ﬁwua.

x -

.' . I. .4 ‘
Exam“ “x
.1
.ﬂ...‘ . ‘
s...

13;
.i t

3::

ﬁstuués

.’

 

- 1151
v

4;

a
v

1 a z. .3
x.» «L 11!). i
.5 3?... 3.3.1} r ,.
a...» ‘»‘(::(t‘ h»!
§ 31.22.; v». : vs...)
. .3: 4.39215:— . .A ..v1~.$ﬂ.s9
.ivx. I. . .
.3. Isl“. Bingu.‘
, .35 x}

Loft!
“ .uMwnqhnii
ind-1 .

.mwvnlunnx 1'

iv... h.

WEE!
kﬁuwﬁvﬁﬁﬂnﬁ
‘ .1. ﬁt...
59w. use!
Sara. .v. :v...

1...}. .thﬁuﬂnﬁnu

10::cvsnal

. 5
. .

 

 

 

3 Li 3RARY

REA-1-: “A..- (MPH-.3
I-Jllbllibrall gram,

i‘ll‘-.~‘i3f$!t}/

 

 

 

This is to certify that the
dissertation entitled

RADIATION RESILIENCE IN REDUCED DIMENTIONALITY
SYSTEMS

presented by
BENJAMIN W. JACOBS

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Electrical and Computer
Engineean

%//W

- Major Professor’s Signature
/L- /O~ 07

Date

 

MSU is an afﬁrmative-action, equal-opportunity employer

 

- - -u-D-I-o-O-D-0-O-I-I-D-o-I-O-Q-I-I.

4 _._.-.-... -

..-.-.—._.-.-.-.-.-u-o-u-o-o-.-.-.-.—

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

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

 

 

RADIATION RESILIENCE IN REDUCED DIMENSIONALITY SYSTEMS
By

Benjamin W. Jacobs

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Electrical and Computer Engineering

2007

ABSTRACT
RADIATION RESILIENCE IN REDUCED DIMENSIONALITY SYSTEMS
By
Benjamin W. Jacobs
Space based electronic systems such as communications satellites and
interplanetary space probes have become increasingly important since their inception in
the mid 20th century. However, as these systems become smaller and more sophisticated,
the ability to carry radiation shielding to protect against the Space radiation environment
becomes more difﬁcult. Therefore, new lightweight, radiation resilient electronics are
needed to meet the demands of new probe designs. Current silicon based transistors rely
on charged separated regions prone to radiation, which can disrupt normal operation.
Novel nanomaterials that may be inherently radiation resilient offer promising
alternatives. Recent research suggests that nanomaterials and nanomaterial based
electronic devices may have inherent radiation resiliency due to their reduced
dimensionalities, and in some cases enhanced self-healing due to their nanoscale sizes.
Low to high-Z heavy ion irradiation experiments, including heavy ion beams of

krypton-86, krypton-78 and calcium-48, were conducted to explore fundamental heavy
ion nanomaterial interactions. Nanomaterials irradiated included carbon nanotubes,
gallium nitride nanowires and carbon onions. Real time gallium nitride nanowire-based
circuit operation in radiation was also investigated using special vault. Results were
highly encouraging and promise high payoff, given their high radiation resilience and

superior materials and electronic performance.

Newly discovered gallium nitride (GaN) nanowires used in radiation
investigations were unique multiphase nanowires that incorporated zinc-blende and
wurtzite crystalline domains that grew in the longitudinal direction simultaneously.
Structural properties of the multiphase nanowires were studied using high-resolution
transmission electron microscopy. Plain-view high-resolution transmission electron
microscopy (HRTEM) was used to identify the presence of both the zinc-blende and
wurtzite crystalline phases present in the nanowire. C ross-section HRTEM was used to
identify the domain orientation relationships within the nanowire. Scanning electron
microscopy (SEM) investigations identiﬁed growth temperature dependence on GaN
matrix features that led to the multiphase nanowire growth. HRTEM investigations of
the sides of GaN hexagonal platelets, from which nanowires grew, revealed a network of
nanoscale ledges. These ledges were identiﬁed as the nanowire nucleation sites.

Electronic properties of individual nanowires using nanomanipulator probes,
cathodoluminescence, and nanowires integrated into devices were investigated. Using
four-point probe techniques the intrinsic nanowire resistance was accurately measured.
Using two-point probe techniques, evidence of single-phase transport was observed,
where a large liquid protrusion emerged from the nanowire, while an outer structure
remained in tact. Cathodoluminescence showed possible electron conﬁnement effects
and stress in the multiphase nanowire system. Gallium nitride nanowire-based ﬁeld
effect transistors were also fabricated and studied. All measurements indicated that the

nanowires were capable of high current densities.

Copyright by
BENJAMIN W. JACOBS
2007

TABLE OF CONTENTS

 

 

 

LIST OF FIGURES .................................................................................................... vii
1. INTRODUCTION - I
1.1 RADIATION RESII II NCY IN RI DUCI D DIMENSIONAI ITY SYSTEMS ......................... 4
1.2 CARBON NANOTUBES AND CARBON ONIONS ......................................................... 7
1.3 MULTIPIIASE GALLIUM NITRIDE NANOWIRES ....................................................... 9
1.4 MULTIDISCIPLINARY RESEARCII APPROACII ........................................................ 1 1
1.5 SUMMARY OF LATER CHAPTERS ......................................................................... 12
2. INSTRUMENTS AND METHODS ..................................................................... 14
2.1 TRANSMISSION ELECTRON MICROSCOPY ............................................................ 16
2.1.1 High-Resolution Transmission Electron Microscopy ................................... 23
2.1.2 Electron Diffraction .................................................................................... 24
2.1.3 Energy Dispersive X -Ray Spectroscopy ....................................................... 2 7
2.1.4 Electron Energy Loss Spectroscopy ............................................................. 29
2.2 FOCUSI'ZI) ION BEAM ........................................................................................... 31
2.3 SCANNING ELECTRON MICROSCOPY ................................................................... 37
2.4 MICROFABRICATION PROCESS ............................................................................ 41
2. 4.1 Photolithography ........................................................................................ 4]
2.4.2 Electron Beam Lithography ......................................................................... 4 7
2.5 DEVICE CIIARATERIZATION TECI-INIOUES ............................................................ 51
2. 5.] Two—Point Probe ......................................................................................... 5 I
2.5.2 Four-Point Probe ........................................................................................ 52
2.6 SCANNING PROBE MICROSCOPY ......................................................................... 55
2. 6.] Atomic Force Microscopy ........................................................................... 55
2. 6.2 Scanning Tunneling Microscopy .................................................................. 58
2.7 RAMAN SPECTROSCOPY ...................................................................................... 60
2.8 CATIIOI)OLUMINESCI-:NCE AND PIIOTOHJMINESCI-ZNCIi ......................................... 63
2.9 TIIERMOGRAVIME'I‘RIC ANALYSIS ....................................................................... 66

3. REDUCED DIMENSIONALITY SYSTEMS: FROM THREE-DIMENSIONS
TO ZERO - ..... - - - --------.7l

4. ZINC-BLENDE WURTZITE MULTIPHASE GALLIUM NITRIDE

NANOWIRES ...... -- - .......... 78
4.1 NANOWIRE STRUCTURE INVI: STIGATIONS ............................................................ 82
4.1.1 Plain- View HRT EM .................................................................................... 82
4.1.2 Cross-Section HR TEM ................................................................................ 94
4.2 NANOWIRE MATRIX INVESTIGATIONS ............................................................... 111
4.2.1 SEM .......................................................................................................... 112
4.2.2 HRTEM ..................................................................................................... [/8
4.3 NANOWIRE GROWTH AND DISCUSSION ............................................................. 122

4.4 ELECTRONIC PROPERTIES ................................................................................. 127
4. 4. 1 Electronic Investigations of Multiphase GaN Nanowire-Based GaNF E Ts. I 27
4. 4. 2 Electronic Investigations of an Individual Multiphase GaN Nanowire ....... 1 3 2
4. 4. 3 Cathodoluminescence ................................................................................ I 4 I

5. FUNDAMENTALS OF RADIATION INTERACTIONS IN REDUCED

 

 

DIMENSIONALITY SYSTEMS -_ __ _ _ - - _ -- _ 149
5.1 THE SPACE RADIATION ENVIRONMENT ............................................................. 149
5.1.1 Single Event Eﬂects and Total Ionizing Dose ............................................. 1 50
5.1.2 Cosmic Rays .............................................................................................. I5 I
5.2 ELECTRONICS IN SPACE .................................................................................... 154
5.2.1 Silicon Based Electronics .......................................................................... 154
5.2.2 An Example: A MOSF E T in Radiation. ...................................................... 156
5.3 THE NATIONAL SUPERCONDUCTING CYCLOTRON LABORATORY ........................ 160
5.3.1 The Coupled Cyclotron .............................................................................. 160
5.3.2 The Single Event Effects Testing Facility ................................................... I 63
5.4 REDUCED DIMENSIONALITY MATERIALS IN HIGH RADIATION ENVIRONMENTS .. 168
5.4.1 Materials ................................................................................................... I 69
5.4.2 Experiments .............................................................................................. I 77
5. 4. 3 Experimenal Results and Discussion ......................................................... 183
6. FUTURE WORK- -- ....... _ -- ......................... 202
6.1 GALLIUM NITRIDE NANOWIRES ........................................................................ 202
6.2 FUNDAMENTALS OF RADIATION INTERACTIONS IN REDUCED DIMENSIONALITY
SYSTEMS .......................................................................................................... 209
6. 2.1 Other Candidate Materialsfor High Radiation Envirnments ..................... 212
6. 2. 2 Other Candidate Devices for High Radiation Envirnments ........................ 213
7. CONCLUSION ................................................................................................... 217

vi

LIST OF FIGURES

Chapter 2

Figure 2.1 JEOL 2200FS ﬁeld-emission transmission electron microscope. (a) This
particular TEM is in part controlled by computer. The computers used for imaging are
shown on the right. (b) A TEM sample holder. The arrow indicates where a TEM grid is
placed. Note its size. This restricts the size of the sample that can be placed in the TEM.
(0) Close up of the TEM. The electron gun is at the top of the TEM and electrons are
accelerated down the beam column and focused by the electromagnetic lenses. The
sample holder indicates where the sample is located in the beam column. Images: B.W.
Jacobs ........................................................................................................................... 19

Figure 2.2 Illustration of Bragg’s law. An incoming wave will interact with the material
and if the angle is correct it will satisfy the Bragg condition and strongly reﬂect at the
same angle as the incident beam. Image: B.W. Jacobs ................................................. 25

Figure 2.3 FEI Quanta 200 3D dual beam FIB with an integrated Omni
Micromanipulator System at the Electron Microprobe Analysis Laboratory at University
of Michigan. (a) The main components of a dual beam FIB are shown here. It is
computer controlled. The electron gun is situated perpendicular to the sample as in a
normal SEM. The ion gun is tilted 52° relative to the electron gun. The Pt source needle
is next to the ion column. Manual stage control is on the front of the specimen chamber
and is used in the cross-section preparation process. The micromanipulator is situated on
the backside of the instrument. (b) A close up of the micromanipulator is shown here.
The XYZ control is shown and consists of computer-controlled screws that move the
needle. Image: B.W. Jacobs .......................................................................................... 32

Figure 2.4 The TEM cross-section preparation method using a FIB. (a) Nanowire
sample. (b) Pt deposition on top of the nanowire to protect it from ion beam damage. A
small amount of extra Pt is deposited on top of the nanowire to mark its exact location.
(c) Creation of the trenches on either side of the Pt, perpendicular to the nanowire. (d)
The U-cut. Two notches on either side are left to hold the sample in place. The
superimposed needle indicates where it should be welded on the cross-section. (e) The
extracted sample. (f) The extracted sample as it nears the TEM grid. (g) The sample is
welded to the TEM grid. (h) Fine milling of the cross-section. It should be thin enough
to be transparent in a TEM electron beam, less than 100 nm. (i) Ion beam image of the
ﬁnal nanowire cross-section. Images: B.W. Jacobs ...................................................... 35

Figure 2.5 Hitachi S-47OOII FESEM at the W. M. Keck Microfabrication Facility at
Michigan State University. (a) The SEM is controlled via software. On the right is the
control computer. The left highlights several components of the SEM. The electron gun
is at the top and accelerates electrons downward through a series of electromagnetic
lenses and apertures. The sample is located at the base of the instrument. When

vii

introducing a sample into the microscope it is ﬁrst placed in an exchange chamber and
then after rough pumping introduced into the specimen chamber shown in (b). (b) A
close up image of the instrument Shows the specimen chamber where imaging occurs and
the lower and upper secondary electron detectors. Image: B.W. Jacobs ......................... 38

Figure 2.6 Electron beam/sample interaction in the SEM. Electrons from the beam
interact with the sample in a teardrop shaped volume. Variations in the surface topology
cause thin points and edges to appear bright because electrons that enter this area can be
scattered out more easily resulting in a higher number of detected electrons from those
areas, and therefore a brighter spot. Image: B.W. Jacobs .............................................. 39

Figure 2.7 AB-M mask aligner at the W. M. Keck Microfabrication Facility at Michigan
State University. (a) The instrument consists of a UV source that moves on rails to
expose the photoresist on a wafer through a mask, located in the center of the instrument.
An optical microscope on the right is used to align other masks used for multiple
exposures for more complex features. (b) An example of a glass/chrome mask used in
this research. Images: B.W. Jacobs .............................................................................. 42

Figure 2.8 Basic photolithography process. (a) A bare silicon wafer. (b) Silicon dioxide
is grown on the surface and is used as an insulating layer. (c) Photoresist is spun on the
wafer and it is exposed to UV light through a mask. (d) The developer removes the
exposed resist. (e) A metal layer is deposited. (f) The remaining unexposed resist is
taken off along with the metal layer that was on top of it, leaving the desired pattern.
Images: B.W. Jacobs .................................................................................................... 43

Figure 2.9 Edwards Auto 306 Thermal Evaporator at the W. M. Keck Microfabrication
Facility at Michigan State University. (a) The cylindrical evaporation chamber is shown
here. The sample holder is at the top of the chamber and is held upside down. The metal
is at the bottom and evaporated metal drifts upward to the sample. (b) The sample holder
is ﬂipped up for easy access for sample mounting. (c) The evaporation mechanism is at
the bottom of the chamber. The boat is screwed down and high current is passed through
it to heat the metal. A detector is used to monitor the evaporation rate. A shutter is used
to block the substrate from the source to stop metal deposition. ((1) The evaporator
controls. The current can be increased and decreased with a knob. The digital readouts
of the rate monitor and vacuum control provide details of the evaporation conditions.
Images: B.W. Jacobs .................................................................................................... 45

Figure 2.10 Tri-layer photoresist stack. (a) After the ﬁrst layer has been deposited and
exposed, a 10 nm layer of AI is thermally evaporated. The second layer of resist is
deposited and the stack is exposed under a mask. (b) After developing an undercut
should be prevalent. (c) Metal is thermally evaporated with a large separation between
the layer of metal on top of the resist stack and the layer on the bare substrate. (d) The
top layers are removed and clean, sharp features are left. Images: B.W. Jacobs ............ 46

Figure 2.11 JEOL 840A SEM with integrated EBL capability at the W. M. Keck
Microfabrication Facility at Michigan State University. (a) An SEM with a tungsten

viii

ﬁlament is used for EBL. (b) A computer that is integrated into the SEM control system
and is used to manipulate the beam according to a pre-deﬁned pattern designed using a
CAD program. Images: B.W. Jacobs ........................................................................... 48

Figure 2.12 Illustration of the electron beam lithography method. The beam flashes the
sample at a point, moves a small distance ﬂashes it again, and continues this way in a
raster pattern. The spacing between the ﬂashes can be deﬁned in the point spacing and
line spacing. Image: B.W. Jacobs ................................................................................ 49

Figure 2.13 The four-point probe resistance measurement technique. This technique
uses four probes that are placed on the sample. The two outside probes pass a known
current through the sample. The two inside probes measure a voltage. Very little current
passes through the voltage probes. By dividing the known current into the measured
voltage, an accurate resistance of the sample can be measured, with the contact
resistances of the probes omitted. Image: B.W. Jacobs ................................................. 53

Figure 2.14 Diagram of basic AFM operation. The cantilever and tip are rastered along
the sample surface. The variation in sample topology is detected by a reﬂected
laser/photodiode system. The signal from the photodiode is then sent to the computer for
digital image construction .............................................................................................. 56

Figure 2.15 Veeco Instruments Nanoprobe III-A Multimode SPM at the Electronic and
Biological Nanostructures Laboratory at Michigan State University. (a) This image
shows an entire SPM setup. The SPM sits on top of several anti-vibration surfaces. An
optical microscope is used to focus the sample surface and roughly focus the laser beam
on the tip. The SPM is controlled via computer. (b) The head contains the laser,
photodiode and provides the electrical connection to the tip. (c) A close up of the
microscope shows the piezoelectric motion control and the head. (d) An AF M tip holder.
(e) STM tip holder. Images: B.W. Jacobs ..................................................................... 57

Figure 2.16 STM operation. A metal tip is brought into close proximity to the sample
surface. A tunneling voltage is applied, creating an electric ﬁeld across the sample, and
electrons tunnel out of the tip into the sample. The amount of tunneling current will
depend on the variations in the electron density in the sample, i.e. the tunneling current
over an atom will be different than the tunneling current between atoms. Image: adapted
from original image, Michael Schmid, TU Wien ........................................................... 59

Figure 2.17 Illustration of the basic theory of Raman scattering. In each case a photon
causes electrons to enter a virtual energy state. The electron will then drop in energy
either returning to the ground state or entering a higher vibrational energy state. Rayleigh
scattering is the most common form of scattering and is caused by inelastic scattering.
Stokes scattering is more common than anti-Stokes scattering. Image: B.W. Jacobs ....61

Figure 2.18 Kaiser Instruments micro Raman Spectrograph at the W. M. Keck

Microfabrication Facility at Michigan State University. (a) The instrument consists of a
mechanical stage control, laser source and detector, and a computer used for analyzing

ix

data. (b) The laser used for exciting materials is focused through a microscope. This
allows for very small areas of a sample to be analyzed. Image: B.W. Jacobs ................. 62

Figure 2.19 Illustration of the process involved in cathodoluminesence. Electrons
represented by the black dots are excited to the conduction band, E, from the valence
band, E, by the SEM electron beam. Electrons can be excited to states within the band
gap, which can consist of impurities and defects in the material. Image: B.W. Jacobs...64

Figure 2.20 TGA instrument. Rigaku 8120 TG—DTA at the Tokura-Hirata Laboratory at
Tokyo Institute of Technology. The sample material is packed into a small Pt pan and
placed on precise balance. The pan is then put into a furnace and heated up. The arrow
below the furnace label indicates the movement of the furnace. The furnace rides on two
rails and encapsulates the samples in the furnace. The inset shows the whole instrument,
with the furnace on the left and the balance mechanism on the right. A small tube for a
desired gas is also indicated. Images: B.W. Jacobs ....................................................... 67

Chapter 3

Figure 3.1 Examples of reduced dimensionality systems. From left to right: three-
dimensional, two-dimensional, one-dimensional, and zero-dimensional. Images: B.W.
Jacobs ............................................................................................................................ 72

Figure 3.2 A quantum well. Particles in a quantum well are trapped in a conﬁning
potential. If the length of the well, L, is small enough, energy levels within the well will
begin to separate and give rise to quantized energy levels. These energy levels are
determined by the equation on the right, where n is the quantized energy level, II is
Planck’s constant, m is the mass of the particle, and L is the length of the well. Image:
B.W. Jacobs .................................................................................................................. 73

Chapter 4

Figure 4.1 TEM and HRTEM images Of a multiphase GaN nanowire that grew at 850°
C. (a) The TEM image shows the width of the nanowire. Contrast variations indicate
changes in thickness as well as phase across the nanowire. (b) An H RTEM image of the
indicated boxed area in (a). The zinc-blende and wurtzite domains are clearly
distinguishable in this image along with the sharp phase transition denoted by the dotted
line. Images: B.W. Jacobs ............................................................................................. 84

Figure 4.2 HRTEM image of a nanowire that grew at 850° C, and SAED and FF T
analysis illustrating how the FFT can augment SAED patterns. (a) An HRTEM image of
a nanowire shows the multiphase structure of the nanowire. (b) An SAED pattern of the
nanowire showing both zinc-blende and wurtzite contributions. The zinc-blende SAED
pattern was along the [001] axis and the wurtzite was along the [0001] axis. (c) FFT of
the wurtzite phase taken from the boxed area in (a). (d) FFT of the zinc-blende phase

taken from the boxed area in (a). These FFTS separate out the zinc-blende and wurtzite
contributions in the SAED pattern, and identiﬁed where these contributions were located
in the nanowire. Images and interpretation: B.W. Jacobs .............................................. 85

Figure 4.3 (a) An HRTEM image of a single-phase wurtzite nanowire grown at 1000° C.
The wurtzite growth direction is the same as the nanowires grown at lower temperatures,
[ 2T T0]. N0 zinc-blende phase was observed. (b) This SAED pattern was solved for the
wurtzite crystal structure along the [0001] zone axis. Images and interpretation: B.W.
Jacobs ............................................................................................................................ 86

Figure 4.4 HRTEM images of nanowires grown at 1000° C and 850° C. (a) An
HRTEM image of a nanowire grown at 1000° C. Zinc-blende and wurtzite phases were
identiﬁed using FFTS taken from the indicated areas. A sharp phase transition was
evident. The arrows indicate the nanowire growth direction for each phase. (b) An
HRTEM image of a nanowire grown at 850° C. A very similar structure was observed,
with similar FFTS and the same growth directions for each phase. Images and
interpretation: B.W. Jacobs ............................................................................................ 87

Figure 4.5 TEM images of a nanowire grown at 1000° C. (a) A TEM images shows the
nanowire end, with its tapered tip. It was about 700 nm wide. Its growth was along the
[0001] direction. (b) An HRTEM image taken from the boxed area in (a) shows a well-
resolved lattice indicating the high crystal quality of the nanowire. (c) An SAED pattern
was solved for the wurtzite structure along the [01 T0] zone axis. Images and
interpretation: B.W. Jacobs ............................................................................................ 88

Figure 4.6 Simulated diffraction patterns highlighting similar features found in some
wurtzite and zinc-blende patterns. (a) [0001] zone a>£is for the wurtzite structure. (b)
[I I 1] zone axis for the zinc-blende structure. (c) [10 l 2] zone axis for the wurtzite

structure. (d) [112] zone axis for the zinc-blende structure. Images and interpretation:
B.W. Jacobs .................................................................................................................. 90

Figure 4.7 EDS spectrum of a GaN nanowire. The nitrogen and gallium peaks indicate
the purity of these nanowires. A copper peak is also present due the copper TEM grid
used to support the nanowire sample. Spectrum and interpretation: B.W. Jacobs .......... 92

Figure 4.8 EELS spectra of a GaN nanowire. (a) The nitrogen core loss is located at 401
eV, and a strong signal was measured there indicating a signiﬁcant presence of nitrogen.
The oxygen core loss at 532 eV is located on this spectrum but no peak was detected
conﬁrming the lack of oxygen. (b) The gallium core loss is located at 1115 eV, and a
strong signal was measured there indicating a signiﬁcant presence of gallium. Spectra
and interpretation: B.W. Jacobs ..................................................................................... 93

Figure 4.9 TEM images of cross-sections fabricated with the FIB. (a) Cross-section
HRTEM image of a nanowire grown at 850° C. (b) Cross-section HRTEM image of a
nanowire grown at 950° C. (c) Cross-section TEM image of multiple nanowires grown at
1000° C. (d) SEM image of the group of nanowires from which the cross-section was

xi

 

taken. The dotted line denotes where the cross-section was taken. (e) A close up of this
group shows each individual nanowire nucleating from the matrix. Both the large and
small nanowires can be seen here. (I) An ion beam images shows the cross-section half
way through the extraction process. The arrow points out where the three large
nanowires were located. The nanowires were on top of a Silicon substrate and under a
thin layer of gold and a thick layer of platinum to protect them from ion beam damage.
(g) A TEM image of a small nanowire. Its location within the nanowire group is
indicated in (c). (h) A TEM image of a second small nanowire. Its location within the
nanowire group is also indicated in (c). Images: B.W. Jacobs ....................................... 96

Figure 4.10 Details of the cross-section taken from the nanowire grown at 850° C. This
nanowire was separated into ﬁve regions corresponding to its distinct crystalline
domains. These regions were divided along domain interfaces. FFTS from each region
are shown on the right. Regions 1-4 were wurtzite and region 5 was zinc-blende. The
arrows in the FF Ts indicate the [0001] and [111] directions in the wurtzite and zinc-
blende domains, respectively. Zinc-blende and wurtzite crystal models show which
planes were responsible for faceting resulting in the triangular shape of the nanowire. In
the wurtzite structure the { IOT I} and (0001) planes faceted and were parallel with the
growth direction, while the {2T TO} planes were perpendicular to the growth direction.
In the zinc-blende structure the {I 1]} planes faceted and were parallel to the growth
direction, while the {011} planes were perpendicular to the growth direction. Images and
interpretation: B.W. Jacobs ............................................................................................ 98

Figure 4.11 HRTEM images of the cross-section taken from the nanowire grown at 850°
C. (a) A close up from the corresponding boxed area in the center inset revealed several
distinct crystalline domains with coherent and incoherent interfaces. as denoted by the
solid and dashed lines, respectively. A dark contrast wurtzite domain was observed and
could be due to a slightly different orientation. (b) The dark contrast wurtzite domain,
from the corresponding boxed area in the center inset, is surrounded by both zinc-blende
and wurtzite domains forming incoherent interface at every edge, as denoted by the
dashed lines. The zinc-blende domains form coherent interfaces with the surrounding
wurtzite domains, as denoted by the solid lines. (c) A close up of an incoherent interface,
denoted by the dashed line, from the corresponding boxed area in the center inset. (d)
This image shows the same incoherent interface at a different location. On the right side
of this ﬁgure a long coherent interface that extends from the center of the nanowire to the
outside edge forms between the (I 1 l) and (0001) planes, as denoted by the solid line.
Images and interpretation: B.W. Jacobs ....................................................................... 100

Figure 4.12 Details of the cross-section taken from the nanowire grown at 950° C. This
nanowire was separated into six regions corresponding to distinct crystalline domains.
These regions were divided along domain interfaces. FFTS from each region are Shown
on the right and bottom. Regions 1-5 were wurtzite and region 6 was zinc-blende. The
arrows point in the [0001] and [l l 1] directions in the wurtzite and zinc-blende domains,
respectively, to emphasize differences in orientation. Images and interpretation: B.W.
Jacobs .......................................................................................................................... 102

xii

Figure 4.13 HRTEM images of the cross-section taken from the nanowire grown at 950°
C. (a) A close up from the indicated area in the center inset revealed two distinct
crystalline domains with incoherent interfaces. Possible stacking faults ran horizontally
across the center of this image. (b) This image shows three distinct wurtzite domains
with two incoherent interfaces. (0) Two wurtzite domains with an incoherent interface
that changed its direction running from the center to the outside edge of the nanowire.

(d) This image shows the center zinc-blende domain and a coherent interface between the
zinc-blende and wurtzite domain. The top and bottom domain surrounding the zinc-
blende domain were possible stacking faults that transitioned to wurtzite crystal
structures. Images and interpretation: B.W. Jacobs ...................................................... 104

Figure 4.14 HRTEM images Of the cross-section taken from the nanowire grown at
1000° C. (a) A large nanowire with a width of roughly 1.5 um is shown. A wurtzite
crystal model shows the facet planes parallel to the growth direction, {IOTO}, that
resulted in the hexagonal cross-section shape, and the growth plane, (000]),
perpendicular to the growth direction. (b) A high-resolution image of region 1 showing
the high crystallinity of the nanowire. An FFT of region I that could be solved for the
wurtzite crystal structure in the [0001] direction is shown in the inset. (c) A close up
image of region 2 revealed a dark contrast wurtzite domain. ((1) TEM image of the entire
cross-section indicating where this nanowire cross-section was located. Images and
interpretation: B.W. Jacobs .......................................................................................... 106

Figure 4.15 HRTEM images of the cross-section taken from the nanowire grown at
1000° C. (a) A small nanowire with a width of roughly 200 nm is shown here. It grew in
the same direction as the nanowire shown in ﬁgure 4.15. It was divided into three
regions. Regions 2 and 3 were lighter in contrast relative to the surrounding region I.
The FFTS Shown on the right were all solved for the wurtzite structure along the [0001]
direction. They were all in the same orientation. (b) A TEM image of the entire cross-
section showing where this nanowire cross-section was located. (c) A close up of region
2 and 3 revealed that the light contrast regions were both hexagonal in shape. ((1) A high-
resolution image of region I showed the high crystallinity of the nanowire. Images and
interpretation: B.W. Jacobs .......................................................................................... 107

Figure 4.16 HRTEM images of the cross-section taken from the nanowire grown at
'IOOO° C. (a) A large “double” nanowire with a width of roughly 2 pm at its widest point
is shown here. It grew in the same direction as the nanowire shown in ﬁgure 4. I 5 and
4.16. It was divided into four regions. Regions 1 and 2 are the left and right nanowires
that made up the double nanowire. Regions 3 and 4 were areas with lighter contrast
relative to the surrounding regions I and 2. The FFTS shown on the bottom of the image
were taken from regions I and 2, and were solved for wurtzite structure along the [000]]
direction. They were in the same orientation. (b) A TEM image of the entire cross-
section indicates where this nanowire cross-section was located. (c) A close up of region
3 indicated that the light contrast region was hexagonal in shape. A small platelet was
partially covering a hole in the center of the nanowire. (d) A close up of region 4 also
shows an area of lighter contrast with a hole at its center. Images and interpretation: B.W.
Jacobs .......................................................................................................................... 109

xiii

 

Figure 4.17 SEM images of the matrix that formed at 850° C. (a) The matrix is
relatively uniform with feature sizes of around 200 nm to 1 pm in size. (b) A close up
image shows the features in detail. There are many small and medium sized GaN crystal
formations. Two nanowires in this image, as indicated by the arrows, have nucleated
from the matrix. Images: K. McElroy ......................................................................... 1 12

Figure 4.18 SEM images of the matrix that formed at 850° C. (a) A nanowire nucleated
from the side of a hexagonal platelet. The arrows Show the probable nanowire growth
direction, [2T T0] and the sides of the hexagonal platelet, {IOTO}. (b) A nanowire
nucleated from the side of a hexagonal platelet. The arrows show the nanowire growth
direction and sides of the platelet. Images: B.W. Jacobs ............................................. I 13

Figure 4.19 SEM images of the matrix that formed at 950° C. (a) The matrix is
relatively uniform with feature sizes of around 200 nm to 1 pm in size, and Similar to the
matrix features formed at 850° C. (b) A close up image Shows the features in detail.
There are many small and medium sized GaN crystal formations. One nanowire in this
image, as indicated by the arrow, has nucleated from the matrix. Images: K. McElroy
.................................................................................................................................... 114

Figure 4.20 SEM images of the matrix that formed at 950° C. (a) A nanowire nucleated
from the side of a hexagonal platelet. The arrows show the probable nanowire growth
direction, [IOTO] and the sides of the hexagonal platelet, {2T TO}. (b) A nanowire is
also shown to have nucleated from the side of a hexagonal platelet. The arrows show the
nanowire growth direction and sides of the platelet. Images: B.W. Jacobs .................. 115

Figure 4.21 SEM images of the matrix that formed at 1000° C. (a) The matrix formed
distinct features that could be divided into three regions. Region 1 indicates where the
smallest features formed, region 2 the larger features, similar to those found in the matrix
that formed at 850° C and 950° C and region 3 with the largest features. (b) A higher
magniﬁcation image of regions I and 3 shows the features in more detail. The small
features of region I, in the foreground, are compared to the larger features of region 3, in
the background. One nanowire in this image, as indicated by the arrow, has nucleated
from region 1. (c) A higher magniﬁcation image of region 2 shows the features in more
detail. This region is very similar to the matrix found to form at 850° C and 950° C.
Three nanowire in this image, as indicated by the arrows, have nucleated from this
region. ((1) A higher magniﬁcation image of region 3 shows the features in more detail.
The large crystal formations are easily seen here. A large rod is seen in the center of the
image. These large rods were observed in the matrix that formed at 1000° C and not in
the matrix that formed at the lower two temperatures. Images: K. McElroy ................ 116

Figure 4.22 SEM and TEM images of the matrix that formed at 1000° C. (a) A SEM
image of multiple nanowires that grew from the matrix. The arrow indicates where a
nanowire nucleated from the top of a hexagonal platelet. (b) A SEM image of a possible
nanowire nucleation site on the top of a hexagonal platelet. (c) A TEM image of the
matrix showing large and small hexagonal holes, which may be sites where nanowires

xiv

 

nucleated and broke off. ((1) A TEM image of the matrix also showing large and small
hexagonal holes indicating possible nanowire nucleation sites. Images: (a),(b) K.
McElroy, (c),(d) B.W. Jacobs ...................................................................................... 117

Figure 4.23 TEM images of a hexagonal platelet and its nanoscale ledges from matrix
that formed at 850° C. (a) A TEM image shows a typical GaN hexagonal platelet. (b)
HRTEM image of the indicated area in (a) revealed a network of nanoscale ledges.
Images: B.W. Jacobs ................................................................................................... l 18

Figure 4.24 HRTEM images of the nanoscale ledges from matrix that formed at 850° C.
(a) An HRTEM image shows the side of a hexagonal platelet revealing the orientation of
the nanoscale ledges. (b) An HRTEM image taken from a different area of the same
hexagonal platelet revealed nanoscale ledges in a different but equivalent orientation as
the ledges shown in (a) (c) An HRTEM image of a different platelet shows no nanoscale
ledges revealing that not all hexagonal platelets have these features. (d) An HRTEM
image of a different area on the same platelet in (c) shows the smooth edges of this
platelet. Images: B.W. Jacobs ..................................................................................... 120

Figure 4.25 TEM images of hexagonal platelets and nanoscale ledges from matrix that
formed at 1000° C. (a) Decomposed platelet resulting in the formation of nanoscale
ledges. The dotted line indicates a fully formed platelet. (b) An HRTEM image of the
area indicated in (a) shows the features and orientation of the ledges. (c) A TEM image
of a different platelet indicates that nanoscale ledge formation can occur on the sides of
platelets leading to a roughened edge. Images: B.W. Jacobs ....................................... 121

Figure 4.26 Model showing the nanoscale ledges. The edges are the {IOTO} planes
and t_he peaks point in the <2T TO> direction. Nanoscale ledge growth occurred on the

{ 10 I 0} planes and nanowire growth occurred along the <2T TO> direction. Image:

B.W. Jacobs ................................................................................................................ 122

Figure 4.27 SEM images indicating the evidence of the {IOTO} growth. (a) The arrow
indicates where the {IOTO} platelet growth occurred, near the nanowire nucleation site,
and the nanowire continuing to grow from there. (b) The arrow indicates a similar
platelet grth in this image. The nanowire is shown to continue to grow from the base
formation. Images: K. McElroy ................................................................................... 123

Figure 4.28 SEM images showing the tapered tips of nanowires that grew at 1000° C
along the [0001] direction. The tips faceted on the { 101 1} planes. Images: K. McElroy
Interpretation: B.W. Jacobs ......................................................................................... 124

Figure 4.29 A typical GaNFET conﬁguration. A highly doped silicon wafer was used as
the substrate. SiOz was grown as the gate dielectric and a layer of metal was thermally

evaporated on the backside of the wafer for the gate electrode. Nanowires were dispersed
on top of the wafer and source and drain contacts were patterned with EBL. Image: B.W.
Jacobs ......................................................................................................................... 128

XV

Figure 4.30 I-V characteristics of a GaNFET. (a) The current was measured as V05 was
swept and V03 stepped. The nonlinearity of the curves indicated Schottky barrier
behavior. (b) This graphs emphasizes the change in current as a result of the change in
V03. These data were extracted from the data in (a). I-V characteristics: B.W. Jacobs et
al. Nanotechnol, 18, 475710 (2007) with permission .................................................. 129

Figure 4.31 I-V characteristics of the circuit shown in the SEM image. (a) The
nonlinear curve indicated that the nanowire-metal junction formed a Schottky barrier. At
6 Vns over 70 uA of current was passing through the nanowire. (b) An SEM image
shows the actual circuit. The SiOz insulating layer and Ti/Au contacts are labeled. The
arrow shows the path current takes through the circuit. I-V characteristics, image and
interpretation: B.W. Jacobs ......................................................................................... 130

Figure 4.32 I-V characteristics of a GaNFETs with three nanowires hooked in parallel.
(a) The measured current was very high, over 400 DA, at 6 VDs. (b) An SEM image
shows the entire circuit. Photolithography (PL) was used to deﬁne the large contact pads
where the contact probes were placed. The left and right boxes denote where close ups
were taken in (c) and (d), respectively. (c) An SEM image showing two nanowires
contacted in parallel. EBL was used to deﬁne the areas where metal was deposited on
top of the nanowires. The bottom left pattern created with EBL is seen on top of an PL
deﬁned pattern. (d) A SEM image shows the third nanowire hooked in parallel and the
EBL deﬁned patterns that connected it. I-V characteristics, images and interpretation:
B.W. Jacobs ............................................................................................................... 131

Figure 4.33 The Zyvex SIOO Nanomanipulator System. (a) The entire probing system is
mounted in an SEM for high resolution imaging essential for nanoprobing and
nanomanipulation. The arrows point out the four piezoelectrically controlled tungsten
nanomanipulators. Images: R. Stallcup, A. Hartman and V.M. Ayres ..................... 133

Figure 4.34 SEM images showing the four-point probe conﬁgurations. Probes 1 and 4
were the current source probes, 2 and 3 were the voltage sense probes. (a) Conﬁguration
A, all probes were contacted to the contact pads. (b) Conﬁguration B; current source
probes were contacted to the contact pads, while the voltage sense probes were contacted
to the nanowire. (c) Conﬁguration C; all probes were contacted to the nanowire. (d)
Conﬁguration D; one current source probe was contacting the end of the nanowire and
the other the contact pad, while the voltage sense probes were contacted to the nanowire.
Images: B.W. Jacobs et al. Nanotechnol, 18, 475710 (2007) with permission ............. 135

Figure 4.35 Four-point probe resistance measurements. (a) Conﬁguration A, total
system resistance measurement. Resistance values were between 3 and 5 MO. Negative
values corresponded to current sourced in the opposite direction. (b) Conﬁguration C,
NW resistance measurement. side contact. Resistance values were between 500 and 800
RH. (c) Conﬁguration D, NW resistance measurement, end contact. Resistance values
were between 375 k!) and 3 MO. Below the resistance diagrams are the corresponding
nanomanipulator nanoprobe placements for each conﬁguration. Resistance
measurements and diagrams: R. Stallcup, A. Hartman, V.M. Ayres and B.W. Jacobs .136

xvi

 

Figure 4.36 Two-point probe results. The graph shows the different voltage
dependences of the current when the nanoprobes are on the Ti/Au contacts (dotted line,
compliance was set to 250 nA) and directly on the nanowire (dashed line). Diagrams of
the probe conﬁgurations are shown on the right. I-V characteristics and diagrams: K.
McElroy and B.W. Jacobs .......................................................................................... 138

Figure 4.37 SEM images of the breakdown measurements, in conﬁguration D. (a) The
fractured nanowire. (b) The pull-apart between the voltage probes. The location of the
pull-apart is shown in the inset. (c) Simultaneous onset of single-phase melt. The
location of the melt is shown in the inset. (d) A larger amount of material melted and
formed a protrusion outside the nanowire. Images: B.W. Jacobs et al. Nanotechnol, 18,
475710 (2007) with permission ................................................................................... 140

Figure 4.38 Cathodoluminescence spectrum showing the double peak indicating the
wurtzite and zinc-blende phases present in the nanowire. The band gap energies for each
phase is blue-shifted on the order of 0.4 eV. Spectra: B.W. Jacobs et al. Nano Lett. 7,
1435-1438 (2007) with permission .............................................................................. 143

Chapter 5

Figure 5.1 The Space Radiation Environment. Galactic cosmic rays constantly bombard
the Earth and surrounding space. The Earth’s magnetic ﬁeld, as indicated by the lines
surrounding the Earth, protects it from harmful radiation. Image: NASA ................... 150

Figure 5.2 n-type MOSFET. Radiation induced damage by an energetic heavy ion. As
the heavy ion traverses through the semiconductor, it leaves a trail of defects. Images:
B.W. Jacobs ............................................................................................................... 158

Figure 5.3 Illustration of the particle acceleration process at the NSCL. Ions are sent
from the source through attenuators to adjust the beam current. They are then injected
into the K500 cyclotron and then the K1200 for acceleration. Degraders and attenuators
adjust the beam properties, which is sent to the speciﬁed beam vault for experiments.

The boxed area in the upper right represents the SEETF vault and its components.

