waveguides.
43
Figure 3-6 Finite-difference time-domain simulation of the electric field intensity bent trapezoidal waveguide with an
inner radius of 75 m at 1550nm. The optical field clearly leaks into the silica membrane in-between the two
waveguide arms. This leakage is a significant source of loss in the serpentine devices.
3.2.3 Experimental setup
The experimental setup was briefly described in section 2.5.1, and here details
any necessary information. A single mode lensed fiber (Oz Optics) with a spot size of
around 2 m is used to couple light into the waveguide. A series of fixed wavelength
diode lasers at 658, 980, and 1550nm were used for the initial optical loss measurements
performed at moderate input powers. To accurately determine the loss, the cut-back
method was used [14]. Both straight and serpentine waveguides were tested using this
setup.
A complementary high input power measurement was also performed on straight
waveguides. For this experiment, a 1550nm laser is connected to an erbium doped fiber
amplifier (EDFA). The output of the EDFA is connected to a 99:1 optical coupler. The 1%
output is monitored by a power meter and the 99% output is connected to the lensed fiber
44
and coupled into the waveguide. The power coming out from the waveguide is focused
into the beam profiler. A free space attenuator is inserted between the focal lens and
beam profiler to avoid saturating the beam profiler.
Finally, at 1550nm, the polarization dependent loss is determined by varying the
polarization of input light by placing an inline polarization controller between the laser
and the lensed fiber. Specifically, the output power from the waveguides and the input
power from the lensed fiber at several polarization states are recorded.
3.2.4 Experimental results and discussion
3.2.4.1 Propagation loss measurement
The propagation loss is measured using the cut-back method which is described
in section 2.2.3. Figure 3-7 is the loss measurement results. The propagation loss is 0.69,
0.59 and 0.41dB/cm at 658, 980 and 1550nm respectively as determined from the slope
of the linear fit. The coupling loss is also determined by the intercept which is [2.3, 1.6,
1.3] dB at [658, 980, 1550] nm. As mentioned earlier, this waveguide supports higher
order modes, so the loss measurement is not for fundamental mode only. It is important
to note that these are straight waveguides which were several mm long, not devices in
ring geometries. Therefore, it is anticipated that the bandwidth would be similar to that
of optical fiber. Additionally, the results from multiple devices are shown, verifying the
reproducibility of the measurements.
45
Figure 3-7 Measured propagation loss of the trapezoidal silica waveguide at three wavelengths: 658, 980 and 1550nm
As proposed by Tien[15], the propagation loss is defined as:
2 2 2
2 0
2
s
k h E
n
E dx
(3.2)
where is the core/cladding interface roughness, k
0
is the free space wavenumber, is
the modal propagation constant, n is the refractive index difference between the core
and the cladding, h is the transverse propagation constant in the core, and E
s
is the
scattered electric field amplitude due to surface roughness. As shown in equation 3.2, the
propagation loss is proportional to the scattered electric field amplitude E
s
. There are two
dominant loss regimes: 1) reduction in optical mode confinement, and 2) surface
scattering due to roughness. At long wavelengths, when the optical mode approaches the
size of the device, the first regime is the dominant loss mechanism of the device. At
shorter wavelengths, interface roughness is the primary loss mechanism. As can be
46
clearly observed from the simulation results, the optical mode is completely confined
within the device. Additionally, the loss increases as the wavelength decreases.
Therefore, the increase in propagation loss is the result of an increase in the scattered
electric field amplitude due to interface roughness ( ). [3]
Figure 3-8 shows the results from the polarization measurements. The
transmission at different polarization states varied slightly, which indicates that the
waveguide has a small polarization dependence over long propagation lengths. However,
it is important to note that the multi-modal behavior of the waveguide also means that
these devices are not polarization-maintaining. Due to this behavior, the polarization
dependence of the devices is not useful, as small changes in things like input coupling
conditions can have an effect on the polarization state of the light within the waveguide.
Although the simulation results for the effective refractive indices for the TE and TM
polarizations indicate polarization-dependent behavior (Table 3-1), experimentally we
see that due to the multimodal behavior of the waveguides, the polarization dependence is
somewhat unpredictable. In Figure 3-8, the transmission for various polarization states
fluctuates somewhat randomly, often overlapping, indicating that this unpredictable
behavior does not have a large effect for standard operation of these waveguides. By
modifying the device dimensions but keeping the overall geometry similar, it might be
possible to achieve single mode operation and make the polarization dependence more
useful.
47
Figure 3-8 Transmission at four different polarization states
3.2.4.2 Bending loss measurement
The bending loss of the curved waveguide was calculated using the following
relationship:
total length coupling bend
(3.3)
where α
total
is the total measured loss of the devices, α
length
is the loss due to waveguide
length, α
coupling
is a constant that consists primarily of coupling loss, and α
bend
is the
bending loss. The total loss is found by measuring the difference between the input and
output powers, as detailed previously. Losses due to length (
length
) are found by
measuring the length of the curved device (R) and measuring the length of the arms using
an optical microscope. This total length is multiplied by the propagation losses of 0.69,
0.59 and 0.41dB/cm at 658, 980 and 1550nm, which are found above. To account for
length differences, the loss was normalized by dividing the measured loss in dB by the
48
arc-length of the waveguide bends. The length loss and the system loss are then
subtracted from the total loss to calculate the bending loss.
Figure 3-9 shows the bending loss as a function of bending radius. Each point
represents a unique device (N>3 for each R). For all three wavelengths, the loss is fit to
an exponential curve. From Figure 3-9, one can see that the exponential fits for all three
wavelengths are very similar, with the shorter wavelengths exhibiting more loss. This
occurs because at shorter wavelengths, the light more easily leaks into the silica
membrane and is therefore more easily lost into the silicon pillar. If the light signal were
to only travel through outside bends, we expect that this trend might be reversed. The
critical bending radius has a slight wavelength dependence, and is below 375 m for all
wavelengths.
Figure 3-9 Measured bending loss of the trapezoidal silica waveguide at 658, 980 and 1550nm and the FDTD results
49
For direct comparison, the simulation results are also plotted in Figure 3-9. While
the general shape of the results is in excellent agreement with the experimental values,
the precise bending loss values are systematically lower. As mentioned previously, the
simulations did not include the silicon pillar. This difference provides further support for
the hypothesis that the leakage into the membrane and the subsequent loss into the silicon
pillar accounts for a significant portion of our bending losses. Therefore, if the light
signal only travelled through outside bends, the bending loss would be much lower and
the critical radii would be significantly reduced, due to the high index contrast of the
devices, and our simulations support this theory.
3.2.4.3 Power dependence measurement
Silica is known to have very low power-dependent, non-linear coefficients [16]
from the visible through the near-IR and thus the trapezoidal waveguide should have a
linear power response at very high input powers. As shown in Figure 3-10, the output
power of the trapezoidal waveguide changes linearly with the input power up to 200mW
which was the maximum power achievable by the laser input system. This linear
behavior is quite an advantage compared with silicon or polymer waveguides which have
large power-dependent, non-linear coefficients. The fabricated silica waveguide thus
shows great potential for high power applications.
50
Figure 3-10 The output power of the trapezoidal waveguide shows a linear dependence on the input power up to
200mW.
3.3 Silica Beam Splitter
3.3.1 Fabrication of the splitter
The suspended 2 ×2 silica splitters are fabricated from 2 m thick thermal oxide
(wet) on a 300 m silicon wafer (Montco Silicon). The fabrication process for the silica
splitter is shown in Figure 3-11. Figure 3-11 (a-c) shows the entire device and Figure 3-
10 (d-f) shows the cross section of the device during fabrication process.
A pair of photolithography steps is performed to get the two silica waveguides
(Figure 3-11a, d). First, photolithography and buffered HF etching are used to form the
80 m wide silica stripes on a silicon wafer. Subsequently, a second photolithography
and buffered HF etching is performed to create the two 14 m wide silica waveguide
channels and also thin down the membrane between the waveguides to less than 1 m
51
except in the tapered or splitting region of the device (Figure 3-11a, d). By thinning the
membrane, any leakage of optical power from the waveguide to the support pillar was
reduced.
To optically isolate the silica splitter from the higher refractive index silicon
wafer, XeF
2
is used to isotropically etch the silicon substrate until the silica waveguides
are 3 to 4 m away from the supporting silicon pillar. At the same time, in the tapered
region, only a few microns (3 to 4 m) thick silicon pillar is left (Figure 3-11b, e).
Then, the silica waveguides are reflowed using a CO
2
laser to create the smooth,
circular (diameter≈5 m) waveguide channels (Figure 3-11c, f). The splitter is then diced
using a dicing saw to form the input and output ends. A final XeF
2
etching step is used to
remove the silicon pillar in the tapered region.
Figure 3-11 Overview of the fabrication process of the splitter (a, d) Rendering and cartoon of the dual
photolithography and buffered HF etching proceedure to achieve the rectangular shaped silica waveguides with a
taperd region, (b,e) Rendering and cartoon of the XeF
2
etching to isotropically undercut the silica structure, (c)
Rendering of the CO
2
laser reflow to form the circular silica waveguide channels shown in operation and (f) Cartoon of
CO2 laser reflow to form the circular silica waveguide channels.
52
Figure 3-12 and figure 3-13 is the microscopic image and SEM image of the
device respectively. The green color in the microscopic image indicates that the
membrane is less than 1 m. The suspended 2 ×2 silica splitter is composed of two
circular waveguides with a spacing of around 60 m in the input and output region,
gradually reducing to 20 m in the tapered region. The silicon pillar in the
tapered/coupling region is completely removed, which eliminates any leakage to the
silicon substrate during coupling. In the input and output ends of the splitter, the
waveguide channels are also elevated off the substrate, which also prevents leakage into
the substrate.
Figure 3-14 shows the input/output facets of the splitter. The circular feature of the
waveguide channels is clearly evident. The silica waveguides are air-clad; this creates an
index difference ((n
2
core
-n
2
clad
)/(2n
2
core
)) of around 25%.
Figure 3-12 Composite optical microscope image of the splitter
53
Figure 3-13 SEM image of the splitter which has two input and output silica circular waveguide channels with a
suspended, tapered region
Figure 3-14 SEM image of the end of the splitter showing two circular silica waveguides
54
Please note that the distance between the two waveguides in the tapered/coupling
region is greater than 20 m, this big distance indicates that this beam splitter works with
a totally different coupling mechanism than the traditional directional couplers. The
simulations about how this beam splitter works is investigated in reference [17]. Actually,
the initial fabrication goal is to fabricate the traditional directional couplers with a micron
or less coupling gap. A lot of effort was done trying to get the two waveguides reflowed
smoothly and straight with submicron coupling gap.
In order to form two waveguides in the coupling region, there has to be a very
thin silicon pillar (less than 1 m) underneath to separate the reflow. First, this thin pillar
is quite difficult to get as it’s very difficult to time the XeF
2
etching to get the right
thickness. Second, the silicon pillar has some thickness nonuniformity which will transfer
to the reflow of the waveguides as the waveguides are going to be really close (almost in
contact) to the pillar. Figure 3-15a shows the splitter before reflow with the silicon pillar
tapered to be very thin at the coupling region. Figure 3-15b shows the reflow result. As a
result of the thickness nonuniformity of the silicon pillar, the reflowed waveguides are
wavy instead of being straight, also the gap of the two waveguides are still bigger than
the desired a micron or so gap. This approach clearly doesn’t look promising to work, so
all of the following testing results are performed using the splitters fabricated with over
20 m gap in the coupling region.
55
Figure 3-15 Microscopic images showing fabrication results of directional couplers (a) the coupler before reflow with
very thin (less than 1 m) silicon pillar (b) reflow result which shows wavy waveguides as a result of the thickness
ununiformity
56
3.3.2 Experimental results and discussion
3.3.2.1 Wavelength dependence
The splitter’s characterization is carried out on the same testing setup for
waveguide. The transmission spectrum of the silica splitter is shown in Figure 3-16.
Figure 3-16a shows the splitting ratio dependence on wavelength: the through port and
the cross port. The splitting ratio was taken at 1nm intervals. Per convention, the splitting
ratio is defined by the division between the output of one channel and the total output
power: P
through
/(P
through
+P
cross
) or P
cross
/(P
through
+P
cross
). The splitting ratio is quite flat over
a wide wavelength range from 1520 to 1630nm.
Figure 3-16 Optical characterization of the 2x2 splitter The a) splitting ratio versus wavelength and b) excess loss versus wavelength
were measured from 1520 to 1630nm. Both metrics are quite smooth and flat over the wavelength range.
