DEVELOPMENT, FABRICATION AND CHARACTERIZATION OF GRAPHENE
AND BISMUTH HALL SENSORS FOR SCANNING HALL PROBE MICROSCOPY
by
SELDA SONUŞEN
Submitted to the Graduate School of Engineering and Natural Sciences
in partial fulfillment of
the requirements for the degree of
Doctor of Philosophy
Sabancı University
January 2015
DEVELOPMENT, FABRICATION AND CHARACTERIZATION OF GRAPHENE
AND BISMUTH HALL SENSORS FOR SCANNING HALL PROBE MICROSCOPY
APPROVED BY:
Prof. Dr. Ahmet Oral ………………………….
(Dissertation Supervisor)
Prof. Dr. Yuda Yürüm ………………………….
Assoc. Prof. Dr İnanç Adagideli ………………………….
Assoc. Prof. Dr Selmiye Alkan Gürsel ………………………….
Assoc. Prof. Dr Özgür Özer ………………………….
DATE OF APPROVAL: ………………………….
© Selda Sonuşen 2015
All Rights Reserved
iv
DEVELOPMENT, FABRICATION AND CHARACTERIZATION OF GRAPHENE
AND BISMUTH HALL SENSORS FOR SCANNING HALL PROBE MICROSCOPY
Selda Sonuşen
Physics, PhD Thesis, 2015
Thesis Supervisor: Prof. Ahmet Oral
Keywords: Graphene Hall Sensor, Bismuth Hall Sensor, Scanning Hall Probe
Microscopy (SHPM), Quartz Tuning Fork (QTF), Graphene Growth by Chemically
Vapor Deposition (CVD), Magnetic Imaging
ABSTRACT
Scanning Hall Probe Microscopy (SHPM) is a powerful magnetic imaging
technique which provides high magnetic field and spatial resolution, simultaneously
with the topography of magnetic and superconducting materials. It is a quantative
method which can be operated under high magnetic fields and a wide temperature
range. The Hall sensor which is sensitive to the perpendicular component of the
magnetic field emanating from the specimen determines resolution. In this work, single
layer Graphene Hall Probes (GHP) were fabricated using Chemical Vapor Deposition
(CVD) grown graphene on copper foils, transferred to Silicon wafers. The Hall
coefficient and field sensitivity of GPHs were measured to be 0.18 Ω/G and 0.20
G/√Hz, respectively, for a 3 µA drive current at room temperature. For the first time,
GHP is successfully used for magnetic imaging in 3–300 K range in SHPM with quartz
crystal Atomic Force Microscopy (AFM) feedback. This study has demonstrated that
graphene is an alternative material to be used for magnetic imaging in SHPM.
Alternatively, Bismuth based Hall probes are also fabricated by advanced lithographic
techniques. Since Bismuth is a semimetal with low carrier concentration, it has been
considered as a promising Hall sensor material. For the first time, SHPM images of
NdFeB demagnetized magnet were acquired using Bi hall sensor fabricated by Electron
Beam Lithography (EBL) with quartz crystal AFM feedback.
v
TARAMALI HALL AYGITI MİKROSKOPLARI İÇİN GRAFEN VE BİZMUT
HALL ALGILAYICI GELİŞTİRİLMESİ, ÜRETİMİ VE KARAKTERİZASYONU
Selda Sonuşen
Fizik, Doktora Tezi, 2015
Tez Danışmanı: Prof. Dr. Ahmet Oral
Anahtar Kelimeler: Grafen Hall Aygıtı, Bizmut Hall Aygıtı, Taramalı Hall Aygıtı
Mikroskopu, Kuvartz Kristal Çatal, Kimyasal Buhar Biriktime Yöntemi ile Grafen
Üretimi,
ÖZET
Taramalı Hall Aygıtı Mikroskobu (THAM) manyetik ve süperiletken
malzemelerin yüksek manyetik ve uzaysal çözünürlükle çalışan bir görüntüleme
tekniğidir. THAM yüksek manyetik alan altında ve geniş sıcaklık aralığında da
çalıştırılabilen nitel bir metoddur. İncelenen numuneden gelen manyetik alanın dik
bileşenine hassas olan Hall aygıtı, çözünürlüğü ve hassasiyeti belirler. Bu çalışmada,
kimyasal buhar biriktirme yöntemi ile elde edilen tek tabakalı grafenden üretilen,
Grafen Hall aygıtının oda sıcaklığında ve 3 µA akım değeri için, Hall katsayısı ve alan
hassasiyeti sırası ile 0.18 Ω/G and 0.20 G/√Hz olarak ölçüldü. Grafen Hall aygıtı
dünyada ilk kez 3–300 K sıcaklık aralığında manyetik görüntülemede, kuvars kristalli
Atomik Kuvvet Mikroskobu (AKM) geribeslemesi ile çalışan THAM ile başarılı olarak
kullanıldı ve de–manyetize edilmiş NdFeB mıknatısın görüntüleri elde edildi. Bu
çalışma, grafenin THAM da kullanılmak üzere alternatif bir malzeme olduğunu
gösterdi. Eş zamanlı olarak, bizmut Hall aygıtı ileri litografi teknikleri kullanılarak
üretildi. Bizmut düşük taşıyıcı yoğunluğuna sahip bir yarımetal olduğu için bizmut da
gelecek vaat eden hall aygıtı malzemesi olarak değerlendirilir. Elektron demeti
lithografisi ile üretilmiş Bizmut Hall aygıtı ilk kez THAM da de–manyetize edilmiş
NdFeB mıknatısın görüntülenmesinde kullanıldı.
vi
To Ayten Sonuşen and Hasan Sonuşen
vii
ACKNOWLEDGEMENTS
First of all, I would like to extend my deepest gratitude to my thesis supervisor
and greatest mentor to Prof. Dr. Ahmet Oral for his unfailing support throughout my
Ph.D studies. This thesis would not have been possible without his outstanding
supervision.
I would like to thank to Prof. Dr. Yuda Yürüm, Assoc. Prof. Dr İnanç Adagideli,
Assoc. Prof. Dr Selmiye Alkan Gürsel,and Assoc. Prof. Dr Özgür Özer for being on my
thesis committee.
I am indebted to Asst. Prof Dr. Seda Aksoy for helping me with the fabrication
and characterization of graphene Hall sensor. I am deeply grateful to Özgür Karcı and
Dr. Münir Dede for sharing their experience with me and for generating the magnetic
imaging in SHPM. I would like to thank to the entire staff of NanoMagnetics
Instruments Ltd. for their support during my training in this company. I am also
thankful to Assoc. Prof. Dr Hidayet Çetin sharing his experience on graphene.
I am grateful to Dr. Anıl Günay Demirkol , the ultimate polymath, both for her
friendship and mentorship. I am thankful to Cenk Yanık for making life easier,
bearable, and enjoyable for me all the time. I would also like to thank to Süleyman
Çelik for his rare competence in performing fine work. Hasan Özkaya and Dr. Sibel
Kasapare to be mentioned for their great support. I would like to acknowledge my
previous and current group members, Derya Gemici, Nihan Özkan Aytekin, Dr. Musa
Mutlu Can, Hüsnü Aslan, Dr. Shumaila Karamat, Yiğit Uysallı, and Ekin Özgönül.
My greatest thanks go to my friends at Sabancı University, especially to Naime
Kınalı, Senem Avaz, Bahriye Karakaş, Utku Seven, Raghu Sharma Mokkapati, Ezgi
Uzun, Ines Karmous and Damla Arifoğlu for their all unforgettable support during my
thesis writing process. I would not have been able to finalize this thesis without their
constant encouragement. I would like also like to thank to my dear friends, Billur,
Umut, Emre, Nurdan, and Gülhis for their friendship.
viii
Finally; no word describes my deepest feelings to my dad, Hasan Sonuşen, and
my mom, Ayten Sonuşen. If there were a hundred million words to describe my feelings
to them in my native language, I would not cease to use these words of indebtedness to
them every single day. My sister; Seda Ibas, the other half of my soul, and Doruk Ibas,
the bringer of luck, happiness and the joy for all of us, are at the top of my gratitude list.
I am also thankful to Arda Ibas for being a part of our family.
ix
TABLE OF CONTENTS
1 INTRODUCTION ....................................................................................................... 1
1.1 Context and Motivation ................................................................................ 1
1.2 Structure of the Thesis .................................................................................. 3
2 SCANNING PROBE MICROSCOPY AND MAGNETIC IMAGNING ................. 5
2.1 Scanning Probe Microscopy ........................................................................... 5
2.1.1 Scanning Tunneling Microscopy ..................................................... 5
2.1.2 Atomic Force Microscopy ................................................................ 7
2.1.3 Magnetic Imaging techniques .......................................................... 8
2.1.3.1 Magnetic Force Microscopy .............................................. 8
2.1.3.2 Scanning Hall Probe Microscopy ...................................... 9
3 SCANNING HALL PROBE MICROSCOPY .......................................................... 13
3.1 Introduction ................................................................................................. 13
3.2 Scanning Hall Probe Microscopy ............................................................... 15
3.2.1 The Microscope Design and Operation .......................................... 15
3.2.2 LT–SHPM Electronics ................................................................... 22
3.2.3 LT–SHPM Software ....................................................................... 23
3.3 Scan Modes of LT–SHPM .......................................................................... 23
3.3.1 Lift–off Mode ................................................................................. 23
3.3.2 Real Time Mode ............................................................................. 24
3.3.3 AFM/STM tracking mode .............................................................. 25
3.4 Quartz Tunning Fork Force Sensor for AFM Tracking Mode in SHPM .... 29
4 FABRICATION OF THE HALL SENSOR AND CONVENTIONAL
FABRICATION METHODS ......................................................................................... 37
4.1 Introduction ................................................................................................. 37
4.2 Nanofabrication Techniques ....................................................................... 38
4.2.1 Substrate Preparation...................................................................... 38
4.2.2 Optical lithography ......................................................................... 39
4.2.3 Electron Beam Lithography ........................................................... 44
x
4.2.4 Etching Process .............................................................................. 55
4.2.4.1 Wet Etching ..................................................................... 55
4.2.4.2 Dry etching ...................................................................... 58
4.2.5 Thermal Evaporation ...................................................................... 62
4.2.6 Dicing and Wire Bonding .............................................................. 63
5 GRAPHENE HALL SENSORS for SHPM ................................................................ 65
5.1 Introduction ................................................................................................. 65
5.2 Mechanic and Electronic Properties of Graphene ...................................... 67
5.3 Graphene Fabrication Methods ................................................................... 68
5.4 Characterization Methods of Graphene ...................................................... 71
5.5 Graphene Hall sensor .................................................................................. 73
5.5.1 Fabrication of GHP ........................................................................ 74
5.5.2 Characterization of GHP ................................................................ 78
5.5.3 Imaging NdFeB Demagnetized Magnet by GHP ........................... 83
5.6 Conclusion and Discussion ......................................................................... 85
6 BISMUTH HALL SENSORS for SHPM .................................................................... 88
6.1 Introduction ................................................................................................. 88
6.2 Bismuth Hall sensor .................................................................................... 90
6.2.1 Fabrication of Bismuth Hall sensor ................................................ 90
6.2.2 Characterization of Bismuth Hall Sensor ....................................... 94
6.2.3 Imaging NdFeB Demagnetized Magnet by Bismuth Hall Sensor . 96
6.3 Conclusion .................................................................................................. 99
7 CONCLUSION AND FUTURE WORK ................................................................ 100
8 BIBLIOGRAPHY .................................................................................................... 102
xi
LIST OF FIGURES
Figure 1.1: Schematic layout of the Scanning Probe Microscope ............................... 2
Figure 2.1: The principle of Scanning Tunneling Microscopy .................................... 6
Figure 2.2: STM topographic image of Si (111) under ultra high vacuum. ................ 7
Figure 2.3: Force as a function of distance between tip and sample .......................... 8
Figure 2.4: Hall effect .................................................................................................. 9
Figure 2.5: Schematic layout of the Scanning Hall Probe Microscope .................. ...12
Figure 3.1: Comparison of magnetic field sensitivity and spatial resolution of the
different imaging techniques. .................................................................. 14
Figure 3.2: Photograph of LT–SHPM used in the experiments ................................ 15
Figure 3.3: Detailed photograph of LT–SHPM ......................................................... 16
Figure 3.4: VHall Out vs Time graph of a 500 nm Bismuth Hall probe for –10 µA
Hall current. ............................................................................................. 17
Figure 3.5: (a) The illustration shows the Hall sensor and the sample mounted on the
slider puck while alignment process is carried out under optical
microscope (b) and tilted Hall sensor with respect to the sample surface
from different views ................................................................................ 19
Figure 3.6: Photograph of scanner and slider piezos in SHPM. ................................ 21
Figure 3.7: SHPM images of Bismuth substituted iron garnet thin film taken by
using real time scanning mode. Values above images represent applied
perpendicular magnetic field to bismuth substituted iron garnet thin film.
................................................................................................................. 25
Figure 3.8: Optical microscope image of the Hall sensor designed and fabricated for
STM tracking mode. ................................................................................ 26
Figure 3.9: (a) SEM images of the Hall sensor integrated AFM tip apex for different
magnifications. (b) High magnification SEM image of Hall cross located
at the end of the silicon cantilever (c) and (d) topography and magnetic
image of 40 MB computer hard disk respectively. ................................. 28
Figure 3.10: (a) The schematic illustrates top and side view of the cantilever which
consists of two sensors for both topography and magnetic imaging. (b)
SEM image of the fabricated sensor. Surface topography (c) and SHPM
image of array of NiFe rectangles. .......................................................... 28
xii
Figure 3.11: AFM image of Si (111) under ultra high vacuum by using a sharp tip
attached to QTF. ...................................................................................... 30
Figure 3.12: The pictures show process steps of fabrication of Hall sensor for AFM
tracking mode by using QTF. (a) Before (top) and after (bottom) QTF is
removed from its metal can. (b) The picture of PCB used throughout
measurements during this study. (c) and (d) pictures of QTF which is
glued on the top of PCB and Hall sensor attached to QTF respectively.
(e) Dedicatedly designed boxes for Hall sensor storage. ........................ 32
Figure 3.13: Picture of dialog box in SPM software for central frequency, quality
factor and maximum phase calculation. .................................................. 34
Figure 3.14: Resonance curve of QTF before (a) and (b) after Hall sensor attached to
QTF in ambient condition respectively. .................................................. 35
Figure 3.15: Resonance frequency curve of QTF attached GaN Hall sensor for one
prong free and both prongs free situations and for two different sizes of
quartz tubes. ............................................................................................ 36
Figure 4.1: The entire mask design which is used for Hall sensor fabrication .......... 40
Figure 4.2: Process flow of the photolithography for both negative and positive resist
................................................................................................................. 41
Figure 4.3: Process flow of the photolithography for both negative and positive resist
................................................................................................................. 42
Figure 4.4: Process flow of the EBL for both negative and positive resist tone. ...... 47
Figure 4.5: Schematic illustration of fabrication steps for bilayer system. ............... 48
Figure 4.6: (a) Screenshot of Hall sensor array designed in DesignCAD program. (b)
SEM image of hall sensor patterns after lift–off process. ....................... 50
Figure 4.7: (a) SEM image of hall sensor patterns arrayed by Run File Editor (b)
after lift–off process. ............................................................................... 51
Figure 4.8: (a) Screenshot of 100 nm Hall sensor designed in DesignCAD program
(b) SEM image of fabricated 100nm Hall sensor pattern. ...................... 51
Figure 4.9: (a) The optical microscope image of 50 nm Hall sensor after
development (b) Simulation of Hall cross pattern after implemented by
PEC.......................................................................................................... 53
Figure 4.10: (a) and (b) the SEM images of 50 nm Hall sensor at different
magnifications after 5nm Cr/30 nm Au evaporation and lift–off in
acetone. .................................................................................................... 54
xiii
Figure 4.11: The SEM image of 21 nm sized Hall cross by using 950 K PMMA–C2.
................................................................................................................. 54
Figure 4.12: (a), (b) The optical microscope and (c), (d) the SEM images of recess
formation. ................................................................................................ 57
Figure 4.13: SEM image of mesa formation on Bismuth Hall sensor. ........................ 58
Figure 4.14: (a) and (b) SEM image of Bismuth Hall sensor without and with oxygen
plasma application before e–beam resist coating respectively. .............. 59
Figure 4.15: The optical microscope image of GHP after mesa formation. ................ 60
Figure 4.16: The SEM images of Bismuth Hall sensor after mesa formation. ............ 61
Figure 4.17: The SEM images of 50 nm Bismuth Hall sensor (a) with and (b) without
evaporation of Cr as adhesion layer before Bismuth metal. ................... 63
Figure 4.18: Diced Bismuth Hall sensor fabricated on GaAs substrate. ..................... 64
Figure 4.19: The SEM image of diced and bonded Bismuth Hall sensor. .................. 64
Figure 5.1: The energy dispersion of graphene and the Dirac cone. ......................... 68
Figure 5.2: (a), (b) and (c) Graphene fabrication steps and (d) single and multi layer
graphene sheets produced by this method. .............................................. 69
Figure 5.3: Calculated optical contrast variation of graphene with changes in SiO2
thickness and of wavelength of light. ...................................................... 71
Figure 5.4: (a) Optical microscope image of graphene sheets produced via
mechanical exfoliation method. Raman spectrum of (b) single and (c)
multi layer graphene film. (d) AFM topography of single layer graphene
sheet. ........................................................................................................ 73
Figure 5.5: The optical microscope images of (a) mechanical exfoliated graphene (b)
after contact pad metallization, (c) O2 plasma etching of graphene (d)
taking contact from graphene to contact pads by implementing EBL and
(e) mesa step formation ........................................................................... 75
Figure 5.6: The optical microscope images of after (a) Hall cross definition by
implementing positive photolithography and (b) after O2 plasma process
and removing photoresist ........................................................................ 76
Figure 5.7: The optical microscope images (a) of the developed sample with
AZ726MIF after positive photolithography by using recess mask pattern
(b) after contact pads metallization ......................................................... 77
xiv
Figure 5.8: Optical microscope images of GHP. (a) After mesa formation by etching
830 nm SiO2/Si layer. (b) Side view of Hall sensor glued on 1 × 1 cm
PCB ......................................................................................................... 77
Figure 5.9: (a) The optical microscope images of as received CVD growth graphene
on 285 nm SiO2/Si wafers. (b) The Raman spectrum of as received CVD
growth graphene from different three points........................................... 78
Figure 5.10: Optical microscope image of GHP (left).Raman map of (I2D/IG ≥ 2)
intensity ratio measured in a 14 µm × 14 µm area of GHP (right–top).
Single Raman spectrum taken from center of the GHP (right–bottom). . 79
Figure 5.11: Hall voltage response of GHP to applied magnetic field at room
temperature .............................................................................................. 80
Figure 5.12: (a) Bmin for different drive current as a function of frequency at 300 K.