Image: adapted from R. Ladbury et al. IEEE T. Nucl. Sci. 51, 3664 (2004) with
permission ................................................................................................................... 164

Figure 5.4 Image of the NSCL SEETF. Each main component of the SEETF is noted.
The ion beam runs through the PPAC for analysis, exits through a zirconium foil, runs
through the plastic scintillator, hits the target, and hits the graphite beam stop. Top
image: B.W. Jacobs Bottom image: R. Ladbury et al., IEEE T. Nucl. Sci. 51, 3664
(2004) with permission ................................................................................................ 166

Figure 5.5 The (n,m) CNT labeling system. T is the tube axis. Ch is the chirality vector
and describes how to roll up the graphene sheet. a] and a; are the unit vectors of

xvii

 

graphene. (n,0) and (n,n) are special chirality vectors denoting zig-zag and armchair
CNTs, respectively. The shaded box denotes the unit cell of the nanotube .................. 170

Figure 5.6 Types of CNTS. There are three main types of CNTS. zig-zag. armchair, and
the rest, chiral nanotubes ............................................................................................. 171

Figure 5.7 HRTEM images of SWCNTS and MWCNTS. (a) A single SWCNT can be
seen in the middle of the image rtmning nearly vertically. The single wall structure is
easily seen. Disordered carbon is also seen on the surface of the nanotube. (b) The

nested MWCNT structure is seen in this ﬁgure where each line indicates a wall. Images:
B.W. Jacobs ................................................................................................................ 172

Figure 5.8 HTREM image of carbon onions. (a) A single round carbon onion is shown
in this ﬁgure. The inset shows a model of a C60 molecule, which is the basis of carbon
onions. (b) These carbon onions formed polygonal sides with some faceting on the outer
layers. Images: B.W. Jacobs ....................................................................................... 174

Figure 5.9 HRTEM image of nanocrystalline diamond with diameters of 5 to 10 nm.

Nanocrystalline diamond is used as a starting material for producing carbon onions.
Images: B.W. Jacobs ................................................................................................... 176

Figure 5.10 The carbon samples were placed in quartz tubes and put in a close packed
conﬁguration. These samples were irradiated simultaneously. The heavy ions exited into
air through a thin zirconium window and passed through the quartz tubes with an on
target beam energy of over 1 1 GeV, as shown in the left image. The samples packed in
the quartz tubes are shown on the right. Images: B.W. Jacobs ..................................... 178

Figure 5.11 The carbon and gallium nitride nanowire samples were placed on a silicon
wafer that was attached to a piece of particleboard with silicone adhesive. The
particleboard was then screwed onto the SEETF movable stage. The samples were held
in place by a small mica sheet that was attached to the silicon wafer with silicone
adhesive. Each sample could then be separately irradiated with a speciﬁc dose. The
heavy ions exited into air through a thin zirconium window and passed through the mica
sheets with an on target beam energy of over 9.7 GeV, as shown in the left image. The
samples that were placed on the silicon substrate are shown on the right. Images: B.W.
Jacobs .......................................................................................................................... 180

Figure 5.12 The electronic characteristics of a GaNFET in radiated were measured. (a)
The circuit was mounted in a ceramic package and mounted on particleboard and
attached to the SEETF moveable stage. (b) A close up of the circuit shows its wire
bonds. Images: B.W. Jacobs ...................................................................................... 181

Figure 5.13 TEM images of irradiated GaN nanowire. (a) A TEM image shows a plume

Of material, which may have resulted from heavy ion radiation. (b) An HRTEM image of
the plume showing its amorphous character. The F FTs of the indicated areas Show the

xviii

 

ring structure associated with amorphous and polycrystalline material. Image: B.W.
Jacobs .......................................................................................................................... 184

Figure 5.14 TEM images of an irradiated nanowire. (a) The TEM image shows a
nanowire with a rough outer surface. which may have resulted from heavy ion radiation.
(b) HRTEM image of the boxed area in (a). An amorphous plume is shown here. An
FFT taken from the boxed area shows a ring pattern associated with an amorphous and
polycrystalline region. (c) HRTEM image taken from the boxed area in (a). The zinc-
blende/wurtzite coherent interface is shown here. No damage is observed in this picture
and FFTS do not Show signiﬁcant ring patterns. Images: B.W. Jacobs ......................... 185

Figure 5.15 Multiple VDS sweep of a GaNFET in radiation. V05 was kept constant at
9V. I-V Characteristics: V.M. Ayres et al. Diam. Relat. Mater. 15, 1117-1121 (2006)
.................................................................................................................................... 186

Figure 5.16 Raman spectra of irradiated SWCNTS. (a) Before and after krypton-78
irradiation. (b) Before and aﬁer calcium-48 irradiation. After all irradiation the SWCNTS
indicated no observable damage. Spectra: B.W. Jacobs ............................................... 188

Figure 5.17 SEM and HRTEM images of single walled CNTS before and after
irradiation. (a) SEM before irradiation. (b) HRTEM before irradiation. (c) SEM after
1,000Gy krypton-78 irradiation. (d) HRTEM after l,OOOGy krypton-78 irradiation. (e)
SEM after 10,0000y krypton-78 irradiation. (I) HRTEM after 10,000Gy krypton-78
irradiation. Images: B.W. Jacobs ................................................................................ 189

Figure 5.18 HRTEM image of SWCNT after 10,000Gy krypton-78 irradiation. This
image shows carbon-60 packed into a SWCNT, a peapod, possibly due to heavy ion
irradiation. Image: B.W. Jacobs .................................................................................. 190

Figure 5.19 Raman spectra of MWCNTS before and after krypton-78 radiation. The D
and G-peaks do not change signiﬁcantly after each radiation dose compared to the un-
irradiated sample. This indicates that the MWCNTS held up very well in the krypton-78
radiation and retained their multi walled structure. Spectra: B.W. Jacobs .................... 191

Figure 5.20 SEM and HRTEM images of multi walled CNTs before and after
irradiation. (a) SEM before irradiation. (b) HRTEM before irradiation. (c) SEM after
1,000Gy krypton-78 irradiation. (d) HRTEM after l,OOOGy krypton-78 irradiation. (e)
SEM after 10,0000y krypton-78 irradiation. (t) HRTEM after 10,000Gy krypton-78
irradiation. Images: B.W. Jacobs ................................................................................ 192

Figure 5.21 Raman spectra of untreated and treated carbon onions before and aﬁer
calcium-48 irradiation. (a) Untreated carbon onions. (b) Treated carbon onions. Spectra:
B.W. Jacobs ................................................................................................................ 193

Figure 5.22 HRTEM images of carbon onions before and after calcium-48 irradiation.
(a) and (b) compares the before and after irradiation of the untreated carbon onions. (c)

xix

and ((1) compares the before and after irradiation of the carbon onions with oxidation
treatment. The arrows in (b) and ((1) denote modiﬁcations due to radiation that took the
form of onion coalescence. Images: B.W. Jacobs ........................................................ 194

Figure 5.23 HRTEM images of carbon onions after calcium-48 irradiation. (a)
Oxidation treated onions. The left most arrows point out outer layer discontinuities and
the right points out onion coalescence. The inset also shows onion coalescence. (b)
Untreated carbon onions. The arrow points out onion coalescence. The inset also shows
onions coalescence. Both images indicate that that the overall onion structure was
maintained. Onion structure modiﬁcations are consistent for both treated and untreated
samples. Images: B.W. Jacobs .................................................................................... 195

Figure 5.24 HRTEM images of untreated carbon onions grown at temperatures (a)
1700°C (b) 2000°C and (c) 2300°C Show the evolution from spherical to polyhedral
onions. The increased onion coalescence and outer layer discontinuities can be seen in
(c). Images: B.W. Jacobs ............................................................................................ 196

Chapter 6

Figure 6.1 Bruker AXS high resolution XRD. (a) The X-ray source originates from the
right. It hits the sample and is detected on the left. The detector sweeps in a
counterclockwise direction detected reﬂected X-rays that meet the Bragg condition. (b)
The X-ray detector. A reﬂected beam from the sample is detected. (c) The X-ray source.
X-rays travel from the source to the sample. Images: B.W. Jacobs ............................. 205

Figure 6.2 Diagram of BEEM operation. The left image shows the physical setup of a
BEEM. The tip interacts with the base metal surface passing electrons through gap
between the tip and surface. 1c is the current coming out of the semiconductor collector,
I. is the current from the base metal resulting from electrons that were scattered in the
metal and did not have enough energy to surmount the barrier, plus the collector current.
The left image shows the energy diagram indicating each barrier. Electrons originate in
the metal STM tip, tunnel through the vacuum gap and pass through the metal and into
the semiconductor collector. E- is the Fermi level of each material, and eVb is the height
of the metal-semiconductor Schottky barrier. The ﬁgure is not to scale. The vacuum gap
is usually on the order of a few nanometers, and the metal base thickness is usually
around IOnm. Image: B.W. Jacobs ............................................................................. 207

Figure 6.3 Basic method of XPS sample spectrum acquisition. The X-rays are aimed at
the sample surface at a known angle and interact with the surface inducing photo-emitted
electrons. The photo-emitted electrons are collected and analyzed to determine their
energy. A spectrum of the photo-emitted electrons is then created indicating what
materials are on the sample surface, and the electronic state they are in ....................... 210

XX

 

II—v-Ir

1. Introduction

Since the mid 20th century nanotechnology has emerged as a highly
multidisciplinary research initiative spanning nearly all branches of science. The science
of nanotechnology took off with the discovery of the C60 molecule by Smalley, Kroto and
Curl in the 1980’s; a discovery that won the three scientists the 1996 Nobel Prize in
Chemistry. The new molecule was named Buckminsterfullerene, or buckyball for Short.
after the architect Richard Buckminster Fuller who popularized the geodesic dome, which
closely resembles the C60 shape. These new molecules sparked a revolution in science
that is still on the cutting-edge of research today, 20 years later. AS a result, novel ideas
and concepts are being developed across the globe that will ﬁnd practical use in the next
several decades.

Spanning many disciplines, nanotechnology has transformed science and
engineering with an eminent inﬂuence in the electronics and materials industries. These
industries in particular are on the cusp of a fundamental paradigm shift that will establish
new ways of thinking about and solving the next generation problems in device design
and materials engineering.

The search for new materials and systems with properties that meet or exceed the
current state-of-the-art has gained signiﬁcant momentum with the advent of the
discoveries within the nanotechnology ﬁeld. Many unexpected characteristics are
continually discovered in these novel structures while working to uncover and understand
their properties. There are several promising materials in particular that are spurring

research for new applications. These materials, or nanomaterials. possess many

interesting properties due to their small Size. The resulting properties can only be

 

 

described using quantum mechanics. These properties lead to amazing new and
unexpected characteristics, and exploiting these properties will drive innovation for many
years.

As stated in the candidate’s original thesis proposal, investigations of fundamental
heavy ion radiation interactions with nanomaterials will be presented. In the ﬁrst
radiation experiments, carbon nanotubes were irradiated and their structural
modiﬁcations, if any, were analyzed with several techniques. The results were very
promising and additional experiments were planned and implemented. In the subsequent
radiation experiments, gallium nitride (GaN) nanowires were also included. The GaN
nanowire material performance, as well as devices that incorporated GaN nanowires as
active elements, were investigated.

While analyzing these GaN nanowires used in the heavy ion radiation
experiments, a previously unreported nanowire structure was discovered by the
candidate. The details of this novel structure warranted further study, and a full
investigation was initiated. A signiﬁcant portion of this thesis presents results of this
study obtained using several cutting-edge instruments and techniques.

Heavy ion radiation research involving reduced dimensionalities is also in its
infancy with only a handful of groups contributing results at this time. This is an area of
potentially national signiﬁcance. There are several issues regarding nanomaterial and
nanodevice heavy ion radiation experiment procedures that should be addressed and
improved upon. Therefore, important problems, as well as recommendations for

improvement, will be brieﬂy discussed.

 

Speciﬁcs on how each investigated nanomaterial used in heavy ion radiation
experiments was grown and characterized, as well as how devices that integrate
nanomaterials were fabricated, are presented. The tools needed to characterize the
nanomaterials and fabricate devices will also be discussed. Without these highly
sophisticated tools, methods, and the ability to operate them and interpret results
correctly, this thesis research would not have been possible. The methods are valuable in

themselves, and are aimed at helping provide a baseline for future work in this area.

1.1 Radiation Resilience in Reduced Dimensionality Systems

Since the dawn of satellite communications and interplanetary space exploration.
organizations from around the world have sent sophisticated electronic systems into
space with increasing regularity. Though highly shielded and equipped with radiation
hardened electronics and high redundancy, these systems are still prone to radiation
induced upset. For satellites in near Earth orbit, the magnetic ﬁeld induced by the Earth‘s
core provides some protection. However, in years of increased solar activity these near
Earth satellites are affected by radiation produced by the Sun, and instances of complete
satellite failure due to solar radiation are not uncommon. Increased shielding and
redundancy are not long-term solutions to this problem. Custom radiation hardened
electronics and redundant systems are expensive to develop and implement. Shielding
adds a substantial amount of weight, which translates to expensive added fuel costs, and
adding increased shielding to new smaller space microprobe designs is increasingly
diﬂicuh;

In order to reduce shielding requirements and the need for highly complex
redundant systems, electronics should be inherently radiation resilient in their individual
device critical components. With current trends in silicon transistor technology, where
gate thickness and conducting channel length are getting ever smaller, already sensitive
areas are becoming more susceptible to radiation induced upset. The gate thickness and
channel length reduction also leads to lower operating voltages, which ampliﬁes device
sensitivity in radiation. As length scales are reduced by one order of magnitude, the
frequency of radiation induced upset, and the magnitude at which damage is incurred,

increases.

 

Novel nanomaterials, on scales two orders of magnitude smaller than current
state-of-the-art feature sizes of Silicon based devices, may be inherently radiation
resilient. These radiation resilient properties may be inherent to their structure
characteristics. Recent investigations suggesting that radiation interactions and damage
propagation mechanisms in nanomaterials differ from bulk materials in ways that result
in enhanced radiation resilience"2‘3.

Promising nanomaterials, which can be defined as a material with at least one
reduced dimensionality component, are ﬁnding use in a variety of applications. They
have been recognized for their excellent electrical, optical, and mechanical properties.
Nanomaterials utilized in devices can enhance system performance due to their reduced
dimensionality components. This enhanced performance is an important consideration in
ﬁnding alternative technologies to the current state-of-the-art. These systems must
outperform current technologies in order to be viable in future systems. Recently
electronic devices have been shown to be faster and operate at lower voltages“, chemical
sensors are much more sensitive to speciﬁc speciess, and optical devices are more
efﬁcient6 compared with current state-of—the-art technologies. Therefore, current silicon
based technologies will inevitably be augmented with or replaced by these emerging
devices that incorporate nanomaterials as key device components. With enhanced device
performance and inherent radiation resilience, nanomaterials, and their use in new
electronic, optical. and mechanical devices, are excellent candidates for next generation
space applications.

Reduced dimensionality systems will inevitably become the mainstream in the

electronics industry. If these systems are proven to be inherently radiation resilient, the

 

ability to use off the shelf electronic parts for new space based systems will signiﬁcantly
reduce the cost of system implementation. Shielding and redundancy could be reduced
and the need for customized radiation hardened electronics will be mitigated.

Several types of nanomaterials and nanodevices have been explored in this
research initiative. Speciﬁcally carbon nanotubes, carbon onions, gallium nitride
nanowires, and gallium nitride nanowire transistors have been tested in heavy ion
irradiation. The results of each nanomaterial and nanodevice performance, when
irradiated with speciﬁc species and doses of highly energetic heavy ions, will be
presented. Discussion of results will follow with possible explanations for these results

will be given.

 

1.2 Carbon Nanotubes and Carbon Onions

Carbon nanotubes are a widely researched nanomaterial with many interesting
properties. Carbon nanotubes are essentially a single sheet of graphene that self-
assemble into a seamless cylinder around 1 nm in diameter with lengths reaching several
millimeters. Ofﬁcially discovered in 19917, nanotubes have drawn an enormous amount
of attention in scientiﬁc research. Due to their small size and unique structure, nearly all
properties of carbon nanotubes, including their electrical, chemical, Optical, and
mechanical properties, have exhibited interesting, and many times very unexpected,
properties that larger, bulk materials simply do not possess.

Carbon nanotubes are excellent candidates for harsh environments due to their
robust structure and wide variety of possible applications. Possibly the most interesting
property important for radiation resiliency, is their ability to self-heal after heavy ion
damage. Carbon nanotubes were irradiated with three heavy ion beams, krypton-86,
krypton-78, and calcium-48. The candidate helped design and. implement these
experiments, and analyzed irradiated nanomaterial.

Carbon onions are similar to carbon nanotubes, but instead Of a seamless cylinder
of graphene, they are nested seamless spheres of graphene. Their name comes from the
similarity in structure of an onion of the vegetable variety, when cut in half, the
concentric onion layers can easily be seen. Similarly, if a carbon onion could be cut in
half, the concentric layers of graphene could easily be identiﬁed.

The round structure of carbon onions gives them the ability to provide a low
friction interface between two surfaces, and they are small enough to ﬁll cracks in those

surfaces. Carbon onions were therefore tested for viability as a solid lubricant. It has

 

 

been shown that they work well as a lubricant in vacuum, and with a similar structure as
carbon nanotubes, they may have the ability to self-heal after heavy ion radiation

damage. Carbon onions were irradiated with calcium-48.

 

1.3 Multiphase Gallium Nitride Nanowires

A long solid structure comprised of a semiconducting or metallic material,
nanowires have recently become a widely researched structure for electronic, mechanical.
and sensor devices. Gallium nitride nanowires in particular are promising due to their
versatility in devices. Gallium nitride (GaN) has a wide, direct band gap, and is
inherently n-type, making it viable in many optical and electronic applications. The GaN
nanowires investigated in this thesis research were found to have an interesting structure.
Grown via a novel catalyst-free technique, the candidate identiﬁed a multiphase nanowire
that incorporated the zinc-blende and wurtzite crystal structures simultaneously. The
multiphase structure spanned the entire length of the nanowire, which can grow to 1003
of microns in length. He identiﬁed possible nanowire nucleation sites responsible for the
multiphase nanowire growth. The candidate has fabricated and tested many devices
using the multiphase nanowire as the active element, and the ﬁrst characterization as an
electronic material and as an electronic element within a nanodevice will be presented.

Structural and electronic investigations were done with several instruments
specially designed for characterizing nanosystems. Using these instruments several new
and exciting results were obtained that would otherwise not have been possible using
traditional characterization tools. High-resolution transmission electron microscopy,
focused ion beam techniques, scanning probe microscopy, and a nanomanipulation
system integrated with a low noise electronic characterization system were crucial in
discovering and understanding the new properties these multiphase GaN nanowires.

GaN nanowires were also irradiated with a krypton-78 beam. Their subsequent

material modiﬁcations, as a result of heavy ion radiation, were investigated by the

v‘ -—v"- '—

 

candidate and will be presented. GaN nanowire based circuits were fabricated by the

candidate and also tested in real time in radiation.

10

 

1.4 Multidisciplinary Research Approach

Research and project development in nanotechnology requires a comprehensive
understanding of a variety of disciplines as well as a highly interdisciplinary team
working closely together in order for complex research initiatives to be successful. This
multidisciplinary approach is quickly becoming the status quo in research labs
investigating nanotechnology issues. Interdepartmental and inter-institutional national, as
well as international, collaborations are becoming more common in nanotechnology
research efforts, as was the case with this thesis research, combining expertise from a
several scientiﬁc backgrounds.

In this thesis several different scientiﬁc disciplines are discussed including
electrical engineering, materials science, chemistry, mechanical engineering, high-energy
particle physics and quantum mechanics. Collaboration with scientists and engineers
from these disciplines was required to accurately understand and describe material and
device phenomena occurring in these new systems. This highly collaborative expertise
was essential in the success of this research. These partnerships included inter-
departmental collaboration with Chemical Engineering and Material Science and the
National Superconducting Cyclotron Laboratory at Michigan State University, inter-
institutional academic collaboration with Howard University and Tokyo Institute of
Technology, government collaboration with NASA Goddard Space Flight Center and Jet
Propulsion Laboratory, and industry collaboration with Keithley Instruments and Zyvex
Corporation. The candidate was essential in the success of this group effort. He learned
a great deal from these collaborations through direct interactions with members as well as

acting as a liaison with all collaborators.

lI

 

1.5 Summary of Later Chapters

In the second chapter several current analysis techniques used in characterizing
nanomaterials, along with devices and systems that integrate nanomaterials, will be
discussed. In the third chapter implications of size restrictions on speciﬁc materials will
be introduced to explain how reduced dimensionalities affect certain material properties.
In the fourth chapter a comprehensive analysis and discussion of a recently discovered
multiphase gallium nitride nanowire homostructure will be presented. This includes
discussions on the growth, structural and electronic characteristics pertaining to
experimental results. In the ﬁfth chapter, an overview of the heavy ion radiation
experiments at the National Superconducting Cyclotron Laboratory (NSCL) will be
discussed. Operation of the NSCL as well as the Single Event Effects Testing Facility,
where the experiments were conducted, will be summarized so that basic knowledge of
each facility is acquired. Results of GaN nanowire, carbon nanotube and carbon onion as
well as GaN-based electronic devices in heavy ion irradiation will be presented. Chapters

6 and 7 will contain future work and a conclusion, respectively.

12

 

 

References

1.

R. Leon, S. Marcinkevicius, J. Siegert, B. Cechavicius, B. Magness, W. Taylor, C.
Lobo, “Effects of proton irradiation on luminescence emission and carrier dynamics
of self-assembled III-V quantum dots,” IEEE Trans. Nucl. Sci., 49, 2844-2851 (2002)

P. G. Piva, R. D. Goldberg, I. V. Mitchell, D. Labrie, R. Leon, S. Charbonneau, Z. R.
Wasilewski, S. F afard, “Enhanced degradation resistance of quantum dot lasers to
radiation damage,” Appl. Phys. Lett., 77, 624-626 (2000)

M. P. Petkov, L. D. Bell, H. A. Atwater, “High total dose tolerance of prototype
silicon nanocrystal non-volatile memory cells,” IEEE Trans. Nucl. Sci., 51, 3822-
3826 (2004)

S. J. Wind, J. Appenzeller, R. Martel, V. Derycke, P. Avouris, “Vertical scaling of
carbon nanotube ﬁeld-effect transistors using top gate electrodes,” Appl. Phys. Lett.,
80, 3817-3819 (2002)

F. Patolsky, C. M. Lieber, “Nanowire nanosensors,” Mater. Today, 8, 20-28 (2005)
F. Qian, S. Gradeak, Y. Li, C-Y. Wen, C. M. Lieber, “Core/Multishell Nanowire
Heterostructures as Multicolor, High-Efﬁciency Light-Emitting Diodes,” Nano Lett.,
5, 2287-2281 (2005)

S. Iijima, “Helical microtubules of graphitic carbon,” Nature, 354, 56-58 (1991)

13

 

 

2. Instruments and Methods

With the increasing complexity and unexpected properties of nanomaterials and
nanosystems, a wide variety of instruments and methods are required to understand the
fundamental characteristics of a given sample. Recently, in response to new instrument
requirements brought about by the need to characterize novel materials and devices,
specialized tools that can measure, probe, manipulate, and create on the nanoscale have
been invented and many existing tools have been further customized and improved upon.
Due to the wide variety of materials and systems investigated in this research, the use of
many different instruments was required. During the development of this thesis research
Special instrumentation at several different sites was used. A description of the
instruments and methods used, and the role they played in the candidate’s thesis research,
will provide necessary background information needed to understand later research
results and discussions.

There are many instruments used for nanomaterial analysis. Several key
instruments and methods used in this research are discussed here. Using these
instruments, material structure, surface topology, elemental composition, etc. can be
analyzed. With this fundamental information as a starting point in the design process,
new devices and systems that utilize these nanomaterials can be successfully
implemented in real world applications.

Many of the instruments discussed were located in different facilities and labs at
Michigan State University. These facilities included the W. M. Keck Microfabrication

Facility, the Center for Advanced Microscopy, the National Superconducting Cyclotron

Laboratory, and the Electronic and Biological Nanostructures Laboratory. Additional

14

 

instrumentation was located at other institutions including the Electron Microbeam
Analysis Laboratory at the University of Michigan, the Tokura-Hirata Lab at Tokyo
Institute of Technology, Keithley Instruments in Cleveland, Ohio, and Zyvex Instruments

in Richardson, Texas.

15

 

 

 

2.1 Transmission Electron Microscopy

The transmission electron microscope (TEM) is a highly versatile tool used for
materials analysis. It is virtually indispensable for material identiﬁcation, structure and
defect analysis, material chemistry, and a host of other applications. This instrument has
improved a great deal over the past few decades and has become a primary tool for
analyzing very small structures. The TEM is one of the most important tools used in this
thesis research. All the materials discussed have been analyzed in some way in the TEM.
The TEM was able to resolve intricate features, leading to an enhanced understanding of
the samples analyzed.

The TEM, as the name implies, uses electrons for image creation and sample
analysis. Electrons can be used for imaging, because like all matter, they possess wave-
particle duality. This means that electrons act as waves and particles simultaneously.
The wavelength of an electron is inversely proportional to its momentum, in accord with

the de Broglie hypothesis',

In this relationship A is the particle wavelength, h is Planck’s constant, and p is the
particle momentum. Since the resolution of an imaging system depends on the
wavelength of the incident beam, be it photons, electrons, neutrons. etc., shorter
wavelengths lead to higher resolutionz. Therefore, if the electron wavelength decreases,
the momentum of the electron increases and the resolution increases accordingly.
Electron wavelengths can be less than .037 A for acceleration voltages of 100 kV and

abOVe, much shorter than the range of wavelengths for visible light, which is around 400-

700 nm.

16

 

v_- - .- .

The electron is also a charged particle. Therefore, electrons can interact with
electromagnetic focusing ﬁelds in the TEM and with the sample, which also contains
charged particles in the form of atomic nuclei and electron clouds surrounding the nuclei.
These atomic interactions allow for speciﬁc sample material information to be obtained.

Electrons in the TEM are produced using one of three sources. A very common
source is the tungsten ﬁlament used for thermionic emission. The ﬁlament is heated by
passing a large current, which in turn produces electrons with enough energy to overcome
the work function of tungsten. The work function is deﬁned as the amount of energy
needed for an electron to escape from a given material. The work function of tungsten is
around 4.5 eV. By applying a high positive voltage to an anode near the ﬁlament, the
electrons are attracted to the anode and pass through a small aperture. The high voltage
on the anode is also responsible for the required electron acceleration. Tungsten
ﬁlaments, however, produce a rather high distribution of electron energies, called the
electron energy spread. This spread can cause electrons with different energies to be
focused slightly different. This is called chromatic aberration and reduces resolution in
the TEM. Tungsten ﬁlaments are quite inexpensive and last about 100 hours. A second
method used for electron generation is the lanthanum hexaboride (LaB6) tip. Similar to
the tungsten ﬁlament in operation, it also uses thermionic emission, but it is brighter.
meaning it emits more electrons due to its lower work function, 2.5 eV. It lasts roughly
10-15 times longer than the tungsten ﬁlament and has a smaller electron energy spread,
but it is more expensive. A third way to produce electrons, and wholly different from
thermionic emission, is to use a ﬁeld-emission source. A ﬁeld-emission source is usually

made of tungsten that has a very sharp tip. When an electric ﬁeld is applied, it is

17

III

concentrated at the sharp point of the ﬁeld-emission tip. In response to this concentrated
electric ﬁeld, electrons at the tip of the tungsten source tunnel out of the material and are
accelerated to the desired energy by the anode. Field-emission sources have become very
popular, since they have a very low electron energy spread, are very bright, and provide
excellent high-resolution images. They are, however, very expensive.

After the electrons have been accelerated through the anode, they are focused
using a series of electromagnetic lenses and apertures. The physical location of these
lenses and apertures are shown in ﬁgure 2.1. An electromagnetic lens essentially consists
of a copper wire winding with a hole in the center that surrounds the electron beam.
When a current is passed through the wire, a magnetic ﬁeld is produced that can interact
with and manipulate the electron beam. Apertures are used to block the transmission of
unwanted electrons from contributing to the ﬁnal image. The system of focusing
electrons in electromagnetic lenses is nearly identical in concept to focusing photons in a
glass optical lens. Using an electromagnetic lens, however, one has the ability to change
the magnetic ﬁeld strength by altering the current passing through the copper windings.
In this way the beam can be manipulated, resulting in, for instance, a change in the

magniﬁcation, without having to switch to a different lens. as in optical microscopy.

18

 

High Voltage

Apertures ' ~
: _ r.
Controls and f ,l, \

. . *. .. ndenser Lens
VIewmg Computers . . ER

\ Sample Holder

Objective Lens

\ Intermediate
- Lens

 

Figure 2.1. JEOL 2200FS ﬁeld-emission transmission electron microscope at the Center for Advanced
Microscopy at Michigan State University. (a) This particular TEM is in part controlled by computer. The
computers used for imaging are shown on the right. (b) A TEM sample holder. The arrow indicates where
a TEM grid is placed. Note its size. This restricts the size ofthe sample that can be placed in the TEM. (c)
Close up of the TEM. The electron gun is at the top of the TEM and electrons are accelerated down the
beam column and focused by the electromagnetic lenses. The sample holder indicates where the sample is
located in the beam column. Images: B.W. Jacobs

In the TEM electromagnetic lens system the electron beam is ﬁrst focused using a
condenser lens so that a parallel beam is created. A parallel beam maximizes coherence
and sample contrast. The condenser aperture is used to block stray electrons. The beam
must also be circular so that astigmatism effects are mitigated. A non—circular beam will
be focused differently since the magnetic ﬁeld is stronger at the edge of the lens. After
the beam passes through the sample, the objective lens is used to focus the image. The
Objective aperture and selected area diffraction aperture below the sample can further

restrict the electron beam. The objective aperture is used to enhance the image contrast

19

and select certain diffracted beams used in advanced imaging techniques. The selected
area diffraction aperture is used to select a certain area for diffraction studies and will be
discussed later. The ﬁnal lens normally used in image formation is the intermediate lens.
which determines the magniﬁcation.

The electromagnetic lenses used in the TEM, however, are far from ideal. Unlike
optical lenses that can be made nearly perfect, maximizing achievable resolution using
the visible light spectrum, electromagnetic lenses fall far short of achieving this
maximum possible resolution using electrons. A primary issue in electromagnetic lenses
is the non-uniformity of the magnetic ﬁeld around the lens. This causes electrons at
different areas in the beam to be focused differently. The further out an electron is from
the center of the beam, the stronger it is deﬂected. This phenomenon is called spherical
aberration. This is a critical problem limiting the resolution of the modern TEM,
restricting the resolution to around .5 A for acceleration voltages of 200 keV and an
electron wavelength of .0251 A. Although atomic scale resolution is possible, the
maximum theoretical resolution using 200 keV electrons would easily be great enough to
PiCk out ﬁne details of atomic structure. The improvement of electromagnetic lenses and
Spherical aberration compensation has made signiﬁcant progress, and new TEMs are able
to resolve ﬁner and ﬁner detail in samples.

Preparing samples for imaging in the TEM poses several Obstacles. The sample
mUSt meet several criteria before images can be taken. First and foremost it must be thin
enough to be transparent to the TEM electron beam, and the thinner the sample is, the
beﬁer. The sample should be less than 100 nm thick depending on the transparency of

the material. The transparency depends on the sample material, i.e. its density, atomic

20

number, etc. The sample must also be small, since the TEM sample holder is only
around 3 mm in diameter. The sample must be able to hold up in the high-energy
electron beam so that imaging can take place. Samples will inevitably be damaged. so
the electron beam Should stay as defocused, or spread out, as much as possible. These
criteria signiﬁcantly restrict the type of samples that can be analyzed, and it complicates
sample preparation.

Several techniques exist for preparing samples, including ion milling, microtome
cross-sections, and chemical etching, to name a few. When imaging many nanomaterials
in the TEM, however, these preparation techniques may not be needed because the
samples are already very small. Usually in preparing a nanomaterial, such as nanowires
or nanotubes, for imaging, an ethanol suspension containing the sample is placed on a
TEM sample support and imaged. The sample support usually consists of a round copper
or nickel grid with a ﬁne mesh of square holes several hundred microns wide. A very
thin carbon ﬁlm with many holes only a few microns wide, which provides the actual
Samme support, spans the copper mesh. Ideally the sample material will span a hole in
the carbon ﬁlm so that maximum beam current will pass through the sample resulting in
brighter, higher resolution images. This is the primary nanomaterial sample preparation
teChnique used in this thesis research. Another primary sample preparation technique
u363d in this research is nanowire cross-sectioning using a focused ion beam. This
teeI'lnique will be discussed in detail later the chapter.

In forming a sample image, electrons interact with atomic nuclei and electrons
Surrounding the nuclei resulting in elastic or inelastic scattering. Elastic scattering is the

priThary scattering process in the TEMZ. In order to understand how images are formed.

21

the wave nature of electrons, from the wave-particle duality concept, must be invoked.
As the electron beam passes through a sample and scatters elastically, usually at very
small angles, the resulting image will typically contain areas of light and dark contrast
corresponding to the speciﬁc properties of that sample. There are two primary types of
contrast that occur when forming an image, amplitude contrast and phase contrast. Phase
contrast is important in high-resolution TEM and will be discussed later in the chapter.
Amplitude contrast can be further divided into two groups, mass-thickness and diffraction
contrast. Mass-thickness contrast primarily arises due to thickness variations in the
sample. When the sample is thicker, there is a higher likelihood of multiple scattering
events that cause electrons to signiﬁcantly deviate from the beam axis and not contribute
to the ﬁnal image. Different elements present within the sample will also affect the
contrast, since the electron beam will scatter differently off different elements. The
amount of electron scatter depends on the atomic number of the speciﬁc element in the
Sample. Scattered electrons that are incoherent form mass-thickness contrast. In other
Words, there is no constructive or destructive interference in forming the image. In
diffraction contrast, however, the sample is tilted so that two strong beams are obtained.
one diffracted beam and the transmission beam, called a two-beam condition. Scattered
eleCtrons are coherent and as a result constructively and destructively interfere.
Diffraction contrast is important for defect analysis and determining thickness variations
on the sample. Diffraction contrast is similar to phase-contrast imaging used for hi gh-

r esOlution TEM, which is discussed next.

22

2.1.1 High-Resolution Transmission Electron Microscopy

High-resolution TEM (HRTEM) is a very useful tool when analyzing
nanomaterials and nanostructures. Atomic scale resolution is obtainable if the sample
material is thin enough. The local structure of materials can then be analyzed in detail.
The JEOL 2200FS at the Center for Advanced Microscopy at Michigan State University
was the primary HRTEM used in this research.

Usually HRTEM is thought to be synonymous with phase-contrast images.
Actually, HRTEM is possible as a result of phase-contrast imaging. Phase-contrast
images are formed when multiple coherent beams are allowed to constructively and
destructively interfere. This is similar to diffraction contrast when two beams are
allowed to interfere, but in phase-contrast images, many more beams interfere. Phase-
contrast does occur in all TEMS, but not all TEMS have the capability to resolve atomic
scale features.

This constructive and destructive interference of coherent beams is described by
the transfer function,

T(u) = 2A(u)sinx(u);
Where A(u) is the aperture function and ﬂu) is the phase-distortion functionz. The
transfer function is a function of spatial frequency, II, which is a reciprocal lattice vector,
and is related to I/x, where x is a distance. Reciprocal space is discussed in the next
SeCtion. The phase-distortion function, xtu), takes the form of a phase shift. If the phase-
distOl‘tion function is compared to the contrast transfer function, which describes the
periOdie nature, or bands, of contrast corresponding to areas of good transmission and

Zero transmission, the contrast transfer function has maximum transfer of contrast when

23

the phase-distortion function has multiple odd values of 221/2 and zero contrast when the
function has values of In. In other words when the value of sinxht) is equal to 1.7172, TM)
is maximized corresponding to bright contrast and when SIIIX(II) is equal to In, T(u) is
minimized corresponding to dark contrast. ﬂu) depends on the defocus of the beam, and
is a complicated function. In short, to achieve maximum resolution and contrast, an
optimum value for Xﬂl) and therefore an optimum value of defocus can be determined.
This is called the Scherzer defocus. The maximum resolution of a TEM is determined

using this value. It depends on the spherical aberration, Cs, and electron wavelength, A.

AfSCh = 4.21/C5}. .

SO it is important to know what this value is and to be able to obtain it, in order to
maximize resolution in the HRTEM.

The interpretation of HRTEM images, however, is not as straightforward as one
might expect. Complex interference patterns give rise to misleading images that are not
directly representative of the sample. Aberrations in the lenses also complicate images.
Therefore, many additional considerations must be taken into account when imaging with
HRTEM that are not issues with normal TEM imaging.

HRTEM has proven very useful in analyzing the nanomaterials used in this
reSearch. Since the materials studied in this research are very small, the HRTEM
Provided essential details of the structure of these materials. Many times certain material
Structures are too small to reliably produce informative diffraction patterns. Therefore
post-processed F F Ts of HRTEM images, further discussed in chapter 4, provide details of
the Sample material. Without the resolution capability of the HRTEM for material and

FFT analysis, much of this research would not have been possible.

24

2.1.2 Electron Diffraction

Diffraction is one of the most important and useful technique used in the TEM.
and is the basis of modern crystallography. Diffraction patterns are extremely useful in

determining. a material’s properties including its structure, orientation, lattice constants,

grain boundaries and so on.

Electrons in the TEM are able to form diffraction patterns due to their wave
nature. When interacting with a given crystalline material, electron waves scatter offof

speciﬁc atomic planes in the material, according to Bragg‘s law.

    
  

Reﬂected

incoming
Beam

Beam
. .......... . u... .......... .
dhkl
. ........... . .......... . ..................... . .......... . .......... ..

n11 = 2(1th sinH

Figu re 2.2. Illustration of Bragg’s law. An incoming wave will interact with the material and if the angle
Is correct it will satisfy the Bragg condition and strongly reﬂect at the same angle as the incident beam.

lmage: B.W. Jacobs

In Bragg‘s law n is any integer, usually equal to 1, It is the electron wavelength, cit,“ is the
interplanar spacing in the crystal lattice, and I9 is the Bragg angle. When a beam of
eleCtrons interacts with a crystalline material, and if the angle is correct, it will reﬂect
Stt‘Ongly off the atomic planes at the same angle as the incident beam. This strong

reﬂ ection results because at angles and interplanar spacing that satisfy Bragg‘s law.

25

constructive interference will occur. This correct combination of incident angle and
interplanar spacing is called a Bragg condition and is the basis of diffraction.

For a crystalline material, the diffraction pattern, a collection of diffracted beams
that meet the Bragg condition, will create a pattern of bright spots. With the exception of
the center spot, which is the transmitted spot, each bright spot corresponds to speciﬁc
atomic planes in the material. The location and relative brightness of each spot is a result
of the crystal properties of the sample material. The information contained in the
diffraction pattern represents the crystal structure in inverse space or reciprocal Space.
The spot pattern represents the atomic plane arrangement in real space. The distance
between diffraction spots is measured in inverse meters and represents the spacing
between the planes in real space. The diffraction pattern effectively represents the crystal
structure in a way that is easier to analyze and quantify than real space images. Using
diffraction techniques many properties of the crystal structure can be identiﬁed, such as
orientation relationships, crystal type, and defects. For further information on reciprocal
Space see references 2 and 3. Diffraction patterns are formed with an aperture called the
selected area diffraction aperture, which is used to select a certain area of interest in the
material. The area selected is then projected to the viewing screen in reciprocal space.

Fast Fourier transforms are another tool useful for determining a material’s
CryStallography. The information contained in a fast Fourier transform (F FT) is similar to
the information contained in a diffraction pattern, but a FFT is a post-processed software
technique. A diffraction pattern is purely physical, while a F F T is purely mathematical.
After a high-resolution, atomic scale image has been taken, a FFT can be used to simulate

a diffraction pattern, or determine the reciprocal space pattern of a real space image.

26

F FTs are excellent companions to diffraction patterns and augment diffraction data.

F FTs are also indispensable when analyzing very small crystalline areas. Diffraction
patterns are somewhat limited spatially, and sometimes diffraction Spots from unwanted
areas can appear. So if high quality, high-resolution TEM images can be Obtained, F FTs
do an excellent job in providing essential crystal data. The FFT is an excellent tool, but
real diffraction patterns are far superior in the information they provide, so whenever
possible, take a diffraction pattern.

Diffraction patterns and FF Ts were a very important tool in this research.
Without the use of diffraction patterns and FF Ts, the crystallography of the GaN
nanowires would have been extremely difﬁcult if not impossible to solve. The small
crystalline domains found in the nanowires were many times too small to use diffraction,

so FFTS were ultimately heavily relied upon, but were backed up indirectly using real

diffraction patterns.

2.] .3 Energy Dispersive X-Ray Spectroscopy

Energy dispersive x-ray spectroscopy (EDS) is a technique used in the TEM to
determine the elemental composition of a given sample. Using the same high-energy
elecu'ons it uses to create images, electrons in the TEM cause electrons in core atomic
0Fbitals of the sample material to be ejected. This causes an electron from a higher
encrgy to drop into the “hole” left by the ejected electron. Since energy must be
co“Served, the electron that drops into the hole must release energy. This energy is equal
to the difference between the ejected electron energy and the energy of the electron that

droDped in energy to ﬁll the hole. Energy is conserved through the emission of a photon.

27

The photon emitted will have a wavelength in the X-ray regime. The energy of the X-ray
is speciﬁc to each element in the periodic table. The X-ray is detected and a computer
plots a spectrum with peaks associated with each element present in the material. Using
this technique one can very quickly and easily obtain the elemental makeup of a given
material. EDS is generally used in conjunction with other techniques such as EELS to
obtain a great deal of information about a given material.

This method is particularly useful when the composition of a material is not
known. Instead of painstakingly solving diffraction patterns and comparing known
diffraction patterns for material identiﬁcation, EDS provides the user with valuable
elemental information almost instantly. It is capable of estimating the elemental
constituents of a material if the thickness and density is known. Small areas of only a

few nanometers of material can also be analyzed since the TEM electron beam can be

highly focused.

