multi-stimuli effects on thin films and devices
Post on 07-Nov-2021
1 Views
Preview:
TRANSCRIPT
The Pennsylvania State University
The Graduate School
MULTI-STIMULI EFFECTS ON THIN FILMS AND
DEVICES
A Dissertation in
Mechanical Engineering
by
Md Zahabul Islam
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Doctor of Philosophy
August 2020
The dissertation of Md Zahabul Islam was reviewed and approved by the following:
Aman Haque
Professor
Department of Mechanical Engineering
Dissertation Advisor & Chair of Committee
Adri van Duin
Kenneth Kuan-Yun Kuo Early Career Professor
Professor of Mechanical Engineering
Professor of Chemical Engineering
Professor of Engineering Science and Mechanics
Director of the Materials Computation Center
Daudi Waryoba
Assistant Professor, Engineering
Coordinator, Applied Materials Option
Penn State DuBois
Douglas E. Wolfe
Professor
Department of Materials Science and Engineering
Department of Engineering Science and Mechanics
Department Head, Advanced Coatings at the Applied Research Laboratory
Karen A. Thole
Professor
Department of Mechanical Engineering
Head of the Department of Mechanical Engineering
iii
Abstract
Physical properties of materials are known to depend on microstructure and defects across
multiple length-scales. The structure-property scaling becomes enhanced at the micro and nano
scales, an example of which is the ‘smaller is stronger’ phenomenon. Size effects also render
nanoscale materials more sensitive to external stimuli such as stress, temperature, electrical
current, light compared to their bulk counterparts. Even more interesting is the observation of the
breakdown of classical physical laws at length scales (grain size, thickness) at or below
characteristic length scales for physical domains. For example, scaling of yield stress (known as
the Hall-Petch law) breaks down below ~25 nm, where a grain cannot accommodate statistically
significant number of dislocations to induce plasticity. Similar breakdown phenomena have been
observed for other (electrical, thermal) domains. Fundamentals of the mechanics and physics of
nanomaterials is a prerequisite for the development of nanotechnology, which makes the length
scale and external stimuli effect on materials behavior as an attractive field of research.
While extensive efforts are ongoing to explore nanoscale structure-properties relationship in
single domains, this dissertation is rooted in processing-structure aspects of materials by
exploiting pronounced coupling exists among physical domains. Since the core of materials
processing relates to the response of the material to external stimuli (such as temperature), our
approach is to explore size or confinement effects that could make materials more sensitive to
external stimuli compared to conventional bulk materials. This lays down our hypothesis, ‘size-
induced coupling of multiple domains is manifested in form of unprecedented synergy of
multiple stimuli, which can be exploited to tailor microstructure or defect density to achieve
control over physical properties’. This hypothesis is aligned to the ulterior goal of this
dissertation, which is to develop novel materials processing techniques that are faster, more
iv
effective and energy efficient compared to the conventional (high temperature) thermal
annealing. Accordingly, the objective of this dissertation is to validate the hypothesis and
demonstrate it on two classes of materials spanning nano to micro scales namely, (i) ~2 nm thick
two-dimensional (2D) and (ii) 100 nm to 100 micron thick metals and additive manufactured
alloys. For each of these cases, we have explored the multi-stimuli synergy to achieve control
over crystallinity, grain size and defect density.
Since nanoscale characterization is challenging even in single domains, a key hurdle for this
research is to develop an experimental setup, which can simultaneously apply multiple stimuli,
or conversely, characterize the materials in multiple domains. We achieved this with a Micro
Electro-Mechanical System (MEMS) based framework, where strain, temperature and electrical
current are simultaneously applied on the specimen inside a high-resolution microscope that
visualizes the microstructural changes in real time. The setup is small enough to fit inside a
transmission electron microscope (TEM), which can provide atomic resolution. We have
compared the magnitude of these stimuli for both cases (i) when they are applied simultaneously
and (ii) separately to quantify the synergistic effect of the stimuli. In addition, we have also
quantified the time rate or the dynamics of microstructural evolution. We have also analyzed the
stimuli magnitude and microstructural dynamics to demonstrate the efficiency of our proposed
multi-stimuli materials processing technique. Our findings shows the enhanced atomic and defect
mobility due to the electrical current. It also reveals the microstructural transformation of near-
amorphous material to nanocrystalline materials. This type of transformation is difficult or
energy extensive for conventional thermal annealing. We have also investigated the pronounced
effect of multi-stimuli (instead of single stimuli) to observe the microstructural changes.
v
In this research, at first we have chosen e-beam deposited thin films and additive
manufactured (AM) alloys as platform materials. Additive manufacturing is a highly non-
equilibrium manufacturing process where laser sintering/melting results in defects spanning
micro to nanoscales. While the pores and voids can be eliminated by conventional thermal
annealing, more challenging tasks are the nanoscale defects, such as sub-grain structures. Thus,
we have explored the effectiveness of the multi-stimuli processing on microstructure control of
additive manufactured alloys as well as thin films (zirconium and palladium, gold). We also used
ion irradiation to generate controlled defects in polycrystalline gold films and then investigated
the effectiveness of multi-stimuli on defects annihilation.
Another platform material is 2D materials for their extremely small length-scale (mono to
few atomic layers configuration). Our study on chemical vapor deposited (CVD) MoS2, with
few atomic layers, inside a TEM shows the effectiveness of the stimulus effects on defects
annihilation and microstructural changes at low temperatures. Later on, we extend our multi-
stimuli synergy on 2D material based back-gated field effect transistor (FET). External stimulus
such as electrical current generates both resistive heating i.e., Joule heating and atomic scale
force also known as electron wind force (EWF) in a material. Study shows that this external
stimulus can induce significant amount of momentum on the defective sites even at low
temperature due to the EWF. Study also reveals that this unique EWF accompanied at low
temperature can enhance device performance in a short period of time span, which indicates this
proposed technique will potentially lead to time and cost-effective post-processing of two-
dimensional materials and their devices.
The scientific contribution of this research will be experimental validation of the hypothesis
that simultaneously applied stimuli are more effective, energy efficient and faster in achieving
vi
control over defects and microstructure compared to conventional thermal annealing process.
The potential impact of successful validation is a novel material processing technique, whose
unprecedented atomic and defect mobility at lower temperatures will open a new horizon in
defect engineering to modulate physical properties to find applications from nanotechnology to
advanced manufacturing.
vii
TABLE OF CONTENTS
List of Figures
……………………………………………………………………………………
x
Acknowledgements
………………………………………………………………………………
………………………………………………………………………………….
xvii
Chapter 1 Introduction
……………………………………………………………………………
1
1.1 Central Theme of this Dissertation
1
1.2 Structure-Property Relationship in Materials
…………………………
2
1.3 Conventional Microstructural Control Approaches
5
1.4 Proposed Multi-stimuli Approach 7
1.5 Objectives and Impacts of this Research
……………………………………………..
12
Chapter 2 External Stimuli Sensitivity in Thin Films and Additive Manufactured
Alloys …..
19
2.1 Temperature-Electron Wind Force Synergy in Thin Films
……………………..
20
2.1.1 Objective and Motivation ………………………………………..
20
2.1.2 Materials and Methods ………………………………………….. 22
2.1.3 Results and Discussion ………………………………………….. 25
2.1.4 Conclusion ………………………………………………………. 29
2.2 Multi-Stimuli (Electron Wind Force and Mechanical Strain) Effects in
Thin Films ….………………………………………………………………
30
2.2.1 Objective and Motivation ………………………………………..
31
2.2.2 Materials and Methods ………………………………………….. 33
2.2.3 Results and Discussion ………………………………………….. 37
2.2.4 Conclusion ……………………………………………………… 45
2.3 Low Temperature Processing of Additive Manufactured Ti64 Alloy 46
viii
2.3.1 Objective and Motivation ………………………………………..
47
2.3.2 Materials and Methods ………………………………………….. 48
2.3.3 Results and Discussion ………………………………………….. 50
2.3.4 Conclusion ……………………………………………………… 56
Chapter 3 Stimuli Effects on Two-Dimensional (2D) Materials and Devices
……………..
58
3.1 Low Temperature-Electron Wind Force Synergy in Molybdenum Disulfide
…...
59
3.1.1 Objective and Motivation ………………………………………..
59
3.1.2 Materials and Methods ………………………………………….. 62
3.1.3 Results and Discussion ………………………………………….. 65
3.1.4 Conclusion ………………………………………………………. 70
3.2 Low Temperature Processing of 2D Material based Thin Film Transistors
…...
71
3.2.1 Objective and Motivation ………………………………………..
71
3.2.2 Materials and Methods ………………………………………….. 73
3.2.3 Results and Discussion ………………………………………….. 76
3.2.4 Conclusion ………………………………………………………. 83
Chapter 4 Synergy of Stimuli On Operation and Degradation of Nanoscale Devices
……..
85
4.1 On-state Degradatation of High Electron Mobility Transistor
……………….
86
4.1.1 Objective and Motivation ………………………………………..
86
4.1.2 Materials and Methods ………………………………………….. 89
4.1.3 Results and Discussion ………………………………………….. 91
4.1.4 Conclusion ………………………………………………………. 97
4.2 Off-state Failure of High Electron Mobility Transistor
………………………
97
ix
4.2.1 Objective and Motivation ………………………………………..
98
4.2.2 Materials and Methods ………………………………………….. 99
4.2.3 Results and Discussion ………………………………………….. 101
4.2.4 Conclusion ………………………………………………………. 107
Chapter 5 Ion Irradiation and External Stimuli Effect at Nanoscale …………………… 108
5.1 Irradiation Damage and Degradation in Nanoscale Transistor
………………….
109
5.1.1 Objective and Motivation ………………………………………..
109
5.1.2 Materials and Methods ………………………………………….. 111
5.1.3 Results and Discussion ………………………………………….. 114
5.1.4 Conclusion …………………………………………………….. 124
5.2 Low Temperature Synergy on Recovery of Irradiation Damage in Thin
Films…………………………………………………………………….......
124
5.2.1 Objective and Motivation ………………………………………..
125
5.2.2 Materials and Methods ………………………………………….. 126
5.2.3 Results and Discussion ………………………………………….. 127
5.2.4 Conclusion ………………………………………………………. 133
6. References ……………………………………………………………………………. 134
x
List of Figures
Figure 1.1. Passage of electrical current creates thermo-electro-mechanical effects that
can be tuned for microstructural control …………………………………………………
......................
1
Figure 1.2. Electrical Annealing showing defects annihilation …………………………. 2
Figure: 1.3 (a) Defect confinement leading to size effect in materials, and (b) The Hall-
Petch effect in materials …………………………………………………………………
3
Figure 1.4. (a-c) Defects spanning multiple length-scales in metallic materials, Defects
in 2D materials: (d) Vacancy, (e) dislocation, and (f) grain boundary structures in high-
resolution transmission electron microscope (TEM)……...………………………………
5
Figure 1.5. (a) Atomistic representation is showing uniform heating of materials during
conventional thermal annealing, and (b) Time and energy intensive conventional
thermal annealing exhibits grain growth of material ……………………………………
7
Figure 1.6. (a) Effect of electrical current on materials, (b) & (c) an alternative route
shows the localized heating at the vicinity of the grain boundaries (GBs) ……………..
8
Figure 1.7. Synergy of temperature, current and stress fields can create high atomic flux
and mobility along the grain boundaries, which are diffusion pathways in a metal. The
phenomenon is pronounced in areas with higher fraction of defects or disorder (such as
grain boundaries) ………………………………………………………………………..