(b) Bmin for different temperature as a function of frequency for 5 µA
Hall current .............................................................................................. 81
Figure 5.13: Comparison of Bmin in vacuum and in Helium exchange gas for different
current values. (d) Serial resistance of two arms of Hall crosses as a
function of temperature ........................................................................... 82
Figure 5.14: The variations of RHall as a function of temperature for ± 2 µA driving
current..................................................................................................... 83
Figure 5.15: Measured resonance frequency at 300 K (a) before and (b) after GHP
was glued on QTF. .................................................................................. 83
Figure 5.16: (a) and (c) magnetic images of NdFeB demagnetized magnet by using
GHP at 300 K. (b) and (d) magnetic field variations along the line drawn
on images ................................................................................................ 84
Figure 5.17: (a) Topographic image of NdFeB. (b) 50 µm × 50µm LT–SHPM magnetic
image of NdFeB at 126 K. 14 µm × 14 µm LT–SHPM magnetic image of
the same sample for (c) IHall = +2 µA (d) IHall = –2 µA at 3 K ................ 85
Figure 5.18: SEM image of as received CVD graphene sheet which shows cracks and
discontinuity on the graphene film surface ............................................. 86
Figure 6.1: The SEM images of 50 nm × 50 nm nano-Bismuth hall probe. SHPM
image of crystalline Bismuth substituted iron garnet thin film taken by
fabricated 50 nm Bismuth Hall sensor at room temperature ................... 89
Figure 6.2: The optical microscope image of (a) symmetric (b) asymmetric ohmic
contacts after 10 nm Cr and 100 nm Au evaporation and lift–off process
................................................................................................................. 90
Figure 6.3: Then SEM images of 50 nm Bismuth Hall cross at different magnifications
................................................................................................................. 92
xv
Figure 6.4: (a)The optical microscope images after the first time EBL is employed for
connection problem (b), (c) and (d) show SEM images after third EBL
process applied at different magnifications and different points of view
................................................................................................................. 93
Figure 6.5: The SEM images of 50 nm Bismuth Hall sensor after mesa formation ...... 94
Figure 6.6: (a) The SEM image of 500 nm Bismuth Hall sensor. Bmin as a function of
driving current at different temperature values (b) 300 K (c) 77 K and
(d) 4 K The SEM images of 50 nm Bismuth Hall sensor after mesa
formation ................................................................................................. 95
Figure 6.7: (a) The SEM image of 200 nm Bismuth Hall sensor Bmin a function of
driving current at different temperature values (b) 300 K (c) 77 K and (d)
4 K ........................................................................................................... 95
Figure 6.8: SHPM images of NdFeB demagnetized magnet at (a) 300 K, (b) 77 K, (c)
4 K taken by using 500nm Bismuth Hall sensor for 50 µA driving current
and at (d) 4 K for –50 µA driving current. .............................................. 96
Figure 6.9: SHPM images of NdFeB demagnetized magnet at 300 K (left) by using
200 nm Bismuth Hall sensor for 500 µA drive current and graph of cross
sections (right top and bottom). .............................................................. 97
Figure 6.10: (a) Topography and (b) SHPM image of NdFeB sample at 77 K by using
200 nm Bismuth Hall sensor for 500 µA drive current ........................... 97
Figure 6.11: (a) Topography and (b) SHPM image of NdFeB sample at 4 K by using
200 nm Bismuth Hall sensor for 500 µA drive current ........................... 98
Figure 6.12: (a) SHPM image of NdFeB sample at 300 K by using 100 nm Bismuth
Hall sensor for 500 µA drive current. (b) Magnetic field variations along
the line drawn on images ......................................................................... 99
xvi
LIST OF TABLES
Table 4.1: Summary of the positive lithography process parameters optimized for
two different mask aligner systems are given ......................................... 43
Table 4.2: Summary of image reversal process parameters optimized for two
different mask aligner systems are given ................................................ 43
Table 5.1: Summary of data measured from InSb, epitaxial graphene and CVD
graphene sensor for 5 µm Hall cross size................................................ 66
xvii
LIST OF ABBREVIATIONS
MO : Magneto Optics
QTF : Quartz Tuning Fork
ADC : Analog to Digital Converter
VTI : Variable Temperature Insert
PCB : Printed Circuit Board
CBT : Cantilever Beam Theory
SHM : Simple Harmonic Motion
EBL : Electron Beam Lithography
CD : Critical Dimension
ACE : Acetone
IPA : Isopropyl Alcohol
DI : Deionized
PAC : Photoactive Compound
UV : Ultraviolet
SEM : Scanning Electron Microscopy
PMMA : Polymethyl Methacrylate
HSQ : Hydrogen Silsesqioaxene
Fox 12 : Flowable Oxide
MIBK : Methyl Isobutyl Ketone
NPGS : Nabity Pattern Generation System
EHT : Extra High Tension
PEC : Proximity Effect Correction
PSF : Point Spread Function
RIE : Reactive Ion Etching
xviii
GHP : Graphene Hall Probe
SHPM : Scanning Hall Probe Microscopy
SPM : Scanning Probe Microscope
LT–SHPM : Low Temperature Scanning Hall Probe Microscopy
FET : Field Effect Transistor
LEED : Low–Energy Electron Diffraction
TEM : Transmission Electron Microscope
ICP–RIE : Inductively Coupled Plasma Reactive Ion Etching
1
CHAPTER 1
INTRODUCTION
1.1 Context and Motivation
The development of Scanning Probe Microscopy (SPM), which can achieve
atomic scale resolution, has been pioneered by the invention of Scanning Tunnelling
Microscopy (STM) in 1981 by Binnig & Rohrer 1. In STM we use an atomically sharp
tip which is kept at very close proximity of the surface by regulating the distance to
keep a constant tunnel current, in the nA range. The image is formed by scanning the tip
across the surface by means of piezolectric scanners, as we record the tip height as
shown in Figure x. The basic working principle of all members of SPM family is similar
to each other. In short we scan a probe at close proximity of the specimen surface, while
e measure the interaction between the tip and sample: this interaction can be chemical
force, magnetic force, electric current etc. One can obtain information on various
properties of samples like surface morphology, topography, magnetic properties etc, by
utilizing different interaction modes between the sample and the tip. The Scanning Hall
Probe Microscopy (SHPM) uses a micro or nano Hall sensor to image the z–component
of local magnetic flux density, Bz, down to 50nm resolution. The main content of this
thesis covers development of new graphene and Bismuth micro and nano Hall sensors
for SHPM.
2
Figure 1.1: Schematic layout of the Scanning Probe Microscope.
The basic operating principle of SHPM is to provide information on magnetic
properties of materials by measuring classical Hall effect. The resolution of SHPM is
determined by the Hall sensor which is manufactured using advanced microfabrication
techniques. Based on the necessities, the probe of the Hall sensor can be fabricated at
different sizes, from different materials. In order to increase the magnetic field
resolution, high mobility materials with low charge carrier concentrations are preferred
2. Moreover, the spatial resolution of the microscope is defined by the
physical dimensions the Hall sensor 3. Various materials have been utilized for the
fabrication of Hall sensors for different applications and temperatures, so far such
as InSb, GaAs/AlGaAs, Silicon–on Insulator, AlGaN/GaN etc 4, 5
. In addition to these
materials, the experimental isolation of graphene made it a new candidate for
fabricating high sensitivity Hall sensors. Graphene is a promising material thanks to its
extraordinary mechanical and electrical properties 6. In literature it has
been indicated that minimum detectable magnetic field of epitaxial–grown graphene is
better than InSb Hall sensor; which has been shown as a low noise and high
3
performance Hall probe at room temperatures 7. This fact shaped the motivation of this
thesis; to use graphene as Hall sensor material for magnetic imagining and also increase
magnetic field resolution and hopefully decrease the size for smaller than 50 nm spatial
resolution. In addition to graphene, bismuth is also an alternative material for SHPM
application thanks to its low charge carrier concentration. Bismuth Hall sensor
fabricated by using focused ion beam (FIB) system is damaged by Ga+
and this results
in having hall sensor with poor performance 8, 9
. It is reported that minimum detectable
magnetic field of Bismuth Hall sensor is improved further by implementing electron
beam lithography in the fabrication process 10
. In this thesis work, we also attempted to
reduce the Bismuth Hall sensor area, in order to obtain higher spatial resolution and
implement electron beam lithography (EBL) in order to prevent damage arising from
FIB technique.
1.2 Structure of the Thesis
In Chapter 1, motivations of this study are underlined. Chapter 2 presents a
general introduction about scanning probe microscopy and provides detailed
information on both theoretical aspects and working principles of different types of
SPM, such as Atomic Force Microscopy (AFM), STM, Magnetic Force
Microscopy(MFM) and SHPM. In particular, this chapter addresses the well–known
phenomena called as classical Hall effect and its utilization in SHPM applications.
Chapter 3 covers the basic principles of SHPM and its components, microscope
design, electronics, operation principles and explain the types of scanning modes that
we used. Among the scanning modes, extra effort was spent to explain the quartz tuning
fork AFM tracking mode.
In Chapter 4, we mainly discuss fabrication of Hall probes which is the most
crucial part that determines the resolution of SHPM. Fabrication techniques used for
Hall sensor production, problems encountered during fabrication process and their
solutions are presented with using SEM and optical microscope images.
4
In Chapter 5 and Chapter 6 present fabrication, characterization, and magnetic
imaging of NdFeB demagnetized magnet by using Graphene Hall sensor and Bismuth
Hall sensor respectively. In chapter 7, we discussed our results and future work.
5
CHAPTER 2
SCANNING PROBE MICROSCOPY AND MAGNETIC IMAGING
2.1 Scanning Probe Microscopy
Scanning probe microscopy (SPM) is a commonly used imaging technique which
provides local properties of an investigated sample surface down to atomic scale. Most
SPMs use an extremely sharp tip which is brought into close proximity of a sample in
order to measure interaction between tip and sample. These interactions are used as a
feedback signal and define type of SPM. Sample is typically scanned by piezoelectric
crystals; which also controls tip–sample distance with high precision. There are various
types of SPM. In the following sections, first STM and AFM will be presented and then,
working principle of the most prominent magnetic imaging techniques and SHPM will
be covered.
2.1.1 Scanning Tunneling Microscopy
SPM has provided scientists to characterize surface of a sample at atomic scale.
Before the 1980's, it was not possible to image of a material in the atomic scale.
However, in 1981 Gerd Binnig and Heinrich Rohrer invented STM, which could image
single atoms, at IBM Laboratories in Switzerland 1. This invention brought Nobel Prize
in 1986 to its inventors and contributed to the development of SPM.
In STM, sharp conductive tip, having a single atom at its closest point of surface,
is brought into close proximity with the conductive sample surface. Although tip and
sample are very close to each other, they are not in physical contact. When a voltage is
6
applied between tip and sample, electrons start tunneling from sample to tip or the other
way around depending on the polarity of voltage; which determine the direction of
tunneling current. Electron tunneling is at the heart of STM which is one of the
fundamental phenomena in quantum mechanics. In classical physics, it is not possible
for a particle to penetrate a potential barrier if the total energy of the particle is less than
the barrier height. However, in quantum mechanics, this particle can go through the
potential barrier and this is known as quantum mechanical tunneling 11. The tunneling
current is an exponential function of distance and it is given by
. (2.1)
. (2.2)
where m is the electron mass, is the average work function of the tip and sample, is
the Planck’s constant, VBias is the applied bias voltage between tip and specimen and z
is the tip–sample distance (Figure 2.1) 12, 13
.
Figure 2.1: The principle of Scanning Tunneling Microscopy 13
.
7
Distance between tip and sample can be controlled by measuring tunneling
current and if tunneling current is fed to a control circuit, it enables to obtain
topographical image of a conductive sample with atomic resolution.
Figure 2.2: STM topographic image of Si (111) under ultra high vacuum.
2.1.2 Atomic Force Microscopy
Even though STM provides high resolution imaging, it only allows us to examine
conductive samples. AFM was developed by Binnig et al in 1986 14
to overcome this
drawback. One of the most important advantages of AFM is that it enables to analyze
insulator samples as well as conductive materials .
AFM is sensitive to interaction between an extremely sharp tip at the end of the
micro–fabricated AFM cantilever and the atoms located on the sample surface. There
are basically two types of tip–sample interactions known as attractive and repulsive
forces. van der Waals interaction, chemical force, capillary and electrostatic forces are
considered as attractive forces while repulsive forces are hard sphere repulsion, Pauli
exclusion interaction and electron–electron Coulomb interaction 15
. Force as a function
of distance between the tip and sample is shown in Figure 2.3 16
. As is seen from the
figure, when tip approaches toward the sample, first of all, attractive forces are
dominated. At large distance, main component of attractive forces is the van der Waals
interactions. These forces result from fluctuations in electric dipoles of atoms and
molecules 17
.
8
Figure 2.3: Force as a function of distance between tip and sample 16
.
When the surface topography of sample changes, interatomic forces between tip
and the sample surface also differ and thus result in cantilever deflection. By monitoring
the deflection of cantilever with optical detection system, topographic images of both
conductive and nonconductive samples are obtained.
2.1.3 Magnetic Imaging Techniques
2.1.3.1 Magnetic Force Microscopy
Different than Atomic force microscopy, Magnetic force microscope (MFM) uses
a sharp magnetic tip to scan a sample with magnetic domains and provides information
regarding its magnetic structure. Attractive or repulsive magnetic forces between the
sample and magnetic tip attached to a flexible microcantilever are measured to obtain
magnetic domains on the specimen. The basic principles of MFM to obtain an image are
as follows:
Magnetic tip is first used in tapping mode to map the topography of
sample’s surface.
Surface is tracked from a certain distance above the surface in lift mode.
Image is produced by the interactions between sample’s surface magnetic
field and magnetic AFM tip.
9
MFM measures the vertical gradient of magnetic force that is established between
magnetic tip and the sample surface. Derivative of the vertical component can either be
measured by slope or modulations in frequency. Sensitivity of MFM therefore mainly
depends on minimum detectable force gradient 18
.
Although MFM’s resolution is dependent on various factors, among them the
most important ones are perhaps the tip size & geometry and sample tip spacing.
Today’s MFM devices are capable to resolve magnetic features up to 10–50 nm
depending on the environmental conditions 19
. However, the demand for improved
MFM resolutions is still attracting researchers to be able to resolve smaller features. For
instance, Han et al. Fabricated a dual–synthetic MFM tip to improve spatial resolution
and compared the images with those obtained by conventional MESP tip and confirmed
superior performance of dual–synthetic tips. The samples were scanned on a 10 μm ×
10 μm area under ambient conditions in tapping/lift mode 20
.
2.1.3.2 Scanning Hall Probe Microscopy
Scanning Hall Probe Microscopy (SHPM) is based on Hall effect principle which
was discovered by Edwin Herbert Hall in 1879. He has found that if a current carrying
conductor or semiconductor is placed in a magnetic field at right angles, a voltage is
created in the perpendicular direction to both to direction of current and the magnetic
field. In Hall Effect, as shown in Figure 2.4, a semiconductor material with dimension ℓ
× w × d is subjected to an electric current in the positive × direction.
10
Figure 2.4: Hall effect.
If a material is n type semiconductor, the great part of charge carriers will be
electrons which move opposite direction of applied electric current. In the presence of
external magnetic field, electrons with drift velocity Ve experience Lorentz force given
by the equation;
. (3.2)
Under influence of Lorentz force, moving electrons are deflected in negative y
direction which makes lower edge of semiconductor negatively charged while upper
edge becomes positively charged. Thus, electric field Ey is generated in the
semiconductor and this field applies a force on electrons given by,
. (3.2)
When the system reaches the equilibrium, the electric force becomes equal to the
magnetic force,
(3.3)
(3.4)
Thus, the electric field can be found by using equation 3.4,
(3.5)
If total charge in this slab is Q, then free carrier density n is given by,
(3.6)
Moreover, current is defined as the amount of charge that flows per unit time. By
combining equation 3.6 with current, one can find current as,
(3.7)
11
If we rearrange equation 3.7 and substitute drift velocity in equation 3.5, electric
field is given by,
(3.8)
It is also possible to write electric field in terms of hall voltage,
(3.9)
By combing equation 3.8 and 3.9, Hall voltage VH is given by,
(3.10)
where RH is equals to
and it is called as Hall coefficient. If charge carriers in
semiconductor are holes, Hall coefficient RH is given by
. In SHPM, by
measuring Hall voltage, knowing applied current and external magnetic field, Hall
coefficient can be calculated easily. Moreover, since sign of Hall coefficient changes
according to the sign of current carrying charges, this enable to determine the type of
semiconductor used in the experiment. On the other hand, if current is through a
semiconductor whose Hall coefficient is already known, magnetic field coming from
this material can be determined by measuring Hall voltage.
In SHPM, distance between hall probe and surface of a sample is a critical
parameter. Since magnetic field decreases with distance, distance between hall probe
and surface must be very close to each other in order to measure small magnetic field 21
.
When a hall probe is brought proximity of a sample and applied a current through two
arms of hall probe, a fabricated Hall probe being a sensitive to perpendicular component
of stray magnetic field produced by sample creates a Hall voltage. This Hall voltage is
measured from other two arms of the Hall sensor. Therefore, unknown magnetic field
coming from sample can be determined. Resolution of this magnetic field is limited
dominantly by Johnson noise due to thermal agitation 3.
12
Figure 2.5: Schematic layout of the Scanning Hall Probe Microscope 2.
13
CHAPTER 3
SCANING HALL PROBE MICROSCOPE
3.1 Introduction
Magnetic domain is a small region of magnetic moments where all magnetic
moments are aligned in the same direction. Magnetic domains structure plays a very
important role in development of magnetic storage technology. Each domain
corresponds to a bit and recorded in a magnetic medium. Size of the bit determines the
storage capacity and it needs to be decreased in order to increase density of recording.
Size and various properties of magnetic domain structure can be investigated and well
understood by using different imaging methods at nano–scale such as MFM, Magneto
Optics (MO), SQUID, and SHPM. Besides, these techniques are not only used for
observation of magnetic domains in magnetic recording media but also vortices in
superconductive materials or any other magnetic properties in magnetic materials. Some
of the common important features of magnetic imaging methods can be listed as
follows: quantitative measurements, fast scan speed, high magnetic field sensitivity and
high spatial resolution.
Both magnetic field sensitivity and spatial resolution are the most critical
parameters for magnetic imaging application and Figure 3.1 illustrates the comparison
of these parameters for different magnetic imaging tools 22
. As seen from this figure, the
highest magnetic field sensitivity can be obtained by SQUID despite the fact that its
spatial resolution is limited around 5–10 µm. On the other hand, MFM provides higher
14
spatial resolution than SQUID and it can be performed at different measurement
condition such as in vacuum condition, which can further improve the sensitivity and at
ambient temperature as well as low temperature 23
. Together with its benefits, MFM has
much poorer magnetic field sensitivity as it is compared to SQUID and SHPM. Among
them, SHPM offers reasonable spatial resolution and magnetic sensitivity at the same
time. By using SHPM, it is also possible to get topographic image of investigated
sample simultaneously. Last but not least, it yields quantitative measurements without
damaging sample surface at both room and low temperature.
Figure 3.1: Comparison of magnetic field sensitivity and spatial resolution of the
different imaging techniques 22
.
In this chapter, all mechanical and electrical components of SHPM will be
discussed in detail. Two cryostat systems are used in order to obtain local magnetic
measurement of a sample and electrical characterization of the fabricated hall sensor in
a wide range of temperature during this work will be also introduced. In the third part of
15
this chapter, all scanning modes of SHPM are explained. Last but not least, theoretical
and experimental aspects of Quartz Tuning Fork (QTF) as force sensor are presented
3.2 Scanning Hall Probe Microscopy
3.2.1 The Microscope Design and Operation
As with other SPM, SHPM is composed of five main parts; the probe,
approaching system, scanning system, electronic controller unit and the software 3.
Figure 3.2 shows the LT–SHPM system manufactured by NanoMagnetics Instruments
Ltd. The microscope insert consists of the LT–SHPM head, the flange and radiation
baffles. All the electrical wires coming from the SPM electronic pass through the
cylindrical tube and they transmit signal from electronic unit to the sensor. Length of
the microscope is adjusted so as to bring the sensor at the center of magnetic field when
it is inserted to the cryostat. Furthermore, there are 6 radiation shields at certain
distances surrounded the microscope body, which reduces heat load coming from top
part of the microscope.
Figure 3.2: Photograph of LT–SHPM used in the experiments 24
.
16
The microscope shield shown in Figure 3.3 protects the SHPM head and its
components from mechanical damages and it also maintains temperature uniformity.
After making sure that the sensor works properly, the shield can be fixed to the head of
SHPM and the microscope can be loaded the cryostat safely. Moreover, a coil which
can apply a constant small magnetic field can be integrated the shield. It allows quick
measurement of the Hall coefficient of the sensor and gives an idea about performance
of the sensor used before the microscope is loaded to a cryostat.
Figure 3.3: Detailed photograph of LT–SHPM 24
.
Checking whether the fabricated hall sensor works properly or not is highly
recommended before mounting the sample to the microscope or before making any
further steps. For this, the Hall sensor is driven by a constant Hall current value and
nulled by using SPM software in order to reduce offset voltage in the absence of the
magnetic field. Then, a permanent magnet is approached to the sensor. Therefore,
increase or decrease in Hall voltage output (VHall Out) signal can be observed in the
SPM software by connecting it to one of the channel located in the spare Analog to
Digital Converter (ADC). If opposite magnet field direction is applied to the Hall
sensor, VHall Out signal is reversed. In addition to this, opposite direction of current
result in changing polarity of VHall Out signal as well for the proper working Hall
sensor. VHall Out signal of 500 nm Bismuth Hall probe versus time for –10 µA current is
17
shown in Figure 3.4. There is offset voltage after nulling the sensor. When the magnet is
approached initially to the sensor, VHall out is increased as seen from this figure. Since
the polarity of magnet is reversed around the fifth second, VHall is increased in the
opposite direction with almost same magnitude on the graph.
Figure 3.4: VHall Out vs Time graph of a 500 nm Bismuth Hall probe for –10 µA
Hall current.
The sample to be measured must be as clean as possible to obtain better
experimental data. For STM feedback the sample must be conductive. On the other
hand, both conductive and non conductive sample can be measured by using AFM
tracking mode. Silver paint is used in order to glue the sample on the sample holder.
After gluing the sample, the holder is attached to the slider puck. There are three quartz
hemispheres on the XY slider. They decarese the frcition and therefore they provide
easier movements of the sample holder. After the sample and the sensor are attached
properly to the microscope, the slider puck is mounted to the quartz glass tube by using
a leaf spring and two screws as seen from Figure 3.3. Mounting the sample slider puck
to the quartz tube is a crucial step. Excessive force should not be applied to in order to
avoid damaging to the glass tube and the leaf spring must be tightening at right amount.
On the other hand, if the screws on the leaf spring are too loose the puck can fall down.
18
Piezoelectric materials are commonly used in nanotechnology. In particular, they
are fundamental tools in SPM technology due to its precise movement of both scanning
and tip–sample approaching mechanism. Furthermore, piezoelectric materials are
suitable for low temperature applications which expand their area of usage. They
basically convert the mechanical energy to electrical energy or vice versa. When a
voltage is applied to a piezoelectric material, it changes its size by extending or
contraction according to polarity of applied voltage. Piezoelectric materials are widely
preferred for approaching system due to its accurate movement. Since they move in
micron scale, an approach mechanism which provides greater steps is required for
course approach process. In SHPM, the sample is approached to the sensor by using
stick/slip coarse approach mechanism, after the sample and the sensor are fixed on the
microscope. In this mechanism, first a voltage with slow ramp is applied to slider piezo
which results in slow movement of the slider puck in forward direction together with
the piezo tube then this voltage value is turned back to its initial value very fast, which
causes to move quartz tube in opposite direction while the slider puck stays stationary.
Consequently, the slider puck is moved in desired direction. Moreover, this saw–tooth
voltage signal applied to the piezo tube can be set up to 400 V. Step size of movements
can be adjusted by changing voltage values although step size are not reproducible for
the same voltage value at each approaching application 3.
After the sensor is brought to the sample surface close enough, first of all the
sample is aligned parallel with respect to the Hall Sensor by using three compression
springs under a light microscope. Alignment is carried out at a safe distance and it is
completed when the Hall sensor and its reflection on the sample surface is exactly
parallel to each other. Alignment accuracy of the sample can be easily understood by
bringing the Hall sensor very close to the sample and checking it from different sides by
using bottom illumination of the optical microscope. Next, one of the compression
springs located back side of the sensor is loosen around one and a half tour so that edge
of the sensor becomes closest point to the investigated sample. Angle between edge of
the Hall sensor and the sample surface should be between 1–1.5° for precise and high
resolution data acquisition. Small number of step sizes has to be used while the piezo
tube is moved in z direction. Besides, before the sample is tilted, the sensor must be
retracted to a few hundred microns away in order to avoid the Hall sensor crashing into
19
the sample surface. The picture of the hall sensor and the sample mounted on the slider
puck during these processes is presented in Figure 3.5 (a). Figure 3.5 (b) and (c)
illustrates tilted Hall sensor with respect to the sample surface from different views.
