ENGINEERING GRAPHENE FOR NANOELECTRONICS
By
Haider Sahib Al-Mumen

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

ABSTRACT
ENGINEERING GRAPHENE FOR NANOELECTRONICS
By
Haider Sahib Al-Mumen
Graphene has attracted research interest since its discovery in 2004 and professor Geim’s
receipt of the Nobel prize in 2010. It has been used for constructing a variety of electronics
and sensors due to its unique electrical, mechanical, thermal, and optical properties.
In this dissertation, we developed several technologies to control the electronic properties of
graphene, which will pave the way for the future development of graphene nanoelectronics.
First, since graphene properties vary depending on its number of layers, we identified a
method for engineering the number of graphene layers using fine-tuned oxygen plasma
etching. With this technique, a single layer of graphene can be removed at a time. In addition,
we demonstrated a template-less nanofabrication technique for batch production of graphene
nanomeshes and multiribbons, and explored the feasibility of using these nanopatterns to
construct field effect transistors (FETs). By introducing nanopatterns into pristine graphene,
we could effectively open the band gap of graphene and convert it from semimetal into
semiconductor. Furthermore, we studied a doping method for making n-type graphene with
long-term chemical stability in air and stability at wide range of temperature. Highly stable ntype graphene with minimal defects was achieved using photo acid generator (PAG) mixed
with SU-8 epoxy resin as an effective electron dopant and encapsulation. The electronic
properties of the as-doped n-type graphene were confirmed by measuring its current transport
characteristics and Fermi level shifts.
Building on the aforementioned engineering techniques, we proposed a new Metal-SU8graphene (MSG) technology, which is compatible with the conventional CMOS fabrication

technology. MSG FETs were fabricated on both rigid and flexible substrates. A graphene
invertor was also constructed as a proof of concept.
Finally, we explored the potential applications of graphene in nanosensors, including
chemical, temperature and flow sensors. We studied the possibility of using an inter-layer
graphene nano configuration to detect the absorption/desorption of different chemical
molecules. Our results show a remarkable enhancement in graphene surface sensitivity,
which can be attributed to extra edges and inter-sheet tunneling effects. We also
demonstrated the capability of using graphene nanowires in temperature and flow rate
sensing.

Dedicated to my parents, my wife Areej, my daughters Rutaj and Jumana,
to my brother Amir and my sister Ban, for their love and support

iv

ACKNOWLEDGEMENTS

I would like to express my gratitude to my adviser, Dr. Wen Li, for her guidance and,
kindness during my PhD work. It was a nice opportunity to work with her and I learned a lot
through our insightful discussions. I am also grateful for fellow research group members in
the microtechnology laboratory for their help. I would especially like to thank Dr. Baokang
Bi, the manager of the KECK Michrofabrication Facility for the training assistance and also
thank Dr. Fubo Rao for his assistance and advice. You all will be an important part of my
memory that I will never forget.
I want to acknowledge the technicians of the Department of the Electrical and Computer
Engineering as well as the technicians of the Center of the Composite Materials for their
technical support.
I would like to extend appreciation to my committee members, especially Dr. Lixin Dong, for
their advice and fruitful discussions
My gratitude also goes to the staff of the Iraqi Cultural Office in Washington, D.C., Specially
Dr. Hadi Alkhalili and Mr. Muhanad Al-Kabi.
Lastly, a special thank to my parents, wife, brother and sister. This dissertation is dedicated to
them for their love and support for achieving my academic goals.

v

TABLE OF CONTENTS

LIST OF TABLES………………………………………………………………………….viii
LIST OF FIGURES……………………………………………………………......................ix
KEY TO ABBREVIATIONS……………………………………………………………….xvi
1

Introduction to Graphene .................................................................................................... 1
1.1 Introduction ................................................................................................................. 1
1.2 Theoretical Background of Graphene ......................................................................... 2
1.2.1
Crystal structure of graphene ............................................................................... 2
1.2.2
Carrier density of graphene .................................................................................. 5
1.2.3
Lattice vibration in graphene ............................................................................... 7
1.2.4
Gate-voltage modulating effect............................................................................ 7
1.2.5
Influences of substrate on graphene mobility ...................................................... 8
1.2.6
Geometrical effects ............................................................................................ 10
1.3 Graphene Applications in Nanoelectronic Circuits ................................................... 12
1.4 Mission and Outline .................................................................................................. 14
REFERENCES .......................................................................................................... 14

2

Graphene Preparing Method ............................................................................................. 20
2.1 Introduction ............................................................................................................... 20
2.2 Methods to Produce Graphene .................................................................................. 20
2.2.1
Micromechanical exfoliation and cleavage ....................................................... 20
2.2.2
Epitaxial graphene from silicon carbide ............................................................ 22
2.2.3
Chemical vapour deposition (CVD) .................................................................. 23
2.2.4
Comparison of three methods ............................................................................ 24
2.3 Methods to Identify Graphene................................................................................... 25
2.3.1
Optical microscopy ............................................................................................ 25
2.3.2
Atomic force microscopy (AFM) ...................................................................... 26
2.4 Raman Spectroscopy ................................................................................................. 27
2.5 Graphene Cleaning Method ...................................................................................... 30
2.6 Conclusion................................................................................................................. 32
REFERENCES .......................................................................................................... 34

3

Engineering of Graphene Sheet ........................................................................................ 37
3.1 Introduction ............................................................................................................... 37
3.2 Layer Engineering of Graphene Sheets ..................................................................... 37
3.2.1
Background ........................................................................................................ 37
3.2.2
Singular sheet etching method ........................................................................... 38
3.2.3
Results of graphene singular sheet etching ........................................................ 40
3.2.4
Application of the SSE method.......................................................................... 48
3.3 Band Gap Engineering of Graphene ......................................................................... 50
3.3.1
Background ........................................................................................................ 50
3.3.2
E-beam patterning method to engineer graphene band gap ............................... 51
3.3.3
Fabrication of p-type graphene FETs ................................................................ 59
3.4 Conclusion................................................................................................................. 65
REFERENCES .......................................................................................................... 67
vi

4

Graphene N-type doping .................................................................................................. 71
4.1 Introduction ............................................................................................................... 71
4.2 Method ...................................................................................................................... 72
4.3 SU-8 Doping Results and Discussions ...................................................................... 74
4.3.1
SU-8 n-type doping effect .................................................................................. 74
4.3.2
SU-8 doped n-type graphene FETs .................................................................... 76
4.3.3
Doping mechanism of SU-8............................................................................... 80
4.3.4
Stability of SU-8 doped n-type graphene........................................................... 84
4.4 Conclusion................................................................................................................. 91
4.5 REFERENCES .......................................................................................................... 93
5 Graphene Logic Circuits Based on a Metal SU-8 – Graphene Technology .................... 97
5.1 Introduction ............................................................................................................... 97
5.2 E-beam lithography of SU-8 ..................................................................................... 97
5.3 Graphene inverter circuit ........................................................................................... 99
5.3.1
Fabrication of graphene circuit .......................................................................... 99
5.3.2
Characterization of graphene circuit ................................................................ 100
5.4 Flexible Graphene Transistor .................................................................................. 102
5.4.1
Top-gated graphene FETs on mechanically flexible Parylene-C substrate ..... 103
5.4.2
Back-gated graphene FETs on mechanically flexible Parylene-C substrate ... 105
5.5 Conclusion............................................................................................................... 108
REFERENCES ........................................................................................................ 109
6

Graphene Applications in Nanosensing ......................................................................... 111
6.1 Introduction ............................................................................................................. 111
6.2 Graphene Inter-sheet Sensor for Chemical Detection ............................................. 111
6.2.1
Graphene inter-sheet and intra-sheet effects .................................................... 111
6.2.2
Method ............................................................................................................. 113
6.2.3
Result and Discussion ...................................................................................... 115
6.2.4
Discussion of sensing mechanism ................................................................... 122
6.3 Graphene Sensor for Flow/Temperature Sensing ................................................... 124
6.3.1
Overview of flow/temperature sensors ............................................................ 124
6.3.2
Heat convection theory .................................................................................... 125
6.3.3
Fabrication method .......................................................................................... 127
6.3.4
Results and discussions .................................................................................... 128
APPENDICES.…………………………………………………………………....138
APPENDIX A: Fabrication Recipes ................................................................. 139
APPENDIX B: Testing ..................................................................................... 146
REFERENCES…………………………………………………………………... 150

vii

LIST OF TABLES

Table 2.1 Three representative methods for producing graphene films…………………….. 25
Table 3.1 The effect of oxygen plasma etching on graphene nanomesh…………………… 58

viii

LIST OF FIGURES

Figure 1.1 Graphene band structure [7]. .................................................................................... 2
Figure 1.2 Conceptually, graphene is a building material for (a) fullerene (b) CNTs and (c)
graphite [8]. ................................................................................................................................ 3
Figure 1.3 (a) Bernal stack of a typical graphite structure. (b) Band structure of graphene
showing the six Dirac neutrality points [6]. ............................................................................... 3
Figure 1.4 (a) The Brillouin zone and (b) simplified band structure of mono-layer graphene.. 4
Figure 1.‎5 Energy and the density of state relationship in graphene ......................................... 6
Figure 1.6 Illustration of two types of GNR edges: (a) zigzag and (b) armchair. (c)
Confinement in the x-direction quantizes the kx quantization of graphene band into 1-D band.
.................................................................................................................................................. 10
Figure 2.1 Metal markers on a SiO2/Si substrate. .................................................................... 21
Figure 2.2 (a) Mechanical exfoliation of graphite using a scotch tape. (b) A 300 nm SiO2/Si
substrate with transferred graphene. The dashed circle indicates a thin graphene sheet. ........ 22
Figure 2.3 Mono-, bi- and tri-layer graphene produced by graphitization of SiC [4]. ............ 23
Figure 2.4 Optical microscopic image of a CVD graphene sample on a SiO2/Si substrate. ... 24
Figure 2.5 Optical microscopic images for graphene samples with different layer numbers:
(a) monolayer; (b) bilayer; (c) 3 layer; (d) and (e) 4 to 6 layers; (f) to (h) more than 10 layers.
.................................................................................................................................................. 26
Figure 2.6 AFM (DI 3100) thickness measurements of four monolayer graphene samples.
The thickness ranges from 0.676 nm to 0.829 nm. .................................................................. 27
Figure 2.7 (a) Raman spectra of different numbers of graphene sheet layers obtained using
HoloProbe Micro-Raman Spectrograph; (b) The 2D band Raman spectrum of a bilayer
graphene sheet, showing 4 edges in one big peak. .................................................................. 29
Figure 2.8 Raman spectra of a same graphene sample before and after introducing defects
using O2 plasma with 70 W radio-frequency (RF) power for 19 sec. The D band appeared as a
result of surface defects. .......................................................................................................... 30
Figure 2.9 AFM images of a monolayer graphene before (left image) and after (right image)
a heat treatment for removing tape residues. The sheet thickness decreased from 1.260 nm to
0.676 nm after cleaning............................................................................................................ 31
Figure 2.10 Raman spectra of a monolayer graphene sample before (black color) and after
(red color) the cleaning treatment. ........................................................................................... 32
ix

Figure 3.1 Conceptual schematic and experimental setup of the layer engineering process
with oxygen plasma. The etching process can be performed with either the ground electrode
oxygen plasma at the ground electrode or the powered electrode. .......................................... 40
Figure 3.2 SEM images show the evolution of a graphene sample after each “singular sheet
etching” step with ground electrode oxygen plasma: (a) as-produced; (b) after the first
“singular sheet etching” step; (c) after the second “singular sheet etching” step. ................... 41
Figure 3.3 Demonstration of producing a monolayer graphene from a bilayer graphene using
the SSE method. (a), (b), and (c) show the SEM image, AFM image (up left panel), 2D band
of its Raman spectrum (up right panel) and its height profile from the AFM image (bottom
panel) before being etched with oxygen plasma (a), after the first “singular sheet etching”
process (b), and after the second “singular sheet etching” process (c), respectively. .............. 42
Figure 3.4 Schematic cartoons show two possible etching mechanisms of ground electrode
oxygen plasma: (a) anisotropic vertical etching and (b) anisotropical horizontal etching. In
the anisotropic vertical etching mode, the defect patterns on the top graphene sheet will be
copied to the bottom graphene sheet with similar sizes. However, in the anisotropical
horizontal etching mode, the defect patterns will expand during etching and be copied to the
bottom graphene sheet in an obviously larger size, while the size of the graphene sheet will
shrink. Illustrations (c)-(e) show the SEM images of monolayer graphene samples before and
after being subjected to oxygen plasma etching for 32 s, 30 s and 28 s, respectively. The
monolayer regions of the graphene samples were marked with arrows. In each sub-figure, the
top and bottom SEM images show the graphene sample before and after being etched with
the corresponding process duration, respectively. ................................................................... 45
Figure 3.5 AFM images, height profiles, and the 2D bands of its Raman spectrum of a bilayer
graphene (a) before and (b) after the SSE with powered electrode oxygen plasma,
demonstrating the formation of a monolayer graphene from a bilayer graphene. ................... 46
Figure 3.6 Raman spectra of bilayer graphene samples before and after the SSE process with
(a) the powered electrode oxygen plasma and (b) the ground electrode oxygen plasma. (c)
Statistic analysis of Raman ID/IG ratios of graphene samples after being etched with the
powered electrode (solid circles) and the oxygen plasma (open circles). The red and blue
dotted lines show the average values of data from solid and open circles, respectively. (d)
Raman spectrum of a monolayer graphene sheet produced from a bilayer graphene sheet after
the annealing treatment. Its Raman spectrum before annealing is shown in the bottom panel
of (a). ........................................................................................................................................ 48
Figure 3.7 An SEM image of a 2-nm-thick suspended graphene film produced from graphite
flake of ~ 15 nm using layer-by-layer oxygen plasma etching. Inset is the Raman spectra. ... 50
Figure 3.8 Conceptual diagram of the proposed template-less technique for fabrication of
graphene nanopatterned FET transistors. ................................................................................. 52
Figure 3.9 Periodic nanomesh patterned by e-beam with a scanning area of 50×50 µm2, an ebeam resist thickness of 900 nm, and one scanning cycle. (a) and (b) SEM images show that
the dimension of the holes increased as the e-beam current increased. (c) Histogram of the
average sizes under different beam currents. ........................................................................... 54

x

Figure 3.10 Nanoholes patterned on different scanning areas: (a) 2500 µm2 with a neck with
of ~150 nm, (b) 900 µm2 with a neck width of ~ 90 nm, (c) 900 µm2 with a neck width of ~
50 nm, and (d) 100 µm2 with overlapped holes. (e) Histogram shows the effect of the
scanning area on the average minimum neck width of the GNMs. ......................................... 55
Figure 3.11 Average neck widths of 75 nm and 25 nm were obtained by two scans (a) and
three scans (b and c), respectively. (d) Histogram shows the relation between the number of
the scanning cycles and the average neck width of the GNMs................................................ 56
Figure 3.12 (a) Concept of the muliribbon fabrication teqnique. (b) A representative graphene
multi-ribbon structure. ............................................................................................................. 57
Figure 3.13 SEM image of an MSU logo with nanoholes on a large area. ............................. 59
Figure 3.14 Process flow for making GNM transistors. .......................................................... 60
Figure 3.15 Representative GNR transistor fabricated by on (a) monolayer graphene sheet (b)
few-layer graphene sheet. ........................................................................................................ 61
Figure 3.16 Drain current (Ids) versus drain–source voltage (Vds) at different gate voltages
(Vg) for a monolayer GNM transistor (b) Ids versus Vg at different Vds voltages (c) Drain
current (Ids) versus drain–source voltage (Vds) at different gate voltages (Vg) for a monolayer
GNM transistor (d) resistance variation as the function of the gate voltage of a few-layer
graphene device. All graphene nanomeshes had p-type semiconductor behavior................... 62
Figure 3.17 Ids-Vds characteristics of a monolayer GNM transistor tested under vacuum
condition and at atmospheric pressure, showing no significant difference. ............................ 64
Figure 3.18 (a) Current voltage transfer characteristic of a graphene multi-ribbon transistor,
and (b) its current transport characteristic. .............................................................................. 65
Figure 4.1 (a) Schematic of an n-type graphene transistor with SU-8 as the n-type dopant and
self-encapsulation. (b) Three basic chemical compounds of the SU-8 resist and the chemical
reaction for generating the photo-acid upon UV exposure. (c) A representative GNM
transistor doped and encapsulated with SU-8. ......................................................................... 73
Figure 4.2 (a) Raman spectra of a representative bilayer graphene sample with different
treatments. (b) Raman spectra of a representative monolayer graphene sample with different
treatments. The spectra from the bottom to the top correspond to the monolayer PG without
nanomesh, undoped monolayer GNM, and SU-8 doped monolayer GNM, respectively. ...... 76
Figure 4.3 A representative GNM transistorwith the Si back gate, 300 nmSiO2 dielectric,
Au/Ti source and drain contacts. (a) Gate-dependent conductivity (σ) of graphene before the
SU-8 doping, after the doping, and after removing SU-8. (b) Current-voltage (Vds-Ids) curve
of the SU-8 doped device, measured at various gate voltages (Vg) ranging from -20 to +20 V.
.................................................................................................................................................. 77
Figure 4.4 Gate-dependent conductivity curves (σ) of the n-type doped devices (a) with
various SU-8 soft-baking durations of 10 min, 30 min, and 90 min, and (b) with various SU-8
thicknesses of ~ 4 µm and ~ 100 µm. ...................................................................................... 79
xi

Figure 4.5 (a) Band diagrams showing the shifts in the Fermi level of p-type and n-type
doped graphene, respectively. (b) Calculated electron concentrations (ne) and Fermi level
shift (ΔEF) of graphene samples, as a function of SU-8 thickness and soft-baking duration. . 79
Figure 4.6 Current transfer characteristcs of graphene FETs doped by two different chemical
mixtures. Case 1 (p-type): graphene doped with the resin-solvent mixture. Case 2 (n-type):
Graphene doped by the resin-solvent-PAG mixture. The inset is the SEM image of the
device. ...................................................................................................................................... 82
Figure 4.7 Vds-Ids relation of the graphene FET after n-doping with the resin-solvent-PAG
mixture. .................................................................................................................................... 82
Figure 4.8 Device characteristics before and after n-doping with two different PAG
concentrations. ......................................................................................................................... 83
Figure 4.9 The positions of the Dirac points of four samples with different PAG
concentrations. The triangle, square, circle and star symbols resemble different samples at
different PAG concentrations. The Dirac point positions of some samples were observed at
different PAG concentrations and some were observed at only one PAG concentration since
sometimes it is hard to totally remove the PAG from graphene after soft baking. .................. 84
Figure 4.10 Air stability testing of a representative n-type GNM transistor doped and selfencapsulated with ~ 4 µm crosslinked SU-8. (a) Gate-dependent conductivity (σ) curves (b)
Gate-dependent resistance of the device measured at different time points. The dash lines
represent the positions of the Dirac points. .............................................................................. 85
Figure 4.11 Temperature stability of two n-type graphene FETs: (a) Vg-Ids curves at a
temperature range from -15 °C to 100 °C. (b) Vg-Ids curves at a temperature range from RT to
100 °C. ..................................................................................................................................... 86
Figure 4.12 (a) Raman spectra of pristine graphene at different temperatures. (b) Temperature
dependence of G-band peak for mono- and bilayer pristine graphene samples. ..................... 87
Figure 4.13 (a) Raman spectra of n-doped graphene at different temperatures. (b) Comparison
of temperature dependence between pristine and n-doped monolayer graphene sheets. (c)
Temperature dependence of bilayer graphene. The slope of fitting curve represents the
extracted temperature coefficient of G-band peak. .................................................................. 89
Figure 5.1 SU-8 patterns with various feature sizes (500 nm to 3 µm), exposed using different
areal doses of: (a) 300 µC/cm2, (b) 15 µC/cm2, (c) 4 µC/cm2. ................................................ 99
Figure 5.2 An integrated graphene inverter: (a) the schematic of the circuit fabrication; (b) the
SEM image of the fabricated inverter; and (c) the circuit layout.(d) Schematic of the
fabrication technique to avoid impinging electrons. .............................................................. 100
Figure 5.3 Resistance curve (R vs Vin) of graphene transistors schematically illustrated in
Figure 5.1, measured at room temperature. Rn and RP corresponded to the transfer resistances
of the n-type and p-type graphene regions, respectively. ...................................................... 101

xii

Figure 5.4 Voltage transfer characteristic, demonstrating a logic inverter function. ............ 102
Figure 5.5 Raman spectrum of graphene transferred on a Parylene substrate. Inset is the
optical microscope image of the graphene............................................................................. 104
Figure 5.6 Schematic of the top-gated, n-type graphene FET before and after removing the
carrier substrate. ..................................................................................................................... 104
Figure 5.7 (a) A fabricated graphene n-type FET and (b) the measured Ids-Vg characteristic of
the device ............................................................................................................................... 105
Figure 5.8 Device configurations of (a) n-type and (b) p-type back-gated graphene FETs. . 106
Figure 5.9 A fabricated p-type back-gated, flexible graphene transistor before and after
removing the carrier substrate. Silver paste was applied to the contacts in order to avoid
punching the substrate during device testing. Dashed area shows the transparent graphene
sheet. (b) The current transfer characteristic of the device, showing the p-type transport
property of the graphene. ....................................................................................................... 107
Figure 5.10 Mechanical flexibility of the as fabricated graphene transistors. ....................... 107
Figure 6.1 Schematic diagrams of (a) an “intra-sheet” and (b) an “inter-sheet” graphene
sensors. ................................................................................................................................... 112
Figure 6.2 (a) Example of a mechanical exfoliated graphene containing both mono- and bilayer regions. (b) Raman spectra of different regions. The top and bottom insets represent the
enlarged 2D bands of region B and A respectively. .............................................................. 114
Figure 6.3 Schematic processes for fabricating (a) “intra-sheet” and (b) “inter-sheet”
graphene sensors. ................................................................................................................... 115
Figure 6.4 (a) Optical image, (b) Raman spectrum and (c) the enlarged Raman 2D band and
its Lorentzian fitting of the bilayer graphene for an “intra-sheet” sensor. (d) Optical image,
(e) Raman spectrum and (f) the enlarged Raman 2D bands and their Lorentzian fittings of the
mono- and bi-layer graphene regions for an “inter-sheet” sensor. ........................................ 116
Figure 6.5 Schematic diagram of the testing system for both “intra-sheet” and “inter-sheet”
ethanol molecule graphene sensors. ....................................................................................... 117
Figure 6.6 Resistance responses of (a) the “intra-sheet” sensor and (b) the “inter-sheet” sensor
upon exposure to 0.5% ethanol vapor in nitrogen environment. ........................................... 118
Figure 6.7 (a) SEM image of an “intra-sheet” sensor (electrodes 1 and 2) and an “inter-sheet”
sensor (electrodes 2 and 3) fabricated on a same graphene sheet. Graphs (b) and (c) are time
dependent responses of the “intra-sheet” and “inter-sheet” resistances before and after
introducing 0.5% ethanol into nitrogen. ................................................................................ 119
Figure 6.8 (a), (b) and (c) SEM images an “inter-sheet” sensor fabricated on different
graphene layers, an “intra-sheet” sensor fabricated on a monolayer graphene, and an “intrasheet” sensor fabricated on a bilayer graphene, respectively. (d), (e) and (f) The
xiii

corresponding Raman 2D bands and their Lorentzian fittings of the graphene samples in (a),
(b) and (c), respectively. The insets show the entire Raman spectra of the graphene. The A
and B curves in (d) correspond to the mono- and bi-layer graphene regions in (a). ............. 120
Figure 6.9 (a) Current response of the “inter-sheet” sensor upon exposure to water vapor at a
pressure of 500 Pa in the sample chamber within one cycle. (b), (c), and (d) The current
responses of the “inter-sheet” sensor, “intra-sheet” sensor on a monolayer graphene and the
“intra-sheet” sensor on a bilayer graphene, respectively, in several cycles of water vapor
turning ON and OFF. ............................................................................................................. 122
Figure 6.10 A typical process flow for making the proposed graphene hot-wire sensors. .... 127
Figure 6.11 SEM images of the fabricated bilayer (a) and few-layer(b) graphene hot-wire
sensors. The insets show the Raman spectra obtained from the bilayer and few-layer graphene
areas. ...................................................................................................................................... 128
Figure 6.12 (a) Device sealed in a DIP carrier with capillaries as upstream and downstream
connections. (b) COMSOL simulation shows temperature profile of the packagevs. distance
from the inlet at different input flowrates.. ............................................................................ 129
Figure 6.13 Transient current responses of a bilayer sensor with (a) different N2 flow rates
and (b) different N2 input pressure......................................................................................... 130
Figure 6.14 Normalized resistance changes at different flow rates for (a) a bilayer graphene
hot wire and (b) a few-layer sensor. (c) Comparison between the normalized resistancesof the
mono-, bi- and few-layer graphene devices, measured as temperature changed between RT
and 80 °C. .............................................................................................................................. 131
Figure 6.15 Transient responses of different graphene devices as the environmental
temperature varied from RT to 70 °C: (a) the response of a bilayer graphene sensor, (b) the
response of a few-layer graphene sensor, (c) the response of the bilayer device when
temperature increased from 70 °C to RT then increased to 70 °C. ........................................ 133
Figure 6.16 (a) Sample 1: Self-heating effects of bilayer graphene device with an applied
voltage of 3V. (b) Sample 2,current variation with supplied voltage of 3.5 V. (c) Sample 3
was damaged by Joul heating after appling 3 V for ~ 2000 sec. The inset is an AFM image of
a failed device due to Joul heating. ........................................................................................ 134
Figure A.1 Microfabrication instruments (a) ABM Mask aligner (b) Auto 306 Thermal
evaporator (c) PX-250 Plasma Etcher....................................................................................140
Figure A.2 Micromechanical exfoliation of graphite. ...........................................................141
Figure A.3 Annealing setup. ..................................................................................................141
Figure A.4 JEOL 840A Electron Beam Lithography instrument (a) Sample loading (b)
instrument during writing. .....................................................................................................142
Figure A.5 Hitashi S-4700 SEM. The SEM screen shows a GNR sample ............................144

xiv

Figure B.1 Testing in air (a) REL-4100A probe station. (b) HP-4145semiconductor parameter
analyser ..................................................................................................................................146
Figure B.2 Testing in an inert gas (a) feed through testing setup (b) set up connected to
semiconductor parameter analyser. ........................................................................................147
Figure B.3 Feed-through for SEM (SEM EVO LS-25) .........................................................147
Figure B.4 Dimension 3100 SPM Atomic Force Microscope ...............................................148
Figure B.5 HoloProbe Raman Spectrograph coupled to an Olympus BX-60 optical
microscope .............................................................................................................................149

xv

KEY TO ABBREVIATIONS

AFM: Field effect transistor
Cox: Oxide capacitance
CNT: Carbon nanotube
d: distance between gate and closest layer (SiO2 thickness)

𝜺o: permittivity of the vacuum
D (E): Density of state
DI: Deionized water
e: Electron charge
α: Resistance temperature coefficient
σs: Sheet resistance
Ef: Fermi energy
Et,b: Electric field produces by top/back gate voltage
EBL: Electron beam lithography
F0 (E): Fermi function
FET: Field effect transistor
FIB: Focused ion beam
FLG: Few layer graphene
λ: Mean free path

υF: Fermi velocity
xvi

Vg: Gate voltage
ħ: Reduced Plank’s constant
h: Heat transfer coefficient
HOPG: Highly ordered pyrolytic graphite
IV: Current voltage relation
k: dielectric constant
K: Kelvin
ns: Charge density
OPE: Oxigen plasma etching
PMMA: polymethyl methacrylate
RIE: Reactive ion etching
RT: Room temperature
RW: Wire resistance
Sccm: Standard cubic centimeter
SEM: Scanning electron microscope
SF6: Sulfur hexafluoride
tox: Oxide thickness
TW: Wire temperature
Tf : Fluid temperature
PAG: Photo acid generator

xvii

1

1.1

NTRODUCTION TO GRAPHENE

Introduction

Graphene has attracted research interest since its discovery in 2004. Unlike semiconductors,
the band structure of graphene is unique. It is a zero band semiconductor, which means there
is no band gap between the valance and the conduction bands, as shown in the Figure 1.1.
The valance band and conduction band are symmetric around the six Dirac points. This
unique band structure promotes high electron mobility of up to 2×10
4

5

2 -1 -1

cm V s

for

2 -1 -1

suspended graphene and around 1.5×10 cm V s for graphene on silicon dioxide (SiO2)
[1-4]. In addition, graphene has a high cutoff frequency theoretically. The reported study
predicts a cutoff frequency of 400-600 GHz for 50 nm gated graphene field effect transistors
(FETs) [5], whereas the maximum value recorded for silicon is only around 40 GHz. This
makes graphene an excellent candidate material for the next generation of radio-frequency
(RF) transistors and analog integrated circuits. Moreover, graphene has high thermal
3

-1 -1

2

conductivity (5×10 Wm K ), a high surface area (theoretically 2630 m /g), and high
transparency (~97% over a wide range of wavelengths from the visible spectrum to the near
infrared (IR)) [6]. These stunning properties provide a fertile ground for a wide variety of

potential applications of graphene, such as electronics, photonics, and sensors. The unique
combination of electrical, thermal, and optical properties of graphene can be better
understood from its physics properties, which will be discussed in details in section 1.2.