9
Figure 1.8. (a) Schematic showing active temperature control using liquid N2, (b)
Gripping on massive heat sinks introduces two distinct temperature zone in the sample ..
11
Figure 1.9. MEMS device mounted on in-situ TEM holder, and (b) MEMS device with
heater, sensor, actuators and biasing capability ………………………………………….
13
xi
Figure 1.10. Mechanical straining: (a) Fabricated MEMS device using standard
nanofabrication techniques, (b) schematic showing sensors and samples integrated with
MEMS device, and (c) spring-equivalence of specimen-device system …………………
14
Figure 1.11 Flowchart showing EWF implemented in MD simulation using LAMMPS
package …………………………………………………………………………………...
15
Figure 2.1. (a) Scanning electron microscope (SEM) micrograph of the MEMS device
showing the current flow through the specimen. Inset shows diffraction pattern at 0
A/cm2 current density, (b) Atomistic model with grains oriented at different angles, and
(c) Electro-thermal simulation of sample with actual geometry, resistance and current
density …………………………………………………………………………………… 24
Figure 2.2 In-situ TEM study indicating grain growth as a function of dc electrical
current density ……………………………………………………………………………
26
Figure 2.3. Comparison between thermal and electrical thermal annealing: (a) TEM
bright field image after thermal loading (b) corresponding SAED pattern, (c) MD
simulation cell showing grain boundary reconstruction in limited locations indicated by
arrows, (d) TEM BF image after current loading, (e) corresponding SAED pattern, and
(f) MD simulation showing grain boundary reconstruction due to the electrical current
loading ……………………………………………………………………………………. 28
Figure 2.4. Time evolution of grain growth obtained from MD simulation trajectory: (a)
initial structure, (b) two triple points before electrical annealing, and (c) two triple points
after electrical annealing ………………………………………………………………… 29
Figure 2.5. In-situ TEM experimental and MD simulation setups: (a) Scanning electron
micrograph of the adopted MEMS device with actuators and electrodes including a
TEM bright field (BF) image and diffraction pattern of as-deposited specimen, (b)
temperature profile along the length of the sample obtained from electro-thermal
simulation mimicking the actual experimental conditions, and (c) atomistic simulation
cell with randomly oriented grains used to mimic the experiments …………………….. 34
Figure 2.6. (a) TEM bright field (BF) image of the as-deposited specimen after
prolonged exposure to the electron beam, and (b) corresponding SAED pattern ………. 37
xii
Figure 2.7. In-situ TEM BF and SAED evidence of grain growth in the specimen center
region at (a, d) 0 x105 A/cm2, (b, e) 6.5x105 A/cm2, and (c, f) 7x105 A/cm2 current
densities ………………………………………………………………………………….
39
Figure 2.8. Specimen microstructure at 7x105 A/cm2 current density and room
temperature, (a) before, and (b) after application of 0.1% strain ……………………….. 40
Figure 2.9. MD simulation model on grain growth: (a) initial structure, (b) electrical
annealing prior to the application of strain, and (c) electrical annealing after strain
application, (Green color indicates face centered cubic (FCC) and red color indicates
hexagonal closed packed (HCP) phase of palladium) ……………………………………
43
Figure 2.10. (a) Strain energy increment during tensile straining of the system, and (b)
grain size as a function of applied tensile strain …………………………………………
44
Figure 2.11. Schematic showing experimental set-up with temperature controlled stage.. 49
Figure 2.12. Optical micrographs of Ti64 specimens in the (a) as-built, and (b) Low
temperature EWF processed conditions ………………………………………………….
52
Figure 2.13. Basal Schmid factor maps for the Ti64 specimens in the (a) as-built, and (b)
electric current processed sample, and (c) Calculated Taylor factors Twining during
electrical annealing: (d) as-built specimen and (e) after applying a current density of
5x103 A/cm2……………………………………………………………………………….
54
Figure 2.14. KAM maps for the Ti64 specimens in the (a) as-built and (b) electric
current processed conditions. A threshold of 5° was used to exclude well-defined grain
boundaries in the analysis ………………………………………………………………..
55
Figure 2.15. (a) Force-displacement plot obtained from nanoindentation experiment, (b)
calculated hardness and Young’s modulus ………………………………………………. 56
Figure 3.1. Schematic shows the transfer process of monolayer MoS2 on to a MEMS
device and subsequent experimental setup for in-situ TEM investigation ………………
63
Figure 3.2. (a) Observation of almost one order of magnitude reduction in electrical
resistance of MoS2 specimens during EWF annealing, and (b) spatial temperature
distribution for the highest current density ………………………………………………
64
xiii
Figure 3.3. In-situ TEM electron wind force annealing results: (a) Bright-field image,
(b) SAED pattern of as-deposited nominally monolayer MoS2 specimen, and (c, d) The
same location after annealing at 9.5x105 A/cm2 current density ……………………….
66
Figure 3.4. (a)-(c) 6|8Mo type dislocation migration during the electrical annealing, (d)-
(e) Transformation of a 6|8 ring to 6|6|4 ring at the Grain boundary (GB) (Individual
grains are shown by different colors, smaller radius sphere indicates Sulphur atoms and
larger sphere indicates Molybdenum atoms) …………………………………………….
68
Figure 3.5. Transformation of vacancy defects at the GB: (a) initial sample, (b) vacancy
transforms to 6|8 S defects, (c) 6|8 S transforms to 6|6|4 ring defects, (d) formation of
6|8 S due to the dislocation motion, and (e) formation of 6|6|4 ring (Individual grains are
shown by different colors, smaller radius sphere indicates sulfur atoms and larger sphere
indicates molybdenum atoms) ……………………………………………………………
69
Figure 3.6. (a) Optical Microscope image of fabricated back-gated WSe2 FET transistor,
(b) schematic diagram of WSe2 transistor, (c) electro-thermal simulation model of WSe2
FET, and (d) temperature profile across the cross-section of the sample obtained from
model ……………………………………………………………………………………..
76
Figure 3.7. (a) Output characteristics of WSe2 FET after annealing at different drain
voltage, and (b) Improvement in drain current after annealing while FET surface was
maintained at 296K ……………………………………………………………………….
78
Figure 3.8. (a) Transfer characteristics showing WSe2 transistors performance after
annealing at different voltage, and (b) maximum drain current obtained after annealing ..
80
Figure 3.9. Monolayer WSe2: (a) prior to the annealing, and (b) after annealing ……… 81
Figure 3.10. (a)-(f) Failure at high biasing condition due to the electrical and thermal
field, and (g)-(j) side view of the sample showing void creation at the cathode side (left)
and mass accumulation at the anode side (right) (Color bar in Figure 3.12 shows the
atomic stress distribution in the sample) …………………………………………………
82
Figure 4.1. Comparison of properties among competing semiconductors ……………..... 87
Figure. 4.2. (a) Optical micrograph of GaN HEMT die, and (b) low magnification SEM
image of the die …………………………………………………………………………... 89
xiv
Figure 4.3. Details of the GaN HEMT specimen preparation and transfer technique
using FIB for in-situ TEM reliability study …………………………………………........ 90
Figure 4.4. In-situ TEM reliability study of GaN HEMT: (a) electron transparent
specimen prior to the application of voltage stimulus, (b) device after failure, and (c)
output characteristic curve obtained from “on-state mode operation” …………………… 92
Figure 4.5. (a) Degradation of the passivation layer, (b) Evaporation of the buffer layer
due to the high thermal field, and (c) SAED indicating the transformation of GaN from
crystalline to amorphous state ……………………………………………………………. 94
Figure 4.6. GaN HEMT device degradation: (a) Hot electron induced failure at the
source side, (b) evaporation of the buffer layer and formation of small crystallites
(nanoparticles) due to the high thermal field, and (c) high-resolution image of a
spherical crystallite ………………………………………………………………………. 95
Figure 4.7. In-situ TEM reliability testing showing real-time device degradation ………. 96
Figure 4.8. (a) The experimental setup showing a MEMS chip on a TEM specimen
holder, (b) The specimen integrated with the MEMS chip, (c) SEM image of the
electron transparent GaN HEMT specimen, and (d) transfer characteristic of the HEMT
die, and (e) Off-state loading of the 100 nm thick HEMT sample ………………………. 100
Figure 4.9. Off-state characterization of (a) die-level transistor, and (b) 100 nm thick
HEMT during phase I loading …………………………………………………………… 101
Figure 4.10. Bright field TEM images acquired at drain voltages: (a) 0V, (b) 7.2V (c)
11.6V and (d) 23V, and (e) corresponding drain current vs. drain voltage data at gate
voltage -5V for a 100 nm thick GaN HEMT specimen …………………………………. 103
Figure 4.11. Real-time operation of GaN HEMT: (a) electron transparent HEMT
specimen prior to the loading, where arrows indicating pre-existing defects, (b)
magnified view of the arrows marked 2 and 3 at the on-set of source-drain leakage, (c)
molten drain metal pool at this instant the drain side at current density of 2000 mA/mm,
(d) metal diffusion through the GaN layer, (e) rapid breaching of the GaN-SiC interface,
105
xv
and (f) degradation of the SiC layer ………………………………………………………
Figure 4.12. EDS mapping of a failed HEMT specimen at the (a-d) gate and (e-h) drain
areas. (i, j) normalized weight percentage of the various elements in the gate and drain
area respectively ………………………………………………………………………….. 106
Figure 5.1. Experimental setup for in-situ TEM experiment of electron transparent
HEMTs: (a) GaN HEMT die, (b) a MEMS chip with the HEMT specimen mounted on
in-situ TEM electrical biasing holder, and (c) FIB lamella of the HEMT before
mounting on to the MEMS chip …………………………………………………………. 113
Figure 5.2. (a) Downward arrowhead in the schematic diagram of the GaN HEMT
showing irradiation direction, (b) displacement per atom (dpa) profile as a function of
depth for different doses of irradiation, (c) TEM image of a pristine HEMT showing
mostly bend contours, and (d) TEM image of an irradiated HEMT at 45 dpa showing
very high dislocation density …………………………………………………………….. 115
Figure 5.3. Die-level HEMTs specimens characterization curve as function of ion
irradiation damage in dpa: (a) transfer characteristics, and (b) output characteristics ....... 116
Figure 5.4. Electron transparent HEMT device: (a) before, and (b) after 2.8 MeV Au4+
ion irradiation for 60 minutes to a fluence of at 4x1014 ions/cm2. The rectangular dashed
box shows contrast change due to point defect accumulation, while the arrows indicate
dislocation activities at the GaN-SiC interface ………………………………………….. 117
Figure 5.5. (a) Drain current vs. drain voltage plot of electron transparent GaN HEMT
specimen during off-state operation inside the TEM. The data labels correspond to the
in-situ TEM images in Figure 5.6., and (b) Comparison between pristine and irradiated
conditions ………………………………………………………………………………… 119
Figure 5.6. TEM BF images showing source, gate and drain at the same time. Drain
voltage: (a) Vd= 0V, (b) TEM image of screw dislocations in buffer layer, (c) Vd= 5V,
(d) Vd= 7V, (e) Vd= 8.5V, and (f) Vd= 10.2V ……………………………………………. 120
Figure 5.7. (a) TEM BF image at the drain side of drain-gate region, (b) Dislocations in
the GaN layer, (c) High-resolution TEM (HRTEM) image of dislocations in GaN layer,
122
xvi
(d, e) atomic strain mapping in the sample showing normal and shear strain field
associated with individual dislocations, and (f) simulated lattice fringes with the
dislocations ……………………………………………………………………………….