After the sample is tilted around 1.5° with respect to the Hall sensor, the sample is
ready to approach. For this, generally automatic approach is preferred. When the
automatic approach is used to bring the sensor close the sample surface, at each step
piezo first fully retracted and check the set interaction value which can be either
tunneling current or frequency shift depending on feedback mechanism used and then
the piezo is extended to a few steps. Automatic approach process continues until
interaction value reaches the set value.
20
Figure 3.5: (a) The illustration shows the Hall sensor and the sample mounted on
the slider puck while alignment process is carried out under optical microscope (b) and
tilted Hall sensor with respect to the sample surface from different views front and (c)
side.
There are three piezos fixed on the microscope. The first one, slider tube piezo, is
used for approaching as it is mentioned above and the second piezo is used for scanning
called as scanner piezo. Piezo tube consists of distinct electrodes which allows
movement in three direction x, y and z with high precision. Expansion or contraction in
desired direction is achieved by giving voltage to the appropriate electrodes. Voltage
values and thus movement of the scanner is well controlled by using SPM software and
electronics in SHPM. Different piezos respond differently to the applied voltage based
on its sensitivity. The sensitivity of piezoelectric material is defined by displacement of
the piezo per applied voltage and it is directly affected by thickness of piezo tube,
temperature condition etc. Furthermore, piezoelectric materials exhibit hysteresis effect.
It means that there is a nonlinear relationship between applied voltage and output
displacement. This can result in image distortion. In addition to hysteresis and
sensitivity, another important issue to be considered is behavior of piezoelectric
materials at cryogenic temperature. Electronic properties of the Hall sensor such as
mobility, conductivity or charge carrier concentration depends on temperature and also
field sensitivity of the Hall probe varies with changing temperature. Therefore, low
temperature measurements play important role in magnetic imaging. Temperature
dependency of both hardware and software components of the microscope have to be
taken into account. For instance, calibrated piezo constants should be entered the SPM
software control according to temperature value before the measurements take place,
because the piezo coefficients are dependent directly on the temperature. They decrease
as the temperature is lowered. If the piezo coefficients are not entered the software
correctly, the size of scan area or step size as the sample is approached to the Hall
sensor differs from their actual value. Maximum scan range is also having temperature
dependency and it is extended with increasing temperature. For the microscope we used,
the maximum scan range at room temperature is 108x108 µm2
while this value is 18x18
µm2 area at 4 K. Scan speed, size of scan area and number of scans are adjustable
parameters. Scanning is initiated from SPM software program and it can be stopped
21
whenever it is required. Scanning modes and feedback mechanism will be discussed
later in this chapter.
Figure 3.6: Picture of scanner and slider piezos in SHPM.
We used two different commercially available cryostats for the low temperature
measurement and high magnetic field application during this study. One of them is,
Teslatron PT, manufactured by Oxford Instruments Nanoscience and located in
SUNUM at Sabancı University. It is a cryogen free system whose temperature can be
controlled between 1.5–300 K. A superconductive magnet integrated the system
generates high magnetic field in both direction up to 8 T with maximum field ramp rate
0.207 T/min. Besides, the other cryostat system, supplied by Cryogenic Limited, located
at NanoMagnetics Instruments Ltd., is used mostly for magnetic imaging and
characterization of the fabricated Hall probes. It is also a closed cycle cryogen free
cryostat and any temperature between 1.6 K and 300 K can be stabilized easily.
Although the systems specification allows changing the sample rod at any temperature,
the microscope must be loaded or removed from the cryostat when the temperature of
the variable temperature insert (VTI) sample space is at 300 K to prevent the
22
microscope from thermal shocks. The rate of temperature must be adjusted between 2–3
K/min in order to protect components of the microscope from mechanical damage when
it is cooling down or warming up to a desired temperature. Furthermore, not only
measurements can be obtained in He exchange gas; but also the sample space can be
pumped up to 10-5
mbar vacuum level by turbo pump and magnetic imaging or electrical
characterization can be performed in vacuum environment. This comes into prominence
when working with the graphene hall sensor.
Last but not least, performance of the fabricated sensor is determined initially
with the help of a test station manufactured by NanoMagnetics Instruments Ltd. Since it
does not include essential mechanical components for scanning such as scanner and
slider piezo, it is not suitable for magnetic imaging. However, it is designed to measure
Hall voltage response, Hall coefficient and noise spectrum of the sensors one after
another by connecting to head amplifier to the appropriate test station head as it is
possible to mount two or three sensors to test station at the same time. The benefit of
using the test station is achieved by that it enables to select the most promising sensor
according to electrical characterization results among the mounted sensors before
scanning step. Besides, it also allows doing measurements in vacuum environment and
in a wide range temperature including cryogenic temperature.
3.2.2 LT–SHPM Electronics
There are various electronic cards which are used for different purpose and
operation in the LT–SHPM control electronics. Some basic properties of these cards are
given as following:
a) POWER SUPPLY CARD
b) DAC CARD
c) CONTROLLER CARD
d) SLIDER CARD
e) HEAD AMPLIFIER
f) PHASE LOCKED LOOP CARD
g) SPARE ADC
h) SCAN DACS
23
i) HIGH VOLTAGE AMPLIFIER
j) CONTROLLER CARD
k) HALL PROBE AMPLIFIER CARD
l) MICRO CONTROLLER AD CARD
3.2.3 LT–SHPM Software
NMI SPM program (version 2.0.16) written by Nanomagnetics Instruments Ltd is
used in LT–SHPM measurement in order to adjust related parameters with the
experiments. Before performing scanning or any other steps, suitable scanning mode
must be selected and other commands like approaching, scanning or finding resonance
frequency of cantilever is set in this program. Besides, the software makes it possible
that the collected data can be edited. For example; obtained images can be corrected by
using plane or line correction. In addition, basic process like cropping image, measuring
distance between two features and measuring roughness can be done via NMI SPM.
Various filtering are available in this program such as mean, median, Gaussian or
parametric low past filter. There are also another program written by NanoMagnetics
Instruments Ltd exists for data editing called NMI SPM which allows the processing of
collected images.
3.3 Scan Modes of LT–SHPM
LT–SHPM can be operated with different modes by mounting appropriate tip to
microscope according to type of modes. Scanning mode of SHPM is preferred by
considering the sample type, required resolution or image acquisition time. There are
three different scanning modes of SHPM: tracking mode with STM or AFM feedback,
lift–off mode, real time mode. Each mode has some advantages and disadvantages on its
way. Details of these modes will be discussed in the following subsections.
3.3.1 Lift–off Mode
Lift–off mode is widely used method in SHPM due to its several advantageous.
The greatest benefit of the lift–off mode is being very fast since it does not use any
feedback mechanism. Scan speed of this mode is approximately 4s/frame 2. Moreover,
24
with this mode flicker noise (1/f noise) can be reduced by taking average of images
obtained 3.
For the lift–off mode operation, first of all the sample–sensor angle is adjusted
and then the sample is approached to the sensor by using STM feedback control. As the
desired tunneling current is established, approaching stops and initially area which will
be scanned must be checked to be sure that it has free of contaminations. For this
purpose, the sensor is lifted off a few microns away from the sample surface and the
sensor is moved to four edges of the scan area to confirm that there is no tunneling
current between the sample and the corner of Hall sensor. If the tunneling current is
observed in any corner, it means there is an obstacle on the sample surface and lift–off
distance must be increased in order to avoid crushing the sensor to the sample surface.
In this case, after increasing distance between the sample and the sensor, corner
checking process should be repeated. When ascertained that there is no contamination in
the scanning area, scanning is performed at certain height from the sample surface. One
of the drawbacks of this mode is that only magnetic imaging can be obtained, since
there is no feedback mechanism it is not possible to get topography image in this mode.
Besides this, as distance between the sample and the Hall sensor is a bit much compared
the other modes such as AFM or STM tracking modes, both spatial and magnetic
resolution are slightly decreased 3, 21
.
3.3.2 Real Time Mode
Another mode used in SHPM application is known as real time scanning mode.
Real time scanning mode does not use any feedback mechanism as in the case of lift–off
mode and scanning is completed with very fast scanning speed. The scan speed of the
real time is reported as 1 s/frame. The major difference between lift off and real time
mode is that obtained image is displayed once scanning is finished 3. Figure 3.7 shows
magnetic images of bismuth substituted iron garnet thin film under different external
magnetic field by using real time scanning mode in SHPM 25
. It is reported that these
images are taken less than 10 s with this mode.
25
Figure 3.7: SHPM images of Bismuth substituted iron garnet thin film taken by
using real time scanning mode. Values above images represent applied perpendicular
magnetic field to bismuth substituted iron garnet thin film 25
.
3.3.3 AFM/STM Tracking Mode
In STM tracking mode, firstly tilted sample is approached to the sensor via stick–
slip approach mechanism until the desired tunneling current flows between the sample
and the sensor. Tunneling current is measured at individual points on scanning area and
this current value is then sent to the SPM electronics as feedback so that the distance
between sensor and the sample is maintained at certain height throughout scanning
process 3. Typical tunneling current value used in STM tracking mode is around 0.1nA.
In the fabrication process of the Hall sensor, a metal structure (mostly gold is
preferred for metallization) is defined by optical lithography which will be used later as
STM tip for STM tracking mode. Figure 3.8 shows optical microscope image of the
Hall sensor which is patterned by photolithography and STM tip located left side of the
Hall cross 2.
26
Figure 3.8: Optical microscope image of the Hall sensor designed and fabricated
for STM tracking mode 2.
In STM tracking mode, both topography and magnetic image can be obtained
simultaneously by choosing related image channel from SPM software. Distance
between the sample and the sensor is reduced in this mode and therefore high magnetic
field resolution can be obtained. One of the disadvantages of this mode is that only
conductive samples can be measured. Moreover, scanning speed is much slower in this
mode as compared the lift–off or real time mode since feedback mechanism is active
during scanning.
The other scanning method which uses feedback mechanism is AFM tracking
mode. The major benefit of this mode is that not only conductive samples but also non–
conductive samples can be investigated. It also provides high magnetic field resolution
as in the case of STM tracking mode since the tip can be brought very close to the
sample surface. In this mode, force between tip and sample can be measured with
various detection methods such as optical and piezoresistive detection systems.
Therefore, tip–sample separation can be precisely controlled 26
.
Operation of AFM tracking mode is very similar to that of STM tracking mode
with some exceptions. In AFM tracking mode, the sample is aligned parallel with
respect to Hall sensor as in the STM tracking. Then, it is pulled back to a safe distance
27
in order to not to crush the sample to the sensor when the sample is tilted. As tilted
sample is approached to the sensor with stick–slip mechanism, resonance frequency of
the sensor shifts because of the tip–sample interactions and shift in the resonance
frequency is used as feedback instead of tunneling current in this mode.
There are two ways to produce sensor for AFM tracking mode. In the first way, an
AFM tip and Hall cross are fabricated in a single cantilever with different designs 27, 28,
29. Figure 3.9 (a) and (b) show the SEM images of such cantilever. In this work, they
fabricate the Hall cross on an AFM tip apex as is shown in bottom and left view in same
Figure 30
. They begin fabrication with defining 20 µm high pyramid of the cantilever by
wet etching process and then they deposited silicon nitride by using plasma–enhanced
CVD in order to provide electrical isolation between silicon and Hall cross region.
Finally, they patterned 400 nm Bismuth Hall cross structure on the center of the
pyramid by EBL and lift off process. Resonance frequency of this cantilever is
measured as 157 Hz. Figure 3.9 (c) and (d) show surface topography and magnetic
imaging results taken by using this cantilever 30
.
Alternatively, piezoelectric materials can be used for designing sensors which
have both magnetic sensing and distance control elements in a single cantilever.
However, piezoelectric material with a large piezoelectric coefficient and high mobility
should be preferred for better scanning results of topography and magnetic imaging. For
this purpose, n–Al0.4Ga0.6As is used in a few of the experimental works 29, 31
. Al
component in this alloy ensures that material have a higher piezoelectric coefficient.
Side and top views of such a cantilever is illustrated in Figure 3.10 (a) 31
. As is clearly
seen from SEM image, Hall cross and sharp AFM tip with 2–2.5 µm height is a few
micron away from each other in this cantilever (Figure 3.10 (b)). In this work, they
reported longitudinal piezoelectric coefficient of the material in [011] direction is 1.35 ×
10-9
Pa-1
at room temperature. They also measured quality factor and resonance
frequency of the sensor between 300–500 and 18–21 kHz respectively in environment
condition while quality factor increased to over 10,000 under a vacuum of
approximately 10-5
mBar. Besides, spatial resolution of the sensor is less than 1.5 µm
and Hall coefficient is measured as 3×103 Ω/T. Magnetic imaging and topography of
NiFe rectangles taken by this cantilever are given in Figure 3.10 (c) and (d) 31
.
28
Figure 3.9: (a) SEM images of the Hall sensor integrated AFM tip apex for
different magnifications. (b) High magnification SEM image of Hall cross located at the
end of the silicon cantilever (c) and (d) topography and magnetic image of 40 MB
computer hard disk respectively 30
.
(c) (d)
(b)(a)
29
Figure 3.10: (a) The schematic illustrates top and side view of the cantilever
which consists of two sensors for both topography and magnetic imaging. (b) SEM
image of the fabricated sensor. Surface topography (c) and SHPM image of array of
NiFe rectangles 31
.
There are different designs and fabrication ways for combined cantilevers in
literature. Although having additional component for distance control is not necessary
and advantageous for the sensors produced with this way, it is not only high costly and
but also difficult method to fabricate. QTF is used as an alternative method in SHPM
for AFM tracking mode. For this purpose, the Hall sensor is mounted on top of the
commercially available QTF which is used as a force sensor for approaching
mechanism and control tip–sensor separation during scanning process in SHPM
application. In the following section, details of the theoretical and experimental aspects
of QTF as force sensor will be presented.
3.4 Quartz Tuning Fork Force Sensor for AFM Tracking Mode in SHPM
Piezoelectric effect is a property of some materials that can generate electrical
energy when they are subjected to mechanical pressure. After this phenomenon is
discovered by Jacques and Pierre Curie, they reported piezoelectricity in their first paper
in 1880 32
. They also designed piezoelectric quartz electrometer which is capable of
measuring very small amounts of current values with high precision. Since then,
piezoelectric materials have been widely used either in engineering or in daily life such
as motors, lighter, musical instruments, keyboards, actuators or various kinds of
sensors.
Nowadays, piezoelectric materials exist naturally as well as they can be
manufactured in any desired sizes and shapes. Quartz, tourmaline and Rochelle salt are
some of the examples of natural piezoelectric materials whilst lead zirconate–titanate,
barium titanate and silicon selenite are examples of synthetic ones 33
. Among them, the
most abundant natural piezoelectric material is known as quartz which is one of the
crystalline forms of silicon dioxide (SiO2). Quartz crystals have been employed in
various applications due to their unique properties such as high young modulus, high
mechanical stress and lack of hysteresis and pyroelectric effect 34
. One of the important
30
and well known applications of quartz crystal is Quartz Tuning Fork(QTF). QTF can be
used in various fields. Most widely, the watch industry uses them since it is not only
low–priced and but also more precise than balance wheel. Besides, they have been used
in spectroscopic gas sensor which enables analyzing of samples with volume of 1 mm3
35. QTF is first implemented successfully in scanning probe microscopy by Günther et.
al. in 1989. They achieved 3 µm lateral resolution and 5 nm vertical resolution of a
grating sample with 8 µm periodicity by Scanning near–Field Acoustic Microscopy
(SNAM) 36
. Later, it has been applied to scanning near field optical microscope
(SNOM) for distance control and measuring shear force quantitatively 37
. It can be also
used in MFM in both room temperature and low temperature applications 38, 39
. In
addition to this, AFM takes advantage of QTF owing to its high quality factor and stable
resonance frequency. Using QTF in AFM both shear force and normal force can be
measured by attaching tip either perpendicular or parallel to oscillation direction
respectively. 100 nm lateral resolution has been reported in the earlier study by
attaching a tip with 150 nm apex to QTF 40
. One of the prominent studies by using QTF
in AFM has been performed by Franz Giessibl in 2000. In this work, he glued one of the
prongs of QTF to a substrate and mounted a sharp tungsten tip to the free prong. Then,
he managed to obtain atomic resolution image of Si (111) under vacuum condition
(Figure 3.11) 41
.
31
Figure 3.11: AFM image of Si (111) under vacuum by using a sharp tip attached
to QTF 41
.
QTF works successfully in different condition such as low temperature, high
magnetic field or vacuum. Furthermore, QTF is superior than Si cantilever in some
respects. For example; it has high quality factor which allow measuring frequency shift
with higher accuracy and it has high stiffness which avoids tip jump into contact with
sample surface 42
. Therefore, it also took important place in SHPM application. In
SHPM, conductive samples are easily measured by STM tracking mode but for
nonconductive samples, AFM tracking mode has to be used. As previously mentioned,
there are different ways to fabricate Hall sensors for AFM tracking mode. Since
fabrication of Hall probe which includes both magnetic sensing and force sensing
element in a single cantilever is so difficult, alternative method is developed by using
QTF. In this method basically, fabricated Hall sensor is attached on top of QTF which
has 32,768 Hz resonance frequency (Figure 3.12 (a)). The details of the fabrication of
Hall sensor integrated QTF are as follows. First of all, electrical connections on printed
circuit board (PCB) should be checked individually by multimeter in order to confirm
that there is no short circuit between adjacent pads. Picture of PCB used in SHPM is
given Figure 3.12 (b). Then, a thin rectangular ceramic sheet is glued on top of PCB by
low temperature epoxy to provide electrical isolation between pads and QTF. Low
temperature epoxy is prepared with hardener and adhesive by mixing them under the
ratio of 1:2 in volume. The mixture should be applied to the desired pieces as soon as
possible before it hardens. Then, pieces are baked at 100°C in an oven to accelerate
drying for 20 minutes. Glued pieces can also be dried by leaving them at room
temperature for 1 day. Once PCB and ceramic sheet is glued together, QTF can be
prepared. Commercially available QTF is enclosed by metal shell under vacuum. QTF
can be extracted from its cylindrical metal package with the aid of pliers. In this step,
one should be very careful in order to inhibit damaging electrodes of QTF or break QTF
itself as it is a delicate material. Legs of QTF are made up of magnetic material.
Therefore, they need to be broken off by using soldering iron, because a magnetic
material around the sensor affects the measurements and scanning results. If soldering
iron touches to soldering joints and a magnet placed close enough to them, then legs are
attracted by the magnet and they can be easily removed from QTF. Extracted QTF
should be controlled under optical microscope before using it to ascertain that electrodes
32
are not damaged. Then QTF is glued on ceramic sheet as flat possible by low
temperature epoxy as shown in Figure 3.12(c). Once epoxy dried, electrical connections
from QTF to PCB can be taken with thin non–magnetic wire by soldering. After all
electrical connections are controlled by multimeter, PCB should be cleaned with ACE
and IPA to remove contaminants on its pads and to provide maximum adhesion for the
wire bonding step. Finally, according to application, sensor to be used has to be
integrated on QTF. Sharp AFM tip is glued by metal epoxy while low temperature
epoxy is used to attach the Hall sensor metal to QTF (Figure 3.12 (d)). Fabricated Hall
sensor should be stored in a specially designed box as is shown Figure 3.12 (e).
33
Figure 3.12: The pictures show process steps of fabrication of Hall sensor for
AFM tracking mode by using QTF. (a) Before (top) and after (bottom) QTF is removed
from its metal can. (b) The picture of PCB used throughout measurements during this
study. (c) and (d) pictures of QTF which is glued on the top of PCB and Hall sensor
attached to QTF respectively. (e) Dedicatedly designed boxes for Hall sensor storage.
Non–contact AFM technique was used to track the surface using a hall sensor that
is mounted on a QTF. In AFM tracking mode by using QTF, first of all, resonance
frequency of QTF is measured before it is approached to the surface. For this purpose,
QTF is oscillated by dither piezo for a predetermined frequency interval and phase and
amplitude of the signal is measured. Frequency value at which maximum amplitude
occurs is called as resonance frequency. Once resonance frequency of QTF is found,
PLL is locked to phase of the signal which comes from QTF. Since QTF is a
piezoelectric material when it is vibrated by dither piezo, it produces an electric signal
which is typically on the order of nano–Amperes. Generated AC current is amplified by
amplifier and converted to the voltage by I–V converter. Therefore, an AC signal which
corresponds to mechanical vibration is obtained. PLL is locked to input signal phase.
PLL shifts the frequency to keep the phase constant. In non–contact mode AFM shifted
frequency (Δf) signal is used as a feedback source. If the QTF is approached to the
sample surface, resonance frequency of the QTF shifts due to short range and long
range forces between tip and sample. Approaching is completed when Δf equals to the
set value entered in the software 3, 21
. Feedback loop keeps the Δf at that set point.
Frequency shift is directly proportional to gradient of force between tip and sample and
given by the equation 43
:
. (3.1)
In this equation, k is the spring constant and f0 is resonance frequency of QTF.
Moreover, quality factor of QTF is calculated by equation 3.2 44
:
. (3.2)
34
Here, Q is the quality factor, f0 is resonance frequency and FWHM is full width half
maximum of resonance curve.