1

Figure ‎1.1 Graphene band structure [7].
1.2

Theoretical Background of Graphene

1.2.1 Crystal structure of graphene
Graphene is a one-atom thick layer of graphite, a honeycomb two dimensional lattice of
2

carbon atoms with sp hybridization. It can be wrapped up into zero dimension fullerenes,
rolled into one dimension CNTs, or stacked into three dimensional graphite [8], as depicted in
Figure 1.2. The crystal structure of graphite consists of multiple layers of carbon sheets in a
Bernal stack. The s, px, and py atomic orbitals on each carbon hybridize to form strong
2

covalent sp bonds within a single layer graphene sheet (Figure 1.3 (a)) [9]. The remaining pz
orbital on each carbon overlaps with its three neighboring carbons to form a band of filled π
orbitals (known as the valence band) and a band of empty π* orbitals (known as the
conduction band). This weak interlayer binding leads to out-of-plane electrical and thermal
conductivities between multilayer graphene, which could be utilized as a sensing mechanism
for detecting molecule adsorption/desorption, mechanical displacement, and atomic forces,
etc., as discussed in Chapter 6. In addition, graphene has zero effective mass and minimum
conductivity for holes and electrons near the six Dirac points (Figure 1.3 (b)). These charge
carriers can travell thousand interatomic distances without scattering, owing to the super-fast
mobility of graphene.

2

Figure ‎1.2 Conceptually, graphene is a building material for (a) fullerene (b) CNTs and (c)
graphite [8].

Figure ‎1.3 (a) Bernal stack of a typical graphite structure. (b) Band structure of graphene
showing the six Dirac neutrality points [6].
In the a single-layer graphene sheet, the crystal structure consists of six carbon atoms in a
unit cell, of which two are distinct atoms and the other atoms can be reached by a reciprocal

3

lattice vector translation (Figure 1.4 (a)). As in Figure 1.4 (b), the band structure of graphene
can be simplified into two cones. The Fermi level is in the intersection of the cones with no
energy gap between the valance band and the conduction band. The shifting in Fermi level
upward to the conduction band or downward to the valance band can be calculated by

Δ𝐸 = √𝜋𝑛 (ћ𝜐 )

where ns is the charge carrier and 𝜐 is the Fermi velocity.

Figure ‎1.4 (a) The Brillouin zone and (b) simplified band structure of mono-layer graphene.
The energy bands can be derived using the following equation [10]:

𝐸( ) =

𝜐

=

𝜐 √

Equation ‎1-1

Where ħ is the reduced Plank’s constant and k is the wave vector. The Fermi velocity of
electrons, υF, can be determined from the slop of k-E(k) [10],

𝜐( ) =

=𝜐

Equation 1-2

From the above equations, it can be seen that the slop of k-E(k) is constant everywhere and
6

the electron velocity is equal to Fermi velocity (υF), where υF ≈ 3×10 m/s [11].

4

As discussed previously, unlike semiconductors, the band gap between the valance and
conduction bands of graphene is zero. Therefore, a two-component wave function can be
defined for mono-layer graphene, as given in Equation 1-3.

(

)=[

]

(

)

Equation ‎1-3

Where, s=sng(E) and Ө=arctan(Ky/Kx). Due to this zero band gap, graphene is expected to
5

2

have very high electron mobility of around 2×10 cm V

-1 -1

s

at room temperature [1] and

-1 -1

exceptional thermal conductivity (5000 Wm K ). In addition, the zero band gap of pristine
graphene makes it a metallic behaving material.
1.2.2 Carrier density of graphene
From the physics point of view, one significant factor affecting the electron mobility in
graphene is the carrier density. Theoretically, the density of state increases linearly with
energy as shown in Figure 1.5, and can be discribed by:

(𝐸) = |𝐸| 𝜋

𝜐

Equation ‎1-4

Where ħ is the reduced Plank’s constant, υF is the Fermi velocity of electrons, and E is the
band energy.

5

Figure ‎1.5 Energy and the density of state relationship in graphene
Therefore, the carrier density can be derived by integrating the density of state (D(E))
multiplied by the Fermi function (f0(E)), as shown in the following equations.

𝑛 (𝐸 ) = ∫

(𝐸) (𝐸) 𝐸

𝑛 (𝐸 ) = ∫

𝑛 (𝐸 ) =

∫

𝑛 (𝐸 ) =

Equation ‎1-5

(𝐸) 𝐸

Equation ‎1-6

𝐸 𝐸

Equation ‎1-7

Equation ‎1-8

Increasing the carrier density leads to the decrease of the mobility, due to the dominant
scatterers. In particular, the electron mobility can decrease significantly when graphene is
transferred onto a dielectric substrate, depending on the nature and impurity of the insulator.
As an example, the electron mobility of graphene on SiO2 is reported to be only a few
2 -1 -1

thousands of cm V s [12].

6

1.2.3 Lattice vibration in graphene
Another reason of graphene’s high mobility at room temperature is the behavior of lattice
vibrations (phonons). In conventional semiconductors such as silicon, the scattering of optical
phonons dominates at room temperature, causing a large decrease of the electron mobility
with an increasing temperature [13]. In graphene, electrons do not interact with optical
phonons due to the high energy (ħω~200 meV) of optical phonons [14]. However, electrons
interact with longitudinal acoustic (LA) phonons of graphene and the emission and the
absorption of acoustic phonon lead to momentum loss and generation of resistance. The
contribution of the resistivity from LA phonons is temperature dependent and can be
described by [15]:

( 𝑛) =

Equation ‎1-9

Where kB is the Boltzmann constant, D is the acoustic deformation potential, T is the
4

temperature, υF is the Fermi velocity of electrons, ʋph=2.1×10 m/s is the sound velocity of
LA phonons in graphene, and ρm=6.5×10

-7

2

kg/m is the areal mass density of graphene. The

Equation 1-12 applies when temperature is greater than the Bloch-Gruneisen temperature
(TBG) where TBG=2ħkFʋph/kB. In fact, the effects of the LA scattering on the intrinsic
mobility of graphene are very little at room temperature because of the high Fermi
temperature at high carrier density of graphene [16, 17]. As a result, graphene is expected to
exhibit much higher electron mobility than conventional semiconductors.
1.2.4 Gate-voltage modulating effect
While pristine graphene is known to be metallic, it is possible to modulate the Fermi level
and thus the semiconducting properties of graphene, usually by applying a gate voltage either
7

though a back-, a top-, or a dual-gate configuration. In these cases, the gate voltage is defined
as the voltage applied to a graphene sheet through an insulation layer, such as SiO2. Through
the gate-voltage modulating effect, the type of charge carriers in graphene can be tuned
13

-2

between electrons and holes with a doping concentration (n) as high as 1×10 cm and an
5

2

-1 -1

electron mobility (μ) of greater than 2×10 cm V s

for suspended exfoliated graphene.

Moreover, the sheet conductance is also affected by the gate voltage and its value, as well as
the charge carrier concentration, and can be determined experimentally:

(𝐸 )

(𝐸 ) (

𝑛 =

(

)

)

Equation ‎1-10

Equation ‎1-11

Specifically, Cox=εoεox/tox is the oxide capacitance, in which εo is the vacuum dielectric
constant and εox is the oxide relative dielectric constant, and tox is the oxide thickness.
Graphene capacitance is negligible since the gate oxide thickness is relatively high (300 nm).
Furthermore, the mean-free-path (λ) of charge carriers is inversely proportional to the Fermi
energy (EF), as described in Equation 1-12, and can be expressed by:

(𝐸 ) =

(

)( √

)

Equation ‎1-12

1.2.5 Influences of substrate on graphene mobility
In the absence of the gate voltage, the mobility, by which can be calculated by µ=t/(dσ/dvg)
where, t is the silicon oxide thickness (300nm), 𝜺 is the permittivity of the silicon dioxide
𝜺=8.85e-12 F/m, σ and VG are the conductance and the gate voltage respectively, is mainly
8

limited by impurities and phonon scattering of the substrate where graphene is transferred
onto. Usually during device fabrication, graphene is transferred onto a SiO2 substrate, which
is a polar substrate. Such substrates allow for existence of polar optical phonons localized
near the graphene-substrate interface. This could be an important scattering source of
graphene carriers that gives rise to additional resistance and temperature dependence. This
type of scattering is called remote oxide phonons scattering [12, 18]. At room temperature
there are two optical modes for a SiO2 substrate, ħω1~60 meV and ħω2~150 meV, which
4

2 -1 -1

together resemble an extrinsic mobility limit of around 4×10 cm V s [18].
2 -

For exfoliated graphene on SiO2, the best reported electron mobility is around 20000 cm V
1 -1

s

[19]. Based on the scanning tunneling microscope (STM) studies, suspension of

graphene minimizes the optical phonon scattering and therefore improves the electron
mobility by an order of magnitude [20]. Moreover, graphene that is achieved from silicon
carbide (SiC) or chemical vapor deposition (CVD) methods typically has lower mobility in
2 -1 -1

the order of several thousands of cm V s . The mobility declines due to the lattice defects
and interfacial roughness, which is another source of extrinsic scattering.
In addition, although graphene is chemically inert and does not form chemical bonds with a
SiO2 surface, it is however subjected to charge transfer and Coulomb scattering as a result of
ionized donors. Previously it has been reported that SiO2 surface can transfer electrons to
graphene, resulting in electron doping of graphene. This n-doping was observed after
annealing graphene in high vacuum at 200 °C for 20 hrs [21]. Therefore, to further enhance
electron mobility, it is required to use a chemical inert substrate such as crystalline boron

9

2 -1 -1

nitride substrate [22], with which the electron mobility of up to 60000 cm V s has been
achieved.
1.2.6 Geometrical effects
As the width of graphene sheet shrinks in one dimension to several times of its unit cell, its
electronic properties change accordingly. Graphene sheets with such a narrow width are
called graphene nanoribbons (GNRs). Figure 1.6 shows schematics of two GNRs with an
equal width but two different kinds of edges, armchair and zigzag.

Figure ‎1.6 Illustration of two types of GNR edges: (a) zigzag and (b) armchair. (c)
Confinement in the x-direction quantizes the kx quantization of graphene band into 1-D band.
The electronic states of both the armchair and zigzag are confined in the x-direction. This
quantum confinement leads to quantized momentum, kx=nπ/W, where n=0, 1, 2… and W is
the width of the GNR (in nm). The kx quantization divides the Brillouin zone into discrete
energy bands as shown in Figure 1.6 (c). With a narrow width, the energy gap between
adjacent bands can be enlarged significantly. Theoretically, the energy gap between adjacent
bands is on the order of ΔE~ħʋFπ/W or 2 eV/W (nm), where W is the width of nanoribbon
[23]. The possibility of opening the band gap in GNRs enables the potential application of
graphene in the field of electronic devices. However, achieving a band gap close to
10

conventional semiconductors (e.g. Si) is still challenging since it would require a GNR with a
width of a few nanometers.
In addition to the influence on the band gap, the width (W) of GNRs can significantly affect
the current ON-OFF ratio of graphene-based transistors. For example, the ON-OFF ratio is
6

around 10 when W is wider than 20 nm, whereas it is increased to 10 when W is approaching
2 nm [24, 25]. While a narrower GNR enables a higher ON-OFF ratio, GNR transistors suffer
2 -1 -1

from low electron mobility in the order of several hundreds of cm V s

when W is ≤ 20 nm

[26]. It has been found that, at W < 60 nm, the electron mobility highly depends on the edge
scattering. The edge roughness induced by the etching process creates band edge fluctuations,
which in turn produce electron-hole scattering and lead to the decrease in the carrier mobility.
In addition to the effect of the GNR width upon its electronic properties, the edge
configuration of the GNR is another significant factor. Depending on the edge orientation, the
GNRs with zigzag edges usually exhibit metallic properties, while the GNRs with armchair
edges may exhibit either metallic or semiconductor properties. In particular, an armchair
GNR is metallic when W=(3n+1)a, where W is the width of the nanoribbon, n is an integer,
and a=0.246 nm is the lattice constant of graphene, and is semiconductor otherwise [27, 28].
Practically, the edges often consist of a mixture of zigzag and arm chair termination.
Theoretical calculations show that zigzag edges dominate when the GNR is wider [23].
Despite the metallic properties of the GNR that exist theoretically, most fabricated GNRs are
semiconductors since the edges of the zigzag GNRs could not survive even with a small
surface disorder.

11

1.3

Graphene Applications in Nanoelectronic Circuits

The extraordinary electrical and thermal properties of graphene make it attractive for
applications in nanoscale transistors and nanoelectronic circuits. While such promising
properties have also been reported for carbon nano tubes (CNTs), the fact that graphene can
be fabricated using conventional CMOS-technology is potentially a humongous advantage
over CNTs. Moreover, in order to use CNTs for electronic devices, it is necessary to select
CNTs that show semiconductor properties, which resembles another difficulty for device
fabrication.
At present, the use of graphene in the design and development of nanoelectronic and
integrated circuits can be divided into three main categories. The first category is radio
frequency (RF) circuits, such as RF mixers [29-31], RF amplifier [32], Gigahertz ring
oscillators [33], and frequency multipliers [34], due to the high cutoff frequency and the
ambipolar transport property of graphene. Second, the high thermal conductivity of graphene
makes it a promising alternative material to replace copper for on-chip interconnection. The
main purpose of using graphene as interconnection is to improve heat dissipation, which
resembles a bottleneck in the CMOS integrated circuit design [35-37]. Finally, graphene has
also been explored in constructing logic gate circuits. However, unlike analog circuit and
interconnect applications, the development of graphene-based logic circuits faces critical
challenges that remain unsolved.
First, logic gate circuits require opening the band gap of pristine graphene to convert it from a
metallic material to a semiconducting one. Several methods have been reported to produce
GNRs as techniques for opening the band gap. For example, electron beam lithography has
been demonstrated in the fabrication of GNRs with the width of around 20 nm, but with
rough edges [26]. Chemical exfoliation of graphite can produce GNRs with the width of less

12

than 10 nm and smooth edges [38]. Unzipping multi-walled carbon nanotubes is another
approach to generate narrow ribbons with a width of below 10 nm, and can be achieved by
oxidative process [39] or by plasma etching of carbon nanotubes partially embedded in a
polymer film [40]. However, these methods have high fabrication cost and non-controllability
of the width of the fabricated GNRs.
The second challenge of making graphene logic circuits is the requirement of both n- and ptype transistors. Unlike p-type graphene that can be doped effectively and stably through
chemical methods, chemically-doped n-type graphene is very unstable in air due to oxygen
and impurity doping effects. Specific processes need to be developed in order to fabricate
both n- and p- type graphene monolithically on a single substrate. Previously, Trarersi et al.
[41] demonstrated an electrical annealing technique to clean up a part of a graphene sheet to
restore the n-type property whereas the other part was left as the p-type region. However, this
technique is unpractical since the electrical annealing greatly limits the operation lifetime of
the transistor due to Joule heat induced damage to graphene. Li et al. fabricated a
complementary-like graphene logic circuit, where the current transport properties of graphene
were controlled by electrostatic doping. However, this technique is limited to bilayer
graphene that usually has lower electron mobility and cutoff frequency than monolayer
graphene. Moreover, the aforementioned techniques could not achieve stable n-type graphene
in air due to its high surface sensitivity.
Therefore, the bottleneck in the fabrication of graphene-based logic circuits is the lack of a
practical and reliable method for fabricating air-stable n- and p-type graphene transistors on
the same sheet.

13

1.4

Mission and Outline

In order to address the above challenges, this PhD work presents the development of
innovative techniques that are required for fabricating highly stable and reliable integrated
graphene circuits. More specifically, we have developed a high-throughput and highly
repeatable technique to engineer the number of graphene layers. We have also demonstrated
the ability to engineer the graphene band gap by introducing nano-structures, such as
nanomeshes and nanoribbons into pristine graphene. Moreover, we have investigated a
polymer doping and encapsulation method to realize air-stable n-type graphene transistors.
As a proof of principle study, we have design, fabricated, and characterized an integrated
graphene invertor circuit on mechanically rigid and flexible platforms. Finally, we extended
the range of graphene applications to the areas of chemical and temperature sensing.
The outline of the dissertation is given as follows. Chapter 2 introduces methods for graphene
preparation, with specific focus on micromechanical exfoliation. Different techniques to
identify the number of graphene layers are also discussed. Chapter 3 discusses engineering
methods to control the number of graphene layers using oxygen plasma etching. In addition,
methods to convert pristine graphene to semiconductors based on nanomesh and multiribbon
structures are covered in this chapter. P-type FETs are designed, fabricated, and characterized
to prove the concept. In Chapter 4, different n-doping techniques of thin graphene films are
studied, followed by the demonstration of highly stable n-type FETs. Chapter 5 reports the
design, fabrication, and characterization of graphene integrated circuits on both rigid and
flexible substrates. Finally, Chapter 6 summarizes several potential applications of graphene,
including gas sensors and flow/temperature sensors.

14

REFERENCES

15

REFERENCES

[1]
W. Han, K. M. McCreary, K. Pi, W. H. Wang, Y. Li, H. Wen, J. R. Chen, and R. K.
Kawakami, "Spin transport and relaxation in graphene," Journal of Magnetism and Magnetic
Materials, vol. 324, pp. 369-381, Feb 2012.
[2]
S. V. Morozov, K. S. Novoselov, M. I. Katsnelson, F. Schedin, D. C. Elias, J. A.
Jaszczak, and A. K. Geim, "Giant intrinsic carrier mobilities in graphene and its bilayer,"
Physical Review Letters, vol. 100, p. 4, Jan 2008.
[3]
C. Soldano, A. Mahmood, and E. Dujardin, "Production, properties and potential of
graphene," Carbon, vol. 48, pp. 2127-2150, Jul 2010.
[4]
N. Tombros, S. Tanabe, A. Veligura, C. Jozsa, M. Popinciuc, H. T. Jonkman, and B.
J. van Wees, "Anisotropic spin relaxation in graphene," Physical Review Letters, vol. 101, p.
4, Jul 2008.
[5]
S. O. Koswatta, A. Valdes-Garcia, M. B. Steiner, Y.-M. Lin, and P. Avouris,
"Ultimate RF performance potential of carbon electronics," Microwave Theory and
Techniques, IEEE Transactions on, vol. 59, pp. 2739-2750, 2011.
[6]
R. Nair, P. Blake, A. Grigorenko, K. Novoselov, T. Booth, T. Stauber, N. Peres, and
A. Geim, "Fine structure constant defines visual transparency of graphene," Science, vol. 320,
pp. 1308-1308, 2008.
[7]
P. Avouris, "Graphene: Electronic and photonic properties and devices," Nano
Letters, vol. 10, pp. 4285-4294, Nov 2010.
[8]
A. K. Geim and K. S. Novoselov, "The rise of graphene," Nature materials, vol. 6, pp.
183-191, 2007.
[9]
D. S. L. Abergel, V. Apalkov, J. Berashevich, K. Ziegler, and T. Chakraborty,
"Properties of graphene: a theoretical perspective," Advances in Physics, vol. 59, pp. 261482, 2010.
[10] A. C. Neto, F. Guinea, N. Peres, K. S. Novoselov, and A. K. Geim, "The electronic
properties of graphene," Reviews of Modern Physics, vol. 81, p. 109, 2009.
[11] D. Elias, R. Gorbachev, A. Mayorov, S. Morozov, A. Zhukov, P. Blake, L. P. I.
Grigorieva, K. Novoselov, F. Guinea, and A. Geim, "Dirac cones reshaped by interaction
effects in suspended graphene," Nature Physics, vol. 7, pp. 701-704, 2011.
[12] J. H. Chen, C. Jang, S. D. Xiao, M. Ishigami, and M. S. Fuhrer, "Intrinsic and
extrinsic performance limits of graphene devices on SiO2," Nature Nanotechnology, vol. 3,
pp. 206-209, Apr 2008.

16

[13] G. Abstreiter, "Electronic-properties of the two-dimensional system at gaas-alxga1xasinterfaces," Surface Science, vol. 98, pp. 117-125, 1980.
[14] A. C. Betz, F. Vialla, D. Brunel, C. Voisin, M. Picher, A. Cavanna, A. Madouri, G.
Feve, J. M. Berroir, B. Placais, and E. Pallecchi, "Hot electron cooling by acoustic phonons
in graphene," Physical Review Letters, vol. 109, p. 5, Aug 2012.
[15] E. H. Hwang and S. Das Sarma, "Acoustic phonon scattering limited carrier mobility
in two-dimensional extrinsic graphene," Physical Review B, vol. 77, p. 6, Mar 2008.
[16] D. K. Efetov and P. Kim, "Controlling electron-phonon interactions in graphene at
ultrahigh carrier densities," Physical Review Letters, vol. 105, p. 4, Dec 2010.
[17] E. H. Hwang and S. Das Sarma, "Surface polar optical phonon interaction induced
many-body effects and hot-electron relaxation in graphene," Physical Review B, vol. 87, p.
10, Mar 2013.
[18] K. Zou, X. Hong, D. Keefer, and J. Zhu, "Deposition of high-quality HfO2 on
graphene and the effect of remote oxide phonon scattering," Physical Review Letters, vol.
105, p. 4, Sep 2010.
[19] S. Morozov, K. Novoselov, M. Katsnelson, F. Schedin, D. Elias, J. Jaszczak, and A.
Geim, "Giant intrinsic carrier mobilities in graphene and its bilayer," Physical Review Letters,
vol. 100, p. 016602, 2008.
[20] E. Stolyarova, K. T. Rim, S. M. Ryu, J. Maultzsch, P. Kim, L. E. Brus, T. F. Heinz,
M. S. Hybertsen, and G. W. Flynn, "High-resolution scanning tunneling microscopy imaging
of mesoscopic graphene sheets on an insulating surface," Proceedings of the National
Academy of Sciences of the United States of America, vol. 104, pp. 9209-9212, May 2007.
[21] H. E. Romero, N. Shen, P. Joshi, H. R. Gutierrez, S. A. Tadigadapa, J. O. Sofo, and P.
C. Eklund, "n-Type behavior of graphene supported on Si/SiO2 substrates," ACS Nano, vol.
2, pp. 2037-2044, Oct 2008.
[22] C. R. Dean, A. F. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K. Watanabe, T.
Taniguchi, P. Kim, K. L. Shepard, and J. Hone, "Boron nitride substrates for high-quality
graphene electronics," Nature Nanotechnology, vol. 5, pp. 722-726, Oct 2010.
[23] K. Nakada, M. Fujita, G. Dresselhaus, and M. S. Dresselhaus, "Edge state in graphene
ribbons: Nanometer size effect and edge shape dependence," Physical Review B, vol. 54, pp.
17954-17961, Dec 1996.
[24] M. Y. Han, B. Ozyilmaz, Y. B. Zhang, and P. Kim, "Energy band-gap engineering of
graphene nanoribbons," Physical Review Letters, vol. 98, p. 4, May 2007.
[25] Z. H. Chen, Y. M. Lin, M. J. Rooks, and P. Avouris, "Graphene nano-ribbon
electronics," Physica E-Low-Dimensional Systems & Nanostructures, vol. 40, pp. 228-232,
Dec 2007.
17

[26] Y. X. Yang and R. Murali, "Impact of size effect on graphene nanoribbon transport,"
IEEE Electron Device Letters, vol. 31, pp. 237-239, Mar 2010.
[27] S. M. M. Dubois, Z. Zanolli, X. Declerck, and J. C. Charlier, "Electronic properties
and quantum transport in graphene-based nanostructures," European Physical Journal B, vol.
72, pp. 1-24, Nov 2009.
[28] M. Ezawa, "Peculiar width dependence of the electronic properties of carbon
nanoribbons," Physical Review B, vol. 73, p. 8, Jan 2006.
[29] Y.-M. Lin, A. Valdes-Garcia, S.-J. Han, D. B. Farmer, I. Meric, Y. Sun, Y. Wu, C.
Dimitrakopoulos, A. Grill, and P. Avouris, "Wafer-scale graphene integrated circuit,"
Science, vol. 332, pp. 1294-1297, 2011.
[30] H. Wang, A. Hsu, J. Wu, J. Kong, and T. Palacios, "Graphene-based ambipolar RF
mixers," Electron Device Letters, IEEE, vol. 31, pp. 906-908, 2010.
[31] L. Liao, J. Bai, R. Cheng, H. Zhou, L. Liu, Y. Liu, Y. Huang, and X. Duan, "Scalable
fabrication of self-aligned graphene transistors and circuits on glass," Nano letters, vol. 12,
pp. 2653-2657, 2011.
[32] S.-J. Han, K. A. Jenkins, A. Valdes Garcia, A. D. Franklin, A. A. Bol, and W.
Haensch, "High-frequency graphene voltage amplifier," Nano letters, vol. 11, pp. 3690-3693,
2011.
[33] E. Guerriero, L. Polloni, M. Bianchi, A. Behnam, E. Carrion, L. G. Rizzi, E. Pop, and
R. Sordan, "Gigahertz integrated graphene ring oscillators," ACS nano, vol. 7, pp. 558-5594
2013.
[34] H. Wang, D. Nezich, J. Kong, and T. Palacios, "Graphene frequency multipliers,"
Electron Device Letters", vol. 30, pp. 547-549, 2009.
[35] X. Chen, D. Akinwande, K.-J. Lee, G. F. Close, S. Yasuda, B. C. Paul, S. Fujita, J.
Kong, and H. Wong, "Fully integrated graphene and carbon nanotube interconnects for
gigahertz high-speed CMOS electronics," Electron Devices, IEEE Transactions on, vol. 57,
pp. 3137-3143, 2010.
[36] Y. Chen, Y. Xu, K. Zhao, X. Wan, J. Deng, and W. Yan, "Towards flexible all-carbon
electronics: Flexible organic field-effect transistors and inverter circuits using solutionprocessed all-graphene source/drain/gate electrodes," Nano Research, vol. 3, pp. 714-721,
2010.
[37] W. J. Yu, S. Y. Lee, S. H. Chae, D. Perello, G. H. Han, M. Yun, and Y. H. Lee,
"Small hysteresis nanocarbon-based integrated circuits on flexible and transparent plastic
substrate," Nano letters, vol. 11, pp. 1344-1350, 2011.