This technique has proved useful on several occasions. The stoichiometry of the

GaN nanowires was initially called into question and EDS was one technique used to
estimate the nanowire elemental constituents. Another occasion in particular was the
discovery of gallium oxide (GazO3) nanowires, when gallium nitride nanowires were
€Xpected. In analyzing these nanowires, the structure looked somewhat different than
What was usually observed for GaN nanowires. An EDS spectrum was taken, and in less
than five minutes the structure was determined to contain gallium and oxygen in a 2:3
ratio. Therefore, taking an EDS spectrum on a given sample to analyze its structure is

CIUick, easy, and usually highly beneﬁcial.

28

2.1.4 Electron Energy Loss Spectroscopy

Electron energy loss Spectroscopy (EELS) is a technique used to analyze the
electronic structure of a given material in the TEM. Electrons in the TEM beam will
undergo some inelastic scattering as a result of electron interactions with the sample
material. These electrons will exit the material with a lower energy that is speciﬁc to the
interaction within the sample. Electrons from the TEM beam can interact with the
sample material electrons in several ways, including inner shell ionizations, plasmon
excitations, phonon excitations, and energy band transitions. The plasmon excitations
and inner shell ionizations are particularly important and provide information about the
elemental composition, chemical bonding, valence and conduction band properties, and
so on. Since these interactions are speciﬁc to the elements present in the material,
quantitative information about the sample can be obtained using this technique.

The detection of electrons that are subjected to inelastic scattering in the sample
material is done with magnetic prism located at the bottom of the microscope or with an
in column omega ﬁlter, so named due to its close resemblance to the uppercase Greek
letter 9. Since electrons have a distribution of kinetic energies after passing through the

sample relative to the incident beam, the energies can be segregated. Electrons with
different kinetic energies can be therefore be detected with resolutions of around 1 eV.
The kinetic energy distribution of the scattered electrons is then plotted in a spectrum
With peaks corresponding to scattered electrons with the same kinetic energy. Elemental

mapping and EDS can be used in conjunction with EELS to obtain spatial mappings of

Speciﬁc elements within a sample.

29

An EELS spectrum can be difﬁcult to obtain, because TEM beam stability and
alignment can be ﬁnicky and thin samples are needed. However, with some patience and
a stable beam, EELS spectra can provide a wealth of information about a sample.

Usually the beam should be at the operating voltage for at least 8 hours for maximum

stability before EELS is to be done.

EELS has proven particularly useful in the analysis of carbon onions. EELS is
highly sensitive to the bond structure present in the carbon system and can easily
distinguish between sp2 and sp3 hybridizations. When combined with Raman
spectroscopy and XPS, both discussed later, EELS provided an excellent quantitative
analysis of the sp2 to sp3 ratio. EELS also provided an independent veriﬁcation of the

presence of gallium and nitrogen in the GaN nanowires, as well as the lack of impurities

such as oxygen.

30

2.2 Focused Ion Beam

A focused ion beam (FIB) system uses an ion beam, usually consisting of gallium
ions, to create sample images. Gallium ions are accelerated and focused in
electromagnetic lenses, just as in the scanning electron microscope (SEM), discussed
next, and can produce images similar to those created in the SEM. Imaging with ions,
however, can be very destructive to the sample. Therefore, a FIB is generally used for
sample manipulation and preparation, due to its ion beam sputtering capability. In the
sputtering process energetic Ga ions bombard the sample and eject atoms off the surface
into a gas phase. In this way, patterns on the sample surface can be made. However, due
to the inherent destructive nature of the ion beam, the dual beam FIB has been developed.
Coupled with an electron beam, i.e. an SEM, a dual beam FIB can use the ion beam for
sputtering and use the SEM for imaging. This minimizes damage imparted on the usually

already delicate sample. An example of a dual beam FIB system is shown in ﬁgure 2.3.

31

 

 

Omni cromanipulator

 

Figure 2.3. “FEI Quanta 200 3D dual beam FIB with an integrated Omni Micromanipulator System at the
Electron Microprobe Analysis Laboratory at University of Michigan. (a) The main components ofa dual
beam FIB are shown here. It is computer controlled. The electron gun is situated perpendicular to the
sample as in a normal SEM. The ion gun is tilted 52° relative to the electron gun. The Pt source needle is
next to the ion column. Manual stage control is on the front ofthe specimen chamber and is used in the
cross-section preparation process. The micromanipulator is situated on the backside of the instrument. (b)
A close up of the micromanipulator is shown here. The XYZ control is shown and consists of computer-
controlled screws that move the needle. Image: B.W. Jacobs

The primary use of the FIB in this research has been TEM sample preparation. It
was used to create cross-sections of GaN nanowires for HRTEM imaging. Using this
cross-section technique the crystallography within the nanowires could be determined.
Samples that have been prepared using the FIB have provided a wealth of information
about the nanowires internal structure. Details of this structure will be discussed at
length in chapter 4. Here the details of FIB sample preparation and implementation
developed by the candidate for cross-section nanowire analysis are discussed.

When preparing a nanowire cross-section sample for HRTEM analysis, a

dispersion of nanowires is ﬁrst placed on a small silicon wafer. A 100-200 nm layer of

32

 

gold is thermally evaporated over the nanowires and silicon to initially protect the
nanowires from ion beam damage in the FIB, ﬁgure 2.4(a). A layer of platinum is then
deposited through reactive gas/ion Pt deposition; a 4-6 pm thick layer of Pt is sufﬁcient.
This layer provides the necessary protection for the nanowire through the entire cross-
section preparation process. A few extra microns of Pt is added to the area directly over
the nanowire to mark its location, see ﬁgure 2.4(b). After the necessary Pt metal is
deposited, two trenches perpendicular to the nanowire and parallel to the Pt metal are
sputtered, ﬁgure 2.4(c). These trenches are around 20-25 pm wide and 10 um long on
one side and 5 pm long on the other. The depth is 10 pm on either side, and tapers
upward away from the deposited Pt. The sample is roughly milled at this time, to around
1 pm in thickness. After the trenches are complete, the sample is tilted and a “U-cut”
around the bottom and sides of the sample is done, ﬁgure 2.4(d). This U—cut should
penetrate through to the other side of the sample, and partially release the sample from
the substrate. This is a very important step. The U-cut must extend completely through
the base and sides of the sample. Small notches are left on each edge of the sample to
hold it in place. A needle attached to a micromanipulator comes down and touches the
top of the sample. The needle is superimposed in ﬁgure 2.4(d) to indicate where it would
be placed directly prior to extraction. The needle is then welded to the Pt protection layer
with additional Pt. The notches are sputtered away with the ion beam and the sample
should be free, attached only to the needle. When sample extraction takes place, there
should not be any material left attached to the substrate. The sample stage is then
lowered using the manual stage control on the front of the microscope, and the sample

extraction is complete. The sample can then be transferred to the TEM grid. A special

33

 

SEM sample holder is used so that the extracted cross-section sample can be transferred
from the substrate to the TEM grid without the need to evacuate the specimen chamber.
The sample Should be brought to the TEM grid with extreme care. It is connected only to
the needle at one delicate point, and any vibration or jarring could easily knock it loose
resulting in a lost sample, see ﬁgure 2.4(e-f). When the sample just touches the side of
the TEM grid it should be immediately welded and secured, ﬁgure 2.4(g). After welding
the needle is cut with the ion beam and the sample is left welded to the TEM grid. Fine
milling of the sample can now take place. The sample thickness must be thin enough to
be transparent to the TEM electron beam. Therefore it must be less than 100 nm thick. It
must be milled to the point where it seems that it cannot be milled further. It must be

milled as absolutely thin as possible, see ﬁgure 2.4(h). A completed sample is shown in

ﬁgure 2.4(i).

34

Nanowire
/

Silicon
"5 Substrate

Extracted

“U-cut"
Sample \

TEM Grid

Fine Milling ,
\ Nanowire

X-section
SIIICOI') (too small to see)

I <100 nm

    

Figure 2.4. The TEM cross-section preparation method using a FIB. (a) Nanowire sample. (b) Pt
deposition on top ofthe nanowire to protect it from ion beam damage. A small amount ofcxtra Pt is
deposited on top ofthe nanowire to mark its exact location. (c) Creation of the trenches on either side of
the Pt, perpendicular to the nanowire. (d) The U-cut. Two notches on either side are left to hold the
sample in place. The superimposed needle indicates where it should be welded on the cross-section. (e)
The extracted sample. (f) The extracted sample as it nears the TEM grid. (g) The sample is welded to the
TEM grid. (h) Fine milling ofthe cross-section. It should be thin enough to be transparent in a TEM
electron beam, less than 100 nm. (i) Ion beam image ofthe ﬁnal nanowire cross-section. Images: B.W.
Jacobs

   

After ﬁne milling is complete, the sample can be analyzed in the TEM. It is
important to thoroughly weld the cross-section to the TEM grid so that it can withstand
the process of taking it in and out of the TEM. Note the amount of Pt deposited on the
right side of the cross-section in ﬁgure 2.4(i). Therefore, extra time and care must be

taken when preparing these samples if quality high-resolution images are to be obtained.

35

The FIB allows for unique samples to be imaged in the TEM that would otherwise
be nearly impossible due to size and thickness constraints. The FIB has traditionally
been used in the semiconductor industry to image such structures as transistors and
heterostructures to determine thin ﬁlm thicknesses and to analyze interfaces and defects.
The dual beam system has added further versatility to the FIB, which has increased in
popularity due to the added SEM functionality and with the need to analyze new types of
materials. The FIB has provided a truly unique twist on how materials analysis in this
research has been conducted. It has helped uncover an interesting phenomenon in the
GaN nanowires, which might not have been completely understood without the use of the
FIB. The primary FIB used in this thesis research was a FEI Quanta 200 3D Dual Beam
at the Electron Microbeam Analysis Laboratory at University of Michigan. These results

will be discussed at length in chapter 4.

36

2.3 Scanning Electron Microscopy

Scanning electron microscopy is a very popular surface imaging technique. Using
electrons for imaging, it is capable of high-resolution, topographical views of a sample.
The electron beam in an SEM is produced via similar means as the TEM, by using a
tungsten ﬁlament, LaBb, or ﬁeld emission tip. An anode accelerates the emitted electrons
at a bias voltage between 1-50 kV depending on the instrument and user requirements.
Although an SEM, like a TEM, uses electrons to image a sample, that is where the
similarity ends.

After acceleration, electrons in an SEM are focused through a condenser lens and
condenser aperture and an objective lens and objective aperture, see ﬁgure 2.5. However,
unlike a TEM, where electrons pass through the sample and are diffracted, electrons are
either backseattered or generate secondary electrons from the sample. The
electromagnetic lenses used in the SEM are only used for incident beam manipulation,
i.e. focusing and scanning the beam. The lenses in the SEM do not directly contribute to
image formation in any way like the objective and intermediate lenses in the TEM.

The incident electron beam interacts with sample electrons and atomic nuclei,
where the size of the interaction volume depends on the sample density, on target energy
of the electron, and atomic number of the sample material. The material should be
sufﬁciently conducting or SEM imaging would be difﬁcult. An insulating material can
be imaged in the SEM if a thin layer of conducting material, such as 20 nm of gold, is
ﬁrst deposited on the sample surface. SEMS that utilize a ﬁeld emission gun. however,

can image insulating samples even at low acceleration voltages.

37

Computer
Control

Exchange Chamber

 

Figure 2.5. Hitachi S-470011 FESEM at the W. M. Keck Microfabrication Facility at Michigan State
University. (a) The SEM is controlled via software. On the right is the control computer. The left
highlights several components ofthe SEM. The electron gun is at the top and accelerates electrons
downward through a series of electromagnetic lenses and apertures. The sample is located at the base of
the instrument. When introducing a sample into the microscope it is ﬁrst placed in an exchange chamber
and then after rough pumping introduced into the specimen chamber shown in (b). (b) A close up image of
the instrument shows the specimen chamber where imaging occurs and the lower and upper secondary
electron detectors. Image: B.W. Jacobs

In forming an image, the SEM electron beam is rastered across the sample, and a
detector collects secondary or backseattered electrons emitted from the sample. This
detector consists ofa scintillator, which converts electrons into photons. The electrons
are attracted to the scintillator using a Faraday cage that is tuned to a certain voltage
potential and is also used to mitigate the scintillator/ incident beam interactions. After the
scintillator converts an electron to a photon, a photomultiplier ampliﬁes the signal and
converts the photon into an electrical signal, which is displayed on a CRT or digital
imaging device.

The two main types of electron scattering in the SEM, backseattered and
secondary, can be used to determine different properties in the sample. When viewing
the secondary electron image, areas of dark and bright contrast indicate edges or points
on the sample. This contrast variation is a result of the incident electron beam interaction

with the sample. If a sample is ﬂat, the teardrop shaped volume of interaction restricts

38

the number of secondary electrons emitted. However, if an edge or point is being
imaged, secondary electrons can escape the sample volume much easier since the volume
of interaction is smaller, see ﬁgure 2.6. Therefore, a larger number of secondary

electrons are collected, creating an area of bright contrast.

Electron Beam

   

Detector

Bright
Area

Dark
Area 97/ Sample Surface

‘i

Figure 2.6. Electron beam/sample interaction in the SEM. Electrons from the beam interact with the
sample in a teardrop shaped volume. Variations in the surface topology cause thin points and edges to
appear bright because electrons that enter this area can be scattered out more easily resulting in a higher
number of detected electrons from those areas, and therefore a brighter spot. Image: B.W. Jacobs

When Viewing the backseattered electron image, areas of dark and bright contrast
indicate different elements within the sample. A backscattered image is formed through
the different interaction characteristics of the incident electron beam with the elemental
constituents in the sample. For example, an electron will more effectively scatter off an
element with a high atomic number, such as gold, and appear as a bright area in the
image, as opposed to a lighter element, such as aluminum, which will appear darker

relative to gold.

39

 

Backseattered electrons can also be used to form diffraction patterns. For
crystalline or polycrystalline materials, electron backseatter diffraction (EBSD) can
determine crystal orientation, be used in defect analysis, crystalline phase determination
and so on. An EBSD image is formed by tilting a sample to a high angle, ~70° relative
to the electron beam. The beam will diffract off the sample with varying intensity
depending on the orientation of the crystal. The beam will diffract strongly when a Bragg
condition is met, as is the case with electron diffraction in the TEM, and weakly when the
condition is not met. EBSD can also be coupled with EDS. resulting in a powerful tool
for materials characterization.

The SEM has been very useful in characterizing materials and devices used in this
research. Located in the W. M. Keck Microfabrication Facility, the Hitachi S-47OOII
ﬁeld emission SEM was the primary SEM used in this research. It was used to analyze
features of fabricated GaN nanowire devices to assess fabrication quality. The SEM was
also essential in materials characterization, including identifying nanowire nucleation
sites, surface features on the sides of nanowires, and helped identify the triangular and
hexagonal cross-section shape of many nanowires. These images resulted in many
important ﬁndings and new information, making the SEM a very important tool in this

research.

40

2.4 Microfabrication Process

Microfabrication encompasses a wide range of techniques for creating devices
and systems on a small scale. Most of these techniques are geared toward the fabrication
of electronics, such as microprocessors and optical devices. However, in recent years the
microfabrication process has lent itself to the creation of a wide variety of mechanical
devices. These devices include, temperature, pressure and acceleration sensors, actuators
and inkjet nozzles for printers. The art of microfabrication is rooted deep in the
semiconductor industry and has become highly developed and precise process through
years of research and development. All devices were fabricated at the W. M. Keck
Microfabrication Facility at Michigan State University, which is a class 100/ 1000 clean

room facility.

2.4.1 Photolithography

Photolithography is by far the most common microfabrication process and is used
almost exclusively by chip manufacturing companies to fabricate intricate electronic
devices. These devices are usually created using silicon. Since the invention of the
integrated circuit in the early 1960’s, silicon has dominated as the material of choice for
nearly all such devices through the present era. Photolithography, along with integrated
circuits, has grown in complexity many times over. The integrated circuit, comprised of
many individual on/off switches (transistors) has shrunk ever smaller as the desire for
faster and faster devices increases with each passing year. Making transistors smaller

increases the speed at which it can change state, i.e. how fast it can turn from off to on

41

and vice versa. A detailed discussion on the operation of a transistor will be discussed in
chapter 5.

Photolithography uses light to expose speciﬁc areas of a special light sensitive
material, called photoresist, which is on the surface of a substrate, usually a silicon wafer.
These exposed areas then break down and become soluble in a developer and wash away
leaving the unexposed photoresist remaining on the surface. Using this method, very
intricate patterns can be made. Photolithography is done using a glass plate, called a
mask, with a pattern of clear windows and opaque areas consisting of the desired pattern,
see ﬁgure 2.7(b). This pattern is then transferred to the photoresist, exposing certain
areas with UV light, ﬁgure 2.7(a), through the clear window on the mask and blocking
certain areas with the opaque areass. This is basically the same process that major

semiconductor manufacturers use to produce highly complex integrated circuits.

 

  

Figure 2.7. AB-M mask aligner at the W. M. Keck
(a) The instrument consists ofa UV source that moves on rails to expose the photoresist on a wafer through
a mask. located in the center ofthe instrument. An optical microscope on the right is used to align other
masks used for multiple exposures for more complex features. (b) An example ofa glass/chrome mask
used in this research. Images: B.W. Jacobs

42

In the basic photolithography process used in this research, a bare silicon wafer is
used as the substrate, see ﬁgure 2.8. A layer of silicon dioxide is grown on the surface
and a layer of photoresist is spun on at 4000 RPM for maximum uniformin and
consistent thickness. The photoresist is exposed to UV light through a mask with
predeﬁned windows, transferring the mask pattern to the photoresist. The exposed
photoresist breaks down and can be removed with a developer. A metal can then be
deposited and the areas with remaining photoresist can be removed along the metal that

was on top of those areas leaving the desired pattern.

 

 

 

 

 

 

 

 

 

(a) 0’)
Si Wafer
Mgggggggg (d)
H

 

 

 

i=‘—="i
i=—==lr———i

Figure 2.8. Basic photolithography process. (a) A bare silicon wafer. (b) Silicon dioxide is grown on the
surface and is used as an insulating layer. (c) Photoresist is spun on the wafer and it is exposed to UV light
through a mask. ((1) The developer removes the exposed resist. (e) A metal layer is deposited. (I) The
remaining unexposed resist is taken off along with the metal layer that was on top of it, leaving the desired
pattern. Images: B.W. Jacobs

AS silicon based integrated circuits continue to shrink, however, photolithography
as well as silicon itself, is reaching a limit. The minimum feature sizes of patterns
created with photolithography are highly dependent on the wavelength of incident light

and the numerical aperture according to:

43

F=kA-;
N

A
where F is the feature size in nanometers, k is process related and is approximately equal
to 0.5, A. is the wavelength of the incident light and NA is the numerical aperture. This is
somewhat similar to the situation in the TEM, where resolving power depends on the
wavelength of the incident beam. In order to achieve smaller and smaller feature sizes,
semiconductor manufacturers have been reducing the wavelength of light used in
photolithography. The use of shorter and shorter wavelengths, however, is proving to be
problematic and expensive due to the light sources needed. The introduction of
immersion lithography has been used in reducing feature sizes. Since the feature size is
also dependent on the numerical aperture, which is dependent on the refraction index of
the medium in which the lens is working, an increase in the index of refraction reduces
the feature size. This development has helped photolithography extend its reign in
integrated circuit fabrication.

When producing patterns with photolithography for creating devices in this
research, it is only used to deposit metal on a substrate. There are several considerations
to keep in mind when carrying out this process. Thermal evaporation is usually used to
deposit metal in predeﬁned patterns corresponding to the mask pattern. Evaporation of
metal is done in high vacuum using a small ﬁlament consisting of a metal with a high
melting point, molybdenum or tantalum for example, called a boat, see ﬁgure 2.9. The
metal to be evaporated is placed on the boat and a high current is passed through heating
the boat to a high enough temperature to evaporate the metal. Usually a thin ﬁlm of

titanium is deposited ﬁrst as an adhesion layer and gold is then subsequently evaporated.

44

(Cl Curient {ﬁts/Shutter
Source-

r

i
g
i
I
it

   
 

8 Mt IS I t Vacuum
OUFCG e a 9 8C __g . .... " .
Sample Holder ‘ » COMIQI

Figure 2. 9. Edwards Auto 306 Thermal Evaporator at the W. M. Keck Microfabrication Facility at
Michigan State University. (a) The cylindrical evaporation chamber 15 shown here The sample holder is
at the top of the chamber and is held upside down. The metal is at the bottom and evaporated metal drifts
upward to the sample. (b) The sample holder is ﬂipped up for easy access for sample mounting. (c) The
evaporation mechanism is at the bottom of the chamber. The boat is screwed down and high current is
passed through it to heat the metal. A detector is used to monitor the evaporation rate. A shutter is used to
block the substrate from the source to stop metal deposition. (d) The evaporator controls. The current can
be increased and decreased with a knob. The digital readouts ofthe rate monitor and vacuum control
provide details ofthe evaporation conditions. Images: B.W. Jacobs

In order for the thermal evaporation process to be successful and to produce clean,
sharp features, a dual resist layer is required, see ﬁgure 2.10. This method is used so that
an undercut separates the unwanted metal layer on top of the resist with the layer that is
to remain on the substrate. The undercut is created a bottom resist layer that develops
faster than the top layer. This undercut provides added separation resulting in much
cleaner, sharper features. The dual resist stack can be achieved in a few different ways.

A special resist that inherently develops faster than the top resist is available, but is
expensive. A tri-layer stack is also possible and was used for device fabrication in this
research. First a layer of normal photoresist is deposited and exposed to light without a
mask, i.e. the entire substrate and photoresist layer is exposed. Then a thin layer of

aluminum, roughly 10 nm, is thermally evaporated. Aluminum is used because the

45

developer can etch it. A top layer of photoresist is deposited and the substrate is exposed
to light under the mask as described above. Since the bottom layer is exposed to light
twice it will develop quicker than the top layer, and the aluminum acts as a buffer to
prevent mixing of the top and bottom layers of photoresist. This process has proven
successful, is a much cheaper alternative to the Special undercut photoresists and a large
undercut can be obtained. After developing, the titanium/gold metal stack can be
thermally evaporated and the remaining resist and aluminum can be removed leaving the

desired pattern.

33333333
t—-—i ”F;

NWT—“m

Figure 2.10. Tri-layer photoresist stack. (a) After the ﬁrst layer has been deposited and exposed, a 10 nm
layer of Al is thermally evaporated. The second layer of resist is deposited and the stack is exposed under a
mask. (b) After developing an undercut should be prevalent. (c) Metal is thermally evaporated with a large
separation between the layer of metal on top of the resist stack and the layer on the bare substrate. ((1) The
top layers are removed and clean, sharp features are left. Images: B.W. Jacobs

 

 

Reproducible patterns made with photolithography techniques are essential for
making devices that function properly. A strict regiment of process steps should be
followed to produce successful results. Since lithography plays such a large role in this
research a process tailored by an individual should always be followed so that the yield is
maximized and the time needed to produce working circuits is minimized.

Photolithography was used to pattern larger areas for placing characterization probes.

46

 

'_ ni-

Smaller more intricate features were patterned with electron beam lithography. discussed

next

2.4.2 Electron Beam Lithography

Instead of using light for exposing photoresist in creating patterns on a substrate,
electron beam lithography uses highly focused electrons. Electron beam lithography
(EBL) is usually done in a SEM, where a highly focused beam of electrons is available.
see ﬁgure 2.1 1. This process is similar to photolithography in that it uses electron beam
sensitive resist (e-beam resist) to produce pre-deﬁned patterns. The EBL process is
mask-less, however. It uses the SEM beam deflectors to manipulate the beam position
and the beam blanker to expose only certain areas of the e-beam resist using a computer
control system. The e-beam resist is not exposed all at once. like in the photolithography
system, where a wide area of parallel light beams is used to expose the masked surface.
The small beam probe meticulously “writes” out the pattern sequentially, point by point.
After the beam has written the desired pattern, a developer is used to wash away the

exposed areas, just as in the photolithography process.

47

 

 

Figure 2.] I. JEOL 840A SEM with integrated EBL capability at the W. M. Keck Microfabrication
Facility at Michigan State University. (a) An SEM with a tungsten ﬁlament is used for EBL. (b) A
computer that is integrated into the SEM control system and is used to manipulate the beam according to a
pre-deﬁned pattern designed using a CAD program. Images: B.W. Jacobs

There are several factors in EBL one must consider to develop consistent,
reproducible results. First the ability to produce sharp, small feature sizes depends on the
electron beam parameters and how much the electron radiation the e-beam resist is
exposed to. In exposing a sample the round electron beam essentially ﬂashes the resist
one small spot at a time. see ﬁgure 2.12. It exposes one round area for a certain amount
of time, called the dwell time, then moves a set distance, called the point spacing. After
writing one line, the beam then moves a set distance to the next line, called the line
spacing, and continues this way in a raster pattern. The SEM electron beam current plays
a very important role in the exposure process and is used to calculate the dose a sample
receives. These parameters: dwell time, point spacing, line spacing. beam current and the
dose, are all essential considerations for obtaining successful results. They should be
used consistently after a successful combination has been found so that device yield is

maximized.

4s

I
I
g Electron Beam

 

Finish ._ .
\U0 0 oo oootLlne
00000 0

O O Spacing
-—-> «9
Start Raster Point-
Direction 59“”

 

 

 

Figure 2.l2. Illustration ofthe electron beam lithography method. The beam ﬂashes the sample at a point.
moves a small distance ﬂashes it again, and continues this way in a raster pattern. The spacing between the
ﬂashes can be deﬁned in the point spacing and line spacing. Image: B.W. Jacobs

EBL has several advantages when compared to photolithography. Since the
electron beam probe in the SEM is highly focused and small, just a few nanometers in
diameter, very small feature sizes on the order of 10 nm or less are possible". Second,
EBL is mask-less. Therefore, custom circuit designs can be created and many variations
can be made with little or no extra cost for fabrication. This makes EBL invaluable for
academic and research laboratories.

EBL has several disadvantages. however. Feature size limitations due to electron
backscatter and secondary electron emissions, the basis of SEM image formation as
discussed in section 2.3, from the substrate causes a proximity effect that exposes
additional resist. This same phenomenon limits the ultimate feature sizes by causing
patterns that are close together to be affected by additional electron exposure when the
adjacent pattern is being written. The electromagnetic lenses are also a source of
problems in EBL. As previously mentioned these lenses are far from perfect and small
variations in their magnetic ﬁeld create inconsistencies across multiple exposed samples.

The low throughput is the main reason why the semiconductor industry has not adopted

49

EBL in mass production of integrated circuits. Integrated circuits need an extraordinarily
high level of consistency, reproducibility and speed in fabrication.

EBL is one of the primary tools used in this research. All of the GaN nanowire
devices studied in this thesis research have been fabricated with EBL, with a J EOL 840A
SEM located in the W.M. Keck Microfabrication Laboratory at Michigan State
University. The mask-less system has made it an invaluable tool for nanowire devices
because nanowires are placed on the substrate in random locations. Because a custom
pattern can be created with EBL each and every time a new circuit is to be made. it is an

excellent tool for fabricating working devices.

50

2.5 Device Characterization Techniques

Characterizing devices that incorporate nanoscale materials pose signiﬁcant
challenges. Many properties of such a device are much different than traditional
macroscopic devices. These properties include fundamental differences in device physics
due to reduced dimensionality components, such as ballistic transport, carrier
conﬁnement, and other quantum mechanical effects. Devices that incorporate these F

reduced dimensionality components usually operate at lower currents and voltages and

 

can operate at higher frequencies. So the need for sensitive, low noise measurement
instruments is essential for achieving accurate and meaningful results.

Probe and contact coupling to these nanoscale materials is an important research
issue and must also be taken into account7. For example, when a macroscopic, three-
dimensional probe contacts a one-dimensional carbon nanotube or nanowire, the junction
characteristics between the nanotube and nanowire are fundamentally different than in a
planar system, such is the case in current transistors. Electron transport through a
macroscopic contact to a nanotube or nanowire is also an issue since the available
electron states in the macroscopic contact is different than in the nanoscopic nanotube or
nanowire. Instrument developers, as well as researchers who seek high-quality, accurate
and meaningful measurements of test devices. should keep these issues in mind when

testing devices.

2.5.1 Two-Point Probe
Several properties of a device can be measured using two probes. By applying a

voltage across two probes in a device and measuring the current, or by passing a current

51

through two probes in a device and measuring the voltage, a resistance can be obtained.
Using this technique several important properties of a device can be measured. Current-
voltage (I-V) measurements are probably the most common two-point probe
measurement. Much information about the device properties can be obtained from an I-V
measurement, and it is one of the main tests used to characterize a transistor. If a voltage
across the transistor gate is also varied, the I-V measure will change according to the
voltage on the gate. Therefore, the gate voltage inﬂuence on the amount of transistor
current allowed to pass can be quantitatively determined. Two-point probe investigations
using a low noise Keithley 4200-SCS at Keithley Instruments in Cleveland, Ohio,
characterized several GaN nanowire-based transistors conﬁrming gate modulation and
their ability to carry a large current density. In addition to I-V, capacitance-voltage (C-V)
measurements also yield important junction capacitance information as a function of

voltage applied.

2.5.2 Four-Point Probe

In determining the resistance of a device element, the four-point probe technique
produces more accurate results when compared to the two-point probe technique. As the
name implies the four-point probe technique uses four probes in taking measurements.
The two outside probes source a known current, while the two inside probes measure a
voltage, as shown in ﬁgure 2.13. The voltage probes carry very little current so only a
voltage drop is measured. Since a known current is passing through the sample, the

measured voltage can be divided by the current and an accurate resistance calculated. By

52

 

using the voltage probes to measure the only the voltage drop across the sample and not

the current. the resistance of the contacts is omitted.

 

 

Probes

 

‘5 V'

Sample

   

Figure 2.13. The four-point probe resistance measurement technique. This technique uses four probes that
are placed on the sample. The two outside probes pass a known current through the sample. The two
inside probes measure a voltage. Very little current passes through the voltage probes. By dividing the
known current into the measured voltage, an accurate resistance of the sample can be measured, with the
contact resistances of the probes omitted. Image: B.W. Jacobs

Four-point probe is a powerful technique when the resistance of just the sample is
required. However, several issues involving probe coupling to the sample arise due to the
small size of the samples used in this thesis research. As mentioned above, if a
macroscopic probe contacts a nanoscopic element, the change in dimensionality will
introduce added effects. In order to address this issue, instrument manufacturers have
designed probes with tip sizes on the nanoscale. This reduces the effects of the probes,
making them less invasive, so a more accurate four-point probe measurement can be
taken.

The opportunity to use a state-of-the-art Zyvex KZ-l 00 nanomanipulation system,
as Zyvex Co. in Richardson, Texas, that incorporated these nanoscale probes connected

to a low-noise Keithley 4200-SCS was afforded to this research. Resistance

53

 

measurements of a GaN nanowire using this nanomanipulation system were taken with
real-time visual inspection with an SEM. Intriguing phenomena was observed, which
would not have been possible with traditional probing instruments. These results will be
discussed at length in chapter 4.

The two and four-point probe techniques represent very important device
characterization methods essential for producing real world devices and systems. New
and custom instruments designed to accurately and precisely measure nanoscale devices

and systems must be utilized to obtain accurate measurements.

54

 

2.6 Scanning Probe Microscopy

The scanning probe microscope is a powerful ultra-high-resolution surface
characterization technique that is highly versatile. The scanning probe microscope
(SPM) encompasses several different techniques designed to obtain different
characteristics of a given material. The ﬁrst SPM was the scanning tunneling microscope
invented in 1972 by Russell Young and colleagues at the National Institute of Standards
and Technology O\IIST), and further developed by Gerd Binnig and Heinrich Rohrer at
the IBM Zurich Research Laboratory in Switzerland. The scanning tunneling microscope
(STM) is used to detect weak electrical currents in a material and will be discussed later
in this chapter. Binnig and Rohrer’s STM won them the Nobel Prize in Physics in 1986.
Since then a wide variety of SPMs have been invented and highly creative ways of using

them have revolutionized the way materials are imaged and manipulated.

2.6.] Atomic Force Microscopy

The atomic force microscope (AF M) is probably the most common form of SPM.
The AFM is a surface scanning technique capable of creating a three-dimensional image
of a sample surface with atomic resolution in the z-direction, or height. In creating an
image, the AFM uses a sharp probe tip at the end of a cantilever that is rastered across the
sample surface in the x and y-direction. The probe tip is raised and lowered following
the height variations of the sample surface, however it does not actually touch the sample
surface. The tip-sample interaction forces can be of several different types, including
electrostatic, Van der Waals, magnetic, and so on. A laser that is focused on the end of

the cantilever above the tip is reﬂected to a photodiode, and changes in sample height

55

 

deﬂect the laser to different angles, see ﬁgure 2.14. The photodiode can detect the
variations in the location of the laser resulting from different tip positions. The
photodiode signal is sent to the computer so that a digital image of the sample surface can
be constructed.

The lateral resolution, or resolution in the x and y—directions, however, is highly
dependent on the size of the tip. A smaller tip radius yields better lateral resolution. The
AFM is not generally used for high-resolution imaging in the x and y-directions. Recent
work, however, using certain molecules8 or carbon nanotubes9 as tips, has drastically

increased the lateral resolution of the AF M, and atomic scale imaging has been achieved.

  
 

Photodiode

  
   

 

'15. . .
Sample Surface . Cantilever & T|p

PZT Scanner

Figure 2.14. Diagram ofbasic AFM operation. The cantilever and tip are rastered along the sample
surface. The variation in sample topology is detected by a reﬂected laser/photodiode system. The signal
from the photodiode is then sent to the computer for digital image construction.

Two basic modes of AF M are used for sample analysis, contact and tapping.

Contact mode uses a cantilever with a known spring constant. It rasters across a sample

56

surface and detects variation in height. The cantilever is deﬂected when a change in
height is encountered, resulting in a deﬂection of the reﬂected laser. which is detected by
the photodetector. Height is also controlled via a feedback loop. which is an essential
element in all modalities of the SPM. Cantilevers with several different spring constants
are available. Certain cantilevers tend to work better on certain surfaces than others. The
cantilever is rastered across the surface and the cantilever is deﬂected at variations on the
surface according to Hooke’s law, F = -kx, where k is the spring constant of the
cantilever and x is the distance of deﬂection. Tapping mode uses a cantilever that
oscillates up and down at a high frequency, usually at few 100 kHz, or close to the
resonance frequency of the cantilever. As the cantilever is rastered across the sample it

taps along the surface detecting variations in sample height.

Optical
microscope

Anti—vibration

marble table ‘
Control
computers

  

Figure 2.15. Veeco Instruments Nanoprobe III—A Multimode SPM at the Electronic and Biological
Nanostructures Laboratory at Michigan State University. (a) This image shows an entire SPM setup. The
SPM sits on top of several anti-vibration surfaces. An optical microscope is used to focus the sample
surface and roughly focus the laser beam on the tip. The SPM is controlled via computer. (b) The head
contains the laser, photodiode and provides the electrical connection to the tip. (c) A close up of the
microscope shows the piezoelectric motion control and the head. (d) An AFM tip holder. (e) STM tip
holder. Images: B.W. Jacobs

57

2.6.2 Scanning Tunneling Microscopy

The scanning tunneling microscope is used for imaging atomic scale features in
electrically conducting samples. The conducting tip, usually platinum iridium, is
atomically sharp, only one atom in diameter, providing an extremely small probe size.
Due to this small probe, the STM is one of the highest resolution microscopes available.
Similar to AF M, the STM tip rasters across the sample to form an image. The tip is

brought close to the sample surface and an electric ﬁeld is applied across the sample, see

 

ﬁgure 2.16. Electrons are then emitted from the tip, tunnel through the gap between the
tip and sample surface and enter the sample.

Variations in tunneling current through the sample are measured and depend on
electron density within the sample. For example, tunneling current will be reduced when
the tip is near an atom and is increased in between atoms. The sample can be imaged
keeping either the current or height constant through the use of the feedback loop.
Tunneling current also depends on the square of the distance from the sample surface

making this an extremely sensitive technique.

58

 

 

“”F‘“

/ u
lSmnMe \\

_l_ \

Tunneﬁng
vokage

 

tunneling voltage is applied. creating an electric ﬁeld across the sample. and electrons tunnel out ofthe tip
into the sample. The amount oftunneling current will depend on the variations in the electron density in
the sample. i.e. the tunneling current over an atom will be different than the tunneling current between
atoms. Image: adapted from original image, Michael Schmid. TU Wien.

The SPM has been used to image nanoscale features of several types of
nanomaterials used in this research, including carbon nanotubes, carbon onions, GaN
nanowires and GaN nanowire growth matrix. All SPM investigations have been done
with a Veeco Instruments Multimode Nanoscope IIIa located at the Electronic and
Biological Nanostructures Laboratory at Michigan State University. It provides surface
data in great detail and augments other material characterization instruments such as the
TEM and SEM. Due to the extreme resolution in height, much information regarding the

surfaces of nanomaterials can be uncovered.

59

2.7 Raman Spectroscopy

Raman spectroscopy is used to study the molecular composition of solid samples.
It uses a beam of laser light, usually in the visible regime, to induce vibrations. or
phonons, in a sample. These phonons are material speciﬁc and depend on atomic mass,
bond order, molecular constituents, molecular geometry, and hydrogen bonding.

When the photons from the laser interact with electrons surrounding atomic nuclei
ofa sample material, photons are usually elastically scattered. This means they will have
the same energy and wavelength before and after the interaction. This is called Rayleigh
scattering, as shown in ﬁgure 2. l 7. In rare instances the photons will be scattered
through an inelastic process and lose some energy to the material. This process can
happen in one of two primary ways: Stokes scattering and anti-Stokes scattering"). In
Stokes scattering an electron will absorb some energy from a photon and go from the
vibrational ground state to a higher virtual energy state. The electron will reduce in
energy and fall back into a vibrational state higher than the ground state. The energy that
is transferred to the electron is equal to the difference in energy of the incident beam and
the scattered beam of laser light. This loss in energy will cause the scattered photon to
have a longer wavelength than the incident photon. This change in wavelength can be
detected, and is referred to as the Raman shift, measured in inverse centimeters.

Anti-Stokes scattering is a very similar process. In anti—Stokes scattering an
electron that is already in an excited vibrational energy state will absorb some energy
from a photon and go from its excited state to a higher virtual energy state. The electron
will reduce in energy and fall back below its initial excited state to the ground state. This

process will give the photon more energy and shorten its wavelength. The magnitude of

60

the change in photon energy for both Stokes and anti-Stokes scattering is the same but
has opposite signs. However. at room temperature anti-Stokes scattering is much less
frequent. Electrons at room temperature are not often at an excited vibrational energy
state above the ground state, so anti-Stokes scattering does not usually signiﬁcantly

contribute to a Raman spectrum. Anti-Stokes scattering can be a measure for how

anvr-s l w L'a'

materials respond to temperature.

 

Virtual
Energy States

 

 

    

’1

 
 
  

Excitation Rayleigh Stokes Anti-Stokes

 

 

 

 

 

 

Energy 3 Scattering Scattering Scattering
% Vibrational
g V Energy States
i I H Ground State

 

Figure 2.17. Illustration of the basic theory of Raman scattering. In each case a photon causes electrons to
enter a virtual energy state. The electron will then drop in energy either returning to the ground state or
entering a higher vibrational energy state. Rayleigh scattering is the most common form of scattering and
is caused by inelastic scattering. Stokes scattering is more common than anti—Stokes scattering. Image:
B.W. Jacobs

Most materials have speciﬁc phonon modes and therefore have speciﬁc,
measurable Raman shifts. Raman spectroscopy is an excellent tool for analyzing organic
materials, such as carbon, and has been extensively used in this research for analyzing the
different allotropes of carbon, such as carbon nanotubes and carbon onions. Figure 2.18
shows a typical micro Raman instrument. All Raman investigations were done using this

instrument, the Kaiser Optical micro Raman Spectrograph at the W.M. Keck

61

Microfabrication Facility at Michigan State University. It uses a frequency doubled

YAG laser with a wavelength of 532 nm.

»‘ ~i‘n

Computer

Stage Control
Laser Source
and Detector

Control

    

Figure 2.18. Kaiser Instruments micro Raman Spectrograph at the W. M. Keck Microfabrication Facility at
Michigan State University. (a) The instrument consists ofa mechanical stage control, laser source and
detector. and a computer used for analyzing data. (b) The laser used for exciting materials is focused
through a microscope. This allows for very small areas ofa sample to be analyzed. Image: B.W. Jacobs

There has been much work done involving carbon nanomaterials and Raman
spectroscopy. Many highly speciﬁc material characteristics can be determined using
Raman spectroscopy and since the discovery of the carbon nanotube, it has proved an

invaluable tool in analyzing and understanding their propertiesl 1.

62

I—J

(1.4

("L

("L/

2.8 Cathodoluminescence and Photoluminescence

Cathodoluminescence (CL) is a technique used to determine the electronic
structure characteristics of a material. It uses energetic electrons, usually from a SEM.
which impart enough energy to excite electrons in the material from the valence band,
where electrons are bound to atomic nuclei, into the conduction band, where they are free
to move about the material. This energy transition is material speciﬁc. The amount of
energy, measured in electron volts (eV), electrons need to be raised into the conduction

band from the valence band is called the band gap. Inside this energy gap there are no

 

available energy states for the electron, so in order to enter the conduction band an E
electron must gain at least the amount of energy equal to the band gap.

In CL electrons are given enough energy through the electron beam-sample
interactions to be excited into the conduction band, as shown in ﬁgure 2.19. An excited
electron will leave an empty state, or hole, in the valence band. The electron will quickly
fall back into the valence band since it will always try to minimize its energy. Because
energy is always conserved, this energy loss is converted into a photon equal to the loss
of energy experienced by the electron. The photon energy will usually equal the band
gap. The emitted photon is detected, and its energy, and therefore the sample material
band gap, can be determined.