Resonance frequency of as received QTF used during study is 32,768 Hz and
resonance frequency of each QTF is measured by microscope and SPM software
program. In this way, QTF which has higher quality factor can be selected for scanning
application to obtain higher resolution as the SPM software calculates quality factor and
maximum phase automatically. Figure 3.13 shows dialog box of the SPM software for
QTF Auto tune. Once QTF with higher quality factor is distinguished, this sensor is
glued on top of QTF. That causes reduction in resonance frequency because of
additional mass coming from glue and Hall sensor on QTF. Resonance frequencies of
QTFs in air before and after fabricated Hall sensor attached are given in Figure 3.14 (a)
and (b) respectively. As is clearly seen from these graphs, central frequency is dropped
from 31,857 Hz to 12, 662 Hz after Hall sensor attached to it.
Figure 3.13: Picture of dialog box in SPM software for central frequency, quality
factor and maximum phase calculation.
35
Figure 3.14: Resonance curve of QTF before (a) and (b) after Hall sensor attached
to QTF in ambient condition respectively.
Dynamics of QTF are examined in detail by Akram et. al 45
. For this purpose, they
described motion of QTF as one dimensional driven damped oscillator and they have
calculated resonance frequency of it by using both cantilever beam theory (CBT) and
simple harmonic motion (SHM). They have found resonance frequency of QTF as 31.7
kHz according to SHO and it was found to be 21.2 kHz according to CBT for QTF
whose mass is 2.13 mg with Hall sensor mass of 1.11 mg. In addition to this, they have
used COMSOL simulation program in order to obtain resonance frequency and then
they compared results with theoretical calculations. For this, they took into account two
situations in their simulation: a) both prongs are free and b) one prong is fixed to
relatively massive material with and without Hall sensor mounted. In the first situation
(both prongs free), there are two eigenmodes including in–phase (symmetric) where
both arms of QTF oscillate in the same direction and anti–phase (asymmetric) where
both arms oscillate in opposite direction 3, 45
. In the second situation (one prong free),
there is only symmetric mode. Simulation results show that resonance frequency of
symmetric mode is close to CBT while it is close to Simple Harmonic Oscillator (SHO)
for asymmetric mode. Finally, they fabricated GaN Quantum Well hall sensor and
mounted it on top of QTF. Figure 3.15 represents resonance frequency curve of QTF
attached GaN Hall sensor for one prong free and both prongs free situations and for two
different sizes of quartz tubes 45
. As is seen from this figure, theoretical calculations and
experimental results of resonance frequency agree with each other. Minor differences
result from mass of glue and Hall sensor 45
.
(a)(b)
36
Figure 3.15: Resonance frequency curve of QTF attached GaN Hall sensor for one
prong free and both prongs free situations and for two different sizes of quartz tubes 45
.
Last but not least, temperature dependency of QTF plays important role in SHPM
since piezo coefficients, resonance frequency and quality factor of QTF changes with
variations of temperature. It was reported that between 77 K and 300 K, resonance
frequency increases till 250 K then it starts to decline 3,21
.
37
CHAPTER 4
FABRICATION OF THE HALL SENSOR AND CONVENTIONAL
FABRICATION METHODS
4.1 Introduction
Micro/nanofabrication allows manufacturing of devices with desired patterns on
a substrate for engineering or research purpose. Techniques to be used are chosen
according to features size, resolution, substrate type, reproducibility or speed of process.
For example; for structures with large feature sized (approximately bigger than 1–2 µm)
generally photolithography is preferred while electron beam lithography (EBL) can be
implemented for smaller features (down to 6–7 nm). Besides, photolithography is more
convenient and practical for mass production application. Each technique has its own
advantages and disadvantages. However, shrinking dimension of feature size is one of
the most prominent goals in all lithographic methods. According to Moore’s law,
number of components on a chip becomes two times more approximately every two
years 46
. This means that critical dimension (CD) produced on a substrate is reduced
with new technology and process development every passing day.
In this chapter, conventional device fabrication methods used in Hall sensor
production will be presented in detail. Device fabrication process consists of EBL,
optical lithography, dry and wet etching, packaging etc will be discussed thoroughly.
38
Problems faced during fabrication and solutions developed are presented in this chapter.
In addition to this, recipes implemented for fabrication of Hall sensor are also provided.
4.2 Nanofabrication Techniques
4.2.1 Substrate Preparation
Wafers to be used in device fabrication must be as clean as possible, since
cleanness of the wafers directly affects production yield and performance of fabricated
device. It is easier to clean the pieces after cutting the wafers. Therefore, a second
cleaning is not needed for debris arising from cutting process.
In this work, we used 4 inch 285–300 nm dry chlorinated thermally grown SiO2
on top of 500–550 µm thick Si (supplied by Nova Electronic Materials) for graphene
Hall sensor and 3–inch semi–insulating GaAs (supplied by American Xtal
Techonology) wafers for Bismuth hall sensor. Surface orientation is [100] for both of
these wafers. As stated above, the wafers are diced into 1.5 × 1.5 cm2 or 2 × 2 cm
2 chips
with a diamond scriber to be compatible with mask used in optical lithography, before
cleaning procedure applied.
There is a number of different ways to clean the wafers and cleaning methods
should be selected according to the type of contaminants on the wafer. After cutting the
wafers into small pieces, we initially applied standard cleaning procedure which
consists of three stages, since the samples are unprocessed yet. During cleaning process,
different beakers are used for each chemical in order to avoid cross–contamination. In
the first step, the samples are placed in a beaker filled with analytical grade of acetone
(ACE) which solves most organic materials from the sample surface and kept in an
ultrasonic cleaner for 5 minutes. Ultrasonic cleaning is required to remove particularly
insoluble particulates. Secondly, the samples are soaked one after another in three
different beakers of acetone for 30 seconds without allowing them drying when it is
transferred from one beaker to another one. This process is repeated by using isopropyl
alcohol (IPA). Finally, the samples are flushed with IPA and dried with nitrogen gas.
39
Completion of the sample cleaning with IPA provides to remove ACE residue from the
sample surface.
The other conventional way to remove organic molecules from the sample surface
is known as piranha cleaning. Even though there exist different mixture ratios for this
solution, we prepare 3:1 ratio of H2SO4 (95–97% concentration) to H2O2 (37%
concentration) mixture and pieces are kept in this solution for 5 minutes. Then, the
samples are immersed in deionized water (DI) and this is followed by nitrogen blow
dry. It must be noted that piranha solution is not generally used for GaAs cleaning,
because it etches GaAs substrate. Moreover, piranha is not only used for the wafer
cleaning but also it is widely used for photomask cleaning after repeatedly use of it in
photolithography, which will be discussed in the next section.
4.2.2 Optical Lithography
Optical lithography also known as photolithography is a technique which enables
transferring of desired patterns into a substrate by using ultraviolet light. Almost all
device fabrication flow includes a photolithography step due to its several benefits. One
the most significant advantageous of the photolithography is that many devices can be
fabricated at a time. Additionally, it can produce micron sized or smaller structure on
the sample even though it is relatively low–cost and easy to use method when you
compare to the other lithography tools such as EBL.
Once the wafer is diced and cleaned as discussed in the previous section, pieces
are initially baked at a temperature between 120–150° C in order to remove water
molecule on the samples surface. This step is known as dehydration bake and it is
necessary to get better adhesion between the substrate and photoresist film. After the
dehydration baking, the substrates are coated with photosensitive material which is
composed of resin, solvent and photoactive compound (PAC). Most of the solvent in the
photoresist is evaporated during spin coating and soft baking steps and these results in
reduction of photoresists film thickness. The soft baking is substantial because it not
only solidifies the photoresist so that resist does not stick to an optical mask but also it
enhances photoresist to substrate adhesion further. Next, the substrate is aligned with a
photomask which contains chromium patterns on a quartz plate. In our experiment, we
40
used the 4 × 4 inch mask to be compatible with mask aligner machines in the facilities
we used. Structures on the mask are designed with AutoCAD program and
manufactured by ML&C GmbH. Figure 4.1 shows the entire photomask having 34
different layers of patterns for Hall sensor fabrication. After alignment of the physical
mask, the substrate is illuminated by the ultraviolet (UV) light. There are several
exposure modes in photolithography such as contact, proximity and projection. Among
them, the best resolution approximately 0.07 µm is possible with projection mode;
nevertheless it is very expensive system. For our experiment, contact printing system is
used and we could achieve around 2–3 µm lithographic resolution. Disadvantage of this
method is that the photomask can be easily damaged and it becomes dirty after almost
every usage because the mask is in hard contact with the resist coated substrate. In order
to remove the resist residue on the mask surface, it was cleaned with acetone in each
usage and piranha cleaning was applied once every two or three months. Furthermore,
having a gap between the photomask and the sample result in diffraction which limits
the resolution of contact lithography. This problem generally arises from particles on
the mask and presence of edge bead on the substrate. To avoid this, edges of the coated
sample are scratched with scalpel or any other sharp material and the mask should be
kept as clean as possible.
41
Figure 4.1: The entire mask design which is used for Hall sensor fabrication.
Photoresist being a polymer changes its structure during UV light exposure and it
can be sorted into two groups: positive tone and negative tone resists. In positive resists,
exposed regions become more soluble in a chemical developer because chain–scission
occurs in polymer while areas under mask patterns remain insoluble. On the other hand,
solubility of a negative resist decreases during exposure because polymer is cross–
linked. Figure 4.2 illustrates comparison of the process flow for both negative and
positive tone resists. There are various commercially available photoresist. Since each
resist is required different amount of energy and consequently different exposure time,
suggested exposure energy given in the technical data sheet for individual resist should
be used initially to determine the optimum exposure time. Exposure time can be
calculated by dividing exposure dose by a constant UV light intensity of the mask
aligner system used. Two mask aligner systems are available at Sabancı University. The
exposure intensity of Karl Suss MA6 located at faculty clean room is 1.6 mW / cm2
whereas it varies up to 30 mW/cm2 for Midas / MDA–60MS placed at Nanotechnology
Research and Application Center (SUNUM). Light intensity of Midas mask aligner
monitored by lamp power supply unit is set to 25 mW/cm2 therefore; less exposure time
is required with this equipment compared to Karl Suss. Figure 4.3 shows an optical
microscope image of Hall sensor design after development by implementing positive
lithography in order to show resolution limits of Midas machine.
Cleaned Substrate
Resist Coating andBaking
Photomask
Photoresist
Substrate
UV Exposure
Development
Positive ResistNegative Resist
42
Figure 4.2: Process flow of the photolithography for both negative and positive resist.
Figure 4.3: Process flow of the photolithography for both negative and positive resist.
AZ 5214E is preferred during all photolithography steps of hall sensor fabrication
due to its numerous advantages. It is an image reversal resist which means that even
though it is a positive tone resist, it enables to produce negative tone images. For this
purpose, after standard mask exposure, the samples are baked at a temperature above
110° C known as post bake step. Later, the samples are exposed to the UV light without
physical mask. This process is called as flood exposure. Flood exposure is required
minimum 150 mJ/cm 2 doses for AZ 5214E photoresist but further optimization is done
by changing the exposure time. Finally, the samples are developed with a compatible
developer to AZ 5214E such as AZ 726 MIF, OPD 4280 or AZ 400K. AZ 726 MIF and
OPD 4280 are ready–to–use developers while AZ 400K is diluted with 1:4 ratio before
use it. Our procedure to determine the development time is as follows for image reversal
lithography: first we dip the sample into the developer until the designed patterns on the
sample can be seen by naked eye. Then, we wait for 10 s in order to obtain better
undercut profile. Development is followed by a dip in DI water in order to stop
development for 15 seconds and N2 blow–dried. Image reversal process plays an
important role in the optical lithography because it generates an undercut sidewall
profile which is desirable especially for lift–off process. Parameters of both positive and
10 µ
43
image reversal process for AZ 5214E photoresist are summarized in Table 4.1 and
Table 4.2 respectively.
Table 4.1: Summary of the positive lithography process parameters optimized for two
different mask aligner systems are given.
Karl Suss MA6
(1.6 mW/cm2)
Midas / MDA–60MS
(25 mW/cm2)
Dehydration Bake 120° C for 2 minutes 120° C for 2 minutes
Spin Coating 6,000 rpm for 45 s 6,000 rpm for 45 s
Prebake 100° C for 2 minutes 105° C for 1 minutes
Exposure 120 s
2 s
Development
AZ 726 MIF: 30 s
AZ 726 MIF: 50s
OPD 4280: 30 s
AZ 400K/DI (1:4): 30 s
Table 4.2: Summary of image reversal process parameters optimized for two different
mask aligner systems are given.
Karl Suss MA6
(1.6 mW/cm2)
Midas / MDA–60MS
(25 mW/cm2)
Dehydration Bake 120° C for 2 minutes 120° C for 2 minutes
Spin Coating 6,000 rpm for 45 s 6,000 rpm for 45 s
Prebake 90 ° C for 2 minutes 90 ° C for 2 minutes
Exposure 20 seconds 2 s
Post Bake 115° C for 2 minutes 115° C for 2 minutes
Flood Exposure 180 seconds 11.6 s
Development
AZ 726 MIF: 90 s
AZ 726 MIF: 25 s
OPD 4280: 20 s
44
4.2.3 Electron Beam Lithography
EBL is a prominent tool in device fabrication since it enables to fabricate
structures with feature size as small as sub–10 nm. There are several advantageous of
EBL over photolithography. For instance, resolution is not diffraction limited. Besides,
no physical mask is required in order to transfer pattern onto the substrate surface,
which makes this technique more flexible in arbitrary pattern designing.
The most remarkable property of EBL is its capability of generating high
resolution nano–patterns. In photolithography, resolution of lithography is primarily
limited by wavelength and it is given by Rayleigh equation 47
:
. (4.1)
where k1 is the constant between 0.4–0.8, is the wavelength of radiation and NA is the
numerical aperture of exposure system 47
. As is seen from the equation 4.1, the shorter
wavelength is used, the better resolution can be achieved. The shortest wavelength of
light used in optical lithography is 193 nm 48
. On the other hand, de Broglie wavelength
of electron is 0.0086 nm for 20 keV electron energy, which is used generally in EBL
system integrated to scanning electron microscopy (SEM). This shorter wavelength
enables to get around 0.52 nm atomic resolution which makes it possible to obtain
nanostructures with much higher resolution when you compare to the resolution
achieved in photolithography 49
. By using high energetic electron beam, wavelength of
electrons can be further reduced.
Process flow of EBL resembles in many aspects to photolithography except that a
focused beam of electrons is used for exposure of the resist coated substrate instead of
UV light. First of all, the samples are cleaned with the standard cleaning procedure as
discussed previously and the cleaned samples are covered by resist which is sensitive to
electron beam by using spin coating method. As in the photolithography, there are two
different classes of resists for EBL application called as positive and negative resist.
Polymethyl methacrylate (PMMA) is the most widely used positive EBL resist, since it
is able to produce high resolution structures and it is suitable for lift–off process. When
the positive resist is exposed by the electron beam, it becomes more soluble in the
45
developer and exposed area can be removed from the substrate surface. PMMA with
different molecular weight is commercially available. However, PMMA having
molecular weight of 495 K and 950 K are extensively used in EBL application.
Sensitivity and contrast of e–beam resist depend on molecular weight. The lower
molecular weight means that the resist is more sensitive but has lower contrast and
provides lower resolution. Beside this, generally chlorobenzene and anisole are used as
a solvent of PMMA. When concentration of solvent increases, the resist becomes more
viscous. The final film thickness of PMMA in chlorobenzene is a bit thicker than
PMMA in anisole for the same amount of solvent concentration. In addition, anisole is
less harmful to environment and human body therefore, it is more preferably. On the
other hand, a focused electron beam exposure of a negative e–beam resists result in
cross–linking in the polymer chain. Consequently, negative e–beam resist becomes
insoluble in the developer after exposure. HSQ (Hydrogen Silsesqioaxene) is the most
common negative resist because of the fact that it not only offers very high resolution
less than 10 nm but also it is used for etch mask because of its high resistance to
chemicals 50
. It can be diluted with MIBK in order to obtain resist with different
concentration and thickness. HSQ as Fox 12 (Flowable Oxide) is preferred as negative
e–beam resist in this work. During this work, resist having different thickness,
molecular weight and tone is used according to necessity. Thickness of e–beam resist
can be measured by the surface profiler by scratching the baked samples with a sharp
tweezers or a scalpel. Step height measurement of the PMMA and any other resist film
thickness are measured by KLA–Tencor P6 surface profiler. Spin speed is another
factor which directly affects the thickness of the resist and it is decreased by increasing
spin speed. After coating the samples, they are baked at a recommended temperature in
order to evaporate solvent in the resist and make the resist harder. HSQ is baked at two
steps: first at 150° C for 2 min subsequently, 220° C for 5 min while PMMA is baked at
180 ° C for 1 hour. The samples become ready for electron beam exposure, after spin
coating and baking.
When the electron beam is passed through the e–beam resist on top of the
substrate, inelastic and elastic electrons scattering take place either in the resist or in the
substrate which limits the resolution of EBL. These electrons scattering result in pearl–
shaped volume in the resist. The volume of the pearl–shaped depends on mainly
electron energy. When electrons are more the energetic, they will penetrate deeper.
46
Some of the incident electrons change their direction with small angle after they collide
with resist electrons. They also transfer a portion of their energy (inelastic scattering)
and this is called forward scattering. On the other hand, very few of the incident
electrons collide with atoms in the substrate without losing their energy (elastic
scattering). If they can get back to the resist from the substrate, they expose the resist
additionally. In this case, they make wide angle scattering and this is called as back
scattering. Undesired exposure of the resist due to back scattering of electrons is known
as proximity effect. It results in getting different structure size and shape from the actual
designed pattern size. In fact, it can cause irrecoverable lift–off problem on fabricated
structures. Range of proximity effect depends directly on many parameters such as: the
substrate type, accelerating voltage of electrons and resist thickness. Proximity effect
becomes one of the challenges which users try to overcome this problem. Various
solutions are developed to minimize the proximity effect. Some of them can be listed as
follows:
Increasing acceleration voltage.
Preferring bi–layer resist process.
Using dose modulation program which adjust dose for particularly large
area dense patterns.
After exposing the resist, development process is carried out according to the
resist tone. When the positive resist is used, regions exposed to electron beam become
more soluble in the developer. For PMMA being a positive resist, different ratios of
Methyl Isobutyl Ketone (MIBK) and IPA mixture can be used as the developer. The
best resolution is obtained with 1:3 MIBK: IPA, even though sensitivity is low at this
concentration. When the negative resist is exposed with focused electron beam, as a
result of cross linking reactions in the e–beam resist, irradiated areas becomes insoluble
in the develop. In this study, HSQ is preferred for negative lithography application and
it is developed by using TMAH based developer. EBL process flow for positive and
negative resist is shown Figure 4.4.
47
Figure 4.4: Process flow of the EBL for both negative and positive resist tone.
Well–defined undercut profile plays a major role in both photolithography and
EBL particularly for lift–off process. Producing undercut profile in EBL is easier than
photolithography since volume of electron interaction become larger in the deeper part
of e–beam resist layer 51
. In EBL, the best method for this profile is achieved by bilayer
lithographic process. In this process, a high molecular weight e–beam resist is spin
coated on top of another e–beam resist which has lower molecular weight. As
mentioned above, the resist with lower molecular weight is more sensitive and develops
faster than higher molecular weight PMMA in the developer after e–beam exposure,
which leads to create undercut shape. Two different recipes are used for bilayer process.
In the first one, the substrate is spin coated with 495 K PMMA–C2 at 5,000 rpm (50 s)
and baked at 180 ° C for 40 minutes. After the sample is cooled down, subsequently the
second layer 950 K PMMA–A2 at 4,000 rpm (50 s) is coated and placed on a hot plate
at 180 ° C for 40 minutes. The resist on the top is dropped to substrate when the spinner
is working at very low speed (100 rpm for 2 second) in order to prevent intermixing of
these two resist layers. This is the most critical step in the bilayer process as it can ruin
final pattern structures. In the second recipe, PMMA and its copolymer MMA (8.5)
EL11 is used in order to get undercut resist profile. In this case, in the beginning MMA
is spin coated at 6,000 rpm for 45 s and baked for 5 minutes at 180° C. Later, A2 950 K
PMMA–A2 is dropped the surface while spin coater is at 1,000 rpm for 4 second , as in
48
the previous transaction, and samples are baked for 5 minutes at 180° C. After coating
is completed, the samples are developed 1:3 MIBK: IPA for 1 minute and followed by a
dip IPA for 15 seconds to stop development. Finally, the samples are blow dried with
N2 and then metal evaporation can be performed. The Figure 4.5 illustrates fabrication
procedure of undercut profile and lift–off process for bilayer system.
Figure 4.5: Schematic illustration of fabrication steps for bilayer system.
Exposure of the focused electron beam is the most crucial step in EBL. Electron
beam exposure is performed with two different systems in this study. The first system is
Nabity Pattern Generation System (NPGS) which is integrated to field–emission SEM
(Zeiss Leo Supra 35VP with Gemini column) located at Sabancı University. SEM
machine and computer of NPGS communicate via Remcon32.exe interface software on
the SEM computer. Process flow starts with spin coating independently from exposure
system. Once the substrate is spin coated, resist residue on backside of the sample
High Molecular Weight PMMA
Low Molecular Weight PMMA
Electron Exposure Development
Spin CoatedSubstrate
Metal Evaporation
Lift-off
49
should be cleaned with ACE so that the sample surface is in level as much as possible.
Then, four corners of the samples are scratched by diamond scriber to obtain tiny
particles on the sample surface which will be used for adjustments of focus and
correction of astigmatism later. The electron beam has to be properly focused before
EBL application for high resolution patterning. For this purpose, image of these small
particles on the surface is initially acquired at low magnification and step by step
magnification is increased up to 3,000k After all the calibration is finished, electron
beam has to be moved to the center of the Faraday cup which is located on the sample
holder in order to measure beam current. The gold electroplated sample holder consists
of a lot of holes with 0.5 mm radius on it. Faraday cup is prepared by dripping liquid
carbon paste on one out of these holes. Beam current depends directly on aperture sizes
and it decreases with smaller aperture size. There are different aperture sizes ranging
from 7.5 µm to 120 µm. But, only 10 µm aperture sized is used in all EBL processes.