18

[38] X. L. Li, X. R. Wang, L. Zhang, S. W. Lee, and H. J. Dai, "Chemically derived,
ultrasmooth graphene nanoribbon semiconductors," Science, vol. 319, pp. 1229-1232, Feb
2008.
[39] D. V. Kosynkin, A. L. Higginbotham, A. Sinitskii, J. R. Lomeda, A. Dimiev, B. K.
Price, and J. M. Tour, "Longitudinal unzipping of carbon nanotubes to form graphene
nanoribbons," Nature, vol. 458, pp. 872-U5, Apr 2009.
[40] L. Y. Jiao, L. Zhang, X. R. Wang, G. Diankov, and H. J. Dai, "Narrow graphene
nanoribbons from carbon nanotubes," Nature, vol. 458, pp. 877-880, Apr 2009.
[41] F. Traversi, V. Russo, and R. Sordan, "Integrated complementary graphene inverter,"
Applied Physics Letters, vol. 94, Jun 2009.

19

2

2.1

GRAPHENE PREPARING METHODS

Introduction

High quality graphene thin films are required for making the proposed graphene transistors
and nanoelectronic circuits. At present, the most commonly used methods for graphene
production are micromechanical exfoliation/cleavage, epitaxial synthesis from silicon carbide
(SiC) substrates, and chemical vapor deposition (CVD). This chapter provides a brief
technical introduction to these three methods and compares their differences from several
aspects such as graphene size, quality, and production yield. The methods to identify the
layer number and quality of graphene are also discussed.
2.2

Methods to Produce Graphene

2.2.1 Micromechanical exfoliation and cleavage
Mechanical exfoliation and cleavage use mechanical energy to break weak van der Waals
bonds between stacked graphene sheets in order to separate individual sheets from highpurity graphite flakes. Micromechanical exfoliation was first introduced by Novoselov and
his group in 2004 [1], which typically involves a repeated peeling process. This method
enables production of high-quality graphene in a very reliable and low-cost way, and thus has
attracted great attention from researchers since it was invented. The exfoliation process is
usually done in a clean environment to avoid any kind of contamination.
We have calibrated a robust process to exfoliate graphene samples using the KECK
cleanroom facility at Michigan State University. Prior to the mechanical exfoliation, a
substrate of 300 nm SiO2 /Si (purchased from WRS Materials) was cleaned with acetone and
rinsed in isopropanol alcohol (IPA) and deionized (DI) water. After drying the substrate with
a nitrogen blower, two metal layers of 70 nm Au/5 nm Ti micro-markers (Figure 2.1) were
20

patterned on the SiO2 substrate via photolithography, thermal evaporation, and metal liftoff
processes. The minimum feature size of the markers was around 3 µm and the spacing
between them was 150 µm. The purpose of these markers is to locate graphene sheets on the
substrate and to use them in the alignment process of the electron beam lithography (EBL) in
subsequent fabrication steps .

Figure ‎2.1 Metal markers on a SiO2/Si substrate.
Graphene samples were exfoliated from commercially available, highly oriented pyrolytic
graphite (HOPG) flakes of around 1 mm. After attaching these graphite flakes to a ~20 cm
long Scotch tape with tweezers, the tape was folded to sandwich the flakes between the sticky
sides, pressed to ensure a good adhesion, and then slowly peeled to cleave graphite into two
layers. The tape was repeatedly folded and peeled until a very smooth layer of fine graphene
sheets was achieved on the tape (Figure 2.2). After that, the tape with cleaved graphene was
laid and pressed on the SiO2/Si substrate in order to transfer graphene sheets onto the
substrate. The tape was then slowly removed, leaving the graphene attached on the substrate.
Finally, transferred graphene was cleaned in acetone, IPA, and DI water, and was ready for
subsequent processes.

21

Figure ‎2.2 (a) Mechanical exfoliation of graphite using a Scotch tape. (b) A 300 nm SiO2/Si
substrate with transferred graphene. The dashed circle indicates a thin graphene sheet.
2.2.2 Epitaxial graphene from silicon carbide
Epitaxial growth is a commonly used technique for producing large-scale graphene from a
SiC substrate. In this method, when SiC is heated to around 1400 °C in vacuum, some of the
silicon atoms sublimate from the surface, which results in the formation of a graphene layer
on top of the SiC substrate [2]. It has been reported that graphene transistors can be
manufactured from a SiC substrate by epitaxial growth of graphene on a wafer-scale size [3].
The advantage of this method is that a device can be fabricated on a SiC substrate itself,
without the need for transferring graphene to another substrate. However, graphene sheets
generated by this method are non-homogeneous so the number of the layer cannot by
precisely controlled, as shown in Figure 2.3. In addition, the electron mobility is limited to
2

-1 -1

few thousands of cm V s

[4]. Therefore, although the epitaxial growth of exitaxial

graphene is capable of preparing graphene with high throughput and large areas, it is
relatively expensive and difficult to obtain uniform and high-quality graphene.

22

Figure ‎2.3 Mono-, bi- and tri-layer graphene produced by graphitization of SiC [4].
2.2.3 Chemical vapour deposition (CVD)
Another alternative method to produce graphene is a CVD method, which is capable of
obtaning large graphene sheets (~1 m) [5] with reasonable good electron mobility (~10000
2

-1 -1

cm V s ) [5, 6]. In the CVD approach, a copper substrate is typically used for growing
graphene because copper does not absorb carbon atoms due to the weak bonds that hold the
carbon atoms to copper. Consequently, carbon atoms can only form on the surface of copper.
During the CVD synthesis, the copper substrate is inserted into a furnace and heated at low
vacuum to around 1000 °C in order to increase the grain size. Methan and hydrogen gases are
flowed through the furnace. The function of the hydrogen is to facilitate the reaction between
the methan molecules and the copper substrate, which enables carbon atoms to deposit on the
substrate. After that, the sample is rapidly cooled to achieve thin graphene sheets [7, 8].
One of the favorable features in the CVD method is the ability of transfering CVD graphene
from coper substrate to different other substrates, such as Si, SiO2, or polymers [5, 9].
Typically CVD graphene transfer can be accomplished by depositing a large area of
polymethyl methacrylate (PMMA) on a graphene/copper foil, followed by etching the copper
substrate in Ferric Cloride (FeCl3). After the copper is completed etched away, the
23

graphene/PMMA film floating on the FeCl3 solution will be cleaned with DI water and then
transferred to another substrate. Finally, the PMMA can be dissolved in acetone, leaving only
the graphene on the new substrate [7].
Compared to the mechanical exforliation and epitaxial growth, CVD synthesis cannot achieve
high quality graphene, mainly due to surface disorder, defects, and different types of
contamination (e.g. Cu residues) that require multiple cleaning processes (Figure 2.4 ).

Figure ‎2.4 Optical microscopic image of a CVD graphene sample on a SiO2/Si substrate.
2.2.4 Comparison of three methods
Table 2.1 compares the three representative methods for obtaining graphene films,
in terms of film size, electron mobility, graphene transfer feasibility, cost, and mass
production capability. Besides the abovementioned main techniques, attempts have been
made to synthesize graphene through other approaches, such as un-zipping carbon nanotubes
(CNTs) [10, 11], reduction of graphite oxide [12], or chemical extraction of graphene films
from graphite [13, 14]. Among these methods, the mechanical exfoliation method is
considered as the most effective and efficient way to produce high quality graphene films for
research prototypes. For this reason, the pilot studies in this PhD work mainly use the
mechanically exfoliated graphene to prove the principles.

24

Table 2.1 Three representative methods for producing graphene films
Method

2.3

Mechanical

Epitaxial growth

CVD synthesis on

Exfoliation

on SiC

Ni, Cu, Fe, or Co

Size

10-100 µm

Mobility

Best

Bad

High

Transfer

Yes

Yes

Yes

Cost

Low

High

High

Mass production

No

Yes

Yes

> 6 inch

> 6 inch

Methods to Identify Graphene

The number of graphene layers is a very important factor that determines the properties of
graphene. The thickness of graphene films can be differentiated using one of the three
techniques: optical microscopy, atomic force microscopy (AFM), and Raman spectroscopy.
Details for each method are given below.
2.3.1 Optical microscopy
Microscale graphene sheets can be visualized under an optical microscope fitted with a ×50
objective lens. The noticeable optical contrast between the graphene and the 300 nm SiO2/Si
allows for a rough estimate on the number of carbon layers [15, 16]. Theoretically, thicker
graphene films have higher optical contrasts between the graphene and the substrate, which
can be calculated by C(λ)=RS(λ)-RG(λ)/RS(λ) [17, 18], where RS(λ) is the reflection spectrum
from the 300 nm SiO2/Si substrate and RG(λ) is the reflection spectrum of the graphene sheet
with a given wavelength λ. With this technique, the number of carbon layers in a graphene
sheet can be determined precisely [17] by C(λ)=0.0046+0.0925-0.00255N2, where N ≤ 10 is
25

the number of carbon layers. Experimentally, the number of graphene layers can be
distinguished just by the color of the graphene sheet under a high magnification optical
microscope. With only little experience, one can differentiate between mono-, bi-, and few(3 to 6) layers, and graphite (typically with more than 9 layers). Figure 2.5 shows the optical
microscopic images of graphene sheets with different layer numbers. While optical
microscopy is a simple and fast way to estimate the number of graphene layers, more
advanced tools will be needed in order to determine the number of layer accurately.

Figure ‎2.5 Optical microscopic images for graphene samples with different layer numbers:
(a) monolayer; (b) bilayer; (c) 3 layer; (d) and (e) 4 to 6 layers; (f) to (h) more than 10 layers.
2.3.2 Atomic force microscopy (AFM)
The tapping mode technique of AFM is a commonly used method to measure the thickness of
graphene on SiO2. It also allows topographic imaging of sample surfaces with high
resolution. Theoretically, since the thickness of the monolayer graphene sheet is known
(0.335 nm), it is easy to determine the number of graphene layers from the total thickness
measured by AFM. However, different groups have reported different thicknesses for a single
graphene layer [1, 19-21] resulted from AFM measurements. We obtained similar results in
our experiments, where the measured thickness of a monolayer graphene sheet was always
higher than the theoretical value. As an example, Figure 2.6 shows the thicknesses of four
typical monolayer graphene samples, measured by the DI 3100 AFM in the KECK
cleanroom. The monolayer configuration of these samples later was confirmed using Raman
spectroscopy, which will be discussed in the next section.
26

Figure ‎2.6 AFM (DI 3100) thickness measurements of four monolayer graphene samples.
The thickness ranges from 0.676 nm to 0.829 nm.
The inaccuracy of the AFM thickness measurements can be attributed to several reasons: the
folded regions in graphene [22], sample contamination, the nanoscale roughness of the SiO2
substrate, the fluctuation in the interaction (or attractive forces) between the AFM tip and the
different surfaces when scanning the across graphene and SiO2 surfaces [23], and finally an
unclean or worn AFM cantilever. Hence, it is relatively difficult to determine the exact
number of graphene layers by just relying on the AFM method.
2.4

Raman Spectroscopy

Raman spectroscopy is based on inelastic scattering of laser light, meaning that the frequency
of photons in laser changes due to interaction with a sample. Particularly, this interaction
generates electric dipole moment that deforms the molecules of the sample. Consequently,
the laser photons reemit with a shift in the original frequency that provides information about
27

the vibration of the molecules in the sample (phonons) [24]. Figure 2.7 (a) shows typical
Raman spectra of several graphene sheets with various thicknesses. The Raman spectra were
obtained from a YAG laser (λ=532 nm) with the laser power of 0.53 mW. The most
significant Raman characteristic peaks of graphene are G band at ~1580 cm
(2D) band at ~2700 cm

-1

-1

and second D

[25]. The G band is generated due to the doubly generated optical

phonons, in both longitudinal and transverse modes, at the center of the Brilloun zone [26,
27]. These modes are high frequency modes, corresponding to the vibration frequency of
atoms in the unit cell against each other. As shown in Figure 2.7 (a), the peak of the G band
in graphene increases as the graphene thickness increases, simply because of the increment of
carbon atoms. Hence, the peak of the G band can be used to determine the number of
graphene layers [25].
The other significant band in the Raman spectrum is the 2D band. As depicted in the same
figure, the peak of the 2D band is the key point to differentiate the mono-, bi-, tri- and fewlayers of graphene sheets. In the monolayer graphene, the 2D peak is sharper and stronger
than the G peak. Whereas the 2D peak of the bi-layer graphene has four edges (Figure 2.7
(b)) and its intensity is close to that of the G peak. It is also observed that the 2D peak of the
tri-layer graphene contains only two edges and the peak intensity is close to that of the G
peak. Furthermore, it is relatively difficult to differentiate graphene sheets with four to six
layers, because they have similar shapes in their 2D peaks, with the typical peak intensity I2D
of around 70% of the G peak intensity IG. For graphene sheets with more than six layers, the
shape of their 2D band becomes closer to the 2D shape of graphite. Basically, as the number
of graphene layers increases, new electronic bands will be generated, which leads to more
electron-phonon interactions and thus different peaks appear accordingly.

28

Figure ‎2.7 (a) Raman spectra of different numbers of graphene sheet layers obtained using
HoloProbe Micro-Raman Spectrograph; (b) The 2D band Raman spectrum of a bilayer
graphene sheet, showing 4 edges in one big peak.
-1

As another important observation in Figure ‎2.8, a new band in ~1350 cm , denoted as D,
arises when the graphene surface has defects or mechanical deformation. The intensity of the
D band increases with the increasing defect level of the graphene surface [28, 29]. The

29

defects can be classified as missing atoms, dangling or dislocation bonds in the carbon lattice.
It is of note that, with the increasing number of graphene layer, the D-band intensity ID can
be seen to dramatically decline relative to the G-band intensity IG, which is possibly due to
the increase in rigidity of the thick graphene samples [30].

Figure ‎2.8 Raman spectra of a same graphene sample before and after introducing defects
using O2 plasma with 70 W RF power for 19 sec. The D band appeared as a result of surface
defects.
2.5

Graphene Cleaning Method

As discussed earlier in this chapter, we mainly used the mechanically exfoliated graphene in
our experiments. One major drawback of the mechanical exfoliation is that it leaves a lot of
tape residues on the graphene surface and the SiO2 substrate, which significantly reduce the
carrier mobility of graphene [31, 32]. We found that the tape residues can be removed
effectively by annealing the as-transferred graphene at 400 °C in Ar (97%) /H2 (3%)
atmosphere for 1 h. The gas mixture was introduced into the annealing furnace with the flow
rate of 2000 sccm. Figure 2.9 shows the AFM images of a monolayer graphene sample before
and after the cleaning treatment. It can be seen that after the annealing, the graphene appeared
30

with a smoother surface on the substrate, mainly due to the removal of surficial
contamination and the recovery of defects. The measurement of the sheet thickness, after
cleaning, is closer to the thickness of the reported monolayer graphene (~0.7 nm).

Figure ‎2.9 AFM images of a monolayer graphene before (left image) and after (right image)
a heat treatment for removing tape residues. The sheet thickness decreased from 1.260 nm to
0.676 nm after cleaning.
After the cleaning process, the sample was checked by a Raman spectroscope to investigate
the effect of Ar/H2 annealing on the graphene sheet. As shown in Figure ‎2.10, the intensity
ratio I2D/IG was decreased from 2.4 to 1.27, indicating that the graphene surface has been
doped after the heat treatment. In addition, the defect peak was suppressed after annealing,
due to the defect recovery in an H2 environment. The graphene doping perhaps is due to the
adsorption of hydrogen atoms on the graphene film [33, 34]. However, this doping source
does not show a significant impact on graphene electronic properties since the H2
concentration is only 3%. The chemical doping of graphene will be discussed in details later
in Chapter 4.

31

`
Figure ‎2.10 Raman spectra of a monolayer graphene sample before (black color) and after
(red color) the cleaning treatment.
2.6

Conclusion

In conclusion, graphene produced via CVD is not as high quality as via mechanical
exfoliation due to surface disorder, defects, in addition to the different types of contamination
that require multiple cleaning processes. Therefore, the studies in this thesis will use
mechanical exfoliation for graphene preparation.
Raman spectroscopy is the best method to identify graphene compared to the AFM and
optical microscopy methods. On the other hand, optical microscopy technique resembles a
fast and cheap tool to visualize thin graphene, but experience is required for identifying the
number of graphene layers.

32

REFERENCES

33

REFERENCES

[1]

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, vol. 306, pp. 666-669, Oct 2004.

[2]

E. Moreau, F. J. Ferrer, D. Vignaud, S. Godey, and X. Wallart, "Graphene growth by
molecular beam epitaxy using a solid carbon source," Physica Status Solidi aApplications and Materials Science, vol. 207, pp. 300-303, Feb 2010.

[3]

J. Kedzierski, P.-L. Hsu, P. Healey, P. W. Wyatt, C. L. Keast, M. Sprinkle, C. Berger,
and W. A. de Heer, "Epitaxial graphene transistors on SiC substrates," Electron
Devices, IEEE Transactions on, vol. 55, pp. 2078-2085, 2008.

[4]

K. V. Emtsev, A. Bostwick, K. Horn, J. Jobst, G. L. Kellogg, L. Ley, J. L.
McChesney, T. Ohta, S. A. Reshanov, and J. Röhrl, "Towards wafer-size graphene
layers by atmospheric pressure graphitization of silicon carbide," Nature materials,
vol. 8, pp. 203-207, 2009.

[5]

S. Bae, H. Kim, Y. Lee, X. Xu, J.-S. Park, Y. Zheng, J. Balakrishnan, T. Lei, H. R.
Kim, and Y. I. Song, "Roll-to-roll production of 30-inch graphene films for
transparent electrodes," Nature nanotechnology, vol. 5, pp. 574-578, 2010.

[6]

Q. Yu, J. Lian, S. Siriponglert, H. Li, Y. P. Chen, and S.-S. Pei, "Graphene segregated
on Ni surfaces and transferred to insulators," Applied Physics Letters, vol. 93, p.
113103, 2008.

[7]

T. F. Chung, T. Shen, H. Cao, L. A. Jauregui, W. Wu, Q. Yu, D. Newell, and Y. P.
Chen, "Synthetic Graphene Grown by Chemical Vapor Deposition on Copper Foils,"
International Journal of Modern Physics B, vol. 27, 1341002, 2013.

[8]

X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner, A. Velamakanni, I. Jung, and
E. Tutuc, "Large-area synthesis of high-quality and uniform graphene films on copper
foils," Science, vol. 324, pp. 1312-1314, 2009.

[9]

A. Reina, X. Jia, J. Ho, D. Nezich, H. Son, V. Bulovic, M. S. Dresselhaus, and J.
Kong, "Large area, few-layer graphene films on arbitrary substrates by chemical
vapor deposition," Nano letters, vol. 9, pp. 30-35, 2008.

[10]

D. V. Kosynkin, A. L. Higginbotham, A. Sinitskii, J. R. Lomeda, A. Dimiev, B. K.
Price, and J. M. Tour, "Longitudinal unzipping of carbon nanotubes to form graphene
nanoribbons," Nature, vol. 458, pp. 872-876, 2009.
34

[11]

L. Jiao, L. Zhang, X. Wang, G. Diankov, and H. Dai, "Narrow graphene nanoribbons
from carbon nanotubes," Nature, vol. 458, pp. 877-880, 2009.

[12]

W. Choi, I. Lahiri, R. Seelaboyina, and Y. S. Kang, "Synthesis of Graphene and Its
Applications: A Review," Critical Reviews in Solid State and Materials Sciences, vol.
35, pp. 52-71, 2012/06/06.

[13]

S. Park and R. S. Ruoff, "Chemical methods for the production of graphenes," Nat
Nano, vol. 4, pp. 217-224, 2009.

[14]

M. Choucair, P. Thordarson, and J. A. Stride, "Gram-scale production of graphene
based on solvothermal synthesis and sonication," Nat Nano, vol. 4, pp. 30-33, 2009.

[15]

P. Blake, E. W. Hill, A. H. C. Neto, K. S. Novoselov, D. Jiang, R. Yang, T. J. Booth,
and A. K. Geim, "Making graphene visible," Applied Physics Letters, vol. 91, p. 3,
Aug 2007.

[16]

S. Roddaro, P. Pingue, V. Piazza, V. Pellegrini, and F. Beltram, "The optical visibility
of graphene: Interference colors of ultrathin graphite on SiO2," Nano Letters, vol. 7,
pp. 2707-2710, Sep 2007.

[17]

Z. H. Ni, H. M. Wang, J. Kasim, H. M. Fan, T. Yu, Y. H. Wu, Y. P. Feng, and Z. X.
Shen, "Graphene thickness determination using reflection and contrast spectroscopy,"
Nano Letters, vol. 7, pp. 2758-2763, Sep 2007.

[18]

D. S. L. Abergel, A. Russell, and V. I. Fal'ko, "Visibility of graphene flakes on a
dielectric substrate," Applied Physics Letters, vol. 91, p. 3, Aug 2007.

[19]

Z. H. Chen, Y. M. Lin, M. J. Rooks, and P. Avouris, "Graphene nano-ribbon
electronics," Physica E-Low-Dimensional Systems & Nanostructures, vol. 40, pp.
228-232, Dec 2007.

[20]

A. N. Sidorov, M. M. Yazdanpanah, R. Jalilian, P. J. Ouseph, R. W. Cohn, and G. U.
Sumanasekera, "Electrostatic deposition of graphene," Nanotechnology, vol. 18, p. 4,
Apr 2007.

[21]

C. Casiraghi, A. Hartschuh, E. Lidorikis, H. Qian, H. Harutyunyan, T. Gokus, K. S.
Novoselov, and A. C. Ferrari, "Rayleigh imaging of graphene and graphene layers,"
Nano Letters, vol. 7, pp. 2711-2717, Sep 2007.

[22]

K. Novoselov, D. Jiang, F. Schedin, T. Booth, V. Khotkevich, S. Morozov, and A.
Geim, "Two-dimensional atomic crystals," Proceedings of the National Academy of
Sciences of the United States of America, vol. 102, pp. 10451-10453, 2005.

35

[23]

P. Nemes-Incze, Z. Osvath, K. Kamaras, and L. P. Biro, "Anomalies in thickness
measurements of graphene and few layer graphite crystals by tapping mode atomic
force microscopy," Carbon, vol. 46, pp. 1435-1442, Sep 2008.

[24]

D. J. Gardiner, P. R. Graves, and H. J. Bowley, Practical Raman Spectroscopy vol.
21: Springer-Verlag Heidelberg,, Germany, 1989.

[25]

F. Tuinstra and J. L. Koenig, "Raman spectrum of graphite," Journal of Chemical
Physics, vol. 53, pp. 1126-&, 1970.

[26]

M. Pimenta, G. Dresselhaus, M. S. Dresselhaus, L. Cancado, A. Jorio, and R. Saito,
"Studying disorder in graphite-based systems by Raman spectroscopy," Physical
Chemistry Chemical Physics, vol. 9, pp. 1276-1290, 2007.

[27]

M. A. Pimenta, G. Dresselhaus, M. S. Dresselhaus, L. G. Cancado, A. Jorio, and R.
Saito, "Studying disorder in graphite-based systems by Raman spectroscopy,"
Physical Chemistry Chemical Physics, vol. 9, pp. 1276-1291, 2007.

[28]

C. Thomsen and S. Reich, "Doable resonant Raman scattering in graphite," Physical
Review Letters, vol. 85, pp. 5214-5217, Dec 2000.

[29]

E. B. Barros, N. S. Demir, A. G. Souza, J. Mendes, A. Jorio, G. Dresselhaus, and M.
S. Dresselhaus, "Raman spectroscopy of graphitic foams," Physical Review B, vol. 71,
p. 5, Apr 2005.

[30]

A. Gupta, G. Chen, P. Joshi, S. Tadigadapa, and P. C. Eklund, "Raman scattering
from high-frequency phonons in supported n-graphene layer films," Nano Letters, vol.
6, pp. 2667-2673, Dec 2006.

[31]

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
Communications, vol. 146, pp. 351-355, Jun 2008.

[32]

J.-H. Chen, C. Jang, S. Xiao, M. Ishigami, and M. S. Fuhrer, "Intrinsic and extrinsic
performance limits of graphene devices on SiO2," Nature Nanotechnology, vol. 3, pp.
206-209, 2008.

[33]

R. Balog, B. Jørgensen, L. Nilsson, M. Andersen, E. Rienks, M. Bianchi, M. Fanetti,
E. Lægsgaard, A. Baraldi, and S. Lizzit, "Bandgap opening in graphene induced by
patterned hydrogen adsorption," Nature Materials, vol. 9, pp. 315-319, 2010.

[34]

B. H. Kim, S. J. Hong, S. J. Baek, H. Y. Jeong, N. Park, M. Lee, S. W. Lee, M. Park,
S. W. Chu, and H. S. Shin, "N-type graphene induced by dissociative H2 adsorption at
room temperature," Scientific Reports, vol. 2, dio:10.1038 ,2012.
36

3

3.1

ENGINEERING OF GRAPHENE SHEET

Introduction

In this chapter, two technologies will be discussed for controlling graphene electronic
properties. The first section describes a controllable post-synthesis method for engineering
the number of graphene layers. The second section presents a fabrication technique of
graphene nanostructures, including graphene nanomesh and graphene multi-ribbon, for
tuning the band gap of graphene.
3.2

Layer Engineering of Graphene Sheets

3.2.1 Background
As mentioned in Chapter 2, significant progresses have been made in producing graphene
with both high yields and big sizes. However, a challenge remains to generate graphene with
the desired number of layers. The ability to control the number of graphene layers is critical
for the reproducibility of graphene-based devices in practical applications as well as for the
fundamental study of graphene characteristics. Especially, the layer engineering of graphene
will contribute greatly to the in-depth understanding of inter-layer transport properties in an
inter-layer structured graphene sensor that will be discussed in Chapter 6. The influence of
the number of graphene layers on graphene’s electronic and optical properties has also been
reported [1].
Over the past years, several efforts have been made towards the realization of the layer
engineering of graphene [2-5]. For example, Tour et al. [6] reported the layer-by-layer
removal of graphene sheets with a wet etching method by sputtering zinc on graphene
surfaces and dissolving the zinc with diluted acid. This method provides a promising method
for engineering the number of graphene layers, but the requirement of acid treatments makes
37

this method incompatible with conventional microelectronics manufacturing techniques.
2

Recently, the atomic layer etching of graphene was also reported by oxidization of carbon sp
3

bonds to sp bonds with oxygen radicals and the consequent bombardment with argon atoms
[7]. Although this technique allows for the singular sheet etching of graphene, a large amount
of defects could be introduced to the graphene due to the combination of radical oxidation
and argon atom bombardment.
We proposed a simpler and less invasive method for engineering the number of graphene
layers with pure oxygen plasma etching. By carefully tuning the variables of the oxygen
plasma etching (e.g. power, oxygen flow rate, operating pressure and process duration), we
were able to achieve “singular sheet etching” (SSE) of graphene, which removed only one
graphene sheet at a time with both the powered electrode and ground electrode oxygen
plasma. Due to the fact that plasma strength at the ground electrode is lower than that at the
powered electrode, the “singular sheet etching” by the ground oxygen plasma can potentially
introduce relatively less amount of defects to graphene. As a technical demonstration,
monolayer graphene was produced from bilayer graphene by the “singular sheet etching”
processes. Technical details will be discussed in the following sections.
3.2.2 Singular sheet etching method
Graphene flakes were produced with mechanical exfoliation from highly oriented pyrolytic
graphite (HOPG) flakes and then transferred onto a Si substrate with a 300 nm thick SiO 2
layer. Optical microscopy and Raman spectroscopy (HoloProbe Micro-Raman Spectrograph)
were used to locate and identify single- and double-layer graphene samples. Prior to oxygen
plasma etching, the samples were cleaned with acetone rinse for 5 minutes, followed by an
annealing process. The annealing treatment was carried out at 400 ºC in an argon and
38

hydrogen (3% in volume percentage) atmosphere for 1 hr. It should be noted that the acetone
rinse is used to remove tape residues and other organic contaminations prior to the annealing.
Following the acetone clean, the graphene samples were rinsed with isopropanol alcohol and
then DI water to remove the acetone residue so the chemical effect on graphene doping is
negligible.
The SSE of graphene with oxygen plasma was conducted in a plasma etching system (March,
PX-250) under the following conditions: RF frequency of 13.56 MHz, RF power of 70 W,
oxygen flow rate of 10 sccm (standard cubic centimeter per minute), base pressure of 70
mTorr, and processing pressure of 313-326 mTorr. The etching durations were fine-tuned
after a large amount of experiments and varied according to the location of graphene samples
(at a powered electrode or ground electrode). Specifically in our plasma etching system, the
powered electrode refers to the electrode plate that is connected to an RF bias, while the
ground electrode plate is connected to the inner wall of the vacuum chamber and also to the
ground of the etching system. Figure 3.1 shows the experimental setup of the SSE of
graphene on the ground electrode. Since the strengths of oxygen plasma at the two electrodes
are different, the etching effects on graphene may vary. The ground and powered electrodes
are made of aluminum and have dimensions of 7 in. × 8 in. ×0.093 in. The distance between a
ground electrode and the powered one is 2 in. It is clear that the intensity of plasma is higher
in the powered electrode than the ground ones based on the brightness. The gentle plasma
generated on the ground electrode explains why it required longer time to etch a single layer
and generated fewer defects as confirmed later.
The process durations for single sheet removal of graphene were carefully tuned after a large
number of experiments and varied according to the location of graphene samples (at powered
electrode or ground electrode). AFM and SEM (scanning electron microscope, Hitachi-4700)

39

were used to inspect the surface topologies of graphene samples before and after being
treated with oxygen plasma. Micro Raman spectroscopy was used to identify the number of
graphene layers as well as to study quantitatively the defects of the plasma treated graphene
samples.