Figure 5.8. EDS mapping of GaN HEMT showing diffusion of chemical elements at
different drain bias: (a) Vd=0V, (b) onset of failure at Vd=10.2V, and (c) relative
changes in diffusion of chemical elements obtained from EDS at these two voltages ….. 123
Figure 5.9. (a) Displacement per atom (dpa) profile for irradiation dose of 6.5×1015
ion/cm2, (b) micro-electro-mechanical system (MEMS) device mounted on in-situ TEM
holder, (c) temperature profile obtained from electro-thermal simulation, (d) TEM BF
image showing irradiation damage, (e) BF image showing dislocation annihilation at
9.5×105 A/cm2, and (f) SAED pattern after EWF annealing …………………………….. 128
Figure 5.10. (a) Irradiated sample before processing, (b) dark field image of pre-
processed sample showing dislocation lines, (c) dislocation lines interaction during
EWF processing at 3.5×105 A/cm2, (d) partial annihilation (pink circle) of dislocation
lines, and migration towards grain boundary at 7×105 A/cm2, (e) migration of
dislocation lines towards GB and partial annihilation of dislocation lines at 9×105
A/cm2, and (f) complete annihilation of dislocations and defects at 9.8×105 A/cm2 …….. 130
Figure 5.11. Computational results on (a)-(c): Annihilation of dislocations and vacancy
clusters, and (d)-(f): SFT annihilation under EWF ………………………………………. 131
Figure 5.12. HRTEM images of low temperature processing of irradiated materials: (a)
defects in the irradiated sample before processing, (b) HRTEM image of SFT, (c)
surface induced annihilation of SFT at a current density of 7×105 A/cm2, (d) SFT-GB
interaction at 9×105 A/cm2, (e) after EWF annealing at 9.5×105 A/cm2, and (f) HRTEM
image after annealing at 9.5×105 A/cm2 …………………………………………………. 132
xvii
Acknowledgement
I would like to express my sincere gratitude to my advisor Dr. Aman Haque for his continuous
support, patience, and encouragement during my Ph.D. research. His resourceful guidance and
positive criticism helped me to accomplish my research goal. I could not have imagined having a
better advisor and mentor for my Ph.D. study.
Besides my advisor, I would like to thank the rest of my doctoral committee: Dr. Adri van Duin,
Dr. Daudi Waryoba and Dr. Douglas E. Wolfe for their time, and willingness to help.
My sincere thanks goes to Dr. Stephen Pearton and Dr. Fan Ren at University of Florida, Dr.
Khalid Hattar at Sandia National laboratories, Dr. Nicholas Glavin at Air Force Research
Laboratory, Dr. Joshua Robinson, and Dr. Saptarshi Das at Penn State, Dr. Huajian Gao at
Brown University, and Dr. Tim Rupert at University of California Irvine who provided me the
opportunity to work with them. Special thank goes to Dr. Daudi Waryoba for sharing his
profound knowledge with me on electron back-scattered diffraction (EBSD) technique.
I thank my former colleagues Dr. Baoming Wang, Dr. Raghu Pulavarthy, Dr. Tun Wang,
Niranjana Sunderasan, James Kidd and Angela Paoletta for sharing their invaluable experience
with me. I am grateful to my friends for their support and encouragement. I would also like to
thank MRI staff at Penn State Nanofab: Guy Lavallee, Shane Miller, Kathy Gehoski, Michael
Labella, Andy Fitzgerald, Chad Eichfeld, Ted Gehoski, Bill Drawl, Bill Mahoney, Bangzhi Liu,
Beth Jones; Penn State Material Characterization Lab: Trevor Clark, Jennifer Gray, Tim Tighe,
Wes Auker, Max Wetherington, Nichole Wonderling, Maria Dicola, Julie Anderson, Ke Wang,
Haiying Wang, Mike Norrell; Penn State Electrical Characterization Lab: Jeffrey Long and Steve
xviii
Perini for helping me with tools’ training regarding the nanofabrication process, microscopy,
and characterization techniques.
I gratefully acknowledge the support from the Division of Civil, Mechanical, & Manufacturing
Innovation (Nanomanufacturing program) of the National Science Foundation through award #
1760931 and US National Science Foundation (DMR 1609060). The views expressed in the
dissertation do not necessarily represent the views of the US National Science Foundation or the
United States Government.
Last but not the least, I would like to thank my parents for their unconditional love and supports
throughout my life. I would also like to thank the rest of my family members for supporting me
throughout my Ph.D. journey and my life in general.
1
Chapter 1 Introduction
1.1. Central Theme of this Dissertation
Mechanical properties such as yield strength and fracture toughness are governed by
microstructure (grain size, crystallinity, phases and precipitates, defects) at all length-scales [1-
6]. Similarly, electrical and thermal properties are also microstructure dependent. This is evident
from a vast literature in mechanics and physics of materials dedicated to quantitative and
qualitative mapping of structure-property relationship [1, 7-12]. Therefore, significant efforts
have been made to develop materials synthesis and processing techniques with the core objective
of achieving control over the materials properties and functionality. The state of the art
microstructural processing techniques predominantly use temperature as the stimulant while
often experimenting with alloying elements, defects and interfaces (such as grain boundaries) to
control microstructure. Even after extensive research, microstructural control has remained
elusive, which is the major motivation for this research.
Figure 1.1. Passage of electrical current creates thermo-electro-mechanical effects that can be
tuned for microstructural control.
2
This dissertation explores a new route where synergy of thermal, electrical and
mechanical stimuli is exploited for faster, energy efficient materials processing. Figure 1.1 shows
experimental evidence of electrical current driven microstructural/defect control in additive
manufactured 316 stainless steel. Passage of electrical current in metallic materials compound
thermal, mechanical and electrical effects that can contribute to the atomic or defect mobility. A
unique aspect of this dissertation is to consider athermal effects by eliminating the thermal field
generated from the Joule heating. This is shown in the shaded region in Figure 1.2. Another
important aspect is the role of any external mechanical (stress) or thermal (temperature)
stimulant superimposed on the electrical current density effects. Localization effects of these
stimuli is proposed to be the predominant factor behind the synergy. The resulting mechanics
and physics involve an intricate relationship between solid mechanics, heat transfer, electro-
migration and materials science, which we aimed to explore in this dissertation.
Figure 1.2. Proposed multi-stimuli synergy for materials processing.
1.2. Structure-Property Relationship in Materials
Figure 1.3 illustrates how microstructure governs mechanical properties at all length-
scales. Here, we define physical size as the thickness or width of a specimen material.
3
Microstructural size, on the other hand, can be grain size (D), diameter of dislocation loop (d),
obstacle spacing (L), spacing between partial dislocations (w) among others [2]. While the
strength does not differ for a steel rod of two different cross-sections, it can significant vary if the
grain sizes are different even for the same cross section. For metals and alloys, the classical Hall-
Petch formulation is shown in Figure 1.3b, which demonstrates the influence of microstructure
over yield strength. It is essentially a scaling law for strength, which is applicable for grain size
as small as 100 nm. Below this length-scale, the Hall-Petch scaling law becomes less reliable. At
the same time, when the physical size becomes comparable with this length-scale, remarkable
compounding effect on properties is observed. For even smaller size, either physical or
microstructural, the classical scaling law breaks down and can reverse. The inverse Hall-Petch
relationship has been observed experimentally which indicates yield strength decreases or
remains constant at or below 100 nm grain size [3]. Thus, we define ‘characteristic’ length-scale
as where classical scaling law breaks down or deviates appreciably.
Figure: 1.3 (a) Defect confinement leading to size effect in materials [2]. (b) The Hall-Petch
effect in materials [4].
4
Microstructural defects in materials spanning from atomic vacancy to three-dimensional
pores/voids pervade over the length-scales (Figure 1.3) and significantly affect physical
properties [1-3, 6, 8, 10, 12-14]. For example, grain size in metals is one of the most influential
parameter that controls physical properties of nanocrystalline materials [5]. Nanocrystalline
materials exhibit pronounced grain size effects (10-100nm) as manifested through the breakdown
of the classical Hall-Petch relation [6] (shown in Figure 1.3b). Likewise, in a two-dimensional
(2D) material, if physical size such as grain size (1-20nm) overlaps characteristics length scale
i.e., mean free path in thermal or electrical domain, then a remarkable effects are observed on
transport properties [15-19]. Hence, electrical, thermal and optical properties are influenced by
grain size. Electrical conductivity decreases with grain size [7], which is prominent around
length-scales (film thickness, surface roughness and/or grain size) comparable to the electron
mean free path [20]. With decreasing interconnect size, electrical conductivity is critical for
electronic devices. In metals, heat is carried by electrons and a similar grain size dependence of
thermal conductivity is observed [8]. Similarly, phonon scattering is also grain size dependent
[21]. In addition to the grain size, other microstructural defects such as interstitials, vacancies,
dislocations and stacking faults [22, 23] also affect physical properties such as mechanical [13,
24-26] , electrical [14, 27, 28], thermal [29-31], and optical properties [32-34], to name a few.
Figure 1.4 represents commonly observed different types of defects in metals and two-
dimensional (2D) materials. Grain size control, elimination of defects and residual stress is
critical from both performance and reliability perspective for real-world applications [35].
Therefore, high temperature annealing is a common post-processing step in fabrication.
However, high temperature annealing does not always result in perfect microstructure.
Depending upon the systems, it can create thermal stress to stagnate the defect diffusion or even
5
may create new defects. In the following sections, we will discuss how we can gain control over
microstructure and defects. To answer this question, we will investigate the available material
processing techniques and their advantages as well as limitations, and how we can overcome
these limitations.
Figure 1.4. (a-c) Defects spanning multiple length-scales in metallic materials [36-38], Defects in
2D materials: (d) Vacancy [39], (e) dislocation [40], and (f) grain boundary structures [41] in
high-resolution transmission electron microscope (TEM).
1.3 Conventional Microstructural Control Approaches
Materials science has evolved around ‘processing-microstructure-property’ relationship,
where the processing or synthesis component is particularly highlighted. Materials processing,
6
for example annealing, requires mobility of defects and atoms, which has been conventionally
imparted by high temperature. Thermal processes such as conventional thermal annealing induce
random diffusion, which is a very slow mechanism [42] as shown in Figure 1.5b. Also, they heat
both defective and crystalline regions uniformly. This is schematically shown in Figure 1.5a. In a
conventional annealing thermal stimulus is used for controlling grain size, phases and defect
density. Conventional thermal annealing is performed by raising the temperature uniformly to a
very high value (homologous temperature exceeding 0.5) at certain rate, and then hold it and
eventually cool it down to room temperature at rates depending on the material [43-45]. The role
of temperature here is to increase atomic and defect mobility to facilitate microstructural
reorganization. However, during the uniform temperature raise of the material most of the energy
we spend to heat up the existing crystalline area (grain interior), hence we consider the uniform
temperature raise of the whole grain as an energy inefficient technique. For example, Figure 1.5b
shows how temperature and time influence the strength and grain size respectively. In addition,
during the thermal annealing of materials and devices with complex interfaces, requirement of
high temperature itself can degrade the materials by creating more defects or stagnating the
existing ones due to the residual thermal stress between materials’ interfaces. Thus, this high
temperature processing might not work for flexible electronics or temperature sensitive devices.