Interference effects of the fundamental mode with higher order modes or radiated modes
are commonly seen in beam splitters which result from distortions of the input field due to an
57
imperfect alignment or after a bent. These effects induce a slight oscillatory variation of the
splitting ratio when the wavelength is changed.[18-20] As shown in Fig. 3-14a, the splitting ratio
variation of the silica splitter is quite small. This indicates both a perfect alignment and also a very
small distortion of the field due to waveguide curvature. The silica splitter is air-clad which gives
a refractive index difference of about 0.4 between core and cladding; this large index contrast
explains the small distortion of the optical field after the curvature in the tapered region.[12, 21]
If the alignment is imperfect, an oscillatory variation of the splitting ratio when the
wavelength is changed will be observed as shown in Figure 3-17. The oscillation is due to
alignment error rather than the bent, as discussed above.
Figure 3-17 Wavelength dependence of splitting ratio showing oscillatory variations when the alignment is imperfect
The excess loss (EL) is defined as the total output power over the input power:
10log
through cross
input
PP
EL
P
(3.4)
58
The excess loss spectrum of the splitter is shown in Fig. 3-14b. Again, the excess loss is also
very constant across the wavelength range. The major contributing loss is the insertion loss from
the lensed fiber as well as the waveguide. In other experiments, the transmission loss of the
waveguide has been shown to be approximately 0.5dB/cm. There are many possible routes to
reducing the insertion loss. For example, polishing the waveguide input ends or using index
matching gels.
By scanning across the laser wavelength range, we can locate the wavelengths at which
the device splits the signal nearly evenly. Figure 3-18 shows an example 3D optical field profile
where the output intensity distribution of the splitter is close to the ideal 50:50 splitting ratio. The
low crosstalk and high uniformity of the optical field is clearly evident.
Figure 3-18 Output intensity profile from the 2x2 silica splitter showing a 50/50 splitting ratio with low crosstalk and high uniformity.
3.3.2.2 Power and polarization dependence
The splitter’s behavior as a function of input power is also studied (Figure 3-19).
Specifically, the splitting ratio does not change, despite an order of magnitude increase in input
59
power, and the output power increases linearly with the input power. This linear behavior is a
direct result of the extremely low non-linear coefficients present in silica. Additionally, it is very
different from behavior observed in other low-loss waveguiding splitter devices fabricated from
silicon and polymeric materials that have non-linear coefficients which are an order of magnitude
or more greater than that of silica[22, 23].
Figure 3-19 The splitting ratio (red circles, black diamonds) and output power (blue squares) versus input power. The
splitting ratio is constant over the entire range and the output power changes linearly with the input power. Both
indicate that the device’s behavior is not being affected by any non-linear effects over this range of input power
We also did a polarization measurement. Using a polarization controller, the
wavelength dependent splitting ratio was determined at several different input
polarizations. As shown in Figure 3-20, the splitting ratio remains nearly constant for
each polarization state at each wavelength. This indicates that the splitter is polarization
independent.
60
Figure 3-20 Polarization dependence of the splitter. The splitting ratio doesn’t change at each polarization state at each
wavelength
3.4 Conclusions
In conclusion, straight and serpentine trapezoidal silica waveguides on a silicon
substrate are fabricated, modeled and characterized. Finite element method and finite
difference time domain simulations accurately model the device behavior from the visible
through the near-IR and predict the observed slight polarization dependent behavior, as
well as excellent optical field confinement. The trapezoidal shape pinches the optical
mode, resulting in a reduction in the effective mode area. As a result of the high effective
refractive index contrast and the low optical loss of silica, the loss measured using the
cut-back method is 0.69, 0.59 and 0.41 dB/cm at 658, 980 and 1550nm, respectively, and
the critical bending radius was below 375 m for all wavelengths. Additionally, the
waveguide has a linear power response up to 200mW. This structure is very simple and
61
inexpensive to fabricate with the advantage of an extremely high core-clad effective
refractive index contrast (25%). This type of suspended trapezoidal waveguide geometry
will find applications in integrated photonics [1, 2], and bio/chemical sensing. [24-26]
A novel 2 ×2 suspended silica splitter on a silicon substrate is designed, fabricated and
demonstrated based on the trapezoidal waveguide. In addition to using silica as the waveguiding
material, two key innovative design features were included in this splitter. First, by leveraging a
residual membrane between the waveguides to enable efficient energy transfer between them, it
was possible to couple power over distances greater than 20microns. Second, to enhance
confinement elsewhere in the structure, it is suspended off of the silicon substrate, thereby
creating a very high refractive index contrast. The splitter was able to divide power evenly with
low crosstalk and high uniformity. The splitting ratio variation is quite small over the entire
wavelength range and the excess loss is also extremely flat over a wide wavelength range from
1520nm to 1630nm. Finally, the splitting ratio did not depend on the input power, and the output
power scaled linearly with input power, indicating that the device did not suffer from non-linear
behavior over the input powers studied. Initial polarization experiment indicates that the splitter
is polarization independent. Based on these performance characteristics, this type of integrated
silica splitter will find applications in optical communications[2, 6, 27] , temperature sensing [28]
and bio/chemical detection[29].
62
Chapter 3 References
1. Kopp, C., et al., Silicon Photonic Circuits: On-CMOS Integration, Fiber Optical Coupling, and
Packaging. Selected Topics in Quantum Electronics, IEEE Journal of, 2011. 17(3): p. 498-509.
2. Himeno, A., K. Kato, and T. Miya, Silica-based planar lightwave circuits. Selected Topics in
Quantum Electronics, IEEE Journal of, 1998. 4(6): p. 913-924.
3. Vlasov, Y. and S. McNab, Losses in single-mode silicon-on-insulator strip waveguides and bends.
Opt. Express, 2004. 12(8): p. 1622-1631.
4. Bauters, J.F., et al., Ultra-low-loss high-aspect-ratio Si3N4 waveguides. Opt. Express, 2011. 19(4): p.
3163-3174.
5. Sum, T.C., et al., Proton beam writing of low-loss polymer optical waveguides. Appl. Phy. Lett.,
2003. 83(9): p. 1707-1709.
6. Miya, T., Silica-based planar lightwave circuits: passive and thermally active devices. Selected
Topics in Quantum Electronics, IEEE Journal of, 2000. 6(1): p. 38-45.
7. Bona, G.L., R. Germann, and B.J. Offrein, SiON high-refractive-index waveguide and planar
lightwave circuits. IBM Journal of Research and Development, 2003. 47(2.3): p. 239-249.
8. Zhang, X. and A.M. Armani, Suspended bridge-like silica 2x2 beam splitter on silicon. Opt. Lett.,
2011. 36(15): p. 3012-3014.
9. Maker, A.J. and A.M. Armani, Low loss silica on silicon waveguides. Optics Letters, 2011. 36(19): p.
3729-3731.
10. Jared F. Bauters, M.J.R.H., Demis D. John, Jonathon S. Barton, Christiaan M. Bruinink, Arne Leinse,
René G. Heideman, Daniel J. Blumenthal, and John E. Bowers, Planar waveguides with less than 0.1
dB/m propagation loss fabricated with wafer bonding. Optics Express, 2011. 19(24): p. 24090-24101.
11. Hansuek Lee, T.C., Jiang Li, Oskar Painter, and Kerry J. Vahala, Ultra-low-loss optical delay line on
a silicon chip. Nature Communications, 2012. 3(867): p. 1.
12. Ladouceur, F. and E. Labeye, A new general approach to optical waveguide path design. Lightwave
Technology, Journal of, 1995. 13(3): p. 481-492.
13. Malitson, I.H., Interspecimen Comparison of the Refractive Index of Fused Silica. Journal of the
Optical Society of America, 1965. 55(10): p. 1205-1208.
14. Derickson, D., Fiber Optic Test and Measurement1997: Prentice Hall. 672.
15. Tien, P.K., Light Waves in Thin Films and Integrated Optics. Appl. Opt., 1971. 10(11): p. 2395-2413.
16. Boskovic, A., et al., Direct continuous-wave measurement of n
2
in various types of
telecommunication fiber at 1.55 . Opt. Lett., 1996. 21(24): p. 1966-1968.
17. Soltani, S. and A.M. Armani, Optimal design of suspended silica on-chip splitter. Opt. Express, 2013.
21(6): p. 7748-7757.
63
18. Eom, J.B., J.-H. Park, and B.H. Lee, 2? photonic crystal fiber splitter based on silica-based planar
lightwave circuits. Opt. Lett., 2009. 34(23): p. 3737-3739.
19. Deri, R.J. and R.J. Hawkins, Polarization, scattering, and coherent effects in semiconductor rib
waveguide bends. Opt. Lett., 1988. 13(10): p. 922-924.
20. Munowitz, M. and D.J. Vezzetti, Numerical modeling of coherent coupling and radiation fields in
planar Y-branch interferometers. Lightwave Technology, Journal of, 1992. 10(11): p. 1570-1574.
21. Gambling, W.A., H. Matsumura, and C.M. Ragdale, Field deformation in a curved single-mode fibre.
Electronics Letters, 1978. 14(5): p. 130-132.
22. Dinu, M., F. Quochi, and H. Garcia, Third-order nonlinearities in silicon at telecom wavelengths.
Applied Physics Letters, 2003. 82(18): p. 2954-2956.
23. Jakubczyk, Z., R. Tremblay, and R. Vallee, Nonlinear curved fiber coupler. Journal of Applied
Physics, 1989. 66(10): p. 5113-5115.
24. Schmitt, K., et al., Interferometric biosensor based on planar optical waveguide sensor chips for
label-free detection of surface bound bioreactions. Biosensors and Bioelectronics, 2007. 22(11): p.
2591-2597.
25. Hunt, H.K. and A.M. Armani, Label-Free Biological and Chemical Sensors. Nanoscale, 2010. 2(9):
p. 1544-1559.
26. Washburn, A.L. and R.C. Bailey, Photonics-on-a-Chip: Integrated Waveguides as Enabling
Detection Elements for Lab-on-a-Chip Biosensing Applications. Analyst, 2011. 136: p. 227-236.
27. Doerr, C.R. and K. Okamoto, Advances in Silica Planar Lightwave Circuits. Lightwave Technology,
Journal of, 2006. 24(12): p. 4763-4789.
28. Guo, H., et al., Temperature sensor using an optical fiber coupler with a thin film. Appl. Opt., 2008.
47(19): p. 3530-3534.
29. Luff, B.J., et al., Integrated-optical directional coupler biosensor. Opt. Lett., 1996. 21(8): p. 618-620.
64
Chapter 4 Ultimate quality factor of silica microtoroid
resonant cavities
4.1 Introduction
Silica microtoroidal optical resonant cavities which can be fabricated on chip are
attractive in many different applications such as bio/chem detection, telecommunications
and fundamental physics investigations.[1-5] One of the primary motivations for using
the microtoroid resonator is its ability to achieve ultra high-Q factors or Q>100 million in
a planar device. These ultra-high Q factors have been attributed to a combination of very
low material loss and very few surface imperfections. However, another ultra-high-Q
structure, the microsphere resonator, has similar features and has demonstrated
significantly higher Q values.[6, 7] In the present work, the reason for this apparent
disparity is explored and an additional underlying loss mechanism which governs the Q
of microtoroids is proposed.
4.2 Theoretical calculations
As discussed in section 2.4.1, the loaded or measured quality factor (Q) of a silica
ultra-high-Q resonant cavity can be described by two sets of losses: intrinsic and
extrinsic. The intrinsic losses include surface scattering (Q
ss
), material loss (Q
mat
),
contamination loss (Q
cont
) and radiation loss (Q
rad
), and the extrinsic losses include the
coupling losses (Q
coupl
). The classic functional forms for Q
ss
and Q
mat
are:[8]
65
3
2 2 2
3
18
ss
eff
KR
Q
K n B
(4.1)
2
eff
m
n
Q
(4.2)
where K defines the internal reflection condition, and B are the surface roughness of
the cavity, n
eff
is the effective refractive index, is the wavelength, is the material loss,
and R is the radius of the cavity. Q
rad
scales exponentially with diameter, and therefore,
by appropriate choice of cavity diameter, it has negligible impact on the Q of the system.
To minimize the extrinsic losses (Q
coupl
), a high efficiency or low loss coupling method
can be used such as optical tapered fibers.[9]
In initial research, it was assumed that the Q
mat
term, which only contained the
bulk absorption coefficient, accurately described both the bulk and the internal scattering
losses of the system. However, it was recently shown that an additional term which
explicitly described the internal scattering within the cavity (Q
is
) is necessary to capture
this behavior. The Q
is
was shown to have the functional form:[8]
3
2 2 7
3
4
is
T eff
K
Q
p T n
(4.3)
where p, T, and
T
are the Pockels coefficient, melting temperature, Boltzman constant,
and isothermic compressibility at this temperature. It is important to note that the
dependences on the refractive index and wavelength are very different in these three
expressions.