Figure ‎3.1 Conceptual schematic and experimental setup of the layer engineering process
with oxygen plasma. The etching process can be performed with either the ground electrode
oxygen plasma at the ground electrode or the powered electrode.

3.2.3 Results of graphene singular sheet etching
To validate the efficacy of the SSE method, both mono- and bi-layer graphene films were
subjected to oxygen plasma etching for different process durations. The etching time was
increased by a step of 2 seconds. AFM, SEM, and Raman spectroscopy were used to
investigate graphene surfaces after each step. Figure 3.2 shows the evolution of a graphene
duration of 38 seconds. In this case, the graphene sample started with both a sample upon the
singular sheet etching with the ground oxygen plasma for optimum process monolayer region
and a bi-layer region that were identified with Raman spectroscopy ( Figure 3.2 (a)). After a
“singular sheet etching” step, the monolayer region was completely removed, while the
bilayer region was thinned down to a monolayer graphene, as shown in Figure 3.2 (b). By
carefully comparing Figure 3.2 (a) and (b), it can be seen that the border (the upper edge) of
40

the bilayer region did not change obviously, which implies that the edges are not a preferred
direction during the etching process. After the second etching, the bilayer region was
completely etched away (Figure 3.2 (c)), leaving the SiO2/Si substrate. Graphene residues
were found at the boundary between the monolayer and bilayer regions, which were resulted
from thicker graphene fragments and folded graphene edges (> 2 layers) around the boundary.

Figure ‎3.2 SEM images show the evolution of a graphene sample after each “singular sheet
etching” step with ground electrode oxygen plasma: (a) as-produced; (b) after the first
“singular sheet etching” step; (c) after the second “singular sheet etching” step.
To further demonstrate the repeatability and controllability of the SSE method with the
ground electrode plasma, we applied the same recipe to several bilayer graphene samples and
successfully obtained monolayer graphene sheets from the bilayer ones, as verified by AFM
and Raman spectroscopy. Figure 3.3 shows the evolution of a representative bilayer sample
when being subjected to the “singular sheet etching” process. After the first “singular sheet
etching” (Figure 3.3 (b)), while the SEM and AFM images had no distinguishable difference
between the untreated graphene and the treated one, their Raman spectra and AFM profiles
clearly showed the changes caused by plasma etching. Particularly, the thickness of the
bilayer graphene decreased from ~1.259 nm to ~0.810 nm, resulting in an apparent difference
of 0.449 nm. Graphene thickness measurement using an AFM is typically not atomically
precise due to sample contamination and/or an unclean or worn AFM cantilever. The step

41

size 0.449 nm is slightly larger than 0.335 nm (one layer) but obviously much smaller than
0.670 nm (two layers), indicating the removal of one atomic carbon layer. The 2D band of its
Raman spectrum comprised only one Lorenzian peak, which is another indicator of a
monolayer graphene. After the second etching step (Figure 3.3 (c)), the SEM and AFM
images taken from the same regions indicates that the bilayer graphene was completely
removed with only folded thicker portions of the graphene left on the substrate. This was
further confirmed with the Raman spectrum taken from the same location, which contained
no specific peak of graphene.

Figure ‎3.3 Demonstration of producing a monolayer graphene from a bilayer graphene using
the SSE method. (a), (b), and (c) show the SEM image, AFM image (up left panel), 2D band
of its Raman spectrum (up right panel) and its height profile from the AFM image (bottom
panel) before being etched with oxygen plasma (a), after the first “singular sheet etching”
process (b), and after the second “singular sheet etching” process (c), respectively.

42

Figure 3.3 (cont’d)

We have demonstrated that the SSE of graphene can be realized with the oxygen plasma, but
it still remains unclear how the oxygen plasma interacts with graphene. An answer to that
question is essential for further understanding the effects of oxygen plasma on the remaining
graphene sheets after the etching process. Two possible etching mechanisms include isotropic
etching and anisotropic etching. If the single sheet removal of graphenes by oxygen plasma is
isotropic, the etching rate should be uniform everywhere on a graphene sheet regardless of
graphene defects and edges. Initiatively, when a single atomic layer is removed from the top,
a single or a couple of chains of atoms at the edge will be removed. This is because removing
an atom from the top will need sufficient energy to break three bonds, while at the edges,
43

only one or two bonds need to be broken due to the existence of dangling bonds. That means
the etching cannot be an isotropic process. For graphene sheets, there are two possible
preferred directions for anisotropic etching: vertical (top surface) and horizontal (edges).
Experimentally, it is possible to detect a removed single layer using Raman or optical
microscopy, but the loss of a couple of chains of atoms at the edge is beyond the resolution of
these instruments. Consequently, the sizes of the graphene sheet and the defective regions
will not change during the etching in the case of anisotropically vertical etching, but will be
projected to the bottom graphene sheet after the etching process, as illustrated in Figure 3.4
(a). While in the anisotropically horizontal etching mode, the etching rates at the graphene
edges and defective regions are anticipated to be higher than that at other intact regions since
the edges/defects of graphene tend to react faster with chemicals (e.g. oxygen plasma) than
pristine graphene [8]. As schematically displayed in Figure 3.4 (b), the defective regions of
the top graphene sheet will expand, resulting in larger defective areas on the bottom sheet
after the etching process. Additionally, the size of the graphene sheet will shrink due to the
anisotropic etching of oxygen plasma.
Experiments were conducted in order to investigate which one of these two possible
mechanisms will better explain the single sheet removal of graphene. In this study, the
etching rate of graphene was relatively fast and the time interval was too short, making it
difficult to control the etching time accurately on the same graphene sheet. Therefore,
different monolayer graphene samples were etched for various plasma durations and
inspected with SEM to evaluate the time evolution of oxygen plasma etching. Etching
conditions, such as RF power, oxygen flow rate and working pressure, except the process
duration, were kept the same for each sample. Figure 3.4 (c)-(e) show the SEM images of
three monolayer graphene samples before and after being etched by the oxygen plasma at the
ground electrode with the plasma durations ranging from 28 s to 32 s. It can be seen that, first,
44

no noticeable area expansions of defective regions in the monolayer samples were observed
during the etching. Second, no remarkable shrinkage of the monolayer graphene dimensions
due to the possible faster etching rates at the graphene edges was observed during the etching
processes. These results suggest that the single sheet removal of graphenes by oxygen plasma
is most likely anisotropically vertical etching.

Figure ‎3.4 Schematic cartoons show two possible etching mechanisms of ground electrode
oxygen plasma: (a) anisotropic vertical etching and (b) anisotropical horizontal etching.
Illustrations (c)-(e) show the SEM images of monolayer graphene samples before and after
being subjected to oxygen plasma etching for 32 s, 30 s and 28 s, respectively. The
monolayer regions of the graphene samples were marked with arrows. In each sub-figure, the
top and bottom SEM images show the graphene sample before and after being etched with
the corresponding process duration, respectively.

45

With a similar means of investigation, singular layer etching of graphene can also be
achieved with oxygen reactive ion etching at the powered electrode for a shorter etching
duration of 17 seconds. Figure 3.5 provides the demonstration of a monolayer graphene
produced from a bilayer graphene using the oxygen reactive ion etching. A ~0.4 nm thickness
decrease of the graphene was observed after one SSE plasma treatment. The evolution of its
Raman 2D band also demonstrated that a monolayer graphene was produced from the bilayer
graphene.

Figure ‎3.5 AFM images, height profiles, and the 2D bands of its Raman spectrum of a bilayer
graphene (a) before and (b) after the SSE with powered electrode oxygen plasma,
demonstrating the formation of a monolayer graphene from a bilayer graphene.
It is known that the strength of oxygen plasma varies at the powered electrode and the ground
electrode (as shown in Figure 3.1), which is anticipated to result in different levels of
physical damage and defects on the post-etch graphene sheets. Therefore, we investigated the
plasma induced defects on a variety of plasma-treated graphene samples by measuring the
intensity ratios of the D band and G band (ID/IG) in their Raman spectra. A higher ID/IG ratio
46

indicates a larger number of defects. Figure 3.6 (a) and (b) show the Raman spectra taken
from two bilayer graphene samples before and after SSE by the oxygen reactive ion etching
and the oxygen plasma etching, respectively. The Raman spectra show that the sample etched
with the oxygen reactive ion etching at the powered electrode had an ID/IG ratio of ~1.18 (the
bottom panel of Figure 3.6 (a)), while the ID/IG ratio of the graphene etched with the oxygen
plasma at the ground electrode was ~0.94 (the bottom panel of Figure 3.6 (b)). The statistical
analysis of the Raman ID/IG ratios of multiple plasma treated samples was summarized in
Figure 3.6 (c). The average ID/IG ratio of graphene samples after etching with the powered
electrode oxygen plasma was ~1.05, while the average ID/IG ratio after etching with the
oxygen plasma was ~0.86. These results indicate that the oxygen plasma introduced fewer
defects compared to the powered electrode oxygen plasma, which is mainly due to the
relatively low strength of oxygen plasma at the ground electrode.
Furthermore, it is found that the amount of defects can be reduced significantly with a postetch annealing treatment. The annealing process was carried out in an Ar environment at 9001000 °C for 1h. Figure 3.6 (d) shows the Raman spectrum of a monolayer graphene, which
was produced with the SSE method, after the annealing process. Compared with its Raman
spectrum before the annealing process, which was shown in the bottom panel of (a), it can be
seen that the intensity of the D band peak has been greatly suppressed by the annealing
process. The recovery of disordered graphene by high temperature Ar annealing may be
related to the thermally induced reconstruction of graphene lattice and dangling bonds [9] as
well as graphene dehydrogenation [10]. It should be mentioned that electron beam irradiation
during the SEM step is known to have effects on the transformation of the crystalline order
and electronic properties of mono/bilayer graphene films [11]. During our quantitative studies
of defects, the as-etched graphene samples were examined only by the micro Raman
47

spectroscopy in order to eliminate additional defect formation due to the electron beam
irradiation. Therefore, the formation and partial recovery of defects on our graphene
substrates are mainly attributed to the oxygen plasma etching and their successive annealing
treatment.

Figure ‎3.6 Raman spectra of bilayer graphene samples before and after the SSE process with
(a) the powered electrode oxygen plasma and (b) the ground electrode oxygen plasma. (c)
Statistic analysis of Raman ID/IG ratios of graphene samples after being etched with the
powered electrode (solid circles) and the oxygen plasma (open circles). The red and blue
dotted lines show the average values of data from solid and open circles, respectively. (d)
Raman spectrum of a monolayer graphene sheet produced from a bilayer graphene sheet after
the annealing treatment. Its Raman spectrum before annealing is shown in the bottom panel
of (a).
3.2.4 Application of the SSE method
Our singular sheet etching process is completely compatible with microelectronics
manufacturing techniques and shows the prospect of being a promising post-synthesis
48

method for engineering the number of graphene layers and producing graphene thin films
with large sizes and high yield. We have also explored the application of this method in
patterning several specific graphene structures, including an inter-layer graphene structure
that will be discussed in Chapter 6 and a suspended thin graphene structure which will be
discussed briefly below.
2

The remarkable electron mobility of suspended graphene (200000 cm /V s) makes it
candidates for future nano electronic devices. To produce suspended graphenes, several
methods have been reported. For example, Tombros et al. [12] constructed suspended
graphene structures using pillars of lift-off resist (LOR) polymer as supports. Wet etching
technique has also been explored to remove SiO2 underneath graphene patterns [13].
However, in these approaches, critical-point-drying must be used in order to avoid the surface
tension-induced collapse of the suspended graphene structure. Presently, a simple and reliable
method to achieve large area suspened graphenes is still unavailable. Motivated by the limits
of the existing methods, we have applied our SSE method in controlling the number of layers
in the case of suspended graphene.

In this approach, tranches were made on the SiO2 (300 nm)/Si substrates through
photolithography and induced coupled plasma reactive ion etching (ICP-RIE), respectively.
First, SiO2 was completely removed and Si substrate was etched down by around 10 µm.
Next, graphene layer was prepared using michromechanical exfoliation and then deposited
over the tranches and the holes. After that, graphenes were located by optical microscope and
Raman spectroscope. Then graphene layer was etched layer-by-layer using oxygen plasma,
showing that the SSE method provides a possible way to produce graphenes directly from
graphite flakes. Figure 3.7 shows a suspended graphene sheet with a thickness of ~2 nm
49

produced through layer by layer etching of a graphite flake of an original thickness of ~15
nm.

Figure ‎3.7 An SEM image of a 2-nm-thick suspended graphene film produced from graphite
flake of ~ 15 nm using layer-by-layer oxygen plasma etching. Inset is the Raman spectra.

3.3

Band Gap Engineering of Graphene

3.3.1 Background
Pristine graphene is known to be semi-metallic with a zero band gap at room temperature, as
discussed in Chapter 1. Therefore, an engineering approach must be developed to open the
band gap in order to achieve desired semiconductor properties of grapheme for electronics
fabrication. Recent studies have shown that the graphene band gap can be tuned by either
modifying pristine graphene with dopants [14] or constructing specific structures such as
nanoribbons [15-17] or nanomeshes [18-25]. However, field effect transistors (FETs) based
on a single graphene nanoribbon often suffer from low driving current and conductivity [15].
Compared to nanoribbon, graphene nanomesh (GNM) is a more practical structure to
effectively open the graphene band gap while allowing a favorably high driving current of the
graphene FETs. Several fabrication techniques have been implemented to make the GNMs.
For example, Bai et al. created sub-100-nm nanoholes on graphene using a block copolymer
lithography method, in which a poly(styrene-block-methyl methacrylate) (P(S-b-MMA))
block copolymer thin film with cylindrical domains normal to the graphene surface was used
50

as the template of reactive ion etching (RIE) [15]. Self-assembled monolayers of colloidal
microspheres were also exploited as the template to fabricate the GNMs [19, 26]. However,
these approaches involve RIE or hydrofluoric acid (HF) to etch the SiO2 mask layer and also
the Si/SiO2 substrate to make the GNMs, which might induce a leakage current into the back
gate or even result in a short circuit between electrodes and the back gate of the GNM
transistor. Alternative methods using chemical vapor deposition (CVD) of nanomesh
graphene on porous MgO or Cu foil templates have been developed by several groups [2022]. These methods, however, still require specific templates and expensive instrumentation
(CVD), thus increasing the complexity and the cost of device fabrication. Recently, Zeng et
al. reported a method for fabricating the GNM on a large area, simply using an anodic
aluminum oxide (AAO) membrane as an etching mask [23]. However, the hole size and neck
width of the GNM were determined by the prefabricated AAO mask and cannot be easily
adjusted without changing the mask. The transistor fabricated with this method also showed a
relatively low ON/OFF ratio. Other methods have also been explored for making the GNMs,
such as local catalytic hydrogenation of carbon by Cu nanoparticles [24] and local photo
degradation of graphene oxide at the tip of ZnO nanorods[25]. Nevertheless, precise control
of the hole size and neck width of the GNMs in the current approaches still remains a
challenge, increasing the demand for a new technology.
3.3.2 E-beam patterning method to engineer graphene band gap
To overcome this challenge, we propose a template-less technique based on direct electron
beam (e-beam) lithography (EBL), which allows precisely controlled, high-throughput, lowcost fabrication of periodic nanomeshes on a large graphene sheet. Typically EBL processes
use nanometer pattern generation system (NPGS) software to optimize the writing parameters
through a run file editor. Two kinds of files are usually prepared: an alignment file and a
51

writing file. Our method is different from the conventional EBL process by utilizing only the
alignment mode to directly write nanoholes on EBL resist. Such nanoholes can then be
transferred onto the graphene sheet using oxygen plasma etching through the EBL resist
mask. Figure 3.8 shows the conceptual schematic of our approach for making nanopatterned
graphene FETs. This approach takes advantages of automation and high precision of EBL
instrumentation and thus enables batch production of graphene nanomeshes or nanoribbons
with high repeatability. The hole size and neck width of the nanomeshes can be controlled
precisely by manipulating EBL processing parameters such as e-beam current, scanning area,
e-beam resist thickness, and number of scanning cycles.

Figure ‎3.8 Conceptual diagram of the proposed template-less technique for fabrication of
graphene nanopatterned FET transistors.

Monolayer and fewlayer (~ 4-9 layers) graphene sheets were used in this study. The graphene
samples were prepared with a micromechanical exfoliation method and transferred onto SiO2
(300 nm)/Si substrates. Ti/Au markers had been previously fabricated on the surface of the

52

SiO2 substrate via thermal evaporation, photolithography patterning, and metal lift-off
processes. These markers allowed one to easily identify the locations of thin graphene sheets
on the substrate (as discussed in Chapter 2). After locating the graphene sheets on the
substrate via optical microscopy, micro Raman spectroscopy was used to analytically
determine the number of layers of the graphene sheets, based on the intensity, shape and
position of G and second-order 2D bands of their Raman spectra. The chips with the
graphene sheets were then annealed at ~ 400 °C in an Ar/H2 environment for 1 hr to remove
the tape residues. The hole engineering of the graphene sheets was performed with an EBL
system (JEOL 840A, JEOL Ltd.) based on the aforementioned unconventional e-beam
writing method.
The hole size and the neck width are the two key factors to control the dimensions of the
GNMs, which determine the size of the band gap and thus the electrical properties of the
nanopatterned graphene semiconductors [15, 16]. Therefore, we investigated several
important processing parameters in order to fully control the hole size and the neck width in
the poly(methyl methacrylate) (PMMA) resist layer. It is well known that the acceleration
voltage affects the electron beam width such that a higher acceleration voltage leads to a
thinner e-beam, thus smaller holes in the EBL process. Specifically in our experiment, the
accelaration voltage remained a constant at 35 kV, which was set based on the previous
calibration of the EBL instrutment. Other EBL parameters included the e-beam current, the
scanning area, the resist thickness, and the number of scanning cycles. The effect of each of
these parameters was studied individually while keeping the others constant. Under each
testing condition, 20 samples were measured and their average minimum value of the hole
size or the neck width was taken. After the PMMA layer was fabricated, transfer patterning
of the nanomeshes or the multi-ribbons into the graphene sheets was conducted in a plasma
53

etching system (PX-250, March Instruments) using oxygen plasma. The plasma etching
parameters including radio-frequency (RF) power and etching duration were also studied to
ensure high yield and repeatability of the graphene nanopatterns.
The first parameter was e-beam current. In this case, the scanning area, the resist thickness,
2

and the number of scanning cycles of the SEM-EBL were tuned at 50×50 µm , 900 nm, and
one cycle, respectively. As shown in , a higher e-beam current resulted in a bigger hole size
and therefore a smaller neck width. The neck width decreased from 300 nm to around 90 nm
as the e-beam current increased from 30 pA to 100 pA. The minimum diameter of the
through holes was around 100 nm obtained at the e-beam current of 30 pA.

Figure ‎3.9 Periodic nanomesh patterned by e-beam with a scanning area of 50×50 µm2, an ebeam resist thickness of 900 nm, and one scanning cycle. (a) and (b) SEM images show that
the dimension of the holes increased as the e-beam current increased. (c) Histogram of the
average sizes under different beam currents.
Small e-beam currents such as 5 pA and 20 pA resulted in incomplete exposure of the e-beam
resist. In our experiments, PMMA-C2 and PMMA-C4 were used to obtain different
thicknesses ranging from 450 nm to 900 nm. Technically, exposure of a thicker resist layer
requires a greater amount of the e-beam current, leading to a larger hole size and a smaller

54

neck width. The experimental results suggested that a minimum e-beam current of 30 pA was
required to fully expose 900 nm PMMA.
The descrease of the scanning area will lead to the descrease of the minimum neck width.
Figure 3.10 (c) shows that a significant change of the neck width can be achieved by
manipulating the scanning area. However, there is an area limit since a very small scanning
area will cause overlapping holes (Figure 3.10 (d)). In this study, the e-beam current, the
resist thickness, and the number of scanning cycles were 35 pA, 450 nm, and one cycle,
respectively.

Figure ‎3.10 Nanoholes patterned on different scanning areas: (a) 2500 µm2 with a neck with
of ~150 nm, (b) 900 µm2 with a neck width of ~ 90 nm, (c) 900 µm2 with a neck width of ~
50 nm, and (d) 100 µm2 with overlapped holes. (e) Histogram shows the effect of the
scanning area on the average minimum neck width of the GNMs.
When the e-beam scanning is repeated on a particular area more than once, the e-beam
instability can cause the drift of the hole size and thus the neck width by a few nanometers
from each scan. However, this phenomenon can be exploited for our purpose to decrease the
neck width in the GNM structures. Figure 3.11 shows the statistical study of the average neck
55

width in relation to the number of the scanning cycles, with the beam current of 35 pA, the
2

scanning area of 2500 µm , and the resist thickness of 900 nm. By increasing the number of
the scanning cycles to three cycles, the minimum neck widths of ~ 25 nm and ~ 15 nm were
achieved in PMMA layers with the thicknesses of 900 nm and 450 nm, respectively.
As the number of the scanning cycles continues to increase (> 4 cycles), overlapping holes
were generated, resulting in long channels across the PMMA layer. The unexposed resist
protected thin strips of the graphene sheet that eventually formed parallel nanoribbons after
oxygen plasma etching. This process has been applied to successful fabrication of nanoscale
graphene multi-ribbon structures with a minimum width ranging from 15 nm to 45 nm.
Figure ‎3.11. Figure ‎3.12 shows the concept of our multi-scan technique for making
nanoribbon patterns and a representative multi-ribbon array fabricated on a monolayer
graphene sheet.

Figure ‎3.11 Average neck widths of 75 nm and 25 nm were obtained by two scans (a) and
three scans (b and c), respectively. (d) Histogram shows the relation between the number of
the scanning cycles and the average neck width of the GNMs.

56

Figure 3.11 (cont’d)

Figure ‎3.12 (a) Concept of the muliribbon fabrication teqnique. (b) A representative graphene
multi-ribbon structure.
During the transfer patterning step, the isotropic nature of the plasma etching can induce
undercut into the underlying graphene sheet. Therefore, the power and duration of the oxygen
plasma etching must be fine tuned to ensure the integrity of the graphene nanopatterns,
particularly the GNMs with 15-40 nm neck widths, and the repeatibility of the fabrication. In
this study, a PMMA nanomesh mask was fabricated using the process described above, with
the resist thickness of 450 nm, the minimum neck width of 15 nm, and the hole size of ~ 150

57

nm. The RF power of the plasma etcher was tuned between 70 W and 150 W and the etching
duration was varied from 17 s to 120 s. Table 3.1 summarizes the yields of the nanomeshes
etched into the single- and few-layer graphene sheets with different powers and durations of
the oxygen plasma etching. For the single-layer graphene samples, the best yield was
obtained atthe 70 W RF power for 17 s. The optimized recipe for etching the few-layer
graphene was at the 150 W RF power for 40 s.

Table 3.1 The effect of oxygen plasma etching on graphene nanomesh

Our fabrication process relies on the EBL alignment process that is independent of sample
size and shape. Therefore, it is scalable to make nanoholes on a large area of the resist for the
formation of nanomeshes or multi-ribbons on large-scale graphene sheets. As a
demonstration, a block “S” logo of Michigan State University (MSU) was patterned on a
450-nm-thick MMA/PMMA layer, with the hole size of 70 nm and the average neck width of
150 nm (Figure 3.13 ).

58

Figure ‎3.13 SEM image of an MSU logo with nanoholes on a large area.

3.3.3 Fabrication of p-type graphene nanomesh FETs
Graphene nanomesh FETs were constructed following the fabrication flow described in
Figure 3.14 and Figure 3.15. The mechanically exfoliated pristine graphene sheets (monoand few-layer) were first transferred onto the SiO2/Si substrate. The graphene sheets were
engineered to regular shapes by a fine-tuned oxygen plasma etching method (150 W, 10
sccm, 315 mTorr, for 1 minute). Ti (5 nm)/Au (70 nm) electrodes were built on the graphene
as source and drain contacts, through thermal evaporation, e-beam lithography, and metal liftoff processes. 140-nm-thick PMMA-C2 served as a protective layer for the oxygen plasma
and a mask for metal lift-off. After dissolving the PMMA resist with acetone and isopropanol
alcohol, the devices were rinsed with de-ionized (DI) water and dried with nitrogen. Metal
was then deposited on the backside of the Si substrate as the back gate contact of the
transistor. To convert the pristine graphene sheets into semiconductors, nanoholes were
written on 450-nm-thick PMMA by e-beam scanning the area of interest, using the above
mentioned unconventional EBL method. The EBL parameters were tuned as follows: 35 kV
acceleration voltage, 30 pA beam current, and one scanning cycle. The developed nanoholes

59

were transferred into the mono- or few-layer graphene sheets using oxygen plasma etching
with 150 W RF power, 10 sccm oxygen flow rate, 315 Torr chamber pressure, and 40 s
duration. Based on the experimental data, this recipe can be used to etch 1 to ~ 10 layers of
graphene sheets.

Figure ‎3.14 Process flow for making GNM transistors.
Figure 3.15 shows the SEM and the optical microscope images of two fabricated GNM FETs.
The same etching parameters and e-beam resist thickness were used for making nanoholes on
the mono- and few-layer graphene sheets, resulting in ~ 100 nm periodic holes. Initially, the
electrodes were deposited on as-patterned GNMs. Some nonfunctioning devices fabricated in
this way did not appear damaged under close inspection with SEM. We suspected the damage
to these devices occurred in the GNM region covered by the electrodes. Depositing electrodes
before nanoholes patterning corrected this problem and greatly improved the fabrication
yield. Additionally, this method improved the performance of the GNM FETs by increasing
the contact area and thus the driving current.