Beam current was measured around 30 pA for 30 kV and 20 pA for 15 kV EHT voltage.
After measuring beam current, firstly x–y–focus should be accomplished in case there is
tilt on the sample surface due to dusts on back side of the sample, inhom*ogeneities on
resist surface etc. NPGS calculates working distance values and sent to microscope for
each point on sample by using x–y–focus mode.
Prepared pattern file is inserted Run File Editor which allows defining writing
conditions. Some of the parameters in the run file are as follows: center to center
distance, line spacing, measure beam current, area (line or point) dose and
magnification. Area dose given in the equation 4.2 is applied for filled polygon shapes
for this study 52
. It is also possible to determine the entity type of the designed file in the
run file editor such as array, alignment or pattern.
(4.2)
Patterns are created according to resist tone in DesignCAD drawing program
which is supplied by NPGS system. In DesignCAD program, patterns can be assigned
to different layers which allow exposing the components in the same design file with
different exposure parameters. It is always better to determine optimal dose for specific
pattern. There are two ways to do this. In the first way, pattern is arrayed by using
50
DesignCAD program. Each element in the design file is assigned with different color
and each color represents different exposure dose. However, when DesignCAD program
was used for finding optimum dose of Hall sensor pattern (Figure 4.6 (a)), lift–off
problem occurred. Figure 4.6 shows SEM image of Hall sensor after 3 nm chromium
and 30 nm gold coated and lifted–off.
Figure 4.6: (a) Screenshot of Hall sensor array designed in DesignCAD program. (b)
SEM image of hall sensor patterns after lift–off process.
Optimum dose can also be found by using array mode in Run File Editor. All
parameters such as array spacing, number of rows and columns are entered in Run file
Editor instead of making it in DesignCAD file. The software calculates required dwell
time once beam current and exposure dose entered. In this case, designed pattern is
exposed at a different dose one after another according specified dosed in editor. It is
supposed that lift–off problem of Hall sensor is solved with this method due to the fact
that Run File allows exposing patterns with smaller writing field and higher
magnification (Figure 4.7 (a)). Applied exposure dose varies from 150 μC/cm2
to 550
μC/cm2 in this test and optimum dose for 100 nm Hall cross sized sensor is determined
as approximately 230 μC/cm2
according to SEM image (Figure 4.7 (b)). As is seen
from this image, although we overcome the lift–off problem, we came up with the
difficulty of discontinuity at two junction legs on hall cross area. This problem is
resolved by dividing Hall cross pattern into two parts as shown in Figure 4.8 (a). Since
a b
51
Hall cross area has smaller feature and delicate, first Hall cross part is exposed and then
legs of Hall sensor is exposed to electron beam separately. 200 nm overlap between
Hall cross and legs was sufficient for desired design. SEM image of fabricated Hall
sensor designed according to this is given in Figure 4.8 (b).
Figure 4.7: (a) SEM image of hall sensor patterns arrayed by Run File Editor (b) after
lift–off process.
Figure 4.8: (a) Screenshot of 100 nm Hall sensor designed in DesignCAD program (b)
SEM image of fabricated 100nm Hall sensor pattern.
(a) (b)
(b)(a)
52
Although NPGS is capable to produce various patterns with different electron
energy and aperture sizes, there exist some restrictions in this system. One of them is
that, specimen current is relatively low because of low high extra high tension (EHT)
voltage, which causes longer exposure time. Also, alignment process takes considerable
amount of time, since it is done one at a time manually. In addition to this, there is no
proximity effect correction (PEC) algorithm embedded on the system. Therefore,
proximity effect is a limiting factor for densely packed patterns. Advanced EBL systems
may overcome these problems successfully. In this study, we have also used dedicated
EBL instrument (EBPG 5000+ ES from Vistec). This system is superior to EBL
integrated to SEM in many aspects. For instance; it has laser interferometer stage which
provides very fast movement with high field stitching accuracy (+/–15 nm at 100 µm.)
Also, range of beam currents is not limited as in the case of SEM converted system.
EBPG system offers successfully working beam current varies from 50 pA up to 150
nA, which is measured Faraday cup placed on substrate holder. Therefore, writing time
can be reduced by using high beam current for large area patterns and low specimen
current can be preferred for structures with smaller feature sizes to obtain higher
resolution as spot size (diameter of electron beam) decreased. Although, this system can
be operated with 50 keV and 100 keV electron beam energy, exposure with this
machine was performed at 100 keV with several different doses. Moreover, it is
possible to manufacture photomasks with this system since it provides not only high
beam currents but also it is equipped with 5 inch mask holder.
AutoCAD program is preferred for pattern generation although patterns to be
fabricated can be designed by various drawing software programs. Once design file is
prepared, LayoutBEAMER supplied by GenISys is used to convert this file into a
readable format in Vistec machine. It is user friendly software which allows not only
fracturing patterns into subfield and but also editing designed file by rotating, extracting
layers, biasing patterns, reversing tone etc. Converted file is loaded a graphical user
interface called CJOB supplied by EBPG5000+ system. By using CJOB interface, one
can set exposure dose, make array of designed patterns, define identifier and alignment
marker etc.
One of the primary factors which significantly affect resolution of fabricated
structure in EBL process is proximity effect. As it is mentioned above, using high EHT
53
voltage and PEC can minimize this problem. Particularly, densely packed patterns are
suffering from proximity effect. PEC is provided by LayoutBEAMER program in
Vistec machine. In this program, a computational technique called Monte Carlo
simulation which models electron beam and substrate interaction is used. Monte Carlo
simulation generates not only trajectory of each electron when electron beam penetrates
a solid according to material type, thickness and electron energy but also calculates
point spread function (PSF). Psf file generated for specific resist thickness, substrate
type and electron energy are inserted in LayouBEAMER program for each pattern.
Figure 4.9 (a) shows the optical microscope image of Hall sensor pattern (50 nm Hall
cross size) after the sample developed without implementing PEC. The EBL parameters
used in this work were 180 pA beam current, 1 nm step size and 750 μC/cm2
exposure
dose (same dose is applied for whole pattern). It is clearly seen from this figure, regions
between pads of Hall sensor are also reluctantly exposed which can cause shortcut
problem after metallization, while tip of pattern seems properly exposed according to
color contrast of the image. This problem is corrected by using PEC function. Hall cross
pattern implemented by PEC is shown in figure 4.9 (b). Color scale shows dose factor
of the layout and it varies from 0.5 to 2 for this layout. According to simulation, dose
increases at the edge of pattern.
Figure 4.9: (a) The optical microscope image of 50 nm Hall sensor after development
(b) Simulation of Hall cross pattern after implemented by PEC.
(a) (b)
54
Determining optimum dose for each pattern is required for also this system in
order to obtain correct sized pattern without having any lift–off problem. For this
purpose, we initially performed dose array which varies from 300 μC/cm2 to 1200
μC/cm2. From this test result, it is found that 800 μC/cm
2 exposure dose was the closest
value to get desired size. The SEM images of 50 nm Hall sensor at different
magnifications after 5nm Cr/30 nm Au evaporation and lift–off in acetone are presented
in Figure 4.10 (a) and (b). 500 μC/cm2
is used as base dose in this sample so that Hall
cross area is exposed with around 800 μC/cm2 according to PEC simulation. Minimum
Hall cross sized obtained with Vistec machine by using 950 K PMMA–C2 (100 nm
thickness) was around 20 nm which is shown in Figure 4.11. Thinner resist could
further reduce cross size.
Figure 4.10: (a) and (b) the SEM images of 50 nm Hall sensor at different
magnifications after 5nm Cr/30 nm Au evaporation and lift–off in acetone.
55
Figure 4.11: The SEM image of 21 nm sized Hall cross by using 950 K PMMA–C2.
Last but not least, alignment procedure is easier in Vistec machine than EBL
system based on SEM. In this case, only three alignment markers are used for entire
sample which significantly shorten alignment process time to a few minutes. Moreover,
focus adjustment and astigmatism correction is not necessary and this makes the system
easy to use.
Although it has several advantageous over photolithography, there are also some
drawbacks of EBL. One of them is that they are high costly machines and maintenance
is required regularly. Besides, evaporated film thickness must be thirty percent less than
resist thickness which also limits application fields, since. In addition to this, proximity
effect limits to obtain high resolution structure though there exists some solution to
avoid it.
4.2.4 Etching Process
Etching process is frequently used micro/nanofabrication for removal of intended
material. There are many fundamental parameters which should be taken into account
for all etching process such as selectivity, uniformity, reproducibility, etching direction
and etch rates. One of the most important parameter is etching profile. In isotropic
etching, etching rates is same for all direction while horizontal etching rate and vertical
etching rate are different to each other in anisotropic etching. Besides, etch rates should
be determined before etching real samples in order to prevent over etching since it is
irreversible process. Each material has different etching ratio for different etcher. This is
called selectivity and generally highly selective etcher is preferred in order to obtain
desired structures properly. Both dry etching and wet etching are used during
fabrication of Hall sensor according to necessity. In the next subsection, detail of
etching methods and recipes used in this study will be explained.
4.2.4.1 Wet etching
56
Wet etching is a liquid based etching method as name referred and it involves
chemical reaction between solution and material to be etched. It gives generally
isotropic profile with some exceptions. It is preferred frequently since it provides high
selectivity and reproducibility. In addition, it is an inexpensive and simple technique
compared to dry etching.
Hall sensor fabrication begins with cleaning of the substrates and recess etch.
Recess etch is required in order to prevent shortcut between electrical connections of
sensor and sample while the sensor is scanning the sample surface. For this purpose,
end of contact pads are etched around 50–60 µm deep before contact pad formation.
This step is very critical, since if undercut profile occur during optical lithography, this
can cause discontinuity in contact pads. In this work, semi–insulating GaAs used as
substrate is etched for recess formation. The sample is first spin coated with AZ 5214E
by spinning 4,000 rpm for 40 s, which provides around 1.5 µm resist thickness. Then,
prebake is done at 105 ° C for 1 minute. The sample is illuminated for 2 seconds by
using Midas mask–aligner system and after exposure, development is performed by AZ
726 MIF for 50s. Developed samples are rinsed in DI water for a longer time in this
case. Because of the fact that developer is a basic solution and if developer residue
remains on the sample surface, after dipping in an acid solution, this creates salt on the
sample surface which cannot be removed occasionally. Then samples are baked at 120 °
C for 2 minutes for hard bake process. Hard baking step also plays critical role for this
progress, since it improves adhesion resist to substrate surface and makes resist more
durable in chemical solution. Finally the sample is immersed in HCl:H2O2:H2O solution
at volume ratios of 4:7:55. Etch rate with this recipe is found as approximately 1
µ/minute. However, it is also observed that, etch rate decreases with passing time. Light
microscope image of etched sample by using this recipe is given in Figure 4.12 (a) and
(b) and etching profile is shown in Figure 4.12 (c) and (d).
57
Figure 4.12: (a), (b) The optical microscope and (c), (d) the SEM images of recess
formation.
Wet etching is also used in mesa step formation for fabrication of Hall sensor. It is
necessary in order to bring tip of the sensor the closest to the sample surface. In this
case, slow etch rate is required since 1.5–2 µm deep etching is fairly enough for our
purpose. EBL is implemented for patterning, just because it ensures higher alignment
accuracy than photolithography. We performed two different recipes. For the first one,
950 K PMMA–C2 which has 100 nm resist thickness by spinning 6,000 rpm for 45 s is
used. Even though hard bake at is applied after EBL, all resist on the sample surface has
been removed before desired etch thickness is achieved. This problem is solved by
using bilayer process. The sample is spin coated by 950 K PMMA–A2 and its
copolymer MMA (8.5) EL11 as explained in the previous section, which gives
approximately 550 µm resist thickness. Next, the sample is exposed with 550 μC/cm2
base dose by using PEC function. After development with 1:3 MIBK:IPA solution for
60 s, the sample is baked at 100 ° C (post bake). Then, the sample is immersed in
H2S4O4:H2O2:H2O solution at volume ratios of 1:8:40 for 60 s, which gives around 1.5
200 µ
(a) (b)
(c) (d)
50 µ
(b)
58
µm etch thickness. Mesa step is generally done after Hall sensor pattern is defined,
because it causes non–uniform resist coating for the next lithography steps if it is done
beforehand. The SEM image of mesa formation performed by this recipe is shown in
Figure 4.13. Dry etching is also implemented for mesa formation, but applications of
dry etch decrease production yields of fabricated sensors which will be discussed in the
next section.
Figure 4.13: SEM image of mesa formation on Bismuth Hall sensor.
4.2.4.2 Dry etching
Dry etching is chemical or physical process where ions in gas introduced to the
chamber interacts with a substrate to be etched 53
. Both isotropic and anisotropic profile
can be obtained with this method. Besides, etch profile and etch rate can be controlled
by changing pressure, temperature, applied power and ratios of gas. Dry etching can be
classified into three groups; physical etching, chemical etching and physical chemical
59
etching. In physical dry etch, highly energetic ions knock out atoms from sample
surface. In chemical dry etch, chemical reactions result in etching.
There are two different reactive ion etching (RIE) systems available in SUNUM
facility: chlorine based and fluorine based. F–based system is used for both oxygen
plasma and SiO2/Si etching. It is equipped with gas cylinder of O2, C4F8, Ar and Sf6.
Oxygen plasma removes organic materials from the sample surface by chemical
reaction and it is used in Hall sensor fabrication for several purposes. In Graphene Hall
probe (GHP) fabrication, after Hall cross pattern is defined by positive optical
lithography or EBL, oxygen plasma is used to etch unprotected graphene region with
resist in order to pattern graphene layer. The process parameters are 20 sccm O2 flow, in
a 100 W RIE power, 37.5 mTorr pressure. It is determined that 9 second completely
removes single–layer graphene by using given recipe. If oxygen plasma is applied
longer than this length of time, it can also etch not only side wall of resist, since resist is
also an organic material, but also graphene itself which results obtaining smaller pattern
size than desired. Furthermore, oxygen plasma is also indispensible for Bismuth Hall
sensor fabrication. In conventional Hall sensor fabrication, Hall cross is defined after a
number of lithography steps, which leaves a few nm resist residue on the sample
surface. Therefore, resist residue causes adhesion problem particularly for smaller size
Hall sensor. We resolved this problem by applying oxygen plasma before spin coating
with e–beam resist. Figure 4.14 (a) presents Bismuth Hall sensor which has not been
applied oxygen plasma before spin coating with e–beam resist. As it is seen from the
figure, there is no Hall cross defined by EBL due to adhesion problem whereas applying
oxygen plasma before spin coating overcame this problem shown in Figure 4.14 (b).
60
Figure 4.14: (a) and (b) SEM image of Bismuth Hall sensor without and with
oxygen plasma application before e–beam resist coating respectively.
Oxygen plasma is also used for enhancing adhesion of contact pads to the
substrate surface. For this purpose, after contact pads formation by photolithography,
oxygen plasma is implemented for 8–9 s by provided recipe before metal evaporation.
This step particularly plays very critical role in wire bonding of both GHP and Bismuth
hall sensor.
F–based system is used also SiO2/Si etching. The mesa which serves as AFM tip
was obtained in GHP by etching SiO2/Si wafer around 830 nm. 285 nm of SiO2 layer
was etched with C4F8/O2: 50/5 sccm; ICP power: 1,750 W; RIE power: 100 W;
pressure: 7 mTorr for 45 s and this is followed by 545 nm Si was etched with
C4F8/SF6:45/25sccm; ICP power: 1500 W; RIE power: 35 W; pressure: 15 mTorr for 3
minutes (Figure 4.15). After plasma etching, photoresist on the sample surface is
removed with ACE and followed by N2 blow dry. If there is still resist residue left on
the sample surface, the sample can be immersed in AZ–100 remover which strips resist
residue by either leaving it in this solution for one day at room temperature or 10
minutes at 80° C.
61
Figure 4.15: The optical microscope image of GHP after mesa formation.
In the early stage of this work, mesa step of Bismuth Hall sensor fabricated on
GaAs substrate is performed by using RIE system located in Advanced Research
Laboratory’s clean room at Bilkent University. Process parameters are 25 sccm CCl2F2
flow, in a 100 W RF power and 0.4 mBar pressure for 15 minutes. The Figure 4.16
shows the (a) top view and (b) tilted SEM images of a Bismuth Hall sensor whose mesa
step is performed by given recipe. As seen from these images, after resist removed from
the sample by dipping it in ACE, Bismuth grains on Hall cross area are also peeled off
which is observed almost all die and causes a big connection problem. In addition,
photoresist residues remained on the surface particularly, near edge of the chip which
significantly affects performance of Hall sensor. Then, Cl based RIE system which is
located in SUNUM and is equipped with gas cylinder of BCl3, Cl2 and Ar are used for
same purpose. Process parameters are 30 sccm BCl3 and 50 sccm Cl2 flow, in a 100 W
RF power, 800 W ICP power and 7 mTorr pressure. 45 seconds etch is performed in
order to etch 1.5 µm GaAs (Figure 4.16 (c) and (d)). Although successful etching is
performed with this recipe, since gas line of BCl3 is easily clogged, we decided using
wet etching for this step.
(a) (b)
(c) (d)
62
Figure 4.16: The SEM images of Bismuth Hall sensor after mesa formation.
4.2.5 Thermal Evaporation
Thermal evaporation is one of the most widely used for metal deposition in device
fabrication. In this method, source material to be deposited is heated until it is
evaporated or sublimated by applying high current under vacuum condition. In this
study, two different box coaters are used. The first one is supplied by Nanovak and it is
used for evaporation of Au, Cr, Ti and Bi. The second evaporation system is supplied
by Torr International which is equipped with both e–beam and thermal evaporator part.
In e–beam evaporation, high energetic electron beam cause evaporation of the source
material to be deposited. One of the advantages of e–beam evaporation over thermal
evaporation is that e–beam heats only the source material and this results obtaining thin
film with higher purity.
In the fabrication of Hall sensor, after contacts pads which provides electrical
connection from Hall cross area are patterned by image reversal optical lithography, 10
nm Cr and 200 nm Au are evaporated to the sample surface for metallization of contact
pads of both GHP and Bismuth Hall probe. Next, the sample is immersed in ACE for
lift–off process. Since adhesion of Au to the substrate surface is quite poor, Cr or Ti are
used before Au evaporation as an adhesion layer. Otherwise, Au can peel off easily
from the sample surface. Deposition rate during evaporation of contact pads is around 1
A/s for Cr while it is 5 A/ s for Au. Besides, evaporation of Bismuth pellets used as Hall
sensor material is performed at very slow evaporation rate approximately 0.1–0.2 A/s.
Since surface of Bismuth pieces is oxidized quickly, Bismuth pellets are immersed 1:5
ratio of HCl : DI water solution in volume for 5 minutes before evaporation. In addition
to this, we have not used Cr as adhesion layer for Bismuth evaporation. The SEM
images of 50 nm Bismuth Hall sensor (a) with and (b) without Cr used as adhesion layer
are given in Figure 4.17. As is seen from these figures, evaporation of Cr before
Bismuth makes grain size of Bismuth much larger. It is also observed that even though
EBL parameters are exactly same for both samples, size of Hall cross is around 20 nm
63
bigger when Cr is used before Bismuth evaporation. Therefore, we preferred to not to
use adhesion layer for Bismuth Hall sensor.
Figure 4.17: The SEM images of 50 nm Bismuth Hall sensor (a) with and (b)
without evaporation of Cr as adhesion layer before Bismuth metal.
4.2.6 Dicing and Wire Bonding
Wafer to be used in device fabrication is cut by a scriber into pieces whereas once
mesa formation of the samples is completed; the sample is diced into 64 pieces by
automatic dicing saw supplied by Disco. Since distance between the sensors have been
designed on mask to be 100 µ, it is compulsory to use dicer at the last step. Dicing blade
is selected specifically for our mask design and substrate type used in device
fabrication. Blade with 25–35 µm kerf width which is thickness of the slot on the
sample after dicing process is used in order to increase production yield of Hall devices.
However, it is observed that after prolonged usage, kerf width becomes thicker (up to
70 µ, after 15–20 usage)
Dicing step starts with spin coating of the sample with AZ 5214E photoresist in
order to protect devices from debris arising during cutting process and the sample is
baked at 80° C on a hot plate to harden photoresist. Process parameters of dicing were
same for both GHP and Bismuth Hall probe except feed speed. Feed speed for GaAs
substrate has to be as much as slow and it was set 0.1 mm/ sn in order to decrease
chipping defects whereas feed speed was set to 1 mm/s for SiO2/Si substrate. The Figure
4.18 shows diced Bismuth Hall sensor fabricated on GaAs substrate.
(a) (b)
64
Figure 4.18: Diced Bismuth Hall sensor fabricated on GaAs substrate
After dicing, resistance of each sensor is measured individually by using
multimeter in the probe station (Cascade Microtech PM5 Probe Station). Then, sensors
with resistance in the range of kΩ are selected and glued on top of QTF as is mentioned
previously. Next, electrical connections from sensor to PCB are provided by Kulicke
and Soffa wire bonder. .Though it provides both ball and wedge bonding, wedge
bonding is performed to take electrical connections from contact pads of Hall sensor to
the pads of PCB with 25 µm Au gold wire. The SEM image of diced and bonded
Bismuth Hall sensor is given in Figure 4.19.
Figure 4.19: The SEM image of diced and bonded Bismuth Hall sensor.