These conventional processing limitations inspire us to investigate an alternative pathway for
active controlling of microstructure and defects. We therefore conclude that the current art of
single (thermal) stimuli microstructural control is not only energy intensive, but also time
consuming, and could be detrimental for flexible or temperature sensitive devices. Thus, in the
following sections we will probe in detail on our newly developed low temperature and time
7
efficient technique, which exploits multi-stimuli synergy to process and control microstructure in
an active manner.
Figure 1.5. (a) Atomistic representation is showing uniform heating of materials during
conventional thermal annealing, and (b) Time and energy intensive conventional thermal
annealing exhibits grain growth of material [42].
1.4 Proposed Multi-Stimuli Approach
While thermal processing has been around since the Bronze Age, this dissertation
considers a new route, where synergy of electrical current, mechanical strain and temperature is
focal point of research. A potential outcome could be non-thermal enhancement of defect or
atomic mobility that could help us to achieve microstructural control. We hypothesize that at
micro to nanoscales (relevant to microstructure but not necessarily the physical dimension) these
stimuli are synergistic, i.e., their effects are not merely additive but compounding. A corollary of
our hypothesis is that significantly higher atomic and defect mobility can be achieved even at
lower temperatures due to the electrical current passage, and the effects will be more pronounced
8
if it is accompanied by mechanical strain. Thus, in our present study, we developed a non-
thermal route to conventional thermal annealing. Figure 1.6 illustrates approach where we pass
electrical current through metallic or semiconducting materials. The literature identifies this
process as ‘electrical current annealing’ (or more commonly electro-pulsing [46-50]), which
induces Joule heating (Figure 1.6a) to raise the temperature. It also develops ‘electron wind
force’ (EWF), which is nothing but the momentum transfer from electrons to defects. Thus, EWF
is inherently a mechanical force that is highly localized around defects where the momentum
transfer (or electron-defect scattering) takes place. This is analogous to a gentle wind blowing
the leaves (defects) in a tree while the limbs (lattice) remain unperturbed. A major contribution
of this dissertation is the investigation of EWF alone on the acceleration of defect mobility. In a
stark contrast to electro-pulsing, we achieve this by actively removing the Joule heating. This not
only keeps the material at near-room or low temperatures, but also triggers mobility ‘just in
location’ or where needed. In other words, our proposed isolation and application of EWF
specifically targets defective atoms, leaving the crystalline lattice alone.
Figure 1.6. (a) Effect of electrical current on materials, (b) & (c) an alternative route shows the
localized heating at the vicinity of the grain boundaries (GBs).
9
To investigate our hypothesis, we start with very small current density (30-50% of the
electromigration failure limit). Electrical current is a very effective stimulus to generate [51, 52]
and flow vacancies. This is due to the electron wind force arising from the momentum transfer
between conduction electrons and metal ions. The most important aspect is that the current
density impacts the defects and grain boundaries significantly more than the crystalline grain
interior. Inside the grains, the uniform lattice structure means there is less momentum transfer.
The opposite is true at the grain boundaries, where most of the vacancy generation and motion
takes place [53]. Figure 1.7a shows this phenomenon, where blue arrows indicates electron flow
and red arrows shows the impact of electron on grain boundary. Thus, electrical annealing is
highly localized to defects around the grain boundary, where we need the atomic mobility and
not the entire grain area. This enhances the energy efficiency.
Figure 1.7. Synergy of temperature, current and stress fields can create high atomic flux and
mobility along the grain boundaries, which are diffusion pathways in a metal. The phenomenon
is pronounced in areas with higher fraction of defects or disorder (such as grain boundaries) [54].
10
Mechanical stress field also has a synergistic effect on the electrical annealing process
[55]. The vacancy flux due to electrical annealing decreases as soon as internal stresses build up
in the metal due to vacancy transport from cathode end to anode end. The role of an external
mechanical field is to counter the effects of the internal stress and the net effect will be higher
vacancy flux and mobility [56]. The most important aspect of stress field is that stresses are
highly localized at the grain boundaries (Figure 1.7c). This means only a small amount of stress
is required as an external stimulus, because it becomes amplified at the defects and grain
boundaries. This effect is very similar to current density, and therefore it also specifically targets
defects and not the material uniformly, which makes it both time and energy efficient. To
summarize, we can express individual contribution from each stimulus such as EWF (JEWF),
temperature (JT) and mechanical stress (Jσ) on the atomic flux mobility as follows [57]:
𝐽𝐸𝑊𝐹 =𝑁𝑒𝑍∗𝐷𝜌
𝑘𝑇𝑗; 𝐽𝑇 = −
𝑁𝐷𝑄𝜌
𝑘𝑇2 ∇𝑇; 𝐽𝜎 =𝑁𝐷𝛺
𝑘𝑇∇𝜎 (1.1)
where, N is vacancy concentration, k is Boltzmann’s constant, T is absolute temperature, e is the
elementary charge, Z* is effective charge number, D is diffusivity, ρ is resistivity, j is current
density, Q is the heat of transport, Ω is the atomic volume, and σ is the hydrostatic stress.
Since our goal is to develop non-thermal low temperature microstructural processing
using multi-stimuli, a distinct feature of our study is that we remove the heat generated by
current to separate the effects of temperature. Thus, to ensure low temperature processing we
took two distinct approaches. In the first approach, we actively control the sample temperature
by passing liquid N2 through the sample stage as shown by the schematic diagram in Figure 1.8a.
In the second approach, we gripped freestanding thin film specimen with massive heatsinks. For
a freestanding specimen, the current flow heats up the mid-point region the most, while the two
11
end regions (anode and cathode) are constrained to lower temperature as shown in Figure 1.8b.
This is because the end regions are more massive and act as heat sinks. Suitable design of the
specimen and the two ends makes it possible to have the ends at ambient condition. The
uniqueness of this setup is that we are able to decouple the effect of Joule heating and the
electron wind force in the same specimen. The mid-point of the specimen shows the effect of
EWF and temperature, while the two end regions show the effects of EWF only.
Figure 1.8. (a) Schematic showing active temperature control using liquid N2, (b) Gripping on
massive heat sinks introduces two distinct temperature zone in the sample.
To validate the low temperature processing, we pass electrical current through the
specimen. The electrons are minimally scattered by the lattice, but they transfer their momentum
wherever they meet defects and grain boundaries [58]. This gives rise to highly defect-specific
atomic scale force EWF. By combining this EWF with other stimuli such as mechanical strain,
we can achieve active microstructural control across multiple length scale, namely bulk-, micro-,
and nano-scale. The uniqueness of this technique is the non-thermal processing of microstructure
and defects using multiple stimuli i.e., mechanical strain and EWF, whereas the latter one is
intense around the defects and non-existent in defect-free lattices.
12
1.5 Objectives and Impacts of this Research
The objective of the present research is to validate the hypothesis that at micro to nanoscales
multi-stimuli synergy can be exploited to achieve time and energy efficient microstructural
processing and control at low temperature to gain control over microstructure and defects. Thus,
impact of this hypothesis is the potential for tailoring defects and microstructures for desired
performance. Our developed multi-stimuli materials processing technique can potentially
outperform the existing approaches in both speed and energy requirement, thereby profoundly
impacting the next generation materials processing technology. The impact of this research can
be realized for the design and development of materials spanning from nano to microscale. For
example, these unique multi-stimuli synergy can be exploited to enhance performance of next
generation advanced manufacturing including but not limited to nanoelectronics, sensors as well
as additive manufactured alloys. Our scalability study of multi-stimuli synergy on additive
manufactured metallic alloys shows potential promise for active microstructural control, which
will be realized by the large potential of metallic applications in civil and military infrastructure,
automotive, aerospace, heavy machinery and cutting tools, oil and gas, protective coatings, bio-
medical implants and devices, to name a few [59-61]. Additionally, discovery of graphene[62]
has brought the concept of single layer atomic device and components such as back-gated field
effect transistor (FET) closer to the real life applications. Thus, 2D materials such as graphene,
h-BN, MoS2, WSe2, phosphorene, silicene etc., have gained significant interest due to their
outstanding electrical, optical, chemical, and thermal properties [63-70]. Potential applications of
2D materials encompasses electronic devices[71], sensors[72], catalysts[73], energy conversion,
storage devices[74], to name a few. Our onboard low temperature rapid processing might be an
13
alternative viable path to enhance device performance without introducing residual thermal
stress.
To accomplish this objective, we have developed a versatile micro-electromechanical
system (MEMS) based experimental setup with sensors, heater and actuators integrated with
micro-, nano-scale specimens. The uniqueness of this set-up allows measuring multi-domain
properties under controlled exposure to multiple stimuli as well as measuring the resulting
properties at high resolution. These fabricated MEMS devices (as shown in Figure 1.9) have
been also tested with electron, infrared, thermoreflectance, and Raman microscopes to access the
various stimuli and or domains. Figure 1.9b shows the SEM micrograph of the fabricated MEMS
device with sensors, actuator and heaters. In order to accomplish electrical annealing, we
supplied electrical current through the electrode pad B1 and C. Mechanical strain can be applied
on the sample by passing current through the actuator pad A1 and A2. Combination of these
stimulus (e.g multi-stimuli effect) can be achived at the same time by powering up the elctrode
pads simultaneously. Figure 1.10 shows our devices’ capability to study multi-stimuli synergy on
microstructural control.
Figure 1.9. MEMS device mounted on in-situ TEM holder, and (b) MEMS device with heater,
sensor, actuators and biasing capability.
14
Figure 1.10. Mechanical straining: (a) Fabricated MEMS device using standard nanofabrication
techniques, (b) schematic showing sensors and samples integrated with MEMS device, and (c)
spring-equivalence of specimen-device system.
To understand the multi-stimuli synergy on microstructure control, we have also developed
molecular dynamics simulation code using LAMMPS [75] package. We started with single
stimulus effect (e.g either mechanical strain, thermal or electrical current) and later on, we have
extended our study to observe the multi-stimuli effects on metallic thin films as well as 2D
materials. The goal of the multi-stimuli study using computational techniques such as molecular
dynamics is to assess the compounding effect on microstructural changes, grain growth and
defect annihilation not quantitatively but qualitatively. To mimic the EWF effects in our MD
simulation, we apply additional force on individual atoms during the simulation. Thus, the net
force on each atom is a combination of force due to the interatomic potential and imposed
electron wind force. A flowchart to express EWF implementation in MD simulation is shown as
follows:
15
Figure. 1.11. Flowchart showing EWF implemented in MD simulation using LAMMPS [75]
package.
Where, r is the position, v is the velocity, U is the potential energy, m is the atomic mass, Δt
is the time step, F is the force, 𝑍∗is the effective valence number, e is the electron charge, 𝜌 is the
16
specific resistivity, j is the current density, and a is the acceleration. Similar approach has been
taken to study electromigration failure at higher current density in metal interconnects using MD
simulation [76, 77]. In our study, we have implemented embedded atom model (EAM) potential
to simulate metal and alloys. EAM potentials are well calibrated and has been used to study
metals and alloys [78, 79]. Functional form of an EAM potential for pure metals requires at least
cohesive energy, lattice parameter and elastic properties calibration. In addition to these
parameters, alloys require heat of fusion and energy differences among phases for the calibration
process. EAM potentials used in our study are well calibrated and reported in literatures [78, 79].
We have also used reactive empirical bond-order (REBO) [80] and Stillinger-Weber (SW) [81]
potentials for molybdenum disulfide (MoS2) and tungsten diselenide (WSe2) samples
respectively. Both REBO [80] and SW [81] potentials have been calibrated and used to study
mechanical and thermal properties of MoS2 and WSe2 sample respectively. Additionally, recently
developed ReaxFF [82] potential has been implemented to study mechanical properties of MoS2.