This technique can also determine the purity of a sample. If the material contains
impurities, the impurities will also emit light with a different wavelength, because usually
the energies of the impurities will lie within the band gap of the material. Defects also
cause luminescence in differing wavelengths depending on where the defect lies within

the band gap.

63

hoIeO Energy A

 

 

electron.
Ec ‘ ¥ 9
‘ i I
, i _'_
SEM e- Beam I
E. W l +
S | I
| l

 

 

m
<
{

Figure 2.19. Illustration of the process involved in cathodoluminesence. Electrons represented by the
black dots are excited to the conduction band, EC, from the valence band, Ev, by the SEM electron beam.
Electrons can be excited to states within the band gap, which can consist of impurities and defects in the
material. Image: B.W. Jacobs

Photoluminescence (PL) is a very similar technique as CL and provides much of
the same information. PL measures electron transitions in a material just as in CL, but it
uses photons instead of electrons to probe the material. There are subtle differences
between the two techniques, however.

In CL the electron beam induces excitations and emission from all the
luminescent mechanisms within the material. This includes band-edge, valence to
conduction band, emission as previously mentioned, defect induced emission, where
defects with energies within the band gap are excited, as well as other recombination
centers present in the material”. This is due to the high energy of the electron beam. CL
also provides depth-resolved luminescence information by varying the electron beam
energy. When using a SEM to perform CL measurements a high spatial resolution is also
available, where small areas of a material can be probed, which provides a spatial map of
the sample luminescence.

PL excitation wavelengths, on the other hand, can be highly selective and can

resolve spec1ﬁc luminescence mechanisms Within a material 3. This is important when

64

the band-edge emission and other luminescence mechanisms such as defects within the
band gap need to be distinguished. Defect energies usually lie within the band gap of a
material. Wavelengths that are too low in energy to excite band-edge luminescence can
be used to distinguish these lower energy mechanisms. Therefore, if a defect with lower
energy than the band gap is to be investigated, the excitation wavelength must be higher
than the defect energy but lower than the band gap.

CL and PL are very important characterization tools in semiconductor
engineering. They provide critical information, such as band gap energies, defect and
recombination center details and the particulars of emission efﬁciency of materials. A
LEO Supra 50 VP SEM in conjunction with a Gatan MonoCL system with a Hamamatsu
R5509 photomultiplier located at Jet Propulsion Lab was used for the CL measurements.
PL has as yet not been used in this thesis research.

Since this research deals heavily with GaN, which is a very promising optical
material for blue and UV devices, CL and PL investigations are of utmost importance in
characterizing the optical properties of the GaN nanowires. CL was also an independent
experiment that identiﬁed both the wurtzite and zinc-blende phases in the multiphase

nanowires, as will be discussed in chapter 4.

65

2.9 Thermogravimetric Analysis

Thermogravimetric analysis is used to determine the thermal stability of a given
material. This is done with a highly precise balance and furnace. In sample testing the
furnace temperature increases at a speciﬁed rate and at a given temperature the sample
will begin to lose mass through vaporization. After a certain temperature is reached the
sample will quickly vaporize. Throughout the temperature cycle, the precise balance
measures the rate at which the material vaporizes, see ﬁgure 2.21. The furnace chamber
can also be purged with an inert gas, water vapor, oxygen or other gas depending on the
desired reaction. The resulting mass loss curves may also need further post processing to
bring out subtle variations. The derivative of the mass curves is often used in uncovering

these variations and to determine the point at which mass loss rate is greatest.

66

Furnace

.4

 

*--~ Gas Flow

Figure 2. 20. TGA instrument, Rigaku 8120 TG- D I A at the Tokura—Hirata Laboratory at Tokyo Institute
of Technology The sample material is packed into a small Pt pan and placed on precise balance. The pan
is then put into a furnace and heated up. The arrow below the furnace label indicates the movement of the
furnace. The furnace rides on two rails and encapsulates the samples in the furnace. The inset shows the
whole instrument, with the fumace on the leﬁ and the balance mechanism on the right. A small tube for a
desired gas is also indicated. Images: B.W. Jacobs

The TGA was primarily used to analyze carbon onion samples. The TGA system
used was a Rigaku 8120 TG-DTA located in the Tokura-Hirata Laboratory at Tokyo
Institute of Technology. Multiple carbon onion samples were produced from a variety of
different starting materials and at different temperatures. In order to gain insight into the
quality of carbon onions produced, and to determine the optimum starting material and
temperature, TGA was used. Early on, the carbon onions were heated without the use of
a ﬂowing gas. The results were inconclusive, and the rate at which the temperature could

be increased was much too low. The rate at which the temperature can be increase must

67

be such that a smooth loss in mass results. That way more consistent results and critical
points such as the point at which mass loss is greatest can be more easily determined.
Therefore puriﬁed air was passed through the furnace chamber at .2 l/min. The oxygen
in the puriﬁed air caused the onions to oxidize into a gas form thereby reducing the mass.
Initial mass reduction occurred at a lower rate and more consistent results were obtained.
which will be discussed in chapter 5.

In addition to the ﬂowing gas, there are several other considerations one must take
into account when doing TGA. A consistent sample weight must be used, which must be
the maximum amount the pan can accommodate, since carbon onions are very light and
ﬂuffy. They can also be pressed so that more material can be placed into the pan. All
procedures must be as consistent as possible if reproducible results are to be obtained.

Using this procedure one can quantify the quality of as grown onions. If the
carbon onions have outer layer defects, such as facet points or discontinuities in the wall
structure, they will oxidize at lower temperatures”. The ﬁndings were compared with
results from such studies as HRTEM imaging or Raman spectra, and an overall picture of
onion growth characteristics were obtained.

Each instrument described above provided essential details of analyzed samples.
These details could then be pieced together to gain an overall picture. Many times in
uncovering and understanding new materials and material phenomena, several
independent tests are need to provide a clear explanation for the observed phenomena.
This was the case, for instance, for the new GaN nanowire structure and carbon onion
modiﬁcations due to heavy ion radiation, results from several instruments were needed to

compare the observations. Each instrument described above provided important results.

68

Additional instruments and methods will be described in chapter 6 that could provide
further important details of these materials and devices discussed in this thesis research.
The candidate investigated the materials and devices using all the instruments and

methods described above.

69

 

References

l.

10.

ll.

12.

l3.

14.

D. J. Grifﬁths, “Introduction to Quantum Mechanics,” Benjamin C ummings. San
Francisco, 2004.

D. B. Williams, C. B. Carter, “Transmission Electron Microscopy,” Plenum Press,
New York, 1996.

R. F. Pierret, “Advanced Semiconductor Fundamentals.” Prentice Hall, Upper Saddle
River, 2003.

R. F. Egerton, Electron Energy Loss Spectroscopy in the Electron Microscope,
Plenum Press. New York, 1986

M. .I. Madou, “Fundamentals of Microfabrication: The Science ofMiniaturization,”
CRC Press, Boca Raton, 2002.

K. A. Valiev, “The Physics of Submicron Lithography,” Plenum Press, New York.
1992.

“Nanotechnology Measurement Handbook,” Keithley Instruments, Inc., Cleveland,
2007.

F. Kuhner , L. T. Costa. P. M. Bisch, S. Thalhammer, W. M. Heckl, H. E. Gaub,
“LexA-DNA Bond Strength by Single Molecule Force Spectroscopy,” Biophys. J.,
87, 2683-2690 (2004)

J. S. Bunch, T. N. Rhodin, P. L. Mchen, “Noncontact—AF M imaging of molecular

surfaces using single-wall carbon nanotube technology,” Nanotechnol., 15, 876-878
(2004)

R. L. McCreery, “Raman Spectroscopy for Chemical Analysis,” John Wiley & Sons.
New York, 2000.

M. S. Dresselhaus, G. Dresselhaus, A. Jorio. “Unusual Properties and Structure of
Carbon Nanotubes,” Annu. Rev. Mater. Res, 34, 247-278 (2004)

B. G. Yacobi, D. B. Holt, “Cathodoluminescence Microscopy of Inorganic Solids.”
Plenum Press, New York, 2002.

“Optical Radiation Measurements, Vol. 3, Measurement of Photoluminescence,” Ed.
K. D. Mielenz, Academic Press, London, 1982.

S.-Osswald, G. Yushin, V. Mochalin, S. O. Kucheyev, Y. Gogotsi, “Control of

sp2/sp3 Carbon Ratio and Surface Chemistry of Nanodiamond Powders by Selective
Oxidation in Air,”J. Am. Chem. Soc, 128, 11635-11642 (2006)

70

3. Reduced Dimensionality Systems: From Three Dimensions to Zero

As current silicon based devices are facing an inevitable end due to size
limitations, novel materials and device concepts are being explored as successors to
silicon technologies. The materials that seem to be the most promising usually have
reduced dimensionality components. Essentially, a reduced dimensionality component.
either engineered or inherently present in a material, has spatial dimensions that are so
small that quantum mechanical conﬁnement effects begin to play a signiﬁcant role in
determining the properties of the material. These quantum mechanical properties can
enhance the performance of a material in a device, invoking properties that are not found
in the same device made from the same material in bulk form. Such enhancements
include faster electronics, brighter more efﬁcient lasers and light emitting diodes, and
sensors with extremely high sensitivity to a given compound.

Starting from a bulk material and working down, reducing one dimension at a
time and discussing important properties at each step, will help clarify the importance of
these systems, and the concept of reduced dimensionalities. Starting with a bulk block of
material, the behavior of electrons will be explained as the spatial dimensions are shrunk
as an illustrative example. This shrinking of dimensions is actually much more complex
than it initially seems with intriguing quantum mechanical induced effects, but these
effects are beyond the scope of this example.

Starting with a three-dimensional n-type semiconducting material, excess
electrons can more or less freely move and bounce throughout the material. When an
electron is a three-dimensional material it acts much like a particle bouncing about

through the lattice, interacting with the larger ﬁxed atoms. If there is an external

7]

perturbation, such as an applied electric ﬁeld, the electron will randomly bounce through

the lattice with an average motion opposite to the electric ﬁeld.

‘2 . z z
'7)’ Y Y
X X

Figure 3.1 Examples of reduced dimensionality systems. From leﬂ to right: three-dimensional, two-
dimensional, one-dimensional, and zero-dimensional. Image: B.W. Jacobs

 

 

 

 

 

 

 

 

 

If one spatial dimension is reduced, a very thin sheet of material will be formed,
similar to a sheet of paper. In this case one spatial dimension is conﬁned, i.e. it has one
reduced dimensionality component. If we assign the z-direction of a three-dimensional
coordinate system normal to this thin ﬁlm, and assign the x and y-directions to the plane
of the ﬁlm, this system is considered to be conﬁned in the z-direction. If this spatial
conﬁnement is sufﬁciently small, the electron energy landscape in this system will
change relative to the three-dimensional, bulk system, i.e. electrons will not be able to
freely move throughout the material. This spatial restriction will reduce the amount of
available energy states accessible to the electrons, and electrons can only exist where
there is available energy. If this spatial dimension is reduced beyond a certain point, the
available electron energies will begin to separate. There will no longer be a continuous
range of energy. A distinguishable set of energies, or energy levels, will begin to arise.
This is called energy level quantization, and as the space gets ever smaller, the energy

levels become more separated. However, let us not forget the energy landscape the in x

72

and y-directions. In plane with the surface of the thin ﬁlm, electrons will move much like
they did in the bulk material. This quantization of energy levels in the z-direction is
purely a quantum mechanical phenomenon and can be explained by a relatively simple
and very common model in quantum mechanics called a quantum well.

The quantum well is used to describe the behavior of trapped particles, including
electrons. Figure 3.2 shows a one-dimensional quantum well with a length, the x-axis,
and energy, the y-axis. This figure shows 4 energy levels stacked on top of one another
depicting how the energy of the electrons arrange in such a conﬁguration. As previously
mentioned, electrons can only exist where there is available energy, so only a speciﬁc
number of electrons can exist in a well. These energy levels depend on the inverse
length, L, of the well. As this length decreases it increases the energy level spacing and
reduces the number of available energy levels within the well. This concept is used in a
variety of models in quantum mechanics and can be used to approximate energy levels in

reduced dimensionality systems.

 

 

 

n=4
>, nznzhz
O) = -
g n 3 E" 2mL2
LIJ n:

n=1

 

 

 

 

Length

Figure 3.2 A quantum well. Particles in a quantum well are trapped in a conﬁning potential. Ifthe length
ofthe well, L, is small enough, energy levels within the well will begin to separate and give rise to
quantized energy levels. These energy levels are determined by the equation on the right, where n is the
quantized energy level, h is Planck’s constant, in is the mass of the particle, and L is the length of the well.
Image: B.W. Jacobs

73

 

Two-dimensional systems are very popular in many systems and devices. In
heterostructures a thin sheet of material sandwiched between two layers of material of a
larger band gap, creates a quantum well in the thin, two-dimensional material. This thin
material can be used as an optical device, efﬁciently creating certain colors of light.
Two-dimensional materials can also be used in high power devices. If a two-dimensional
material is again sandwiched between two materials with a larger band gap, a large ll

amount of current can pass through the thin area. This device is called a high electron

 

mobility transistor (HEMT) and is described in detail in chapter 6.

If a spatial dimension in the two-dimensional system were reduced, a one-
dimensional system would result. This system consists of a line or very skinny wire of
material, usually called a quantum wire or nanowire. This wire will restrict the energy
levels in two-dimensions, i.e. it has two reduced dimensionality components. Electrons
in this system are only free to move in the x-direction if the spatial dimension is reduced
in the y direction. Two quantum wells, one in the z-direction, as in the thin ﬁlm case, and
in the y-direction. dictate the energy landscape. This has signiﬁcant implications on
electron conduction in such a system. The nanowire can only accommodate a set number
of quantized energy levels in which electrons can travel, because of the quantum wells in
the y and 2 directions. These quantized energy levels are called conduction channels.
Carbon nanotubes are considered to be one of the best models of a one-dimensional
system and are discussed in chapter 5. The GaN nanowires used in this thesis research
are a pseudo one-dimensional system, because they do possess some one-dimensional

quantum phenomena, but are quite large compared to a nanotube.

74

Electron injection into the one-dimensional system from a three-dimensional
metal, which is usually used to connect the nanowires in devices, is fundamentally
different than in a classical case due to the finite number of conduction channels in the
reduced dimensional system. The three-dimensional metal essentially has an inﬁnite
number of conduction channels, so when electrons are injected into a one-dimensional
system there will be a resistance. There will also be a resistance when the electron
travels out of the one-dimensional system into the three-dimensional metal. For efﬁcient
junctions to low-dimensional materials, a so-called conduction channel-matching
situation would be ideal. If the both the contact metal and one-dimensional wire had the
same number of conduction channels there would theoretically be no resistance'. This
situation simply does not exist in macroscopic electronics. It is important to note, this is
one of the main problems facing nanoelectronics, namely creating good contacts from the
macroscopic world to the reduced dimensional system.

In reducing the ﬁnal dimension, we are left with a zero-dimensional system, or a
small box, called a quantum dot. This system has quantum wells in each of the three
directions, conﬁning electrons to a very small area with a small number of available
energy levels. Quantum dots have excellent optical properties and have been shown to be
efﬁcient LEDs2 and lasers3. Quantum dots have also been shown to be good photovoltaic
cells.

Most future electronic devices will have some reduced dimensionality, or
quantum component. Each reduced dimensionality has unique properties due to quantum
mechanical induced effects, and speciﬁc devices can exploit these properties in niche

applications. These properties are too numerous to discuss in full. This thesis discusses

75

 

original work in GaN nanowires, carbon nanotubes and carbon onions. all reduced

dimensionality systems, and their applicability in next generation systems and devices.

76

 

3“: r"

 

References

l.

b.)

S. Datta, Electronic Transport in Mesoscopic Systems, Cambridge: Cambridge
University Press 1995

R. Schmidt, U. Scholz, M. Vitzethum, R. Fix, C. Metzner, P. Kailuweit, D. Reuter, A.
Wieck, M. C. Hiibner, S. Stuﬂer, A. Zrenner, S. Malzer, G. H. Dohler, “Fabrication

of genuine single-quantum-dot light-emitting diodes,” Appl. Phys. Lett., 88, I21 115
(2006)

M. Sugawara, N. Hatori, M. Ishida, H. Ebe, Y. Arakawa, T. Akiyama, K. Otsubo, T.
Yamamoto, Y. Nakata, “Recent progress in self-assembled quantum-dot optical
devices for optical telecommunication: temperature-insensitive 10 Gb s’l directly

modulated lasers and 40 Gb 3" signal-regenerative ampliﬁers,” J. Phys. D, 38, 2126-
2134 (2005)

77

4. Zinc-Blende Wurtzite Multiphase Gallium Nitride Nanowires

This chapter discusses all original research contributed by the candidate. Recent
discovery of the multiphase nanowires by the candidate and subsequent investigations of
the nanowire structure, growth and electronic characteristics will be thoroughly
discussed.

Semiconducting nanowires have gained enormous popularity due to their
versatility in multiple applications. Semiconductor nanowire versatility translates
directly from the bulk semiconductor counterparts, providing a wide assortment of band
gaps, hole and electron mobilities, and mechanical properties that can be utilized in
device engineering. Many nanowire properties are even improved by conﬁnement effects
in comparison to semiconductor bulk properties. Various research groups have grown a
wide variety of semiconductor nanowire structures including ZnO', Siz, Ge3, GaAs4 and
InN5 nanowires. Nanowire heterostructures“7 and nanotubesg‘9 also represent important
steps forward toward practical device applications in nanowire engineering.

Gallium nitride has gained much interest in recent years, due to its desirable
characteristics in electronic and optical devices. It is a direct, wide band gap
semiconductor, 3.299 eV in bulk zinc-blende and 3.437 eV in bulk wurtzite"), and is
inherently n-type with a doping concentration of around 10'8 cm'3 present due to nitrogen

- l l.l2
vacanCIes

. Gallium nitride (GaN) nanowires are currently being used in and
developed for a wide variety of applications. GaN nanowire ﬁeld effect transistors‘3 have
demonstrated high transconductance and good switching behavior. Complementary

logic'4 and optical'5 devices have been fabricated using p-type dopants in nanowire

rowth'6, further enhancin the versatili of these nanowires in devices. Mechanical
g g

78

 

properties of GaN nanowires have also been investigated for use in mechanical
resonators”. GaN also has the ability to form in two different crystalline phases, zinc-
blende, the metastable phase, and wurtzite, the stable phase.

In order to develop GaN nanowire devices with high levels of performance,
nanowires must be of high crystalline quality and material purity. Therefore, a precisely
tuned nanowire growth process must be implemented in order to achieve this quality and
purity. Many research groups have successfully grown GaN nanowires using several
methods of growthm‘lg‘zo. These methods can be split into two basic categories, vapor-
liquid-solid (VLS) and vapor-solid (VS)2'. In VLS growth a catalyst particle is used for
nanowire nucleation. The catalyst, usually a transition metal such as nickel or iron in
liquid form, decomposes vapor-phase growth constituents on its surface, then
supersaturates with the radicalized constituents, which is followed by re-assembly and
continuous extrusion of a solid crystalline nanowire with correct stoichiometry. In VS
growth, nanowire constituents decompose from pre-existing solid crystalline elements
and at active sites on the surface of the same crystalline elements. These pre-existing
crystalline elements are called a matrix.

The GaN nanowires that were investigated in this thesis research are distinctly
unique; they naturally grow incorporating both the zinc-blende and wurtzite crystalline
phases simultaneously in longitudinal crystal domains. This multiphase nanowire
structure was obtained through a catalyst-free VS growth mechanism in a direct reaction
of gallium metal with ammonia (N H3) in a quartz tube fumace22‘23. Growth temperatures
ranged between 850-1000° C with the ammonia ﬂow rates around 20 sccm and pressures

between 5 and 15 Torr. Nanowire grth can be broken into three basic stages. In the

79

 

ﬁrst stage, an amorphous GaN matrix forms. In the second stage, GaN hexagonal crystal
platelets form within the amorphous matrix. In the third and ﬁnal stage the nanowires
grow from the sides and corners of the GaN hexagonal platelets. By tuning growth
parameters, different nanowire structures have been obtained, including the zinc-
blende/wurtzite multiphase nanowires as well as wurtzite single-phase nanowires.
Temperature seemed to be the strongest controlling parameter and extensive SEM and
HRTEM analysis based on growth temperatures of 850° C, 950° C and 1000° C revealed
distinct differences in growth matrix properties. The differences in growth matrix
properties resulted in variations in nanowire structure as will be discussed.

The structure of the multiphase nanowires was studied using plain-view HRTEM
images to understand the longitudinal structure conﬁguration. Understanding and
reconstructing this multiphase structure using plain-view HRTEM, which included
diffraction patterns and F FTs, proved difﬁcult. So further investigations using cross-
section HRTEM nanowire samples fabricated with a FIB were invaluable in
understanding the nanowire internal structure.

The electronic properties of these multiphase nanowires were also investigated
using several techniques. Nanowires were used as the semiconducting element in ﬁeld
effect transistors, and the performance of these devices was assessed. Nanowire
resistance measurements were done using nanoscale probes so an accurate, non-invasive
measurement could be obtained. Unexpected transport characteristics were also observed
while measurements took place, and indicated phase-speciﬁc transport.
Cathodoluminescence studies were carried out to obtain electronic structure information

including the band gap and if impurities were present.

80

 

In this chapter two terms that distinguish speciﬁc crystal elements of the nanowire
will be used: phase and domain. Phase is used to distinguish between the zinc-blende and
wurtzite crystalline structures. Therefore the term multiphase is used to describe the two
different phases that exist in the nanowire. The term domain will be used to distinguish
different crystalline structures within the nanowire that are separated by a boundary or
interface. Domain can be used to distinguish, for instance, two wurtzite, two zinc-blende
or one wurtzite and one zinc-blende crystalline structure that are separated by an

interface.

81

4.1 Nanowire Structure Investigations

The nanowire structure was primarily investigated with TEM techniques,
including plain-view HRTEM, selected area electron diffraction (SAED) patterns, fast
Fourier transforms (FFTS), energy dispersive x-ray spectroscopy (EDS), electron energy
loss spectroscopy (EELS), and cross-section HRTEM. Investigations of nanowires that
grew at 850° C, 950° C, and 1000° C were carried out. Multiphase nanowires were
observed at each temperature, but wurtzite single-phase nanowires were also observed at

1000° C. This was conﬁrmed by cross-section HRTEM investigations where results

 

from multiphase nanowires from 850° C and 9500 C as well as wurtzite single-phase L
nanowires that grew at 1000° C were observed and will be presented. EDS and EELS

were used for elemental composition conﬁrmation.

4.1.1 Plain-View TEM

Extensive plain-view TEM studies were done in order to solve the crystallography
of the multiphase nanowire structure“. In these investigations the multiphase structure of
the nanowires and the nanowire growth directions of both the zinc-blende and wurtzite
phases were identiﬁed. This nanowire structural analysis was done using HRTEM
images, SAED patterns, FF Ts, EDS, and EELS. Each technique was critical in assessing
the crystallography and stoichiometry of the nanowire. SAED patterns and F FTs of
HRTEM images were used to distinguish the zine-blende and wurtzite phases and to
determine the growth direction of each phase. EDS was used to estimate the elemental

composition. EDS spectra consistently showed, within an acceptable margin of error of

82

:10%, that the nanowires contained gallium and nitrogen in a 1:1 ratio. EELS was used
to conﬁrm the presence of gallium and nitrogen along with the lack of oxygen.

Nanowire TEM sample preparation consisted of just a few steps. First the
nanowire/matrix sample was placed in a small plastic vial ﬁlled with ethanol.
Ultrasonicating the nanowire/matrix/ethanol mixture for l-3 seconds released nanowires
from the matrix into suspension. Once in suspension the nanowires were petted out and
dropped onto a copper TEM grid with carbon lacy support ﬁlm spanning the holes on the
grid. The TEM grid was allowed to dry for a few minutes and was left with several
nanowires as well as many bits of matrix on the support ﬁlm.

A typical HRTEM image of a multiphase nanowire that grew at 850° C is shown
in ﬁgure 4.1. The multiphase structure is clearly evident in this image. Figure 4.l(a)
shows a TEM image indicating the width, roughly 120 nm, of the nanowire. Figure
4. l (b) shows an HRTEM image of the area indicated in 4.l(a). A distinct phase
transition between the zinc-blende and wurtzite crystalline phases can be seen and is
indicated by the dashed line. It was very sharp, only a few atomic layers thick, with no

observable defects.

83

Phase Transition

Zinc-Blende

Wurtzite

  

20nm . ‘ 3nm

Figure 4.1. TEM and HRTEM images ofa multiphase GaN nanowire that grew at 850° C. (a) The TEM
image shows the width ofthe nanowire. Contrast variations indicate changes in thickness as well as phase
across the nanowire. (b) An HRTEM image ofthe indicated boxed area in (a). The zinc-blende and
wurtzite domains are clearly distinguishable in this image along with the sharp phase transition denoted by
the dotted line. Images: B.W. Jacobs

An example of multiphase nanowire crystal phase and orientation determination
using an SAED pattern and FFTs is shown in ﬁgure 4.2. Using diffraction. different
crystalline domains can be distinguished and their orientations relative to the nanowire
growth direction and their orientations relative to each other can be identiﬁed. A typical
HRTEM image of a nanowire that grew at 850° C is shown in ﬁgure 4.2(a). It indicates
the complicated structure of the nanowire where multiple crystalline domains in different
orientations can contribute to the image. The wave-like features in the bottom right are
likely moire’ fringes resulting from similar stacked periodic structures”. Figure 4.2(b)
shows an SA ED pattern taken of this nanowire. A zinc-blende phase and a wurtzite
phase contributed to this pattern. The zinc-blende pattern is along the [001] zone axis
and the wurtzite pattern is along the [0001] zone axis and it has been solved accordingly.
Figures 4.2(c) and 4.2(d) are F F Ts taken from the indicated areas in ﬁgure 4.2(a). These

FF Ts identiﬁed the zinc-blende and wurtzite domains that contributed to the SAED

84

pattern. The growth direction of this nanowire was along the [2T T0| direction in the

wurtzite phase and along the [011] direction in the zinc-blende phase as indicated by the

SAED pattern.

      

image fa nan wire that grew at 8500 C, and SAED and FFT analysis illustrating
how the FFT can augment SAED patterns. (a) An HRTEM image ofa nanowire shows the multiphase
structure of the nanowire. (b) An SAED pattern of the nanowire showing both zinc-blende and wurtzite
contributions. The zinc—blende SA ED pattern was along the [001] axis and the wurtzite was along the
[0001] axis. (c) FFT of the wurtzite phase taken from the boxed area in (a). (d) FFT of the zinc-blende
phase taken from the boxed area in (3). These FFTs separate out the zinc-blende and wurtzite contributions
in the SA ED pattern, and identiﬁed where these contributions were located in the nanowire. Images and
interpretation: B.W. Jacobs

 

Every nanowire grown at 850° C imaged to date has been multiphase. Figure 4.3
shows an HRTEM image of a nanowire grown at 1000° C with a wurtzite single—phase
structure. The growth direction of this nanowire is the same wurtzite growth direction for
the nanowires grown at 850° C, [2T T0]. The moire' fringes may result from different
domain orientations within the nanowire. The formation of the single-phase nanowire

may have to do with different growth kinetics present at higher temperatures.

85

 

[0001] . _ ,. _. _
Figure 4.3. (a) An HRTEM image ofa single-phase wurtzite nanowne grown at l000"
growth direction is the same as the nanowires grown at lower temperatures. IZTTOI. No zinc-blende
phase was observed. (b) This SAED pattern was solved for the wurtzite crystal structure along the [000i]
zone axis. Images and interpretation: B.W. Jacobs

C The wurtzite

Multiphase nanowires with very similar structures to those found at 850° C were
observed at 1000" C in addition to the single-phase nanowires. In comparing a
multiphase nanowire grown at 1000" C, ﬁgure 4.4(a), to one grown at 850° C, ﬁgure
4.4(b), they were very similar. The sharp transition between the zinc-blende and wurtzite
phases is evident in each ﬁgure. The nanowire growth directions are the same, and

similar FFTS indicate similar crystal orientations.

86

    

Figure 4.4. HRTEM images of nanowires grown at 10000 C and 850° C. (a) An HRTEM image ofa
nanowire grown at 10000 C. Zinc-blende and wurtzite phases were identiﬁed using FFTs taken from the
indicated areas. A sharp phase transition was evident. The arrows indicate the nanowire growth direction
for each phase. (b) An HRTEM image ofa nanowire grown at 8500 C. A very similar structure was
observed, with similar FFTs and the same growth directions for each phase. Images and interpretation:
B.W. Jacobs

Single-phase wurtzite nanowires grown at 1000° C, which had a different grth
direction than the nanowire shown in ﬁgure 4.3. were also observed. With growth along
the [0001] direction, these nanowires tended to be much thicker in width, ranging from
200 nm to 10 um. Figure 4.5(a) shows a single-phase nanowire grown at 1000° C with
growth along the [0001] direction. Note the tapered tip of the nanowire; this feature will
be discussed in section 4.3. Figure 4.5(b) shows an HRTEM image taken from the boxed
area in 4.5(a) with a well-resolved lattice indicating the high crystal quality of the
nanowire. Figure 4.5(c) shows an SAED pattern and was solved for the wurtzite
structure along the [OI TO] zone axis. Nanowires with this growth direction were
speciﬁc to this growth temperature, and no nanowires have been observed to date with

this growth direction at the lower temperatures.

87

“Ill: 5I\ 3.;

[mt-1' :3.
all: ndni iii

In!“ an.

eugfsil

mine; i;

HILL

  

A.

r:

     

.\‘

[0001 ]

 

101101 . , _ .-

Figure 4.5. TEM images ofa nanowire grown at IOOOC C. (a) A TEM images shows the nanowire end.
with its tapered tip. It was about 700 nm wide. Its growth was along the [000]] direction. (b) An HRTEM
image taken from the boxed area in (a) shows a well-resolved lattice indicating the high_crystal quality of
the nanowire. (c) An SAED pattern was solved for the wurtzite structure along the l()| l0] zone axis.
Images and interpretation: B.W. Jacobs

Obtaining HRTEM images of the nanowires, as shown in the ﬁgures above. was
not usually difficult. Many nanowires were very thin, roughly 60-150 mm in width, with
edges thin enough to obtain high-resolution images. The center of the nanowires is
normally too thick to obtain clear HRTEM images. Usually. when imaging in the TEM,
tilting to a low index zone axis. such as [l I l] or [001] in zinc-blende. is done to
understand the orientation of the crystal and to obtain the clearest high-resolution images.
This is very difﬁcult to do when imaging nanowires, however. As mentioned, the

nanowire was usually very thin, so when tilting, the nanowire tended to move out of the

viewing area and above or below eucentric height. Eucentric height is when the sample

88

 

is in focus and the objective lens current is at its optimum value. Therefore, the nanowire
must be continually moved back into the viewing area and its location within the
objective lens continually adjusted to eucentric. Many times, simply due to the
multiphase structure of the nanowire, the orientation of one domain will allow for clear
high-resolution images. Tilting to a low index zone axis can be done but requires
patience on the part of the user.

Adding to the difﬁculty of obtaining clear images at a speciﬁc zone axis, the zinc-
blende and wurtzite crystalline structures are quite similar, and the distinction between
certain diffraction patterns may not always be clear. Several important diffraction
patterns in the zinc-blende and the wurtzite phases are very similar, as shown in ﬁgure
4.6. When analyzing similar patterns, the spot intensity can be used as a clue as to which
pattern is which. However, since the TEM beam is usually not exactly perpendicular to a
set of low index planes, the Bragg condition may not be exactly met, giving different spot
intensities than expected. Stacked domains can complicate the pattern. Also multiple
reﬂections that have the same interplanar spacing and are at the same orientation will add
to the spot intensity making it much brighter, as shown in ﬁgure 4.2(b). So after a
diffraction pattern is obtained, special care must be taken in determining which phase it

corresponds to.

89

[0001]

[1072]
Figure 4.6. Simulated diffraction patterns highlighting similar features found in some wurtzite and zinc-
blende patterns. (3) [000l] zone axis for the wurtzite structure. (b) [l l 1] zone axis for the zinc-blende
structure. (c) | IO 1 2| zone axis for the wurtzite structure. (d) [1 l2] zone axis for the zinc-blende structure.
Images and interpretation: B.W. Jacobs

 

FFTS play an important role in multiphase nanowire crystal analysis since many
times the different crystalline domains within the nanowire are too small to easily
distinguish in a SAED pattern. Therefore, using FFTS taken from HRTEM images.
SAED patterns can be linked to speciﬁc phases and domains within the nanowire, as
shown in ﬁgure 4.2. However, FFTS do have their drawbacks. No speciﬁc spatial
information is contained in an FFT. In real diffraction patterns, information about the
spatial scale, based on the magniﬁcation and how far the diffraction pattern image is
projected below the sample, called the camera length, is provided. Using this information
the interplanar spacing for a given material can be calculated with a high degree of
accuracy and compared against known values for conﬁrmation. This information is key

in determining the crystallography of a material. FF Ts do not provide this information

90

directly. In order to successfully conﬁrm the pattern generated with an FFT, it must be
backtracked to a known diffraction pattern.

If the FF T can be backtracked to a known diffraction pattern, an FFT can be a
powerful tool. For example, if an FFT is backtracked to a real diffraction pattern it can
be compared against other FFTS that are taken at the same magniﬁcation and from the
same size area. To take an FFT of an HRTEM image, a certain area of an HRTEM image
is selected in a program such as Imagelzé. This size of this area, usually a square,
determines the size of the FF T along with the magniﬁcation of the image. As long as the
same size square at the same magniﬁcation is used spatial information can be extracted
from FFTS, making them an invaluable tool for crystal analysis, augmenting SAED
patterns.

In addition to structural analysis, EDS in the TEM was used to investigate the
elemental composition of the nanowires. It was essential to verify the purity of the GaN
nanowires. Possible nanowire impurities, such as oxygen, were purposely removed from
the growth chamber, and of the lack of these impurities in the nanowires needed
veriﬁcation. A common nanowire is gallium oxide (GazO3), and oxygen was not
observed in the EDS spectra. Figure 4.7 shows an EDS spectrum with peaks occurring at
speciﬁc energies corresponding to the elements present within the nanowire. Other peaks
including copper and carbon were present as a result of the copper TEM grid and carbon
lacy support ﬁlm spanning the copper grid holes. The EDS spectra measured gallium and

nitrogen in a 1:1 ratio.

91

Cu

1

Ga Ga

1

L l

Intensity (Arbitrary Units)

 

 

 

 

 

 

 

 

Cu
ll Jill
,

"77'

 

 

T ‘ f V I ' I ' l T I

o 2 Erlergy (NEW 10
Figure 4.7. EDS spectrum of a GaN nanowire. The nitrogen and gallium peaks indicate the purity of these
nanowires. A copper peak is also present due the copper TEM grid used to support the nanowire sample.
Spectrum and interpretation: B.W. Jacobs

EELS was also used in elemental conﬁrmation. EELS is very sensitive technique
that can detect small amounts of a given element within a material. Using EELS, the
presence of nitrogen, ﬁgure 4.8(a), and gallium, ﬁgure 4.8(b), were detected along with
the absence of oxygen. Oxygen was an unwanted impurity and great care was taken to

eliminate oxygen in nanowire growth.

92

       

 

 

 

 

 

. (a) . (b)
.. ‘ Nitrogen Edge "‘ .
ﬁg ‘ 401w '5 .
a ‘ 3 .
r: E j Gallium Edge
g E ‘ 11150V

1 e .
3 s
z. “ E -i
=- 2
it 3.
- "l n-

3001400Y420TJ0'400'480Y50015201SEOVGSO‘ 1080 1120 1160 1200 1240
Energy (eV) Energy (eV)

Figure 4.8. EELS spectra of a GaN nanowire. (a) The nitrogen core loss is located at 40l eV, and a strong :
signal was measured there indicating a signiﬁcant presence of nitrogen. The oxygen core loss at 532 eV is
located on this spectrum but no peak was detected conﬁrming the lack of oxygen. (b) The gallium core

loss is located at I I IS eV, and a strong signal was measured there indicating a signiﬁcant presence of '
gallium. Spectra and interpretation: B.W. Jacobs

Plain-view HRTEM analysis proved very beneﬁcial in determining the structure
and stoichiometry of the multiphase nanowires. The multiphase structure was identiﬁed
in HRTEM images and using SAED patterns and FFTs. The nanowire purity was
veriﬁed using EELS and EDS. However, several questions remained after plain-view
HRTEM analysis. Namely, how the nanowire crystalline domains were oriented relative
to each other, and if sharp defect-free zinc-blende to wurtzite interfaces always formed.
Usually the HRTEM images were too complicated to answer these questions and only so
much information could be extracted from them. Therefore, other sample preparation
techniques were discussed and explored. C ross-section HRTEM analysis proved the best

possibility for obtaining further structural information and was pursued.

93

4.1.2 Cross-Section HRTEM

The internal structure of the multiphase nanowires was difﬁcult to determine
using only plain-view HRTEM techniques. Complicated SAED patterns were often
difﬁcult to solve since multiple crystalline domains could contribute to the pattern.
SAED patterns could also be somewhat distorted because the low index crystal
orientation may not be perpendicular to the electron beam resulting in a Bragg condition
that was not exactly met. Therefore, in order to gain a better understanding on how these
nanowires form, on axis, or cross-section HRTEM nanowire studies were considered.

Several techniques can be used for cross-section TEM sample preparation. A
popular method is embedding a sample in hard resin and using a diamond tipped blade to
out very thin cross-sections. Another method involves placing the sample in liquid
nitrogen and using a mortar and pestle to grind the sample in hopes of getting a small
enough cross-sectioned piece of material. The former technique was attempted in
making a cross-section of a GaN nanowire, but proved difﬁcult. A third technique was
then explored: using a focused ion beam to prepare individual nanowire cross-sections.
This sample preparation technique has been explained in detail in chapter 2. This method
proved difﬁcult at ﬁrst, but after many failed attempts successful samples were fabricated
with spectacular results.

Cross-sections were fabricated from the three different growth temperatures
mentioned above, 850° C, 9500 C and 1000° C. Figure 4.9 shows a cross-section from
each growth temperature. The cross-sections taken from the nanowires grown at 850° C
and 950° C were single nanowires, and the cross-sections from each temperature had a

unique structure. but both had a triangle shaped. cross-section. A multiple nanowire

94

 

cross-section was obtained from nanowires grown at lOOO° C. This was made possible
by a small group of nanowires that held together at the matrix base after ultrasonication.
These nanowires also had unique features, as will be discussed.

Figure 4.9(a) shows a cross-section taken from a nanowire grown at 850° C.
Figure 4.9(b) shows a cross-section taken from a nanowire grown at 950° C. These two
multiphase nanowires had the same growth directions, [2T T0] for the wurtzite phase and
[01 l] zinc-blende phase. They had multiple wurtzite and zinc-blende domains in unique
conﬁgurations, which will be discussed in detail. Figure 4.9(c) shows a cross-section
taken from a group of multiple nanowires grown at lOOO° C. There are ﬁve individual
nanowires all with hexagon-shaped cross-sections. The very large nanowire on the left
was likely two nanowires that grew together. The individual nanowires in this cross-
section ranged between about 200 nm to 1.5 pm in width. These nanowires were single-
phase wurtzite with growth in the [0001] direction. Figure 4.9(d) shows an SEM of the
group of nanowires. A dotted line denotes where the cross-section was taken. Figure
4.9(c) shows a close up of where the cross-section was taken. The silicon substrate on
which the nanowires were placed, was tilted 52° relative to the electron beam in this
image. The small nanowires are more easily seen here and the nanowire nucleation sites
are also apparent. Figure 4.9(t) shows the nanowires embedded within the Pt/Au stack
midway through the cross-section extraction process. The trenches on either side of the
cross-section have been sputtered away, and at this stage the cross-section was ready for
the U-cut. An arrow points out the location of the three large nanowires. The ion beam
was used to create this particular image. Figures 4.9(g) and 4.9(b) are close ups of the

two small nanowires, and their locations in the cross-section are indicated in 4.9(c).

95

      

   

Figure 4.9. TEM images of cross—sections fabricated with the FIB. (a) Cross-section HRTEM image ofa
nanowire grown at 8500 C. (b) Cross-section HRTEM image ofa nanowire grown at 950° C. (c) Cross-
section TEM image of multiple nanowires grown at 10000 C. (d) SEM image ofthe group of nanowires
from which the cross-section was taken. The dotted line denotes where the cross—section was taken. (e) A
close up ofthis group shows each individual nanowire nucleating from the matrix. Both the large and
small nanowires can be seen here. (D An ion beam images shows the cross-section halfway through the
extraction process. The arrow points out where the three large nanowires were located. The nanowires
were on top ofa silicon substrate and under a thin layer ofgold and a thick layer of platinum to protect
them from ion beam damage. (g) A TEM image ofa small nanowire. Its location within the nanowire
group is indicated in (c). (h) A TEM image ofa second small nanowire. Its location within the nanowire
group is also indicated in (c). Images: B.W. Jacobs

96

Detailed HRTEM analysis of a cross-section taken from a nanowire grown at
850° C is shown in ﬁgure 4.10. The nanowire was divided into ﬁve distinct regions that
separated each crystal domain present in the nanowire. These regions were distinguished
by domain interfaces denoted by the dotted line in the HRTEM image in ﬁgure 4.10.
FFTs taken from each region are shown on the right. FFTS of regions 1-4 were solved for
the wurtzite structure along the |2T TOI direction, which coincided with the growth
direction for the wurtzite phase found in the plain-view TEM investigations described in
the previous section. The FFT from region 5 was solved for the zinc-blende structure
along the [01 1] direction, which also coincided with the growth direction for the zinc-
blende phase found in the plain-view TEM investigations. The arrows in the FF Ts point
in the [0001] and [111] directions in the wurtzite domains and the zine-blende domain,
respectively, to emphasize the different orientation of each domain.