To calculate Q
is
, Q
mat
and Q
ss
, previously determined values for , B, n
eff
, p, T,
and were used, along with the experimental parameters which define K and .[6, 8]
Additionally, it was assumed that thermal oxides have material losses similar to silica.[6,
66
7, 10] The model was verified by comparing its calculated Q values with those found in
other publications for microspheres. [6-8] Based on these calculations, the dominant loss
mechanism in a silica microtoroid resonant cavity with a radius above 40 m is Q
mat
, and,
for the specific geometry of cavities used in the present research, this threshold is
determined to be near 10
10
in the visible. However, several of the previously stated
assumptions are incorrect in the case of microtoroids.
Unlike the microsphere, the microtoroid is fabricated from thermal oxide not silica
fiber.[11] Thermal oxide is grown by oxidizing a silicon wafer.[12] Previous research
has shown that dopants diffuse from the silicon into the oxide, at a rate which is
dependent on the growth conditions.[12, 13] These dopants can modify the refractive
index, which in turn will affect the Q of the cavity. In the case of boron, the refractive
index is decreased. Assuming that Q
-1
=Q
is
-1
+Q
ss
-1
+Q
mat
-1
, the dependence of the Q on the
refractive index and the resonant wavelength was calculated as shown in Figure 4-1. It is
clear from Figure 4-1 that small changes in the refractive index can significantly affect
the Q factor of the cavity.
67
Figure 4-1 The calculated dependence of the quality factor on the refractive index and the resonant wavelength. The Q
factor decreases as the refractive index decreases and the wavelength increases.
4.3 Experimental verification
4.3.1 Fabrication of microtoroids
To fabricate the devices, a combination of photolithography and buffered HF
etching are used to form circular silica pads on a silicon wafer (Figure 4-2a). Then, the
silicon substrate is isotropically etched with XeF
2
to form silica microdisks (Figure 4-2b).
Finally, the microdisk is reflowed using a CO
2
laser to create the microtoroid resonator
(Figure 4-2c).[11] Figure 4-3 shows the optical microscopic images of the device at each
of these three fabrication steps. The wedged shape in Figure 4-3a is a natural result of
isotropic buffered HF etching; the bright circle in Figure 4-3b which is in the center of
the microdisk is the silicon pillar underneath. Figure 4-4 is the side view scanning
electron microscope (SEM) image of a microtoroid.
68
Figure 4-2 Renderings of the fabrication process: (a) photolithography and buffered HF etching are used to form
circular silica pads on a silicon substrate, (b) XeF2 undercuts the oxide, forming silica microdisks sitting on silicon
pillars, (c) a CO
2
laser reflows the silica, forming silica microtoroids on silicon pillar
Figure 4-3 Microscopic images of the device at each step: (a) silica pad, the wedged shape is a result of isotropic
buffered HF etching, (b) silica microdisk, the bright circle in the middle is the silicon pillar underneath (c) silica
microtoroid
Figure 4-4 A scanning electron micrograph of the fabricated silica microtoroid resonator
69
In order to verify the effect of the dopant on the quality factor of the cavity, a
series of thermal oxide (wet) grown on a set of silicon wafers with increasing boron
doping concentrations (Montco Silicon) are used. The specific boron doping
concentrations are summarized in Table 4-1 (the first column; other columns contain data
which will be discussed in the following sections). Microtoroids are then fabricated on
these different wafers using the fabrication method mentioned above. To minimize any
fabrication errors, all of the devices are fabricated using the exact same fabrication recipe.
Boron
concentration
(cm
-3
)
Refractive index Quality factor (Q)
1
x 10
8
630nm 850nm 980nm 630nm 850nm 980nm
1.63E14
1.32E15
1.47E16
7.98E18
7.68E19
1.4657
1.4694
1.467
1.4695
1.3834
1.4608
1.4678
1.463
1.4648
1.3799
1.4582
1.449
1.4609
1.4626
1.3659
3.0
4.53
2.3
1.94
1.31
3.39
1.85
1.27
1.13
0.35
2.0
1.4
1.5
1.16
0.43
Table 4-1 The refractive indices of different thermal oxide with different doping concentrations in the substrate at three
wavelengths and the corresponding measured quality factors of the microtoroids fabricated on these different wafers.
1
The error in a single measurement of an individual resonant linewidth is ± 1x10
7
. The variation in Q
between a significant number of cavities is much larger, as a result of imperfections in fabrication.
Therefore, values reported in this table are the highest achieved with each type of cavity.
70
4.3.2 Refractive index measurements
The refractive indices of the thermal oxides are determined by variable-angle
spectroscopic ellipsometry using a J.A. Woollam Co. VASE
®
instrument and are reported
in Table 4-1. Data is measured from 500 to 1100nm at three angles of incidence (64, 69
and 74 ) near 75
o
to enhance sensitivity in the data analysis.[14] The refractive indices
are then determined by fitting the acquired ellipsometric parameters and to a Si-SiO
2
material system. The parameters for Si and SiO
2
are taken from the database provided by
J.A. Woollam. Figure 4-5 is an example fit for a wafer with a doping concentration of
1.63× 10
14
cm
-3
.
Figure 4-5 The acquired ellipsometric parameters and fit to a Si-SiO
2
material system. The solid (open) squares,
triangles and circles represent ( ) recorded at 64, 69 and 74 degrees. The solid lines are fit data.
71
4.3.3 Quality factor testing and analysis
The quality factor of the fabricated microtoroids with different doping
concentrations in the substrate is determined using the testing method descried in section
2.5.2. To thoroughly characterize the effect of the substrate dopant concentration on the
resulting thermal oxide, the quality factor is determined at three different wavelengths:
630, 850 and 980nm. In the measurement, a single mode tapered optical fiber is used to
couple light from a single mode, tunable narrow linewidth CW laser into the microtoroid.
To align the taper to the microtoroid, a nm-resolution motorized stage is used
(Optosigma). The quality factor of the microtoroid is calculated from the linewidth of the
resonance which was recorded by a high-speed NI digitizer.
In these experiments, it is particularly important to use very low loss tapered
optical fiber waveguides and to optimize the coupling conditions into the microtoroid in
order to minimize any coupling-associated losses.[9, 15] All of the quality factors are
taken in under-coupled regime. To eliminate any thermal broadening or other nonlinear
effects which can distort the resonant linewidth, an input power of less than 5 W is used.
Additionally, the scanning frequency is optimized so that neither the scan frequency nor
scan direction affected the lineshape. Finally, to remove any possible experimenter
induced biases, several of the Q factors were reproduced in a single-blind comparison.
The size of the microtoroids, both major and minor diameters, is also kept constant
throughout the experiments.
Figure 4-6 is a typical resonance spectra (forward scan) at 850nm for a silica
microtoroid fabricated on a silicon wafer with a boron concentration of 1.63 ×10
14
cm
-3
.
72
The splitting of the resonant linewidth results from coupling into the clockwise (CW) and
counter-clockwise (CCW) modes.[16] Therefore, in order to determine the linewidth of
the resonance, it is necessary to fit the spectra to a dual Lorentizian. It is notable that the
Q factors of both modes are in excess of 100 million, indicating the low coupling losses
present in the system and the high quality of the optical cavity.
Figure 4-6 A fine scan (the forward scan direction) of the fundamental transverse mode of the microtoroid fabricated
from the film with boron concentration of 1.63 ×10
14
cm
-3
in silicon at 848.8nm with a dual-Lorentz fit (red). The
resonance shows splitting and the quality factor of the left and right peak is 3.39 ×10
8
and 2.67 ×10
8
separately.
Figure 4-7 shows the measured quality factor as a function of refractive index or
boron dopant concentration at the three wavelengths. The specific values which were
measured at each wavelength and refractive index are included in Table 4-1. The Q
factor decreases as the refractive index decreases and as the wavelength increases. This
dependence of the Q on refractive index is qualitatively the same as that shown in the
73
model plotted in Figure 2. This trend confirms that the diffusion of the dopant from the
silicon to the oxide is significantly affecting the Q in these devices.
Figure 4-7 Measured quality factor as a function of refractive index or boron dopant concentration at three wavelengths:
630, 850 and 980nm. The Q factor decreases as the refractive index decreases and the wavelength increases.
4.4 Conclusions
In summary, a mechanism which explains the difference between the Q factors
achieved by microtoroid and microsphere resonators is proposed and verified.
Specifically, the Q of microtoroidal cavities is negatively impacted as a result of dopants
diffusing from the silicon into the oxide layer during thermal growth. Our model, which
includes Q
is
, Q
ss
and Q
mat
, predicted the resonator device behavior across all
combinations of wavelengths and refractive indices. Therefore, to optimize the
performance of the device, microtoroid resonant cavities should be fabricated on an
undoped or intrinsic silicon wafer. Such a performance improvement could impact
74
numerous areas of research and technology, including bio/chem-detection[1, 4], nonlinear
optics[17, 18], fundamental physics[2, 3, 18], and telecommunications.[3]
75
Chapter 4 References
1. Zhu, J.G., et al., On-chip single nanoparticle detection and sizing by mode splitting in an
ultrahigh-Q microresonator. Nature Photonics, 2010. 4(1): p. 46-49.
2. Anetsberger, G., et al., Ultralow-dissipation optomechanical resonators on a chip. Nature
Photonics, 2008. 2(10).
3. Rokhsari, H. and K.J. Vahala, Ultralow loss, high Q, four port resonant couplers for
quantum optics and photonics. Physical Review Letters, 2004. 92(25): p. -.
4. Armani, A.M., et al., Label-Free, Single-Molecule Detection with Optical Microcavities.
Science, 2007. 317(5839): p. 783-787.
5. Choi, H.-S., X. Zhang, and A.M. Armani, Hybrid Silica-Polymer Ultra-High-Q
Microresonators. Optics Letters, 2010. 35(4): p. 459-461.
6. Gorodetsky, M.L., A.A. Savchenkov, and V.S. Ilchenko, Ultimate Q of optical microsphere
resonators. Opt. Lett., 1996. 21(7): p. 453-455.
7. Vernooy, D.W., et al., High-Q measurements of fused-silica microspheres in the near
infrared. Opt. Lett., 1998. 23(4): p. 247-249.
8. Gorodetsky, M.L., A.D. Pryamikov, and V.S. Ilchenko, Rayleigh scattering in high-Q
microspheres. J. Opt. Soc. Am. B, 2000. 17(6): p. 1051-1057.
9. Knight, J.C., et al., Phase-matched excitation of whispering-gallery-mode resonances by a
fiber taper. Opt. Lett., 1997. 22(15): p. 1129-1131.
10. Hale, G.M. and M.R. Querry, Optical-Constants of Water in 200-Nm to 200-Mum
Wavelength Region. Applied Optics, 1973. 12(3): p. 555-563.
11. Armani, D.K., et al., Ultra-high-Q toroid microcavity on a chip. Nature, 2003. 421(6926): p.
925-928.
12. Streetman, B.G. and S. Banerjee, Solid State Electronic Devices. 5 ed1999: Prentice Hall.
558.
13. Fair, R.B., Physical Models of Boron Diffusion in Ultrathin Gate Oxides. Journal of The
Electrochemical Society, 1997. 144(2): p. 708-717.
14. Snyder, P.G., et al., Variable angle of incidence spectroscopic ellipsometry: Application to
GaAs-Al[sub x]Ga[sub 1 - x]As multiple heterostructures. Journal of Applied Physics, 1986.
60(9): p. 3293-3302.
15. Spillane, S.M., et al., Ideality in a fiber-taper-coupled microresonator system for application
to cavity quantum electrodynamics. Physical Review Letters, 2003. 91(4): p. -.
16. Weiss, D.S., et al., ‘Splitting of high-Q Mie modes induced by light backscattering in silica
microspheres. Optics Letters, 1995. 20: p. 1835-1837.
76
17. Kippenberg, T.J., et al., Ultralow-threshold microcavity Raman laser on a microelectronic
chip. Opt. Lett., 2004. 29(11): p. 1224-1226.
18. Hsu, H.-S., C. Cai, and A.M. Armani, Ultra-low threshold Er:Yb sol-gel microlaser on
silicon. Optics Express, 2009. 17(25).
77
Chapter 5 Integrated microtoroid-waveguide system
5.1 Introduction
High quality factor (Q) whispering gallery mode optical resonators can confine and
store light for tens to several hundreds of nanoseconds, allowing the build-up of high
intensity optical fields. As a result of this performance, they have impacted fundamental
physics and biological studies, and are frequently a critical component in optical
communications systems, performing the role of an optical buffer or a laser.[1-3] One
reason for their influence is that many of these phenomena scale with Q
2
.[1, 4, 5] In
order to achieve the enabling ultra-high-Q (Q>100 million) factor, it is necessary to both
use a micro-resonators which intrinsically have low optical loss and efficiently inject the
photons into the device using a waveguide. Most of the whispering gallery mode optical
resonators mentioned in Figure 2-11 are standing-alone devices, which means that an
external waveguide such as an optical fiber taper has to be used to excite the optical
modes of the resonator.