60

Figure ‎3.15 Representative GNR transistor fabricated by on (a) monolayer graphene sheet (b)
few-layer graphene sheet.

The fabricated GNM transistors were characterized by measuring the drain current (Ids)
versus drain voltage (Vds) at different gate voltages (Vg), using a semiconductor parameter
analyzer (HP4145B, Hewlett Packard). All the devices were annealed in an H2/Ar
environment to remove resist residues and fabrication contaminants prior to the measurement.
shows the Vds-Ids characteristics of the monolayer and few-layer GNM FETs. As shown in
Figure 3.16 (b) and (d), the drain current decreases with the increase of the gate voltage. As
discussed previously in Chapter 1, the Dirac point of graphene is defined as a certain gate
potential where the valance and conduction bands meet. After the Dirac point, the
conductivity type of graphene changes from p-type to n-type. Thus one can realize the
ambipolar properties of graphene, which is different from conventional semiconductors, such
as silicon, where only one ON state and one OFF state exist. In addition, the Dirac points of
the as-fabricated GNM transistors were located in the positive region of the Vg axis,
indicating that all the devices exhibited a p-type semiconductor property. From the physics
point of view, this can be explained by the downward shift of the Fermi level into the valance
band. The p-type properties of GNMs can be attributed to the high sensitivity and reactivity
of abundant graphene edges exposed by introducing the nanopatterns into the pristine

61

graphene. The perimeters of these nanoholes will readily adsorb oxygen from the
environment, particularly during oxygen plasma etching [27, 28]. Impurities generated during
the fabrication processes may also have contributed to the p-type doping.

Figure ‎3.16 Drain current (Ids) versus drain–source voltage (Vds) at different gate voltages
(Vg) for a monolayer GNM transistor (b) Ids versus Vg at different Vds voltages (c) Drain
current (Ids) versus drain–source voltage (Vds) at different gate voltages (Vg) for a monolayer

62

Figure 3.16 (cont’d)
GNM transistor (d) resistance variation as the function of the gate voltage of a few-layer
graphene device. All graphene nanomeshes had p-type semiconductor behavior.

The ON-OFF ratio of the monolayer GNM transistor was around 15. However, in order to
achieve higher ON-OFF ratio of GNM transistors, the neck width must be very few
nanometers, which is not easy to achieve by our fabrication process. From the experimental
data, we also calculated the conductance of the monolayer graphene nanomesh, which was
around 3.3 mS at the gate voltage (Vg) of 60 V, whereas the ON-OFF ratio and the
conductance of the few-layer graphene FETs were ~ 2.8 and ~ 0.13 mS, respectively.
The electron mobility (µe) was calculated using the equation: µe=(t/ε)×(dσ/dVg), where
t=300 nm is the thickness of the SiO2, ε=3.9×ε0=3.45×10

-11

F/m is the permittivity of the

SiO2, and σ=L/RW is the conductivity. The maximum electron mobility achieved was 3332
2

2

cm /V s for monolayer device, while it is 1343 cm /V s for the few-layer one.
5

The performance of a monolayer GNM device under the vacuum condition of 5×10 Torr
was investigated inside an environmental SEM (Carl Zeiss variable pressure SEM EVO
LS25), in order to study the effect of environmental contaminations on the semiconductor
properties of the GNMs. Measured Ids-Vds curves showed no significant deviation from those
obtained at atmosphere (Figure 3.17), indicating device performance was resistant to
contaminations or doping associated with normal atmospheric condition.

63

Torr

Figure ‎3.17 Ids-Vds characteristics of a monolayer GNM transistor tested under vacuum
condition and at atmospheric pressure, showing no significant difference.
We also fabricated graphene p-type transistors based on nanoriboon configurations that were
patterned using the aforementioned EBL method. The devices were characterized by
measuring the drain current Ids versus drain voltage Vds at different gate voltages Vg
(Figure 3.18) in addition to the current transport characteristic. The ON-OFF ratio of the
multi-nanoribbon FET was ~ 1.5 and the conductance was ~ 0.04 mS at the gate voltage of 60
V. Fermi level positions relative to the Dirac point (ΔEF = │EF-ED│) were calculated for
nanomesh and graphene multi-ribbon to be 203 meV and 904 meV respectively.

64

Figure ‎3.18 (a) Current voltage transfer characteristic of a graphene multi-ribbon transistor,
and (b) its current transport characteristic.

3.4

Conclusion

In this chapter, we presented a method for engineering the number of graphene layers with
fine-tuned oxygen plasma parameters. By this method, only one layer of graphene sheet can
be etched at a time. Although this technique produces surface defects, it can be suppressed by
annealing process. Additionally, we demonstrated a template-less nanofabrication technique
that can be used for batch production of the graphene nanomesh and multi-ribbon structures.
The EBL processing parameters have been investigated to precisely control the hole size and
the neck width of the nanomeshes. The minimum hole size and neck width achieved was
around 100 nm and 15 nm, respectively. Moreover, the same fabrication method has been
65

successfully applied to the fabrication of the graphene nanomesh and multi-nanoribbon FETs.
Although we have realized p-type graphene transistors using the nanopatterning approach,
fabrication of a digital nanocircuit will require both p-type and n-type graphene
semiconductors. In the next chapter, a doping method to achieve n-type graphene will be
introduced.

66

REFERENCES

67

REFERENCES

[1]

Y. Zhang, T.-T. Tang, C. Girit, Z. Hao, M. C. Martin, A. Zettl, M. F. Crommie, Y. R.
Shen, and F. Wang, "Direct observation of a widely tunable bandgap in bilayer
graphene," Nature, vol. 459, pp. 820-823, 2009.

[2]

G. Zhao, D. Shao, C. Chen, and X. Wang, "Synthesis of few-layered graphene by
H2O2 plasma etching of graphite," Applied Physics Letters, vol. 98, p. 183114, 2011.

[3]

H. Wang, Y. Wu, Z. Ni, and Z. Shen, "Electronic transport and layer engineering in
multilayer graphene structures," Applied Physics Letters, vol. 92, pp. 053504-0535043, 2008.

[4]

C. Cong, T. Yu, H. Wang, K. Zheng, P. Gao, X. Chen, and Q. Zhang, "Self‐limited
oxidation: A route to form graphene layers from graphite by one‐step heating," Small,
vol. 6, pp. 2837-2841, 2010.

[5]

K. Hazra, J. Rafiee, M. Rafiee, A. Mathur, S. Roy, J. McLauhglin, N. Koratkar, and
D. Misra, "Thinning of multilayer graphene to monolayer graphene in a plasma
environment," Nanotechnology, vol. 22, p. 025704, 2011.

[6]

A. Dimiev, D. V. Kosynkin, A. Sinitskii, A. Slesarev, Z. Sun, and J. M. Tour, "Layerby-layer removal of graphene for device patterning," Science, vol. 331, pp. 11681172, 2011.

[7]

W. S. Lim, Y. Y. Kim, H. Kim, S. Jang, N. Kwon, B. J. Park, J.-H. Ahn, I. Chung, B.
H. Hong, and G. Y. Yeom, "Atomic layer etching of graphene for full graphene
device fabrication," Carbon, vol. 50, pp. 429-435, 2012.

[8]

M. Batzill, "The surface science of graphene: Metal interfaces, CVD synthesis,
nanoribbons, chemical modifications, and defects," Surface Science Reports, vol. 67,
pp. 83-115, Mar 1 2012.

[9]

J. Campos-Delgado, Y. A. Kim, T. Hayashi, A. Morelos-Gomez, M. Hofmann, H.
Muramatsu, M. Endo, H. Terrones, R. D. Shull, M. S. Dresselhaus, and M. Terrones,
"Thermal stability studies of CVD-grown graphene nanoribbons: Defect annealing
and loop formation," Chemical Physics Letters, vol. 469, pp. 177-182, Feb 3 2009.

[10]

D. C. Elias, R. R. Nair, T. M. G. Mohiuddin, S. V. Morozov, P. Blake, M. P. Halsall,
A. C. Ferrari, D. W. Boukhvalov, M. I. Katsnelson, A. K. Geim, and K. S. Novoselov,
"Control of graphene's properties by reversible hydrogenation: Evidence for
graphane," Science, vol. 323, pp. 610-613, Jan 30 2009.

68

[11]

D. Teweldebrhan and A. A. Balandin, "Modification of graphene properties due to
electron-beam irradiation," Applied Physics Letters, vol. 94, Jan 5 2009.

[12]

N. Tombros, A. Veligura, J. Junesch, M. H. Guimarães, I. J. Vera-Marun, H. T.
Jonkman, and B. J. van Wees, "Quantized conductance of a suspended graphene
nanoconstriction," Nature Physics, vol. 7, pp. 697-700, 2011.

[13]

K. I. Bolotin, K. Sikes, Z. Jiang, M. Klima, G. Fudenberg, J. Hone, P. Kim, and H.
Stormer, "Ultrahigh electron mobility in suspended graphene," Solid State
Communications, vol. 146, pp. 351-355, 2008.

[14]

F. Schedin, A. Geim, S. Morozov, E. Hill, P. Blake, M. Katsnelson, and K.
Novoselov, "Detection of individual gas molecules adsorbed on graphene," Nature
Materials, vol. 6, pp. 652-655, 2007.

[15]

L. Jiao, L. Zhang, X. Wang, G. Diankov, and H. Dai, "Narrow graphene nanoribbons
from carbon nanotubes," Nature, vol. 458, pp. 877-880, 2009.

[16]

J. Bai, X. Duan, and Y. Huang, "Rational fabrication of graphene nanoribbons using a
nanowire etch mask," Nano Letters, vol. 9, pp. 2083-2087, 2009.

[17]

D. V. Kosynkin, A. L. Higginbotham, A. Sinitskii, J. R. Lomeda, A. Dimiev, B. K.
Price, and J. M. Tour, "Longitudinal unzipping of carbon nanotubes to form graphene
nanoribbons," Nature, vol. 458, pp. 872-876, 2009.

[18]

J. Bai, X. Zhong, S. Jiang, Y. Huang, and X. Duan, "Graphene nanomesh," Nature
Nanotechnology, vol. 5, pp. 190-194, 2010.

[19]

A. Sinitskii and J. M. Tour, "Patterning graphene through the self-assembled
templates: Toward periodic two-dimensional graphene nanostructures with
semiconductor properties," Journal of the American Chemical Society, vol. 132, pp.
14730-14732, 2010.

[20]

G. Ning, C. Xu, L. Hao, O. Kazakova, Z. Fan, H. Wang, K. Wang, J. Gao, W. Qian,
and F. Wei, "Ferromagnetism in nanomesh graphene," Carbon,vol. 51, pp. 390-396,
2012.

[21]

G. Ning, Z. Fan, G. Wang, J. Gao, W. Qian, and F. Wei, "Gram-scale synthesis of
nanomesh graphene with high surface area and its application in supercapacitor
electrodes," Chemical Communications, vol. 47, pp. 5976-5978, 2011.

[22]

M. Wang, L. Fu, L. Gan, C. Zhang, M. Rümmeli, A. Bachmatiuk, K. Huang, Y. Fang,
and Z. Liu, "CVD growth of large area smooth-edged graphene nanomesh by
nanosphere lithography," Scientific Reports, vol. 3, dio:10.1038, 2013.

69

[23]

Z. Zeng, X. Huang, Z. Yin, H. Li, Y. Chen, H. Li, Q. Zhang, J. Ma, F. Boey, and H.
Zhang, "Fabrication of graphene nanomesh by using an anodic aluminum oxide
membrane as a template," Advanced Materials, vol. 24, pp. 4138-4142, 2012.

[24]

J. Liu, H. Cai, X. Yu, K. Zhang, X. Li, J. Li, N. Pan, Q. Shi, Y. Luo, and X. Wang,
"Fabrication of graphene nanomesh andimproved chemical enhancement for Raman
spectroscopy," The Journal of Physical Chemistry C, vol. 116, pp. 15741-15746,
2012.

[25]

O. Akhavan, "Graphene nanomesh by ZnO nanorod photocatalysts," ACS Nano, vol.
4, pp. 4174-4180, 2010.

[26]

R. K. Paul, S. Badhulika, N. M. Saucedo, and A. Mulchandani, "Graphene nanomesh
as highly sensitive chemiresistor gas sensor," Analytical Chemistry, vol. 84, pp. 81718178, 2012.

[27]

T. Li and S.-P. Lu, "Quantum conductance of graphene nanoribbons with edge
defects," Physical Review B, vol. 77, p. 085408, 2008.

[28]

F. Sols, F. Guinea, and A. Neto, "Coulomb blockade in graphene nanoribbons,"
Phisical Review Letters, vol. 99, 166803, 2007.

70

4 GRAPHENE N-TYPE DOPING

4.1

Introduction

Chemical doping has been proven to be an effective solution to open the band gap and
modulate the electrical properties of graphene. In this case, the absorption of chemical
molecules and/or charge exchange on graphene surface can cause the shift of the initial Fermi
level position upward (as shown in n-type graphene) or downward (as shown in p-type
graphene) with respect to the charge neutrality point (the Dirac point). Compared to p-type
doping, graphene n-type doping with long-term chemical stability is more challenging due to
the rapid degradation of n-type properties upon exposure to air. The degradation of the n-type
properties is mainly attributed to the absorption of oxygen molecules, which are known as
effective hole-dopant for graphene. Electron transfer from the n-type graphene to the
absorbed oxygen molecules will result in the reduction of the n-type doping concentration,
which can be well quantified by the electronic band structure and the downward shift in the
Fermi level of the doped graphene.
To date, a number of doping approaches have been developed to achieve n-type graphene: for
example, substitution doping with nitrogen dopants [1-4], covalent functionalization [5, 6],
and radiation-induced doping [7, 8]. However, these existing methods usually increase
defects in carbon lattices, deteriorate the carrier mobility of as-doped graphene, and are
limited in chemical stability. To minimize defects, electrostatic field tuning has been
proposed to modulate the electronic properties of graphene devices [9-11] but this method
does not open the band gap of monolayer graphene and thus has limited applications.
Alternative methods, such as chemical modification of graphene surface with polymer [12,
13] and metallic thin films [14, 15], have also been reported. These n-type graphene devices
still have limited thermal and chemical stability, and are vulnerable to the influence of the
71

surrounding environment. Recently, Ho et al. demonstrated the self-encapsulated doping of
n-type graphene with high carrier mobility and extended air-stability using titanium suboxide
thin films. However, the fabrication of such devices requires sol-gel and hydrolysis
processes, which are incompatible with conventional microfabrication technology [16]. To
mitigate these drawbacks, this study proposed a new technique for n-type graphene doping
with minimum defects and enhanced chemical stability, using SU-8 photoresist as a doping
and encapsulating material. The n-type graphene doped with uncrosslinked SU-8 can be
reversed to the p-type graphene by removing SU-8 with organic solvents. The fabrication of
the n-type graphene transistors is fully compatible with microfabrication technology, which
provides possibilities for mass production of large-scale graphene-based electronics.
4.2

Method

Pristine graphene (PG) samples were mechanically exfoliated from highly oriented pyrolytic
graphite (HOPG) flakes and then transferred onto SiO2 (300 nm)/Si substrates.. The graphene
samples were then annealed at ~ 400 °C in an Ar/H2 environment for 1 hour to remove tape
residues and contaminations.
Figure 4.1 (a) shows the schematic representation of a back-gated graphene field-effect
transistor (FET) with SU-8 as an n-type doping and encapsulating material. The transistors
were constructed prior to the SU-8 doping, using the following fabrication method. PG
samples were patterned into regular shapes via oxygen plasma etching with 150 W RF power,
10 sccm oxygen flow rate, 315 mTorr chamber pressure, and 1 min duration. Ti (5 nm)/Au
(70 nm) electrodes were fabricated on top of the graphene as source and drain contacts, via
thermal evaporation, electron beam lithography (EBL), and metal lift-off processes. A 140nm polymethyl methacrylate (PMMA) layer served as a protective layer for the oxygen
plasma and a mask for the lift-off. The PMMA resist was dissolved with acetone, rinsed with
isopropanol alcohol and de-ionized (DI) water, and then dried with nitrogen. Metal was
72

deposited on the backside of the highly doped Si substrate as the back gate of the transistor.
After metallization, graphene nanomeshes (GNMs) with a hole diameter of ~ 100 nm and a
neck width of ~ 300 nm were selectively patterned on the PG at the location of the transistor
channel, using an unconventional EBL method [18]. GNM transistors are expected to have
relatively higher on/off current ratios compared to GNR transistors and allow for favorably
high driving current of the channel. Furthermore, these nanomeshes can enhance the p-type
doping level of graphene due to the high sensitivity and reactivity of abundant edge defects,
facilitating the reversibility study of the SU-8 doping effect.

(b)

SU-8 epoxy resin

Triaryliumsulfonium salt

GBL solvent

(a)

(c)

Ti/Au

SU-8 coated
n-type GNM
Ti/Au

5 µm

Figure ‎4.1 (a) Schematic of an n-type graphene transistor with SU-8 as the n-type dopant and
self-encapsulation. (b) Three basic chemical compounds of the SU-8 resist and the chemical
reaction for generating the photo-acid upon UV exposure. (c) A representative GNM
transistor doped and encapsulated with SU-8.
The n-type doping of PG or p-type GNMs was performed by spinning 4-µm-thick SU-8 2
negative photoresist (NANO™, MicroChem) on the graphene surface. The SU-8 resist
contains three basic components (Figure 4.1 (b)): an EPON SU-8 epoxy resin, a gammabutyrolactone (GBL) solvent, and a photo-acid generator taken from the family of the
triarylium-sulfonium salts [19]. Following the spin coating, the resist was pre-baked at 95 °C
73

for various durations to evaporate the GBL solvent and solidify the resist. Optionally, some
SU-8-coated samples were exposed to ultraviolet (UV) radiation for 1 min and then postbaked at 105 °C. The UV exposure converts the photo-acid generator into the photo-acid,
which readily reacts with polymer chains to enable the crosslinking of the SU-8 epoxy during
the post-baking step [20]. During this n-type doping procedure, SU-8 is expected to offer
several functions. First, it is hypothesized that the photo-acid generator in the SU-8
compound contains free radicals [21] as efficient electron donors for graphene. Second, SU-8
has good dielectric properties (1 GHz dielectric constant of 4.1 and dielectric strength of 112
V/µm), which make it an effective dielectric medium. Owing to its relatively low water vapor
and gas permeability, SU-8 is also an excellent encapsulating material to protect n-type
graphene and transistors from environmental interference. A fabricated prototype of the SU-8
doped and encapsulated GNM transistor is given in (c).
4.3

SU-8 Doping Results and Discussions

4.3.1 SU-8 n-type doping effect
The n-type effect of the SU-8 doped graphene was studied using Raman spectra taken from
graphene samples with different treatments (). For the SU-8 doped samples, the characteristic
peaks of the SU-8 resist were subtracted from the Raman spectra. The G band of the bilayer
-1

PG was located at ~ 1581 cm , while the G band of the pristine monolayer PG shifted
upward to ~ 1592 cm

-1

due to the decrease in layer number [22]. From the Raman spectra of

all the SU-8-coated samples, the G peak always stiffened and the intensity of the 2D peak
decreased with respect to that of the G peak, which agrees with those observed in typical
electron doping of graphene [13, 16]. The n-type doping effect is also evident in the slight
upshift of the G peak frequency [13, 23, 24]. Referring to the electrostatic gating of graphene
on a Si substrate [13], the change in the 2D to G intensity ratio (I2D/IG) has a strong
74

dependence on the doping level, and a smaller I2D/IG ratio indicates a higher concentration of
electron doping. In this study, after the SU-8 doping, the I2D/IG ratios were reduced from 1.7
(PG) to ~1 (GNM coated with uncrosslinked SU-8), ~0.9 (graphene coated with crosslinked
SU-8), and ~0.6 (graphene coated with uncrosslinked SU-8), as the indication of increasing
doping levels. The doping concentration of the crosslinked SU-8 was lower than that of the
uncrosslinked SU-8, mainly due to the reduced concentration of the photo-acid generator
caused by the UV exposure and post-baking treatments. The rise of the D peak intensity
indicates the substantial distortion of local carbon lattices, as shown in the Raman spectrum
of the GNM (Figure 4.2). The changes in the D peak intensities of all the SU-8 doped
samples were almost negligible compared with their Raman spectra before the doping
treatment, suggesting that the SU-8 doping occurs mainly through surface change transfer
doping at the SU-8/graphene interface [16]. Therefore, our doping method will not induce
permanent damage in the local carbon lattice of graphene, unlike the covalent
functionalization or the substitution doping of graphene in which the increased lattice
disorder results in enhanced D band intensity. In addition, the Raman spectrum of the SU-8coated graphene recovered to a shape similar to that of the PG (Figure 4.2 (a)) after removing
the uncrosslinked SU-8, demonstrating the reversibility of our SU-8 doping process. It is of
note that the crosslinked SU-8 epoxy cannot be attacked by general organic and inorganic
chemicals or plasma etching and thus provides an excellent encapsulation for the n-type
doped graphene.

75

2D

G

D

2D

G

Figure ‎4.2 (a) Raman spectra of a representative bilayer graphene sample with different
treatments. (b) Raman spectra of a representative monolayer graphene sample with different
treatments. The spectra from the bottom to the top correspond to the monolayer PG without
nanomesh, undoped monolayer GNM, and SU-8 doped monolayer GNM, respectively.

4.3.2 SU-8 doped n-type graphene FETs
After studying the n-type doping effect of SU-8, we explored the feasibility of using SU-8
doped GNM graphene to fabricate FETs and verified the n-type transport properties of the
SU-8 doped and encapsulated GNM FETs. In this case, a ~ 4-µm-thick SU-8 layer was
coated and soft baked at 95 °C for 10 min, followed by UV exposure using an UV curing
lamp with ~ 365 nm wavelength for 1 minute. The devices were tested using a semiconductor
parameter analyzer (HP4145B, Hewlett Packard). Figure 4.3 shows the transport properties of
76

a GNM FET, fabricated using the above method. It is observed that the SU-8 doping shifted
the Dirac point from > 20 V to -17 V, representing typical n-type transport behavior of the
graphene. The n-type doping effect is also evident in the gate-dependent characteristics of the
transistor drain current (Figure 4.3 (b)). As the result of the SU-8 removal, the Dirac point
returned to a positive gate voltage, suggesting the recovery of the p-type semiconducting
behavior of the graphene.

Figure ‎4.3 A representative GNM transistorwith the Si back gate, 300 nmSiO2 dielectric,
Au/Ti source and drain contacts. (a) Gate-dependent conductivity (σ) of graphene before the
SU-8 doping, after the doping, and after removing SU-8. (b) Current-voltage (Vds-Ids) curve
of the SU-8 doped device, measured at various gate voltages (Vg) ranging from -20 to +20 V.

The doping level, and hence the Fermi energy (EF), of the graphene, can be tuned by
controlling the soft-baking time and/or the thickness of the SU-8 coating, as shown in .
Longer soft-baking shifted the Dirac point of the n-type doped graphene more negatively,
corresponding to a higher doping level. This shifting is mainly caused by the increased
concentration of the doping agent (the photo-acid generator) in the SU-8 compound, since
more solvent was evaporated during the extended soft-baking treatment. Similarly, thick SU8 coating contains more free radicals and thus enables a higher doping level compared to thin
77

SU-8 coating, which can be demonstrated by the substantial shift of the Dirac point from ~ 50
V to ~ -75 V ( (b)).The electron areal concentrations (ne) were estimated by ne = η(Vg-Vd)
10

with η=Cox/e=(ε0εr)/(de)=7.2×10

-2 -1

cm V , where Cox is the capacitance per unit area

calculated at zero gate bias, d is the thickness of the back-gate oxide (300 nm), ε0 is the
permittivity of free space, εr is the relative permittivity of SiO2, Vg is the back-gate voltage,
Vd is the gate voltage at the Dirac point, and e is the elementary charge. Given the electron
areal concentration, the Fermi level position relative to the Dirac point (ΔEF = │EF-ED│) can
be calculated using the equation: 𝐸 = √𝜋𝑛 ( 𝜐 ) , where υF is the Fermi velocity of
8

electrons (≈ 10 cm/s), and EF and ED are the energy positions of the Fermi level of the ntype graphene and the Dirac point, respectively [27]. The table in Figure 4.5 summarizes the
calculated electron concentrations (ne) of graphene samples under different SU-8 doping
treatments and their corresponding Fermi level shift ΔEF above the Dirac point. The band
diagrams show the energy position of the Dirac point and the shifts in the Fermi level of ptype and n-type doped graphene, respectively. As expected, the increased absorption of the
donors (the photo-acid generator) can effectively enhance the electron doping concentration,
resulting in the upward movement of the Fermi level and thus larger ΔEF.

78

0.022
0.021

(a)

(V)
VV
g g(V)

(mS)

σ (mS)
σ (mS)

10 min
1.5
1.5
1.4
1.4
30 min
1.3
1.3
90 min
1.2
1.2
1.1
1.1
1.0
1.0
0.9
0.9
0.8
0.8
-20
-20
00

20
20

100 µm

0.020
0.019
0.018
-100 -80

-60

(b)

-40

-20

0.86
0.84
0.82
0.80
0.78
0.76
0.74
4 µm 0.72
0.70
0.68
0
20

Vg (V)

Figure ‎4.4 Gate-dependent conductivity curves (σ) of the n-type doped devices (a) with
various SU-8 soft-baking durations of 10 min, 30 min, and 90 min, and (b) with various SU-8
thicknesses of ~ 4 µm and ~ 100 µm.

Figure ‎4.5 (a) Band diagrams showing the shifts in the Fermi level of p-type and n-type
doped graphene, respectively. (b) Calculated electron concentrations (ne) and Fermi level
shift (ΔEF) of graphene samples, as a function of SU-8 thickness and soft-baking duration.

79

The positive slopes of the gate-dependent conductivity (σ-Vg) curves can give a measure of
the electron mobility (µe) of the n-type graphene, which can be expressed using the equation:
µe=Δσ/(CoxΔVg) in which the conductivity (σ) is estimated by σ = (Ids/Vds)(L/W), and L and
W are the length and the width of the graphene channel, respectively. Typically the electron
mobility of the as-fabricated p-type GNM FETs can reach a maximum value of ~ 3332 ± 500
2 -1 -1

cm V s , obtained from a monolayer GNM device. After the SU-8 doping, the electron
mobility of the devices was calculated to be over 3 times lower than that of the undoped ones
2 -1 -1

and varied in a wide range from ~ 382 cm V s

2 -1 -1

to 745 cm V s , depending on electron-

doping concentration and the presence of nanoholes. The maximum electron mobility was
observed in the graphene sample doped with 4 µm uncrosslinked SU-8. Since the SU-8
doping does not increase lattice disorder as demonstrated by the above Raman spectra
(Figure 4.2), the reduction of the electron mobility is mainly attributed to the long-rang
Coulomb scattering induced by impurities from the polymer. This finding also explains the
reduced conductivities of the n-type graphene, in agreement with those observed in typical
polymer doping of graphene [13]. As expected, the presence of nanomeshes further reduced
the electron mobility due to the increased defects in the local carbon lattices. In addition, an
asymmetric effect on electron and hole mobilities was observed in all the devices, which
could be caused by the imbalanced injection of electrons and holes from the source- and
drain-channel interfaces, respectively [12].
4.3.3 Doping mechanism of SU-8
While we have demonstrated that the SU-8 doping can achieve n-type graphene, more studies
are necessary to validate that PAG is indeed the n-type doping source in SU-8. Two
experiments were conducted successively. In the first experiment, only two of the three SU-8
80

components were prepared by dissolving Epon SU-8 (Epikote 157) in organic solvent
(cyclopentanone). A composition of 40 wt% SU-8 resin and 60 wt% solvent was used. The
compound was mixed for 1 hour and then left in air at room temperature for 2 days to mature.
The mixture was spin-coated on the graphene surface, followed by soft baking at 65 ⁰C for
10 min and then at 95 ⁰C for another 10 min. The blue curve in Figure 4.6 shows the Vg-Ids
characteristic of the as-doped graphene, which exhibited a typical p-type property.
After the device characterization, the resin-solvent coating was dissolved in acetone in order
to convert the p-type graphene to the pristine graphene that was immediately subjected to the
second experiment. In this case, the doping chemical was prepared by mixing 40% SU-8
resin, 59% solvent, and 1% PAG. The mixture was baked and matured using the
aforementioned procedures, and then coated on the same pristine graphene. The graphene
transistor doped with the new mixture was characterized and its Vg-Ids and Vds-Ids properties
were given in the red curve of Figure 4.6 and Figure 4.7, respectively. The Dirac point of the
graphene shifted to –29 V, indicating that the pristine graphene was converted to an n-type
one. This result demonstrated that the PAG is an electron donor for the n-type doping of
graphene.