Low energy planes parallel to the nanowire grth direction faceted, which
resulted in the overall triangular cross-section shape as well as in the individual triangular
domains. In each wurtzite domain these planes were the {IOT l }, which made up two
sides of the triangle and the (0001) plane, which made up the third side. Note, however,
that the (000]) planes were always found as internal interfaces, they were never found as
external facet planes. In the zinc-blende structure the {l l 1} planes faceted, and in region
5 are responsible for the diamond shaped domain. The crystal models at the bottom of
this ﬁgure indicate which planes were responsible for faceting as well as the growth
planes in both the zinc-blende and wurtzite structures perpendicular to the nanowire
growth direction. The metal layers on top of the nanowire are also indicated. Titanium,

gold and platinum were deposited on the nanowire to protect it from ion beam damage.

97

 

Figure 4.10. Details ofthe cross-section taken from the nanowire grown at 8503 C. This nanowire was
separated into ﬁve regions corresponding to its distinct crystalline domains. These regions were divided
along domain interfaces. FFTs from each region are shown on the right. Regions l-4 were wurtzite and
region 5 was zinc-blende. The arrows in the FFTs indicate the [0001] and [l l 1] directions in the wurtzite
and zinc-blende domains. respectively. Zinc-blende and wurtzite crystal models show which planes were
responsible for faceting resulting in the triangular shape of the nanowire. In the wurtzite structure the

{ IOT I} and (000l ) planes faceted and were parallel with the growth direction, while the { ZTTO} planes
were perpendicular to the growth direction. In the zinc-blende structure the { l l l } planes faceted and were
parallel to the growth direction. while the {OI l } planes were perpendicular to the growth direction. Images
and interpretation: B.W. Jacobs

Figure 4.1 I shows four HRTEM images of certain areas of the cross-section taken
from the nanowire grown at 850° C, as indicated by the boxed areas in the center inset.
These images highlight the coherent and incoherent interfaces present within the
nanowire, as well as the different orientations of each crystalline domain. Multiple

interfaces seen within the nanowire in previous plain-view HRTEM investigations were

98

clearly delineated in these cross-section HRTEM images. Multiple coherent interfaces
were seen, which explained the extremely sharp, defect free phase transitions observed in
the plain-view images. In ﬁgure 4.1 1(a) several coherent and incoherent interfaces are
denoted by the solid and dashed lines, respectively. A crystalline domain with dark
contrast is apparent in this ﬁgure and in ﬁgure 4.1 l(b), and consisted of a wurtzite
structure, possibly in a slightly different orientation, giving it darker contrast than the L-
surrounding areas. There are two zinc-blende domains on the top and left of the dark
wurtzite domain that formed coherent interfaces with the surrounding wurtzite domains

denoted by the solid lines. The dark wurtzite domain formed incoherent interfaces with

 

‘5”5-. ‘1

all domains that surround it as denoted by the dashed lines. In ﬁgure 4.1 l(c) a long
incoherent interface between a wurtzite and zinc-blende domain is denoted by the dashed
line. Figure 4.1 l(d) is an image taken further along this incoherent interface and also
shows a long coherent interface between the [11 l] and [000]] planes in the zinc-blende
and wurtzite domain, respectively, as denoted by the solid line. This interface extended
from the center of the nanowire 80 nm to the external edge of the nanowire. Each
HRTEM image had well resolved lattices indicating the high crystal quality of each

domain.

99

 

  

Figure 4.1 I. HRTEM images of the cross-section taken from the nanowire grown at 850° C. (a) A close
up from the corresponding boxed area in the center inset revealed several distinct crystalline domains with
coherent and incoherent interfaces. as denoted by the solid and dashed lines. respectively. A dark contrast
wurtzite domain was observed and could be due to a slightly different orientation. (b) The dark contrast
wurtzite domain. from the corresponding boxed area in the center inset, is surrounded by both zinc-blende
and wurtzite domains forming incoherent interface at every edge. as denoted by the dashed lines. The zinc-
blende domains form coherent interfaces with the surrounding wurtzite domains. as denoted by the solid
lines. (c) A close up of an incoherent interface, denoted by the dashed line, from the corresponding boxed
area in the center inset. (d) This image shows the same incoherent interface at a different location. On the
right side ofth is ﬁgure a long coherent interface that extends from the center of the nanowire to the outside
edge forms between the (I I I) and (000I ) planes. as denoted by the solid line. Images and interpretation:
B.W. Jacobs

Detailed HRTEM analysis of the cross-section taken from the nanowire grown at

950° C is shown in ﬁgure 4.12. The nanowire was divided into six distinct regions that

100

separated each crystal domain present in the nanowire. These regions were distinguished
by domain interfaces denoted by the dotted line in the HRTEM image in ﬁgure 4.12.

F F Ts taken from each region are shown on the right and bottom. FFTs of regions 1-5
were solved for the wurtzite structure along the |2T TOI direction. The F F T from region
6 was solved for the zinc-blende structure along the [01 1] direction. These directions
corresponded to the growth directions found for multiphase nanowires in the plain-view
HRTEM investigations. The arrows in the FFTs point in the [0001] and [l 11] directions
in the wurtzite domains and the zine-blende domain, respectively, to emphasize the
different orientation of each domain.

Similar to the cross-section shown in ﬁgure 4.10, this nanowire faceted on the
same wurtzite planes, the {IOT I} and (0001). There was also a zinc-blende domain in
the center of the nanowire that faceted internally on the {111} planes and was also
diamond shaped. Similar to the cross-section shown in ﬁgure 4.10, the (000]) planes

were all found as internal interfaces.

lOI

 

.
u..
o
'-
a
~
.
N
a u
5:
.....
'l-
I.

n.
.u u
........

'-
o
‘‘‘‘‘‘‘
..

 

Figure 4.12. Details ofthe cross—section taken from the nanowire grown at 9500 C. This nanowire was
separated into six regions corresponding to distinct crystalline domains. These regions were divided along
domain interfaces. FFTs from each region are shown on the right and bottom. Regions l-5 were wurtzite
and region 6 was zinc-blende. The arrows point in the [0001] and [I l I] directions in the wurtzite and zinc-
blende domains, respectively, to emphasize differences in orientation. Images and interpretation: B.W.
Jacobs

Figure 4. I 3 shows four HRTEM images of certain areas of the cross-section taken
from the nanowire grown at 950° C. as indicated by the center inset. These images
highlight the coherent and incoherent interfaces present within the nanowire, as well as
the different orientations of each crystalline domain. Figure 4. l 3(a) shows an incoherent
interface between two wurtzite domains. Stacking faults were also seen in this image,
and ran horizontally through the center of the ﬁgure. In ﬁgure 4. l 3(b) two incoherent
interfaces are shown between 3 wurtzite domains. The areas shown in this image also
show possible stacking faults on the left and bottom of the ﬁgure that transitioned to

wurtzite crystal structures. Figure 4.l3(c) shows two wurtzite domains with an

102

incoherent interface. Possible stacking faults were seen in the right hand side of this
image. The amount of zinc-blende crystal structure relative to the wurtzite structure was
much lower than in the cross-section shown in ﬁgure 4.10, and was only observed in
region 6, which was found in the very center of the nanowire, as shown in ﬁgure 4.1 3(d).
This zinc-blende domain formed a coherent interface with region 1 between the (l l l) and
(0001) planes. A region of light contrast was observed in the right side of this ﬁgure.
This may be due to a thinner area resulting in lighter contrast relative to the surrounding
areas. Each HRTEM image had well resolved lattices indicating the high crystal quality

of each domain.

103

 

Figure 4.13. IIRTEM images ofthe cross-section taken from the nanowire grown at 950‘” C. (a) A close
up from the indicated area in the center inset revealed two distinct crystalline domains with incoherent
interfaces. Possible stacking faults ran horizontally across the center ofthis image. (b) This image shows
three distinct wurtzite domains with two incoherent interfaces. (c) Two wurtzite domains with an
incoherent interface that changed its direction running from the center to the outside edge ofthe nanowire.
(d) This image shows the center zinc-blende domain and a coherent interface between the zinc-blende and
wurtzite domain. The top and bottom domain surrounding the zinc-blende domain were possible stacking
faults that transitioned to wurtzite crystal structures. Images and interpretation: B.W. Jacobs

Detailed analysis of three nanowires in the cross-section taken from nanowires
grown at lOOO° C revealed a different nanowire growth direction from those grown at the
two lower temperatures. With hexagon-shaped cross-sections these nanowires grew in

the [0001] direction. and were all wurtzite single-phase nanowires. Nanowires that grew

104

in this direction had a larger range of observed widths, roughly 200 nm to several
microns, than the nanowires grown at the lower two temperatures. The single-phase

nanowires that grew in the [0001] direction also tended to have larger widths on average

than the multiphase nanowires grown at this temperature as well as the lower two

temperatures.

Figure 4.14 shows details of a large nanowire, roughly 1.5 pm in width. The
nanowire was divided into two regions, as shown in ﬁgure 4.14(a). The wurtzite crystal
model at the right of this ﬁgure showed that the planes parallel to the growth direction,
the { 1010} planes, faceted. The (0001) planes were perpendicular to the growth
direction. Figure 4.14(b) shows an HRTEM image of region I with a well-resolved
lattice that indicated the high crystal quality of the nanowire. The F F T of region 1 was
solved for the wurtzite structure along the [0001] direction. Figure 4.14(c) shows a close
up image of region 2, which revealed a small wurtzite domain with darker contrast than
the surrounding region 1. This small domain had a hexagonal shape. However, an

HRTE M image not obtained, so its actual crystal structure could not be determined.

Figure 4.14(d) shows the entire cross-section and indicated the location of this nanowire.

105

/ ( (mu/1

 

Figure 4.14. HRTEM images ofthe cross-section taken from the nanowire grown at lOOO° C. (a) A large
nanowire with a width ofroughly 1.5 pm is shown. A wurtzite crystal model shows the facet planes
parallel to the growth direction, { IOTO}, that resulted in the hexagonal cross-section shape. and the growth
plane. (000l ). perpendicular to the growth direction. (b) A high-resolution image of region 1 showing the
high crystallinity ofthe nanowire. An FFT of region I that could be solved for the wurtzite crystal
structure in the [0001] direction is shown in the inset. (c) A close up image ofregion 2 revealed a dark
contrast wurtzite domain. (d) TEM image ofthe entire cross-section indicating where this nanowire cross-
section was located. Images and interpretation: B.W. Jacobs

Figure 4.15 shows details of a smaller nanowire, roughly 200 nm in width. The
nanowire was divided into three regions, as shown in ﬁgure 4.15(a). Regions 2 and 3 at
the center of the nanowire were areas of lighter contrast relative to the surrounding region
1. FFTS of each region shown on the right hand side of this ﬁgure all indicated that the
nanowire was single—phase wurtzite growing along the [000]] direction. Figure 4.15(b)
shows the entire cross-section and indicates the location of this nanowire. Figure 4.15(c)
shows an HRTEM image of regions 2 and 3. These regions were observed to be

hexagonal in shape. The F FTs of these two regions were in the same orientation of

106

region 1 and no detectable interfaces were observed between these regions. Figure
4.15(d) shows an HRTEM image of region 1 with a well-resolved lattice that indicated

the high crystal quality of the nanowire.

 

‘

Figure 4.15. HRTEM images ofthe cros taken from the nanowire grown at 1000° C. (a) A small
nanowire with a width of roughly 200 nm is shown here. It grew in the same direction as the nanowire
shown in ﬁgure 4.15. It was divided into three regions. Regions 2 and 3 were lighter in contrast relative to
the surrounding region I . The FFTS shown on the right were all solved for the wurtzite structure along the
[0001] direction. They were all in the same orientation. (b) A TEM image of the entire cross-section
showing where this nanowire cross-section was located. (c) A close up of region 2 and 3 revealed that the
light contrast regions were both hexagonal in shape. (d) A high-resolution image of region 1 showed the
high crystallinity of the nanowire. Images and interpretation: B.W. Jacobs

 

Figure 4.16 shows details of a larger “double" nanowire, roughly 2 pm in width at
its widest point. The nanowire was divided into four regions, as shown in ﬁgure 4.16(a).
Regions 1 and 2 comprised the left and right nanowires that made up the double

nanowire, respectively. FFTs of regions 1 and 2 are shown at the bottom of this ﬁgure.

107

They were solved for the wurtzite structure along the [0001] direction. They were also in
the same orientation. Figure 4.16(b) shows the entire cross-section and indicated the
location of this nanowire. Figure 4.16(c) shows an HRTEM image of region 3. This
image shows a small platelet that seemed to partially cover a hole in the center of the
nanowire. 1f the small platelet were to be removed, the hole would be hexagonal in
shape, indicating that it faceted on the {1010} planes, the same facet planes as on the
external surface of the nanowire. Figure 4.16(d) shows an HRTEM image of region 4.
This region also had a hole surrounded by an area of light contrast that is probably due to

a thinner area.

108

     

Figure 4.l6. HRTEM images of the cross-section taken from the nanowire grown at 1000" C. (a) A large
"double" nanowire with a width of roughly 2 pm at its widest point is shown here. It grew in the same
direction as the nanowire shown in ﬁgure 4.l5 and 4.16. It was divided into four regions. Regions I and 2
are the leﬁ and right nanowires that made up the double nanowire. Regions 3 and 4 were areas with lighter
contrast relative to the surrounding regions I and 2. The FFTs shown on the bottom of the image were
taken from regions I and 2. and were solved for wurtzite structure along the [000 1] direction. They were in
the same orientation. (b) A TEM image of the entire cross-section indicates where this nanowire cross-
section was located. (c) A close up of region 3 indicated that the light contrast region was hexagonal in
shape. A small platelet was partially covering a hole in the center ofthe nanowire. (d) A close up of
region 4 also shows an area oflighter contrast with a hole at its center. Images and interpretation: B.W.
Jacobs

Each cross-section investigated with HRTEM revealed a somewhat different
nanowire structure. The cross-sections taken from nanowires grown at 8500 C and 950°
C were different in the domain orientations as well as the wurtzite to zinc-blende ratios.
Each crystalline phase in these nanowires. however. grew in the same direction and
faceted on the same planes. The nanowires grown at 10000 C had more similarities in
their structure but were different than the nanowires grown at the lower two temperatures.
The nanowires were single-phase wurtzite, with growth along the [0001] direction. There

were subtle differences between these nanowires also. Small anomalous domains with

109

light or dark contrast were observed that can not as yet be fully explained. They may be
due to thickness variations or slightly different crystal domain orientations resulting in
contrast differences.

The evidence for a new multiphase crystalline homostructure in GaN nanowires
based on the analysis by HRTEM with SAED, FFT, EDS, EELS, and cross-section
analysis has been conﬁrmed. Wurtzite and zinc-blende domains were observed with
coherent and incoherent interfaces between the domains. Facet planes were identiﬁed
that are responsible for the cross-sectional shape of each nanowire. Reliable growth of
the multiphase nanowires as well as single-phase nanowires is a result of vapor solid

growth, and will be discussed next.

110

4.2 Nanowire Matrix Investigations

In order to understand how the multiphase and single-phase nanowires grow, the
matrix on which the nanowires nucleate was investigated. Using SEM, nanowires have
been shown to nucleate from the sides and points of hexagonal platelets that form early in

(122‘23 ‘24. The SEM investigations presented in

the growth process as previously reporte
this thesis agreed with those previously reported ﬁndings for matrix that formed at 850°
C, 950° C and 1000° C, in that nanowires nucleated from the sides of hexagonal platelets.
Further investigations of matrix that formed at lOOO° C revealed three distinct regions of
matrix with different feature sizes. In the matrix that formed at 1000° C, in addition to
the nucleation sites on the sides of the hexagonal platelets, a second type of nucleation
site was identiﬁed. This nucleation site was located on the tops of hexagonal platelets.

After SEM investigations established where nanowire nucleation was occurring.
at the sides of hexagonal platelets, HRTEM investigations were carried out to determine
if the nucleation sites had features smaller than those resolved in the SEM. The HRTEM
investigations revealed a network of structures on the sides of many hexagonal platelets
in the form of nanoscale ledges that were in the same orientation and coherent with the
hexagonal platelet.

The SEM and HRTEM investigations of the matrix proved essential in
understanding the growth mechanisms involved for multiphase and single-phase

nanowire formation. Without the study of the grth matrix, most information about the

mechanism involved in multiphase nanowire growth would be unknown.

11]

4.2.1 SEM

Matrix features and nanowire nucleation sites that formed at 850° C. 950° C and
10000 C were investigated with SEM. Distinct differences in the matrix features that
formed from each temperature were noted and will be discussed.

Matrix features that formed at 850° C are shown in the SEM images in ﬁgure
4.17. These feature sizes were around 200 nm to 1 pm in size. Figure 4. 1 7(a) shows an
SEM image of the matrix at low magniﬁcation. Figure 4.l7(b) shows another SEM
image at higher magniﬁcation so that subtle features could be resolved. A range of

feature sizes is apparent in this image. Two nanowires, indicated by the arrows. have

nucleated from the matrix.

   

    

   

   

L» ,
liar . k" ,-

' -‘ i- . . ' < ' A- i 71' .'
Figure 4.17. SEM images ofthe matrix that formed at 850° C. (a) The matrix is relatively uniform with
feature sizes of around 200 nm to 1 pm in size. (b) A close up image shows the features in detail. There
are many small and medium sized GaN crystal formations. Two nanowires in this image. as indicated by
the arrows, have nucleated from the matrix. Images: K. McElroy

Figure 4.18 shows SEM images of nanowire nucleation sites on the matrix that
formed at 850° C. In ﬁgure 4.18(a) a nanowire has nucleated from the comer of a
hexagonal platelet. In ﬁgure 4.18(b) a nanowire has nucleated from the side of a

hexagonal platelet. Although the nanowires shown in these images have nucleated from

112

different areas of the platelet, they grew in the same direction relative to the hexagonal

platelet orientation, assumed to be along the <2T T0> direction.

(a) r,

     

k ISOOnm , ' 500nm

Figure 4.18. SEM images ofthe matrix that formed at 850° C. (a) A nanowire nucleated from the side of
a hexagonal platelet. The arrows show the probable nanowire growth direction, [ 21 T0] and the sides of
the hexagonal platelet. { IOTO}. (b) A nanowire nucleated from the side ofa hexagonal platelet. The
arrows show the nanowire growth direction and sides of the platelet. Images: B.W. Jacobs

Matrix features that formed at 950° C are shown in the SEM images in ﬁgure
4.19. These feature sizes were around 200 nm to 1 pm in size. Matrix that formed at this
temperature was very similar to the matrix that formed at 950° C. Figure 4.19(a) shows
an SEM image of the matrix at low magniﬁcation. Figure 4.19(b) shows another SEM
image at higher magniﬁcation so that subtle features could be resolved. A range of
feature sizes is apparent in this image. A nanowire, indicated by the arrow, has nucleated

from the matrix.

113

    

         

‘. n.

. - -‘ .: '-., . 5": j§. ‘ .. . .A
Figure 4J9. SEM images of the n atrix that formed at 9500 C. (a) The matrix is relatively uniform with
feature sizes of around 200 nm to 1 pm in size, and similar to the matrix features formed at 850° C. (b) A
close up image shows the features in detail. There are many small and medium sized GaN crystal

formations. One nanowire in this image, as indicated by the arrow, has nucleated from the matrix. Images:
K. McElroy

Figure 4.20 shows SEM images of nanowire nucleation sites for matrix that
formed at 950° C. In ﬁgure 4.20(a) a nanowire has nucleated from the side of a
hexagonal platelet. The arrows show the probable nanowire growth direction, [2T T0],
and the sides of the hexagonal platelets, { IOTO}. In ﬁgure 4.20(b) a nanowire has also
nucleated from the side of a hexagonal platelet. The arrows show the probable nanowire
growth direction, [2T T0], and the sides of the hexagonal platelets, { lOTO}. These

nucleation sites are very similar to those found for matrix that formed at 850° C.

114

       

 

Figure 4. 20. SEM images ofthe matrix that formed at 950° C. (a) A nanowire nucleated from the side of
a hexagonal platelet. The arrows show the probable nanowire growth direction. [2 T IO] and the sides of
the hexagonal platelet, { 10 T 0}. (b) A nanowire is also shown to have nucleated from the side ofa
hexagonal platelet. The arrows show the nanowire growth direction and sides ofthe platelet. Images:
B.W. Jacobs

The matrix features that formed at lOOO° C were quite different from the matrix
features that formed at the lower two temperatures. Three distinct matrix regions were
divided according to their feature size. as shown in ﬁgure 4.21. The smallest features of
around 100 nm to 200 nm in size were observed in region 1. Larger features of around
200 nm to 1 pm in size. similar to the feature sizes found in the matrix that formed at
850° C and 950° C, were observed in region 2. The largest features of around 1 pm to 5
pm in size were observed in region 3. In ﬁgure 4.21(a) an SEM image shows how all
three regions can form together in one small area. Figure 4.21(b) shows an SEM image
of region 1 in the foreground and region 3 in the background to give a sense of scale of
how different these matrix features are in size. The arrow denotes where a nanowire
nucleated from region 1. Figure 4.21(c) shows and SEM image of region 2 located atop
an area of region 3 matrix. This region was comprised of similar features and sizes found
in matrix that formed at 850° C and 950° C. The arrows denote where nanowires
nucleated from this region. Figure 4.21(d) shows an SEM image of region 3. Large

micron size structures are readily seen. This ﬁgure shows a large rod that was not found

115

at the lower growth temperatures. This type of rod is frequently seen in the matrix that

forms at 1000° C.

       

Figure 4.2]. SEM images of the matrix that formed at lOOO° C. (a) The matrix formed distinct features

that could be divided into three regions. Region 1 indicates where the smallest features formed, region 2
the larger features. similar to those found in the matrix that formed at 8500 C and 950° C and region 3 with
the largest features. (b) A higher magniﬁcation image of regions I and 3 shows the features in more detail.
The small features of region I, in the foreground, are compared to the larger features of region 3, in the
background. One nanowire in this image, as indicated by the arrow, has nucleated from region I. (c) A
higher magniﬁcation image of region 2 shows the features in more detail. This region is very similar to the
matrix found to form at 850° C and 950° C. Three nanowire in this image, as indicated by the arrows. have
nucleated from this region. (d) A higher magniﬁcation image ofregion 3 shows the features in more detail.
The large crystal formations are easily seen here. A large rod is seen in the center ofthe image. These rods
were observed in the matrix that formed at 1000“ C and not in the matrix that formed at the lower two
temperatures. Images: K. McElroy

Figure 4.22 shows SEM and TEM images of nanowire nucleation sites on matrix
that formed at lOOO° C. The nucleation sites shown in this ﬁgure were unique to this
growth temperature. Nanowires shown in the SEM image in ﬁgure 4.22(a) have

nucleated from the top of hexagonal platelets, instead of from the side as shown in the

116

ﬁgures above. Figure 4.22(b) shows an SEM image of what may be a nanowire that
started to nucleate from the top of a hexagonal platelet. Figure 4.22(c) and 4.22( d) shows
a TEM image where large and small hexagon-shaped light contrast regions can be seen,
denoted by arrows. These are identiﬁed as holes in the platelet, which can result from
TEM sample preparation during which the matrix is sonicated to release the nanowires.

Large and small hexagon-shaped holes were observed.

           

-. . ...... ’5: 35' ' ........ . . '
Figure 4.22. SEM and TEM images ofthe matrix that ormed at . (a) A SEM image of mu iple
nanowires that grew from the matrix. The arrow indicates where a nanowire nucleated from the top ofa
hexagonal platelet. (b) A SEM image ofa possible nanowire nucleation site on the top ofa hexagonal
platelet. (c) A TEM image ofthe matrix showing large and small hexagonal holes. which maybe sites
where nanowires nucleated and broke off. (d) A TEM image ofthe matrix also showing large and small
hexagonal holes indicating possible nanowire nucleation sites. Images: (a),(b) K. McElroy, (c).(d) B.W.
Jacobs

 

117

Nanowire nucleation sites on the sides and comers of hexagonal platelets were
also observed in the matrix that formed at 1000" C, not shown. As discussed next.
features that were common to matrix formations from each temperature that give rise to

similar multiphase nanowire growth were identiﬁed.

4.2.2 HRTEM

Further investigation of the matrix with HRTEM revealed features on the sides of
the hexagonal platelets that were not resolvable in the SEM. These features consisted of
a network of nanoscale ledges as shown in ﬁgure 4.23. A typical hexagonal GaN platelet
from matrix that formed at 850° C is shown in ﬁgure 4.23(a). A close up HRTEM image

shown in ﬁgure 4.23(b) of the area indicated in 4.23(a) revealed the nanoscale features

that were present on the sides of the hexagonal platelet.

    

Figure 4.23. TEM images ofa hexagonal platelet and its nanoscale ledges from matrix that formed at 850°
C. (a) A TEM image shows a typical GaN hexagonal platelet. (b) HRTEM image ofthe indicated area in

(a) revealed a network ofnanoscale ledges. Images: B.W. Jacobs

118

Figure 4.24 shows HRTEM analysis of the nanoscale ledges from hexagonal
platelets that formed at 850° C. FFTS and HRTEM images indicated that the nanoscale
ledges were in the same orientation and coherent with the hexagonal platelets. Figure
4.24(a) shows an HRTEM image of nanoscale ledges and their orientation. The
nanoscale ledges formed peak-like structures that resembled a row of teeth on a saw
blade. The two edges of each nanoscale ledge corresponded to the { IOTO} planes as
indicated in each image in ﬁgure 4.24. The <2T T0> directions are also shown because
this was the wurtzite growth direction in the multiphase nanowires. The signiﬁcance of
this will be discussed in section 4.3. Figure 4.24(b) shows an HRTEM image of
nanoscale ledges that were on the same platelet as those shown in 4.24(a). These
platelets were in a different but equivalent orientation. Not all platelets had nanoscale
ledges on their sides, however. Figure 4.24(c) shows an HRTEM image of a smooth
platelet with no observable nanoscale features in the form of ledges. Figure 4.24(d) is an
HRTEM image taken in a slightly different area but on the same platelet where it
intersected another hexagonal platelet. Smooth edges are readily seen with no indication

of nanoscale ledges.

119

 

{1010}
[12101l

[21101l

{1010}

 

Figure 4.24. HRTEM images of the nanoscale ledges from matrix that formed at 850° C. (a) An HRTEM
image shows the side ofa hexagonal platelet revealing the orientation of the nanoscale ledges. (b) An
HRTEM image taken from a different area of the same hexagonal platelet revealed nanoscale ledges in a
different but equivalent orientation as the ledges shown in (a) (c) An HRTEM image ofa different platelet
shows no nanoscale ledges revealing that not all hexagonal platelets have these features. ((1) An HRTEM
image ofa different area on the same platelet in (c) shows the smooth edges ofthis platelet. Images: B.W.
Jacobs

Figure 4.25 shows TEM and HRTEM analysis of the nanoscale ledges from
hexagonal platelets that formed at lOOO° C. Figure 4.25(a) is a TEM image showing
evidence of hexagonal platelet decomposition leading to the formation of nanoscale
ledges. Platelet decomposition will be further discussed in section 4.3. Figure 4.25(b) is
an HRTEM image of the boxed area in (a). This image shows the nanoscale ledge

orientations relative to the hexagonal platelet. Each ledge was in the same orientation

120

and coherent with the platelet. An FFT of the nanoscale ledge, shown in the inset,
conﬁrmed its orientation. Figure 4.25(c) shows the side of a different platelet. It shows
that the nanoscale ledges can form on the long edges of hexagonal platelets. This leads to
very asymmetric nanoscale ledges with one { IOTO} plane much longer than the other

similar to a staircase rather than a saw blade.

__ 1010
{1010} 4110} { lg

    

.4, ' tOnm

Figure 4.25. TEM images of hexagonal platelets and nanoscale ledges from matrix that formed at 1000°
C. (a) Decomposed platelet resulting in the formation of nanoscale ledges. The dotted line indicates a fully
formed platelet. (b) An HRTEM image of the area indicated in (a) shows the features and orientation of the
ledges. (c) A TEM image ofa different platelet indicates that nanoscale ledge formation can occur on the
sides of platelets leading to a roughened edge. Images: B.W. Jacobs

These nanoscale nucleation sites were key in understanding how the nanowires
nucleated. This type of nucleation site was expected to lead to the formation of
multiphase nanowires with multi-domain structures extending in the longitudinal

direction. Details of nanowire nucleation on the nanoscale ledges will be discussed next.

121

4.3 Nanowire Growth and Discussion

SEM investigations of the growth matrix indicated that the multiphase nanowires
grew from the sides and corners of hexagonal platelets. It was later discovered that the
hexagonal platelets formed at all three growth temperatures had networks of nanoscale
ledges in the same orientation and coherent to the hexagonal platelets. The edges and
points of the nanoscale ledges were expected to be active and responsible for nanowire
nucleation. The peaks of the nanoscale ledges were oriented in the <2T T0> direction,
the same direction as the wurtzite phase in the multiphase nanowire growth, as shown in
ﬁgures 4.7-4.9 and in ﬁgure 4.26. However, if only the point contributes to nanowire

nucleation, the nanowire should be much thinner that it was observed to be.

a <2110>

 

A; A
V ‘7

~60nm
Figure 4.26. Model showing the nanoscale ledges. The edges are the { IOTO} planes and the peaks point

in the < 2TTO> directipn_. Nanoscale ledge growth occurred on the { IOTO} planes and nanowire growth
occurred along the < 21 10> direction. Image: B.W. Jacobs

Growth was likely along the active {IOTO} ledges as well as at the ﬁeld-
enhanced active points, as shown in ﬁgure 4.26. However, NH3 is known to break down
at step edges”, and nitrogen may be abstracted forming molecular nitrogen, leading to a
decomposition process on the surface of the nanoscale ledges, as shown in ﬁgure 4.25.
These processes are known to occur in GaN thin ﬁlms. Bombardment of the nanoscale

ledges by energetic gallium atoms may also contribute to the decomposition process.

122

This in turn leads to a line of dangling bonds on the {2T T0} planes, which is expected
to be very active, possibly resulting in nanowire nucleation.

Competitive { IOTO} wurtzite growth versus <011>/<2T T0> nanowire growth.
such as is shown in ﬁgure 4.27, was typical near the nanowire nucleation sites. Platelet
and nanowire orientations are inferred from the shape of the platelet and direction of the
nanowire, which for the wurtzite phase has always been observed to grow along the
<2T T0> direction at this growth temperature. However, growth along the <01 1> and
< 2T TO> directions eventually dominate growth.

Nucleation at multiple nanoscale ledge sites can also explain the multiple wurtzite
domain orientations because multiple nanoscale ledges could contribute to a single
nanowire. These observations support the hypothesis that the nanoscale ledges act as the

nucleation sites. All nanoscale ledges observed have been wurtzite. Therefore evidence

of zinc-blende nucleation from wurtzite GaN was inferred, rather than directly shown.

  
   

300nm

 

_ t. 1.. >.

Figure 4.27. SEM images indicating the evidence ofthe { IOTO} growth. (a) The arrow indicates where
the { IOTO} platelet growth occurred, near the nanowire nucleation site, and the nanowire continuing to
grow from there. (b) The arrow indicates a similar platelet growth in this image. The nanowire is shown to
continue to grow from the base formation. Images: K. McElroy

123

1n the multiphase nanowires, the observed triangular wurtzite domains within the
nanowires form regions with {IOT l} and (0001) faces. However. only the { IOT 1}
planes were observed as external facet planes. The (0001) planes have typically been
observed to terminate within the internal nanowire volume. Further, a tapered tip is
observed on nanowires that grew along the [0001] direction at 1000° C. as shown in
ﬁgure 4.14 and 4.28. The external facet planes at the tip are the {IOT 1} planes based on
the angle relative to the [0001] grth direction.

GaN (0001) planes are known to have multiple surface reconstructions”29

. In the
multiphase nanowire system the instability of the (000]) planes may be compensated
through the incorporation of the zinc-blende phase, as a second stabilizing aspect of the
multiphase conﬁguration. This is readily apparent in the cross-sections in ﬁgures 4.12
and 4.14, where the zinc-blende phase could be a convenient option for internal nanowire
domain conﬁgurations. Also as previously mentioned, nanowires with multiple wurtzite
domains could nucleate off multiple nanoscale nucleation sites. The instability of the

(0001) planes could cause the wurtzite domains to twist inward after nucleation resulting

in (0001) planes that only terminate internally.

[0001 }

[0001}
/

 

Figure 4.28. SEM images showing thE tapered tips ofnanowires that grew at 10000 C along the [0001]
direction. The tips faceted on the { 101 1} planes. Images: K. McElroy Interpretation: B.W. Jacobs

124

The multiphase nanowires and nanoscale ledges that acted as nucleation sites
were found at all growth temperatures investigated. This can explain the presence of the
multiphase nanowires observed at each growth temperature. However, at lOOO° C single-
phase wurtzite nanowires with growth along the [2T T0] direction, in addition to the
multiphase nanowires, were observed, as shown in ﬁgure 4.12. These single-phase
nanowires were typically skinnier than the multiphase nanowires. The single-phase
nanowire growth may have to do with growth kinetics in addition to the type of
nucleation site.

A second type of nucleation site developed at 10000 C, located on the (0001)
plane of the hexagonal platelets. Nanowires from this type of nucleation site grew along
the [0001] direction, and had hexagonal cross-sections. The reason for nanowire
nucleation at these sites is as yet not known, and further investigations are underway in
order to understand how nanowires nucleate of the tops of hexagonal platelets.

A formation with a difference in contrast in the HRTEM images was also
typically observed at the middle of each nanowire, as shown in ﬁgures in section 4.2.
Contrast differences may be due to a small variation in thickness with respect to the rest
of the nanowire cross-section or a slight crystal domain orientation difference. These
features are common in nearly all nanowire cross-sections observed. Continuing
investigations are underway to determine if these features propagate the entire length of
the nanowire.

Details of the nanowire nucleation sites, including the formation of nanoscale
ledges whose orientation dominates the nanowire growth direction, has been discussed.

The formation of a totally coherent interface between the zinc blende and wurtzite

125

phases, which was sustained over the entire length of the nanowire, may provide stability
that renders the multiphase conﬁguration energetically attractive. There are several
implications associated with this newly discovered ability to consistently grow zinc-
blende/wurtzite multiphase GaN nanowires. Multiphase nanowires represent a new class
of waveguide structures with important applications in quantum transport and phase
speciﬁc transport, as will be discussed in the next section. Temperature has been shown
to be an important factor in nucleation site formation. Nucleation site control leads to
single and multiphase GaN nanowires with high crystalline quality that have tailored and
reproducible crystallographic growth directions, which is essential in producing high
performance devices. The vapor-solid one-step method described here simpliﬁes growth,

making new and complex nanowire structures easier to achieve.

126

4.4 Electronic Properties
4.4.1 Electronic Investigations of Multiphase GaN Nanowire-Based GaNFETs

Gallium nitride nanowire-based ﬁeld effect transistors (GaNFETs) have been
shown to possess interesting electronic and optical properties due to their wide, direct
band gap as discussed in the introduction of this chapter. GaNFETs using multiphase
nanowires as active elements were fabricated using electron beam lithography (EBL) and
their electronic properties tested.

A highly doped p-type silicon wafer (5 mQ-cm) was used as the GaNF ET
substrate and back gate. A 100 nm layer of thermally grown silicon dioxide was used as
the gate dielectric, as shown in ﬁgure 4.29. The backside of the wafer was stripped of
silicon dioxide using diluted HF and Ti/Au (IO/70 nm) was thermally evaporated to form
a layer of high conductivity for the back gate contact of the GaN F ET”. GaN nanowires
were then dispersed from an ethanol solution onto the substrate, and source and drain
contacts were patterned via EBL. Ti/Au(10/30 nm) was thermally evaporated for the
conducting source and drain contacts after being exposed to a 100 W oxygen plasma
(March Instruments PX-250) for 30 s to l min to remove any electron beam resist

residue. Subsequent metal lift-off was performed in heated acetone.

127

 

 

Figure 4.29. A typical GaNFET conﬁguration. A highly doped silicon wafer was used as the substrate.
SiOz was grown as the gate dielectric and a layer of metal was thermally evaporated on the backside of the

wafer for the gate electrode. Nanowires were dispersed on top of the wafer and source and drain contacts
were patterned with EBL. Image: B.W. Jacobs

Current—voltage (I-V) measurements of a GaNFET were done using a Keithley
4200-Semiconductor Characterization System (SCS)3 1. The GaNFET was connected to
the 4200-SCS through macroscopic probes via photolithographically patterned contact
pads, with EBL patterned electrodes contacting the nanowire. The global back gate was
contacted to the probe station chuck. Measurements were done in ambient air at room
temperature in a shielded enclosure.

During these measurements the drain—source (DS) voltage was continually swept
from —5 to +5 VDs as the gate—source (GS) voltage stepped from -30 to +30 V03 in steps
of 10 V. I—V curves, ﬁgure 4.30(a), indicated Schottky barrier behavior since the current
changed non-linearly as VDs was swept and VGS stepped. Current modulation from +2 to
+12.6 uA was observed as V03 was stepped from —30 to +30 Vgs at +5 VDs. Current
modulation from -6 to -l 3.7 uA was observed as the gate voltage was stepped from ~30
to +30 V03 at —5 V05. This behavior was typical for an n-type device. Figure 4.30(b)

shows V03 swept as VDS was stepped, and was extracted from the VDS swept data in

128

ﬁgure 4.30(a). VDS was stepped in increments of 1 V. The calculated transconductance,

gm. for this device was ~4 nS, where gm=A1 Ds/AVus.

10.011 < 30V

.1

5.0}1 «

0.0a

 

05 (Amps)

'05 (Amps)

‘ .I; 4‘
”ﬁt”;

/‘ -. /
5 0 '1". ' ll
. r_../

" . “4 /x,v'
/ J ' — -
/

t /‘.‘Z'

 

-100H‘l -10“ 1 ’5V \.. __ .\\.‘
45.0w ' 1 \.\\
. . . . . T . . 1 ~15}; . e r - . . . . . - Y . .
-6 -4 -2 0 2 6 -30 -20 -10 0 10 20 30
V08 (Volts) vGS (Volts)

Figure 4.30. l-V characteristics of a GaNFET. (a) The current was measured as VDs was swept and Vus
stepped. The nonlinearity of the curves indicated Schottky barrier behavior. (b) This graphs emphasizes

the change in current as a result of the change in VGS. These data were extracted from the data in (a). l-V
Characteristics: B.W. Jacobs el al. Nanotechnol., 18, 475710 (2007) with permission.

 

 

 

 

_4

Additional I-V characteristics of GaNFETs were measured at NASA Goddard
Space Flight Center. VDg was varied and the resulting current was measured, but the gate
terminal was not contacted in these experiments. Figure 4.31(a) shows a typical I—V
curve of a circuit investigated in this study. Over 70 uA was measured at 6 VDS. The
nonlinearity of the curve can be attributed to the Schottky barrier that formed at the
nanowire-metal junction as mentioned above and further discussed below. Figure 4.31 (b)
shows an SEM image of the actual circuit. The nanowire was on top of a SiOz insulating
layer and Ti/Au contacts were thermally evaporated on top of the nanowire and 8102
lay er. The arrow shows the path of the induced current through the Ti/Au contacts and

the nanowire.

129

 

(a) .
/ _Nanow1re
60th /
s “
-- 201.1- /-’
,-
/./r
011— V'
o 1 ' ' ' 6 7

 

 

2 3 i

VDs (Volts)
Figure 4.3]. I-V characteristics ofthe circuit shown in the SEM image. (a) The nonlinear curve indicated
that the nanowire-metal junction formed a Schottky barrier. At 6 VDs over 70 11A of current was passing
through the nanowire. (b) An SEM image shows the actual circuit. The SiOz insulating layer and Ti/Au
contacts are labeled. The arrow shows the path current takes through the circuit. I-V characteristics. image
and interpretation: B.W. Jacobs

The circuit shown in ﬁgure 4.32 was fabricated with three nanowires contacted in
parallel. This showed that the current roughly scaled accordingly with the addition of
two nanowires. Over 400 uA was measured at 6 V93, as shown in 4.32(a). Figure
4.32(b) shows the entire circuit and gives an overall picture of how the circuit was
fabricated. First the large Ti/Au pads were patterned with photolithography (PL). The
nanowires were then deposited on the surface, and the Ti/Au contacts were patterned with
EBL. Figure 4.32(c) shows two contacted nanowires. The EBL contact on the right was
on top of the PL pad, thereby completing the circuit. Figure 4.32(d) shows the third

nanowire contacted with EBL.

130

 

 

Ti/Au (PL)

i‘.

’ Ti/Au (EBL)

 

Figure 4.32. l—V characteristics of a GaNFETs with three nanowires hooked in parallel. (a) The measured
current was very high, over 400 uA, at 6 VDS- (b) An SEM image shows the entire circuit.
Photolithography (PL) was used to deﬁne the large contact pads where the contact probes were placed.