The results obtained from the ReaxFF [82, 83] are in good agreement with experimental data,
and this agreement could be attributed to the inclusion of higher number of parameters and
probably more accurate functional form of ReaxFF potentials. However, in this dissertation we
choose REBO and SW potentials to study 2D materials such as MoS2 and WSe2 to perform
simulations within affordable computational time with large simulation cells containing grain
boundaries (GBs).
In this dissertation, we have investigated the effects of various external stimuli on the
microstructure control and properties of different material system across multiple length scale,
and chapters’ summarization are as follows:
17
In chapter 2, we have addressed multi-stimuli effects such as electrical current, electron wind
force (EWF), Joule heating, and mechanical strain effects at nanoscale. To accomplish this goal
we have investigated electrical current flow effects below the electromigration failure limit in
nanocrystalline 100nm thin films. We observe one order of magnitude higher grain growth
indicating concurrent effects of electron wind force and Joule heating specifically target the grain
boundaries, producing much higher grain boundary mobility compared to high temperature
annealing alone. Afterwards, we have investigated multi-stimuli synergy i.e., a combined effect
of electron wind force, joule heating and mechanical strain effects. Our in-situ TEM study
reveals that application of mechanical strain even in the elastic range (about 0.1%) dramatically
increased the grain size almost instantaneously at the low temperature regions. This is
unprecedented because the annealing and recrystallization literature is essentially founded on the
concept of plastic strain initiated defects that act as nuclei of recrystallization. In the subsequent
section, we have explored multi-stimuli synergy on micro- and bulk-scale samples considering
scaling up issues. Our studies show that the accelerated atomic and defect mobility induced by
multi-stimuli can be exploited for microstructural control of additive manufactured alloys.
In chapter 3, we have considered electrical current effect on non-thermal microstructure
controlling process at extreme length scale i.e., <10nm. To assesses the effectiveness of low
temperature processing at this extreme length scale, we choose two-dimensional (2D) material
with few layers thin (2~3 layers) nanocrystalline molybdenum disulfide (MoS2). Our study
reveals that moderate current density gives rise to atomic scale mechanical force whenever the
electrons encounter defects in the lattice or grain boundaries. The effectiveness of the defect
annihilation is reflected on the physical properties improvement of 2D materials after processing.
Later on, we implemented this low temperature processing on back-gated tungsten diselenide
18
(WSe2) field effect transistor (FET) for real world application. Our study indicates that EWF
driven low temperature processing can effectively annihilates defects and results in two order of
magnitude higher output current.
In Chapter 4, we shifted our material system to thin films transistor with more complex
geometry and shape. We chose GaN based high electron mobility transistor (HEMT). Due to the
specimen geometry and delicate interface, thin film HEMT inherently possesses residual stress.
Again, depending on the mode of operation external stimuli such as voltage biasing can induce
high thermal, mechanical and electrical field in HEMT devices. Thus, HEMT devices are ideal
candidates to assess the effectiveness of the multi-stimuli effects on thin films, devices and
interfaces. This study paves the path for the real-time operation of HEMT inside a high-
resolution electron microscope. Though this study is not directly related to the micro-structural
control process, it explains the degradation mechanism of complex material system under the
combined effects of multiple stimuli.
In chapter 5, we have considered real-world application of our newly developed low
temperature processing technique to eradicate irradiation-induced defects. The objective is to
recover or enhance the physical properties by actively controlling defects and microstructure.
Irradiation is considered as an energetic process where highly energetic particles (ions, neutrons,
electrons) collide with atoms in materials, energizing and displacing them from their original
lattice positions, thereby generating various types of defects. In the first section of this study, we
explore the effects of irradiation on HEMT and their subsequent failure mechanism under the
effects of external stimuli. In the next section, we explore the effectiveness of electron wind
force in annihilating defects originating from irradiation damage. Study reveals that EWF can
efficiently eliminate defects in irradiated materials even at low temperature.
19
Chapter 2
External Stimuli Sensitivity in Thin Films and Additive Manufactured Alloys
Contents of this chapter are based on the following journal articles:
Zahabul Islam, Baoming Wang, Aman Haque, Current density effects on the microstructure of
zirconium thin films, Scripta Materialia,Volume 144, Pages 1359-6462, 2018.
Author of this dissertation designed the experiment, performed the sample preparation,
device design and fabrication, experimentation, data analysis, computational modeling as
well as computational data analysis and manuscript writing. Baoming wang assisted on
experiment design and manuscript writing. Aman Haque guided on experiment design,
and involved in data analysis as well as manuscript preparation.
Zahabul Islam, Huajian Gao, Aman Haque, Synergy of elastic strain energy and electron wind
force on thin film grain growth at room temperature, Materials Characterization, Volume
152,Pages 85-93, 2019.
Author of this dissertation designed the experiment, performed the sample and device
preparation, experimentation, data analysis, computational modeling as well as
computational data analysis and manuscript writing. Huajian Gao guided on
computational data analysis, discussion and manuscript preparation. Aman Haque guided
on experiment design, and involved in data analysis as well as manuscript preparation.
Daudi Waryoba, Zahabul Islam, Ted Reutzel, Aman Haque, Electro-Strengthening of the
Additively Manufactured Ti-6Al-4V Alloy, (submitted)
20
Author of this dissertation designed the experiment, performed the sample preparation,
experimentation, computational modeling as well as computational data analysis and
manuscript writing. Daudi Waryoba also conducted the experiment, data analysis and
manuscript writing. Ted Reutzel was involved in additive manufacturing of the Ti64
alloy and manuscript writing. Aman Haque guided on experiment design, and involved in
data analysis as well as manuscript preparation.
2.1 Temperature-Electron Wind Force Synergy in Thin Films
In this section, we investigate the multi-stimuli effects such as electrical current flow in
nanocrystalline zirconium thin films using in-situ Transmission Electron Microscope (in-situ
TEM) and molecular dynamics (MD) simulation. We observed at least one order of magnitude
higher grain growth at current density of 8.5x105 A/cm2 (Joule heating temperature 710 K) in 15
minutes compared to conventional thermal annealing at 873 K for 120 minutes. Both experiment
and simulation results support our hypothesis that the concurrent effects of electron wind force
and Joule heating can produce much higher grain boundary mobility compared to high
temperature annealing alone, and lead to grain growth.
2.1.1 Objective and Motivation
Nanomaterials show strong grain size dependence of their physical properties [4, 5]
across length scales [2], which has motivated the pursuit for microstructural optimization and
control. In a conventional thermal annealing temperature is used as the stimulus for controlling
grain size, phases and defect density. For most of the metallic materials, this temperature is in the
range of 0.3-0.4 𝑇𝑚 (where 𝑇𝑚 corresponds to homologous temperature). However, the applied
temperature field is uniform, targeting both crystalline and defective regions, thus making the
21
process time consuming and energy inefficient. Present study proposes that electrical current
could potentially achieve similar or higher grain boundary and defect mobility at lower energy
and time input. This is due to the pronounced scattering at the grain boundaries and defect sites
[51, 52], effectively enhancing atomic mobility exactly where it is needed for grain growth, and
not uniformly across the sample.
High current density effects are typically considered to be a cause for degradation of
microelectronic interconnects through electromigration [52, 84]. Beyond a critical density, mass
transport takes place due to the electron momentum transfer, particularly intensified at the
defective areas such as grain boundaries (GBs). Other current density studies have focused on
electro-plasticity [85], a phenomenon where electrical current flow induces plasticity in materials
that are otherwise very hard and brittle. To study the fundamentals of electrical current density
effects on microstructures, we adopted a combined experiment-simulation approach. The
experiments were performed inside a Tecnai-LaB6 Transmission Electron Microscope (TEM).
The high resolution imaging and selected area diffraction (SAD) modes make TEM first choice
in visualization and characterization of microstructural changes [86]. However, the challenge in
this technique is the very small work envelope of the TEM chamber, typically accommodating 3
mm diameter grids for specimens [9]. Molecular dynamic (MD) simulation code has been widely
implemented to study mechanical properties [87-89] and electro-migration failure [90]. The
primary modelling challenges are incorporating electron-matter interaction during transport
directly. To overcome this barrier, in our MD modeling approach, we represent the effect of the
electrical current by applying an equivalent electron wind force (EWF) and observe the resulting
atomic/defect migration. The discrepancy in time and length-scales between experiment and
modeling makes it impossible to reach a quantitative agreement. We therefore pursue qualitative
22
and mechanistic contributions from the computational modelling efforts to interpret the
experimental observations.
2.1.2 Materials and Methods
In this present study, we investigate the grain growth mechanism due to the electrical
current flow in zirconium thin films. Zirconium is a transition metal with a hexagonal closed
pack (hcp) lattice structure with high melting point (2128 K), biocompatibility, good corrosion
and radiation resistance, making it a popular choice in nuclear, aviation and surgical implant
applications. We used physical vapor deposition (PVD) technique to deposit about 140 nm thick,
99.97% pure zirconium films on silicon-on-insulator (SOI) substrates. The as-deposited films
showed near- amorphous structure. We used SOI substrate to co-fabricate a micro-electro-
mechanical (MEMS) device with the specimen. The specimen dimension was about 100 microns
long, 5 microns wide. We used standard photo-lithography, lift-off and deep reactive ion etching
on a 100mm (i.e., 4 inch) wafer so that the actuator and heater structures were co-fabricated with
the specimen which ensures perfect specimen alignment and gripping. The chip fabrication
process started with photolithography to transfer a lift-off pattern of the device structure. The
pattern features 100 microns (μm) long and 5 μm wide dog-bone shaped specimens, electro-
thermal actuators and reference points to measure the applied strain. The SOI wafers with 20 μm
device and 2 μm buried oxide (BOX) layers allowed these features to be electrically insolated, so
the mechanical grips also worked as electrodes. The zirconium thin film was evaporated using e-
beam on the lift-off patterns. The wafer was then dry etched with deep reactive ion etching,
which realized the actuators and electrodes with vertical side walls in the device layer. The wafer
was then patterned and the entire handle layer is subsequently etched from the backside. When
the BOX layer was dry etched, all the micromachined silicon structures became freestanding. We
23
then used a physical shadow mask to isotropically etch the silicon beneath the specimen gauge
section, rendering it freestanding. Details of the device design and fabrication are given
elsewhere [91]. Figure 2.1a shows the freestanding zirconium thin film on MEMS device, where
the heavily doped silicon structures act as electrodes. The device fits in a TEM specimen holder
with electrical biasing capability. In our experiment, we passed electrical dc current below
electromigration failure limit (i.e., <106 A/cm2) through the electrodes A and B as shown in
Figure 2.1a to conduct the electrical annealing without damaging the specimen. The inset of
Figure. 2.1a shows the selected area electron diffraction (SAED) pattern of a specimen before
passing electrical current, where the completely diffused rings suggest near-amorphous (< 5 nm
grain size) microstructure without any porosity (Fig. 2.3a). MD simulation model (Figure 2.1b)
was prepared with ten grains with similar size but oriented at different angle. We also performed
electro-thermal simulation of Joule heating using COMSOL® to determine the temperature field
along the sample length (Figure 2.1c). In our in-situ TEM experiments, we passed dc current
through the specimen in a stepwise fashion. Since TEM cannot measure temperature field
directly, this information was obtained from multiphysics simulation of the specimen with actual
geometry, current density and resistance. Simulated temperature profile along the specimen at a
current density of 8.5x105 A/cm2 under vacuum condition mimicking the TEM chamber is
shown in Figure 2.1c. The highest temperature developed at the mid-section of the sample and
was about 710 K.