65
CHAPTER 5
GRAPHENE HALL SENSOR for SHPM
5.1 Introduction
Scanning hall probe microscopy (SHPM) belongs to scanning probe microscope
(SPM) family, which provides quantitative and non–invasive local magnetic field
imaging of magnetic and superconducting materials 54
. It provides high magnetic field
and spatial resolution, simultaneously with the topography of the sample surface 55, 9
.
The most important element of SHPM is the Hall sensor which is sensitive to the
perpendicular component of the magnetic field on the specimen. Hall sensors can be
fabricated from various materials such as GaAs/AlGaAs heterostructure, Bi, InSb and
graphene. GaAs/AlGaAs two–dimensional electron gas heterostructure is one of the
most popular materials for Hall probe production. Even though it has high mobility
which increases Hall coefficient, the minimum cross size is limited with this sensor
because of surface depletion effect, which results in low spatial resolution 56
. It is
reported that GaAs/AlGaAs Hall sensor with 250 nm cross sized fabricated by
implementing EBL has very high resistance which does not allow driving current at
room temperature 57
. In Bismuth Hall probes, the spatial resolution was decreased up to
50 nm, on the other hand minimum detectable magnetic field was increased due to low
mobility of Bismuth film 9. In order to overcome this problem, InSb has been used as
sensing element. It has high mobility (55,500 cm2/ Vs) at room temperature and low
carrier concentration. It was reported that minimum detectable magnetic field was 6–10
mG/ at 50 µA drive current in InSb Hall sensor with a size of 1.5 µm
2.
66
After the discovery of high mobility in graphene, it is also considered to be a good
candidate for Hall probe material 58, 59
. It has been demonstrated that minimum
detectable magnetic field (Bmin)of epitaxial graphene Hall sensors was 3.9 µT
which is better than with same sized InSb and CVD graphene Hall sensors. In table 5.1,
some of data taken from InSb, epitaxial graphene and CVD graphene sensors are given
for 5.0 µm Hall cross sized 7. They also demonstrated size dependency of Bmin and Hall
coefficient (RH) for epitaxial graphene Hall sensor and according to their results, Bmin
increases exponentially with descending cross size for the same driven current applied
(IBias=10 µA).
Table 5.1: Summary of data measured from InSb, epitaxial graphene and CVD graphene
sensor for 5 µm Hall cross size 7.
InSb Device
Epitaxial
Graphene Device
CVD Graphene
Device
R (kΩ) 12 22 100
RH (Ω/T) 974 711 310
µe (cm2/Vs) 8322 2643 –
Bmin (µT ) 6.5
(IBias=10 µA)
3.9
(IBias=10 µA)
43.0
(IBias=3 µA)
Sn
(nV/ )
55.3
(IBias=0 µA)
19.9
(IBias=0 µA)
–
Materials used in Hall sensor fabrication are selected according to its electronic
properties and ease to manufacture. For example; material having low carrier density
results in obtaining higher Hall coefficient and thereby provides better field sensitivity.
In graphene, carrier concentration decreases as Fermi level of graphene is brought to
closer Dirac point by applying gate voltage. Tang et. al produced 5 µm sized Hall
probes from CVD graphene and showed that Hall coefficient initially decreases when
gate voltage is very close to Dirac point , then it starts to increase when it is moved
away from Dirac point 60
. Therefore, as in the other areas, it has been a promising
material for Hall probes in SHPM applications thanks to its commensurable
67
characteristics to conventional Hall Probes. However, utilization of GHP for magnetic
imaging has not been shown in the literature yet.
In this chapter, we represent fabrication, characterization and performance of
graphene based micro–Hall devices for Low Temperature Scanning Hall Probe
Microscopy (LT–SHPM). Graphene produced by CVD method was preferred in order
to increase the production yield of the fabricated Hall sensors. In this work, for the first
time, we used GHP integrated with QTF for imaging localized magnetic field of NdFeB
in a wide temperature range of 3–300K. This study is published as “Single Layer
Graphene Hall Sensors for Scanning Hall Probe Microscopy (SHPM) in 3−300 K
Temperature Range” S. Sonusen, O. Karci, M. Dede, S. Aksoy, and A. Oral, Applied
Surface Science 308, 414–418 (2014) 61
.
5.2 Mechanic and Electronic Properties of Graphene
Graphene is a mono layer of sp² bonded carbon atoms packed into a two–
dimensional (2D) honeycomb lattice. In graphene, zero band gap semiconductors also
known as semi–metal, conduction and valance bands meet at a point called as Dirac
point in which density of states is zero in an ideal graphene (Figure 5.1) 62
. Furthermore,
in graphene both electrons and holes have a linear energy–momentum relationship and
it is given by 63
:
(5.1)
In this equation, Vf is Fermi velocity which is approximately 1.106
m/s (Vf ≅ c/300) and
positive sign represents electrons in this equation whereas negative sign corresponds to
holes 64
. This linear energy–momentum dispersion relation enables us to describe
charge carriers in graphene as massless Dirac fermions 65
.
68
Figure 5.1: The energy dispersion of graphene and the Dirac cone 62
.
One of the most striking properties of graphene is that it has very high carrier
mobility even at room temperature due to zero effective mass of its charge carriers 66
.
This property makes graphene promising material for electronic devices such as field
effect transistor (FET) 67
. It was reported that mobility of mechanically exfoliated
graphene on top of SiO2 is mostly over 10,000 cm2 V
−1 s
−1 at room temperature, while
mobility of suspended graphene exceeded 200,000 cm2 V
−1 s
−1 at low temperature
68 ,69,
58. Moreover, owing to high carrier mobility together with linear energy–momentum
dispersion relationship, graphene permits for observation of quantum hall effect under
high magnetic fields 62,
.
Graphene consists of carbon atoms arranged in a hexagonal pattern and each
carbon atom has 4 valence electrons 70
. Three of these valance electrons form covalent
bonds with its nearest neighbours and the forth electron is delocalized which makes
graphene conducts electricity 70, 71
. Graphene sheets are hold together by weak van der
Waals force and distance between two neighbour sheets is between 3.35 Å and 3.4 Å 72
.
On the other hand, bond length for two carbons in plane is about 1.42 Å and this strong
σ bonds allows graphene to exhibit superior mechanical properties 71
. It is very strong
material with high Young Modulus. It was reported that Young’s modulus of single
layer suspended graphene is 0.5 TPa whereas it is 1 TPa for graphite 73
.
5.3 Graphene Fabrication Methods
Since it was known that two dimensional crystals were thermodynamically
unstable, it was believed that graphene didn't exist in the free state 6. However, graphene
was first prepared via mechanical exfoliation of graphite crystals by Professor Andre
69
Geim's research group at the University of Manchester 58
. It has recently attracted
attention of the research communities, because not only graphene has unique electrical
and mechanical properties, but also the thinnest material ever fabricated
Graphene can be produced by various methods such as; mechanical exfoliation
also known as mechanical cleavage, CVD, epitaxial growth, anodic bonding dry
exfoliation and chemical synthesis 74
. In this study, mechanical exfoliation and CVD
techniques are used.
The most popular and easiest way to produce graphene is mechanical exfoliation.
It is basically repeated peeling of bulk graphite. Different sizes of graphite used during
exfoliation process. However, it is observed that large–scale graphene is obtained with
3–10 mm sized graphite flake supplied by NGS Naturegraphit (India origin). Once
graphite is peeled by using scotch tape, it is transferred to 285–300 nm dry chlorinated
thermally grown SiO2 on top of 500–550 µm thick Si pieces by using tweezers (Figure
5.2). Figure 5.2 (d) presents graphene sheets produced by this method.
Single Layer Multi Layer
(a) (b) (c)
(d)
70
Figure 5.2: (a), (b) and (c) Graphene fabrication steps and (d) single and multi
layer graphene sheets produced by this method.
Mechanical cleavage method is not only the simplest and cheapest way among the
existing methods of graphene production, but also it provides obtaining graphene films
with high mobility even at room temperature 75
. However, large area single crystal
graphene is particularly desired for graphene based devices in order to increase yield of
fabrication and this has led to development of different methods. One of them is CVD
which is a promising method for production of continuous graphene film in large area.
In CVD method, hydrocarbon gases such as methane (CH4), acetylene (C2H2) or
ethylene (C2H4) used as carbon source are introduced to quartz tube, either in vacuum or
ambient condition, located in a furnace 76
. A transition metal such as copper, nickel or
platinum is used as catalyst. When temperature is around 900°C–1100°C, hydrocarbon
gases are decomposed on metal substrate, which results in formation of graphene
lattice on metal susbtrate 77
. Once graphene is formed on the metal, graphene/metal
surface is covered by protactive layer (polymer). Then, the metal is etched using
different etcher according to metal type. Next, floating graphene is transferred to desired
substrate such as SiO2/Si, glass or polymeric surface 78
. Finally, protective and
supportive resist layer is removed from the substrate by a suitable solvent 77
. Although
Cu and Ni substrate is widely used as metal substrate in CVD technique, Pt allows
obtaining larger single crystal grain size than graphene grown on Cu. We used Pt as
metal substrate. Experimental details of graphene growth on Pt and results will be
discussed in the subsequent chapter.
Another widely used technique for large scale graphene production is epitaxial
growth on silicon carbide 79
. This method includes two steps. For the first step, SiC has
to be etched by hydrogen gas in order to prepare suitable SiC surface by removing
scratches for epitaxial growth 80
. In this process, cleaned SiC samples are annealed in a
chamber with %5 H2 and %95 Ar gas flow. For the second step, SiC is heated in
ultrahigh vacuum to temperatures around 1250–1350°C in order to sublimate Si and this
results in formation of carbon rich surface 81
.
71
5.4 Characterization Methods of Graphene
Distinguishing mono–layer and multilayer graphene, determining thickness, size
and electronic properties of graphene film play very important role in fabrication of
graphene based devices and their performance. There are several methods for electrical,
surface morphology, optical characterization of graphene sheet such as low–energy
electron diffraction (LEED), photon–electron spectroscopy, transmission electron
microscope (TEM), Raman spectroscopy, optical microscopy, AFM and SEM. Among
them, optical microscope is the simplest and nondestructive characterization tool. It is
particularly practical for graphene films produced via mechanical exfoliation method in
order to identify mono–layer and to locate graphene region for further measurement
since initially entire substrate surface is scanned. Optical contrast provides to
distinguish single and multi layer of graphene by naked eye under optical microscope.
Blake et al. reported contrast dependency of graphene with respect to SiO2 thickness
and wavelength of light (Figure 5.3) by Fresnel diffraction law 82
. According to their
result, contrast can be maximized by selecting specific SiO2 thickness and using suitable
filtering. Besides, the best visibility of graphene film on 280–300 nm SiO2 thickness is
provided by green light illumination as seen from Figure 5.3 82
.
Figure 5.3: Optical contrast variation of graphene with changes in SiO2 thickness
and of wavelength of light 82
.
72
Even though optical microscope is the easiest way, it is not quantitative method.
Raman spectroscopy is a powerful technique which provides quantitative and sensitive
measurements for graphene characterization. By using Raman spectrum, number of
graphene layers, defects and edge arrangement of graphene can be determined 83, 84
. In
Raman spectroscopy measurements, intense monochromatic light is produced by laser
to illuminates specimen and when a light interact with material, inelastic (Rayleigh) or
elastic (Raman) light scattering occur 85
. In inelastic scattering, frequency of a few
amounts of incoming photons may alter, after scattering. This shift is generally
associated with vibration frequency of the molecule and it results from absorption or
emission energy from the molecule 86, 87
.
There are two main peaks, G and 2D, in carbon based materials. Single and multi–
layer graphene are distinguished according to their intensities and shape of the peaks. In
single layer graphene, sharp and narrow 2D peak which appears at around 2700 cm-1
is
more intense than G peak which is 1580 cm-1
and this situation vice versa in graphite 83
.
In addition, D peak arises at 1350 cm-1
when there are defects on the film 83
.
Furthermore, Raman mapping gives information about uniformity of graphene over film
surface. Last but not the least, Raman spectrum of monolayer graphene can be well
distinguished also intensity ratio of I2D/IG . If this ratio is equal or bigger than 2, it refers
to presence of single layer graphene while I2D/IG decreases with increasing number of
graphene layer 88
.
AFM is another quantitative method for graphene characterization which provides
exact thickness of films even though it is slightly slow when it is compared to Raman
spectroscopy and optical microscope. Even though interplanar spacing of graphene film
is 0.34 nm, thickness of graphene is measured by AFM greater than this value due to
species like nitrogen, oxygen, or water molecule between graphene layer and SiO2
surface in air 89
.
73
Figure 5.4 shows characterization results of graphene sheets produced via
mechanical exfoliation method. It is clearly seen from optical microscope image, region
marked as 1 is single layer graphene while region 2 is multilayer graphene sheet (Figure
5.4 (a)). During this study, Raman spectroscopy revealed these results. Raman spectrum
and mapping are taken from Renishaw inVia Reflex Raman Microscope and
Spectrometer with 532 nm laser excitation. Finally, Figure 5.4 (d) shows AFM
(supplied by NanoMagnetics Instruments Ltd) measuremens of the single layer
graphene. As it is expected, thickness of graphene sheet is more than 0.34 nm since
measurement is performed in air condition. Scanning is employed by AFM tapping
mode with 5 µ/s .
Figure 5.4: (a) Optical microscope image of graphene sheets produced via
mechanical exfoliation method. Raman spectrum of (b) single and (c) multi layer
graphene film. (d) AFM topography of single layer graphene sheet.
5.5 Graphene Hall Sensor
74
5.5.1 Fabrication of GHP
Throughout this work, graphene produced by both mechanical exfoliation and
CVD growth method are used for fabrication of graphene Hall sensor. In the earlier
stage of this study, mechanical exfoliated graphene is used for device fabrication.
Optical microscope images of each step are given in Figure 5.5. Process flow of
fabrication as follows:
1. Once graphene to be fabricated is determined, pattern consists of contact
pads and markers are defined for EBL step by implementing image
reversal photolithography and lift–off process (Figure 5.5(a)).
2. The sample is spin–coated with 950 K PMMA–C2 at 6,000 rpm (45 s) and
baked at 180° C for 5 minutes before HSQ coating as protective layer.
Because, it is difficult to remove HSQ from graphene surface after e–beam
and oxygen plasma process.HSQ is baked with two steps: first, it is baked
at 150° C for 2 min subsequently, at 220° C for 5 min.
3. The sample exposed to 300 µC/cm2 dose at 30 kV for Hall cross definition
by using NPGS system and exposed sample is developed with AZ 726
MIF for 70 s.
4. PMMA and graphene layers are etched by oxygen plasma and the sample
is dipped in ACE to remove both HSQ and PMMA layers.
5. Metal pads between graphene Hall cross and contact pads are defined by
EBL and metal evaporation (3nm Cr/ 30 nm Au).
6. Mesa of the sensor is patterned by positive photolithography and etched by
SF6/O2 plasma.
75
Figure 5.5: The optical microscope images of (a) mechanical exfoliated graphene (b)
after contact pad metallization, (c) O2 plasma etching of graphene (d) taking contact
from graphene to contact pads by implementing EBL and (e) mesa step formation.
Although graphene produced by mechanical cleavage method has superior
electronic properties than CVD growth graphene, in each time only one sensor can be
manufactured due to small sized graphene flake. Fabrication yield was increased by
using large area graphene. For this purpose, Graphene Hall probes have been fabricated
using commercially available 1×1 cm CVD single layer graphene on 285 nm SiO2/Si
wafers supplied by Graphene Supermarket. Approximately, 50 individual chips were
successfully fabricated by using this size of graphene covered substrate. Fabrication
process flow is described as following:
1. Fabrication starts with positive photolithography in order to define Hall
cross patterns which were protected by photoresist (Figure 5.6 (a)). This
was followed by oxygen gas plasma RIE (20 sccm O2 flow, in a 100 W
76
RIE power, 37.5 mTorr, for 9 s) in order to etch the unprotected parts of
CVD–graphene sheet. Next, the sample is immersed into ACE to remove
photoresist. Figure 5.6 (b) presents patterned graphene.
Figure 5.6: The optical microscope images of after (a) Hall cross definition by
implementing positive photolithography and (b) after O2 plasma process and
removing photoresist.
2. Entire surface of purchased 1 cm × 1 cm substrate is covered by graphene.
Although it is an advantage for high yield fabrication, it can cause
adhesion problem after metal evaporation. To prevent this, mask pattern
normally utilized for recess etch process is used in order to etch graphene
region to be used for wire bonding process. Optical microscope image of
the developed sample with AZ 726MIF after positive photolithography by
using recess mask pattern is given in Figure 5.7 (a). This step plays crucial
role in bonding. Because, if organic residue left under metal structure, it
can cause peeling off contact pads from the surface during wire bonding
step.
3. Optical lithography was used to pattern for contact pads on graphene, and
it was followed by thermal evaporation of 10 nm Cr/200 nm Au
evaporation and lift–off (Figure 5.7 (b)).
(a) (b)
77
Figure 5.7: The optical microscope images (a) of the developed sample with AZ
726MIF after positive photolithography by using recess mask pattern (b) after
contact pads metallization.
4. The mesa which serves as AFM tip was obtained by etching SiO2/Si wafer
∼830 nm in an inductively coupled plasma reactive ion etching (ICP–
RIE) system (Figure 5.8 (a)). The dry etching parameters are given in
previous chapter.
5. GHP patterns on the wafer were diced into individual chips and glued
with low temperature epoxy on one of the tines of the QTF which is used
as a force sensor in a wide temperature range. Next, QTF was glued to the
ceramic plate on non–magnetic PCB and electrical connections of GHP to
PCB were established with 25 μm gold wires using wedge wire bonder
(Figure 5.8 (b)).
(a) (b)
(a) (b)
78
Figure 5.8: Optical microscope images of GHP. (a) After mesa formation by
etching 830 nm SiO2/Si layer. (b) Side view of Hall sensor glued on 1 × 1 cm
PCB.
5.5.2 Characterization of GHP
Optical microscope is the simplest and quickest way for graphene
characterization. It is particularly important for mechanical exfoliated graphene as
scanning graphene sheet is initially done with it. Nowadays, it is very reliable tool
which distinguish mono and multilayer graphene accurately for all production methods.
Optical microscope is almost used in each step of Hall sensor fabrication from CVD
growth graphene film to understand how process proceeds. On the other hand, Raman
spectrum which provides obtaining quantitative measurements determines the number
of graphene layers and uniformity of graphene thickness over the GHP surface. Before
the fabrication we have checked graphene substrate not only by using optical
microscope but also Raman spectrum (Figure 5.9). After fabrication is completed
Raman spectrum of the center of Hall Cross which is shown on bottom right of Figure
5.10 verifies that the GHP was fabricated from single layer graphene and the left of this
figure shows optical microscope image of fabricated sensor. Moreover, it was reported
that I2D/IG ≥ 2 is associated with the presence of monolayer graphene 90
. In top right of
Figure 5.10, large scale (14×14 μm) Raman map of intensity ratio gives further
evidence.
50 µ
79
Figure 5.9: (a) The optical microscope images of as received CVD growth graphene on
285 nm SiO2/Si wafers. (b) The Raman spectrum of as received CVD growth graphene
from different three points.
Figure 5.10: Optical microscope image of GHP (left).Raman map of (I2D/IG ≥ 2)
intensity ratio measured in a 14 µm × 14 µm area of GHP (right–top). Single Raman
spectrum taken from center of the GHP (right–bottom).
Electrical and magnetic characterization of GHP was performed in Helium
exchange gas and vacuum conditions by using LT–SHPM initially at room temperature.
Graphene exhibits p–type behavior due to environmental effects and water absorption
when it is exposed to air and resist residues from fabrication process 91, 92, 93, 94
. This
results in a high charge carrier density and low Hall coefficient. Thereby, we tried to
overcome this problem by performing magnetic imaging and characterization in vacuum
condition as we did not have a back gate contact. Most of the experimental data for
electrical characterization is performed in NanoMagnetics Instruments Ltd by using
cryostat supplied by Cryogenic Limited. All experimental results, which will be
80
presented following parts, belong to a specific GHP which has been utilized for
magnetic imaging. A uniform external magnetic field (Bext) was applied to GHP up to
5,000 G and the generated Hall voltage (VHall) was simultaneously measured under
constant driving current. Figure 5.11 shows the linear relationship
between VHall and Bext for a 2 μA driving current at room temperature.
Figure 5.11: Hall voltage response of GHP to applied magnetic field at room
temperature.
Room temperature series resistance and Hall coefficient of GHP were measured to
be 82.20 kΩ and 0.18 Ω/G respectively in vacuum condition for 3 µA drive current. A
spectrum analyzer was used to obtain a noise spectrum of the single layer graphene Hall
probe. We determined minimum detectable magnetic field (B min) from voltage noise
spectrum, which is defined as equation:
(5.2)
where VNoise, RH, IH and G are the total measured voltage noise, Hall coefficient, Hall
current, and Hall probe pre–amplifier gain respectively 95
. The Hall probe pre–amplifier
gain was 1001 for the system used in this experiment. Figure 5 (a) represents Bmin as a
function of frequency for different Hall currents in zero magnetic field and zero back
gate voltage 96
.
(5.3)
81
In this equation, kB is Boltzmann’s constant, T is temperature, Rs is the resistance
of the sensor and Δf is measurement bandwidth 96
.
Figure 5.12: (a) Bmin for different drive current as a function of frequency at 300
K. (b) Bmin for different temperature as a function of frequency for 5 µA Hall current
The graphs in Figure 5.13 (a), (b) and (c) compare the Bmin in vacuum and in
exchange gas (He) for different current values. As seen from these graphs, for 1 µA and
2 µA current values, Bmin is slightly better than in exchange gas condition whereas Bmin
for 3 µA driving current in vacuum condition can detect clearly smaller magnetic field
which provides better magnetic field resolution. Temperature characterization of GHP is
performed in exchange gas by Teslatron PT cryostat located in SUNUM. The variation
of serial resistance of two arms of Hall crosses as a function of temperature has been
shown in Figure 5.13 (d). As seen from this figure, resistance is decreasing with
increasing temperature. This result is consistent with previous work 97
.