The left and right boxes denote where close ups were taken in (c) and (d), respectively. (c) An SEM image
showing two nanowires contacted in parallel. EBL was used to deﬁne the areas where metal was deposited
on top of the nanowires. The bottom left pattern created with EBL is seen on top of an PL deﬁned pattern.
(d) A SEM image shows the third nanowire hooked in parallel and the EBL deﬁned patterns that connected
it. l-V characteristics, images and interpretation: B.W. Jacobs

Investigations by some groups suggest that it is the Schottky barriers between the
Ti/Au source and drain and the GaN nanowire that are the source of the transistor
behavior32‘33‘34. A putative model for similar devices utilizing carbon nanotubes has also
been developed35‘36‘37. In the case of these GaNFETs, raising and lowering of the barrier
through an electrostatic interaction between the source and drain contacts and the back
gate electrode explain the ability of the gate to modulate the current. The current arises
from both tunneling and thermionic emission injection in varying proportions, and was

dependent upon the height and thickness of the barrier”.

131

In the FET experiments the current was modulated via the gate voltage, with less
current ﬂowing when -30 V05 gate was applied, 2 (1A, relative to the current ﬂowing
when +30 V05 gate was applied, 12.6 uA. both at +5 V03. The gate voltages needed to
signiﬁcantly modulate the current were somewhat high, and the on—off ratio was
therefore low. This may be due to the high contact resistance between the nanowire and
Ti/Au contact, as will be discussed in the two and four-point probe investigations. This I
resistance may be improved with further processing steps such as rapid thermal

annealing.

 

The electronic transport characteristics between metal electrodes and GaN i:
nanowires in such devices are of great importance when maximizing device performance.
as these junctions play a critical role in device behavior. Therefore, two and four-point
probe investigations were carried out to quantify the contact resistances in a GaNFET
device using a state-of—the-art probing instrument, as discussed next. However, carrier
injection into nanowires has yet to be completely explained at a fundamental level, which

impedes the effectiveness of device design.

4.4.2 Electronic Investigations of an Individual Multiphase GaN Nanowire

The electronic characteristics of an individual GaN nanowire were investigated
using two-point and four—point probe techniques. A GaNFET was used as the test sample
in these experiments. The back gate was not used. The experiments were performed
using a Zyvex K2100 Nanomanipulator system. The K2100 is a hybrid instrument
interfacing the Keithley 4200-SCS and the Zyvex 8100 Nanomanipulator, as shown in

ﬁgure 4.33. It is equipped with probes that are electrochemically etched tungsten

132

polycrystalline wires with a nominal tip diameter of 50 nm. The KZI 00 is capable of
probe positioning resolution under 5 nm. This small tip diameter and precise probe
positioning allows direct probe-nanowire connections. The electronic characterization
with the KZIOO was performed in a LEO 1530 FESEM at room temperature. allowing
real-time, visual inspection with the SEM during the experiments. During measurements
the SEM electron beam was turned off, and during visual inspection a low energy beam

was used to minimize charge penetration effects. The tungsten probes were electrically

cleaned in situ to provide a clean tungsten surface for measurements.

    

Figure 4.33. The Zyvex SI 00 Nanomanipulator System. (a) The entire probing system is mounted in an
SEM for high resolution imaging essential for nanoprobing and nanomanipulation. The arrows point out
the four piezoelectrically controlled tungsten nanomanipulators. Images: R. Stallcup, A. Hartman and
V.M. Ayres

4.4.2.1 Four-Point Probe

In the four-point probe measurements, the intrinsic nanowire resistance and the
Ti/Au-nanowire contact resistances were determined“. This was done using floating
voltage sense probes that were situated between two current source probes. This
conﬁguration is used to eliminate probe contact resistance effects, as described in chapter

2, so that the intrinsic nanowire resistance can be measured. The nanoscale voltage

133

probes are important for four-point probe measurements of single-nanowire systems since
they are less invasive than macroscopic probes and contacts. The 4200-SCS is also
ideally suited for measurements in such nanoprobing conﬁgurations because it can make
measurements with very low noise and can have input impedances greater than 10'6 (2.
Four separate four-point probe conﬁgurations were carried out to determine both
total system and intrinsic nanowire resistance, as shown in ﬁgure 4.34. In conﬁguration
A, the current source and voltage sense probes were placed on the Ti/Au contacts, as
shown in ﬁgure 4.34(a), to measure the total system resistance. In conﬁguration B, the
current source probes were placed on the Ti/Au contacts and voltage sense probes were
directly contacted to the nanowire, as shown ﬁgure 4.34(b), to investigate the nanowire
resistance. In conﬁguration C, the current source and voltage sense probes were directly
contacted to the nanowire, as shown in ﬁgure 4.34(c), also to investigate the nanowire
resistance. In conﬁguration D, one current source probe was directly contacted at one
end after breaking the nanowire, the voltage sense probes directly contacted the
nanowire, and the other current source probe was placed on the Ti/Au contact, as shown
in ﬁgure 4.34(d). The open nanowire end contact was achieved through probe impact on
the nanowire top surface, which resulted in a brittle fracture. This conﬁguration was
possible due to the probe tip diameter, which is around 50 nm, as it is comparable to the

open nanowire cross-section width.

134

Nanowire

 

4Em 8' I A 1E")

Figure 4.34. SEM images showing the four-point probe conﬁgurations. Probes | and 4 were the current
source probes, 2 and 3 were the voltage sense probes. (a) Conﬁguration A, all probes were contacted to the
contact pads. (b) Conﬁguration B; current source probes were contacted to the contact pads. while the
voltage sense probes were contacted to the nanowire. (c) Configuration C: all probes were contacted to the
nanowire. (d) Conﬁguration D; one current source probe was contacting the end of the nanowire and the
other the contact pad, while the voltage sense probes were contacted to the nanowire. Images: B.W. Jacobs
clal. Nanotechnol, I8, 475710 (2007) with permission.

The four-point resistance measurements from conﬁguration A, used to determine
the total system resistance, consistently produced a resistance of between 3 and 6 M9, as
shown in ﬁgure 4.35(a), which is comparable to other reported ﬁndings”. The graph also
shows “negative“ resistance, however this was a result of taking measurements in the
opposite direction, i.e. a negative voltage, not a negative resistance. Measurements made
in conﬁgurations B and C were used to determine the intrinsic nanowire resistance. In
both configurations the nanowire resistance was between 500 and 800 kQ, as shown in

ﬁgure 4.35(b). Probe coupling directly to the nanowire in these conﬁgurations appeared

135

noisy at low source currents, but resistance values reduced to more consistent values at
higher source currents. For a length of 16 um between the voltage sense probes, and
using a triangular base width and height of 100 nm, the resistivity was calculated to be
15.6 mQ-cm. Conﬁguration D was also used to determine nanowire resistance. After the
nanowire fracture, the voltage sense probes were situated closer together at 10 um. and
the nanowire resistance was between 375 k!) and 3 M9, as shown in ﬁgure 4.35(c). It
was noted that these data were the nosiest among all conﬁgurations, and was like a result
of the end-contacted conﬁguration where coupling may be somewhat different than in the

side—contacted conﬁguration.

 
      
 

20M - (a) 50M (b)
@15M §AOM ,
8 ‘IOM ...... ‘ 8 30M .
C C
I!
.Q
§ “we

 

   

”a. 4: _-._r_+-aﬁas
O 4 0 6 0 8 1

0 2
Source Voltage (V)

   

Figure 4.35. Four-point probe resistance measurements. (a) Conﬁguration A, total system resistance
measurement. Resistance values were between 3 and 5 M9. Negative values corresponded to current
sourced in the opposite direction. (b) Conﬁguration C. NW resistance measurement, side contact.
Resistance values were between 500 and 800 k9. (c) Conﬁguration D, NW resistance measurement, end
contact. Resistance values were between 375 k!) and 3 M9. Below the resistance diagrams are the
corresponding nanomanipulator nanoprobe placements for each conﬁguration. Resistance measurements
and diagrams: R. Stallcup, A. Hartman, V.M. Ayres and B.W. Jacobs

4.4.2.2 Two-Point Probe

Two-point probe experiments were also conducted in conﬁgurations A, C, and D.

136

The purpose was to study the change in dynamic resistance in each conﬁguration due to
size differences of the probe tips and the Ti/Au contacts. The dynamic resistance of the
total system including the probe—Ti/Au—nanowire contact was investigated in
conﬁguration A. The dynamic resistance of the probe—nanowire contact was investigated
in conﬁguration C. Two-point probe—nanowire breakdown investigations were
conducted in conﬁguration D.

In conﬁguration A, two-point probe measurements were conducted to investigate
the dynamic resistance, R = A V/Al, of the total system. The nonlinear character observed
for the I—V characteristics corresponded to a total system resistance, R, which was
comprised of the contact, RC1 and RC2, nanowire. RNw. and probe, Rm and sz,
resistances, where

R = Rm + RC1 + RNW + RC2 + RPZ-
The probe resistances, Rp. and sz, were assumed to be negligible in conﬁguration A
since the tungsten probes and Ti/Au contacts should couple well due to the metal on
metal contact. This was conﬁrmed in the four-point probe measurements in this
conﬁguration, where the source voltage and sense voltage were close in value. The total
dynamic resistance measured in this conﬁguration followed a power law as shown in
ﬁgure 4.36, where

[(V) a V 1.7.

In conﬁguration C, two-point probe measurements were conducted to investigate the
dynamic resistance of the probes and nanowire. RC) and RC; were omitted by directly
contacting the probes to the nanowire. Therefore RNw, RP). and Rm were measured. The

total dynamic resistance in this case followed a different power law, also shown in ﬁgure

137

 

4.36, where

p)
I‘d

 

I( V ) 0: I"
3000:: - 10.0n
----- Nanoprobes on TI/Au Contacts
- - - - Nanoprobes directly on Nanowire A
zso.on - 4:
A 4 3.0" _
< 0
6 2000 E
. I'I -‘ '—
.E ‘O
s ‘1 8.0" 2
:3} 150m ~ g
v - 4.0n 3
o-v o—o
:1 100m - c
o o
t t
:3 ~ 2.0n 3
O 50.0» - Q
o.o~ - 0.0

 

 

 

 

I Y I

0.0 0.2 0.4 0.0 0.8 1 .0

Voltage (V)
Figure 4.36. Two-point probe results. The graph shows the different voltage dependences of the current

when the probes are on the Ti/Au contacts (dotted line, compliance was set to 250 nA) and directly on the
nanowire (dashed line). Diagrams of the probe conﬁgurations are shown on the right. I-V characteristics
and diagrams: K. McElroy and B.W. Jacobs

The dynamic resistance behavior from the two-point probe investigations suggests
the formation of a Schottky barrier at the probe—nanowire contact as well as at the Ti/Au—
nanowire contact. The different exponents could result from variations in coupling, and
indicate different barriers between the probe—nanowire and Ti/Au—nanowire interfaces.
It has also been shown that for a carbon nanotube system a smaller metal contact area
concentrates the electric ﬁeld at the junction“. Electron conduction channel variations
through the contact may also play a role in contact resistance, where microscopic to
microscopic contacts may increase the carrier injection efﬁciency“).

After measurements were made in conﬁguration D, where the nanowire was

broken in the middle. as shown in ﬁgure 4.37(a), the breakdown behavior was

investigated. The source current was increased from 10 nA through 50 (LA, with

138

breakdown occurring around 50 uA. After each measurement the nanowire was visually
inspected with the SEM. After breakdown, a pull-apart of the nanowire at a point
approximately midway between the voltage sense probes was observed in the SEM,
ﬁgure 4.37(b). Also, a new phenomenon was observed prior to breakdown. A liquid
protrusion formed at the directly contacted end while the external structure remained
intact. A small liquid ball was clearly observable by 10 uA, as shown in ﬁgure 4.37(c), T
and continued to grow in size to over 100 nm in diameter, as shown in ﬁgure 4.37(d).

The liquid nature of the protrusion was veriﬁed by mechanically pushing into it with the

 

probes.

I39

nanowire

 

Figure 4.37. SEM images ofthe breakdown measurements, in conﬁguration D. (a) The fractured
nanowire. (b) The pull-apart between the voltage probes. The location ofthe pull-apart is shown in the
inset. (c) Simultaneous onset ofthe single-phase melt. The location ofthe melt is shown in the inset. (d)
A larger amount of material melted and formed a protrusion outside the nanowire. Images: B.W. Jacobs er
al. N0!10IL’L'I1I10[., I8. 475710 (2007) with permission.

The observed structural disintegration may be a result of the multiphase nature of
the GaN nanowire. A single-phase volume was observed to melt while an outer layer
retained its structural integrity. The diameter of the initial melt formation. which
progressed smoothly. as observed in the SE M, to the ﬁnal melt formation was consistent
with melting of a single phase. The electronic structure of the multiphase interface and
the possibility of phase speciﬁc electron transport and phonon conﬁnement are currently

under investigation by our group.

The multiphase nanowire has the ability to carry high current densities as shown

I40

in the GaNF ET and two and four-point probe experiments. This may be due to the
multiphase nature of the nanowire. The actual path electrons take in the nanowire is still
an open question. They could ﬂow in equal amounts throughout the nanowire, however
the two-point breakdown experiments indicated that current might follow a speciﬁc
crystalline phase. The interfaces that exist in the nanowire as shown by the cross-
sections may have barriers high enough to conﬁne electrons in one phase. They may also
form a longitudinal quantum well at the interface, into which the electrons can fall and
travel in. Further theoretical and experimental investigations may answer this question.
However, the multiphase structure opens possibilities for enhanced device performance
due to its ability to carry high current and if single-phase transport can be reliably

achieved, the implications would be signiﬁcant.

4.4.3 Cathodoluminescence

Cathodoluminescence (CL) experiments were performed to investigate the
electronic structure of the multiphase GaN nanowires, including the band gap and if a
signiﬁcant number of impurities exist in the nanowire. The experiments were performed
in a LEO Supra 50 VP SEM in conjunction with a GatanMono CL system with a low-
noise Peltier-cooled Hamamatsu R5509 photomultiplier detector. The GaN signature in
the 290-360 nm wavelength range was used to optimize the CL signal. The best peak-to-
background ratio was achieved at approximately 10 keV incident electron energy. The
slit widths were set at 1 mm, which was the optimum for achieving high count rates,
while maintaining sufﬁcient resolution to differentiate between the wurtzite and zinc-

blende band gaps. Narrower slits did not change the peak widths signiﬁcantly. In these

141

initial investigations, detailed analysis of the near band gap exciton electronic structure
was not considered, and the experiments were carried out at room temperature.

The CL spectra of various GaN nanowires exhibited either a single peak, ascribed
to the wurtzite structure, or two distinct peaks, implying the simultaneous presence of
both wurtzite and zinc-blende structures. It should be noted that the spectra had no
observable contribution from luminescence that could be attributed to defects (stacking
faults or impurity mediated), which indicated high-quality samples. The observed CL
spectrum shown in Figure 4.38, indicated the coexistence of zinc-blende and wurtzite

structures in a single isolated GaN nanowire. The peak maxima, observed at 3.64 and

 

ll

3.88 eV, were identiﬁed as the zinc-blende and wurtzite peaks based on their similarity to
those observed for bulk GaN. The increase in energy of ~0.4 eV, measured by CL,
relative to bulk values, may be attributed to both electron conﬁnement and strain effects

as discussed below.

142

Energy (eV)

 

 

 

 

 

 

4.5 4 3.5 3 2.5

[YII'VIIIIITjIIIUII ' I I T I I I I T I I

_ Wurtzite
7; ~3.889V —— Nanowire Signal
;: .—
5 Zinc-Blende
a. ' ~3.64ev
2
g _
.0 P
L-
s ..
>
g ..
«n
g _
E

I. ___. L “-J_. .- r -_...J-_.. 4: H
300 400 500 600

Wavelength (nm)

Figure 4.38. Cathodoluminescence spectrum showing the double peak indicating the wurtzite and zinc-
blende phases present in the nanowire. The band gap energies for each phase is blue-shifted on the order of
0.4 eV. Spectra: B.W. Jacobs et al. Nano Lett. 7, 1435-1438 (2007) with permission.

A 0.4 eV band gap shift was observed in the CL measurements. An electron
conﬁnement calculation for the nanowires based on an inﬁnite potential well model
accounts for only a few meV of the observed band gap shift. Calculations based on
experiments for pure wurtzite GaN nanowires have identiﬁed band gap shifts as high as

- - 41.42
0.3 eV due to compressrve and tensrle stresses

. Therefore the band gap shift observed
in the present experiments may have more contribution from stress than from electron
conﬁnement, but conﬁnement effects may still play a signiﬁcant role due to the
multiphase nanowire structure, and their effects are under further investigation.

The discovery of the multiphase GaN nanowire was a signiﬁcant step forward in

the ﬁeld of nanowires. It is a new structure with properties that may augment or be

superior to single phase nanowires. The plain-view HRTEM investigations identiﬁed the

143

two phases of the nanowires and the cross-section HRTEM investigations identiﬁed their
orientations and revealed details about how this structure can form. SEM and HRTEM
investigations of the matrix revealed details on how they form and how nanowires with
different growth directions can grow. Electronic properties showed that these nanowires
have a large amount of strain and electron conﬁnement effects based on the blue shifted
band gap results. The multiphase nanowires may have phase speciﬁc transport, begging
the question of how and where electron transport actually occurs in these nanowires. The
ability to carry high currents could lead to high power devices. Their radiation resilience
was also tested as discussed in the next section. which may open up many new

application possibilities for these unique structures.

144

 

References

l.

10.

ll.

12.

Y. W. Heo, D. P. Norton, L. C. Tien, Y. Kwon, B. S. Kang, F. Ren, S. J. Pearton, .1.
R. LaRoche, “ZnO nanowire growth and devices,” Mater. Sci. Eng. R, 47, 1-47
(2004)

Y. Shan, A. K. Kalkan, C-Y. Peng, S. J. Fonash, “From Si Source Gas Directly to
Positioned, Electrically Contacted Si Nanowires: The Self-Assembling "Grow-in-
Place" Approach,” Nano Lett., 4, 2085-2089 (2004)

A. B. Greytak, L. J. Lauhon, M. S. Gudiksen, C. M. Lieber, “Grth and transport
properties of complementary germanium nanowire ﬁeld-effect transistors,” App].
Phys. Lett., 84, 4176-4178 (2004)

A. I. Persson, M. W. Larsson, S. Stenstroom, B. J. Ohlsson, L. Samuelson, L. R.

Wallenberg, “Solid-phase diffusion mechanism for GaAs nanowire growth,” Nat.
Mater., 3, 677-681 (2004)

S. Vaddiraju, A. Mohite, A. Chin, M. Meyyappan, G. Sumanasekera, B. W.
Alphenaar, M. K. Sunkara, “Mechanisms of 1D Crystal Growth in Reactive Vapor
Transport: Indium Nitride Nanowires,” Nano Lett., 5, 1625-1631 (2005)

L. J. Lauhon, M. S. Gudiksen, C. M. Lieber, “Semiconductor nanowire
heterostructures,” Phil. Trans. R. Soc. A, 362, 1247-1260 (2004)

C. P. T Svensson, W. Seifert, M. W. Larsson, L. R. Wallenberg, J. Stangl, G. Bauer,
L. Samuelson, “Epitaxially grown GaP/GaAsl -xPx/GaP double heterostructure
nanowires for optical applications,” Nanotechnol. , 16, 936-939 (2005)

J. Goldberger, R. He, Y. Zhang, S. Lee, H. Yan, H-J. Choi, P. Yang, “Single-
crystal gallium nitride nanotubes,” Nature, 422, 599-602 (2003)

J. Q. Hu, Y. Bando, J. H. Zhan, F. F. Xu, T. Sekiguchi, D. Golberg, “Growth of
Single-Crystalline Cubic GaN Nanotubes with Rectangular Cross-Sections,” Adv.
Mater. 16, 1465-1468 (2004).

I. Vurgauftman, J. R. Meyer, “Band parameters for nitrogen-containing
semiconductors,” .1. Appl. Phys, 94, 3675-3696 (2003)

R. G. Powell, N. E. Lee, Y. W. Kim, J. E. Greene, “Heteroepitaxial wurtzite and zinc-
blende structure GaN grown by reactive-ion molecular-beam epitaxy: Growth

kinetics, microstructure, and properties,” .1. App]. Phys, 73, 189-204 (1993)

P. Boguslawski, E. L. Briggs, J. Bemholc, “Native defects in gallium nitride,” Phys.
Rev. B, 51. 17255-17258 (1995)

145

 

13.

14.

15.

Y. Huang, X. Duan, Y. Cui, C. M. Lieber, “Gallium Nitride Nanowire Nanodevices,”

Nano Lett., 2, 101-104 (2002)

Y. Huang, X. Duan, Y. Cui, L. J. Lauhon, K. H. Kim, C. M. Lieber, “Logic Gates and
Computation from Assembled Nanowire Building Blocks,” Sci., 294, 1313-1317
(2001)

H-M. Kim, T. W. Kang, K. S. Chung, “Nanoscale ultraviolet-1ight-emitting diodes
using wide-bandgap gallium nitride nanorods,” Adv. Mater., 15, 567-569 (2003)

16.2. Zhong, F. Qian, D. Wang, C. M. Lieber, “Synthesis of p-Type Gallium Nitride

17.

18.

19.

20.

21.

22.

23.

Nanowires for Electronic and Photonic Nanodevices,” Nano Lett., 3, 343-346 (2003)

C. Y. Nam, P. Jaroenapibal, D. Tham, D. E. Luzzi, S. Evoy, J. E. Fischer, “Diameter-
Dependent Electromechanical Properties of GaN Nanowires,” Nano Lett., 6, 153-158
(2006)

X. Duan, C. M. Lieber, “Laser—assisted catalytic growth of single crystal GaN
nanowires,”J. Am. Chem. Soc, 122, 188-189 (2000)

H. Y. Peng, X. T. Zhou, N. Wang, Y. F. Zheng, L. S. Liao, W. S. Shi, C. S. Lee, S. T.
Lee, “Bulk-quantity GaN nanowires synthesized from hot ﬁlament chemical vapor
deposition,” Chem. Phys. Lett., 15, 263-270 (2000)

C. C. Chen, C. C. Yeh, C. H. Chen, M. Y. Yu, H. L. Liu, J. J. Wu, K. H. Chen, L. C.
Chen, J. Y. Peng, Y. F. Chen, “Catalytic Growth and Characterization of Gallium
Nitride Nanowires,” J. Am. Chem. Soc, 123, 2791-2798 (2001)

M. Law, J. Goldberger, P Yang, “Semiconductor Nanowires and Nanotubes,” Annu.
Rev. Mater. Res, 34, 83-122 (2004)

M. He, P. Zhou, S. N. Mohammad, G. L. Harris, J. B. Halpem, R. Jacobs, W. L.
Samey, L. Salamanca-Riba, “Growth of GaN nanowires by direct reaction of Ga with
NH3,” .1. QfCryst. Grow, 231, 357-365 (2001)

M. He, 1. Minus, P. Zhou, S. N. Mohammed, J. B. Halpem, R. Jacobs, W. L. Samey,
L. Salamanca-Riba, R. D. Vispute, “Growth of large-scale GaN nanowires and tubes
by direct reaction of Ga with NH3,” Appl. Phys. Lett., 77, 3731-3733 (2000)

24.B. W. Jacobs, V. M. Ayres, M. P. Petkov, J. B. Halpem, M. He, A. D. Baczewski, K.

McElroy, M. A. Crimp, J. Zhang, H. C. Shaw, “Electronic and Structural
Characteristics of Zinc-Blende Wurtzite Biphasic Homostructure GaN Nanowires."
Nano Lett., 7, 1435-1438 (2007)

146

 

25

26.

27.

28.

29.

30.

31.

32.

33.

35.

36.

37.

38.

D. B. Williams, C. B. Carter, “Transmission Electron Microscopy,” Plenum Press.
New York, 1996.

http://rsb. info. ni h. gov/{j/
R. Held, B. E. Ishaug, A. Parkhomovsky, A. M. Dabiran, P. 1. Cohen, “A rate

equation model for the growth of GaN on GaN (0001) by molecular beam epitaxy.” .1.
Appl. Phys, 87, 1219-1226 (2000)

r1

K. Rapcewicz, M. B. Nardelli, J. Bemholc, “Theory of surface morphology of
wurtzite GaN (0001) surfaces,” Phys. Rev. B, 56, 12725 (1997)

J. F ritsch, O. F . Sankey, K. E. Schmidt, J. B. Page, “Ab initio calculation of the
stoichiometry and structure of the (0001) surfaces of GaN and AIN,” Phys Rev. B,
57,15360(1998)

J. Appenzeller, J. Knoch, R. Martel, V. Derycke, S. J. Wind, P. Avouris, “Carbon
nanotube electronics,” IEEE T. Nanotechnol., 4, 184-189 (2002)

B. W. Jacobs, V. M. Ayres, R. E. Stallcup, A. Hartman, M. A. Tupta, A. D.
Baczewski, M. A. Crimp, J. B. Halpem, M. He, H. C. Shaw, “Electron transport in

zinc-blende wurtzite biphasic gallium nitride nanowires and GaNFETs,”
Nanotechnol., 18, 475710 (2007)

C. Y. Nam, D. Tham, J. E. Fischer, “Disorder effects in focused-ion-beam-deposited
Pt contacts on GaN nanowires,” Nano Lett., 5, 2029-2033 (2005)

H. Y. Yu, B. H. Kang, C. W. Park, U. H. Pi, C. J. Lee, S. Y. Choi, “The fabrication
technique and electrical properties of a free-standing GaN nanowire,” App]. Phys. A,
81, 245-247 (2005)

. M-H. Ham, J-H. Choi, W. Hwang, C. Park, W-Y. Lee, J-M. Myoung, “Contact

characteristics in GaN nanowire devices,” Nanotechnol., 17, 2203-2206 (2006)

F. Leonard, A. A. Talin, “Electrical contacts to nanotubes and nanowires: why size
matters A rX iv Preprint cond-mat/0602003 (2006)

R. Martel, V. Derycke, C. Lavoie, J. Appenzeller, K. K. Chan, J. Tersoff, P. Avouris,
“Ambipolar electrical transport in semiconducting single-wall carbon nanotubes,”

Phys. Rev. Lett., 87, 256805 (2001)

S. Heinze, J. Tersoff, R. Martel, V. Derycke, J. Appenzeller, P. Avouris, “Carbon
nanotubes as Schottky barrier transistors,” Phys. Rev. Lett., 89, 106801 (2002)

J-Y. Yu, S-W. Chung. J. R. Heath, “Silicon nanowires: preparation, device
fabrication, and transport properties,” J. Phys. Chem, 104, 11864-11870 (2000)

147

 

39. J-R. Kim, H. M. So, J. W. Park, J-J. Kim, J. Kim, C. J. Lee, S. C. Lyu, “Electrical
transport properties of individual gallium nitride nanowires synthesized by chemical-
vapor-deposition,” Appl. Phys. Lett., 80, 3548-3550 (2002)

40. S. Datta, Electronic Transport in Mesoscopic Systems, (Cambridge: Cambridge
University Press) 1995

41. H. W. Seo, S. Y. Bae, J. Park, H. N. Yang, K. S. Park, S. Kim. “Strained gallium
nitride nanowires,” J. Chem. Phys, 116, 9492-9499 (2002)

42. T. Kuykendall, P. J. Pauzauskie, Y. Zhang, J. Goldberger, D. Sirbuly, J. Denlinger, P.
Yang, “Crystallographic alignment of high-density gallium nitride nanowire arrays,”
Nat. Mater., 3, 524-528 (2004)

 

148

5. Fundamentals of Radiation Interactions in Reduced Dimensionality Systems

The candidate’s research has been supported by grants and contracts from NASA
and by the award of a NASA Graduate Student Researchers Program Fellowship with
two subsequent competitive renewals. NASA’s interest is in circuit and material
resilience in the extreme environment of space. The candidate‘s research is also relevant

for other extreme environment situations.

5.] The Space Radiation Environment

 

The harsh radiation conditions in space have been a formidable problem in

l T

designing suitable electronic devices that are able to sustain desired operation in such
environments. The environment close to the Earth is somewhat more protected due to its
magnetic ﬁeld that deﬂects most of the harmful radiation away from the Earth, and serves
to shield life on Earth as well as electronic devices. Figure 5.1 is an exaggerated
illustration of how the magnetic ﬁeld effects incoming solar radiation and galactic cosmic
rays, where incoming radiation hits the field and is swept past Earth. As one ventures
beyond the realm of this protective magnetic ﬁeld, the threat of harmful radiation
dramatically increases. This radiation is highly energized and highly penetrating, which
even the thickest of radiation shielding is helpless to stop. Sources of this radiation
emanate from across the universe as well as close to home in our own galaxy and solar

system.

149

Galactic Cosmic Rays

Heavier Ions

t
l
.
1

l rapped l’artit‘lus

 

Figure 5.1. The Space Radiation Environment. Galactic cosmic rays constantly bombard the Earth and
surrounding space. The Earth’s magnetic ﬁeld, as indicated by the lines surrounding the Earth. protects it
from harmful radiation. Image: NASA

5.1.1 Single Event Effects and Total Ionizing Dose

Since space based systems and solar system exploration began in the mid 20th
century, space radiation effects have been of primary concern in system design.
Radiation effects on electronic systems has been an ongoing research issue at space
agencies and institutions across the globe. Due to this focused research effort, radiation
effects on electronic systems is well deﬁned and is divided into two categories, total
ionizing dose (TID) and single event effects (SEES). TID is a long-term measure of the
degradation of an electronic device exposed to ionizing radiation. SEES are the effect
that one particle of ionizing radiation has on a device. SEES can be further broken down
into subgroups, delineating speciﬁc affects SEES can induce. These subgroups fall into
two main categories, hard errors and soft errors'l. Hard errors consist of SEES that cause
unrecoverable damage to the device. Soft errors consist of SEES that can be corrected in

ﬂight, through a power reset, error correction, etc.

150

Hard Errors

Single—Event Hard Error (SEH) - An unalterable change of operation that is
typically associated with permanent damage to one or more of the device
elements.

Single-Event Burnout (SEB) — A single ion induced condition that results in the
destruction of the device due to the activation of a localized high current state that
results in catastrophic failure.

Single-Event Gate Rupture (SEGR) — A single ion induced condition that
results in a breakdown and subsequent conducting path through the gate oxide.
Typically seen in a MOSFET device.

Soft Errors

Single-Event Latchup (SEL) — A loss of device functionality due to a single-
event. When SEL occurs a power cycle is needed to return to normal operation.

Single-Event Upset (SEU) — A single latched logic state from one to zero, or vice
versa, i.e. a bit flip.

Single-Event Functional Interrupt (SEFl) — The loss in functionality of the
device that does not require cycling of the device’s power to restore operability.
unlike SEL, and does not result in permanent damage as in SEB.

These two categories, TlD and SEES, provide the framework in which electronic devices

can be tested and benchmarked. The entities that cause these errors are also subdivided

into well deﬁned categories and are now discussed.

5.1.2 Cosmic Rays

Cosmic rays are the primary source of harmful radiation found in space. These

rays come in a variety of forms and species and are the source for many upsets in space

based, as well as terrestrial electronic systems. Cosmic rays can be grouped into two

categories, galactic, originating in the Milky Way including the Sun, and extragalactic.

151

radiation originating from outside the Milky Way3. The atomic composition of galactic
and intergalactic cosmic rays consists of about 90% protons, 9% helium nuclei and 1%
heavier nuclei. Other species of radiation include electrons and photons. such as gamma
and x—rays.

Radiation sources are scattered throughout the universe, with our Sun being the
primary source of high-energy protons and heavy ions in the Solar System4. The Sun is
constantly spewing highly energized heavy ions, protons, neutrons and photons from its
surface that travel through the Solar System and out to interstellar space. Radiation flux
emanating from the Sun varies with the 11-year solar cycle with higher levels of harmful
radiation occurring during the solar maximum. It is during the solar maximum that
orbiting satellites, interplanetary space probes, and other space based electronic systems
are especially susceptible to SEES. Despite the Earth’s protective magnetic ﬁeld,
communication and global positioning (GPS) satellites have been prone to failure during
solar maximums. At times these solar events are so energetic that electronic systems on
Earth, such as the power grid, are affected.

Non-solar galactic cosmic rays, emanating from sources within the Milky Way,
also cause damaging high-energy radiation upsets. These species of radiation are similar
to solar rays, which include protons, electrons, heavy ions, and photons. There are
signiﬁcant differences in non-solar rays, however. Their energies are signiﬁcantly higher
due to the speed at which these particles travel, which is often highly relativistic.
Therefore, they are usually completely ionized, i.e. the electrons are completely stripped
from the nucleus, due to the high the speed at which the particle is traveling. The

energies of these particles can range from GeV to as high as PeV. or quadrillions of

152

electron voltss‘. These high energies are a consequence of the highly energetic events
occurring in interstellar space. Originating from such cataclysmic events as supernovae.
these particles generally travel freely in the high vacuum of space, once free of local
magnetic ﬁelds. These two forms of radiation, which emanate from galactic and
intergalactic sources, are of primary concern when designing systems. They are highly

penetrating and highly destructive.

153

5.2 Electronics in Space

Current silicon based integrated circuits in interplanetary exploration probes and
satellites within Earth’s magnetic ﬁeld must be designed with SEE and TID effects in
mind. High levels of redundancy and radiation hardening are usually employed to cope
with the effects of highly energized radiation species. Radiation hardening includes
several techniques for increasing a circuit‘s resiliency in radiation. Using an insulator.
such as sapphire, as a substrate can greatly increase the Circuit’s lifetime in radiation
environments. Using static random access memory (SRAM) instead of dynamic RAM
(DRAM) for memory modules is also a common practice, since DRAM is more
susceptible to SEES. Error checking design schemes have proved highly effective in
reducing upset, where added circuit elements, for example, check bit parity or identical
redundant elements are queried and checked against each other for errors. Even with
these techniques, problems still persist.

The primary building block used in current device design is the metal-oxide-
semiconductor ﬁeld effect transistor (MOSFET), which have become ubiquitous in the
electronics industry due to its low power and large scale integration capabilities.
MOSFETs as well as other devices such as metal semiconductor FETs (MESFETs) and
bipolar junction transistors (BJTs) have a high susceptibility to radiation6‘7‘8. There are

several key areas of these devices that are especially sensitive and are discussed below.

5.2.1 Silicon Based Electronics

MOSFETS, MESFETS and BJTs rely on charge separated regions, orjunctions, to

operate. It is well documented that these junctions are prone to radiation damage.

154

whereby heavy ions), protons”), electronsl ', photons”, cause a buildup of charge
resulting in the malfunction of the transistor. This malfunction can vary in terms ofthe
degree of damage caused, and these different malfunctions are described in the
subcategories of SEES and TID as described above. Radiation may only cause a
temporary soft error remedied by a reset, or do signiﬁcant physical damage, resulting in a
hard error, rendering the transistor useless.

SEES, as well as TID, affect these sensitive regions through two basic
mechanisms, lattice displacement and ionization. Lattice displacement frequently occurs
in highly energized heavy ions and device interactions. These high-energy heavy ions are
especially problematic because they are highly penetrating and in most cases shielding is
not effective. One SEE upset with a heavy ion and the device could be rendered
inoperable. As the heavy ion interacts with the device it leaves charged ions in its wake
due to displacement damage. This line of charged atoms can occur directly through
sensitive components of the device. This displacement damage is usually permanent.
signiﬁcantly altering device operation characteristics.

Ionization damage can occur with all types of radiation, but signiﬁcant device
operation alterations generally take longer periods of time. While heavy ion lattice
displacement damage can happen nearly instantly, ionization effects can take years to
accumulate. Ionization effects limit the lifetime of a device and design considerations are
essential to limit these effects. However, even with modern designs, ionization, as with
heavy ion damage, can alter the device characteristics so drastically that the functionality

of the device can be irrevocably altered.

155

5.2.2 An Example: A MOSF ET in Radiation

There are many ways that a device can be affected by lattice and ionization
damage. MOSFETS are by far the most popular choice in devices today, so they will be
used as an example. A MOSF ET is comprised of three terminals, a source, drain, and
gate, and is simply an ampliﬁer acting as an on/off switch. The on/off states of the
MOSF ET are determined by the voltage applied to the gate and drain relative to the
source, which is usually regarded as the ground terminal, or reference. The gate can
induce a conduction channel between the source and drain via an electric ﬁeld, and is
separated from the channel by a thin insulating material, usually silicon dioxide. By
applying a voltage across the drain and source, and across the gate and source, a current
can be modulated-

Using this basic three terminal design, two variations of MOSFETs have
primarily been used in devices: the n-type and p-type MOSFET. In an n-type MOSFET
electrons are used to conduct current, and in a p-type MOSFET holes are used to conduct
current. Holes can be thought of as the opposite of electrons, as they have an opposite.
positive, charge. For n-type conduction to occur, a positive gate voltage relative to the
source is applied, and for p-type conduction, a negative voltage is applied. This provides
the fundamental ingredients for modern complimentary metal oxide semiconductor, or
CMOS, technology, which forms the basis for most electronic systems in operation
today.

In MOSF ET operation, the MOS, which is comprised of the metal or polysilicon
gate, the silicon dioxide insulating layer, and the conducting channel, is the essential

element in the on/off functionality of the device. The MOS layer is extremely thin to

156

maximize device response to applied voltages. The oxide layer is usually less than 10nm
thick and the induced conduction channel is around 10nm to 20nm thick. These elements
are highly sensitive to variations in the materials. The gate voltage induced conduction
channel can be affected by substrate dopant atoms, which can scatter electrons. Surface
states between the gate oxide and conduction channel can also cause scatter electrons.
The gate oxide can be affected by unwanted charge, usually due to impurities such as
sodium atoms. Therefore, current MOSFET technologies require an extremely stringent.
highly reproducible fabrication processes to mitigate these unwanted impurities and
defects in devices.

Due to these stringent criteria the MOS area is highly prone to radiation induced
upset. Charged particles of radiation can cause ionization and lattice displacement
damage in this area in a variety of ways. Radiation can cause arcing over the gate oxide,
essentially shorting out the gate terminal to the substrate, activate oxide impurities.
increase surface states between the oxide, and conducting channel and cause deterioration
of the gate oxide itself. Figure 5.2 shows an illustration of an energetic heavy ion passing
through sensitive areas in a MOSFET leaving behind displaced atoms and a line of
dangling bonds. In effect all these factors play a roll in shifting the threshold voltage. V7.
of the transistor, above which in an n-type device is considered on and below which the
device is considered off, and vice versa for a p-type device. In general this radiation
damage will push the threshold voltage down, making it easier to turn n-type devices on
and harder to turn p-type devices off. The threshold voltage'3 for n-type transistors is

given by

"I _\

Q,- Q.
VT=¢ -E-?1+2¢I;1

i i

157

where (1),,“ is the metal—semiconductor work function potential difference. (1),. is the
Fermi level relative to the center of the semiconductor band gap, and C 1. is the MOS
interface capacitance. Q, and Q. the effective MOS interface charge and depleting
region charge, respectively, are the values directly affected by radiation. These values
tend to increase when the transistor is exposed to radiation, causing the threshold voltage
decrease. For CMOS devices this threshold voltage shift is very sensitive, and is
compounded by the inherent design of CMOS: the use of both p-type and n-type

transistors.

 

p-type substrate

 

0 Heavy Ion
Figure 5.2. n-type MOSFET. Radiation induced damage by an energetic heavy ion. As the heavy ion
traverses through the semiconductor, it leaves a trail of defects. Image: B.W. Jacobs

Other sensitive areas in the MOSFET include junctions between dissimilar
materials. These junctions, comprised of lattice-mismatched material ﬁlms and metal on
semiconductors, are very thin and the accumulation of surface states is favorable. This
situation can sometimes cause a physical disruption of the interface resulting in a loss of
adhesion. This build up of interface charge can also result in shifts in the threshold

voltage.

158

As MOSFETS continue to decrease in size, they become more susceptible to SEE
and TID upset. Gate oxide thicknesses are becoming extremely thin, just a few
nanometers, so the conduction channel modulation is more sensitive to changes in gate
voltage. Any small disruption in this ultra thin gate oxide area may result in erroneous
functionality. Transistor operating voltages are also concurrently decreasing, which
increases the sensitivity of threshold voltage shifts due to radiation upset. Therefore
silicon technology is not only reaching its size limits in modern transistors, but its
viability in space applications is becoming increasingly compromised. Space-based
silicon electronics have come a long way, with redundant systems, radiation hardened
devices and shielding, but problems still persist. Therefore, new materials that are
inherently radiation resilient that can be used in electronic devices that meet or exceed the

current state-of—the-art silicon technologies are needed.

159

5.3 The National Superconducting Cyclotron Laboratory

The need to test new nanomaterials and nanocircuits in simulated space
conditions is essential to determine their viability in such environments. There are only a
handful of terrestrial facilities that can simulate the energies and representative species of
heavy ion radiation found in space. One such facility is the National Superconducting
Cyclotron Laboratory (NSCL) at Michigan State University”. The NSCL is particularly
suited for these studies because its available beam energies and species are a good

approximation of the energy spectra of abundant charged particles in cosmic rays. The

 

ability to accelerate any element on the periodic table makes it highly versatile for a

variety of heavy ion radiation experiments.

5.3.] The Coupled Cyclotron

The NSCL uses a coupled cyclotron system, and is a truly unique design. Each
cyclotron uses large superconducting magnets that bend charged particles accelerated by
the cyclotron radiofrequency system. The ﬁrst cyclotron used in acceleration is the
smaller K500, which was the world’s ﬁrst cyclotron to use superconducting magnets.
The second, larger cyclotron is the K1200, which until very recently was the most
powerful in the world, also uses superconducting magnets. Since 2001 these coupled
cyclotrons have produced highly energized, high current particle beams.