The development of a fully integrated ultra-high-Q platform, with both the
waveguide and the resonator on the same substrate, is extremely challenging and has not
been possible due to the sensitive nature of these devices. Here, we demonstrate a novel
approach for fabricating a fully-integrated silica ultra-high Q toroidal whispering gallery
mode resonators integrated with on-chip waveguides. To achieve this performance, we
have developed several new fabrication techniques which are based on a combination of
top down and bottom up fabrication methods.
The quality factor of the device is controlled by numerous intrinsic loss
mechanisms, such as the scattering of photons from surface defects and material loss, and
78
extrinsic loss mechanisms, such as a poor coupling of photons into the cavity[6].
Therefore, to maximize the Q, it is necessary to minimize each one. Currently, all high Q
(Q>1 million) devices which are integrated with a waveguide on a silicon substrate are
limited by surface defects which arise from the fabrication process.[7, 8] Additionally,
because they are fabricated from materials like silicon and silicon nitride, their high-Q
performance is not constant from the visible through the near-IR. One device, the silica
toroidal resonator, solved both of these problems by employing a laser-based reflow
process which removes these defects and by leveraging the inherent low material loss of
silica which allows ultra-high-Q resonator from the visible through the near-IR[9].
However, because the reflow process reduces the diameter of the device by several
microns, the gap between the device and an adjacent waveguide increases beyond the
requisite sub-micron coupling distance.
In this thesis, these issues have been overcome by employing several novel
fabrication methods and leveraging a unique material effect, allowing fully integrated
devices with Q>1 million to be fabricated.
5.2 The design concept
Since we want to integrate a waveguide to the microtoroid, we can look into the
fabrication process for a microtoroid first. The fabrication process for a standing alone
microtoroid is discussed in section 4.3.1. First, thermal oxide of 2 m thickness on a
silicon wafer is used to make microdisks with a diameter of tens of microns and a height
of 2 m. Then the CO
2
laser reflow process melts part of the microdisk and transforms it
to a microtoroid with a certain minor diameter, at the same time, the major diameter
79
reduces to a smaller size due to volume conservation (Figure 5-1). For example, an
80 m microdisk is usually reflowed to a microtoroid with a major diameter of 40 m
and a minor diameter of 8 m. As such, if there is also a waveguide fabricated alongside
the microdisk, after reflow, the gap between the microtoroid and reflowed waveguide will
be tens of microns, which inhibits light coupling. Even if the reflow is just a few microns,
the microdisk still ‘shrinks’ to a microtoroid and the gap between the microtoroid and an
adjacent waveguide still increases beyond the requisite sub-micron coupling distance. So
it is impossible to integrate a waveguide with a microtoroid on the same substrate using
the fabrication method for a stand alone microtoroid.
Figure 5-1 schematics of the cross section of a microdisk (a) and a microtoroid (b), the reflow process transforms the
microdisk to a microtoroid with smaller manor diameter
80
As such, the major concern in the integration of a waveguide to a microtoroid is
the gap control. In this thesis, a new approach is proposed. Since the volume of silica
doesn’t change before and after reflow, if a silica with a rectangular shape with
height>>width as shown in figure 5-2 is reflowed, the rectangle will be transformed to a
circle with a diameter bigger than the width of the rectangle. So the size of the
microtoroid after reflow will be even bigger than the size of the microring before reflow
(Figure 5-2). If we also have a waveguide with similar size features alongside the
microring, the gap of the microtoroid and reflowed waveguide will be 2× (Diameter-
Width) smaller than the gap before reflow. By carefully choosing appropriate height,
width and gap values, we should be able to get sub-micron gaps after reflow.
Figure 5-2 schematics of the proposed design concept, (a) cross section of a microring with a waveguide (b)
microtoroid and reflowed waveguide with a sub-micron gap as a result of reflow and volume conservation
81
As part of this thesis, several different fabrication routes were developed to
fabricate this structure, and each will be detailed in the subsequent sections along with an
analysis of the pros/cons of the different methods.
5.3 Initial fabrication tries
To get the microring and waveguide with shape shown in figure 5-2a, a
photolithography mask with width of 4 m and 6 m and gap of 3 m and 5 m
respectively is designed as shown in Figure 5-3. The fabrication process will be
performed on a 12 m thick thermal oxide on silicon substrate.
Figure 5-3 Mask design schematic: the blue layer is designed for the membrane and the grey layer is for the microring
and waveguide; two different gaps of 3 m and 5 m respectively is designed to optimize the gap
82
Given the thickness (height) of the oxide and the rectangular shape (anisotropic)
we want, it would be unrealistic to use buffered HF etching to etch the oxide. So an oxide
dry etching method should be used. Advanced oxide etching (AOE) system is used as it is
a system designed especially for etching oxide and has relatively high etching rate for
oxide, resulting in an anisotropic etch profile. Now we need to select the appropriate
protective etching mask for the dry etching. As photoresist will not be able to sustain
12 m thick oxide etching, chrome is chosen as the main protective etching mask as it has
good adhesion to oxide and is inexpensive. To get the shape shown in figure 5-2a, a two-
step AOE etching is needed. Initially, the first AOE etching is used to etch a deep trench
with a 1-3m thick membrane left (Figure 5-4a), then a second AOE etching is used to
carve out the rectangular shape (Figure 5-4b). Based on these considerations, the initial
process recipe is developed shown below.
1. Cr deposition
2. Blue layer Photolithography Cr wet etching
3. First AOE etching until 1-3 m oxide is left (Figure 5-4a)
post AOE cleaning, (Cr removal, BOE dipping)
4. Cr deposition
5. Grey layer photolithography and Cr wet etching
6. Second AOE oxide etching(Figure 5-4b)
post AOE cleaning (O2 plasma cleaning, Cr removal, BOE dipping)
7. XeF2 etching (Figure 5-4c)
8. reflow (Figure 5-4d)
83
Figure 5-4 schematics of fabrication process: (a) shape after first AOE etching (b) shape after second AOE etching (c)
XeF
2
undercut (d) CO
2
laser reflow
5.3.1 Fabrication with 4 m thick thermal oxide on silicon
This recipe is first tried out with 4 m thick thermal oxide on silicon while waiting
for the 12 m thick thermal oxide from the vendor. Figure 5-5 shows the device after two
step AOE etching, XeF
2
etching, and CO
2
laser reflow. While the general process was
successful, several issues became apparent.
84
Figure 5-5 optical microscopic images of the device after two step AOE etching (a), XeF
2
etching(b), and CO
2
laser
reflow(c) and SEM image of the device
First, a black residue is left on the surface of oxide after the two step AOE etching
(Figure 2-5a). The residue could be caused by photolithography and Cr etching or it
could be that the AOE chamber is not clean. This could be improved by careful
fabrication and AOE chamber cleaning before etching. Second, the XeF
2
etching is very
asymmetric and non-uniform; this asymmetry is because the silicon substrate is not clean
before placing in the chamber and has some sort of residue on it which may come from
the AOE etching as it tends to deposit polymers onto samples during etching (Figure 2-
5b). This can be solved by extra cleaning of the device such as piranha cleaning or
oxygen plasma. Third, during reflow, with certain reflow power, the microrings get
85
reflowed but the waveguide alongside it doesn’t get reflowed; this means that the heat
sink (the silicon pillar) is too big for the waveguide compared to the microring (Figure 2-
5c). In a future mask design, the total width of the waveguide and the diameter of the
microring should be optimized so that they can reflow at approximately the same power.
In the SEM image, black residues on the surface of silicon substrate can be clearly seen
(Figure 2-5d).
With cleaning AOE chamber before etching and piranha cleaning of the sample
after AOE etching, clean samples has been successfully fabricated with 4 m thick
thermal oxide on silicon. As shown in figure 5-6a, the XeF
2
etching is very uniform and
symmetric, and the microtoroids are perfectly rounded after CO
2
laser reflow (Figure 5-
6b). Additionally, higher power was used to reflow the waveguide and the microrings.
Figure 5-6 improved fabrication result with 4 m thick thermal oxide on silicon: (a) microscopic image of the device
after XeF
2
etching, (b) after CO
2
laser reflow
However, as shown in figure 5-6b, the gap between the microtoroid and the
reflowed waveguide is over 10 m which is impossible to couple light. However, given
the numerous steps in the fabrication process, it is important to check the Q of the
86
microtoroids to make sure that they still maintain high performance. This measurement
can be performed using a tapered optical fiber. The quality factor is in the order of 10
7
which indicates that the fabrication process didn’t degrade the device quality very much
compared to the conventional microtoroids (Figure 5-7).
Figure 5-7 Quality factor tested with an optical fiber taper of the microtoroids fabricated using AOE etching
5.3.2 Fabrication with PECVD silica
Now that very good quality factor microtoroids and waveguides have been
successfully fabricated with 4 m thick thermal oxide on silicon, we need to further
optimize this process. One key change is using thicker oxide to hopefully achieve the
submicron gap. Around 10 m thick oxide is grown on a silicon substrate with PECVD
first and then used to fabricate the system using the improved recipe. Due to the poor
87
quality of PECVD silica, after reflow, the silica is porous, but the gap looks optimistic
(Figure 5-8).
Figure 5-8 microtoroid and waveguide system fabricated using 10 m thick PECVD silica (a) microscopic image of the
device before reflow, (b) after reflow
5.3.3 Fabrication with 12 m thick thermal oxide
At first, same recipe as the one for 4 m thick thermal oxide is used for 12m thick
thermal oxide. Figure 5-9 shows the fabricated device using this recipe. The pattern on
the mask isn’t transferred well to the device. Some regions of waveguides or rings have
gone missing. This is mainly caused by the trenches created after first AOE etching and
second photolithography. As mentioned in section 5.3, the first AOE etching will create
around 10 m deep trenches, this deep trench made the second photolithography very
hard as the thickness of the photoresist is normally one or two microns. The photoresist is
very thin at the edges of these trenches and this won’t be able to protect the Cr
underneath during Cr etching, so a lot of patterns are ruined and don’t preserve the
patterns on the masks any more.
88
Figure 5-9 microscopic images of the device fabricated using 12 m thick thermal oxide (a) before reflow (b) after
reflow
But if we look at the SEM images of the reflowed device, at least we can tell that
the device has smooth surface where it is reflowed and the gap of the waveguide and
microtoroid looks very promising although the shape of the device isn’t good looking.
Figure 5-10 SEM images showing the side view of the reflowed device (a) and the gap between the microtoroid and the
waveguide (b)
To solve the problem of the second photolithography transfer, the thick
photoresist AZ4620 is used to protect the edges of the trenches. With 5000rpm spin speed,
around 6 m thick photoresist can be achieved. So using this thick photoresist for the
89
second photolithography and Cr etching, devices with a better shape have been fabricated
as shown in figure 5-11. Figure 5-11a is the microscope image of the device right after
the second AOE etching with Cr and photoresist residues. The patterns are transferred
significantly better with no missing regions compared to using thin photoresist such as
S1813 or AZ5214. Figure 5-11b is how the device looks after it’s been cleaned.
Figure 5-11 microscopic image of the device right after second AOE etching (a) and after additional cleaning (b)
But if we look at the SEM images of the device, it is clear that there is some
residue on the inner and outer sidewalls of the waveguides and microrings even after
cleaning (indicated by arrows in Figure 5-12a). This is a type of polymeric residue caused
by AOE etching of the photoresist and is very hard to remove. Figure 5-12b shows the
reflowed device which looks a lot better compared to the one using thin photoresist.
90
Figure 5-12 SEM images showing the top view of the reflowed device (a) and the reflowed microtoroid and waveguide
system (b)
From the side view SEM image, we can tell that the microtoroid and the waveguides
are coplanar which is necessary for evanescent coupling (Figure 5-13a). The gap after
reflow is around 2 m which is not small enough for light coupling but by optimizing the
fabrication process and mask design, sub-micron gap should be able to achieve with this
concept (Figure 5-13b).
91
Figure 5-13 SEM images showing the side view of the reflowed device (a) and the gap between the microtoroid and the
waveguide (b)
5.4 One of the fabrication routes which enables microtoroid-waveguide
system with sub-micron gap
As discussed above, the deep trench after the first AOE etching causes lots of
problems for the subsequent fabrication steps, so a new fabrication process is developed
with a new mask.