(a) (b)
82
Figure 5.13: Comparison of Bmin in vacuum and in Helium exchange gas for
different current values. (d) Serial resistance of two arms of Hall crosses as a function of
temperature.
Temperature dependency of RH is also investigated. For this purpose, 2 T external
magnetic fields and ±2 µA driving current are applied to GHP in exchange gas.
Generated VHall out is measured by SPM electronic unit and software. RH is calculated
for different temperature ranging from 1.5 K to300K. As is seen from graphs in Figure
5.14, RHall is slightly increased by increasing temperature. After 100 K it is almost
constant with increasing temperature and current with positive polarity cause obtaining
slightly higher RHall.
(a) (b)
(c) (d)
83
Figure 5.14: The variations of RHall as a function of temperature for ± 2 µA driving
current.
5.5.2 Imaging NdFeB Demagnetized Magnet by GHP
We operated a LT–SHPM system manufactured by NanoMagnetics Instruments
Ltd. in AFM tracking mode, in order to acquire topography and magnetic image of a
NdFeB demagnetized magnet surface in a wide range of temperature (3–300K).
Resonance frequency of QTF used in this experiment was around 31.9 kHz when one
prong is free and the other is fixed to the ceramic plate on top of sensor holder PCB.
After GHP was glued on top of QTF, this resonance frequency decreased to
approximately 17 kHz due to mass of chip and glue (Figure 516).
(a) (b)
84
Figure 5.15: Measured resonance frequency at 300 K (a) before and (b) after GHP
was glued on QTF.
The sample is tilted ~1o
with respect to GHP to ensure that the corner of Hall
sensor mesa is the closest point to the sample surface, which is used as an AFM tip. The
sample was brought in to close proximity of GHP by means of slip–stick coarse
approach mechanism. The tip–sample interaction results in a shift in resonance
frequency of the QTF which is measured by a PLL for AFM feedback 45
. Figures 5.16
(a) and (c) show the SHPM images of magnetic domains of NdFeB demagnetized
magnet at room temperature. The magnetic field variations along a horizontal line of
these images are also shown in Figure 5.16 (b) and (d). The scan parameters are 5 µm/s
scan speed, 50 µm × 50 µm scan area (Figure 5.16 (a)) and 40 µm × 40 µm scan area
(Figure 5.16 (c)) , 512 pixels × 512 pixels resolution and 2 µA drive current for
measured 0.22 Ω/G RH.
(c)
(d)
85
Figure 5.16: (a) and (c) magnetic images of NdFeB demagnetized magnet by
using GHP at 300 K. (b) and (d) magnetic field variations along the line drawn on
images.
We have also investigated the performance of GHP at cryogenic temperatures.
Surface topography and magnetic image of NdFeB sample for IHall= 2 µA at 126 K are
shown in Figure 5.17 (a) and (b), respectively. Figure 5.17 (c) and 5.17 (d) show the
SHPM image of the same sample at 3 K for IHall= 2 µA and IHall = –2 µA.
Figure 5.17: (a) Topographic image of NdFeB. (b) 50 µm × 50µm LT–SHPM
magnetic image of NdFeB at 126 K. 14x14 µm LT–SHPM magnetic image of the same
sample for (c) IHall = +2 µA (d) IHall = –2 µA at 3 K.
5.6 Conclusion and Discussion
In this work, Graphene Hall sensor produced, fabricated, characterized and used
for magnetic imaging. Initially, mechanical exfoliated graphene is used for fabrication.
86
Because not only has it high mobility, but also it is very easy to produce. Even though
mechanically exfoliated graphene provides superior properties than CVD in terms of
both mechanically and electronically, fabrication yield is very low with this method. We
manufacture a few of Hall sensors produced via mechanical cleavage method. Due to
the contact problems arising from several reasons (residual resist between contact pads
and patterned graphene or fabrication process issues), resistance of Hall cross was
measured drastically high around MΩ order. Therefore, we preferred to use CVD
growth graphene which provides large scale graphene film and high yield production.
Raman spectroscopy and optical microscopy is used for characterization of the
sensor during manufacturing process. Raman spectrum results show as received CVD
samples are graphene monolayer. However, on the surface of the graphene film (without
employing any process) there are discontinuities and bilayer or multilayer graphene
islands. White areas in SEM image of as received CVD graphene sheet show cracks on
the films (Figure 5.18). Submicron sized cracks cannot be seen in optical microscope
which cause electrical contact problem. Another problem is that there are a lot of
Bismuth or multilayer graphene islands which is marked as red circles in the same
figure. These multilayer graphene regions result from impurities on copper metal which
act as nucleation sites 98
. Since activation energy is higher on impurities, carbon atoms
dissolve more on these regions and cause forming multilayer graphene regions 99
.
Film discontinuity
Multiple layer
graphene
islands
87
Figure 5.18: SEM image of as received CVD graphene sheet which shows cracks and
discontinuity on the graphene film surface.
Once fabrication is completed, electrical characterization for different condition is
done by using the SHPM. Voltage noise spectrum which limits Bmin is measured by
spectrum analyzer. Effect of temperature and driving current on Bmin is investigated.
According to magnetic flux noise spectrum results Bmin decreases with increasing drive
current and it is found to be 0.20 G/ for a 3 µA drive current at 1 kHz (Figure 5.12
(a)). We also measured the magnetic field noise of GHP at 5 K, 77 K and 300 K for
IH =5 µA and calculated Johnson noise levels for each temperature for the same current
value. Smaller magnetic fields can be detected at lower temperatures by GHP, as shown
in Figure 5.12 (b). Furthermore, serial resistances of Hall sensor change between 60–80
kΩ depending on temperature.
We have successfully used the Graphene Hall Probes for magnetic imaging in 3–
300 K range for the first time in SHPM. This study has demonstrated that graphene is an
alternative material to be used for magnetic imaging. Additionally, it is also possible to
decrease Hall cross dimensions to a few tens of nanometer by employing electron beam
lithography, so that spatial resolution of the sensor can potentially reach sub–100 nm
resolution. Currently, we are working on the improvement of our results by reducing
Hall cross area and removing residues arising from the fabrication process for both
higher magnetic field and spatial resolution.
88
CHAPTER 6
BISMUTH HALL SENSOR for SHPM
6.1 Introduction
As a pentavalent metal, Bismuth has attracted much attention of researchers in
sensing applications due to its remarkable electrical properties arising from its
anisotropic Fermi level in surface, relatively large mean free path in the order of
microns and small effective mass 100, 101, 102
. The exact equal number of free electrons
and holes on the surface, along with a low scattering rate gives rise to the
magnetoresistance, which is known to be very high in semi–metals 103
. In addition to its
electrical and magnetic properties, Bismuth has an extremely low melting point of
271.3º C and its thermal conductivity is one of the lowest among the metals 104
.
The most important element of an SHPM is the Hall sensor that is sensitive to
perpendicular component of the magnetic field on surface of sample. This part can be
fabricated by various materials. However, to achieve a high magnetic resolution,
materials with high carrier mobility and low carrier density are preferred to help
maximize the Hall coefficient and signal to noise ratio. On the other hand, size of the
active area is the key factor that determines spatial resolution of microscope 2. As being
a semimetal with a concentration of five orders of magnitude lower than metals and
with negligible surface charge depletion effect, Bismuth is an alternative material for
Hall probe in the area of magnetic field sensing elements 9.
The first study of Bismuth thin film as Hall sensor was carried out by Broom et
al. in 1962 where they scanned 1,000 or 2,000 Å thick 15 mm x15 mm superconducting
thin and lead films using 100 µm Hall cross. By placing the probe 0.05 mm away from
film surfaces, a current of 2 mA was applied through the probe producing a Hall voltage
89
of 0.6 μV/G at liquid Helium temperatures. The thermal noise was measured as ± 0.1 G
and the resolutions were 0.2 and 0.25 mm, respectively 105
.
The smallest fabricated Bismuth hall probe so far was reported in 2004 by Sandhu
et al., where nano–sized Bismuth sensor was used in a room temperature SHPM, and
images of magnetic domains of low coercivity garnet thin film samples were recorded 9.
This 50x50nm nano–Bismuth hall probe fabricated by optical lithography and focused
ion beam milling were 60 nm in thickness and was located about 4 μm away from the
tip yielded a noise level of 0.85 G/ . The logic behind employing FIB technique in
here was to assure thicker hall crosses that allowed lower resistance values and hence
less Johnson noise. Room temperature Hall coefficients of nano–Bismuth hall probes
were measured as 4.0 x10-4
Ω/G. Feasibility of nano–Bismuth hall probes were tested
on a 5 mm thick crystalline Bismuth substituted iron garnet thin film. Images were
obtained with a drive current of 43 µA and at a tilt angle of 1.2° by measuring the
changes in Hall voltages coming from stray magnetic fields. The SEM image of
fabricate Hall sensor and magnetic imaging of iron garnet thin film taken by fabricated
50 nm Bismuth Hall sensor at room temperature are given in Figure 6.1.
Figure 6.1: The SEM images of 50nm × 50nm nano-Bismuth hall probe. SHPM
image of crystalline Bismuth substituted iron garnet thin film taken by fabricated 50 nm
Bismuth Hall sensor at room temperature.
By considering the studies mentioned above, the scope of this chapter is to
enhance the spatial resolution of SHPM images by employing Bismuth hall sensors and
(a)(b)
90
to observe its behavior at low temperatures. In this context, present chapter describes
optimized fabrication process of Bismuth Hall sensor for SHPM applications in detail
by covering all the fabrication steps, problems raised throughout the process and
solutions developed. We have fabricated Hall crosses in three different sizes of 100 nm,
200 nm and 500 nm, respectively. Along with imaging of NdFeB demagnetized magnet
using these sensors, temperature characterization was also carried out. Since the spatial
resolution is directly proportional to cross size, we reduced the size of the probe
fabricated was to 100 nm and gathered successful images at 300 K. To our knowledge,
this study was the first reported imaging at low temperature using Bismuth hall sensor.
6.2 Bismuth Hall Sensor
6.2.1 Fabrication of Bismuth Hall Sensor
Two different optical mask pattern designs are used in the fabrication of Bismuth
Hall sensors. One of them has 30 µm × 50 µm active area with asymmetric contact pads
whereas the other design has 10 µm × 10 µm areas with symmetrically ordered pads
(Figure 6.2). To reduce area shortens exposing time while EBL is employed. It is
particularly important when NPGS system is used since it works with low current
values.
Figure 6.2: The optical microscope image of (a) symmetric (b) asymmetric ohmic
contacts after 10 nm Cr and 100 nm Au evaporation and lift–off process.
20µ 20µ (a) (b)
91
Many problems have been encountered during fabrication process of Bismuth
Hall sensor. Some of them with their solutions are discussed previously in Chapter 5.
Optimized process recipe is given as following:
1. Fabrication of Bismuth Hall sensor starts with wafer cutting into 1.2 cm ×
1.2 cm pieces. Then, pieces are cleaned by standard cleaning procedure as
mentioned previous chapter.
2. After spin coating with AZ 5214E (4,000 rpm for 40 s), the sample is
baked 105 ° C for 1 minute. Positive optical lithography is implemented
by using recess etch mask with zigzag shapes for 2 s illumination by using
Midas mask aligner. Then, the sample is developed with AZ 726 MIF and
washed by DI water. Hard bake is done at 120 ° C for 2 minutes on a hot
plate and the sample is etched in HCl:H2O2:H2O solution at volume ratios
of 4:7:55 for 90 s which provides around 1.5 µm etch thickness (Figure
4.12).
3. Contact pads are defined by image reversal recipe. First, the sample is
spun at 6,000 rpm for 45 s. Then, it is baked at 90 ° C for 2 minutes
(prebake). The sample is illuminated for 2 s by Midas mask aligner and
post bake is done at 115° C for 2 minutes on a hot plate. Flood exposure
(without mask) is employed for 11.6 s and the sample is developed with
AZ 726 MIF for 25 s. Next, 10 nm Cr and 100 nm Au layer is evaporated
by spin coater and lift off process is performed by immersing the sample
in ACE solution (Figure 6.2).
4. Once contact pads are defined, oxygen plasma etching is performed to the
samples in order to etch resist residue left from optical lithography steps.
The oxygen plasma parameters are 20 sccm O2 flow, in a 100 W RIE
power, 37.5 mTorr pressure for 5 minutes. Then, bi–layer resist recipe is
used (First layer is coated with 495 K PMMA–C2 at 5,000 rpm for 50 s
and baked at 180 ° C for 40 minutes. The second layer is spun with 950 K
PMMA–A2 at 4,000 rpm 50 and placed on a hot plate at 180 ° C for 40
minutes.) Patterns are designed by AutoCAD program and inserted in
92
Layout BEAMER to convert this design file into a readable format in
Vistec machine and utilize from PEC function of this program. 500μC/cm2
base dose is used to obtain 50 nm Hall cross size (Hall cross area is
exposed with 800 μC/cm2) where specimen current is 180 pA beam
current and step size and resolution are 1 nm. Development is done 1:3
MIBK:IPA solution for 60 s. After the sample is lifted–off and 3 nm
chromium and 30 nm Bismuth are thermally evaporated. Figure 6.3 shows
SEM images of 50 nm Bismuth Hall cross at different magnifications by
using this recipe.
Figure 6.3: Then SEM images of 50 nm Bismuth Hall cross at different magnifications.
5. After Hall cross definition, the sample is characterized by SEM and
successfully patterned elements are determined and their resistance are
measured by multimeter in the probe station. Patterns that are in a
magnitude of a few kΩ resistances are selected as suitable candidates for
Hall sensor application. However, we sometimes measure drastically high
resistance values around MΩ scale even though SEM images suggest that
Hall cross area are successfully lifted–off. It has been realized that there
occurs a gap between Au contact pads and legs of Bismuth Hall cross. As
a solution, we apply a second EBL process in order to assure that there is a
full contact. Further EBL processes can be required until contact problem
93
is completely resolved. Figure 6.4 (a) shows optical microscope images
after the first time EBL is employed for connection problem (b), (c) and
(d) show SEM images after third EBL process applied at different
magnifications and different points of view.
Figure 6.4: (a) The optical microscope images after the first time EBL is
employed for connection problem (b), (c) and (d) show SEM images after third EBL
process applied at different magnifications and different points of view.
1. For mesa step formation, the sample is spun with 950 K PMMA–A2 and
its copolymer MMA (8.5) EL11. Then, EBL is employed with 550 μC/cm2
base dose. Development is performed with 1:3 MIBK:IPA solution for 60
s and post bake is done at 100 ° C on a hot plate. Etching solution is
prepared with H2S4O4:H2O2:H2O at volume ratios of 1:8:40 for 60 s. by
using this recipe around 1.5 µm etch thickness is obtained successfully
(Figure 6.5).
(a) (b)
(c) (d)
94
2. The sample is diced by Disco dicer and glued on top of QTF which has
been already attached to PCB. Finally electrical connection is taken by
wedge bonding process with 25 µm gold wire.
Figure 6.5: The SEM images of 50 nm Bismuth Hall sensor after mesa formation.
6.2.2 Characterization of Bismuth Hall Sensor
We carried out electrical characterization of Bismuth Hall probes with different
sized (200 nm and 500 nm) at various temperatures and driving current. The SEM
image of fabricated 500 nm Bismuth Hall sensor is given in Figure 6.6 (a). Room
temperature serial resistance and RH of 500 nm Bismuth Hall sensor are measured as
18.8 kΩ and 3.2 × 10-4
Ω/G respectively. Noise spectrum of fabricated sensors are
measured by spectrum analyzer and converted into magnetic field unit in order to
determine Bmin at different conditions. Figure 6.6 (b), (c) and (d) compare Bmin as a
function of frequency for different driving currents at 300 K, 77 K and 4 K temperature
respectively. The graphs in Figure 6.6 show that Bmin decreases as driving current is
increased for each temperature. This is correlated with theory since Bmin is inversely
proportional with driving current as is shown equation 5.2. Same trend is observed for
200 nm Bismuth Hall sensor by means of relation between Bmin and drive current
(Figure 6.7). Room temperature serial resistance and RH of 200 nm Bismuth Hall sensor
are measured as 13.8 kΩ and 4.2 × 10-4
Ω/G respectively.
95
Figure 6.6: (a) The SEM image of 500 nm Bismuth Hall sensor. Bmin as a function
of driving current at different temperature values (b) 300 K (c) 77 K and (d) 4 K.
300 K
77K 4 K
(a)(b)
(c) (d)
300 K
77 K 4 K
(d)
(b)
(c)
(a)
96
Figure 6.7: (a) The SEM image of 200 nm Bismuth Hall sensor. Bmin as a function of
driving current at different temperature values (b) 300 K (c) 77 K and (d) 4 K.
6.2.3 Imaging NdFeB Demagnetized Magnet by Bismuth Hall Sensor
NdFeB demagnetized magnet is scanned with 500 nm, 200 and 100 nm Bismuth
Hall sensors at different temperatures by using QTF AFM feed–back mode. For this
purpose, fabricated Hall sensor is attached to the microscope and NdFeB is tilted around
1o with respect to Hall sensor in order to use mesa corner as AFM tip. Figure 6.8 shows
SHPM images of NdFeB demagnetized magnet at 300 K, 77 K and 4 K for 50 µA
acquired by using 500 nm Bismuth Hall sensor. Resonance frequency and quality factor
of QTF are measured as 10,170 Hz and 80 at room temperature. Δf is set to 10 Hz
during scan. The scan areas of images are 30 µm × 30 µm for 300 K and 77 K whereas
it is set to 20 µm × 20 µm for 4 K.
T =300 K I = 50 µA T =300 K I = - 50 µA
T =77 K I = 50 µAT =4 K I = 50 µA
(a) (b)
(c)(d)
97
Figure 6.8: SHPM images of NdFeB demagnetized magnet at (a) 300 K, (b) 77 K, (c) 4
K taken by using 500nm Bismuth Hall sensor for 50 µA driving current and at
(d) 4 K for –50 µA driving current.
Figure 6.9 shows magnetic image and cross sections taken from the same image
of NdFeB demagnetized magnetic sample at room temperature by using Bismuth Hall
sensor with 200 nm cross sized. The sample is tilted 1.25o in this scan and room
temperature serial resistance of the sensor is measured as 13.86 kΩ. Topography and
SHPM image (30 µm × 30 µm scan area) of NdFeB sample at 77 K are given in Figure
6.10.
Figure 6.9: SHPM images of Nd,FeB demagnetized magnet at 300 K (left) by using 200
nm Bismuth Hall sensor for 500 µA drive current and graph of cross sections
(right top and bottom).
98
Figure 6.10: (a) Topography and (b) SHPM image of NdFeB sample at 77 K by using
200 nm Bismuth Hall sensor for 500 µA drive current.
Magnetic image of NdFeB with scan area 20 µm × 20 µm at 4 K is also obtained
by using 200 nm Bismuth Hall sensor (Figure 6.11). Serial resistance of the probe at 4
K is measured as 13.86 kΩ.
Figure 6.11: (a) Topography and (b) SHPM image of NdFeB sample at 4 K by using
200 nm Bismuth Hall sensor for 500 µA drive current.
Finally, fabricated 100 nm Bismuth Hall sensor which has 13.86 kΩ series
resistance are used for magnetic imaging of NdFeB demagnetized magnet at room
temperature (Figure 6.12).
(a) (b)
(a) (b)
99
Figure 6.12: (a) SHPM image of NdFeB sample at 300 K by using 100 nm Bismuth
Hall sensor for 500 µA drive current.(b) Magnetic field variations along the line
drawn on images.
5.2 Conclusion
In this chapter, fabrication, electrical characterization and temperature dependency
of Bismuth Hall sensor is explained in detail. Hall crosses with 100 nm, 200 nm and
500 nm in sizes were fabricated and SHPM images of NdFeB demagnetized magnet
were recorded using these sensors with QTF AFM feedback mode. Bmin as a function of
frequency is deduced from voltage noise spectrum for different temperatures and
various driving currents. The smallest Hall sensor used in SHPM was 100 nm. We
could obtain magnetic imagining with this Hall sensor only at room temperature.
100
CHAPTER 7
CONCLUSION AND FUTURE WORK
There are two novel outcome of this thesis. The first one is that GHP was used in
SHPM application for the first time. Although GHP is very well characterized in the
literature in terms of its electrical properties, our study has moved graphene a step
further in its applications. To achieve this, we fabricated GHP by implementing
conventional lithographic techniques. Temperature dependence of electrical
characteristics was determined and an increasing trend in Bmin is observed with
increasing temperature and decreasing driving current. RH of GHP was measured as
0.18 Ω/G for 3 μA Hall current at room temperature in vacuum, whereas Bmin was found
to be 0.20 G/ at 1 kHz. These results show that graphene has become a promising
candidate as Hall sensor material in SHPM application. We have fabricated GHP by
using graphene produced by two different methods; mechanical exfoliation and CVD
growth. However, due to contact problems arising from fabrication process, we could
not obtain expected results using exfoliated graphene. By using CVD graphene we
could overcome this problem. Although CVD graphene increased the sensor production
yield, the noise levels in the sensors were relatively high, as the CVD graphene is
polycrystalline. Electrical properties can be further improved by using large flake,
single crystal graphene with minimal defects. To retain better electrical properties of
CVD graphene, one can use CVD graphene growth on platinum and transferred by
bubbling transfer method.
We have also fabricated Bismuth nano Hall sensors by EBL and lift–off process
and imaged NdFeB demagnetized magnet at wide range temperature range 4–300K. –
During the fabrication of Bismuth Hall sensor, in spite of many difficulties in
101
fabrication process, we were successful to fabricate the devices down to 100 nm by
using advanced lithographic techniques and obtained SHPM images. By using EBL, we
were able to reduce Hall cross size down nanometer scale which resulted in obtaining
better spatial resolution of SHPM images. It might be possible to produce Hall probes
with less than 25 nm sizes by using EBL, which might allow obtaining images with
higher spatial resolution.