In order to achieve such highly uniform particle beams that are accelerated to
relativistic velocities, near the speed of light, a complex array of state-of—the-art

instruments, many of which were ﬁrst designed and developed at the NSCL, are used for

160

initial particle preparation, acceleration, and beam delivery. These instruments must
work in precise synchronicity for the entire process to work.

The ﬁrst step in preparing a beam includes extracting a particular element from a
source. This source is usually purchased in highly pure gas form, such as oxygen or
nitrogen. Sources that are initially in solid form are heated into a gas. These atoms are
usually electrically neutral and if an atom with a neutral charge, with the same number of
positively charged protons in the atom’s nucleus as negatively charged electrons
surrounding the nucleus, enters a magnetic ﬁeld nothing happens. This is similar to
touching a magnet to a non-magnetic material such as wood or plastic. There must be a
charge on the material that the magnetic ﬁeld can inﬂuence. Therefore, initial steps in
ionizing the atom, or stripping electrons to give it an overall positive charge, must be
taken so that it can interact with the magnetic ﬁeld.

This initial electron stripping is done using electron cyclotron resonance (ECR)
ion sources. In an ERC source the movement of electrons is used to prepare a particular
species of heavy ion gas. The movement of electrons can be inﬂuenced by magnetic
ﬁelds due to the Lorenz force. Electrons will move in a helical pattern around static

magnetic ﬁeld lines at a certain frequency given by:

e8.
(1) ’ =—,
CC m

where e is the charge of a single electron, 1.6x10"9 C, B is the magnetic ﬁeld strength

and m is the mass on an electron, 9.1x10'3' kg. An ERC ion source takes advantage of

this and introduces microwaves at the ECR frequency equal to ma, to heat the electrons.

The heated electrons will collide with the gas inside the chamber, which is the element to

be accelerated, and begin to strip electrons of the atoms, ionizing them. This group of

161

 

ionized atoms is called a plasma. After sufﬁcient ionization has occurred, a strong
electric ﬁeld pushes the plasma out of the ECR source down a beam pipe to the ﬁrst
cyclotron so the acceleration process can begin. The ECR source produces ions with
many charge states, and usually intermediate charge states are the most abundant and are
sent to the K500 to maximize the beam current.

Once inside the K500, a magnetic ﬁeld perpendicular to the trajectory of the ions
conﬁnes them while an oscillating electric ﬁeld accelerates them. In this process the
charged ion accelerates in a spiral trajectory running through the accelerating electric

ﬁeld many times over each time gaining more energy. As the particle increases in

 

velocity, the radial path of the particle grows while the frequency of oscillation remains
constant. The equation for circular motion for a particle having charge q and mass m in a

uniform magnetic ﬁeld of strength B is given by,

E _ (BR)2 612

A 2 m2.

 

 

E is the kinetic energy of the particle, A is the atomic number, B is the magnetic ﬁeld, R

is the radius of the particle’s path, q is the charge of the particle, and m is the mass.

 

(3102
2

is called the bending power. The increase in kinetic energy can either be

compensated by increasing the radius of the particle or by increasing the magnetic ﬁeld.
However, in order to keep the cost and size of equipment down, a small radius is
required. Therefore the magnetic ﬁeld must increase if high-energy particles are needed
for experiments. Note, higher values of q/m, result in higher values of E/A.

It was the discovery of the superconductor that has made high-energy. compact

cyclotrons possible. A superconductor is able to carry very large currents with no

162

resistance. Employing F araday’s law of induction, which states that an electric current
will induce a perpendicular magnetic ﬁeld, small, high-power magnets can be realized by
utilizing superconducting coils that carry extraordinary amounts of current with no
resistance. The superconducting coils that produce the magnetic ﬁelds, however, must be
kept very cold. Liquid helium at 4.2 K or -452° F is used to cool the coils so that they
maintain their superconducting properties.

After injection from the ECR ion source into the K500 superconducting magnet.
which can be tuned to between 3 and 5 Tesla (T), the charged particles are accelerated to
around 10% the speed of light, or around 20 MeV/nucleon. After the ions exit the K500,
they are injected into the larger K1200 cyclotron for further acceleration. The K1200 has
a stripper foil near the entrance of the beam line from the K500 cyclotron, which
typically increases the charge of the particle 2.5 times. The ion is accelerated from 10%
the speed of light to around 50% the speed of light, or from around 20 MeV/nucleon to
around 200 MeV/nucleon. The K1200 utilizes the higher charge state of the particle in
acceleration. With a higher charge, the electric ﬁeld used for acceleration can further
boost the ions to higher energies while the stronger magnetic ﬁelds can effectively
conﬁne the ions during its acceleration. After the acceleration process is complete, the
highly energized ions are extracted, and can be sent down a variety of beam lines where

different experiments take place.

5.3.2 The Single Event Effects Testing Facility

Multiple beam lines, which steer the beam to application speciﬁc vaults, allow for

a wide variety of experiments to be done”. One such beam vault is specially designed

163

 

for ion radiation experiments on electronics for space applications. The Single Event
Effects Testing Facility (SEETF)'°"7 is a NASA sponsored facility where energetic heavy
ions can be delivered to a remote controlled testing system. Figure 5.3 shows how a
beam of ions begins from the ECR source and is conditioned and steered to the SEETF
vault. The beam passes through a series of attenuators and apertures before it reaches its
destination.

Quads X'Y :Dosimetry Scint.

  

 

 

 

 

 

Aperture Steerers: E
. -{] =
D . . :
. Alignment .
=.Y9§.Win99~.v.........=
.._.,

Attenuator " D < ll K1200

ll 20-200 MeV/u

Upstream degrader .
to adjust ion energy

K500

 

13-17 MeV/u ( Ion Source
— _
e ' a
Attenuator to Blam stop
adjust ﬂux 0(200411555

Figure 5.3. Illustration of the particle acceleration process at the NSCL. Ions are sent from the source
through attenuators to adjust the beam current. They are then injected into the K500 cyclotron and then the
K1200 for acceleration. Degraders and attenuators adjust the beam properties, which is sent to the
speciﬁed beam vault for experiments. The boxed area in the upper right represents the SEETF vault and its
components. Image: adapted from R. Ladbury et al. IEEE T. Nucl. Sci. 51, 3664 (2004) with permission.

164

 

 

There are several components to the SEETF used for beam preparation, delivery
and analysis. First the beam must be of the correct energy and be uniform. Aluminum
degraders are used to lower the beam energy and make it uniform. The parallel plate
avalanche counter (PPAC) is used to obtain beam details. In order to achieve a desired
dose, the ﬂux, uniformity and position of the beam are all important for a successful
experiment. The PPAC is used to count beam particles and obtain position and
uniformity information, essential for beam tuning. The ions then travel out of vacuum
into air through a thin zirconium foil. After exiting, the particles pass through a plastic
scintillator, which is similar to the PPAC in that it counts the particles passing through it
and notes their position. The scintillator veriﬁes beam uniformity and particle ﬂux
without signiﬁcantly degrading the beam. It can be used during experiments to obtain
real time beam information to determine target dose throughout the run. If the maximum
particle ﬂux possible is desired, the scintillator can also be retracted. The particles then
travel through air to reach the beam target. The on target beam size is usually determined
by the user, with a maximum size of 5 cm by 5 cm. If a larger beam is needed, rastering
the beam is also possible to achieve a larger irradiated area. The target is mounted on a
stage capable of moving 10 inches in the x-direction and 20 inches in the y-direction with
70° of rotational freedom, and can be controlled by an adjacent laptop computer or a
remote computer using LAPLINK'B. Target positioning is determined by a laser
crosshair that denotes the center of the beam. If the beam passes through the target and
stage it will hit a graphite block, stopping the beam. Figure 5.4 shows the equipment
setup in the SEETF. A photo shows the primary equipment components. The illustration

below the photo also shows the components in a more readable form.

165

 

 

seen" PPAC ....
Gate Valve Degrador DUT

 

 

 

 

 

 

 

 

 

 

 

 

 

Qt
— " a:
U ——’ / ' E
2"x2“ Colllmator } - m
75 pm Zr 3 ' ' ’
0 [2W4] IEEE Winm Pas

Figure 5.4. Image ofthe NSCL SEETF. Each main component of the SEETF is noted. The ion beam runs
through the PPAC for analysis, exits through a zirconium foil. runs through the plastic scintillator, hits the
target, and hits the graphite beam stop. Top image: B.W. Jacobs Bottom image: R. Ladbury etu1., IEEE T.
Nucl. Sci., 51, 3664 (2004) with permission.

When testing devices in the SEETF, electronic characterization equipment can be
placed inside the vault near the target. The beam will not disrupt the characterization
equipment, usually supplied by the user, since it is highly focused on the target sample.

Several electrical outlets are available within the vault for all required equipment.

166

User and SEETF equipment can be controlled remotely through the NSC L
Ethernet. This signiﬁcantly enhances the capabilities of the facility. Multiple target
samples can be irradiated by remotely moving the stage. A remote camera allows precise
target placement using a laser crosshair as a reference. Characterization equipment
controlled via LabVIEW, for example, can also be remotely controlled using the
LabVIEW software or Windows Remote Desktop. This adds increased capability for real m
time testing and data acquisition while the beam is on target.

The versatility of the SEETF also lends itself to other possible user built test
environments. A low or high temperature system could be integrated into the stage
system, or a small vacuum device could also be incorporated. With these add-ons, the
SEETF could better simulate conditions found in space. Given the SEETF capabilities
and experiment possibilities, along with the high quality primary heavy ion beams
available at the NSCL, an ideal terrestrial setting for electronic and material testing and

benchmarking for current and promising systems for use in space environments, as well

as radiation environments found on Earth, can be realized.

167

5.4 Reduced Dimensionality Materials in High Radiation Environments

 

Novel reduced dimensionality materials, or nanomaterials, have already changed
the face of modern science. In recent years a wide variety of materials with inherent or
engineered reduced dimensionality components have gained significant interest. The
quantum phenomena the small spatial dimensions invoke can lead to enhanced device
performance. Many of these systems closely resemble theoretical reduced dimensionality
systems and provide excellent experimental evidence of many quantum theories.

Nanomaterials may open up a wide array of electronic as well as mechanical

applications for use in space environments. They have been shown to possess an inherent

 

increased radiation resiliency, whereby damage propagation mechanisms that meet
fundamental conservation law requirements change in reduced dimensionality systems in
ways that may result in enhanced radiation resilience”. Therefore, the use of nanomaterials
in electronics devices and mechanical systems may result in overall enhanced system
performance without the need for heavy shielding and high redundancy.

Experiments to test nanomaterials and devices in a heavy ion radiation
environment was essential in assessing how well these materials could maintain their
structural integrity in a space environment. Candidates chosen for radiation testing
represented nanomaterials with a high likelihood of success in harsh environments and
nanomaterials that were very versatile for potential use in variety of applications. These
three candidates were carbon nanotubes, single and multi walled, gallium nitride
nanowires and carbon onions. The properties of these materials will ﬁrst be discussed

and then the results of the material testing in heavy ion radiation will be presented.

168

 

5.4.1 Materials
5.4.1.1 Gallium Nitride Nanowires and GaNFETs

Multiphase gallium nitride nanowires were chosen as potential candidates due to
their pseudo one-dimensional structure that leads to the variety of promising electronic
and optical properties as discussed in chapter 4. The bond between gallium and nitrogen
is strong and its radiation resiliency has been shown to be high compared to other
semiconductor materials such as gallium arsenidezo. The nanowires used were the same
as described in chapter 4. The GaNFETs were also fabricated using the methods

described in chapter 4.

5.4.1.2 Carbon Nanotubes

Carbon nanotubes have sparked much interest since their discovery in 19912', as
they posses remarkable electrical, mechanical and chemical behavior”. A carbon
nanotube (CNT) is essentially a rolled up single sheet of graphite, or graphene, consisting
of hexagonal rings of carbon atoms arranged in a honeycomb lattice. The wall of a CNT
is only one atom thick, with a diameter of ~1 nm. It can reach millimeters in length, and
possibly longer, resulting in very large aspect ratios. This system is therefore very close
to a theoretical one-dimensional system.

Although CNTs are grown as tubes, it is convenient to describe a CNT based on
how it is “rolled up” from a sheet of graphene, as shown in ﬁgure 5.5. The chiral vector.
Ch, points down the axis of the nanotube, and deﬁnes the orientation of the hexagonal
carbon rings with respect to the nanotube axis. Perpendicular to CI, is the translation
vector, T, which deﬁnes the tube axis. The Ch and T vectors also deﬁne the unit cell of

the CNT, or the basic building block of the tube. CI, is a combination of the unit vectors

169

 

 

of graphene, and is a multiple of these vectors, where. Ch 2 na/ + ma; 5 (n,m). Generally

only the (n,m) notation is used to describe a CNT structure.

 

\ .
(n,n) armchair
Figure 5.5. The (n,m) CNT labeling system. T is the tube axis. Ch is the chirality vector and describes
how to roll up the graphene sheet. a] and a2 are the unit vectors ofgraphene. (n,0) and (n,n) are special

chirality vectors denoting zig-zag and armchair CNTS, respectively. The shaded box denotes the unit cell
ofthe nanotube.

CNTs can be characterized as being achiral or chiral based on their structural
symmetry”. There are only two types of achiral nanotubes, zig-zag and armchair,
corresponding to CI, = (0,n) and Ch : (n,n), respectively. All other nanotubes are chiral.
Figure 5.6 shows three examples of speciﬁc nanotubes including a (0,10) zig-zag, a
(10,10) armchair and a (7,10) chiral. CI, can also be used to calculate the diameter, (1,. of

a nanotube with,

C , ,
J = ﬁa(._(.(m‘ + mn + nz)'/‘/Jr;
Jr

d:

t

170

where 80C is the nearest neighbor carbon-carbon bond distance (1.421 A in graphite) and
O, is the length of C1,. These two nanotube properties, chirality and diameter, inﬂuence
the behavior of nanotubes in signiﬁcant ways. With only subtle changes in these
variables, a nanotube can have quite different and surprising properties. For instance, the
band gap of the nanotube is dependent on its chirality. For a given (n,m) nanotube, when
n-m is a multiple of 3 it is metallic, if it is not a multiple of 3 it is semiconducting. The
band gap of semiconducting nanotubes can also vary and depends on the nanotube

diameter.

(0, I 0)
(zig-zag)

(10,10)

(armchair)

(7,1 0)
(chiral)

 

Figure 5.6. Types of CNTS. There are three main types of CNTs, zig-zag, armchair, and the rest, chiral
nanotubes.

Carbon nanotubes are being investigated for use in high strength materials”,

25‘2”, and chemical sensorszns. Simple carbon

metallic interconnects in integrated circuits
nanotube logic circuits”‘30 have been fabricated and have shown levels of performance

that meet or exceed those of state-of-the-art integrated silicon based logic devices3 '.

l7l

 

 

‘1?

 

CNTs are very promising candidates for use in space applications due to their
versatility in many devices and systems. The carbon-carbon bond in the sp2
conﬁguration as found in nanotubes is possibly the strongest bond found in nature.
Therefore physical durability of a CNT is inherently high. It has also been theoretically
shown that carbon nanotubes can self-heal after irradiation damagen‘”. In heavy ion
radiation, damage in CNTs occurs via a direct knock on and secondary knock on
displacement of carbon atoms“. Atomic vacancies are ﬁlled through dangling bond
saturation and the lattice is reorganized to compensate for the loss of an atom.

Both single walled carbon nanotubes (SWCNT), which consist of one atomic
layer of graphene as described above, and multi walled carbon nanotubes (MWCNT),
multiple SWCNTS in a nested coaxial structure, were tested. MWCNTs share many
properties of the SWCNT in their high strength and electron conduction ability. Figure

5.7 shows HRTEM images of typical single, 5.7(a) and multi walled CNTs 5.7(b).

   

'1 " -- ‘ . 10nm 5mm
'-V"'.. ‘e 3 _‘3 .i _ —:
Figure 5.7. HRTEM images of SWCNTS and MWCNTS. (a) A single SWCNT can be seen in the middle
ofthe image running nearly vertically. The single wall structure is easily seen. Disordered carbon is also
seen on the surface of the nanotube. (b) The nested MWCNT structure is seen in this figure where each
line indicates a wall. Images: B.W. Jacobs.

172

SWCNTS obtained from Tubes@Rice, now Carbon Nanotechnologies, Inc.35 and
NASA Johnson Space Center were grown via the high-pressure carbon monoxide
conversion (HiPco) method”. This method uses heated, high-pressure CO as the source
of carbon and uses a catalyst particle, usually a transition metal, to initiate nanotube
growth. The HiPco method yields high quality nanotubes with only small amounts of
disordered carbon and catalyst impurities. The sample from Tubes@Rice came in the
form of buckypaper, which is a thin sheet of CNTs around 10 mm2 in area and 1 mm

thick. This ensured maximum heavy ion beam-sample interactions due to the relatively

 

large cross-sectional area of the buckypaper.

Multi walled CNTs were obtained from NASA Goddard Space Flight Center and
were grown via a catalyst-free arc discharge method”. In this growth method a welder
creates an electric arc across a carbon anode and carbon cathode causing the carbon
anode to vaporize. An inert gas is pumped through the area of vaporization so that
oxygen does not interfere with the process causing oxidation of the carbon samples.
During vaporization the cathode is cooled in water. This method, while creating
abundant multi walled CNTS. also creates a large amount of amorphous carbon in which
the CNTs are embedded. A post-grth treatment can be done to reduce the amount of
amorphous carbon using nitric acid or toluene, but no such treatment was done for these
MWCNT samples. The MWCNTs were of high quality as shown by the HRTEM image

above.

173

5.4.1.3 Carbon Onions

A carbon onion is a similar form of carbon as a MWCNT in that it consists of
nested single layers of graphene rolled into a ball, as shown in ﬁgure 5.8. These
graphene balls that make up an onion are called buckyballs, and are variations on the C60
molecule, which is shown in ﬁgure 5.8(a) in the inset. Carbon onions, like CNTs, consist
of carbon-carbon bonds the sp2 conﬁguration, which may be the strongest bond in nature.
In radiation carbon onions may exhibit self-healing effects seen in nanotubes due to the

similar structure.

  

Figure 5.8. HTREM image of carbon onions. (a) A single round carbon onion is shown in this ﬁgure.
The inset shows a model of a C60 molecule, which is the basis of carbon onions. (b) These carbon onions
formed polygonal sides with some faceting on the outer layers. Images: B.W. Jacobs

Carbon onions were chosen as potential candidates for space applications because
they exhibit excellent lubrication properties”. Lubricants in space must be able to
maintain a low coefﬁcient of friction over a long period of time in very harsh conditions.
Lubricants in space-based systems cannot be easily changed in the event of material

failure due to their remote locations. Radiation is but one factor effecting lubricants in

174

 

space-based systems, drastic temperature change and the vacuum environment are also
primary factors effecting lubrication performance. Carbon systems such as nanotubes
and onions have been shown to respond very well to temperature variations due to their
high thermal conductivity and durability. Carbon onions also act as a solid lubricant so
the vacuum environment does not affect the onions like it does with liquid lubricants,
where leaks can sometimes occur.

Vacuum conditions have been shown to affect the performance of carbon onions.
Studies have shown that carbon onions maintain low coefﬁcients of friction in air39‘40.
However, when tested in vacuum, the coefﬁcient of friction increases. A loss in ﬁlm
stability is also seen, where the carbon onions slide away from the moving surfaces,
resulting in very high coefﬁcients of friction. Treatment in hydrogen peroxide (H202)
drastically improved the lubrication performance of carbon onions in vacuum, reducing
the coefﬁcient of friction and increasing ﬁlm stability“). Graphite is also a potential
candidate for use as a solid lubricant in space, but the increase of the graphite friction
coefﬁcient in vacuum is a well-known effect, and heavy metal compounds are usually
added for solid lubrication applications in vacuum conditions. Carbon onions do not
require heavy metal additives, therefore representing a more environmentally benign
solid lubricant.

Carbon onion samples were produced at Tokyo Institute of Technology from
nanocrystalline diamond with diameters of around 5 nm, as shown in ﬁgure 5.9. The
nanocrystalline diamond was heated to temperatures between 1700° C to 23000 C in an
infrared radiation furnace in argon at atmospheric pressure. This procedure can produce

relatively large amounts of carbon onion sample quickly and easily. Thermogravimetric

175

J

 

analysis revealed that onions produced at higher temperatures have a higher vaporization

temperature meaning onions with fewer defects or amorphous carbon were produced.

   

Figure 5.9. HRTEM image of norystalline diamond ith diameters of5 to 10 nm. This is used as a
starting material for producing carbon onions. Image: B.W. Jacobs

Each material investigated for heavy ion radiation resilience has potentially high
impact return if they can consistently perform in heavy ion radiation. These three types
of materials have a wide array of potential applications. GaN nanowires and carbon
nanotubes can be used in electronic, optical, mechanical, and sensor applications. Carbon
onions could be used as a solid lubricant for moving parts on mechanical space based
systems such as a moving antenna or a roll out mechanism for solar panels. Each
material, however, must be tested in real world radiation conditions. These materials
must maintain high levels of performance in harsh environments before they can be

implemented in space-based devices and systems.

176

 

5.4.2 Experiments

Heavy ion radiation experiments carried out at the NSCL consisted of a series of
experiments done to investigate the low to high Z heavy ion interactions with
nanomaterials and electronic devices with integrated nanomaterials as active elements.
This series of experiments was funded through an NSCL internal competitive grant that
included 92 hours beam time and collaboration with two senior NSCL physicists. Three
separate radiation experiments were done at the NSCL with three different heavy ion
beams: krypton-86, krypton-78 and calcium-48 at energies that were close to the energy

spectra of heavy ions found in space.

5.4.2.1 Krypton-86: The First Experiment

In the ﬁrst experiment a buckypaper sample consisting of SWCNTS and a powder
sample consisting of MWCNTs embedded in amorphous carbon were irradiated with a
krypton-86 beam at 142 MeV/nucleon, or 12.212 GeV/atom. The experiment was done
in the N4 vault at the NSCL. The beam was focused to uniformly irradiate an area
having dimensions of 20 x 20 mm, as measured from known dimensional markings on
the beam-viewing scintillator placed at the beam line exit vacuum window. The vacuum
window was a 0.022 mm thick zirconium foil. In this experiment carbon samples were
put in close packed in quartz tubes, as shown in ﬁgure 5.10, and placed as close as
possible to the zirconium exit window in the beam line. Thin walled quartz tubes were
used in order to minimize secondary ion production. Each individual quartz tube had an
outer diameter of 3 mm and an inner diameter of 2 mm. The beam current was measured

at the exit point using a Faraday cup, which measured a stable current of 0.3 particle

177

 

 

picoamps (ppA), or 1.87 x 10‘5 particles per second (pps). Therefore, the particle ﬂux was

4.68 x 103 pps/mmz.

      

"it
Figure 5.10. The carbon samples were placed in quartz tubes and put in a close packed conﬁguration.
These samples were irradiated simultaneously. The heavy ions exited into air through a thin zirconium
window and passed through the quartz tubes with an on target beam energy of over I 1 GeV, as shown in
the left image. The samples packed in the quartz tubes are shown on the right. Images: B.W. Jacobs

To quantify the amount of radiation a sample receives, a dose calculation is done.
The total dose of radiation depends upon incident beam energy, beam current, sample
volume and density and all losses. The ﬂuence of the beam determines the number of
particles passing through a cross-section of the beam per unit time. In conjunction with
the cross-sectional area of the samples, this determines the number of particles passing
through the sample under test per unit time. From this and the energy loss per nucleon,
the amount of energy deposited in the sample for each ion passing through the cross-
section of the sample can be determined, giving an effective dose per unit time for a
given beam ﬂuence and energy. From the dose per unit time, the duration of the

exposure of the target to the beam for a desired total dose of radiation can be determined.

178

To estimate the on target beam energy, the energy loss in each stage of the beam
line was calculated using Stopping and Ranges of Ions in Matter (SRIM)‘“. The energy
loss due to the Zr exit window was about 63.13 MeV/atom. In the 63 mm air gap
between the vacuum window and the ﬁrst target samples, the energy loss was about 502.3
MeV/atom. The energy loss after traversing the ﬁrst 0.5 mm of quartz (density = 2.65
g/cm3) was 359.2 MeV/atom. The on target beam energy for the ﬁrst row of material was
approximately 11.29 GeV/atom or 131.2 MeV/nucleon. The irradiated area of each
sample was estimated to be 16.92 mmz. The density of CNTs was estimated to be 2.24

g/cm3, which is the density of single crystal hexagonal planar graphite.

 

5.4.2.2 Krypton-78: The Second Experiment

In the second experiment, SWCNTS, MWCNTs, GaN nanowires, and GaNFETs
were irradiated with a krypton-78 beam at 140 MeV/nucleon, or 10.92 GeV/atom. The
beam was focused to uniformly irradiate an area having dimensions of 20 (height) x 10
mm (width). The estimated beam current during the experiment was .3 ppA and the
current was varied from 1.6x105-106 pps. The particle ﬂux was calculated to be 5x103
pps/mmz.

Samples were packed between a silicon wafer and thin (~0.25 mm) mica covers
that were glued to the wafer using silicone and placed in the beam line, as shown in
ﬁgure 5.] l. The silicon wafer was glued to a piece of particleboard that was mounted to
the SEETF movable stage. This sample conﬁguration proved much easier to work with
than the quartz tube samples. The beam passed through a 0.27 mm four-quadrant plastic

scintillator plate, used to continuously monitor the incoming beam current, a .075 mm Zr

179

exit window, a 432 mm air gap, and the 0.25 mm mica sheets. The energy loss was
calculated to be 226.2 MeV/atom through the Zr foil, 210.6 MeV/atom through the
scintillator plate, 358.8 MeV/atom through the air gap, and 421.2 MeV/atom through the
mica cover. This resulted in an on target beam energy of 9.703 GeV/atom or 124.4
MeV/nucleon.

In this experiment the sample dose calculations were more carefully considered.
Given the ability to irradiate each sample individually, a more accurate dose could be
imparted on each sample. Therefore, doses based on the on target beam energy and each
individual sample mass, accurately measured with a highly precise scale (Denver
Instruments M-220D), and cross-sectional area were used to calculate the time needed to

reach the predetermined dose.

 

 

 

 

V
Figure 5.11. The carbon and gallium nitride nanowire samples were placed on a silicon wafer that was
attached to a piece of particleboard with silicone adhesive. The particleboard was then screwed onto the
SEETF movable stage. The samples were held in place by a small mica sheet that was attached to the
silicon wafer with silicone adhesive. Each sample could then be separately irradiated with a speciﬁc dose.
The heavy ions exited into air through a thin zirconium window and passed through the mica sheets with an
on target beam energy of over 9.7 GeV, as shown in the left image. The samples that were placed on the
silicon substrate are shown on the right. Images: B.W. Jacobs

180

A GaNFET was also tested in the krypton-78 beam. Its electronic characteristics
were measured in real time while the circuit was being irradiated by energetic heavy ions.
This was made possible by the NSCL Ethernet and LabVIEW computer control of the
characterization equipment. Using Windows Remote Desktop, the computer running
LabVIEW in the vault was remotely controlled and parameters such as the gate-source
voltage and drain-source voltage were varied at will during irradiation. The GaNFET
was fabricated as described in chapters 2 and 4. It was placed in ceramic package and

wire bonded. Wire bonding was done at Howard University. David Liu at NASA

Goddard Space Flight Center wrote the LabVIEW program.

Moveable Stage 1 Wire Bonds
\

Figure 5.12. The electronic characteristics ofa GaNFET in radiation were measured. (a) The circuit was
mounted in a ceramic package and mounted on particleboard and attached to the SEETF moveable stage.

(b) A close up ofthe circuit shows its wire bonds. Images: B.W. Jacobs Wire bonding and package mount:

James Grifﬁn, Howard University

The I-V characteristics of the GaNFET were tested before, during and after heavy
ion irradiation (Keithley 487 Picoammeter/Voltage Source and HP 6633A DC Power
Supply). In measuring the I-V characteristics the drain-source voltage was continually

swept from 0 V to 22 V, while the gate-source voltage was stepped through from 0 V,

181

    

3V, 6V and 9V. No dose was calculated for the GaNFET but rather a ﬂuence of the
beam was started low at 102 pps over a beam area of 200 mm2 and was gradually
increased to 104 pps, 105 pps and ﬁnally 106 pps, corresponding to 3x104 to 3x107

particles/cmz. The beam was on target for 600 s for each ﬂuence.

5.4.2.3 Calcium-48: The Third Experiment

In the third experiment, SWCNTS and MWCNTs in addition to carbon onions
were irradiated with a calcium-48 beam at 140 MeV/nucleon and 70 MeV/nucleon, or
6.72 GeV/atom and 3.36 GeV/atom, respectively. This experiment was also done in the
SEETF.

The SWCNT, MWCNT and carbon onion samples were individually mounted in
a similar conﬁguration as the previous experiment behind thin 0.2 mm quartz cover slips
on a silicon substrate with a native oxide layer. The SEETF beam path passed through a
0.075 mm zirconium foil vacuum exit window, a 432 mm air gap, and the 0.2 mm quartz
cover slips. Beam energy calculations were performed using the SRIM software
package. At an incident beam energy of I40 MeV/nucleon, SRIM indicated a loss of 430
MeV/atom, across the zirconium foil vacuum exit window, the air gap, and the quartz
restraining sheet. This loss was 759.84 MeV/atom for the 70 MeV/nucleon beam. The
resulting on target beam energies were 6.29 GeV/atom or 131.04 MeV/nucleon and 2.6
GeV/atom or 54.17 MeV/nucleon.

There are no published guidelines for high-energy heavy ion beam exposure times
for nanomaterials and nanocircuits. The experiments were therefore conducted with

doses and exposure times comparable to those that can be expected to damage or destroy

182

 

 

current state-of—the-art silicon circuits. The failure under irradiation for a diverse range
of silicon devices and technologies indicate that damage thresholds are at about 10 Gray
(Gy), where 1 Gy = 1 Joule/kg. Thresholds for radiation-hardened devices are at about
1000 Gy. Almost all devices will fail before doses reach 105 Gy. Therefore, 100 Gy,
1,000 Gy and 10,000 Gy were set as the target doses. Corresponding times to achieve
these doses were calculated based on the estimated on target beam energy, sample
weight, density and cross sectional area. The density of each nanomaterial is also
essential in calculating an accurate dose. These nanomaterials do not have well deﬁned
densities and are many times very fluffy and difﬁcult to work with. Therefore a rough

estimation of the density of each nanomaterials was used in calculating the dose.

5.4.3 Experimental Results and Discussion
5.4.3.1 Gallium Nitride Nanowires and GaNFETs

GaN nanowires were irradiated at the SEETF with highly energetic krypton-78
heavy ions with an on target energy of 128.4 MeV/nucleon. After irradiation, GaN
nanowires were analyzed with HRTEM and exhibited minimal radiation coupling. The
majority of nanowires analyzed showed no radiation damage. When damage was
observed radiation coupling occurred via a previously unreported short track mechanism.
This radiation coupling took the form of a crystalline to amorphous transition in the outer
layer of the nanowire, as shown in ﬁgure 5.13. An amorphous plume was observed and
was attributed to heavy ion radiation damage. An FFT of this and the surrounding areas.
shown in ﬁgure 5.13(b), Show ring patterns as well as spot patterns, indicating that some

crystallinity was retained after irradiation. Not all nanowires irradiated. however, showed

183

 

 

damage. Only a minority of the nanowires observed via HRTEM showed the crystalline

to amorphous transition.

    

Figure 5.13. TEM images of irradiated GaN nanowire. (a) A TEM image shows a plume ofmatcrial.
which may have resulted from heavy ion radiation. (b) An HRTEM image ofthe plume Showing its
amorphous character. The FFTS of the indicated areas show the ring structure associated with amorphous
or polycrystalline material. Images: B.W. Jacobs

A second nanowire observed with the TEM after heavy ion radiation shows rough
outer layers that may have been caused by the radiation, as shown in ﬁgure 5.14(a).
HRTEM analysis indicated that the rough surface was comprised of amorphous plumes
of GaN, as shown in ﬁgure 5.14(b), where the high crystallinity of the nanowire normally
seen in the non-irradiated samples was not observed. An FFT of the boxed area has a
ring pattern and a spot pattern indicating that the crystallinity in this area was retained.
Other images of the internal structure of the nanowire, shown in ﬁgure 5 .14(c), revealed
what looked to be areas not affected by the heavy ions. The coherent interface between
the zinc-blende and wurtzite phase normally seen in these nanowires was not damaged in
any observable way. FFTS that did not have signiﬁcant ring patterns conﬁrmed the high

crystallinity of these areas.

184

 

 

 

    

.- 4..
Figure 5.14. TEM images of an irradiated nanowire. (a) The TEM image shows a nanowire with a rough
outer surface. which may have resulted from heavy ion radiation. (b) HRTEM image of the boxed area in
(a). An amorphous plume is shown here. An FFT taken from the boxed area shows a ring pattern
associated with an amorphous and polycrystalline region. (c) HRTEM image taken from the boxed area in
(a). The zinc-blende/wurtzite coherent interface is shown here. No damage is observed in this picture and
FFTS do not show signiﬁcant ring patterns. Images: B.W. Jacobs

The GaN nanowire sample was exposed to a uniform 2 cm2 beam centered on a
~0.l6 cm2 sample using the laser alignment. The highest radiation dose was 6.4x105 pps
irradiation for 900 s. As shown in the GaN matrix SEM images in chapter 4, the
nanowires were somewhat dispersed throughout the matrix. Therefore. an actual dose for
the nanowires was roughly estimated to be about 300-400 Gy.

The heavy ion radiation damage mechanisms in nanowires are also not well
deﬁned. Damage in traditional semiconductors is usually seen through a long track of
displaced atoms and dangling bonds in the material corresponding to the path of the
energetic heavy ion. No tracks were observed in the nanowires, only local areas of
disordered material, and some nanowires did not show indications of heavy ion radiation
damage. This could be a result of the strong bond gallium and nitrogen form in GaN, or
the nanowire cross-section could be small enough that heavy ion interactions do not
frequently occur.

GaNFETs were also irradiated in a krypton-78 heavy ion beam in order to study

device operation in real time under radiation. The GaNFET was exposed to highly

185

 

 

energetic heavy ion radiation that would cause traditional silicon based MOSFETS to
malfunction. No dose was calculated for the GaNF ET but rather a ﬂuence of the beam
was started low at 102 pps over a beam area of 200 mm2 and was gradually increased to
104 pps, IOS pps, and ﬁnally 106 pps, corresponding to 3x104 to 3x107 particles/cmz. The
beam was on target for 600 s for each ﬂuence. While the beam was on target, real time I-
V curves were measured. The GaNF ET showed minimal damage after every radiation

exposure and continued to operate after irradiation was complete, see ﬁgure 5.15.

250n1 4
3 - 10 pps 9VGS .
200n . ' 105995 Was ‘
A . . 106pps 9vGS :
é 150m v Pre-Rad 9VGS .
< I ,
‘7, 100n ~ '
.0 I
50h ~ , g I
'v"
I v
0n '. . , . ,

 

 

o 5'1'0'1'5‘20 25
VSD (Volts)

Figure 5.15. Multiple VDS sweep ofa GaNFET in radiation. V05 was kept constant at 9V. l-V
Characteristics: V.M. Ayres et al. Diam. Relat. Mater., 15, 1 1 17-1 121 (2006) with permission.

In addition to the material and small size of the nanowire, the GaNFET design
may also make it inherently radiation resilient. When comparing a GaNFET to a
traditional silicon based MOSFET design, there are several key differences in
architecture that may enhance its radiation resilience. There are fewer junctions in a
GaNFET that are susceptible to radiation damage. The gate oxide layer is not as sensitive

to radiation damage in a GaNFET. Since modern MOSFETS employ a very thin gate

186

 

 

 

oxide layer to modulate an inversion layer directly underneath it, it is very sensitive to
current leakage through the gate oxide. It is therefore very sensitive to ionizing damage.
GaNF ET operation is different in that it uses the gate ﬁeld to modulate the height of the
Schottky barrier formed between the metal-nanowire junctions. This junction is,
however, susceptible to radiation damage, as is any junction of dissimilar materials.
Combined with the radiation resilience of GaN nanowires, and their excellent material

properties GaNFETs may be a viable solution for space-based electronics.

5.4.3.2 Single Walled Carbon Nanotubes

SWCNTS were irradiated with krypton-86, krypton-78 and calcium-48. Post
irradiation analysis using Raman spectroscopy, HRTEM and SEM indicated that the
samples sustained no observable damage. Raman spectroscopy studies across all samples
showed the characteristic double peak at ~1580 cm'l corresponding to semiconducting
and metallic nanotubes, as shown in ﬁgure 5.16. This showed that the CNTs retained
their overall tubular structure. The ~1330cm'l peak is the disorder carbon peak (D-peak),
and after large doses of heavy ions, was expected to increase as the nanotube structure
was broken down due to heavy ion knock on displacement damage. This did not occur,
and in every instance, only a small disordered peak was observed, with no correlation
between the peak characteristics, including the number of counts or its full width half
max (FWHM), and the radiation dose. Subtle variations in the double peak structure
were observed, but these variations are as yet unexplained. Figure 5.l6(a) shows the
SWCNTS Raman spectra before and after the krypton-78 irradiation. The characteristic

double peak is still present after both doses. Figure 5. l 6(b) shows the SWCNTS Raman

I87

 

 

spectra before and after the calcium-48 irradiation. The characteristic double peak is still
present after both doses. These doses were much higher and the SWCNT structure was
maintained. The SWCNTS from the krypton-86 experiment are not shown, because the

calculated dose was much smaller, and very similar spectra to those shown below

 

 

 

 

 

 

resulted.
(a) I M170 OOOG 140M v
10000 Gy .‘ ‘ . y. e
7,; - 1563 2cm ‘fl1590-3cm 7}? 1 1554c,“ 158351ch
+- ..‘ :f: - ‘
g. _1 .I 5 ___.1.5a2c.m,. KL..-

. _ ,_ ,,,,, 1.3338“ - ...."~' \_ . 214000Gy 70Mev55 .1 33 58656
3 was“ ‘ " . 2 ' are 5‘ “
v - 'I15935cm v -1 15452: " '\

3,, 1549,10maI w , .1_342_.5cm..._...._ CLLg.
who .1 1342.1CITI ‘ K +4 ,

C ....................... '7 ._..... ..-.:.....: C (I.

8 « 1 3 5 .
0 Pre Rad 155520m'1i'158226m 8 « Pre Rad 1555 Zem g ‘.1582 2cm‘

4 a) I .1 .
-. 13...... 1___‘i3_‘_‘.°_'i‘..'~>./a x--.“ - w 15.3913» ‘x .....
1200 1300 1400 1500 11600 1700 12100 13100 1400 15100 16100 17'OO
Raman 311'“ (cm ) Raman Shift (cm")

Figure 5.16. Raman spectra of irradiated SWCNTS. (a) Before and after krypton-78 irradiation. (b)
Before and after calcium-48 irradiation. After all irradiation the SWCNTS indicated no observable damage.
Spectra: B.W. Jacobs

SEM and HRTEM images shown in ﬁgure 5.17 also showed no observable
structural damage after krypton-78 irradiation. Figures 5.17(a) and 5.17(b) are SEM and
HRTEM images of SWCNTS before radiation. Figures 5.17(c) and 5.17(d) are SEM and
HRTEM images of SWCNTS after a 1000 Gy dose of krypton-78 radiation. Figures
5.17(c) and 5.17(f) are SEM and HRTEM images of SWCNTS after a 10,000 Gy dose of

krypton-78 radiation. The images show pristine nanotube structures.

188

 

 

Figure 5.17. SEM and HRTEM images ofsingle walled CNTs before and aﬁer irradiation. (a) SEM
before irradiation. (b) HRTEM before irradiation. (c) SEM after l,OOOGy krypton-78 irradiation. (d)
HRTEM after l,OOOGy krypton-78 irradiation. (e) SEM after 10.000Gy krypton-78 irradiation. (t)
HRTEM after 10,0000y krypton-78 irradiation. Images: B.W. Jacobs

One single walled CNT observed, however, did show one alteration. After a high
dose of krypton-78 ion irradiation, the nanotube was packed with C60 molecules, see
ﬁgure 5.18. When buckyballs are packed within nanotubes they are called peapods. It is
believed that these peapods may be a direct result of ion irradiation. This result is under

further investigation.

189

 

Figure 5.18. HRTEM image of SWCNT after 10,0000y krypton-78 irradiation. This image shows
carbon-60 packed into a SWCNT, a peapod, possibly due to heavy ion irradiation. Image: B.W. Jacobs

The cylindrical wall structure of the SWCNTS held up very well at all radiation
doses, with doses exceeding 200 kGy of calcium-48 radiation as indicated by the
HRTEM images and the Raman spectra. This is very high radiation resiliency. These
results are very promising for possible integration of carbon nanotubes in a wide variety

of devices for space applications.

5.4.3.3 Multi Walled Carbon Nanotubes

Multi walled CNTs were also irradiated with krypton-86, krypton-78 and
calcium-48 heavy ion beams. The results of these experiments closely parallel those of
the single walled CNTs in that no beam damage has thus far been observed using Raman
spectroscopy, HRTEM, and SEM analysis.

Figure 5.19 shows Raman spectra of MWCNTs before and after krypton-78 heavy
ion radiation. The D-peak in these spectra, ~1337 cm", is much larger than in the
SWCNT case. This was due to the process in which these MWCNTS were grown, where

a by-product was amorphous carbon. This amorphous carbon can be removed with an

190

acid treatment, but no such treatment was done on these samples. The graphitic peak (G-
peak) at ~1580 cm'I is observed after all radiation doses. This is a characteristic of the
multi walled tubular structure of this material. Overall, the Raman spectra do not change

signiﬁcantly indicating that the MWCNTS held up very well in heavy ion radiation.