81

Figure ‎4.6 Current transfer characteristcs of graphene FETs doped by two different chemical
mixtures. Case 1 (p-type): graphene doped with the resin-solvent mixture. Case 2 (n-type):
Graphene doped by the resin-solvent-PAG mixture. The inset is the SEM image of the
device.

Figure ‎4.7 Vds-Ids relation of the graphene FET after n-doping with the resin-solvent-PAG
mixture.

To further investigate the effect of the PAG, its percentage concentration in the mixture was
doubled. Figure 4.8 shows that the shift in the Dirac point increased with the increase of the
82

PAG concentration, whereas the Vg-Ids slope decreased, corresponding to a higher electron
doping level and lower electron mobility, respectively. The electron mobility dropped from
2 -1 -1

2388 to 690 cm V

s

when the concentration of the PGA increased from 5.5% to 10%,

while the charge carrier increased from 0.36×10

12

to 1.08 ×10

12

-1

cm . It should be noted

that, since the concentration of the amount of solvent diminished during the soft baking step,
it was difficult to determine the concentration of the PGA with respect to the other chemical
components precisely after soft baking. Figure 4.9 observes the positions of the Dirac points
of several graphene samples, which were doped using compounds with different PAG
concentrations measured before soft baking.

Figure ‎4.8 Device characteristics before and after n-doping with two different PAG
concentrations.

83

Figure ‎4.9 The positions of the Dirac points of four samples with different PAG
concentrations. The triangle, square, circle and star symbols resemble different samples at
different PAG concentrations. The Dirac point positions of some samples were observed at
different PAG concentrations and some were observed at only one PAG concentration since
sometimes it is hard to totally remove the PAG from graphene after soft baking.
4.3.4 Stability of SU-8 doped n-type graphene FETs
The chemical stability of the n-type graphene in air was investigated by monitoring the
dynamic change in the transport properties of the GNM transistors, in which the UV exposed
SU-8 was post-baked at 105 °C for at least 30 min to enable complete crosslinking of the
resist. Then the devices were kept under ambient condition for a week and their gatedependent resistances were recorded daily. As shown in Figure 4.10, the gate-dependent
conductivity curves exhibited only a small upshift of the Dirac points by ~ 3 V from around 9.0 V to - 6.0 V, after 2 days of exposure in ambient air, and then remained stable over the
11

course of testing. Quantitatively, the electron areal concentration was reduced by ~ 2×10
-2

cm , corresponding to the downshift in the Fermi level by only ~ 15 meV from 143 meV to
128 meV. This slight change may be attributed to the reduced concentration of the photo-acid
generator due to the slow crosslinking of SU-8 at room temperature in air. The slopes of the
84

conductivity curves showed no remarkable changes (within 15%) after 7 days of exposure in
air, indicating no major decline in the carrier mobility of the n-type graphene.
Our result demonstrates that the SU-8 encapsulation effectively enhances the air stability of
the n-type doped graphene, by minimizing impurity absorption and scattering associated with
normal ambient condition. We also found that the post-baking was a critical step for the UVexposed SU-8 to achieve good encapsulation of the graphene, without which the n-type
graphene reverted to p-type after being exposed in air for several hours. One possible
explanation for this phenomenon is that the photo-acid generated upon the UV exposure
slowly interacts with the polymer chains at room temperature, resulting in the partial
crosslinking of the local epoxy. The non-uniformly crosslinked SU-8 contained pores to
enable the diffusion of p-type impurities (e.g. oxygen) through the encapsulating layer, which
leads to the rapid degradation of the n-type graphene. In addition, the thin SU-8 coating of ~
4 µm was desired for electron doping of graphene, since thermally induced stress in thick
SU-8 films could damage the local carbon lattices.

2 hrs
1 day
2 days
7 days

σ (mS)

0.42
0.37
0.32
0.27
-20

-10

0
Vg (V)

10

20

Figure ‎4.10 Air stability testing of a representative n-type GNM transistor doped and selfencapsulated with ~ 4 µm crosslinked SU-8. (a) Gate-dependent conductivity (σ) curves (b)
Gate-dependent resistance of the device measured at different time points. The dash lines
represent the positions of the Dirac points.

85

The temperature stability or operation temperature range of graphene transistors is one of the
important factors that has to be taken into account. Therefore, we studied the current transfer
characteristics of graphene FETs doped with cross-linked SU-8, in a wide temperature range
from ~ -15 °C to + 100 °C. Figure 4.11 demonstrates the temperature stability of the n-type
properties of two different devices within the testing temperature range, in which only a small
shift in the Dirac point was observed.

Figure ‎4.11 Temperature stability of two n-type graphene FETs: (a) Vg-Ids curves at a
temperature range from -15 °C to 100 °C. (b) Vg-Ids curves at a temperature range from RT to
100 °C.

86

To better understand the temperature effect on SU-8 doped graphene crystalline structure,
such as atomic bonds, phonons, and thermal expansion, we also studied the temperature
dependence of Raman spectra of both pristine and n-doped graphene samples. During the
measurements, the samples were placed on a hot stage under a HoloProb Raman
spectroscope. All the spectra were excited using a visible (532 nm) laser with a relatively low
laser power (0.53 mW) to avoid graphene heating by the excitation laser. A ×100 objective
lens was used to focus the laser beam on the desired location of the sample. The temperature
dependence of the G-band peak position was measured at several temperatures varying from
room temperature (RT) to 150 °C for both the n-doped and pristine graphene FETs.
First, the temperature dependence of pristine mono- and bi-layer graphene was investigated.
The experimental results show that the elevating temperature leads to the downward shift of
the Raman G-band peak to a lower frequency (Figure 4.12). The extracted negative
-1

temperature coefficients of G-mode for mono- and bi-layer graphene were -0.042 cm /°C
-1

and -0.023 cm /°C, respectively.

Figure ‎4.12 (a) Raman spectra of pristine graphene at different temperatures. (b) Temperature
dependence of G-band peak for mono- and bilayer pristine graphene samples.

87

Figure 4.12 (cont’d)

Second, the temperature dependence of SU-8 doped n-type graphene FET was investigated,
as given in Figure 4.13. Another peak appeared at 1607 cm-1, which was the characteristic
peak of the SU-8 resist. One remarkable observation is that the G-band peak of the n-type
graphene became wider as the temperature elevated. The corresponded G-mode temperature
-1

coefficients of n-doped mono- and bi-layer were determined to be -0.069 and -0.024 cm /
⁰C, respectively (Figure 4.13 (b) and (c)). The results indicate that the G-mode temperature
coefficient of the n-doped monolayer graphene had a highest sensitivity to temperature. This
-6

can be attributed to the effect of the thermal expansion difference between SU-8 (~52×10 /
-6

⁰C) and graphene (~ -0.03×10 / ⁰C).

88

(a)(a)

(b)

Figure ‎4.13 (a) Raman spectra of n-doped graphene at different temperatures. (b) Comparison
of temperature dependence between pristine and n-doped monolayer graphene sheets. (c)
Temperature dependence of bi-layer graphene. The slope of fitting curve represents the
extracted temperature coefficient of G-band peak.

89

Figure 4.13 (cont’d)
(c)

The mechanism of the temperature-dependent downshift of the G-band peak is the elongation
of the C-C bonds due to thermal expansion or an harmonic coupling of phonon modes [28].
The temperature dependence of the G-band can be represented by ω‎ = ω0+XT, where ω is
the frequency of the G-mode, X is the first-order temperature coefficient defined by the slope
of the temperature dependent curve, T is the initial temperature value, and ω0 is the G shift at
the initial temperature. The measured frequency change Δω= ω-ω0 can be written as [29]:

Δω=(XT+XV)

=(

)

(

)

, where XT and XV are the frequency shifts due to

the intrinsic temperature effect as well as the thermal expansion that induces a volume
change. The similar effect was also reported in [30], where the G-band peak of suspended
graphene shifted to a lower value as a tension force was applied to its surface. Therefore, the
G-band shift can be used as an indicator not only for thermal impact but also for mechanical
impact on graphene.

90

It is of note that the Si/SiO2 substrate does not strongly affect the thermal coefficient (X)
since the measured G-band at ~ 1582 cm

-1

is made up of optical phonons due to the in-plane

vibrations. In addition, the E2g symmetry of the G-band confines the motion of the atoms to
the plane of the carbon atoms [31]. The out-of-plane vibrations (ZO phonons) in graphene are
not coupled to the in-plane motion that defines the G-band spectrum position. The out-ofplane vibrations are expected to be more associated with the substrate influence [32].
Therefore, the G-band temperature dependence is mostly the properties of the graphene layer
rather than the substrate characteristic. Furthermore, the thermal expansion coefficient for
-6

SiO2 is low (2.6×10 ), which does not induce thermal stress to the graphene layer.
4.4

Conclusion

In summary, we demonstrated a doping technique for making n-type GNM transistors with
high chemical stability and low defect density, using SU-8 as the electron dopant and
encapsulation simultaneously. The n-type semiconducting properties of the SU-8 doped
grapheme were confirmed by the Raman spectra of the as-doped graphene, the negative shift
of the Dirac point of the transistors, and the change in the energy position of the Fermi level.
The influence of the SU-8 doping on the transport properties of the GNM transistors was also
studied, and results were found to be consistent with the previously reported data.
The n-type effect of the as-doped devices was demonstrated to be chemically stable in air and
completely reversible. Additionally, the n-doped graphene FET was stable between - 22 °C
and 100 °C with no significant shifting in the Dirac point. The temperature effect on the
crystalline structures doped and pristine graphene was also studied. The G-mode of the doped
mono-layer graphene was demonstrated to have a greater sensitivity to temperature in

91

comparison to the pristine graphene, whereas the doped bi-layer graphene showed a relatively
lower sensitivity.
Our developed fabrication of the GNM FETs is compatible with conventional
microfabrication technology, which makes it a promising technique for the mass production
of graphene electronics. Further development in flexible graphene electronics with SU-8 selfencapsulated doping layers also presents tremendous opportunities for future applications in
various fields, such as flexible integrated circuits that will be discussed in the next chapter.

92

REFERENCES

93

REFERENCES

[1]

Y.-C. Lin, C.-Y. Lin, and P.-W. Chiu, "Controllable graphene n-doping with
ammonia plasma," Applied Physics Letters, vol. 96, pp. 133110-133110-3, 2010.

[2]

D. Wei, Y. Liu, Y. Wang, H. Zhang, L. Huang, and G. Yu, "Synthesis of n-doped
graphene by chemical vapor deposition and its electrical properties," Nano Letters,
vol. 9, pp. 1752-1758, 2009.

[3]

B. Guo, Q. Liu, E. Chen, H. Zhu, L. Fang, and J. R. Gong, "Controllable n-doping of
graphene," Nano Letters, vol. 10, pp. 4975-4980, 2010.

[4]

L. Zhao, R. He, K. T. Rim, T. Schiros, K. S. Kim, H. Zhou, C. Gutiérrez, S.
Chockalingam, C. J. Arguello, and L. Pálová, "Visualizing individual nitrogen
dopants in monolayer graphene," Science, vol. 333, pp. 999-1003, 2011.

[5]

X. R. Wang, X. L. Li, L. Zhang, Y. Yoon, P. K. Weber, H. L. Wang, J. Guo, and H. J.
Dai, "N-doping of graphene through electrothermal reactions with ammonia,"
Science, vol. 324, pp. 768-771, May 8 2009.

[6]

J. M. Englert, C. Dotzer, G. A. Yang, M. Schmid, C. Papp, J. M. Gottfried, H. P.
Steinruck, E. Spiecker, F. Hauke, and A. Hirsch, "Covalent bulk functionalization of
graphene," Nature Chemistry, vol. 3, pp. 279-286, Apr 2011.

[7]

I. Childres, L. A. Jauregui, M. Foxe, J. F. Tian, R. Jalilian, I. Jovanovic, and Y. P.
Chen, "Effect of electron-beam irradiation on graphene field effect devices," Applied
Physics Letters, vol. 97, Oct 25 2010.

[8]

Z. Luo, N. J. Pinto, Y. Davila, and A. T. C. Johnson, "Controlled doping of graphene
using ultraviolet irradiation," Applied Physics Letters, vol. 100, pp. 253108-253108-4,
2012.

[9]

N. Stander, B. Huard, and D. Goldhaber-Gordon, "Evidence for Klein tunneling in
graphene p-n junctions," Physical Review Letters, vol. 102, Jan 16 2009.

[10]

A. F. Young and P. Kim, "Quantum interference and Klein tunnelling in graphene
heterojunctions," Nature Physics, vol. 5, pp. 222-226, Mar 2009.

[11]

H.-Y. Chiu, V. Perebeinos, Y.-M. Lin, and P. Avouris, "Controllable p-n junction
formation in monolayer graphene using electrostatic substrate engineering," Nano
Letters, vol. 10, pp. 4634-4639, 2013/07/09 2010.

94

[12]

D. B. Farmer, R. Golizadeh-Mojarad, V. Perebeinos, Y. M. Lin, G. S. Tulevski, J. C.
Tsang, and P. Avouris, "Chemical doping and electron-hole conduction asymmetry in
graphene devices," Nano Letters, vol. 9, pp. 388-392, Jan 2009.

[13]

A. Das, S. Pisana, B. Chakraborty, S. Piscanec, S. K. Saha, U. V. Waghmare, K. S.
Novoselov, H. R. Krishnamurthy, A. K. Geim, A. C. Ferrari, and A. K. Sood,
"Monitoring dopants by Raman scattering in an electrochemically top-gated graphene
transistor," Nature Nanotechnology, vol. 3, pp. 210-215, Apr 2008.

[14]

J. H. Chen, C. Jang, S. Adam, M. S. Fuhrer, E. D. Williams, and M. Ishigami,
"Charged-impurity scattering in graphene," Nature Physics, vol. 4, pp. 377-381, May
2008.

[15]

K. M. McCreary, K. Pi, and R. K. Kawakami, "Metallic and insulating adsorbates on
graphene," Applied Physics Letters, vol. 98, PP. 192101-192101-3, May 9 2011.

[16]

P. H. Ho, Y. C. Yeh, D. Y. Wang, S. S. Li, H. A. Chen, Y. H. Chung, C. C. Lin, W.
H. Wang, and C. W. Chen, "Self-encapsulated doping of n-type graphene transistors
with extended air stability," ACS Nano, vol. 6, pp. 6215-6221, Jul 2012.

[17]

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," Physical Review Letters, vol. 97, Nov 2006.

[18]

H. Al-Mumen, F. Rao, L. Dong, and W. Li, "Tunable graphene nanomesh
semiconductor: Design, fabrication, and characterization," in The 8th Annual IEEE
International Conference on Nano/Micro Engineered and Molecular Systems (IEEE
NEMS 2013), Suzhou, China, 2013.

[19]

F. Ru and J. F. Richard, "Influence of processing conditions on the thermal and
mechanical properties of SU8 negative photoresist coatings," Journal of
Micromechanics and Microengineering, vol. 13, p. 80, 2003.

[20]

K. Stephan, B. Gabriela, L. Michael, H. Daniel, and B. Anja, "Processing of thin SU8 films," Journal of Micromechanics and Microengineering, vol. 18, p. 125020, 2008.

[21]

Y. Wang, M. Bachman, C. E. Sims, G. P. Li, and N. L. Allbritton, "Simple
photografting method to chemically modify and micropattern the surface of SU-8
photoresist," Langmuir, vol. 22, pp. 2719-2725, 2013/07/09 2006.

[22]

D. Graf, F. Molitor, K. Ensslin, C. Stampfer, A. Jungen, C. Hierold, and L. Wirtz,
"Spatially resolved raman spectroscopy of single- and few-layer graphene," Nano
Letters, vol. 7, pp. 238-242, Feb 2007.

95

[23]

Y. C. Lin, C. Y. Lin, and P. W. Chiu, "Controllable graphene N-doping with ammonia
plasma," Applied Physics Letters, vol. 96, PP. 133110-133110-3, Mar 29 2010.

[24]

C. H. Zhang, L. Fu, N. Liu, M. H. Liu, Y. Y. Wang, and Z. F. Liu, "Synthesis of
nitrogen-doped graphene using embedded carbon and nitrogen sources," Advanced
Materials, vol. 23, pp. 1020-1024, Feb 22 2011.

[25]

T. C. Li and S.-P. Lu, "Quantum conductance of graphene nanoribbons with edge
defects," Physical Review B, vol. 77, p. 085408, 2008.

[26]

F. Sols, F. Guinea, and A. H. C. Neto, "Coulomb blockade in graphene nanoribbons,"
Physical Review Letters, vol. 99, p. 166803, 2007.

[27]

Z. Chen, I. Santoso, R. Wang, L. F. Xie, H. Y. Mao, H. Huang, Y. Z. Wang, X. Y.
Gao, Z. K. Chen, D. Ma, A. T. S. Wee, and W. Chen, "Surface transfer hole doping of
epitaxial graphene using MoO3 thin film," Applied Physics Letters, vol. 96, p.
213104, 2010.

[28]

L. Ci, Z. Zhou, L. Song, X. Yan, D. Liu, H. Yuan, Y. Gao, J. Wang, L. Liu, and W.
Zhou, "Temperature dependence of resonant Raman scattering in double-wall carbon
nanotubes," Applied Physics Letters, vol. 82, pp. 3098-3100, 2003.

[29]

I. Calizo, A. Balandin, W. Bao, F. Miao, and C. Lau, "Temperature dependence of the
Raman spectra of graphene and graphene multilayers," Nano Letters, vol. 7, pp. 26452649, 2007.

[30]

G. Tsoukleri, J. Parthenios, K. Papagelis, R. Jalil, A. C. Ferrari, A. K. Geim, K. S.
Novoselov, and C. Galiotis, "Subjecting a graphene monolayer to tension and
compression," Small, vol. 5, pp. 2397-2402, 2009.

[31]

F. Tuinstra and J. L. Koenig, "Raman spectrum of graphite," The Journal of Chemical
Physics, vol. 53, p. 1126, 1970.

[32]

N. Mounet and N. Marzari, "High-accuracy first-principles determination of the
structural, vibrational and thermodynamical properties of diamond, graphite, and
derivatives,"Codense Matter-Material Science, 0412643v1, 2004.

96

5

GRAPHENE LOGIC CIRCUITS BASED ON A METAL SU-8 – GRAPHENE
TECHNOLOGY

5.1

Introduction

Building on previously developed graphene engineering and doping techniques, this chapter
will discuss the design, fabrication, and characterization of complementary two, n-type and ptype, transistor graphene integrated logic circuits on both rigid and flexible substrates. A
technology, metal – SU-8 – graphene (MSG) was proposed for building graphene-based
integrated circuits as an alternative to complementary metal-oxide-semiconductor (CMOS)
technology.
5.2

E-beam Lithography of SU-8

In Chapter 4, we have demonstrated that SU-8 is an effective and reliable doping and
encapsulation material for making n-type graphene field effect transistors (FETs). However,
graphene logic circuits requires making both n-type and p-type transistors monolithically on a
single substrate. Therefore, it is very important to develop a technique capable of selective
doping and patterning n-type graphene using SU-8. Previously, photolithography of SU-8 has
been widely used in microfabrication processes that required thick photoresist structures, but
this method usually suffers from limited resolution and is not suitable for patterning
nanoscale features. It has been reported that SU-8 with a thin thickness can be exposed using
e-beam lithography to achieve small feature sizes down to 30 nm [1, 2]. The chemical
reaction of SU-8 during e-beam exposure is similar to that during ultraviolet (UV) exposure.
In both cases, SU-8 behaves as a negative resist. Upon irradiation, the photo-acid generator
(PAG) decomposes to a photo-acid (hexaflouroantimonic acid) that causes a highly crosslinked and insoluble polymer.

97

In this section, we carefully calibrated the parameters of SU-8 e-beam lithography to achieve
sharp and high-resolution patterns with minimal disturbance to the properties of graphene.
For all experiments, we used the commercially available SU-8 2000.2 and SU-8 3005. The
resist was spin-coated on the substrate with a spin speed of 6000 rpm to realize a thin layer of
SU-8 of less than 1 µm. Different SU-8 thicknesses can be achieved by selecting different
SU-8 models (2000 and 3000 series) and changing the spin speed. If a thinner SU-8 layer is
desired, SU-8 thinner such as cyclopentanone can be added to the original SU-8 to further
reduce the viscosity of the resin. Following spin coating, the SU-8 film was soft baked at 65
°C for 2 min and then at 95 °C for 2 min. The soft-baked SU-8 was exposed to e-beam using
JEOL 840, which was operated at 35 kV with beam current of 25 pA. Then the exposed resist
was developed by SU-8 developer for 1 min, followed by immersion in isopropanol alcohol
(IPA) and a rinse with deionized (DI) water. Finally, the samples were post-baked at 95 °C
for 3 min.
Since the e-beam voltage and current are fixed for the instrument, the minimal line-width is
mainly determined by the beam dose of the e-beam lithography. In order to study the
resolution limit of e-beam exposed SU-8 patterns, the SU-8 resist were exposed using
2

different areal doses between 3 and 300 µC/cm . Figure 5.1 shows the optical microscope
images of the resulting patterns. It can be seen that the e-beam writing resolution was
significantly affected by the beam dose. Large areas of undeveloped SU-8 residues were
2

observed around the desired patterns at high e-beam doses of over 270 µC/cm (Figure 5.1
(a) and (b)), which broadened randomly the desired patterns. This was mainly caused by the
scattering of impinging electrons at the resist interface, which leads to the over exposure of a
larger SU-8 area. Therefore, using a smallest effective dose is considered to be the best
solution to significantly decrease SU-8 residues and improve pattering resolution.
98

Preliminary result shows that the minimum effective areal dose for full exposure of ~ 0.5 µm
2

thick SU-8 was ~ 3 µC/cm . It is of note that these residues were hard to remove even before
the post-baking treatment.

Figure ‎5.1 SU-8 patterns with various feature sizes (500 nm to 3 µm), exposed using different
2

2

2

areal doses of: (a) 300 µC/cm , (b) 15 µC/cm , (c) 4 µC/cm .

5.3

Graphene Inverter Circuit

5.3.1 Fabrication of graphene circuit
One great benefit of using e-beam lithography to expose SU-8 is the ability to selevtively
pattern and n-dope a graphene sheet at the nanoscale. In this section, we developed a method
to realize a graphene inverter circuit (logic NOT gate), by integrating both n- and p- type
graphene FETs on a single substrate. The graphene transistors were constructed
monolithically using a single graphene sheet, where the SU-8 coated graphene was an n-type
region and the uncoated graphene was a p-type region, as schematically shown in Figure 5.2
(a). The SU-8 doping and patterning followed the aforementioned methods. Three metal
electrodes were constructed on the graphene sheet to form two transistors, via e-beam
lithography, metal deposition, and the metal lift-off patterning. The two transistors shared a
same back gate as an input voltage terminal (Vin). The source and drain contacts of the two

99

transistors were connected together for use as an output voltage terminal (V0). Moreover, to
avoid impinging electrons, a sacrificial layer of PMMA or Cu is required. After SU-8
development, the sacrificial layer can then be dissolved by acetone or FeCl3 for PMMA and
Cu respectively.

Figure ‎5.2 An integrated graphene inverter: (a) the schematic of the circuit fabrication; (b) the
SEM image of the fabricated inverter; and (c) the circuit layout.(d) Schematic of the
fabrication technique to avoid impinging electrons.
5.3.2 Characterization of graphene circuit
To verify the formation of the n-type and p-type transistors on the same graphene sheet, the
transfer resistances Rp and Rn between the source and drain contacts of the individual FETs
were measured as a function of the applied voltage Vin (Figure 5.3). The measured Dirac
points of both the n-type and p-type graphene sheets were at -27 V and 17 V, respectively. By
connecting the p-type FET to VDD and the n-type FET to the ground, a Dirac point splitting

100

along the Vin axis and then the complementary configuration was achieved within the region
between two of Dirac points.

Figure ‎5.3 Resistance curve (R vs Vin) of graphene transistors schematically illustrated in
Figure 5.1, measured at room temperature. Rn and RP corresponded to the transfer resistances
of the n-type and p-type graphene regions, respectively.
The inverter function was demonstrated by connecting the source of the n-type FET to the
ground, the source of the p-type FET to a supply voltage (VDD), and the output (V0) to the
common drain of the FETs. The output voltage for this configuration was given by V0=VDD
Rn/(Rp+Rn). The voltage transfer characteristic of the fabricated inverter is given in
Figure 5.4. The two FETs operated in the range between the Dirac points, where the increase
in Vin caused the decrease of Rp and the increase of Rn, thus the increase in the ratio of Rn/Rp.
As a result, V0 decreased with the increase in Vin, leading to the voltage inversion. Away
from the Dirac points, the saturation of the output voltage V0 resulted in the approximately
101

constant ratio of Rn/Rp, when Vin < - 27 V for the n-type FET and Vin > 17 V for the p-type
FET. Due to the small band gap of the doped gaphene, the output voltage could not saturate
to zero or VDD, indicating that the as-fabricated graphene inverter could not be completely
turned off. The voltage gain of the inverter was much less than 1, as determined by A =
dV0/dVin. The small voltage gain of the fabricated inverter could be mainly due to the very
small change of the transistor resistances around the Dirac points. The voltage gain can be
relatively increased by decreasing the oxide thickness. The swing voltage of the inverter was
3 mV at VDD = 10 mV. The threshold voltage (VTH), which is the input voltage at the
maximum voltage gain of the inverter, was approximately 5.6 mV (Vin=VTH) and slightly
greater than VDD/2, which is similar to the value of conventional CMOS inverters.

Figure ‎5.4 Voltage transfer characteristic, demonstrating a logic inverter function.
5.4

Flexible Graphene Transistor

Our developed metal - SU-8 - graphene technology can also be applied in the development of
flexible graphene transistors. Flexible transistors have recently received wide attention due to
its multiple potential applications including flexible screen [3], flexible solar cells [4], and
102

biomedical applications [5]. Conventionally, the fabrication of flexible transistor has used
amorphous silicon or organic polymers [6]. However, the electron mobilities of these
materials are low, which constrain their applications in making low frequency transistors and
integrated circuits. The key advantages of graphene are its fast electron mobility and
excellent mechanical properties, which are essential for fabricating flexible and stretchable
electronic circuits [7-9].
In this work, two different approaches were explored to fabricate graphene FETs on
mechanically flexible Parylene-C substrates. Parylene has been widely used as a polymeric
material for micromachining [10]. It can be conformally coated at room temperature by
chemical vapor deposition (CVD). Due to its electrical insulation, chemical stability, and
biocompatibility, it has also been used as a structural and encapsulating material for
implantable electronic devices and embedded electrodes [11].
5.4.1 Top-gated graphene FETs on mechanically flexible Parylene-C substrates
The first approach used a top gate configuration in the design and fabrication of the flexible
graphene FETs. The fabrication of the trasistor began with the deposition of a 10 µm
Parylene layer on a silicon substrate, which was used as a mechanical carrier during the
subsequent fabrication processes. Graphene was transferred onto the Parylene surface
through the Scotch tape method, located with optical microscope, and confirmed by Raman
spectroscope (Figure 5.5). After that, source and drain electrodes were fabricated on the
graphene sheet with e-beam lithography, Ti/Au metal thermal evaporation, and metal lift-off.
Then the graphene sheet was tailored to a regular shape by oxygen plasma etching. To
fabricate a FET gate, ~ 200 nm SU-8 was spun on the whole chip, followed by soft-baking at
65 °C for 2 min and then at 95 °C for 2 min. In this case, SU-8 was used as a gate dielectric
layer and an n-type doping source. The chip was exposed using photolithography for 20 sec

103

and then hard-baked at 110 °C for 30 min. Next, a metal contact was deposited and patterned
on top of the SU-8 dielectric layer as a top gate electrode. Finally, the flexible Parylene film
was separated from the Si substrate, carrying the fabricated graphene transistors. Figure 5.6
shows the schematic of device fabrication.