The effect of electrical current on grain growth mechanism in zirconium thin film was
studied using classical MD simulation conducted by LAMMPS [75] software using Embedded
atom method (EAM) [78] potential. Voronoi tessellation-based models of hcp zirconium were
built with 10 numbers of grains with an average size of 8nm. These grain sizes were chosen to
24
Figure 2.1. (a) Scanning electron microscope (SEM) micrograph of the MEMS device showing
the current flow through the specimen. Inset shows diffraction pattern at 0 A/cm2 current density,
(b) Atomistic model with grains oriented at different angles, and (c) Electro-thermal simulation
of sample with actual geometry, resistance and current density [92].
mimic the as-deposited specimen in the earlier phases of electrical annealing. In our model we
orient the grain at different angle namely 0˚, 5˚, 10˚, 15˚, 30˚and 45˚ as shown in Figure 2.1b,
whereas 0˚ angle lies along [1210] direction and [0001] direction corresponds to film normal i.e
c-axis. We checked the model for overlapping of atoms at the grain boundaries. At first, we
performed energy minimization using conjugate-gradient (CG) method followed by NPT
dynamics for several thousand steps in LAMMPS. We used Huntington-Grone [93] ballistic
model to apply equivalent EWF on each atom. The EWF on each atom is calculated using the
following equations [94]:
𝐹𝑤𝑖𝑛𝑑 = 𝑍∗ × 𝑒 × 𝑗 × 𝜌 (2.1a)
25
Where, 𝑍∗is effective valence number, e is electron charge, j is the current density and 𝜌 is the
specific resistivity of zirconium. In our present simulation, we consider 𝑍∗ as 3.4 [95] and 𝜌 as
421 nΩ.m[96]. During our simulation, we applied periodic boundary conditions in all directions.
Verlet algorithm was employed for time integration during the NPT dynamics with a time step of
0.5fs. EWF was applied on individual atom followed by energy minimization and NPT dynamics
run. We set the simulation temperature at 710 K obtained from electro-thermal simulation by
considering Joule heating effect during the current flow through the sample.
2.1.3 Results and Discussion
Figure 2.2 shows the experimentally observed grain growth during the dc current passage
through the specimen inside a Tecnai LaB6 TEM. We allowed 5 minutes delay between two
consecutive current increments. We continuously monitored grain growth and took TEM bright
field (BF) and selected area electron diffraction (SAED) to probe the grain growth. TEM BF and
associated SAED images indicating microstructural evolution are shown in Figure 2.2a-2.2c. In
our experiment, we observed very fast grain growth dynamics at a current density of 8.5x105
A/cm2 (Figure 2.2b), where the microstructural changes were discernible within few minutes.
Vigorous grain growth was observed at an accelerated current density loading of 1.1x106 A/cm2,
discernible in few seconds. Inset micrograph in Figure 2.2a and 2.2d represent TEM diffraction
patterns for the initial and final conditions in only 15 minutes time span.
We also performed thermal annealing on specimens to assess the effectiveness of
electrical current annealing. In case of thermal annealing, we had to anneal the specimen at 873
K to see any appreciable growth, which is higher than the Joule heating temperature due to the
current flow. The process was very slow, taking 8 times as much time as allowed in the current
26
Figure 2.2 In-situ TEM study indicating grain growth as a function of dc electrical current
density [92].
annealing experiment. Figure 2.3d shows electrical current annealing to produce at least one
order of magnitude larger grain size compared to thermal annealing (Fig. 2.3a). This is also
reflected by the more resolvable spots in Figure 2.3e compared to Figure 2.3b, where the
diffraction pattern of thermally annealed specimen shows only diffused ring patterns. To explain
the observed phenomena, we hypothesize that current annealing efficiently eliminates defects
and dislocations localized around the defective regions such as grain boundaries (GBs). It is well
known that electrical current annealing involves both EWF and Joule heating. Conventional
thermal annealing involves bulk heating of materials whereas electrical current induced EWF
initiates defects annihilation at the targeted locations such as GBs. Thus, thermal annealing is
more energy intensive compared to the electrical current annealing due to the heating up of the
entire sample. We also did not observe any damage in Figure 2.3d, which confirms that electrical
annealing below electro-migration failure limit (i.e., current density < 106 A/cm2) can potentially
be a time and energy efficient path towards active microstructure control.
27
We achieve a qualitative validation of our hypothesis by performing MD simulation that
indirectly captures current flow effect by imposing EWF on individual atoms. Figure 2.3c and
2.3f show MD simulation cell after thermal and electrical annealing respectively. The electrical
annealing was conducted for 25 ps followed by energy minimization and 50 ps NPT dynamics
run. While total thermal annealing takes 1.1 ns, which included first stage heating from 300K to
873K with a temperature ramp rate of 0.012 K/fs, second stage annealing by holding temperature
at 873K for 1.0ns and final stage cooling from 873K to 300 K at a cooling rate of 0.012 K/fs. We
then equilibrated the system for 100 ps at room temperatures i.e 300 K. Thermal annealing
occasionally led to grain boundary (GB) reconstruction (indicated by arrows in Figure 2.3c),
while other grains remain mostly intact. Due to the presence of defects, GBs are at higher energy
state compared to the interior crystalline regions. Thus, any external driving force such as
temperature, strain or electrical current will increase grain size by reducing the GBs area.
External stimuli such as temperature increases the kinetic energy of atoms, which also increases
atomic vibrations at the GBs. Above recrystallization temperature, the thermal driving force
minimizes the GB energy by annihilating defects at the GBs. However, thermal annealing
requires uniform heating of entire material. On contrary, electrical current loading involves both
EWF and Joule heating [97, 98]. EWF accompanied by Joule heating generates driving force that
eliminates GBs defects. Our simulation results as shown in Figure 2.3f shows defect annihilation
and high atomic diffusion due to the pronounced scattering at the GBs during electrical
annealing, which yields larger grain size.
28
Figure 2.3. Comparison between thermal and electrical current annealing: (a) TEM bright field
image after thermal loading (b) corresponding SAED pattern, (c) MD simulation cell showing
grain boundary reconstruction in limited locations indicated by arrows, (d) TEM BF image after
current loading, (e) corresponding SAED pattern, and (f) MD simulation showing grain
boundary reconstruction due to the electrical current loading [92].
Atomistic simulation qualitatively validates our hypothesis. For example, it is evident
from Figure 2.3f that atomic re-orientation and diffusion are dominant at the GBs during the
electrical current flow. Merging of the GBs are seen to be a result of diffusional motion of the
defects under the impetus of the EWF. GBs experience localized stress field as shown in Figure
2.4a, which could be attributed to the local disorder in atomic position, orientation and defect
density at GBs. The localized stresses in the GB regions also indicate higher potential energy
29
states compared to the grain interior atoms. The mechanical stress field around the defects is
another reason behind the localized and targeted enhancement of atomic mobility at the GBs.
Figure 2.4b shows two triple points with an initial orientation at 10°, 15°, 30° and 45° as
indicated by arrows. After electrical annealing, we observed disappearance of GBs as shown in
Figure 2.4c. The reconstruction of original hcp-crystalline structure of zirconium from different
grain sites (as shown in Figure 2.4c) clearly indicates that the grains grow in zirconium thin film
due to the synergy of EWF and Joule heating at the grain boundaries. We also noticed inter
granular diffusion, which provides an evidence of higher mobility of atom due to the pronounced
electron scattering at the GBs. Additionally, we also noticed that all the GBs are oriented along
the same direction (~0°) as shown by arrows in Figure 2.4c after electrical annealing. These
evidence indicate that electrical current could significantly eliminate defects and increase the
grain size in thin films.
Figure 2.4. Time evolution of grain growth obtained from MD simulation trajectory: (a) initial
structure, (b) two triple points before electrical annealing, and (c) two triple points after electrical
annealing [92].
2.1.4 Conclusion
To summarize, we have performed in-situ TEM experiments on near-amorphous
30
zirconium thin films to gain fundamental understanding of external stimuli such as electrical
current density effects on grain growth. We noticed almost two orders of magnitude grain growth
in less than 15 minutes at a current density of 8.5x105 A/cm2. This was about one order of
magnitude higher than what we observed in similar specimens but under convention thermal
annealing at 873 K for 120 minutes. MD simulation results show that the effect of EWF is to
impart very high atomic mobility that is localized to the defects and GBs, which makes the
electrical annealing more energy efficient compared to the conventional thermal annealing where
the crystalline grain interiors are heated to the same temperature as the defective areas. Localized
stress fields around the defects also increase the atomic mobility under EWF thus increase grain
size by annihilating defects at the GBs. The findings of this study may play vital role in
developing novel energy and time efficient techniques for active microstructural reorganization
and control in the near future.
2.2 Multi-Stimuli (Electron Wind Force and Mechanical Strain) Effects in Thin Films
In this section, we will investigate low temperature grain growth using a synergy of
electron wind force (EWF) and mechanical strain. It is well known that thermal annealing is
commonly used for defect elimination and grain growth in polycrystalline materials. Here, we
propose an alternate route through a synergy of electrical current and tensile strain at or near
room temperature. Our experiments involve flow of electrical current below the electromigration
limit in 100 nm thick freestanding palladium films with approximately 5 nm initial grain size.
Electro-thermal simulation shows that Joule heating increases the temperature up to 470 K
(homologous temperature of 0.25) at the middle section of the specimen, whereas, the massive
heat sinks at the two ends of the specimen constrain them to remain at room temperature. In-situ
transmission electron microscopy (in-situ TEM) study shows that more than two orders of
31
magnitude grain growth in the high temperature regions (middle section of the sample) and little
growth at the room temperature regions. However, application of elastic strain (about 0.1%)
dramatically altered the scenario and increased the grain size by more than 10 times in a few
seconds near the room temperature regions. Our finding indicates that the multi-stimuli synergy
of elastic strain energy and electrical current density may achieve grain growth in metallic
materials even at room temperature. Molecular dynamics (MD) simulation of this phenomenon
reveals that the externally applied strain is localized at the grain boundaries (GBs) in
nanocrystalline metals, which promotes the effects of electron wind force on the GB’s atoms.
Thus, we conclude that synergy of two or more stimuli can achieve grain growth at room or even
lower temperatures.
2.2.1 Objective and Motivation
In this section, we wish to study the role of externally applied strain in microstructural
control. For example, grain size is considered to be one of the most influential parameters
controlling mechanical properties of polycrystalline materials [5]. Due to the high fraction of
grain boundaries (GBs) nanocrystalline materials exhibit pronounced grain size effects as
manifested through the breakdown of the classical Hall-Petch relation [6]. Grain size also
influences electrical, thermal and optical properties. [7] [8] [21]. It is well known that GB effects
are usually stronger than those of line or point defects. Additionally, elimination of residual
stress in a sample is critical for both performance and reliability perspective of interconnects
[35]. Therefore, high temperature annealing is a common post-processing step in fabrication.