Figure 5-14 is the schematic of the new mask. The minor diameter of the ring and
waveguide are designed to be 8 m. This way, there will be more oxide to be reflowed to
hopefully decrease the gap. Also, in addition to straight waveguides, curved waveguides
are designed with 100 m and 20 m bend radii respectively. Curvature can help push
the optical modes towards the microtoroid to aid in coupling. The gap between the
waveguide and the microring is designed to be 3 m as from the earlier fabrication. In
addition, the first layer (blue layer) has a 4 m overlap with the second layer (grey layer),
this overlap is used to compensate the misalignment from the second photolithography
and prevent the structure to be etched through at some misaligned points.
92
Figure 5-14 the schematics of the new mask
The improved fabrication process is shown in figure 5-15. Instead of etching down
the blue layer big trenches during the first AOE etching, the blue layer is protected in the
first AOE etching and the region except the blue layer region is etched down for 2 to 3 m.
Then the grey layer is protected and all other region is etched down another 9 to 10 m.
This way, the biggest step handled is only 2 to 3 m which makes the subsequent process
much easier and more reliable. The improved process recipe is shown below. A more
detailed fabrication process for each step is elaborated in the following sections.
93
Figure 5-15 schematic of the fabrication process for the two time AOE etching (a) after first time AOE etching (b) after
second time AOE etching
1. Cr deposition
2. Blue layer Photolithography, Cr wet etching
3. AOE etching for 2-3 m oxide (Figure 5-15a)
post AOE cleaning, (O2 plasma cleaning, Cr removal, piranha
cleaning, BOE dipping)
4. Cr deposition
5. Grey layer photolithography and Cr wet etching
6. AOE oxide etching for another 9-10 m oxide (Figure 5-15b)
post AOE cleaning (O2 plasma cleaning, Cr removal, piranha
cleaning , BOE dipping)
7. XeF2 etching
8. Reflow
94
5.4.1 First Cr deposition
Even though the first AOE etching removes only 2 to 3 m, Cr is still used as the
etching mask. Since the AOE etching will damage the photoresist and will burn the
photoresist to the oxide which will degrade the oxide, it’s not used in direct contact with
oxide.
For the first Cr deposition, only 100 to 200nm Cr are needed. Cr is deposited using
electron beam evaporator. During the deposition, it is important to keep the wafer holder
moving around to get a more uniform deposition. Also adjust the beam position to the
middle of the deposition area to prevent the beam hitting a spot for too long. If the beam
is hitting a spot for a long time, the beam will melt through the Cr and reach the bottom
of the crucible which in turn may contaminate your deposition by melting the crucible.
The deposition rate is set to 3 Angstrom/s, so for 200nm deposition, it’ll take about
11minutes. A single run for the evaporator should be less than 20minutes, so if the
deposition time exceeds that time limit, multiple runs should be used.
5.4.2 First photolithography and Cr wet etching
The first photolithography is performed using S1813 photoresist. The recipe that is
used is shown below.
95
Once the photolithography is done, hard bake the photoresist at 115 degrees for
2:30min to harden the photoresist for the following Cr wet etching. Put the wafers in
chrome etchant (CR-7 from Cyanteck Corporation), make sure to stir the etchant
constantly to accelerate the etching. For 200nm thick Cr, it takes about 2:30 to 3 min to
etch away. Make sure to rinse the wafer thoroughly with running DI water and blow dry.
When Cr etching is finished, remove the remaining photoresist using acetone, and
then clean the wafer using methanol, IPA, then blow dry. Check under the microscope to
see if the photoresist is all cleaned, if not, acetone rinse or O
2
plasma cleaning may be
needed to remove it. Now, the blue layer is covered by Cr only as the etching mask for
the subsequent AOE etching as shown in Figure 5-16.
1. Take photoresist (S1813) from fridge and allow it to get back to room
temperature.
2. Bake at 120 degree for 5min to dehydrate the wafers, let wafers cool down.
3. Put the wafers inside HMDS setup for 2min.
4. Spin coat PR at 3k rpm for 30s. Clean the backside of the sample if there is
PR residue
5. Soft bake on hotplate at 100 degree for 2min. Let the PR to cool down.
6. Expose. Use 80mJ/cm2
7. Develop in MF-321. Stir while developing and watch the development to
make sure it’s developed properly. Usually 30s should be enough. Wash
with running DI water. Blow dry. Check under microscope to make sure
the patterns are developed well.
96
Figure 5-16 Microscopic image of the device after first photolithography and Cr wet etching
5.4.3 First AOE etching
The AOE is used to etch deep and fast SiO
2
, glass or quartz and fast silicon nitride
etch using fluorine and oxygen chemistries. Inductively coupled plasma (ICP) generates
very dense plasma near the top of the electrode. A second RIE generator which is
capacitively coupled to the wafer chuck is used to independently bias the substrate. In
this way, high selectivities and high etch rates can be obtained.
Small pieces wafer (< 4 inches) must be glued to a 4 inch Si carrier wafer in order
for the loading mechanism to work. The Si carrier wafer or your process wafer must be
very clean (like a new wafer that you just pull out from the box). It is extremely
important that the back of the wafer is extremely clean. The wafer has to be within 450 to
500um thickness.
As the 12 m thermal oxide wafer is 2 inches, it has to be glued to a 4 inch silicon
wafer. Use an eyedropper or Q-tip to deposit a small drop of cooling grease (or
equivalent) on the backside of your wafer. Then carefully center the backside of your
97
piece onto the front-side of the carrier wafer and press down firmly. Do not use too much
grease as this will be squeezed out from the back when you press down and can cause
problems during etching. Bake your sample on a hot plate at 120° C for at least 3 minutes.
This hardens the resist. Do not allow the backside of the carrier wafer to become
contaminated (eg from the hot plate).
It is very important to make sure the chamber is clean before the etching. So
usually an O
2
cleaning process is run for 30min to clean the chamber first. After the
cleaning process, run the recipe that is to be used for 5 to 10min to stabilize the system
using test samples.
The etching recipe used has CHF
3
as the process gas with a flow rate of 33sccm.
The ICP and RIE power is 700W and 100W respectively. And the process pressure is
5mTorr. With this recipe, the etching rate is about 300nm/min. So for the first AOE
etching, usually 10minutes of etching is used. Also measure the thickness of the wafer
before and after etching using the Nanospec system for reference.
When the AOE etching is done, remove the wafer from the Si carrier and clean
the backside and front side of the wafer using acetone, methanol and IPA, and then blow
dry. Put the wafer in O
2
plasma system to clean any photoresist residue. Remove the
remaining Cr using Cr etchant. Put the wafer in piranha to remove any residue caused by
AOE etching such as polymeric residues. Or alternatively, use the high power Matrix O
2
plasma system to strip off everything (Figure 5-17). Finally, a buffered oxide etching
(BOE) dipping is needed to get a fresh oxide surface for the following process.
98
Figure 5-17 Microscopic image of the device after the first AOE etching and Matrix O
2
plasma stripping
It’s very important to clean the wafer after the first AOE thoroughly. If there is
any residue left on the surface; it will cause cracking or peeling off of the Cr during the
next Cr deposition.
5.4.4 Second Cr deposition
As mentioned above, if the surface of the wafer is not clean, the Cr will peel off or
crack. Figure 5-18 shows an example of the Cr peeling because the photoresist is not
thoroughly cleaned off the surface after the first photolithography.
99
Figure 5-18 (a) microscopic image of a waveguide with photoresist residue on top and caused (b) the subsequent Cr
deposition peeling off
The first AOE etching creates a 2 to 3 m step,and the step is very anisotropic.
During Cr deposition ,the Cr thickness on the sidewall will be much thinner than the
thickness actually deposited. As shown in Figure 5-19, for a 550nm thick Cr deposition,
the Cr thickness on the sidewall is only around 50nm. 50nm of Cr won’t be able to
protect the sidewall during the second AOE etching. So a protective barrier is created
using a thick photoresist (AZ 4620). This requires a second photolithography step to
protect only the sidewall.
100
Figure 5-19 (a) SEM image of the side view of the waveguide deposited with 550nm Cr, (b) zoomed in view of the
sidewall
5.4.5 Second photolithography and Cr etching
The second photolithography is done with AZ 4620, the thick photoresist. The
recipe that is used is shown below. The AZ 4620 photoresist is very thick and viscous, so
handle it with care to keep it from contamination. During the alignment, leave a gap of at
least 15 m between the wafer and the mask to make sure they are not in contact. Align
using the split view function of the microscope and look at two the alignment marks at
different locations at the same time for easier alignment. Do alignment checks, make sure
it’s aligned well and then expose. Figure 5-20 shows the SEM side view image of the
developed photoresist on top of Cr.
101
Figure 5-20 SEM image of the second photolithography with AZ4620
1. Take photoresist (AZ4620) from fridge and allow it to get back to room
temperature.
2. Bake at 120 degree for 5min to dehydrate the wafers, let wafers cool down.
3. Put the wafers inside HMDS setup for 10min.
4. Spin coat PR at 5k rpm for 40s. Clean the backside of the sample if there is PR
residue
5. Soft bake on hotplate at 115 degree for 2:30 min. Let the PR to cool down.
6. Align the substrate with the second layer.
7. Expose. Use 240mJ/cm2
8. Develop in AZ400K diluted with DI water (1:4). Stir while developing and
watch the development to make sure it’s developed properly. Usually it takes
bout 3min to develop. Rinse thoroughly with running DI water. Blow dry.
Check under microscope to make sure the patterns are developed well. Pay
special attention to the gap region as some gaps may not be developed enough.
If so, add a few more seconds in the developer.
102
After the photoresist is developed well, hard bake it at 120 degree for 5min to
harden the photoresist for the following Cr wet etching. After the hard baking, it is
recommended to do an O
2
plasma descum for a few minutes. The descum helps remove
any residual photoresist at the pattern boundaries, which will benefit the following Cr
etching. 120 degree is chosen (similar to the hard baking temperature). The photoresist
has good adhesion to Cr which leads to a small undercut and the photoresist gets
reflowed a little bit but not too much (Figure 5-20a). As shown in Figure 5-20b, with 130
degree hard baking for 5min, the photoresist is reflowed too much and changed to a
circular shape although the undercut is also small. With 115 degree hard baking for 5min,
the photoresist isn’t quite reflowed and the amount of undercut is bigger compared to the
other two temperatures (Figure 5-20c).
103
Figure 5-21 SEM images of the photoresist at different hard baking temperatures and Cr wet etching profile at that
temperature: (a) 130 degree for 5min (b) 120 degree for 5min(c) 115 degree for 5min
104
When Cr etching is done, keep the photoresist on top of Cr to serve as an additional
protective mask for the sidewall during the second AOE etching. Hard bake the
photoresist again to remove any moisture from the Cr etching and the DI water rinse.
5.4.6 Second AOE etching
Follow the same procedure of preparing samples for the first AOE etching, do O
2
clean run to clean the chamber and do a test run.
The second AOE etching needs to etch 9 to 10 m oxide which takes about 30min.
It is recommended to split the etching into several shorter runs such as 15min plus 13min
plus 2min to get a more uniform etch. Check the thickness after each short run using
Nanospec system. It’s more difficult for the plasma to reach the gap region compared to
other regions, so pay special attention to the gap region and check under the microscope
to make sure that all the gaps are etched enough (with no oxide left).
Figure 5-22 shows the cross section of the waveguide right after the second AOE
etching. The photoresist is all etched away but Cr still has some left both on top of
waveguide and on the sidewall which means that the 550nm Cr and thick photoresist
were able to protect the oxide during the second AOE etching.
After the second AOE etching, the same cleaning process as the one after the first
AOE etching is used to clean the wafer. If some of devices have oxide left in the gap
region, put the device in BOE for a longer time until the oxide is gone (Figure 5-23).
Sometimes, if the oxide is too thick, then it would be not recommended to get it off with
BOE etching as it’ll also etch the oxide of the whole device.
105
Figure 5-22 SEM side view of the cross section of the waveguide right after the second AOE etching
Figure 5-23 Oxide not etched away in the gap region as indicated by the red arrows
5.4.7 XeF
2
etching
When the wafer is cleaned after the second AOE etching, use a diamond scribe to
cut the wafer into single devices, then clean these devices using acetone, methanol, IPA,
then blow dry. The devices are now ready for XeF
2
etching.
106
The XeF
2
etching results are affected by the all the previous fabrication steps. For
example, if the post AOE cleaning didn’t work well, this will cause the XeF
2
etching to
be very non-uniform and unsmooth. A good way to tell is to look at the reflections after
XeF
2
etching, if the surface of the silicon is clean, it’ll have reflections of the devices
(Figure 5-24). If there is oxide remaining in the gap region, the microring and the
waveguide will be bridged together (Figure 5-25). If the gap is too small, then it’s more
difficult for the XeF
2
to reach under the gap region which will cause the etching to be
asymmetric (Figure 5-26).