Figure ‎5.5 Raman spectrum of graphene transferred on a Parylene substrate. Inset is the
optical microscope image of the graphene.

Figure ‎5.6 Schematic of the top-gated, n-type graphene FET before and after removing the
carrier substrate.

104

Figure 5.7 show a representative flexible graphene transistor fabricated using the above
method and the Ids-Vg characteristic of this device. The Dirac point of the SU-8 doped
graphene was around – 38 V, suggesting a strong n-type transport property that was
consistent with the previous results in Chapter 4. While we realized flexible graphene
transistors using the top gate technique, this device configuration is only suitable for making
n-type graphene FETs on flexible substrates.

Figure ‎5.7 (a) A fabricated graphene n-type FET and (b) the measured Ids-Vg characteristic of
the device
5.4.2 Back-gated graphene FETs on mechanically flexible Parylene-C substrate
In the second approach, a back gate configuration was designed, with which both n- and ptype transistors can be constructed on flexible substrates, as shown in Figure 5.9 (a) and (b),
105

respectively. During the fabrication, a layer of a metal was deposited as the back gate
electrode, followed by the deposition of an insulated layer (SU-8) as the gate dielectric layer.
SU-8 was completely cross-linked by soft-baking, UV exposure, and long post-baking
subsequently. After that, mechanically exfoliated graphene was transferred onto the SU-8
substrate. Then Ti/Au electrodes were fabricated on top of graphene as the source and drain
contacts, through thermal evaporation, e-beam lithography, and metal lift-off processes. After
that, n-type graphene could be obtained by covering the channel region with SU-8 electron
dopant, while p-type graphene could be realized by leaving the channel region uncovered.
Finally, the flexible transistors were released by separating the Parylene film from the silicon
substrate. A representative p-type back-gated transistor and its current transfer property are
given in Figure 5.10 (a) and Figure 5.10 (b). demonstrates the flexibility of the as-fabricated
graphene transistor, which could sustain mechanical twisting and bending (> 90°).

Figure ‎5.8 Device configurations of (a) n-type and (b) p-type back-gated graphene FETs.

106

(a)

(b)

Figure ‎5.9 A fabricated p-type back-gated, flexible graphene transistor before and after
removing the carrier substrate. Silver paste was applied to the contacts in order to avoid
punching the substrate during device testing. Dashed area shows the transparent graphene
sheet. (b) The current transfer characteristic of the device, showing the p-type transport
property of the graphene.

Figure ‎5.10 Mechanical flexibility of the as fabricated graphene transistors.

107

5.5

Conclusion

Graphene doping at a dedicated area represents a practical method for MSG technology.
Building on the graphene doping technique, a graphene-based inverter has been fabricated
and characterized, which exhibits clear voltage inversion property. However, it is difficult to
completely turn off the inverter due to the small band gap of the doped graphene, in addition
to the small voltage gain compared to CMOS inverters. Flexible graphene electronics with
SU-8 doping have also been demonstrated, which provide tremendous opportunities for
future applications of graphene electronics in various fields, such as biomedical systems and
flexible integrated circuits.

108

REFERENCES

109

REFERENCES

[1]

M. Aktary, M. O. Jensen, K. L. Westra, M. J. Brett, and M. R. Freeman, "Highresolution pattern generation using the epoxy novolak SU-8 2000 resist by electron
beam lithography," Journal of Vacuum Science & Technology B: Microelectronics
and Nanometer Structures, vol. 21, pp. L5-L7, 2003.

[2]

B. Bilenberg, S. Jacobsen, M. Schmidt, L. Skjolding, P. Shi, P. Bøggild, J.
Tegenfeldt, and A. Kristensen, "High resolution 100kV electron beam lithography in
SU-8," Microelectronic engineering, vol. 83, pp. 1609-1612, 2006.

[3]

P. Mach, S. Rodriguez, R. Nortrup, P. Wiltzius, and J. Rogers, "Monolithically
integrated, flexible display of polymer-dispersed liquid crystal driven by rubberstamped organic thin-film transistors," Applied Physics Letters, vol. 78, pp. 35923594, 2001.

[4]

F. C. Krebs, "Fabrication and processing of polymer solar cells: a review of printing
and coating techniques," Solar Energy Materials and Solar Cells, vol. 93, pp. 394412, 2009.

[5]

D.-H. Kim, N. Lu, R. Ma, Y.-S. Kim, R.-H. Kim, S. Wang, J. Wu, S. M. Won, H.
Tao, and A. Islam, "Epidermal electronics," Science, vol. 333, pp. 838-843, 2011.

[6]

K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, and H. Hosono, "Roomtemperature fabrication of transparent flexible thin-film transistors using amorphous
oxide semiconductors," Nature, vol. 432, pp. 488-492, 2004.

[7]

A. K. Geim, "Graphene: status and prospects," Science, vol. 324, pp. 1530-1534,
2009.

[8]

J. A. Rogers, "Electronic materials: Making graphene for macroelectronics," Nature
Nanotechnology, vol. 3, pp. 254-255, 2008.

[9]

K. S. Kim, Y. Zhao, H. Jang, S. Y. Lee, J. M. Kim, K. S. Kim, J.-H. Ahn, P. Kim, J.Y. Choi, and B. H. Hong, "Large-scale pattern growth of graphene films for
stretchable transparent electrodes," Nature, vol. 457, pp. 706-710, 2009.

[10]

H.-S. Noha, Y. Huangb, and P. J. Hesketha, "Parylene micromolding, a rapid and
low-cost fabrication method for parylene microchannel," Sensors and Actuators B:
Chemical, vol. 102, pp. 78-85, 2004.

[11]

S. Satyanarayana, D. T. McCormick, and A. Majumdar, "Parylene micro membrane
capacitive sensor array for chemical and biological sensing," Sensors and Actuators
B: Chemical, vol. 115, pp. 494-502, 2006.
110

6

6.1

GRAPHENE APPLICATIONS IN NANOSENSING

Introduction

As discussed in Chapter 1, graphene has several unique properties. First, it has a large
2

specific surface area (2630 m /g), meaning that all atoms can be considered as surface atoms
that are capable of adsorbing different kind of molecules. Second, graphene has inherently
low electrical noise [1] because of its high quality crystal lattice and two dimensional
5

architecture. Third, graphene has high electron mobility (2×10

2 -1 -1

cm V s ) at room

2 -1 -1

temperature and high thermal conductivity (15000 cm V s for graphene on SiO2). These
unique properties make graphene an ideal material for building not only nanoelectronics, but
also nanosensors. Therefore, in this chapter, we explored several potential applications of
graphene in nanosesing for the detection of chemical molecules, temperature, and gas flow
rate.
6.2

Graphene Inter-sheet Sensor for Chemical Detection

6.2.1 Graphene inter-sheet and intra-sheet effects
To date, a variety of graphene-based nanosensors have been demonstrated, where the sensing
mechanism mainly relied on the surface doping and edge effects on their intra-sheet transport
properties [2-6]. Few-layer graphenes (FLGs), with their sub-nanometer inter-sheet spacing,
can enable new transduction mechanisms if inter-sheet effects can be exposed. The merits of
this exposure include inter-sheet tunneling, doping, and other effects besides the enhanced
edge effects due to the extra sides. Such inter-sheet effects can be a function of the spacing
change caused by molecule adsorption/desorption, force/displacement, pressure, surface
tension, or thermal energy, and can potentially enrich the applicable transduction mechanisms
remarkably. All these effects will be available if electrical contacts can be made on different
111

layers of FLGs. However, little effort has been made to the development of inter-layer
graphene sensors due to the difficulties in structuring exposed inter-sheet steps in FLGs.
Figure 6.1 schematically illustrates the sensing mechanisms of the “intra-sheet” and “intersheet” structures.

Figure ‎6.1 Schematic diagrams of (a) an “intra-sheet” and (b) an “inter-sheet” graphene
sensors.
The commonly used “intra-sheet” structure (Figure 6.1 (a)) is based on the change of the
transportation properties inside a single layer of a graphene, induced by external stimuli.
Whereas in the “inter-sheet” version (Figure 6.1 (b)), both the top and bottom layers of a
bilayer graphene (or a FLG) are used as active elements, so the sensing response is based on
the change of the electrical properties between the adjacent graphene layers (e.g., tunneling
current, doping, and edge effects), upon exposure to the external stimuli. For example, when
it is used as gas molecule sensors, the electrical resistance of “intra-sheet” sensor is changed
by the absorption or desorption of the gas molecules that act as donors or acceptors. Its
response is proportional to the quantity of the molecules that absorbed to or desorbed from
the graphene surfaces. However, for the “inter-sheet” sensor, an exponential relationship
between the tunneling current through the two layers of a bilayer graphene and the inter-sheet
distance makes it possible to significantly increase the sensitivity of the “inter-sheet” sensor.
With few molecules attached onto the top sheet, local separation of the bilayer graphene can
be shortened due to van der Waals forces, and the tunneling current will increase accordingly.
112

As the amount of the attached molecules increases, inter-molecular forces can induce surface
tension, which may locally distort the top sheet and result in the increment/decrement of
tunneling current.
Motivated by the potential increase of sensitivity, we designed, fabricated, and characterized
“inter-sheet” sensors. The sensing feasibility of the fabricated devices was demonstrated upon
exposure to various types of vapor molecules, including ethanol (donor-type) and water vapor
(acceptor-type) molecules. The sensing behavior of the “inter-sheet” sensors was compared
with “intra-sheet” sensors fabricated on the same or different graphene sheets. Details are
discussed in the following sections.
6.2.2 Method
To make the inter-sheet graphene sensor, high quality mono-/bilayer graphene sheets were
prepared with the mechanical exfoliation method and then transferred onto SiO2/Si substrates.
After locating the graphene sheets on the substrate, Raman spectroscopy was used to further
identify their thicknesses based on the characteristics of G and the second-order Raman 2D
bands. Figure 6.2 (a) shows a mechanical exfoliated graphene sample that consisted of both
monolayer (region A) and bilayer (region B) regions on a same sheet. The Raman spectra and
Lorenzian fittings of the 2D bands are given in Figure 6.2 (b). For the region A, the G and 2D
-1

-1

band appeared at 1586 cm and 2673 cm , respectively. The intensity ratio of the 2D/G was
approximately 2.68. Combined with the 2D/G ratio, the single peak of its 2D band (the
bottom inset in Figure 6.2 (b)) confirmed that the region A was a monolayer graphene region.
While for the region B, the intensity ratio of G and 2D bands, which appeared at 1578 cm
-1

-1

and 2682 cm , was around 1.02. The 2D band of its Lorentzian fitting contained four peaks
(the top inset in Figure 6.2 (b)), demonstrating a bilayer graphene on the area B.

113

Figure ‎6.2 (a) Example of a mechanical exfoliated graphene containing both mono- and bilayer regions. (b) Raman spectra of different regions. The top and bottom insets represent the
enlarged 2D bands of regions B and A, respectively.
Devices were fabricated using conventional planar micro/nanomachining techniques. The
processes for both “intra-sheet” and “inter-sheet” structured graphene sensors are
schematically depicted in Figure 6.3. Briefly, graphene sheets prepared from the mechanical
exfoliation were located on the SiO2 (300 nm)/Si substrates. Then, the graphene sheets were
engineered to regular shapes by a fine tuned oxygen plasma etching (OPE) method. This OPE
process was developed not only to tailor the edge shapes but also to selectively thinner the
thick graphene regions to a target number of layers (mono- or bi-layer) by removing an
individual layer at each time, as discussed in Chapter 3. After that, Ti (5 nm)/Au (100 nm)
electrodes were fabricated on the same (Figure 6.3 (a)) or different ( Figure 6.3 (b)) graphene
sheets through e-beam lithography and metal thermal evaporation processes. Poly(methyl
methacrylate) (PMMA) served as a protective layer for the OPE and a mask for metal lift-off
step. After dissolving PMMA with acetone and isopropanol alcohol (IPA), the devices were
rinsed with deionized (DI) water and then dried with nitrogen. Finally, the sensors were
-7

degassed at 1×10 Torr for 12 hrs prior to testing.

114

Figure ‎6.3 Schematic processes for fabricating (a) “intra-sheet” and (b) “inter-sheet”
graphene sensors.
6.2.3 Result and discussion
“Intra-sheet” and “inter-sheet” sensors were initially fabricated on different graphene sheets
to study for their vapor sensing behavior. Optical images of the fabricated “intra-sheet” and
“inter-sheet” sensors are given in the Figure 6.4 (a) and (d), respectively. Raman spectra of
the graphene areas of these sensors are also given in Figure 6.4 (b) and (e), respectively. The
corresponding G and 2D bands of the graphene in the “intra-sensor” sensor were located at
1581 cm

-1

and 2688 cm

-1

with close intensities. The four Lorentzian peaks fitting of the

Raman 2D band shown in Figure 6.4 (c) indicates a bilayer graphene structure. The Raman
spectra of the graphene in the “inter-sheet” sensor are shown in Figure 6.4 (e) and (f). In area
-1

A, the intensity of the 2D band (2676 cm ) was near 3 times stronger than that of its G band
-1

(1585 cm ) and the 2D band (Figure 6.4 (f)) contained only one Lorentzian peak. Whereas in
-1

-1

area B, the intensity ratio of the 2D band (2693 cm ) and the G band (1582 cm ) was about
1, and four Lorentzian peaks were needed to fit its 2D band (Figure 6.4 (f)). These confirmed
that the graphene areas A and B are mono- and bi-layered graphene, respectively.

115

Figure ‎6.4 (a) Optical image, (b) Raman spectrum and (c) the enlarged Raman 2D band and
its Lorentzian fitting of the bilayer graphene for an “intra-sheet” sensor. (d) Optical image,
(e) Raman spectrum and (f) the enlarged Raman 2D bands and their Lorentzian fittings of the
mono- and bi-layer graphene regions for an “inter-sheet” sensor.
Using the testing setup shown in Figure 6.5, the time dependencies of sensors, illustrated in
Figure 6.6, were monitored using a picoameter Keithley 6487, before and after introducing
0.5% ethanol vapor (molar concentration) in nitrogen, with a 0.5 V voltage applied to both

116

the sensors. Data were collected after the resistance of each sensor reached its steady state,
which can be attributed to the equilibrium of the vapor molecules adsorption and desorption.

Figure ‎6.5 Schematic diagram of the testing system for both “intra-sheet” and “inter-sheet”
ethanol molecule graphene sensors.
Figure 6.6 shows the resistance responses of these sensors after the introduction of 0.5%
ethanol vapor. It can be seen that the resistance of the “intra-sheet” sensor decreased from
12.95 kΩ to 11.98 kΩ when 0.5% ethanol was introduced. This resistance decrease of the
“intra-sheet” sensor was consistent with the previously reported results [6] because ethanol
molecules serve as electron donors to graphene. As for the “inter-sheet” sensor, it was found
from Figure 6.6 (b) that the resistance change had an opposite trend, which increased from
10.50 kΩ (black curve) to 14.96 kΩ (red curve). As the measured resistances highly depend
on the occupied area of the graphene between two electrodes, a normalized resistance change
(ΔR/R) has been derived as the metrics of the device sensitivity, to eliminate the interference
due to sensing area variance. Based on the measured raw data, the ΔR/R values for the intralayer and the inter-sheet devices were calculated to be approximately 7.5% and 42.5%
respectively, demonstrating that the inter-sheet senor exhibited a much higher sensitivity.

117

Figure ‎6.6 Resistance responses of (a) the “intra-sheet” sensor and (b) the “inter-sheet” sensor
upon exposure to 0.5% ethanol vapor in nitrogen environment.
To further validate the sensitivity enhancement from the “inter-sheet” structure, we fabricated
both “intra-sheet” and “inter-sheet” sensors on a same graphene sheet, as shown in Figure 6.7
(a). This graphene had both a monolayer area A and a bilayer area B, which were also
identified with their Raman spectra and Lorenzian fittings of the 2D bands. An “intra-sheet”
element was formed between electrodes 1 and 2 and an “inter-sheet” one was formed
between electrodes 2 and 3. Similar test was performed by applying a 0.01 V bias to both
sensors to avoid Joule heating damage and electrical breakdown of graphene. Figure 6.7(b)
and (c) show the resistance responses of the “intra-sheet” and the “inter-sheet” configurations
before and after introducing 0.5% ethanol to nitrogen environment. From Figure 6.7 (b), it
can be seen that the current of the “intra-sheet” sensor increased from 2.2 μA to 2.67 μA
upon exposure to 0.5% ethanol-nitrogen mixture, resulting the ΔR/R value of ~ 28.3%. In the
“inter-sheet” situation, the measured current decreased from 0.55 μA to 0.22 μA, which
agreed with the result of the stand-alone inter-sheet sensor. The ΔR/R was calculated to be ~
59.6%, which was more than two times higher than that achieved from the on-chip “intra
sheet” sensor.

118

Figure ‎6.7 (a) SEM image of an “intra-sheet” sensor (electrodes 1 and 2) and an “inter-sheet”
sensor (electrodes 2 and 3) fabricated on a same graphene sheet. Graphs (b) and (c) are time
dependent responses of the “intra-sheet” and “inter-sheet” resistances before and after
introducing 0.5% ethanol into nitrogen.

Both “intra-sheet” and “inter-sheet” sensing elements were also tested using water vapor in a
vacuum environment to investigate their responses to the adsorption/desorption of water
molecules. (a) shows an “inter-sheet” sensor by fabricating Ti/Au electrodes on two different
graphene layers. The light gray and darker regions of the graphene underneath the electrodes
were identified to be monolayer and bilayer structures, respectively (Figure 6.8 (d)).
Figure 6.8 (b) and (c) show two other “intra-sheet” sensors fabricated on the monolayer and
bilayer graphene sheets, as confirmed with their Raman spectra (Figure 6.8 (e) and (f)).

119

Figure ‎6.8 (a), (b) and (c) SEM images an “inter-sheet” sensor fabricated on different
graphene layers, an “intra-sheet” sensor of fabricated on a monolayer graphene, and an
“intra-sheet” sensor fabricated on a bilayer graphene, respectively. (d), (e) and (f) The
corresponding Raman 2D bands and their Lorentzian fittings of the graphene samples in (a),
(b) and (c), respectively. The insets show the entire Raman spectra of the graphene. The A
and B curves in (d) correspond to the mono- and bi-layer graphene regions in (a).

120

The water vapor responses of the aforementioned “intra-sheet” and “inter-sheet” sensors were
characterized in the SEM (Zeiss EVO LS25) chamber. During the experiments, the sample
chamber was first pumped to a base pressure of less than 3.3×10

-3

Pa, in order to remove

possibly absorbed vapor molecules on the graphene surfaces of the sensors. Then, water
vapor was introduced to the sample chamber and the pressure inside the sample chamber was
maintained at 500 Pa. After several minutes, the water vapor source was turned off and the
sample chamber was pumped down again to the base pressure. The sensor responses to water
vapor in several ON-OFF cycles were monitored. Figure 6.9 (a) plots the current response of
the “inter-sheet” sensor in one cycle. It can be seen that the current of the “inter-sheet” sensor
had a dramatic decrease when water vapor was introduced into the sample chamber. When
the water vapor supply was turned off, the current gradually increased to its original value
before the introduction of water vapor. A slow recovery was observed, which is mainly due
to the slow desorption of the physically absorbed vapor molecules on the graphene surface at
room temperature. This recovering time can be significantly reduced by heating the device to
increase the speed of desorption. Similar tests were performed on the “inter-sheet” sensor and
the “intra-sheet” sensors in several ON-OFF cycles. Figure 6.9 (b), (c) and (d) show the
typical current responses of the “inter-sheet” sensor, the “intra-sheet” sensor on a monolayer
graphene, and the “intra-sheet”.

121

Figure ‎6.9 (a) Current response of the “inter-sheet” sensor upon exposure to water vapor at a
pressure of 500 Pa in the sample chamber within one cycle. (b), (c), and (d) The current
responses of the “inter-sheet” sensor, “intra-sheet” sensor on a monolayer graphene and the
“intra-sheet” sensor on a bilayer graphene, respectively, in several cycles of water vapor
turning ON and OFF.
6.2.4 Discussion of sensing mechanism
The higher sensitivity achieved from the “inter-sheet” configuration is promising for a new
group of sensors. This can be attributed to the strong exponential relationship between the
tunneling current through the two adjacent layers of the graphene and the inter-sheet distance,
due to the absorption of ethanol vapor molecules. Since ethanol molecules act as electron
donors, while water molecules act as electron acceptors, it is anticipated that the resistance of
the “intra-sheet” sensor decreased when ethanol molecules were adsorbed on the graphene
surface, and increased when water molecules were adsorbed [6], as shown in Figure 6.7

122

(b) and Figure 6.9 (c) and (d). However, it is noticed that, differs from the decreasing on the
resistance of the “intra-sheet” sensor, the resistance of the “inter-sheet” sensor increases no
matter whether the introduced vapor samples are donor-type (ethanol) or acceptor-type
(water) molecules, as shown in Figure 6.7 (c) and Figure 6.9 (a) and (b). These results
indicate that the “inter-sheet” sensor exhibits a different sensing mechanism compared to the
“inter-sheet” sensor.
Although the exact sensing mechanism of such inter-sheet sensors needs further
investigations, we believe that the current tunneling through the top and the bottom graphene
layers plays an important role that contributes to the higher sensitivities and the increased
resistances of the devices. First, due to the unavoidable absorbates such as photoresist
residues and particles on the graphene surface, the as-fabricated graphene are intrinsically
acceptor-type [7]. Therefore, the absorption of donor-type ethanol molecules decreases the
hole tunneling current through the two layers of the graphene, resulting in the increase of the
inter-layer resistance. While in our intra-layer sensor, the monolayer graphene has less
activated open edges compared to the bilayer graphene in the inter-sheet sensor, and thus has
less chance of being contaminated by those absorbates. In this case, a large portion of the
“intra-layer” graphene possibly remains as the intrinsic monolayer graphene. So the
resistance decreases when ethanol molecules are absorbed onto the monolayer graphene
surface [6]. The other causation for the suppression to the tunneling current through two
different graphene layers of the “inter-sheet” sensors could be the increase of the inter-sheet
distance. When a large number of molecules are absorbed on the graphene surface at the
overlap region of the “inter-sheet” sensors, the transverse intermolecular forces among the
vapor molecules would made some changes to the microscopic corrugations [8] of the top
graphene layer. This reconstruction of the microscopic corrugation would cause the
significant increase on the inter-sheet distance, and thus the suppression on the tunneling
123

current. For our “inter-sheet” devices, the inter-layer structures possess extra open
edges/steps to allow for the adsorption of vapor molecules. Therefore, the increase of the
inter-sheet distance could be more prominent compared to doping effect, resulting in the
increased resistance.
As a conclusion, although our results clearly demonstrate a completely different sensing
characteristic in the “inter-sheet” graphene configuration, further investigations using in-situ
measurements can be envisioned to uncover the exact inter-sheet sensing mechanism and
improve sensor performance. The inter-sheet effects significantly enrich the applicable
transduction mechanisms and provide a promising candidate for sensing various stimuli, such
as chemical/biological molecules, force/displacement, pressure and surface tension, with high
sensitivity.
6.3

Graphene Sensor for Flow/Temperature Sensing

6.3.1 Overview of flow/temperature sensors
Micro/nanoscale flow sensor offers tremendous opportunities for detecting flow rate in a
wide variety of applications, such as microfluidics and lab-on-a chip systems. Modern flow
sensor technologies can be classified into two basic categories: thermal-based [9-15] and
mechanical-deflection-based [16-21] sensors. The thermal-based sensors usually involve a
hot wire configuration at the millimeter scale, which is mainly made of materials with
positive temperature coefficients (PTCs), such as tungsten or platinum [15]. However, PTC
sensors suffer from self-heating effect and slow responding time. The sensing principle of the
mechanical-deflection-based sensors relies on either cantilever deflection or lift force [1618]. While these devices can achieve fast response and high sensitivity, they are often very
fragile and not mechanically stable. Therefore, improved device design and careful selection

124

of materials are required to maximize sensor sensitivity and sensing range, enhance device
reliability, and reduce device size suitable for use in micro and nanoscale systems.
To address these challenges, we propose to use graphene as the sensing material of the
thermal-based flow sensor. Graphene is well known to have unique properties that make it an
excellent choice for the thermo-flow and temperature sensing applications [22-24]. In
particular, graphene has a large surface-to-mass ratio and a higher temperature coefficient of
resistance (TCR) than tungsten and platinum, resulting in low thermal inertia and high
sensitivity to temperature variation. In addition, the negative temperature coefficient (NTC)
of graphene means that its resistance decreases as the environmental temperature increases
[23, 25], which enables self-protection of graphene sensors and minimizes sensing error from
current-induced Joule heating. Furthermore, the planar design of hot-wire sensors is
mechanically more stable because it has no fragile elements such as cantilevers. Finally, our
proposed graphene sensors can be built using conventional micro/nanofabrication techniques
with nanoscale sizes, low power, and low cost, and thus can be easily integrated with
microfluidic components to achieve complex functionality. In this study, we focused
primarily on bi- and few-layer graphene devices and secondarily on mono-layer graphene
because of its relatively lower temperature dependence due to the elevated intrinsic mobility
of charge carriers [25, 26]. The subsequence sections describe the theoretical backgound,
design, fabrication process, packaging, as well as characterization of the graphene hotwire
flow/thermistor sensors.
6.3.2 Heat convection theory
Graphene has an NTC property and its resistance (R) at a certain temperature (T) can be
represented by

( )= ( )

[25] where ћ is the Plank constant, e is the

charge of an electron, v is the velocity, Ef is the Fermi energy, 0 is a backscattering rate from
125

atomically sharp defects in graphene lattice, and V0 is the characteristic interaction constant.
Heat conduction in graphite and graphene is achieved by both phonons and electrons but the
2

contribution of electrons is very little because of the strong sp bonds of carbon atoms [27].
Therefore, graphene sheet resistance (Rs) is approximately proportional to the number of
scattered phonons (n0):

, where ћw is the phonon energy, KB

𝑛 and 𝑛 =

is the Boltzmann constant, and T is the temperature [23]. Acoustic phonons have small
energy (ћw<KBT) so the number of scattered phonons can be approximated as: 𝑛 =
and then

𝑛 =

and then

. Whereas for optical phonons, ћw≈KBT which leads to
. If graphene is deposited on a SiO2 substrate,

scattered phonons would be the combination of graphene phonons and SiO2 phonons. In case
of suspended graphene, the scattered phonons will be contributed only by graphene.
However, the suspended graphene is not preferable in our study since it has little surface
disorder, which makes it less sensitive to temperature [28-30]. For this reason, this work uses
a planar hot wire configuration as the building block of the graphene sensors.
Theoretically, when a hot wire is heated by an electrical current, the thermal energy generated
by the wire is equivalent to the energy loss due to convective heat transfer at thermal
2

equilibrium, I Rw =Aw (Tw-Tf) h, where Aw denotes the surface area of the wire, Tw and Tf
2

are the wire and fluid temperatures respectively, h is the heat transfer coefficient, and I Rw
resembles the thermal energy stored in the wire plus the heat transfer from the surrounding.
The resistance of the hot wire (Rw) depends on the wire temperature (Tw). At steady state,
this resistance can be expressed using a linear approximation of the wire temperature [31],
126

Rw= Rref[(1+α‎(Tw-Tf)], Where Rref is the baseline wire resistance and α is the temperature
coefficient of resistance (TCR). In addition, heat transfer coefficient (h) can be expressed as a
c

function of fluid velocity (Vf) according to the King’s Law: h=a+bVf , where a, b and c are
constants obtained from experimental calibrations. By combining the equations above, one
could get the relation between the fluidic flow velocity and the temperature gradient of the
hot wire:

= {[

[

(
(

)
)

𝑎 𝑏}

.