102
BIBLIOGRAPHY
1 G. Binning, H. Rohrer, C. Gerber, and E. Weibel, "Surface Studies by
Scanning Tunneling Microscopy," Phys Rev Lett 49 (1), 57-61 (1982).
2 A. Oral, M. Kaval, M. Dede, H. Masuda, A. Okamoto, I. Shibasaki, and A.
Sandhu, "Room-temperature scanning Hall probe microscope (RT-SHPM)
imaging of garnet films using new high-performance InSb sensors," Ieee T
Magn 38 (5), 2438-2440 (2002).
3 M. Dede, Doctoral Dissertation, Bilkent University, 2008.
4 R. Akram, M. Dede, and A. Oral, "Imaging capability of pseudomorphic high
electron mobility transistors, AlGaN/GaN, and Si micro-Hall probes for
scanning Hall probe microscopy between 25 and 125 degrees C," J Vac Sci
Technol B 27 (2), 1006-1010 (2009).
5 A. Sandhu, H. Masuda, A. Oral, and S. J. Bending, "Direct magnetic
imaging of ferromagnetic domain structures by room temperature scanning hall
probe microscopy using a bismuth micro-hall probe," Jpn J Appl Phys 2 40 (5B),
L524-L527 (2001).
6 A. K. Geim and K. S. Novoselov, "The rise of graphene," Nat Mater 6 (3), 183-
191 (2007).
7 V. Panchal, K. Cedergren, R. Yakimova, A. Tzalenchuk, S. Kubatkin, and
O. Kazakova, "Small epitaxial graphene devices for magnetosensing
applications," J Appl Phys 111 (7) (2012).
8 A. Sandhu, H. Masuda, K. Kurosawa, A. Oral, and S. J. Bending, "Bismuth
nano-Hall probes fabricated by focused ion beam milling for direct magnetic
imaging by room temperature scanning Hall probe microscopy," Electron Lett
37 (22), 1335-1336 (2001).
9 A. Sandhu, K. Kurosawa, M. Dede, and A. Oral, "50 nm Hall sensors for
room temperature scanning Hall probe microscopy," Jpn J Appl Phys 1 43 (2),
777-778 (2004).
10 H. A. Mohammed and S. J. Bending, "Fabrication of nanoscale Bi Hall sensors
by lift-off techniques for applications in scanning probe microscopy," Semicond
Sci Tech 29 (8) (2014).
11 A.F.J. Levi, Applied Quantum Mechanics. (Cambridge University Press, 2012).
12 K. Sergei and G. Alexei, Scanning Probe Microscopy : Electrical and
Electromechanical Phenomena at the Nanoscale (Springer, 2007).
13 D. S. Klaus, Handbook of Nanophysics: Principles and Methods. (CRC Press,
2010).
103
14 G. Binnig, C. F. Quate, and C. Gerber, "Atomic Force Microscope," Phys Rev
Lett 56 (9), 930-933 (1986).
15 Y. Seo and W. Jhe, "Atomic force microscopy and spectroscopy," Rep Prog
Phys 71 (1) (2008).
16 X. Ning and L. King, Nano Optoelectronic Sensors and Devices:
Nanophotonics from Design to Manufacturing (William Andrew, 2011).
17 H. Peter, Advances in Imaging and Electron Physics. (Academic Press, 2003).
18 L. Luan, Doctoral Dissertation, Stanford University, 2011.
19 P. Grutter, T. Jung, H. Heinzelmann, A. Wadas, E. Meyer, H. R. Hidber,
and H. J. Guntherodt, "10-Nm Resolution by Magnetic Force Microscopy on
Fendb," J Appl Phys 67 (3), 1437-1441 (1990).
20 G. Han, Y. H. Wu, and Y. K. Zheng, "Dual-synthetic magnetic force
microscopy tip and its imaging performance," J Magn Magn Mater 315 (1), 46-
52 (2007).
21 Ü. Koray, The Master of Science, Bilkent University, 2005.
22 S. J. Bending, "Local magnetic probes of superconductors," Advances in
Physics 48 (4), 449-535 (1999).
23 S. Morita, Roadmap of Scanning Probe Microscopy. (Springer, 2007).
24 "http://www.nanomagnetics-inst.com/tr".
25 A. Sandhu, H. Masuda, A. Oral, and S. J. Bending, "Room temperature sub-
micron magnetic imaging by scanning Hall probe microscopy," Jpn J Appl Phys
1 40 (6B), 4321-4324 (2001).
26 B. Aktaş, L. Tagirov, and F. Mikailov, Magnetic Nanostructures. (Springer,
2007).
27 S. Mouaziz, G. Boero, R. S. Popovic, and J. Brugger, "Polymer-Based
Cantilevers With Integrated Electrodes," J. Microelectromech. Syst 15 (4), 890-
895 (2006).
28 T. Schweinbock, D. Weiss, M. Lipinski, and K. Eberl, "Scanning Hall probe
microscopy with shear force distance control," J Appl Phys 87 (9), 6496-6498
(2000).
29 A. J. Brook, S. J. Bending, J. Pinto, A. Oral, D. Ritchie, H. Beere, M.
Henini, and A. Springthorpe, "Integrated piezoresistive sensors for atomic
force-guided scanning Hall probe microscopy," Appl Phys Lett 82 (20), 3538-
3540 (2003).
104
30 B. K. Chong, H. Zhou, G. Mills, L. Donaldson, and J. M. R. Weaver,
"Scanning Hall probe microscopy on an atomic force microscope tip," J Vac Sci
Technol A 19 (4), 1769-1772 (2001).
31 A. J. Brook, S. J. Bending, J. Pinto, A. Oral, D. Ritchie, H. Beere, A.
Springthorpe, and M. Henini, "Micromachined III-V cantilevers for AFM-
tracking scanning hall probe microscopy," Journal of Micromechanics and
Microengineering 13 (1), 124-128 (2003).
32 J. Curie and P. Curie, "Development, via compression, of electric polarization
in hemihedral crystals with inclined faces," Bull. Soc. minéral. Fr 3 (90), 383-
386 (1880).
33 M. Shahinpoor and H. J. Schneider, Intelligent Materials. (Royal Society of
Chemistry, Cambridge, 2008).
34 V. Medved, Measurement of Human Locomotion. (CRC Press, 2000).
35 A. A. Kosterev, F. K. Tittel, D. V. Serebryakov, A. L. Malinovsky, and I. V.
Morozov, "Applications of quartz tuning forks in spectroscopic gas sensing,"
Rev Sci Instrum 76 (4) (2005).
36 P. Gunther, U. Fischer, and K. Dransfeld, "Scanning near-Field Acoustic
Microscopy," Applied Physics B-Photophysics and Laser Chemistry 48 (1), 89-
92 (1989).
37 K. Karrai and R. D. Grober, "Piezoelectric Tip-Sample Distance Control for
near-Field Optical Microscopes," Appl Phys Lett 66 (14), 1842-1844 (1995).
38 M. Todorovic and S. Schultz, "Magnetic force microscopy using nonoptical
piezoelectric quartz tuning fork detection design with applications to magnetic
recording studies," J Appl Phys 83 (11), 6229-6231 (1998).
39 F. D. Callaghan, X. Yu, and C. J. Mellor, "Variable temperature magnetic force
microscopy with piezoelectric quartz tuning forks as probes optimized using Q-
control," Appl Phys Lett 87 (21) (2005).
40 F. J. Giessibl, "High-speed force sensor for force microscopy and profilometry
utilizing a quartz tuning fork," Appl Phys Lett 73 (26), 3956-3958 (1998).
41 F. J. Giessibl, "Atomic resolution on Si(111)-(7x7) by noncontact atomic force
microscopy with a force sensor based on a quartz tuning fork," Appl Phys Lett
76 (11), 1470-1472 (2000).
42 Y. Qin, Doctoral Dissertation, Purdue University, 2007.
43 T. Ludwig, "Casimir force experiments with quartz tuning forks and an atomic
force microscope (AFM)," Journal of Physics a-Mathematical and Theoretical
41 (16) (2008).
105
44 R. Oria, J. Otero, L. Gonzalez, L. Botaya, M. Carmona, and M. Puig-Vidal,
"Finite Element Analysis of Electrically Excited Quartz Tuning Fork Devices,"
Sensors 13 (6), 7156-7169 (2013).
45 R. Akram, M. Dede, and A. Oral, "Variable Temperature-Scanning Hall Probe
Microscopy With GaN/AlGaN Two-Dimensional Electron Gas (2DEG) Micro
Hall Sensors in 4.2-425 K Range Using Novel Quartz Tuning Fork AFM
Feedback," Ieee T Magn 44 (11), 3255-3260 (2008).
46 G. E. Moore, "Cramming more components onto integrated circuits (Reprinted
from Electronics, pg 114-117, April 19, 1965)," P Ieee 86 (1), 82-85 (1998).
47 Y. Nishi and R. Doering, Handbook of Semiconductor Manufacturing
Technology. (CRC Press, 2007).
48 H.S. Nalwa, Deposition and Processing. (Academic Press, 2002).
49 Z. Zalevsky and I. Abdulhalim, Integrated Nanophotonic Devices. (Elsevier,
2010).
50 A. E. Grigorescu, M. C. van der Krogt, C. W. Hagen, and P. Kruit, "10 nm
lines and spaces written in HSQ, using electron beam lithography,"
Microelectron Eng 84 (5-8), 822-824 (2007).
51 Z. Cui, Nanofabrication: Principles, Capabilities and Limits (Springer, 2010).
52 M. T. Moneck, Doctoral Dissertation, Carnegie Mellon University, 2008.
53 V. K. Varadan, K. J. Vinoy, and S. Gopalakrishnan, Smart Material Systems
and MEMS. (WILEY, 2006).
54 A. Oral, S. J. Bending, and M. Henini, "Scanning Hall probe microscopy of
superconductors and magnetic materials," J Vac Sci Technol B 14 (2), 1202-
1205 (1996).
55 G. D. Howells, A. Oral, S. J. Bending, S. R. Andrews, P. T. Squire, P. Rice,
A. de Lozanne, J. A. C. Bland, I. Kaya, and M. Henini, "Scanning Hall probe
microscopy of ferromagnetic structures," J Magn Magn Mater 196, 917-919
(1999).
56 A. Sandhu, H. Masuda, A. Oral, S. J. Bending, A. Yamada, and M. Konagai,
"Room temperature scanning Hall probe microscopy using GaAs/AlGaAs and
Bi micro-hall probes," Ultramicroscopy 91 (1-4), 97-101 (2002).
57 A. Sandhu, A. Okamoto, I. Shibasaki, and A. Oral, "Nano and micro Hall-
effect sensors for room-temperature scanning hall probe microscopy,"
Microelectron Eng 73-4, 524-528 (2004).
58 K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V.
Dubonos, I. V. Grigorieva, and A. A. Firsov, "Electric field effect in atomically
thin carbon films," Science 306 (5696), 666-669 (2004).
106
59 C. Berger, Z. M. Song, X. B. Li, X. S. Wu, N. Brown, C. Naud, D. Mayou,
T. B. Li, J. Hass, A. N. Marchenkov, E. H. Conrad, P. N. First, and W. A.
de Heer, "Electronic confinement and coherence in patterned epitaxial
graphene," Science 312 (5777), 1191-1196 (2006).
60 C. C. Tang, M. Y. Li, L. J. Li, C. C. Chi, and J. C. Chen, "Characteristics of a
sensitive micro-Hall probe fabricated on chemical vapor deposited graphene
over the temperature range from liquid-helium to room temperature," Appl Phys
Lett 99 (11) (2011).
61 S. Sonusen, O. Karci, M. Dede, S. Aksoy, and A. Oral, "Single layer
graphene Hall sensors for scanning Hall probe microscopy (SHPM) in 3-300 K
temperature range," Appl Surf Sci 308, 414-418 (2014).
62 S. Das Sarma, S. Adam, E. H. Hwang, and E. Rossi, "Electronic transport in
two-dimensional graphene," Rev Mod Phys 83 (2), 407-470 (2011).
63 J. Irudayaraj, Biomedical Nanosensors. (CRC Press, 2012).
64 D. Dragoman, M. Dragoman, and A. A. Muller, "Graphene - A One-Atom-
Thick Material for Microwave Devices," Rom J Inf Sci Tech 11 (1), 29-35
(2008).
65 D. Dragoman and M. Dragoman, "Negative differential resistance of electrons
in graphene barrier," Appl Phys Lett 90 (14) (2007).
66 Z. Liu and X. Zhou, Graphene: Energy Storage and Conversion Applications.
(CRC Press, 2015).
67 Y. M. Lin, C. Dimitrakopoulos, K. A. Jenkins, D. B. Farmer, H. Y. Chiu, A.
Grill, and P. Avouris, "100-GHz Transistors from Wafer-Scale Epitaxial
Graphene," Science 327 (5966), 662-662 (2010).
68 K. I. Bolotin, K. J. Sikes, Z. Jiang, M. Klima, G. Fudenberg, J. Hone, P.
Kim, and H. L. Stormer, "Ultrahigh electron mobility in suspended graphene,"
Solid State Commun 146 (9-10), 351-355 (2008).
69 K. Nagashio, T. Nishimura, K. Kita, and A. Toriumi, "Mobility Variations in
Mono- and Multi-Layer Graphene Films," Appl Phys Express 2 (2) (2009).
70 C. S. S. R. Kumar, Raman Spectroscopy for Nanomaterials Characterization.
(Springer, 2012).
71 E. Elhajjar, V. L. Saponara, and A. Mulaiana, Smart Composites: Mechanics
and Design. (CRC Press, 2014).
72 J. C. Charlier, P. C. Eklund, J. Zhu, and A. C. Ferrari, "Electron and phonon
properties of graphene: Their relationship with carbon nanotubes," Top Appl
Phys 111, 673-709 (2008).
107
73 I. W. Frank, D. M. Tanenbaum, A. M. Van der Zande, and P. L. McEuen,
"Mechanical properties of suspended graphene sheets," J Vac Sci Technol B 25
(6), 2558-2561 (2007).
74 F. Bonaccorso, A. Lombardo, T. Hasan, Z. P. Sun, L. Colombo, and A. C.
Ferrari, "Production and processing of graphene and 2d crystals," Mater Today
15 (12), 564-589 (2012).
75 J. H. Chen, C. Jang, S. D. Xiao, M. Ishigami, and M. S. Fuhrer, "Intrinsic
and extrinsic performance limits of graphene devices on SiO2," Nat
Nanotechnol 3 (4), 206-209 (2008).
76 R. Munoz and C. Gomez-Aleixandre, "Review of CVD Synthesis of
Graphene," Chem Vapor Depos 19 (10-12), 297-322 (2013).
77 Y. Zhang, L. Y. Zhang, and C. W. Zhou, "Review of Chemical Vapor
Deposition of Graphene and Related Applications," Accounts Chem Res 46
(10), 2329-2339 (2013).
78 G. X. Ni, Y. Zheng, S. Bae, C. Y. Tan, O. Kahya, J. Wu, B. H. Hong, K.
Yao, and B. Ozyilmaz, "Graphene-Ferroelectric Hybrid Structure for Flexible
Transparent Electrodes," Acs Nano 6 (5), 3935-3942 (2012).
79 R. Yakimova, C. Virojanadara, D. Gogova, M. Syvajarvi, D. Siche, K.
Larsson, and L. I. Johansson, "Analysis of the Formation Conditions for Large
Area Epitaxial Graphene on SiC Substrates," Silicon Carbide and Related
Materials 2009, Pts 1 and 2 645-648, 565-568 (2010).
80 F. Dulot, L. Mansour, A. Leycuras, W. Wulfhekel, D. Sander, F. A.
d'Avitaya, and M. Hanbucken, "Structure and morphology of concave-shaped
surfaces on 6H-Si(0001) after H-2 etching," Appl Surf Sci 187 (3-4), 319-325
(2002).
81 J. Hass, W. A. de Heer, and E. H. Conrad, "The growth and morphology of
epitaxial multilayer graphene," J Phys-Condens Mat 20 (32) (2008).
82 P. Blake, E. W. Hill, A. H. Castro Neto, K. S. Novoselov, D. Jiang, R. Yang,
T. J. Booth, and A. K. Geim, "Making graphene visible," Appl Phys Lett 91
(6) (2007).
83 A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri,
S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth, and A. K. Geim, "Raman
spectrum of graphene and graphene layers," Phys Rev Lett 97 (18) (2006).
84 Y. M. You, Z. H. Ni, T. Yu, and Z. X. Shen, "Edge chirality determination
of graphene by Raman spectroscopy," Appl Phys Lett 93 (16) (2008).
85 F. Han, A Modern Course in the Quantum Theory of Solids. (World Scientific
Publishing Company, 2012).
86 V.V. Tuchin, Handbook of Optical Sensing of Glucose in Biological Fluids and
Tissues. (CRC Press, 2009).
108
87 J. P. Dakin and R. G. W. Brown, Handbook of Optoelectronics. ( CRC Press,
2006).
88 A. A. Green and M. C. Hersam, "Solution Phase Production of Graphene with
Controlled Thickness via Density Differentiation," Nano Lett 9 (12), 4031-4036
(2009).
89 M. Ishigami, J. H. Chen, W. G. Cullen, M. S. Fuhrer, and E. D. Williams,
"Atomic structure of graphene on SiO2," Nano Lett 7 (6), 1643-1648 (2007).
90 S. A. Dyakov, T. S. Perova, C. Q. Miao, Y. H. Xie, S. A. Cherevkov, and A.
V. Baranov, "Influence of the buffer layer properties on the intensity of Raman
scattering of graphene," J Raman Spectrosc 44 (6), 803-809 (2013).
91 O. Kazakova, T. L. Burnett, J. Patten, L. Yang, and R. Yakimova, "Epitaxial
graphene on SiC(000(1)over-bar): functional electrical microscopy studies and
effect of atmosphere," Nanotechnology 24 (21) (2013).
92 Z. G. Cheng, Q. Y. Zhou, C. X. Wang, Q. A. Li, C. Wang, and Y. Fang,
"Toward Intrinsic Graphene Surfaces: A Systematic Study on Thermal
Annealing and Wet-Chemical Treatment of SiO2-Supported Graphene Devices,"
Nano Lett 11 (2), 767-771 (2011).
93 D. W. Shin, H. M. Lee, S. M. Yu, K. S. Lim, J. H. Jung, M. K. Kim, S. W.
Kim, J. H. Han, R. S. Ruoff, and J. B. Yoo, "A Facile Route To Recover
Intrinsic Graphene over Large Scale," Acs Nano 6 (9), 7781-7788 (2012).
94 T. T. Feng, D. Xie, Y. X. Lin, H. Tian, H. M. Zhao, T. L. Ren, and H. W.
Zhu, "Unipolar to ambipolar conversion in graphene field-effect transistors,"
Appl Phys Lett 101 (25) (2012).
95 M. Bando, T. Ohashi, M. Dede, R. Akram, A. Oral, S. Y. Park, I. Shibasaki,
H. Handa, and A. Sandhu, "High sensitivity and multifunctional micro-Hall
sensors fabricated using InAlSb/InAsSb/InAlSb heterostructures," J Appl Phys
105 (7) (2009).
96 A. Oral, S. J. Bending, and M. Henini, "Real-time scanning hall probe
microscopy," Appl Phys Lett 69 (9), 1324-1326 (1996).
97 V. Skakalova, A. B. Kaiser, J. S. Yoo, D. Obergfell, and S. Roth,
"Correlation between resistance fluctuations and temperature dependence of
conductivity in graphene," Phys Rev B 80 (15) (2009).
98 S. M. Kim, A. Hsu, Y. H. Lee, M. Dresselhaus, T. Palacios, K. K. Kim, and
J. Kong, "The effect of copper pre-cleaning on graphene synthesis,"
Nanotechnology 24 (36) (2013).
99 W. Liu, H. Li, C. Xu, Y. Khatami, and K. Banerjee, "Synthesis of high-
quality monolayer and bilayer graphene on copper using chemical vapor
deposition," Carbon 49 (13), 4122-4130 (2011); I. Vlassiouk, P. Fulvio, H.
Meyer, N. Lavrik, S. Dai, P. Datskos, and S. Smirnov, "Large scale
109
atmospheric pressure chemical vapor deposition of graphene," Carbon 54, 58-67
(2013).
100 X. Du, S. W. Tsai, D. L. Maslov, and A. F. Hebard, "Metal-insulator-like
behavior in semimetallic bismuth and graphite," Phys Rev Lett 94 (16) (2005).
101 C. M. B. Hincapie, M. J. P. Cardenas, J. E. A. Orjuela, E. R. Parra, and J. J.
O. Florez, "Physical-Chemical Properties of Bismuth and Bismuth Oxides:
Synthesis, Characterization and Applications," Dyna-Colombia 79 (176), 139-
148 (2012).
102 A. Kumar and O. P. Katyal, "Thickness Dependence of Temperature-
Coefficient of Resistivity of Polycrystalline Bismuth-Films," Appl Phys a-Mater
52 (4), 265-267 (1991).
103 S. G. Bompadre, C. Biagini, D. Maslov, and A. F. Hebard, "Unambiguous
determination of the g factor for holes in bismuth at high B/T," Phys Rev B 64
(7) (2001).
104 J. Puthussery, T. H. Kosel, and M. Kuno, "Facile Synthesis and Size Control
of II-VI Nanowires Using Bismuth Salts," Small 5 (10), 1112-1116 (2009).
105 R. F. Broom and E. H. Rhoderick, "Studies of the Intermediate State in Thin
Superconducting Films " PROC. PHYS. SOC. 79 (1962).