Grain size and residual stress in interconnects depend on parameters related to the
deposition process, temperature, pressure, substrate etc., and are difficult to control during
32
deposition. Additional parameters such as intrinsic stress, micro-texture, defect density and
surface roughness need to be considered. To overcome these issues, thermal annealing is widely
used, where temperature is raised (> 0.3 homologous temperature) to promote atomic diffusion
or defect mobility. This thermal stimulus activates the microstructural evolution process that is
governed by the minimization of surface, interface and mechanical strain energy [99]. Grain
growth is a mechanism for such energy minimization and is not a monotonic function of
temperature. Rather, the initial grain size and texture, residual stress, annealing environment and
film thickness are vital to grain growth. However, for complex multi-layer systems it is not
possible to raise temperature arbitrarily without introducing thermal stresses.
Thus, motivation for this study comes from the above-mentioned challenges in grain size
and residual stress control. In particular, we propose a non-thermal route towards grain size
control using multi-stimuli such as electron wind force (EWF), temperature and mechanical
strain. Our basic premise is that, acceleration of grain growth kinetics may be achievable even at
lower temperatures with the synergy of stimuli such as electrical current and mechanical strain.
The effect of electrical current is to (a) raise the temperature (i.e., Joule heating) and (b) apply
EWF at the defective areas (such as GBs and triple junction) due to the transfer of the electron
momentum. EWF creates directed-diffusion of atoms that can take different paths, such as grain
interior, boundary and external surface [100], which decreases atomic migration energy barrier
[101], and facilitates atomic rearrangement for energy minimization. Since our goal is to develop
non-thermal microstructural processing, a unique feature of this study is to remove the heat
generated by current in order to separate the effects of temperature. We achieved this through
gripping freestanding thin film specimen with massive heatsinks. Thus, when we pass current,
the center region remains heated, while the edge regions are constrained to the ambient
33
temperature. By doing this we decouple the effect of Joule heating and the EWF in the same
specimen.
Without temperature effects, it is expected that the electron wind force may not have the
enough atomic mobility needed for annealing or grain growth. We therefore attempted to utilize
mechanical strain as a promoting stimulus. For nanocrystalline metals stress assisted grain
growth is observed due to the higher GB mobility and low activation energy [102]. These have
been also reported by literatures at temperatures as low as ambient [103-105]. However, stress-
assisted grain growth requires relatively high levels of stresses and dependent on types of
material. Nevertheless, the focus of this study is the synergy of mechanical strain and EWF,
which we hypothesize to have the potential to provide the atomic mobility even at lower
temperatures.
2.2.2 Materials and Methods
To investigate the synergy of electric current and mechanical strain on grain growth
kinetics we choose 100nm thick palladium films. Palladium melts at 1555 ºC and is used in
catalysis [106] and electronics applications [107]. The experimental setup consists of 5mm x
3mm silicon-on-insulator (SOI) chips, each containing 100 nm thick palladium film specimens
which are integrated with electrodes, micro-heaters and mechanical actuators as mentioned in
section 2.1.2. Figure 2.5a shows a zoomed-in view of the chip, which is designed to fit in an in-
situ transmission electron microscope (TEM) specimen holder. We also develop molecular
dynamics (MD) models mimicking the in-situ TEM experiments to understand the fundamental
mechanisms behind the multi-stimuli annealing process qualitatively.
34
Figure 2.5a shows a TEM image and diffraction pattern of the as-deposited specimen.
Figure 2.5b shows a COMSOL® multi-physics model for a simulation of the temperature
distribution at current density of 7x105 A/cm2. The relatively massive silicon grips act as heat
sinks, bringing down the temperature to the ambient. This unique boundary condition allows us
to isolate the temperature effects from EWF.
Figure 2.5. In-situ TEM experimental and MD simulation setups: (a) Scanning electron
micrograph of the adopted MEMS device with actuators and electrodes including a TEM bright
field (BF) image and diffraction pattern of as-deposited specimen, (b) temperature profile along
the length of the sample obtained from electro-thermal simulation mimicking the actual
experimental conditions, and (c) atomistic simulation cell with randomly oriented grains used to
mimic the experiments [108].
35
The in-situ TEM experiments were performed inside a FEITM Talos F200X
scanning/transmission electron microscope (S/TEM) with a resolution of 0.12 nm using field
emission gun and 200 kV acceleration voltage. In-situ TEM enables real-time high-resolution
imaging as well as selected area electron diffraction (SAED). In a typical experiment, we first
bias the specimen while measuring the current. About 5 minutes of time gap is allowed between
the current increments to discern the ensuing microstructural changes. We increased the bias as
soon as the relative magnitude of grain growth per current increment decreases. At this point, we
activated the thermal actuators to apply a strain on the palladium thin film. The resulting strain is
measured by comparing the extension of the fixed end of the specimen with respect to the
moving end. We observe the two ends as well as the middle section of the specimen during the
straining.
We prepared MD simulation (Figure 2.5c) cell with 22 randomly oriented grains with
approximate grain size of 5nm. We performed electro-thermal simulations of electric current
flow and subsequent temperature field due to Joule heating using COMSOL®. During this multi-
physics simulation, we considered actual geometry, resistance and current density of the sample
and later on, we fed temperature in the MD simulation model. Figure 2.5b shows the temperature
profile along the specimen at a current density of 7x105 A/cm2 mimicking the TEM chamber.
The highest temperature 470 K, which is < 0.25Tm (where Tm is 1828 K for palladium) was
observed at the middle section of the sample. Additionally, we employed classical MD
simulation to investigate electrical current induced grain growth mechanism in palladium thin
film using LAMMPS [75] package. In our present simulation, we chose a time step of 0.5 fs.
Voronoi tessellation-based simulation cells of face-centered cubic (fcc) Palladium were built as
36
shown in Figure 2.5c. This grain size was chosen to mimic the as-deposited specimen as well as
grain size distribution in the earlier phases of electrical annealing.
During our simulation, periodic boundary condition was applied along the longitudinal
direction (x-direction) while keeping free surface boundary conditions along transverse and film
normal directions (y and z directions respectively). Energy minimization was carried out using
conjugate-gradient (CG) method followed by NVT dynamics at 470K for several thousand steps.
EWF was applied on individual atoms followed by energy minimization and NVT dynamics run.
To describe the interatomic forces among atoms in a polycrystalline palladium thin film we
employ the embedded atom method (EAM) [79] potential. In a classical MD simulation it is
difficult to directly employ electron effects for a large system, thus we first quantify the
equivalent wind force on each atom using the Huntington-Grone [93] model using the following
equations [94]:
𝐹𝑤𝑖𝑛𝑑 = 𝑍∗ × 𝑒 × 𝑗 × 𝜌 (2.2a)
where 𝑍∗is the effective valence number, e the electron charge, j the current density and 𝜌 the
specific resistivity of palladium. In our present simulation, we consider 𝑍∗ as -5.1 [95]. Thus
total force on an individual atom during the electrical current flow can be written as follows:
𝐹𝑇𝑜𝑡𝑎𝑙 = 𝐹𝐸𝐴𝑀 + 𝐹𝑤𝑖𝑛𝑑 (2.2b)
In order to mimic the experiment and investigate the effect of strain on grain growth
under EWF, we apply tensile strain on the palladium polycrystalline simulation cell along its
longitudinal direction (i.e. the x-direction) at a strain rate of 5×108𝑠−1. Visualization of the grain
37
growth was conducted on OVITO [109] software. During the simulation, we also monitor
potential energy and stress distribution on individual atoms.
2.2.3 Results and Discussion
Figure 2.6a shows TEM bright-field image of a 100 nm thick physical vapor deposited
(PVD) palladium films. SAED pattern indicates that as-deposited specimens initially had a near-
amorphous (<5 nm grain size) structure (Figure 2.6b). To investigate electron beam irradiation
effect, we kept the sample for prolonged period (i.e., 20 hours) under the electron beam and we
did not observe any discernible grain growth. Apparently, electron beam has no/insignificant
effect on grain growth or microstructural changes as shown in Figure 2.6a. Experimentation
outside TEM (i.e ex-situ) also yields similar order of magnitude grain growth, which indicates e-
beam does not have any significant effect on grain growth.
Figure 2.6. (a) TEM bright field (BF) image of the as-deposited specimen after prolonged
exposure to the electron beam, and (b) corresponding SAED pattern [108].
38
Figure 2.7 shows the experimental results on electrical current induced grain growth at
the middle section of the palladium specimen. We noticed grain growth is a non-linear function
of current density and we did not increase current density beyond 7 × 105 𝐴/𝑐𝑚2 to avoid any
electromigration damage. Theory indicates that Joule heating is a nonlinear function of current
density whereas EWF is a linear function of current density. Thus, 7% increment of current
density can increase the Joule heating by 15% as shown in Figure 2.7b and 2.7c. After a current
density of 6.5x105 A/cm2, nonlinear increment of Joule heating accompanied by electron wind
force induced significant grain growth as shown in Figure 2.7b and 2.7c. The average grain size
after electrical annealing was 550 nm, which is more than two order of magnitude higher than
the as-deposited grain size. Grain growth is mostly uniform, however the SAED pattern in Figure
2.7f indicates significant change in texture. Electrical current induced grain growth kinetics is an
order of magnitude faster than thermal annealing, and takes only 5 minutes at each current
values.
It is important to note that electrical current effect in conductive materials has been
consistently demonstrated to change microstructure of metallic materials. Thus, electrical current
accompanied by temperature field is known to play a dominant role in enhancing atomic
mobility. In the previous section, we have argued that the role of EWF is pronounced when the
specimen has appreciable volume fraction of defects [110] (such as grain boundaries), where
momentum transfer is immensely amplified. Thus, the decoupling of thermal contribution will
allow us to access fundamental understanding of electrical flow related phenomena in metals
[111]. In this respect, configuration of our freestanding thin film specimen allows us to study the
role of temperature (in the middle section) and electron wind (at the edges, connected to the grips
that act as massive heat sinks) separately in the same specimen. The electro-thermal simulation
39
as shown in Figure 2.5b shows the temperature profile along the specimen length due to the
Joule heating during the electrical current flow. From the TEM BF it is clear that we have very
large grain growth at the middle region of the sample, while very little effect is observed due to
the near room temperature at the edges.
Figure 2.7. In-situ TEM BF and SAED evidence of grain growth in the specimen center region at
(a, d) 0 x105 A/cm2, (b, e) 6.5x105 A/cm2, and (c, f) 7x105 A/cm2 current densities [108].
Figure 2.8a shows the microstructure at the anode end of the specimen, where the current
density is 6.7x105 A/cm2, but the temperature remains at the ambient. After annealing, the
average grain size is about 10 nm. This is significantly lower compared to the center region,
suggesting that the EWF alone cannot induce significant amount of grain growth at room
temperature. This observation motivates us to study the synergy of mechanical strain and
40
electrical current on grain growth at room temperature. While it is well known that surface and
strain energies are primary driving force for the annealing and recrystallization processes.
However, most of the studies in the literature involve very large, plastic strain [112] in the form
of cold work or field-assisted sintering. Here, we explore the effect of strain energy in the elastic
regime, which is yet to be studied. We hypothesize a synergy between elastic strain and EWF
[113] might enhance grain growth. Experimental finding on synergy study is shown in Figure
2.8b, where even as small as 0.1% mechanical strain exhibits profound impact on the grain size
at room temperature.
Figure 2.8. Specimen microstructure at 7x105 A/cm2 current density and room temperature, (a)
before, and (b) after application of 0.1% strain [108].