6.3.3 Fabrication method

Figure ‎6.10 A typical process flow for making the proposed graphene hot-wire sensors.
The fabrication process of the proposed graphene hot wires used conventional
micro/nanofabrication techniques, as depicted in Figure 6.10. Graphene thin films were
prepared using a micromechanical exfoliation technique from highly oriented pyrolytic
graphite (HOPG) flakes and then were transferred onto a 300 nm SiO2/Si substrate. A micro
Raman spectroscope was used to determine the number of layers of graphene sheet. The
mechanical exfoliated bi- and few-layer graphene films were patterned through electronicbeam lithography (EBL) and then an 80 nm copper layer was deposited and patterned as a
mask for graphene etching. The graphene sheet area that was not protected by the copper
mask was etched in oxygen plasma for 40 s with 150 W radio-frequency (RF) power, 516 mT
127

pressure, and 10sccm flow rate. To dissolve the copper mask, the chip was immersed in a
ferric chloride solution for one minute. After that, conductive electrodes were fabricated on
the two terminals of the graphene wire through EBL, thermal evaporation of Ti (5 nm)/Au
(75 nm), and then metal lift-off processes. Finally, the device was annealed in an H2/Ar
environment at 400 °C in order to improve electrode contact resistance and remove
fabrication residues. Figure 6.11 shows the SEM images of as-fabricated bi- and few- layer
graphene hot wires. The bilayer graphene wire had a length of ~ 53 μm and the average width
of ~ 0.5 μm. For the few-layer device, the length was about 50 μm and the average width was
about 10 μm.

Figure ‎6.11 SEM images of the fabricated bilayer (a) and few-layer(b) graphene hot-wire
sensors. The insets show the Raman spectra obtained from the bilayer and few-layer graphene
areas.
6.3.4 Results and discussions
To measure the flow sensitivity of the devices, the graphene hot-wires were placed in a dual
in-line package (DIP) socket and covered by a glass lid with capillaries (350 µm in outer
diameter) as upstream and downstream flow interconnections ( Figure 6.12 (a)). The gas
chamber was sealed with epoxy to minimize side leakage. During the measurement, the
temperature of the package was maintained at a constant temperature (70 °C) whereas the
temperature of nitrogen (N2) inflow was at room temperature (RT). Prior to flow testing, to
128

determine the proper placement of the graphene sensors, the temperature profile inside the
package was simulated with COMSOL Multiphysics 4.3 at different inlet flow rates. As
shown in Figure 6.12 (b), the temperature varied from RT to 70 °C within a narrow distance
close to the inlet capillary, due to the cooling effect of N2 inflow. The simulation result
suggests that the sensor should be placed in a particular distance and orientation from the
inlet in order to generate detectable response. For example, for a flow rate of 0.01 L/min, the
sensor should be localized at less than 2 mm from the inlet capillary tube.

(b)

(a)

Figure ‎6.12 (a) Device sealed in a DIP carrier with capillaries as upstream and downstream
connections. (b) COMSOL simulation shows temperature profile of the packagevs. distance
from the inlet at different input flowrates.
The flow sensing responses of the as-fabricated devices was studied by monitoring the
current and resistance changes of the graphene hot wires as the N2 flow was introduced into
the test chamber. Figure 6.13 shows the transient response of current of a representative
bilayer graphene wire at various N2 flow rates and also N2 input pressures. When the gas
flowed through the device, the current decreased rapidly with a significant undershoot and
then stayed constant with little variations. The steady-state current decreased as the flow rate
increased. The current undershoot could be attributed to the sudden rise of the graphene
resistance, which was caused by the temperature drop on the graphene surface, due to the
cooling effect of the N2 flow. As the current continuously passed through the device, the
129

temperature of the graphene wire increased due to current-induced Joule heating, thereby,
reducing the wire resistance until the equlibruim point was reached for a particular flow rate.

Figure ‎6.13 Transient current responses of a bilayer sensor with (a) different N2 flow rates
and (b) different N2 input pressures.
Similar testing was performed for the few-layer graphene sensors. Figure 6.14 (a) and (b)
summarize the normalized resistance variations of the bi- and few-layer devices as functions
of the N2 flow rate. In particular, the flow sensing resolutions of around 0.07 L/min and 0.1
L/min were obtained from the bilayer and few-layer graphene hot wires, respectively.
Moreover, the negative TCRs of the bi- and few-layer graphene films were determined based

130

on a linear approximation method, resulting in the TCR ranges of -0.00619 K
-1

-1

to -0.008 K

-1

-1

for the bilayer graphene and -0.0014 K to -0.00175 K for the few-layer graphene.

Figure ‎6.14 Normalized resistance changes at different flow rates for (a) a bilayer graphene
hot wire and (b) a few-layer sensor. (c) Comparison between the normalized resistances of
the mono-, bi- and few-layer graphene devices, measured as temperature changed between
RT and 80 °C.

131

From the histogram of Figure 6.14 (c), compared to the mono- and few-layer graphene, the
bilayer graphene has the highest negative TCR, measured as the temperature changed
between RT and 80 °C, which is favorable to temperature sensing applications. Consequently,
the bilayer device is expected to have a higher current capacity and a higher sensitivity than
those of the mono- and few-layer ones.
Time responses to temperature variation for the packaged devices were also measured.
During the experiment, each device was heated from RT to 70 °C and the corresponding
current change was monitored and plotted as a function of time. From the experimental data,
it can be found that the device response was faster with a greater applied voltage. For
example, for the bilayer device, the time constant (the time frame when the current reached
63.2% of the steady-state value) was decreased by 18% as the applied voltage increased from
0.5 V to 3 V, whereas for the few-layer the time constant decreased by 40% as the applied
voltage increased from 0.1 V to 1 V. The temperature sensitivity of the bilayer device was
about 5 times greater than that of the few-layer device, with the applied voltages of 3 V and
0.3 V (Figure 6.15 (a) and (b)) for the bi- and few-layer devices, respectively. Figure 6.15 (c)
shows the time response of the bilayer device during a heating-cooling cycle from RT to
70 °C and then 70 °C to RT. The time constant of the cooling (T1) step was slightly smaller
than the time constant of the heating (T2) step because the heat source was a hot plate
operated in an open environment. The other remarkable thing was that the settling levels of
the currents of all the samples were around 10% to 20% lower at 70 °C than the currents at
RT. This can be attributed to the variation of the metal-graphene contact resistance with
temperature.

132

Figure ‎6.15 Transient responses of different graphene devices as the environmental
temperature varied from RT to 70 °C: (a) the response of a bilayer graphene sensor, (b) the
response of a few-layer graphene sensor, (c) the response of the bilayer device when
temperature increased from 70 °C to RT then increased to 70 °C.
Although an increased voltage can enhance the time response of temperature sensing, the
applied voltage was governed by the resistance of the as-fabricated graphene. This indicates
that the device with a lower resistance has a lower allowable applied voltage, since a high
voltage would damage a device by Joule heating. Therefore, we studied thecurrent-induced
self-heating effect in the bilayer graphene sensor by applying a known voltage across the
133

graphene hot wire and measuring the resulting current change of the device with time. The
self-heating can result in the temperature increase of the graphene and lead to the current
increase due to the NTC of the bilayer graphene. shows the current variation with time due to
self heating. For the sample 1 (Figure 6.16 (a)), with the applied voltage of 3 V, the currrent
was increased by ~ 20% after ~ 55 seconds. The sample 2 (Figure 6.16 (b)) was supplied by 3
V and the corresponding current was increased from from 0.0072 A to 0.0092 A within 1000
sec. The sample 3 was burned due to Joule heating, after applying the 2.5 V input voltage for
2000 sec ( Figure 6.16(c)). Since the self heating effect can cause imprecise temperature
measurement and even device damage, the applied voltage should be selected properly during
the sensing applications.

Figure ‎6.16 (a) Sample 1, Self-heating effects of bilayer graphene device with an applied
voltage of 3V. (b) Sample 2 was damaged by Joul heating after applying 3 V for ~ 2000 sec.
The inset is an AFM image of a failed device due to Joul heating. (c) Sample 3, current
variation with supplied voltage of 3.5 V

134

Figure 6.16 (cont’d)

(b)

(c)

Based on the experimental results, the heat capacity of the device can be extrapolated using
=

, where V is the applied voltage, I is the sensor current,

is the time constant

obtained from the graph, and ∆T is the temperature gradient due to power of Joule heating
(P=I×V). Since we had two heat sources, Joule heat and a hot plate, it is important to
measure the equivalent power of heating at 70 °C. To do this, we first measured the graphene
resistance at 70 °C using the hot plate as the heat source. Then we replaced the hot plate heat
with Joule heat and increased the supply voltage of the graphene wire until the same
135

resistance was obtained. The corresponding current (I) and voltage (V) were recorded and the
equivalent power (P=IV) was calculated to determine the heat capacity of the overall system
that contained the graphene hotwire, the SiO2 substrate, and the microfluidic carrier by the
above expression. The thermal capacity was ~7.35 µWs/°C for a bilayer graphene device
2

with an overall area of 53 × 0.5 µm at 2.42 V, which was the highest compare to the monoand bi-layer devices.
Since the thermal capacity is one of the significant factors affecting the response time of the
device, an alternative sensor package with a low thermal capacity and a low thermal mass
will be needed to speed up the sensor response. For the flow/temperature sensing applications
using graphene, device design can be tailored to obtain desired sensitivity and time response,
by optimizing the dimensions of graphene wire (length and width), the number of graphene
layers, supply voltage, and packaging technique.
6.3

Conclusion

This chapter has discussed the concept, fabrication, and characterization of graphene-based
nanoscale sensors for chemical molecule, temperature, and flow detection. Chemical sensing
of the inter-sheet graphene sensors has been studied, demonstrating that the inter-layer
configuration enables higher sensitivities compared to the most commonly used intra-layer
configuration. There is a high likelihood that this sensitivity improvement can be attributed to
the variance of the tunneling current through the stacked graphene sheets due to the
adsorption of vapor molecules, either mechanically by surface-tension-induced distortion or
electronically by inter-sheet electron donation.
In addition, the surface heat convection properties of the bi- and few-layer graphene wires
were studied and applied in the development of the flow and temperature sensors. The
136

graphene hot wires were fabricated with specific dimensions to achieve high resistances and
uniform temperature distribution. The flow sensing resolutions of approximately 0.07 L/min
and 0.1 L/min were achieved from the bi- and few-layer graphene hot wires, respectively.
The bilayer graphene sensor exhibited greater sensitivity than the few-layer one because of
the larger NTC of the bilayer graphene. Furthermore, the time response of the current through
the graphene hot wires was studied, showing that the time constant decreased as the
resistance of the hot wire decreased, due to the reduced Joule heat.
Future research efforts are necessary to produce large-scale graphene films with the desired
number of layers and high quality. It is expected that, with the development of advanced
surface functionalization and nano-processing technologies, graphene sensors will become
more attractive for applications in the detection of various stimuli, such as
chemical/biological molecules, force/displacement, pressure, and surface tension.

137

APPENDICES

138

APPENDIX A : Fabrication Recipes

A.1 Markers Fabrication
1- Prepare a substrate of 300 nm SiO2 /Si (purchased from WRS Materials).
2- Spin on S1813 photo resist at 4000 rpm for 50 sec.
3- Bake at 110 °C for 1 min on hot plate.
4- Align and expose for 9 sec using ABM mask aligner (Figure A.1(a))
5- Develop in MF325 for 30 sec.
6- Rinse for 30 sec in DI water and then dry by N2.
7- Check under optical microscope that development was successful.
8- Place the sample in Edward Auto-306 thermal evaporator with Ti and Au sources
(Figure A.1 (b)).
9- Deposit 5 nm of Ti layer at a rate of ~0.02 nm/s.
10- Deposit 70 nm of Au layer at a rate of ~0.2 nm/s.
11- Remove the sample from the thermal evaporator and place in acetone for 24 h for
liftoff.
12- Rinse in IPA, DI water and then dry by N2.
13- Etch the photo resist residues by PX-250 O2 plasma, Figure A.1 (c) (300 W, 22 sccm
for 1 min).

139

Figure A.1 Microfabrication instruments (a) ABM Mask aligner (b) Auto 306 Thermal
evaporator (c) PX-250 Plasma Etcher.
A.2 Mechanical Exfoliation of Graphite
1- Place graphite flake onto adhesive tape.
2- Fold the tape next to the flake and then slowly pull apart in order to smoothly cleave
graphite.
3- Repeat steps 2 for around 10 times.
4- Gently place adhesive tape with cleaved graphite onto substrate.
5- Press out any air for 2 min using the handle of a pair of tweezers.
6- Very slowly peel off the tape.
Figure A.2 shows the exfoliation process.

140

Figure A.2 Micromechanical exfoliation of graphite.
A.3 Residue Cleaning Process
1- Insert the samples into a tube furnace (). Flow argon (97%) and hydrogen gases (3%)
at 2000 sccm for 5 min to remove O2 from the glass tube.
2- Heat up the furnace to ~400 °C under H2/Ar flow for 1 hr.
Figure A.3 shows the annealing setup.

Figure A.3 Annealing setup
A.4 Metal Electrodes Deposition
1- Design electrodes using CAD.
2- Spin on MMA-EL9 at 4000 rpm for 50 s.
3- Bake at 180 °C for 90 sec on hot plate and then place on cooling block for 2 min.
4- Spin on PMMA-C2 at 4000 rpm for 50 sec.
141

5- Bake at 180 °C on hot plate for 90 sec.
6- Use JEOL 840A Electron Beam Lithography instrument (Figure A.4) to pattern
2

electrodes (acceleration voltage=35 KV, dose=280 µC/cm and beam current =30
PA).
7- Develop MMA/PMMA for 20 s in 1:3 methyl-isobutyl-ketones (MIBK) to
isopropanol.
8- Rinse for 20 s in IPA and then N2 dry.
9- Check under optical microscope that development was successful. Repeat steps 6 and
7 if under-developed.
10- Deposit 5 nm of Ti as a sticking layer at a rate of ~0.2 Å/s.
11- Deposit 70 nm of Au at a rate of ~2 Å/s.
12- Remove sample from thermal evaporator and place in acetone for 24 h for liftoff.
13- Rinse in IPA, DI water and then dry by N2.
Figure A4. Shows photo images of the e-beam lithography instrument instrument.

(b)

(a)

Figure A.4 JEOL 840A Electron Beam Lithography instrument (a) Sample loading (b)
instrument during writing.

142

A.5 Graphene Patterning and Etching
1- Pattern e-beam resist by electronic-beam lithography (EBL) as in A.4 step 1-9.
2- Deposit 80 nm of copper layer as a mask for graphene etching.
3- Use O2 plasma for 40 s with a power of 150 W and 10 sccm to etch the unprotected
part of graphene sheet.
4- Dissolve the copper mask by immersing the chip in a Ferric Fluoride (FeCl3) for 1
min.
5- Fabricate metal electrodes on the two terminals of the graphene sheet through EBL
(step 1) then deposit Ti (5 nm)/Au (75 nm) layers.
6- Place the samples in acetone for 24 h for metal liftoff process.
7- Anneal in an Ar/H2 environment at 400 °C in order to remove fabrication residues.
A.6 Graphene Surface Recovering after Plasma Etching
The amount of defects that usually produce due to plasma etching can be reduced
significantly with a post-etch annealing treatment, as in the following steps:
1- Insert the samples into a tube furnace.
2- Flow argon gas into a tube furnace at 2500 sccm for 100 min to dismiss O2 from the
furnace to avoid graphene burning at high temperature.
3- Heat up the furnace to 900-1000 °C for 1 hr.
A.7 Graphene Nanomesh FET Fabrication Process
1- Prepare the 300 nm SiO2/Si substrate.
2- Transfer graphene using micromechanical exfoliation.
3- Deposit metal electrodes (as in A.4).
4- Spin on a MMA/PMMA (as in A.4, steps 2-5).
143

5- Use EBL (alignment file only), to introduce nano holes, with the following recipe:
beam current=40 PA, center to center distance=50 nm, spacing 500 nm, and
magnification=×400, acceleration voltage=35 KV and the number of scans=1.
6- Develop (as in A4 steps 7-8).
7- Etch O2 plasma to etch unprotected part of graphene to generate nanomesh.
8- Rinse by acetone, IPA and DI water and then dry with N2.
9- Check graphene nanomesh under Hitachi S-4700 SEM (Figure A.5).

Figure A.5 Hitashi S-4700 SEM. The SEM screen shows a GNR sample.
A.8 Solving the CVD/Substrate Adhesion Problem:
After deposition PMMA (without baking), just leave it for 24 h to dry and then:
1- Dissolve cu in FeCl3 salt.
2- Transfer graphene/PMMA to a beaker of DI water for 5 min.
3- Transfer to another beaker of water for another 5 min.
4- Transfer graphene/PMMA to RCA1 (1:1:20) NH4OH, H2O2, H2O for 10 to 15 min.
5- Transfer to RCA2 (1:1:20) HCL, H2O2, H2O for 10 to 15 min.
6- Transfer to a beaker of DI water.
144

7- Before transfer to substrate, treat it by O2 plasma 150W, 22 sccm for 3 min.
8- Transfer graphene/PMMA to a substrate.
9- Leave it in ambient to dry.
10- Bake at 150 ⁰C for 10 to 15 min.
11- Dissolve PMMA by acetone.
12- Rinse by IPA and dry by N2. Then Bake at 180-200 °C in vacuum for 30min.

145

APPENDIX B: Testing

B.1 Testing methods
B.1.1 In air
REL-4100A probe-station which is connected to semiconductor parameter analyzer was used
for testing in air (Figure B.1).
(b)

(a)

Figure B.1 Testing in air (a) REL-4100A probe station. (b) HP-4145semiconductor parameter
analyser
B.1.2 Testing in inert enviroment
We have used Ar and Hellium enviroment in testing. A feed though testing setup (Figure B.2)
was designed and them made in the MSU chemistry scientific glass shop. The set up made of
Borosilicate-7740 with electrodes were made of three parts copper, tingistin and nichel
matels. The tingistin part was sealed to glass.

146

(a)

(b)

Figure B.2 Testing in an inert gas (a) feed through testing setup (b) set up connected to
semiconductor parameter analyser.
B.1.3 Testing in vacuum
Feed through tool connected to EVO LS-25 SEM was used for vacuum testing. A
plastic chip-holder with aluminum probes was designed and constructed in a CNC
machine shop (Figure B.3).

Figure B.3 Feed-though for SEM (SEM EVO LS-25)
B.2 Testing Instruments
B.2.1 Atomic Force Microscope (Figure B.4)
Dimension 3100 SPM atomic force microscope. Taping made was used to avoid
damaging graphene sheet. The system has the following specifications:


Noise Level: < 0.5Å RMS in vertical (Z) direction

147



X-Y imaging area approx. 90 µm × 90 µm



Z range approx. 6 µm



Lateral accuracy typically within 1%, maximum 2%



Provides full 16-bit resolution on all axes for all scan sizes and offsets.

Figure B.4 Dimension 3100 SPM Atomic Force Microscope.
B.2.2 Raman Spectroscope (Figure B.5)
HoloProbe Raman Spectrograph coupled to an Olympus BX-60 optical microscope. The
System has the following specifications:


Excitation wavelengths: 532 nm



Laser spot size: ~5-10 µm (with an 100× objective)



Spectral resolution: 5 cm



Spectral coverage: ~250 - 4000 cm

-1

(1 cm

-1

with High-Resolution grating)

-1

Raman shift

148

Figure B.5 HoloProbe Raman Spectrograph coupled to an Olympus BX-60 optical
microscope

149

REFERENCES

150

REFERENCES

[1]

T. Palacios, A. Hsu, and H. Wang, "Applications of graphene devices in RF
communications," Communications Magazine, IEEE, vol. 48, pp. 122-128, 2010.

[2]

W. Yang, K. R. Ratinac, S. P. Ringer, P. Thordarson, J. J. Gooding, and F. Braet,
"Carbon nanomaterials in biosensors: Should you use nanotubes or graphene?,"
Angewandte Chemie International Edition, vol. 49, pp. 2114-2138, 2010.

[3]

Y. Dan, Y. Lu, N. J. Kybert, Z. Luo, and A. C. Johnson, "Intrinsic response of
graphene vapor sensors," Nano Letters, vol. 9, pp. 1472-1475, 2009.

[4]

J. T. Robinson, F. K. Perkins, E. S. Snow, Z. Wei, and P. E. Sheehan, "Reduced
graphene oxide molecular sensors," Nano Letters, vol. 8, pp. 3137-3140, 2008.

[5]

J. D. Fowler, M. J. Allen, V. C. Tung, Y. Yang, R. B. Kaner, and B. H. Weiller,
"Practical chemical sensors from chemically derived graphene," ACS Nano, vol. 3, pp.
301-306, 2009.

[6]

Z. Cheng, Q. Li, Z. Li, Q. Zhou, and Y. Fang, "Suspended graphene sensors with
improved signal and reduced noise," Nano Letters, vol. 10, pp. 1864-1868, 2010.

[7]

X. Wang, X. Li, L. Zhang, Y. Yoon, P. K. Weber, H. Wang, J. Guo, and H. Dai, "Ndoping of graphene through electrothermal reactions with ammonia," Science, vol.
324, pp. 768-771, 2009.

[8]

V. Geringer, M. Liebmann, T. Echtermeyer, S. Runte, M. Schmidt, R. Rückamp, M.
C. Lemme, and M. Morgenstern, "Intrinsic and extrinsic corrugation of monolayer
graphene deposited on SiO2," Physical Review Letters, vol. 102, p. 076102, 2009.

[9]

R. G. Johnson and R. E. Higashi, "A highly sensitive silicon chip microtransducer for
air flow and differential pressure sensing applications," Sensors and Actuators, vol.
11, pp. 63-72, 1987.

[10]

A. W. van Herwaarden, D. C. van Duyn, B. W. van Oudheusden, and P. M. Sarro,
"Integrated thermopile sensors," Sensors and Actuators A: Physical, vol. 22, pp. 621630, 1989.

[11]

D. Moser, R. Lenggenhager, and H. Baltes, "Silicon gas flow sensors using industrial
CMOS and bipolar IC technology," Sensors and Actuators A: Physical, vol. 27, pp.
577-581, 1991.

151

[12]

S.-T. Hung, S.-C. Wong, and W. Fang, "The development and application of
microthermal sensors with a mesh-membrane supporting structure," Sensors and
Actuators A: Physical, vol. 84, pp. 70-75, 2000.

[13]

K. A. A. Makinwa and J. H. Huijsing, "A smart wind sensor using thermal sigmadelta modulation techniques," Sensors and Actuators A: Physical, vol. 97-98, pp. 1520, 2002.

[14]

J. Chen, Z. Fan, J. Zou, J. Engel, and C. Liu, "Two-dimensional micromachined flow
sensor array for fluid mechanics studies," Journal of Aerospace Engineering, vol. 16,
pp. 85-97, 2003.

[15]

N. Sabate, J. Santander, L. Fonseca, I. Gracia, and C. Cane, "Multi-range silicon
micromachined flow sensor," Sensors and Actuators A: Physical, vol. 110, pp. 282288, 2004.

[16]

N. Svedin, E. Kalvesten, E. Stemme, and G. Stemme, "A lift-force flow sensor
designed for acceleration insensitivity," Sensors and Actuators A: Physical, vol. 68,
pp. 263-268, 1998.

[17]

N. Svedin, E. Stemme, and G. Stemme, "A new bi-directional gas-flow sensor based
on lift force," in Solid State Sensors and Actuators, 1997. TRANSDUCERS '97
Chicago., 1997 International Conference on, 1997, pp. 145-148 vol.1.

[18]

N. Svedin, E. Kalvesten, and G. Stemme, "A lift force sensor with integrated hotchips for wide range flow measurements," Sensors and Actuators A: Physical, vol.
109, pp. 120-130, 2003.

[19]

Y.-H. Wang, C.-P. Chen, C.-M. Chang, C.-P. Lin, C.-H. Lin, L.-M. Fu, and C.-Y.
Lee, "MEMS-based gas flow sensors," Microfluidics and Nanofluidics, vol. 6, pp.
333-346, 2009.

[20]

Y. Su, A. G. R. Evans, A. Brunnschweiler, and G. Ensell, "Characterization of a
highly sensitive ultra-thin piezoresistive silicon cantilever probe and its application in
gas flow velocity sensing," Journal of Micromechanics and Microengineering, vol.
12, p. 780, 2002.

[21]

Y.-H. Wang, C.-Y. Lee, and C.-M. Chiang, "A MEMS-based air flow sensor with a
free-standing micro-cantilever structure," Sensors, vol. 7, pp. 2389-2401, 2007.

[22]

A. K. Geim, "Graphene: status and prospects," Science, vol. 324, pp. 1530-1534, Jun
19 2009.

152

[23]

J. H. Chen, C. Jang, S. Xiao, M. Ishigami, and M. S. Fuhrer, "Intrinsic and extrinsic
performance limits of graphene devices on SiO2," Nature Nanotechnology, vol. 3, pp.
206-209, 2008.

[24]

H. Al-Mumen, R. Fubo, D. Lixin, and L. Wen, "Characterization of surface heat
convection of bilayer graphene," in Nanotechnology (IEEE-NANO), 2012 12th IEEE
Conference on, 2012, pp. 1-4.

[25]

Q. Shao, G. Liu, D. Teweldebrhan, and A. A. Balandin, "High-temperature quenching
of electrical resistance in graphene interconnects," Applied Physics Letters, vol. 92,
pp. 202108-202108-3, 2008.

[26]

T. Fang, A. Konar, H. Xing, and D. Jena, "High-field transport in two-dimensional
graphene," Physical Review B, vol. 84, p. 125450.

[27]

A. B. Alexander, "Thermal properties of graphene and nanostructured carbon
materials," Nature Materials, vol. 10, pp. 569-581, 2011.

[28]

V. Skakalova, A. B. Kaiser, J. S. Yoo, D. Obergfell, and S. Roth, "Correlation
between resistance fluctuations and temperature dependence of conductivity in
graphene," Physical Review B, vol. 80, p. 153404, 2009.

[29]

K. I. Bolotin, K. J. Sikes, J. Hone, H. L. Stormer, and P. Kim, "Temperaturedependent transport in suspended graphene," Physical Review Letters, vol. 101, p.
096802, 2008.

[30]

V. V. Cheianov and V. I. Fal'ko, "Friedel oscillations, impurity scattering, and
temperature dependence of resistivity in graphene," Physical Review Letters, vol. 97,
p. 226801, 2006.

[31]

H. H. Bruun, Hot wire anemometry principle and signal analysis, 1 ed.: Oxford
Science Publications, 1995.

153