Figure 5-24 Miroscopic images showing XeF2 etching result of (a) not well cleaned device (b) well cleaned device
107
Figure 5-25 SEM image showing the microring and the waveguide bridged together
Figure 5-26 Optical microscopic image showing the remaining silicon near the gap region (indicated by red arrows)
If all of the above mentioned problem are not present, the device will be etched
both smoothly and symmetrically. Figure 5-27 shows the SEM image at different view
angles.
108
Figure 5-27 SEM images of the device after XeF
2
etching (a) top view (b) angled view (c) side view
The amount of undercut is very important and it’s suggested to undercut until the
silicon pillar is right at the inner sidewall as shown in Figure 5-28. Neither under nor over
109
etch would be able to get the submicron gap after reflow. Therefore, it is important to
monitor the Si etch process.
Figure 5-28 SEM image showing the amount of proper undercut, red arrows indicate where the inner sidewall and the
silicon pillar are
5.4.8 CO
2
laser reflow
When the devices are XeF
2
etched, they are ready for CO
2
laser reflow. A CO
2
laser is selected as it reflows the silica device without affecting the silicon substrate. The
reflow result is directly related to the quality of the fabricated device. Any residue on the
device or an imperfect XeF
2
undercut will all be translated to the overall result of the
reflow. The ends of the waveguide are not reflowed and form the input and output photon
injection ports. There is an optimum value of CO
2
laser power intensity at which the gap
is smallest, lower power intensity won’t reflow the device thoroughly and higher power
intensity will over reflow the device and in either case, the gap will be bigger than the
optimum value.
110
Figure 5-29 shows the optical microscopic images of the device before reflow and
after reflow. The reflowed microtoroid and waveguide has a submicron gap which is
evident from Figure 5-29b. Figure 30 is the SEM image of a reflowed microtoroid-
waveguide system with submicron gap. The major and minor diameter of the microtoroid
is around 80 m and 10 m respectively. The waveguide also has a diameter around 10 m.
the radius of curvature of the waveguide in the coupling region is 20 m. Both the
microtoroid and the waveguide has ultra smooth surface resulting from CO
2
laser reflow.
Figure 5-31a is the microscopic image of a reflowed device with submicron gap,
but after reflowing it with a higher power again, the gap increases beyond submicron.
Sometimes the microtoroid and the reflowed waveguide can be reflowed together. As
shown in Figure 5-32a, the microtoroid isn’t reflowed thoroughly at a certain reflow
power; however, after increasing the reflow power, the microtoroid and the waveguide
are reflowed together (Figure 5-32b).
111
Figure 5-29 microscopic image of the device (a) before reflow (b) after reflow with a submicron gap
112
Figure 5-30 SEM image of a microtoroid-waveguide system with submicron gap
113
Figure 5-31 Microscopic images of a reflowed device (a) with submicron gap, (b) after reflow with a higher power
again, the gap increases beyond submicron.
114
Figure 5-32 Microscopic images of a device (a) reflowed at certain power which didn’t reflow thoroughly (b) after
reflow with a higher power again, the microtorod and the waveguide are reflowed together.
5.5 The other fabrication routes which enables microtoroid-waveguide
system with sub-micron gap
An alternative fabrication process which involves only one Cr deposition is also
developed. The same mask as shown in Figure 5-14 is used. The recipe is shown below.
The resulting shape after each AOE etching is shown in Figure 5-33.
115
Figure 5-33 schematics of the device after (a) first AOE etching (b) second AOE etching
1. Grey layer Photolithography with image reversal recipe for AZ5214
2. Cr deposition and lift off
3. Blue layer Photolithography (PR=AZ5214)
4. AOE etching for 2 to 3um (Figure 5-32a)
post AOE cleaning-remove the remaining photoresist (AZ5214
stripper, O
2
plasma cleaning)
5. AOE etching for 9 to 10um(Figure 5-32b)
post AOE cleaning (O2 plasma cleaning, Cr removal, piranha
cleaning, BOE dipping)
6. XeF2 etching
7. Reflow
116
5.5.1 Detailed fabrication process
The first step is to do photolithography for the grey layer in Figure 5-14 with AZ
5214 with image reversal recipe (negative mask). The detailed parameters used are
shown below. When photolithography is done, deposit 550nm Cr using ebeam evaporator.
After the deposition, put the wafers in acetone to liftoff the metal except on the grey
patterns.
A second photolithography with AZ 5214 but with a normal recipe is used to cover
the blue region. The parameters used are shown below.
Spin coating: 500rpm for 5sec then 4000rpm for 30sec
Soft bake: 100 degree for 2min
Exposure: use vacuum contact, 8mW/cm2 for 3.6sec
Reversal bake: 118 degree for 50sec
Flood exposure: 8mW/cm2 for 80sec
Develop: in AZ400K 1:4 for 30sec
Before Cr evaporation: De-scum 30s @ 100W, 200mT
117
Once the second photolithography is done, do the first AOE etching for 10min to
etch about 3 m of oxide except the blue and grey region in Figure 5-14 (Figure 5-33a).
Then, AZ 5412 stripper is used to strip off the photoresist in the blue region. If the
stripper can’t remove the photoresist thoroughly, put the wafers in an O
2
plasma asher for
a few minutes. This will remove the photoresist completely.
After this, only the grey region in Figure 5-14 is covered with Cr. A second AOE
etching is then used to etch another 9 m oxide (Figure 5-33b). Post AOE cleaning is
then performed to clean the wafer, followed by XeF
2
etching and reflow which is
discussed above.
5.5.2 Fabrication result
With this fabrication method, devices with good reflowed shape can also be
achieved. Figure 5-34a is the device before reflow, and Figure 5-34a is the device after
reflow. But there is one question about this fabrication method: after the first AOE
etching, no proper process can be used to remove the possible residue from AOE etching
(appears as the rainbow colors on the devices). Whether this will affect the performance
of the device is as yet unknown.
Spin coating: 500rpm for 5sec then 3000rpm for 30sec
Soft bake: 100 degree for 2min
Exposure: use vacuum contact, 8mW/cm2 for 9sec
Develop: in AZ400K 1:4 for 30sec
Hard bake: 115 degree for 2:30min
Before AOE etching: De-scum 30s @ 100W, 200mT
118
Figure 5-34 Microscopic images of the device (a) before reflow (b) after reflow
5.6 Characterization of the microtorid-waveguide system
The characterization setup is shown in section 2.5.3. The quality factor and free
spectral range is discussed in section 2.4.1 and 2.4.2. To characterize the mode structure
and the quality factor of the device, a single mode, narrow linewidth tunable laser (1270
to 1330 nm) was used. The laser is coupled into and out of the waveguide injection ports
using a pair of single mode lensed fibers which are aligned using motorized nano-
positioning stages. As the laser is scanned over a series of wavelengths, the output signal
is recorded on a high speed digitizer/oscilloscope. From this data, the free spectral range
of the device and the quality factor can be determined. The free spectral range
corresponds to the distance ( ) between sequential optical resonances and can be
calculated (will be shown in the next section). The experimentally measured or loaded
quality factor is determined by fitting a single resonance to a Lorentzian and using the
expression Q= , where is the wavelength and is the full width at half
maximum[9].
119
The transmission spectra of a 70 m diameter integrated device with a bent
waveguide (100 m radius of curvature) is shown is Figure 5-35a. The free spectral
range which corresponds to the equatorial mode number (l-index) is 5.53 nm, which
agrees very well with the theoretically predicted value of 5.49 nm. The transmission
spectra shows that the resonator supports very few additional radial and azimuthal (m-
index or transverse) modes.
A resonance spectra with the lorentzian fit is shown in Figure 5-35b. The full
width half maximum from the fit is 3.1 × 10
-4
, yielding a loaded Q value of 4.3× 10
6
.
This measurement was repeated on several different devices, varying the waveguide
radius of curvature from 20 m to 100 m. All devices yielded Q values in excess of 1
million. Additionally, after storing the devices in ambient environments for over a week
(room atmosphere and temperature), the Q values and spectra were measured again and
no statistically significant change occurred
2
.
Figure 5-35 The optical characterization of the integrated ultra-high-Q microtoroid resonant cavity. a) Transmission
spectra of the 70 µm diameter integrated microtoroid with a bent waveguide with a radius of curvature of 100 µm. The
free spectral range of this device is 5.53nm. b) An example resonance spectra with the lorentzian fit. The full width at
half maximum (dl) from the fit is 3.1 × 10
-4
, yielding a loaded Q value of 4.3× 10
6
.
2
The quality factor was the same, to +/- 3%, which is within the experimental error of the measurement.
120
5.7 Analytical calculation of free spectral range
The free spectral range (FSR) of an optical cavity is the separation between two
resonant wavelengths: =
2
/(L
cav
n
eff
)=
2
/( Dn
eff
), where is the FSR, is the
wavelength, L
cav
is the length of the cavity, n
eff
is the effective refractive index, and D is
the cavity diameter. Therefore, in order to accurately calculate the FSR, it is necessary to
determine n
eff
. In toroidal cavities, this requires using finite element method simulations
to model the optical field distribution in the cavity and in the environment. In the present
work, we used COMSOL Multiphysics to determine the optical field distribution. The
optical constants of the system used in the simulation are taken from the COMSOL
material library. A mesh size of less than 0.021 m
2
is used and the fundamental mode is
modeled.
The effective refractive index was calculated using the following expression:
n
eff
= n
cavity
+ n
air
where are the percent of the field in the cavity and in air
respectively and n
cavity
and n
air
are the refractive indices of silica and air. The optical field
distribution is dependent on the device geometry (major diameter D and minor diameter
d). Therefore, the n
eff
was calculated for a range of major and minor diameter
combinations (Figure 5-36). The combination which most closely matched the
experimental geometry was reported above in Figure 5-35.
121
Figure 5-36 a) Calculated effective refractive index for a range of major and minor diameter combinations. b)
Calculated free spectral range for a range of major and minor diameter combinations.
5.8 Conclusions
In conclusion, an ultra-high-Q microtoroid-waveguide system has been
successfully demonstrated. Because they are fabricated from silica, the ultra-high-Q
factors will be maintained from the visible through the near-IR, paving the way for high
performance, fully integrated photonic systems which operate across a wide range of
wavelengths. Finally, the device has an extremely well-defined free spectral range which
agrees very closely with theoretical calculations. This combination of ultra-high Q and
wavelength flexibility will enable these integrated systems to accelerate research in the
fields of systems biology and biodetection[10-12], optical computing and
telecommunications[2, 13], and fundamental physics[4].
122
Chapter 5 References
1. Armani, A.M., et al., Label-Free, Single-Molecule Detection with Optical Microcavities.
Science, published online July 5, 2007 [DOI: 10.1126/science.1145002]. 317: p. 783 (2007).
2. Levy, J.S., et al., CMOS-compatible multiple-wavelength oscillator for on-chip optical
interconnects Nature Photonics, 2010. 4(1): p. 37-40.
3. Hunt, H.K. and A.M. Armani, Label-Free Biological and Chemical Sensors. Nanoscale,
2010. 2(9): p. 1544-1559.
4. Verhagen, E., et al., Quantum-coherent coupling of a mechanical oscillator to an optical
cavity mode. Nature, 2012. 482(7383): p. 63-67.
5. Matsko, A.B., et al., Hard and soft excitation regimes of Kerr frequency combs. Physical
Review A, 2012. 85(2): p. 023830.
6. Gorodetsky, M.L., A.D. Pryamikov, and V.S. Ilchenko, Rayleigh scattering in high-Q
microspheres. J. Opt. Soc. Am. B, 2000. 17(6): p. 1051-1057.
7. Gondarenko, A., J.S. Levy, and M. Lipson, High confinement micron-scale silicon nitride
high Q ring resonator. Optics Express, 2009. 17: p. 11366-11370.
8. Shah Hosseini, E., et al., High quality planar silicon nitride microdisk resonators for
integrated photonics in the visible wavelength range. Optics Express, 2009. 17: p. 14543-
14551.
9. Armani, D.K., et al., Ultra-high-Q toroid microcavity on a chip. Nature, 2003. 421(6926): p.
925-928.
10. Washburn, A.L. and R.C. Bailey, Photonics-on-a-Chip: Integrated Waveguides as Enabling
Detection Elements for Lab-on-a-Chip Biosensing Applications. Analyst, 2011. 136: p. 227-
236.
11. He, L.N., et al., Detecting single viruses and nanoparticles using whispering gallery
microlasers. Nature Nanotechnology, 2011. 6(7): p. 428-432.
12. Dominguez-Juarez, J.L., G. Kozyreff, and J. Martorell, Whispering gallery microresonators
for second harmonic light generation from a low number of small molecules. Nature
Communications, 2011. 2: p. 254.
13. Hafezi, M., et al., Robust optical delay lines with topological protection. Nature Physics,
2011. 7(11): p. 907-912.