< -1.
5 10,000 Gy ‘11504.50m 1582.4cm“
1337cm \ \\
—1 ’ - ‘. \, 4.3, l‘.‘
‘ 1276”“ ., \ ‘,_1€50£3).30m'1
.‘:‘..‘;."..;';::.E;:;-.;2::;;:E;:'.‘S;‘.'.‘..'-...’-‘ ....... '

. 1.000 Gy 1535.7cm"
, 1336.8cm"
. 1261.9cm“

       
 
 
 
 
 
 

Counts (Arb. Units)

 

 

; Pre Rad
1338.1cm'1 ,~
- 1269.9cm’1 ’
1000 1200 1400 1600 1800 2000
Raman Shift (cm")

Figure 5.19. Raman spectra of MWCNTs before and after krypton-78 radiation. The D and G-peaks do
not change signiﬁcantly after each radiation dose compared to the un-irradiated sample. This indicates that
the MWCNTs held up very well in the krypton-78 radiation and retained their multi walled structure.
Spectra: B.W. Jacobs

SEM and HRTEM images shown in ﬁgure 5.20 also showed no observable
structural damage after krypton-78 irradiation. Figures 5.20(a) and 5.20(b) are SEM and
HRTEM images of MWCNTS before radiation. The nanotubes can be seen sticking out
of the amorphous carbon material. Figures 5.20(c) and 5.20(d) are SEM and HRTEM
images of MWCNTs after a 1000 Gy dose of krypton-78 radiation. Figures 5.20(c) and
5.20(t) are SEM and HRTEM images of MWCNTs after a 10,000 Gy dose of krypton-78
radiation. All images show pristine nanotube structures, but some amorphous carbon can

be seen on the surface of some nanotubes.

191

 

 

 

 

Figure 5.20. SEM and HRTEM images of multi walled CNTs before and after irradiation. (a) SEM before
irradiation. (b) HRTEM before irradiation. (c) SEM aﬁer l,OOOGy krypton-78 irradiation. (d) HRTEM
after l,OOOGy krypton-78 irradiation. (e) SEM after 10,000Gy krypton-78 irradiation. (f) HRTEM after
10,0000y krypton-78 irradiation. Images: B.W. Jacobs

The resiliency of the MWCNT was also a surprising result. All observed samples
retained their wall structure with no detectable damage. Results may indicate that the
heavy ion radiation possibly reduced the amount of amorphous carbon surrounding the
MWCNTS. This is immediately apparent in ﬁgure 5.20(f), where only a small amount of
amorphous carbon is seen on the outer walls of the pristine MWCNT. Like SWCNTS,
MWCNTS are excellent candidates for potential applications in high radiation

environments.
5.4.3.4 Carbon Onions

Carbon onions were irradiated with a calcium-48 heavy ion beam. Post

irradiation analysis using Raman spectroscopy and HRTEM indicated that the sample

192

 

sustained some observable damage. The Raman spectra of irradiated carbon onions,
shown in ﬁgure 5.21, indicated that the onions retained their cylindrical Shape and multi
wall structure. Comparing before and after irradiation Raman spectra of onions that did
not receive an oxidation treatment and onions that did receive an oxidation treatment, the

D and G-peaks remained similar after all radiation doses.

  

 

 

 

(a)- 70MeV/u (b) 70MeV/u ‘ . _,
~10.000Gy , . 1576 7cm I ~10.000Gy 511575.2cm

‘ ' . 1338.5cm" W 1' . '-
A 13411cm"./ A : f '. .
g . f 1506 5cm 9 ,1 .1; I. 1604.7cm‘
'E E . _ ’ ‘ ll \ ‘.
D _ . .............. D ........ ., I <
3 Pre Rad , . 1579 80",- '2 Pre-Rad A 1579 60m“
V ‘1 .1: '\ (I, v _ :1 '1 V. '1
.23 , 134519“ A/. 1 .2 1341.2cm‘I/ I}. ft
3 - I\\ I: 1614cm" g ‘ ‘/ ‘\‘\ 11" I? 1509 9cm

5' " 5‘ I ' I ‘

O ‘ / \\\/ < ...... I """"" O - .. // \4

_ J. 3.. ,.__:__________..-—»/ J <

 

 

 

 

 

1000 1200 1400 1600 1300 2000 1000 1200 1400 .1600 41800 2000
Raman Shift (cm") Raman Shlft (em )

Figure 5.21. Raman spectra of untreated and treated carbon onions before and after calcium-48 irradiation.
(a) Untreated carbon onions. (b) Treated carbon onions. Spectra: B.W. Jacobs

HRTEM investigations of irradiated onions coincided with the Raman results
indicating that the onions retained their overall spherical shape. Figure 5.22 shows
HRTEM images of irradiated onions that did not receive an oxidation treatment, ﬁgure
5.22(a) and 5.22(b), and onions that did receive an oxidation treatment, ﬁgure 5.22(c) and
5.22(d). The overall onion shape is preserved after heavy ion irradiation, with only minor
changes in onion structure. The arrows in ﬁgures 5.22(b) and 5.22(d) point out some

modiﬁcations that took the form of onion coalescence.

193

 

 

Figure 5.22. HRTEM images of carbon onions before and after calcium-48 irradiation. (a) and (b)
compares the before and after irradiation of the untreated carbon onions. (c) and (d) compares the before
and after irradiation of the carbon onions with oxidation treatment. The arrows in (b) and (d) denote
modiﬁcations due to radiation that took the form of onion coalescence. Images: B.W. Jacobs

Figure 5.23 shows additional carbon onions after heavy ion radiation and the
structural modiﬁcations that resulted for untreated, ﬁgure 5.23(a), and treated, ﬁgure
5.23(b), onions. Arrows in ﬁgure 5.23 denote these modiﬁcations. Faceting on the outer
carbon onion layers was frequently seen as well as onion coalescence where two carbon
onions began to form one larger onion. Carbon onion coalescence is shown in both insets

in ﬁgure 5.23.

194

 

 

Figure 5.23. HRTEM images of carbon onions after calcium-48 irradiation. (a) Oxidation treated onions.
The left most arrows point out outer layer discontinuities and the right points out onion coalescence. The
inset also shows onion coalescence. (b) Untreated carbon onions. The arrow points out onion coalescence.
The inset also shows onions coalescence. Both images indicate that that the overall onion structure was
maintained. Onion structure modiﬁcations are consistent for both treated and untreated samples. Images:
B.W. Jacobs

When comparing HRTEM images of carbon onions before radiation, onions that
underwent a high dose of radiation resembled onions grown at 2300°C, as shown in
ﬁgure 5.24(c). At this high growth temperature onion faceting and onion coalescence
occurred. Ribbons of graphite could also be seen in the post irradiated onions as well as
in the onions grown at 2300°C. Onions grown at lower temperatures are also shown in
ﬁgure 5.24(a) and 5.24(b) for comparison and to highlight the structural changes as a
result of growth temperature. This suggested that heavy ion radiation reduced the
disordered carbon component of the carbon onions, or the carbon onion structure changed

and annealed. or possibly a combination of both.

195

 

  

21m 21m ' ' mm

Figure 5.24. HRTEM images of untreated carbon onions grown at temperatures (a) l700°C (b) 2000°C

and (c) 2300°C show the evolution from spherical to polyhedral onions. The increased onion coalescence
and outer layer discontinuities can be seen in (c). Images: B.W. Jacobs

The nanomaterials performed very well under each radiation species and energies
at all doses. The carbon nanotube samples had no damage that could be observed with
Raman spectroscopy or HRTEM. The peapods found in one nanotube could be attributed
to heavy ion radiation, and is a new result, but warrants further investigation. The carbon
onions did have minor damage in the form of outer layer discontinuities and carbon onion
coalescence. This result was identiﬁed with HRTEM. Raman spectroscopy did not
indicate signiﬁcant peak alterations before and after radiation. Thermogravimetric.
analysis could also be used to further quantify the damage of carbon onions incurred
through radiation.

The radiation damage in GaN nanowires occurred through a short track
mechanism where highly crystalline material turned amorphous. This was identiﬁed with
HRTEM and F FTs of HRTEM images. Not all nanowires observed were damaged. The
GaNFETs tested also performed well in radiation with minimal deviation from normal
operation.

These results are signiﬁcant. Heavy ion radiation has been a serious problem in

space-based systems. These materials represent a great leap forward toward the solution

196

of this problem. Shielding in space-based applications could be minimized, and the need
for redundant systems mitigated. Reliable electronic and mechanical operation of space-
based systems would be greatly increased. These materials could revolutionize space

applications as they have already revolutionized many ﬁelds of science and engineering.

J97

 

References

l.

10.

11.

K. LaBel “Single Event Effect Criticality Analysis,” conference of single event effect
criticality analysis, http://radhome. gsfc. nasa. g0v/radhome/papers/seecai. htm.

EIA/JEDEC Standards No. 57, “Test procedures for the measurement of single event
effects in semiconductor devices from heavy ion irradiation,” Electronic Industries
Association, 1996.

J. Barth, “Ionizing Radiation Environment Concerns,” conference of single event
effect criticality analysis, http://radhome. gsfc.nasa.gov/radhome/papers/seeca3.htm.

D. V. Reames, “Solar Energetic Particles: Sampling Coronal Abundances.” Space
Sci. Rev., 85, 327-340 (1998)

J. Cronin, T. K. Gaisser, S. P. Swordy, “Cosmic rays at the energy frontier,” Sci.
A mer., 276, 44-49 (1997)

H. J. Barnaby, S. K. Smith, R. D. Schrimpf, D. M. Fleetwood, R. L. Pease,
“Analytical model for proton irradiation effects in bipolar devices,” IEEE T. Nucl.
Sci. , 49, 2650-2649 (2002)

P. Lenahan, J. F. Conley, Jr., “A comprehensive physically based predictive model
for radiation damage in MOS systems,” Appl. Phys. Lett., 71, 2413-2423 (1998); A.
Y. Yang, P. M. Lenahan, J. F. Conley, Jr., “The radiation response of the high
dielectric-constant hafnium oxide/silicon systems, IEEE T. Nucl. Sci., 49, 2844-2851
(2002)

T. P. Ma, P. V. Dressendorfer, Ionizing Radiation Effects in MOS Devices & Circuits,
chap. 1, John Wiley and Sons, Toronto, 1989

L. W. Massengill, B. K. Choi, D. M. Fleetwood, R. D. Schrimpf, K. F. Galloway, M.
R. Shaneyfelt, T. L. Meisenheimer, P. E. Dodd, J. R. Schwank, Y. M. Lee, R. S.

Johnson, G. Lucovsky, “Heavy-ion-induced breakdown in ultra-thin gate oxides and
high-k dielectrics,” IEEE T. Nucl. Sci., 48, 1904-1912 (2001)

E. Simoen, A. Mercha, A. Morata, K. Hayama, G. Richardson, J. M. Raﬁ, E.
Augendre, C. Claeys, A. Mohammadzadeh. H. Ohyama, A. Romano-Rodriguez,
“Short-channel radiation effect in 60 MeV proton irradiated 0.13 /spl mu/m CMOS
transistors,” IEEE T. Nucl. Sci., 50, 2426-2432 (2003)

K. Kobayashi, H. Ohyama, M. Yoneoka, K. Hayama. M. Nakabayashi, E. Simoen, C.

Claeys, Y. Takami, H. Takizawa, S. Kohiki, “Radiation damage of N-MOSFETS
fabricated in a BiCMOS process,” J. Mater. Sci: Mater. Elect, 12, 227-230 (200])

198

 

 

 

12.

l3.

14.

15.

16.

17.

18.

19.

20.

21

22.

23.

24.

25.

B. Djezzar, A. Smatti, A. Amouche, M. Kechouane, “Channel—Length Impact on
Radiation-Induced Threshold-Voltage shift in N-MOSF ET’ 8 Devices at Low Gamma
Rays Radiation Doses,” IEEE T. Nucl. Sci., 47, 1872-1878 (2000)

B. Streetman, S. Banerjee, Solid State Electronic Devices. Fifth Edition, Prentice
Hall, Upper Saddle River, NJ, 2000.

http://www. nscl. msu. edu/tech/index. html

http://www. nscl. msu. edu/tech/devices/index. html

R. Ladbury, R. A. Reed, Member, P. Marshall, K. A. LaBel, R. Anantaraman, R. Fox.
D. P. Sanderson, A. Stolz, J. Yurkon, A. F. Zeller, J. W. Stetson, “Performance of the
High-Energy Single-Event Effects Test Facility (SEETF) at Michigan State
University’s National Superconducting Cyclotron Laboratory (N SCL),” IEEE T.
Nucl. Sci., 51, 3664-3668 (2004)

N. Anantaraman, “NSCL Single Event Effects Test Facility Users Manual,” National
Superconducting Cyclotron Lab., Michigan State Univ., East Lansing, MI.

http://www. laplink. com

R. Leon, S. Marcinkevicius, J. Siegert, B. Cechavicius, B. Magness, W. Taylor and C.
Lobo, “Effects of proton irradiation on luminescence emission and carrier dynamics of
self-assembled III-V Quantum dots,” IEEE T. Nucl. Sci., 49, 2844-2851 (2002)

A. Ionascut-Nedelcescu, C. Carlone, A. Houdaycr, H. J. von Bardeleben, J .-L. Cantin,
S. Raymond, “Radiation Hardness of Gallium Nitride,” IEEE T. Nucl. Sci., 49, 2733-
2738(2002)

. S. Iijima, “Helical microtubules of graphitic carbon,” Nature, 354, 56-58 (1991)

R. H. Baughman, A. A. Zakhidov, W. A. de Heer, “Carbon Nanotubes - the Route
Toward Applications,” Science, 297, 787-792 (2002)

R. Saito, G. Dresselhaus, M. S. Dresselhaus, Physical Properties of Carbon
Nanotubes, London, Imperial College Press, 1998.

R. Andrews, D. Jacques, A. M. Rao, T. Rantell, F. Derbyshire, Y. Chen, J. Chen, and
R. C. Haddon, “Nanotube composite carbon ﬁbers,” Appl. Phys. Lett, 75, 1329-1331
(1999)

Z. Chen, J. Appenzeller, Y-M. Lin, J. Sippel-Oakley, A. G. Rinzler, J. Tang, S. J.

Wind, P. M. Solomon, P. Avouris, “An Integrated Logic Circuit Assembled on a
Single Carbon Nanotube,” Science, 311, 1735 (2006)

199

 

 

 

26

27.

29.

30.

31.

33.

34.

35.

36.

37.

38.

S. Li, Z. Yu, S-F. Yen, W. C. Tang, P. J. Burke, “Carbon Nanotube Transistor
Operation at 2.6 GHz,” Nano Lett., 4, 753-756 (2004)

P. W. Barone, S. Baik, D. A. Heller, M. S. Strano, “Near-infrared optical sensors
based on single-walled carbon nanotubes,” Nat. Mater., 4, 86-92 (2005)

. E. S. Snow, F. K. Perkins, E. J. Houser, S. C. Badescu, T. L. Reinecke, “Chemical

Detection with a Single-Walled Carbon Nanotube Capacitor,” Science, 307, 1942-
1945(2005)

P. R. Bandaru, C. Daraio, S. Jin, A. M. Rao, “Novel electrical switching behaviour
and logic in carbon nanotube Y-junctions,” Nat. Mater., 4, 663-666 (2005)

S-H. Hur, M-H. Yoon, A. Gaur, M. Shim, A. Facchetti, T. J. Marks, J. A. Rogers,
“Organic Nanodielectrics for Low Voltage Carbon Nanotube Thin Film Transistors
and Complementary Logic Gates,” J. Amer. Chem. Soc., 127, 13808-13809 (2005)

P. Avouris, “Supertubes [carbon nanotubes],” IEEE Spect, 41, 40-45 (2004)

. A.V. Krasheninnikov and K. Nordlund, “Irradiation effects in carbon nanotubes,”

Nucl. Instrum. Meth. B, 216, 355-366 (2004)

A.V. Krasheninnikov, K. Nordlund, and J. Keinonen, “Production of Defects in
Supported Carbon Nanotubes under Ion Irradiation,” Phys. Rev. B. , 65, 165423-1,
(2002)

F. Banhart, “Irradiation effects in carbon nanostructures,” Rep. Prog. Phys, 62, 1 181-
1221(1999)

http://www. cnanotech. com

M. J. Bronikowski, P. A. Willis, D. T. Colbert, K. A. Smith, R. E. Smalley, “Gas-
phase production of carbon single-walled nanotubes from carbon monoxide via the
HiPco process: A parametric study,” J. Vac. Sci. Technol, A, 19, 1800-1805 (2001)

J. M. Benavides, “Method for manufacturing high quality carbon nanotubes,” US
Patent 7,008,605 (2006)

A. Hirata, M. I garashi, T. Kaito, “Study on solid lubricant properties of carbon onions
produced by heat treatment of diamond clusters or particles,” Tribo. Inter. , 37, 899-
905 (2004)

200

 

 

39. K. W. Street, M. Marchetti, R. L. Vander Wal, A.J. Tomasek, “Evaluation of the
tribological behavior of nano-onions in Krytox 143AB,” Tribal. Lett., 16, 143-149
(2004)

40. A. Hirata, R. Yamada, B. W. Jacobs, A. D. Baczewski, V. M. Ayres, “Correlation of
Fundamental Properties with Tribological Performance of Carbon Onions with
Oxidation Treatment,” accepted for publication in Japanese Journal of Applied
Physics

41. J. F. Ziegler, “SRIM-2003,” Nucl. Instrum. Methods Phys. Res. Sect. B, 219-220.
1027-1036 (2004)

 

 

201

6. Future Work

The research presented in this thesis represents an ongoing research effort with
long-term goals. There are still many possible research avenues that can be taken in
order to understand the properties of the multiphase GaN nanowires and fundamental
heavy ion interactions with nanomaterials. Further experimental investigations using the
discussed techniques and largely unexplored techniques, discussed below, may uncover
more interesting properties of the multiphase nanowires and heavy ion-nanomaterial

interactions.

6.]. Gallium Nitride Nanowires

Cross-section HRTEM analysis will be essential in further investigations of the
structure and growth of the multiphase as well as the single-phase nanowires.
Observations of nanowires in the SEM and TEM have shown a variety of different
features that are unique to each individual nanowire. The nanowire may have different
features based on the nucleation site characteristics. Since the multiphase characteristic
is an important feature for many possible future applications, the ability to control how
zinc-blende and wurtzite form, would have considerable implications. Cross-sections of
the nanowires grown at different conditions would also be beneﬁcial to understand how
these conditions affect nanowire growth. Additional cross-sections of nanowires grown
along the [0001] direction could help determine the nature of the anomalous features in
the center of these nanowires, including the holes and dark and light contrast areas

discussed in chapter 4. The origins of the light and dark contrast areas may shed light on

202

 

 

how these nanowires nucleate and grow. If holes do occur along the entire length of the
nanowire. the application possibilities would be intriguing.

The investigations with the Zyvex Nanomanipulator system only scratched the
surface of what can be done with this instrument. Thorough electronic transport studies
of individual nanowires could be carried out to determine if single-phase transport is
reproducible. The transport characteristics of the nanowires that grew along the [000]]
direction should also be investigated because the transport characteristics might be
different due to the different growth direction. Single-phase wurtzite nanowires that
grew along the [2T T0] direction could be studied to see how its transport differs from
the single-phase nanowires that grew along the [0001] direction and the multiphase
nanowires. The probes could be used to study the mechanical properties of the different
nanowires.

The electronic structure of these nanowires are very intriguing, and as yet have
not been investigated in detail. The large band gap shift measured using CL showed that
these nanowires can absorb and emit light in the UV range. Further investigations with
CL could be done to ascertain whether the band gap shift varies from nanowire to
nanowire or if it persists in all multiphase and single-phase nanowires.
Photoluminescence could also uncover a wide variety of properties. With the ability to
vary the wavelength of light, PL could pick out subtle mid-gap luminescence
mechanisms in the nanowire. These luminescence mechanisms could be traced to the
interface states between domains. PL could also further assess the feasibility of

multiphase nanowires in optical devices.

203

 

 

X-ray diffraction (XRD) could be used to investigate the structure of the nanowire
and matrix material. It is in many ways similar to electron diffraction in the TEM.
Instead of using electrons, an XRD utilizes the wave nature of photons to interact with a
crystal structure. The photons are diffracted from planes in the crystal and strong
reflections result when the Bragg condition is satisﬁed. An XRD system has some
advantages compared to using a TEM for diffraction, in that the size of the sample is not
as restricted and X-rays are not nearly as destructive as a high-energy electron beam.

An XRD system creates and displays diffraction information somewhat
differently than the TEM. Figure 6.1 shows a typical XRD system. On the right the X-
ray source emits photons that are directed at the sample, in the center. A detector on the
left side collects the diffracted X-rays. The detector is swept out in a circular path while
the source remains stationary. As the detector sweeps out it detects diffracted beams that
satisfy the Bragg condition and records the associated angle at which the reflections
occur. This information is then graphed on a computer. In the generated graph the x-axis
is the swept angle and the y-axis is the measured intensity. Certain crystals will diffract
more strongly at certain angles resulting in strong peaks at those angles. The angle is
based on the interplanar spacing of a given crystal material and the wavelength of the X-
ray, just as in electron optics in the TEM. A XRD system can measure the interplanar
spacing of a given crystal sample to a high degree of accuracy. Due to this high-
resolution many crystal system are measured for possible strain in the material. By
measuring the shift in the angles of diffraction peaks the amount of strain present in a

material can be quantiﬁed.

204

X-Ray Detectqg.

i z I l H. Li A
DetectorsafrjlzleSource . My 1 l I M .

I "I. he .. Incident 've

m ,. 5 " Reflected Wave ~" .1— . W?

W

A

    

Figure 6.1. Bruker AXS high resolution XRD at the W. M. Keck Microfabrication Facility at Michigan
State University. (a) The X—ray beam originates from the right. It hits the sample and is detected on the
left. The detector sweeps in a counterclockwise direction and records the reflected beam. (b) The X-ray
detector. A reflected beam from the sample is detected. (c) The X-ray source. X-rays travel from the
source to the sample. Images: B.W. Jacobs

The XRD could further quantify the amount of strain present in the multiphase
nanowires. CL detected a large amount of strain in the nanowires through the measure of
a large blue shift in the band gap of the nanowires. XRD could provide a more accurate
measure of this strain.

Further investigations of the GaNFET characteristics, including the behavior of
the metal-nanowire junction, would be very beneﬁcial for nanowire circuit design. This
junction is not well understood, and the nature of the Schottky barrier that forms there has
not been well deﬁned. Using a STM, a special capability called scanning probe
recognition microscopy and ballistic electron emission microscopy. the Schottky barrier
of a metal-nanowire junction can be directly measured. These techniques are discussed
below.

Scanning Probe Recognition Microscopy is a new scanning probe capability
developed by the Ayres’ research group in partnership with Veeco Instruments”. A
standard Veeco Instruments Nanoscope IIIa scanning probe microscope system is equipped

with a special modiﬁcation developed by the Ayres’ research group by which the SPM

205

system itself is given the ability to auto-focus on regions of interest through incorporation
of recognition—based tip control. This is described as Scanning Probe Recognition
Microscopy (SPRM). The recognition capability is realized using algorithms and
techniques from computer vision, pattern recognition and signal processing ﬁelds.
Adaptive learning and prediction are also implemented to make detection and recognition
procedures quicker and more reliable. The integration of recognition makes the SPRM

system more powerful and ﬂexible in investigating speciﬁc properties of samples.

Ballistic electron emission microscopy is a promising technique that is used with
a ST M3 ‘4. When determining the Schottky barrier between a metal and semiconductor. a
thin layer of metal is deposited on top of a clean semiconductor surface. The sample is
placed in the STM metal side up, i.e. the metal side interacts directly with the tip.
Connections to the metal, called the base, and to the semiconductor, called the collector.
are made. One ammeter is connected to the collector and another to the base, see ﬁgure
6.2. The STM tip is then brought down to the surface and normal STM imaging
commences. Electrons tunnel through the barrier between the tip and sample, and when
the emitted electrons hit the metal surface they can either pass through the metal without
scattering, or scatter and lose energy and pass through the attached ammeter. The
electrons can pass through the metal without scattering because the metal is thinner than
the mean free path of the electron, which is the average distance an electron travels
through a material before scattering and losing a signiﬁcant portion of its energy. If the
un-scattered electron has enough energy to overcome the metal-semiconductor Schottky

barrier, eVb, the electron can enter the semiconductor and will eventually pass through

206

 

 

the attached ammeter. The Schottky barrier is determined by the measured currents, I,

and 1c.

 

 

 

Figure 6.2. Diagram of BEEM operation. The left image shows the physical setup ofa BEEM. The tip
interacts with the base metal surface passing electrons through gap between the tip and surface. IC is the
current coming out of the semiconductor collector, I. is the current from the base metal resulting from
electrons that were scattered in the metal and did not have enough energy to surmount the barrier, plus the
collector current. The left image shows the energy diagram indicating each barrier. Electrons originate in
the metal STM tip, tunnel through the vacuum gap and pass through the metal and into the semiconductor
collector. E.- is the Fermi level of each material. and eV., is the height of the metal-semiconductor Schottky
barrier. The ﬁgure is not to scale. The vacuum gap is usually on the order ofa few nanometers, and the
metal base thickness is usually around 10nm. Image: B.W. Jacobs

Used in conjunction with SPRM the STM will be able to automatically track the
nanowire/metal interface creating a spatial map as well as an energy spectrum of the
Schottky barrier present at the interface. Investigations on the nanowire-metal junctions
could be studied in different designs, and different contact metals could be deposited to
see how the Schottky barrier and carrier injection differs.

Using STM to image the electronic surface states of the nanowire would also help
understand the energy landscape of the nanowire. Surface investigations as well as cross-
section STM investigations would shed light on the electronic states of the nanowire
surface, which is important information for understanding the nature of nanowire-metal

junctions. STM of the cross-section would help uncover electronic properties of the

207

interface and provide information on the electronic barrier characteristics of the interface.
The SPRM modality of the STM would also prove highly effective in these
investigations. because it would be able to track an interesting feature, such as an
interface between a zinc-blende and wurtzite phase, at will, mapping out the energy
landscape at that interface along the way. These experiments would further assess the

viability of the multiphase GaN nanowires used in real world devices. I.

 

 

208

6.2 Fundamentals of Radiation Interactions in Reduced Dimensionality Systems

The radiation resilience of nanomaterials and nanocircuits is a very promising
area of research with high probability of success. The ideas and results presented in this
thesis were critical in getting this project off the ground, but more experimental as well as
theoretical testing can still be explored with different devices, materials and radiation
beams.

First, further heavy ion radiation investigations of the nanomaterials already
tested must be done. An in depth analysis of low to high Z heavy ions with multiple
charge-to-mass-ratios should provide further details of nanomaterial-heavy ion
interactions. A future experiment using oxygen-16 is currently being planned. Sample
density, weight and area must be highly precise so that dose calculations are accurate.
Standard methods and calculations for irradiating nanomaterials should be developed. A
reproducible procedure must be followed so accurate results are continually produced.
Since the nanomaterials were not signiﬁcantly damaged, higher radiation doses should be
administered. Without signiﬁcant damage to the materials, the mechanism in which they
interact with the heavy ion beam cannot be described.

X-ray Photoelectron Spectroscopy (XPS) could also be utilized as an additional
carbon structure characterization technique for before and after irradiation analysis. XPS
is a surface characterization technique used to determine the elemental composition,
chemical or electronic state of a given material. In an XPS system X-rays are at high
enough energy to interact with core electrons in sample atoms. These core electrons in
the sample can gain enough energy through the absorption of the incident X-rays to

overcome the work function of the material, see ﬁgure 6.3. The removal of an electron

209

- can.“

 

_.1

 

from a material through the absorption of a photon is called the photoelectric effect. Core
electron energies are highly speciﬁc to each element in the periodic table. Therefore,
measuring the energy of the emitted electrons, called photoelectrons or photo-emitted
electrons, speciﬁc elements and their bonds can be detected and characterized in a given
material. The beam of incident X-rays penetrate into the material up to several microns,
however only surface electrons are usually emitted, generally up to 10 nm in depth, due
to recombination within the material. Therefore, this technique is only used to analyze

surfaces.

Electron Energy
Photo-Emitted Electrons (<1.5 kV) Analyzer (0-15 kV)

Top 70—110A of the sample surface

Electron

, Electron Detector
Collection § \ 7
Lens M, - *

  
 
  
  

  

Focused Beam of

X-rays (1.5 kV)
Electron
Takeoff
Angle

~ ~ ~ swam-am;
Samples usually solid XPS Si (2p) XPS
requires UHV (>10a torr) signals from sample

Figure 6.3. Basic method of XPS sample spectrum acquisition. The X-rays are aimed at the sample
surface at a known angle and interact with the surface inducing photo-emitted electrons. The photo-emitted
electrons are collected and analyzed to determine their energy. A spectrum of the photo-emitted electrons
is then created indicating what materials are on the sample surface. and the electronic state they are in.

An XPS system uses one of two X-ray sources: a monochromatic aluminum

source or a non-monochromatic magnesium source, both of which are often integrated

210

into the same system. The Al source emits X-rays with energy of 1486.7 eV, and
because the energy spread of X-rays emitted from this source is quite small, 0.4-0.6 eV, it
is used for high-resolution XPS. The intensity is rather small compared to the Mg non-
monochromatic source, so the time required to gather a spectrum with an acceptable
signal-to-noise ratio is longer. The Mg source emits X-rays with energy of 1253 eV, and
has a larger energy spread of around 1 eV. Therefore, it has a lower resolution but more
intense source. Since the photoelectron detector is far from the sample surface, the
sample chamber must be under ultra-high vacuum, 10'8-10'IO Torr. This allows the
photoelectrons to travel to the detector without losing much energy. After the electrons
have been detected they are sent to a computer where an energy spectrum is created.

XPS is a powerful technique that can provide very beneﬁcial information about
sample surfaces. Acquiring spectra is usually very quick, only a few minutes, after the
lengthy sample preparation and pump down procedures have been completed. Some
samples. however, can take several hours if the monochromatic Al source is used.

In this thesis research, the XPS system was used to determine carbon onion
properties. XPS can distinguish between the sp2 and sp3 hybridizations and determine if
oxygen exists in the carbon system. When taking a XPS spectrum of the carbon onions,
sample preparation is of great importance. Many times samples will be placed on a metal
such as copper with an adhesive layer on one side. This adhesive layer contains carbon
and oxygen and therefore interferes with the spectrum. An adhesive that does not contain
carbon and oxygen should be used, or if an alternate powder sample mounting technique

could be developed, it would create more reliable XPS spectra.

21]

 

 

GaNFET operation in heavy ion radiation must also be further explored. Many
circuits must be tested so that the failure mechanisms of the GaNFETs as a result of
irradiation can be identiﬁed and improved upon. The optical properties of the GaN
nanowires are thought to be conducive to operation in UV detectors and LEDs.
Therefore, real time operation in the heavy ion beam testing the optical properties of GaN

nanowires might also yield promising results.

6.2.] Other Candidate Materials for High Radiation Environments
6.2.1.1 Quantum Dots

Quantum dots represent a relatively new device concept. A quantum dot is a
zero-dimensional system where electrons in the system are conﬁned in every direction.
Quantum dots are primarily used for optical applications since they are very efﬁcient due
to their high quantum yields. This allows for bright light emitting devices. which do not
require much electrical power.

Ion irradiation experiments on quantum dots have shown the possibility of an
inherently reduced dimensionality radiation resilient system". When the optical
performance of a quantum well (two-dimensional) structure was compared to a quantum
dot structure in phosphorus radiation, the quantum dot substantially outperformed the
quantum well in optical emission7. These results are very promising for quantum dot
performance in radiation. This result for the zero-dimensional performance reinforces the
results obtained with the one-dimensional systems described in this research. Reduced

dimensionalities do seem to play a role in radiation resilience.

212

6.2.1.2 Semiconducting Nanowires of Various Materials

There have been many reports of successful growth of a variety of different
semiconducting nanowires. These nanowires, like their bulk material counterparts,
possess a variety of different band gaps, electron and hole mobilities and mechanical
properties making them highly versatile. This versatility could be used for radiation
resilient devices. GaN nanowires have shown a high resilience in radiation either due to
its material properties, its small size, or both. Since radiation interactions in
nanomaterials may fundamentally change as discussed in chapter 5, this opens up a wide
variety of application possibilities that exploit the versatility of semiconducting

nanowires.

6.2.2 Other Candidate Devices for High Radiation Environments
6.2.2.1_Carbon N anotube—Based Field Effect Transistors

Carbon nanotubes showed very high heavy ion radiation resilience in all
experiments discussed in chapter 5. The results are promising in that carbon nanotubes
can be used for a variety of electronic, optical and chemical devices. It is probable that
these devices that integrate carbon nanotubes as the primary element will be radiation
resilient. To date, however, no carbon nanotube—based device has been tested in this
research effort. Therefore, nanotube devices must be tested in heavy ion radiation, as

they may outperform GaNFETs already tested.

213

 

6.2.2.2 High Electron Mobility Transistors

High electron mobility transistors have also been fabricated8 using a two-
dimensional GaN thin ﬁlm as a quantum well conduction channel, and are being used in
high power, high frequency microwave communications applications. They operate via
source, drain and gate voltage modulation with a gate modulated conduction channel
consisting of a thin intrinsic semiconductor layer sandwiched between two insulating
layers. In a GaN/AlGaN HEMT a heavily doped AlGaN layer, connected to a source and
drain voltage. is the electron source. Since AlGaN has a larger band gap than GaN, when
electrons enter the GaN conduction channel they will be trapped at the interface. They
are drawn to the interface because the conduction band energy is reduced due to the band
bending resulting from Fermi level alignment, and a triangular quantum well forms as a
result. Although there are only a few allowed energy levels in the quantum well,
electrons are able to move with very high motilities due to the reduced number of
scattering centers, which are normally present in most devices due to dopants. With no
dopants acting as scattering centers, the electrons scatter less and very high levels of
current can be achieved. The HEMT can also operate at very high frequencies.

This conduction channel represents a two-dimensional system with a constricted
z-direction of only ~10nm. Electrons are conﬁned in the z-direction so they can only
have certain quantized values on energy, but are free to move in both the x- and y-
directions. With this conﬁguration, HEMTS and their reduced dimensionality may be
more resilient in radiation.

This ongoing project that assesses the viability of reduced dimensionality systems

for heavy ion radiation environments will need further investigation. This thesis has laid

214

the groundwork for future studies of these materials and systems. The materials and
methods discussed in this chapter may uncover applications and properties that were

previously unknown, taking these materials and systems a step closer to real world

apphcaﬁons.

215

 

References

to

. http://wmr. Its/[gov/awardsearch/shoWA ward. do ?A wardN umber = 04 002 98

Y. Fan, Q. Chen. V. M. Ayres, A. D. Baczewski, L. Udpa S. Kumar, “Scanning Probe
Recognition Microscopy Investigation of Tissue Scaffold Properties”, in press Int. .1.
Nanomed., 2, (2007)

V. Narayanamurti, M. Kozhevnikov, “BEEM imaging and spectroscopy of buried
structures in semiconductors,” Phys. Rep, 349, 447-514 (2001)

R. E. Martinez 11, I. Appelbaum, C. V. Reddy, R. Sheth, K. J. Russell, V.
Narayanamurti, J-H. Ryou, U. Chowdhury, R. D. Dupuis, “Electron transport through
strongly coupled AlInP/GalnP superlattices,” Appl. Phys. Lett. 81, 3576-3578 (2002)

l. Gryczynski, J. Malicka, W. Jiang, H. Fischer, W. C. W. Chan, Z. Gryczynski, W.
Grudzinski, J. R. Lakowicz, “Surface-Plasmon-Coupled Emission of Quantum Dots,”
J. Phys. Chem, 109, 1088-1093 (2005)

S. Marcinkevicius, J. Siegert, R. Leon, B. Cechavicius, B. Magness, W. Taylor, C.
Lobo, “Changes in luminescence intensities and carrier dynamics induced by proton
irradiation in lnxGal-xAs/GaAs quantum dots,” Phys. Rev. B, 66, 235314 (2002)

P. G. Piva, R. D. Goldberg, l. V. Mitchell, D. Labrie, R. Leon, S. Charbonneau, Z. R.
Wasilewski, S. Fafard, “Enhanced degradation resistance of quantum dot lasers to
radiation damage,” Appl. Phys. Lett., 77, 624-626 (2000).

H. Xing, S. Keller, Y-F. Wu, L. McCarthy, 1. P. Smorchkova, D. Buttari, R. Cofﬁe,
D. S. Green, G. Parish, S. Heikman, L. Shen, N. Zhang, J. J. Xu, B. P. Keller, S. P.
DenBaars, U. K. Mishra, “Gallium nitride based transistors,” J. of Phys. .' Cond. Matt,
13. 7139-7157 (2001)

216

7. Conclusion

A signiﬁcant discovery was made during the exploration of nanomaterials for use
in space applications. GaN nanowires were thought to possess high radiation resilience
due to their bond strength as well as their small size. In analyzing the nanowire with
HRTEM, a multiphase structure was uncovered with longitudinal crystal domains that
extended the entire nanowire length.

The matrix on which the nanowires grew was analyzed with SEM and TEM.
Nanowire growth from the sides of GaN hexagonal platelets that formed on the matrix
was observed. At different temperatures different matrix formations led to nanowires
with different growth directions and morphologies. Nanowires that grew at lower
temperatures always exhibited a zinc-blende/wurtzite multiphase structure with growth
along the <01. l>/<2T TO> direction. HRTEM analysis linked nanoscale platelets. which
formed on the hexagonal platelet sides in the same orientation and coherent to the
platelets, to the multiphase nanowire nucleation. However, when the growth temperature
was increased, two additional types of nanowires were observed. Single-phase wurtzite
nanowires were identiﬁed with growth in the same direction as the wurtzite phase in the
multiphase nanowires, <2T T0>. Nanoscale nucleation sites were linked this single-
phase nanowire growth. Single-phase wurtzite nanowires were identiﬁed with growth
along the [0001] direction. Hexagonal platelet tops were linked to nucleation sites for
this single-phase nanowire.

Cross-section HRTEM analysis revealed how the multiphase structure formed and
the orientations of the longitudinal crystalline domains. Several coherent and incoherent

interfaces formed between crystalline domains. A coherent interface was seen in all

2l7

 

multiphase nanowire cross-sections analyzed and explained the coherent interface seen in
the plain-view HRTEM analysis. This coherent interface could provide a stabilization
mechanism that renders the multiphase nanowire energetically attractive. The (0001)
face was always observed to terminate on an internal interface, possibly due to its
instability. In the multiphase nanowire system, the instability of the (0001) planes may
be compensated through the incorporation of the zinc-blende phase as a second
stabilizing aspect of the multiphase conﬁguration. The growth directions identiﬁed in
plain-view HRTEM analysis corresponded to the on axis F F Ts from each crystalline
domain found in the cross-sections. Cross—section analysis revealed holes in the center of
some single-phase nanowires.

The electronic properties of a multiphase nanowire were tested using a
nanomanipulator. Two and four-point probe investigations were carried out to determine
the resistance of an individual nanowire with non-invasive probes, and to study coupling
differences between nanoscale probes and macroscopic metal contacts. Differences in
coupling were linked to macroscopic vs. microscopic contacts to the nanowire, where
microscopic probes that had a tip diameter on the order of the nanowire width. injected
carriers more effectively.

Using two-point probe techniques the nanowire was end-contacted and
breakdown measurements were carried out. As a result of passing a large current through
the nanowire, a round agglomeration of material formed at nanowire end where the probe
was contacted and began to grow. This material looked as though it came from the
internal structure of the nanowire, while an external structure remained intact. Probing

the agglomeration showed that it was liquid. This was evidence of possible phase-

218

specific transport, and has signiﬁcant implications on the question of how and where
electron transport actually occurs within the multiphase nanowires.

GaN nanowire-based transistors were fabricated using electron beam lithography.
I-V measurements of these devices showed that multiphase nanowires were viable in
devices. Gate modulation of the current was achieved in measurements. l-V
characteristics indicated Schottky barrier behavior. Two-point probe and l-V
measurements of a GaN device showed the ability of the nanowire to may very high
currents. important for high power applications.

The radiation resilience of nanomaterials and devices were investigated using
multiple heavy ion radiation beams including krypton-86, krypton-78, and calcium-48 at
the NSCL at Michigan State University. The nanomaterials tested include single and
multi wall carbon nanotubes, GaN nanowires, carbon onions and GaN nanowire-based
FETs. Each nanomaterial performed very well after each radiation dose. The carbon
nanotubes did not exhibit any observable damage as a result of heavy ion radiation.
Carbon onions had some structural modiﬁcations as a result of heavy ion radiation
including increased outer layer discontinuities and multiple onion coalescence. GaN
nanowires showed damage in the form of a short track crystalline to amorphous
transition. All materials performed above expectation and are very promising candidates
for future space—based systems.

The results presented in this thesis mark signiﬁcant steps forward in
understanding GaN nanowire multiphase growth and applications, and fundamental
radiation interactions with nanomaterials and devices that incorporate a reduced

dimensionality system. This thesis research also laid the groundwork for further study of

219

the multiphase nanowire system and radiation experiments, and suggested future
experiments and analysis methods. The nanomaterials and devices that incorporate them
have great promise for high performance devices and systems as well as extraordinary

potential for future space as well as terrestrial applications.

220

   

lllilliligtlitglﬂl1111111l