123
Chapter 6 Integrated microtoroid-waveguide system sensors
6.1 Introduction
Whispering gallery mode resonators are especially suited for sensing applications
due to their long photon lifetime or high Q factor. The long photon lifetime enables
extreme sensitivity down to the single molecule level. Various research has been done
using whispering gallery mode resonators as biological, chemical or environmental
sensors[1, 2].
Whispering gallery mode resonators can confine light at certain wavelengths which
are also called the resonant wavelength of the cavity. The optical field of these
resonances extends into the surrounding environment and can interact with the
surrounding environment. If anything binds to the surface of the resonator, the refractive
index of the resonator will be changed and thus cause the resonance to shift. The higher
the quality factor, the longer sampling time, and thus the higher sensitivity. Additionally,
since the circumference of the resonator fits an integer multiple of resonant wavelength,
so any changes in the size of the device will also cause the resonant wavelength to shift.
To demonstrate the utility of this device in sensing applications, two different
sensing experiments are performed. First, the device is used to detect small changes in
temperature in the environment, and then the device is used to detect UV light.
One of the advantages of using silica over other material systems is its linear
power-dependent response at high input powers. The primary mechanism is based on the
thermo-optic effect, in which optical power is converted to heat which induces a
refractive index change. However, because the thermo-optic coefficient of silica is very
low, the overall response of silica devices is also small. As a result, high optical powers
124
can be coupled into the cavity with minimal impact on the resonant linewidth and
position. So in this thesis, power dependence experiments is also performed..
6.2 Power dependence
To determine the effect of circulating power on the resonant wavelength and
lineshape, the 1550nm tunable laser is connected to an EDFA to amplify the power. The
output from the EDFA is connected to the input lensed fiber. The power from the EDFA
can be increased manually. To determine the input power to the microtoroid, the insertion
loss and the percent of coupling are taken into account.
Figure 6-1 shows the power dependent wavelength response of the device in terms
of both input power and circulating power (P
circ
). The resonant wavelength increases
linearly in direct proportion to the input power, indicating that thermal effects are solely
responsible for the change. It is important to note that even with a change in circulating
power of 10W (eg. P
circ
increases from 9W to 19W), the change in the resonant
wavelength is extremely small, approximately 5pm, which corresponds to 10 cavity
linewidths.
125
Figure 6-1 As the input power is increased, the resonant wavelength slightly shifts. However, because of the low
thermo-optic coefficient of silica, several hundred microwatts of input power, which corresponds to several watts of
circulating power, are needed to induce a multi-linewidth shift.
6.3 Temperature and UV sensing demonstrations
6.3.1 Detection mechanism and testing set-up
Because the specific location of the resonant wavelength is determined by both the
geometry of the cavity and the device’s material properties, changes to either value will
induce a resonant wavelength shift. This effect can be summarized in the below
expression:
/ / / n n R R (6.1)
where is the resonant wavelength, is the change in resonant wavelength, n is the
refractive index, n is the change in refractive index, R is the radius of the device, and
R is the change in radius. In the present work, the primary detection signal is due to a
126
refractive index change caused by the thermo-optic effect. Therefore, the above
expression can be simplified to:
0
/
()
dn dT
T
n
(6.2)
where T is the change in temperature and is the thermal expansion coefficient. For
fused silica, dn/dT is 1.2E-5 C
-1
and is 0.55E-6 C
-1
at room temperature [3]. However,
because the optical field is not entirely confined within the device, the above expression
provides an upper bound for the measurement.
The same basic testing set-up to the one shown in section 2.5.1 is used with two
modifications. To perform the temperature experiments, the sample stage is replaced with
one that has an integrated heater and thermocouple (Figure 6-2a). Specifically, the heater
(Omega CSH-102100/120 V) is embedded directly into the sample holder. Silver
conductive epoxy (MG Chemicals) is used between the heater and the sample holder to
ensure effective heat transfer, to minimize heat lost to the environment and to hold the
heater in place. A fast response thermocouple sensor (Omega SA1XL) is attached
immediately adjacent to the optical device to accurately read the temperature in real time.
The response time for the thermocouple sensor is less than 0.15 s. Both the heater and the
sensor are connected to a benchtop controller (Omega CSC32 series) to control and read
the temperature.
127
Figure 6-2 The testing set-ups for the detection experiments. (a) The cylindrical heater is integrated directly under the
integrated resonator, and the thermocouple is adjacent to the sample. (b) The UV lamp is position directly above
(13mm gap) the resonant cavity.
To perform the UV detection experiments, a 385 nm LED UV lamp with an
integrated 5mm lightguide is used (Bluewave LED DX-1000) (Figure 6-2b). The
lightguide can deliver up to 15W/cm
2
in a high intensity spot. The intensity is adjustable
from 1% to 100% of the total power using a digital controller. During the present series
of experiments, a sub-set of this range is used (54mW/cm
2
to 100mW/cm
2
). The
lightguide is mounted vertically 13mm above the device.
In both sets of experiments, a 1550nm tunable laser connected to a lensed optical
fiber is used to couple light into the cavity. Once the resonant wavelength is located, the
change in the resonant wavelength can be tracked and recorded using a customized
Labview peak tracking program. Both the transmission and the resonant wavelength are
simultaneously recorded. By monitoring both values, fluctuations in coupling can also be
detected. For comparison, a baseline signal is taken for three minutes, and a Gaussian is
fit to the signal fluctuation. The noise level is set as the value which is 3σ from the center
of the Gaussian.
128
6.3.2 Temperature sensing
Representative results from the temperature sensing experiments are shown in
Figure 6-3. As indicated in Figure 6-3a, the temperature increase increment is changed
from 0.5
o
C to 1.7
o
C to check the linearity of the sensor. The slight dip at the end of
equilibrium of each increment is due to cooling while the next temperature increment is
manually set on the digital controller. However, after this initial dip, the resonant
wavelength shift is easily identifiable and stable.
A histogram of the noise distribution in the measurement is shown in the inset of
Figure 6-3a. This measurement was taken with a T of 0.5
o
C. As such, it includes
laser fluctuations, detector noise, and resonator noise as well as noise from the
temperature stage. However, even with these numerous potential sources of noise, the
total noise (3 σ) in the measurement is 0.4952pm. Typically, the theoretical noise limit is
the linewidth of the device, which is 0.49pm in the present measurement. Therefore, this
noise level is excellent.
As shown in Figure 6-3b, the resonant wavelength shift increases linearly as the
temperature is changed, and the slope is 14.6pm/° C. Using this value, we can calculate a
theoretical sensitivity limit which incorporates both the noise and the sensing
performance of the device. Setting the signal to noise threshold at 2, the minimum T
which is detectable is 0.0678
o
C. This value is particularly impressive as it assumes no
noise correcting algorithms or subsequent data processing has been used.
Assuming that the dominant contribution to this shift is the change in refractive
index, the slope is directly proportional to the dn/dT of the material and corresponds to a
129
dn/dT value of 1.27 × 10
-5
° C, which is in excellent agreement with the established dn/dT
value of silica (1.2× 10
-5
° C).
Figure 6-3 Temperature sensing experiments. (a) Sensor response when the temperature is increased. Inset: The
histogram from the noise measurement with a Gaussian fit. (b) The results from part (a) are re-plotted to highlight the
relationship between the resonance shift and the temperature change. The solid line is the linear fit.
6.3.3 UV sensing
While the detection of a temperature change is clearly based on the thermo-optic
effect, it is not immediately apparent that a similar mechanism is responsible for the
detection of UV light. However, because of the high material absorption ( of silica in
the UV range and the heat capacity of silica (c), the same mechanism applies.
Specifically, the temperature change ( T) upon exposure to UV light can be analytically
described by: T=E l/mc, where c is 703 J/kgC, is 3.262m
-1
at 385nm, l is the
absorption length, m is the mass of the device, and E is the energy from the UV source
[4]. Both l and m are determined by the specific toroid dimensions while E is an
experimentally controlled parameter.
Figure 6-4a shows the sensor response to several different UV intensities,
increasing from 54mW/ cm
2
to 100mW/ cm
2
and then decreasing back to 54mW/ cm
2
.
For each cycle, we first exposed the microtoroid-waveguide system, left the UV on for
130
15minutes, then turned the UV off and waited for the resonant wavelength to recover.
The sensor system exhibited very little hysteresis, and the wavelength completely
returned to its original position after all 5 cycles. This result indicates that the UV sensing
response produced by the microtoroid-waveguide system is stable, and its sensing
performance is not degraded due to UV-induced damage to the silica. Figure 6-4b is an
enlarged version of the highest intensity signal (100mW/ cm
2
). Once the UV light is
turned on, the resonant wavelength undergoes a large shift and then stabilizes. When the
UV is turned off, the resonant wavelength recovers quickly. In this measurement, the
total noise (3 ) is 0.5587pm (Figure 6-4b, inset). This value is much smaller than the
device linewidth (2.5pm) which indicates excellent noise level.
Figure 6-4 UV sensing results. (a) Sensor response with several different exposure intensity cycles, increasing from
54mW/ cm
2
to 100mW/ cm
2
and then decreasing to 54mW/ cm
2
(b) The characteristic UV sensing curve showing both
the forward and reverse response at 100mW. The resonance undergoes a large, rapid wavelength shift once the UV is
turned on. When the UV is turned off, the resonant wavelength returns to its original value. Inset: The histogram from
the noise measurement with a Gaussian fit.
6.4 Conclusions
As a result of the low thermo-optic coefficient of silica, the resonant wavelength is
extremely stable, even at high input powers, and experiences a predictable, linear
131
increase. Additionally, sensing experiments are performed using the microtoroid-
waveguide system. Specifically, the system is able to detect temperature changes, and
the wavelength shift changes linearly as the temperature increases. This system has a
characteristic UV sensing response, and the signal produced is stable and is not degraded
by UV exposure.
132
Chapter 6 References
1. Hunt, H.K. and A.M. Armani, Label-Free Biological and Chemical Sensors. Nanoscale,
2010. 2(9): p. 1544-1559.
2. Armani, A.M., et al., Label-Free, Single-Molecule Detection with Optical Microcavities.
Science, published online July 5, 2007 [DOI: 10.1126/science.1145002]. 317: p. 783 (2007).
3. Wakaki, M., K. Kudo, and T. Shibuya, Physical properties and data of optical materials2010:
CRC Press.
4. Kitamura, R., L. Pilon, and M. Jonasz, Optical constants of silica glass from extreme
ultraviolet to far infrared at near room temperature. Applied optics, 2007. 46(33): p. 8118-
8133.
133
Appendix A: Hybrid Silica-Polymer Ultra-High-Q
Microresonators
A.1: Introduction
Ultra-high-quality factor (UHQ) optical cavities have numerous applications
throughout engineering and science. Incorporating active elements into these UHQ
cavities to create dynamic devices would extend their applicability; however, it is
inherently difficult to develop an active UHQ device. Ultra thin films formed from
optically active polymers provide one route to overcome this limitation. In the present
work, hybrid devices comprised of UHQ planar optical cavities with ultra-thin films are
fabricated on a silicon wafer. Using finite element method simulations, the optical field
overlap between the cavity and the polymer film is modeled and experimentally verified
using two polymers, polymethylmethacrylate and polystyrene. These hybrid devices
have demonstrated material-limited Q factors above 10 million.
A.2 Theory
A.2.1 Loss mechanism
The Q factor of an optical cavity is described by a series of loss mechanisms
which include Q
mat
, Q
ss
, Q
rad
, and Q
coupl
, which are the material loss, surface scattering
loss, radiation loss, and coupling loss of the cavity. The first three are the intrinsic loss
(Q
o
) of the cavity and the last is the extrinsic loss (Q
ext
) of the system [1]. In the present
hybrid system, it is proposed that material absorption is the dominant loss mechanism,
yielding Q
o
~ Q
mat
. Under this condition, the quality factor of the device is described by
Q
mat
=2πn
eff
/λα
eff
where n
eff
and α
eff
represent the effective refractive index and optical
absorption coefficients. These are expressed as n
eff
=βn
silica
+γn
Polymer
+δn
air
, and
134
α
eff
=βα
silica
+γα
polymer
+δα
air
where β, γ and δ represent the percentage of the optical field in
silica, polymer, and air, respectively. The refractive index of the PMMA and PS is
verified using ellipsometry. The material loss of the silica is taken from reference [2] and
the material loss of the polymers is measured using spectrophotometry. All values agree
with commonly referenced values[3]. Therefore, to calculate Q
mat
, it is necessary to
determine the values for , , and .
A.2.2 FEM simulations
To determine and , the optical field distribution is modeled using COMSOL
Multiphysics finite element method [4]. The mesh size in the region of the optical field
was 0.016μm
2
(area divided by # of mesh), and the field distribution was modeled
holding either the geometry of the toroid or the film constant. Setting the PMMA film
thickness at 500nm, the optical intensity profile was calculated for major and minor
diameters of [40, 60, 80, 100]μm and [2.5, 3, 4, 5, 6, 7, 8, 9, 10] m, respectively. Then,
setting the major (minor) diameter of the toroid to 40(8) m, the field was modeled for
[20, 50, 100, 200, 350, 500]nm thick PMMA films. For the PS case, the same toroid
geometry was used, and the optical field was modeled with [20, 50, 100, 200]nm thick
films. The percentage of the optical field was determined by Power
In (Silica, Polymer,
Air)
/Power
Tot
, where Power
In
and Power
Tot
represent the portion of the optical power in
each region and total optical power in the cavity and its surrounding environment,
respectively. The resonant wavelength was set to 850nm or 980nm in each simulation by
controlling mode number (M), where M is azimuthal mode order in the cavity. Figure 1
shows FEM simulation results for a 40(8) m major(minor) diameter microtoroid. As the
135
polymer film thickens, the optical field shifts from the silica into the polymer film
(Figure A-1).
Figure A-0-1 (a) Scanning electron micrograph of a toroidal microresonator. Finite element method simulation results
for the optical field intensity distribution: (b) silica resonant cavity, (c) hybrid resonant cavity with a 100nm thick PS
film, and (d) 200nm thick PS film. Note that the optical field shifts from the silica towards the polymer film as the
thickness of the polymer film increases. The toroid major (minor) diameter is 40(8) m and λ=850nm.
By systematically tuning all of these variables (film thickness, major/minor
diameter, wavelength, polymer material properties), it is possible to map out the
interaction of the field with the polymer layer. Figure A-2a and Figure A-2b summarizes
these results for the two polymers studied: PMMA and PS. In Figure A-2a, the PMMA
film thickness is fixed at 500nm and the percentage of the optical field is determined as a
function of minor diameter from 2.5 to 10μm with the change of the major diameter from
40 to 100μm. From this simulation result, changing the major diameter of the device is
found to have a greater impact on the location of the optical field compared changing the
minor diameters. In Figure A-2b, the results from the inverse set of simulations is shown,
136
in which the toroid geometry was fixed at 40(8) m major(minor) diameter, and the film
material and thickness are varied.
Figure A-0-2 Finite element method results. (a) Percentage of the optical field in the polymer layer as a function of the
minor diameter as the major diameter increases from 40 to 100μm at λ=980nm. The PMMA film thickness is fixed at
500nm. (b) Percentage of the optical field as a function of the polymer film thickness for PMMA and PS at λ=850,
980nm.
Due to improved confinement of the optical field at the shorter wavelength, a
greater percentage of the optical field resides in the polymer layer compared to the longer
wavelength. Simulation results also show that there is a larger percent of the optical field
in the PS coating layer due to the higher contrast of refractive indices between silica and
PS, compared to silica and PMMA. This is because the polymer film, with a refractive
index higher than silica, acts as a poor cladding layer. It is important to note that while
micron-scale changes in the major and minor diameter of the toroid do not significantly
impact the optical field overlap, nm-scale changes in the film thickness do result in a
significant change in the optical field. Therefore, the experimental investigations will
focus on this aspect.
137
A.3 Fabrication of the hybrid device
The hybrid silica-polymer ultra-high Q microtoroids were made by first
lithographically fabricating the silica microtoroid device, and then applying the polymer
film. There are three major steps in the lithographic process: 1) pattern circular oxide
pads on a silicon wafer, 2) undercut the oxide with XeF
2
to form a microdisk structure,
and 3) use a CO
2
laser-induced reflow to form the ultra high Q microtoroids [5]. Two
different polymers were used throughout this project, PMMA (Sigma Aldrich, 15k MW)
and PS (Sigma Aldrich, 200k MW). To enable spin-coating deposition, the polymer was
dissolved in ultra-high purity toluene solvent (J.T. Baker, 99.5% min) at 0.5, 2, and 4%
for PMMA and 0.5, 1, 2% for PS. After spin-coating, the film is thermally reflowed for
30 minutes above the polymer’s glass transition temperature (T
g
) in a gravity oven at
115° C (PMMA) and 105° C (Polystyrene). This thermal reflow allows the hybrid
microtoroids to maintain the smooth surface of the silica devices. Film thickness
measurements for PMMA and for Polystyrene were performed by ellipsometry on bare
silicon wafers and compared to the value in a reference [6].
A.4 Experimental result and discussions
To measure the Q of the hybrid device, it is coupled to a narrow-linewidth, tunable,
CW laser (850nm or 980nm) using a tapered optical fiber. Tapered fibers have
demonstrated high efficiency/low-loss coupling to optical resonant cavities, thus
minimizing the coupling or extrinsic losses if the Q is measured in the undercoupled
regime [7]. To accurately align the tapered optical fiber and the resonant cavity, a two
axis (top and side) machine vision system in combination with a nm-resolution motorized
stage is used. Data is automatically recorded on a computer for ready analysis. The scan
138
frequency and laser power was optimized (forward and backward scans gave similar
linewidth measurements). In the present experiment, where thermal effects are especially
prominent, this is a particularly important aspect and was monitored closely. It should be
noted that the Q of every device was taken twice: before and after the polymer film was
applied. Only if a microtoroid had an initial Q factor which was above the theoretical Q
threshold was that microtoroid used.
The Q of the hybrid devices was measured with four different film thicknesses,
including without a film, for PMMA and for Polystyrene and at two different
wavelengths (850nm, 980nm). The modeling results and the experimental data are
presented in Figure A-3. It should be noted that Q factors above 10
7
were measured for
hybrid devices fabricated using either PS or PMMA. To the authors’ knowledge, these Q
factors are the highest to date for a hybrid device. The theoretical and experimental data
was fit to an equation of the form y=ax
b
, which is the appropriate form for Q
mat
. As seen
in Table 1, there is very good agreement between the values for (a, b) predicted by the
model and the experimental results. Therefore, the Q factor of the hybrid device is
material loss limited.
139
Figure A-0-3 Quality factor (Q) and the percentage of the optical field for the hybrid polymer resonant cavity devices
as a function of coating thickness. PMMA at (a) 850nm and (b) 980nm. PS at (c) 850nm and (d) at 980nm. The results
were fit to an equation of the form y=ax
b
, which is included in the figure as a solid (dashed) red line for the theoretical
(experimental) Q results. The parameters (a,b) are summarized in Table 1. The black dotted line indicates the highest
Q demonstrated with a silica toroidal resonant cavity to date, setting an upper bound on Q [17].
Model Experiment
Polymer
Wavelength
(nm) a (x10
9
) b a (x10
9
) b
PS
PS
PMMA
PMMA
850
980
850
980
5.24
4.76
3.11
1.37
-1.49
-1.42
-1.32
-1.19
4.96
4.02
1.52
0.398
-1.53
-1.47
-1.21
-1.06
Table A-1 Summary of Model and Experimental Fit Parameters.
A-5 Conclusions
In summary, hybrid microresonators with Q factors in excess of 10
7
have been
demonstrated. These devices have material-limited Q factors, as verified by excellent
agreement between the experimental results and the finite element method simulations.
These types of hybrid UHQ devices will find applications in telecommunications as
140
optical filters and modulators [8]. Additionally, as a result of the increased Q factor,
these devices will have reduced lasing threshold over previous polymer microlasers [9]
and improved detection sensitivity [10, 11] for bio/chem sensing applications.
141
Appendix A References
1. Gorodetsky, M.L., A.A. Savchenkov, and V.S. Ilchenko, Ultimate Q of optical microsphere
resonators. Optics Letters, 1996. 21(7): p. 453-455.
2. Pinnow, D.A., et al., FUNDAMENTAL OPTICAL ATTENUATION LIMITS IN LIQUID
AND GLASSY STATE WITH APPLICATION TO FIBER OPTICAL WAVEGUIDE
MATERIALS. Applied Physics Letters, 1973. 22(10): p. 527-529.
3. Kasarova, S.N., et al., Analysis of the dispersion of optical plastic materials. Optical
Materials, 2007. 29(11): p. 1481-1490.
4. Oxborrow, M., Traceable 2-D finite-element simulation of the whispering-gallery modes of
axisymmetric electromagnetic resonators. Ieee Transactions on Microwave Theory and
Techniques, 2007. 55(6): p. 1209-1218.
5. Armani, D.K., et al., Ultra-high-Q toroid microcavity on a chip. Nature, 2003. 421(6926): p.
925-928.
6. Walsh, C.B. and E.I. Franses, Ultrathin PMMA films spin-coated from toluene solutions.
Thin Solid Films, 2003. 429(1-2): p. 71-76.
7. Knight, J.C., et al., Phase-matched excitation of whispering-gallery-mode resonances by a
fiber taper. Optics Letters, 1997. 22(15): p. 1129-1131.
8. Rabiei, P., et al., Polymer micro-ring filters and modulators. Journal of Lightwave
Technology, 2002. 20(11): p. 1968-1975.
9. Tulek, A., D. Akbulut, and M. Bayindir, Ultralow threshold laser action from toroidal
polymer microcavity. Applied Physics Letters, 2009. 94(20).
10. Dong, C.H., et al., Fabrication of high-Q polydimethylsiloxane optical microspheres for
thermal sensing. Applied Physics Letters, 2009. 94(23).
11. Chao, C.Y. and L.J. Guo, Biochemical sensors based on polymer microrings with sharp
asymmetrical resonance. Applied Physics Letters, 2003. 83(8): p. 1527-1529.
Abstract (if available)
Abstract
High qualify factor (Q) integrated micro-cavities are not only one of the key elements in integrated photonics but also have numerous other applications such as fundamental physics studies and biosensing. Since the invention of silica ultra-high quality factor micro-disks and micro-toroids, they have proven to be very efficient in these applications. However, currently, optical fiber tapers are used to couple light into the micro-disks and toroids. In order to use them in practical applications, they have to be fabricated with a waveguide on the same substrate. The focus of this dissertation is the design of a new waveguide which enables the development of a fully integrated waveguide-resonator system. ❧ The silica integrated waveguides which are directly fabricated on a silicon substrate are demonstrated in the first part of the thesis. Silica splitters based on the silica waveguides are demonstrated as another essential element in photonic circuitry. The loss mechanisms which impact the quality factor of micro-toroids are investigated both experimentally and theoretically. Then, a novel approach for fabricating fully-integrated silica ultra-high Q toroidal resonators integrated with on-chip waveguides is demonstrated. Temperature sensing, UV sensing and power dependence experiments are performed using this fully integrated ultra-high Q toroidal resonator system.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Developing improved silica materials and devices for integrated optics applications
PDF
Silicon integrated devices for optical system applications
PDF
Active integrated photonic devices in single crystal LiNbO₃ micro-platelets and a hybrid Si-LiNbO₃ platform
PDF
Titania and hybrid titania - silica sol-gel thin films and their applications in integrated optical devices
PDF
Nonlinear optical nanomaterials in integrated photonic devices
PDF
Development of novel optical materials for whispering gallery mode resonator for nonlinear optics
PDF
Integration of practical high sensitivity whispering gallery mode resonator sensors
PDF
Development of integrated waveguide biosensors and portable optical biomaterial analysis systems
PDF
Application of optical forces in microphotonic systems
PDF
Integrated photonics assisted electron emission devices
PDF
Optical simulation and development of novel whispering gallery mode microresonators
PDF
Integrated silicon waveguides and specialty optical fibers for optical communications system applications
PDF
Fabrication and characterization of yoroidal resonators for optical process improvement
PDF
Analysis and design of cleaved edge lithium niobate waveguide resonators
PDF
Development of hybrid optical microcavities for Plasmonic laser and improving biodetection
PDF
Optical studies in photonics: terahertz detection and propagation in slot waveguide
PDF
Silicon micro-ring resonator device design for optical interconnect systems
PDF
Development of organic-inorganic optical microcavities for studying polymer thin films
PDF
Surface-enhanced on-chip whispering gallery mode resonators for improved nonlinear optics
PDF
Development of hybrid microsensors for environmental monitoring and biodetection
Asset Metadata
Creator
Zhang, Xiaomin
(author)
Core Title
Development of optical devices for applications in photonic integrated circuit and sensing
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Materials Science
Publication Date
11/21/2013
Defense Date
08/08/2013
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
OAI-PMH Harvest,photonic devices,resonators,silica,waveguides
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Armani, Andrea M. (
committee chair
)
Creator Email
xmzhang01@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c3-349704
Unique identifier
UC11296597
Identifier
etd-ZhangXiaom-2175.pdf (filename),usctheses-c3-349704 (legacy record id)
Legacy Identifier
etd-ZhangXiaom-2175.pdf
Dmrecord
349704
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Zhang, Xiaomin
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
photonic devices
resonators
silica
waveguides