To study the effect of elastic strain energy, we applied electrical bias on the on-chip
actuator pad. To stay in the elastic region, we keep the applied strain in the specimen at about
0.1%. Strain was measured by tracking the moving end of the specimen with respect to a fixed
reference frame on MEMs device. While tracking a region near the edge (and hence around the
41
room temperature), we applied this strain in a single step. We almost instantaneously observed a
dramatic increase in the grain size as shown in Figure 2.8. The significant (>10x) increase in
grain growth suggests that even at very low temperature (i.e., room temperature), there exists a
strong synergistic effect of elastic strain and electrical current density in promoting grain growth
dynamics. Stress induced grain growth at high deformation have been reported [114, 115].
However, in our present study, instead of applying high stress, we coupled lower strain (in the
elastic regime) and EWF. This coupling yields significant amount of grain growth even at room
temperature. This grain growth is attributed to the pronounced interaction between multi-stimuli
(strain and current density) and pre-existing high-volume fraction of defects at the GBs. Defects
(which could be an outcome of plastic deformation) are required for recrystallization. Due to the
high-volume fraction of GBs in nanocrystalline materials, they contain large number of defects
and stored elastic energy in terms of defects and disorders compared to their bulk
counterpart. Additionally, in a nanocrystalline material stresses are highly localized at the grain
boundaries (GBs). Due to this localized stress, a small amount of applied strain could amplify its
effects at the GBs, which indicates lower external strain could specifically targets the grain
boundary atoms to enhance atomic mobility. Due to the pronounced scattering of conducting
electrons at the GBs, EWF is also higher at the GBs. Thus, pronounced effects of strain and
electrical current on the GBs atoms could recrystallize the nanocrystalline materials even at
lower strain and temperature.
The literature indicates that plastic strain energy is the core requirement for nucleation of
recrystallization. For example, high density of dislocations in a sub-grain structure possess
higher plastic strain energy compared to the lattice. Therefore, a sub-grain is the first region to
start recrystallization. Therefore, there is no evidence in the literature that elastic strain energy
42
could drive grain growth. We propose a conceptual model and study its effectiveness using a MD
simulation model. Here, the fundamental premise is that the grain boundaries are not just high-
volume fraction of surface defects in the nanocrystalline thin films; they also experience
localized, higher mechanical stress compared to the grain interior. The effect of EWF is
pronounced at the grain boundaries, where the electrons impart their momentum to enhance the
mobility of the atoms. However, an additional driving force is still required for the atoms to
migrate towards minimizing their energy at low temperature. We propose that small amount of
external elastic strain energy is sufficient and the synergy can be strong enough to induce
significant grain growth at room temperature. It is important to note that while the strain energy
is crucial for the process, it has little or no effect without EWF. This indicates that at room
temperature single stimulus might not induce significant atomic mobility for the grain growth.
To comprehend the synergy of electrical current and strain on grain growth mechanism in
palladium thin films, we performed MD simulation as shown in Figure 2.9. Figure 2.9 shows
interior crystalline region of individual grains by hiding the GB atoms. Figure 2.9b and 2.9c
show 1nm wide fixed electrode area at the two edges colored by red bands. As mentioned
earlier, initial sample contains randomly oriented 22 grains with an average grain size of 5nm
prior to the annealing. After the first stage annealing, number of grains reduces to 11 due to the
grain growth. Once we apply the strain and anneal the sample, the grains grow further, and the
total number of grains reduces to 7. GBs contain defective atoms that are at a higher energy state
compared to their interior crystalline regions. Due to the high energy state of GBs, external
stimuli such as electrical current, temperature or mechanical strain could potentially annihilate
some of these defects and increase the grain size.
43
Figure 2.9. MD simulation model on grain growth: (a) initial structure, (b) electrical annealing
prior to the application of strain, and (c) electrical annealing after strain application, (Green color
indicates face centered cubic (FCC) and red color indicates hexagonal closed packed (HCP)
phase of palladium) [108].
44
Recent studies show that [97, 116] electrical current loading could annihilate GBs’
defects efficiently and thus lead to the grain growth. Electrical current which introduces both
EWF and Joule heating [97, 98] could create pronounced electron scattering at the GBs’
defective sites which further reduces GBs’ energy by annihilating randomness at the GBs. As
mentioned earlier, EWF is effective at the GBs, where they could exchange their momentum
with disordered atoms. Due to the imparted EWF and Joule heating effects we notice significant
amount of grain growth compared to the pre-strained sample. The localized stresses around the
grain boundary regions are also an indicator to the higher energy states of the system. The initial
higher stress around the GBs enhances the atomic mobility during the electrical current flow and
allows defective atoms to relax and rearrange at the GBs.
Figure 2.10. Computational study: (a) Strain energy increment during tensile straining of the
system, and (b) grain size as a function of applied tensile strain [108].
Figure 2.10a shows the strain energy increment in the sample during tensile loading of
the sample. However, in our present simulation we applied 1% strain (0.01 nm/nm; Figure 2.10a)
as shown by dotted blue color vertical line in Figure 2.10a. This small amount of strain confirms
45
that the applied strain is within the elastic regime. Due to the time and length scale effects in MD
simulation we applied higher strain (1%) compared to the experimental strain (0.1%). With
several thousand of atoms, small time step and complex forcefield, MD simulations could be
highly time intensive. Thus, we adopted an exaggerated computational strain to capture the
physical phenomenon within affordable computational time. Study shows that elastic strain
contributes ~1meV strain energy per atom into the system. Figure 2.10b shows the synergistic
effect of the electron wind force and mechanical strain on grain growth, and represents the grain
size as a function of applied tensile strain (for the given experimental value of applied EWF). We
notice grain size increment with the applied strain accompanied by EWF. Thus, our study
indicates there exists a strong synergy between EWF and mechanical strain even at room
temperature.
2.2.4 Conclusion
Electrical current induced phenomena such as electro-plasticity and electro-pulsing are
receiving increasing attention due to the rapidness in materials processing compared to the
conventional heat treatment processing. Prior to the commercialization, the fundamentals of the
contributions from the current, temperature and strain must be understood. Unfortunately,
decoupling of these effects is often arduous. To overcome this issue, we developed a unique
experimental setup, where the same specimen could be divided into two distinct regions. The
center region experiences a high temperature while the two edges of the freestanding thin film
specimens are subject to the same current density, but remain near room temperature. In this
case, insignificant amount of grain growth near the edges of the sample suggests the dominant
role of temperature on grain boundary mobility.
46
Present study reports the experimental evidence of multi-stimuli synergy i.e., mechanical
strain and EWF effects on grain growth even at lower temperature. It is well known that surface
and strain energy can act as driving force for grain growth. However, very little is known about
elastic strain driven grain growth. This is because fundamental electro-plasticity or dynamic
electropulsing studies introduce significant amount of mechanical strain during the passage of
current. On the other hand, in this study we propose a new direction where a relatively small
(0.1%) elastic strain is applied on the specimen in conjunction with EWF, resulting in
remarkable grain growth. Interestingly, the EWF manifests at the grain boundaries due to the
higher localized stress at the GBs. The experimental evidence as well as the computational
modeling indicates that the synergy of EWF and mechanical strain can be exploited to achieve
grain growth at near room temperatures.
2.3 Low Temperature Processing of Additive Manufactured Ti64 Alloy
In this section, we will examine scaling up feasibility of electron wind force (EWF)
induced low temperature processing. To accomplish this goal we passed electrical current
through the additive manufactured Ti64 alloy to annihilate defects and reduction of residual
stress without sacrificing mechanical properties. Our results show that both grain size and nano
hardness increased by 15% and 16% respectively after EWF processing. This is attributed to the
pronounced dislocation interactions as well as defect healing during low temperature processing.
Reduction in the residual strain and an increase in the intrinsic strength obtained from electron
back-scattered diffraction (EBSD) study corroborate the effectiveness of the EWF annealing at
microscale sample.
47
2.3.1 Objective and Motivation
Microstructures in an additive manufacturing (AM) process are governed by a complex
combination of processing parameters, which include the scan rate of input energy source, power
of the energy source, deposition rate, and the dwell time between individual layers [1,2]. Thus,
mechanical properties [3,4] of AM parts show large scatter and inconsistency due to the
numerous process parameters, which further results into a wide range of microstructures. Among
many other materials, titanium alloys especially Ti-6Al-4V (i.e., Ti64) alloy has received
considerable interest among material science community due to their applications in aerospace
and biomedical industries [15-17]. Commercially available Ti64 alloy possesses high corrosion
resistance and high specific strength [18]. Owing to the superior properties such as corrosion
resistance, tensile strength, elongation, Young's modulus and fatigue properties of commercially
available Ti64 alloy; AM fabricated Ti64 alloy contains different types of internal defects
[19,20] which negatively affect their properties. Typical AM fabricated Ti64 materials contains
columnar prior -grains that are aligned with the build direction [5-7]. Within these columnar
prior -grains, a mixture of colony -laths, ʹ-martensite, and retained phases is present [5-7].
Both the size and fraction/proportion of the -laths, ʹ-martensite, and size of grains determine
the mechanical properties. Aside from these microstructural parameters, internal macro- and
microscopic defects have also shown to significantly affect the mechanical properties.
Microscopic defects include micropores, residual/transformation stresses, and microstructural
defects such as grain boundary (GB), cracking and non-equilibrium microstructures due to
intrinsically high-solidification rate of the process [8,9]. Likewise, the spatial distribution of
these defects creates unwanted anisotropy in tensile properties [21,22]. Thus, in order to ensure
high quality of AM Ti64 alloy, microstructural defects further need to be reduced.
48
To overcome this issue post-processing methods such as hot-isostatic pressing (HIP) or
heat treatment are often employed to reduce defects and residual stress in AM parts. However,
HIP and other heat treatment operations often result in inferior mechanical properties. For
example, porosity is a common defect in metal AM parts and currently, the most acceptable post-
processing technique to mitigate its effects is to apply hot-isostatic pressing (HIP). HIP is often
applied to AM structures in order to cure process-related voids and pores within the builds.
However, the application of this post-processing treatment has been shown to coarsen the
microstructure and decrease the strength of AM materials [12]. Heat-treatment is another post-
processing method that is used for stress relieving and microstructural modification/enhancement
of AM parts [2,13,14]. However, high temperature annealing that is typically used for optimizing
the microstructure of AM parts results in mechanical properties degradation due to thermally
activated processes such as recovery, recrystallization, and grain growth.
Unlike conventional heat treatment or HIP, in this section, we propose a novel technique
to eliminate microstructural defects at room temperature by employing electric current annealing.
This method uses EWF to promote an enhanced atomic mobility near the defect cores to
annihilate microstructural defects without compromising the mechanical properties. The
investigation was performed on powder bed fusion additive manufactured Ti64 alloy using
optical microscopy, electron backscatter diffraction (EBSD), and nanoindentation
characterization techniques to assess the effectiveness of the method.
2.3.2 Materials and Methods
The Ti64 sample was manufactured at Penn State’s Center for Innovation Material
Processing thru Direct Digital Deposition (CIMP-3D). The samples were fabricated using an
49
EOS M280 Laser-Powder Bed Fusion machine. The material feedstock for the build was EOS
Titanium Ti64 (which conforms to ASTM B348 Grade 23 powder material specifications), with
particle size ranging between 15 μm to 45 μm. The maximum current density during the low
temperature processing was 5 x105 A/cm2. A custom stage with electrical biasing capability was
built to perform this experiment (as shown in Figure 2.11). To ensure the sample remains at low
temperature processing, we passed liquid nitrogen (N2) through the stage attached with the
sample as shown in Figure 2.11. We controlled the sample temperature at 20°C. During the
experiment, we gradually increased the current by 0.1A per step while continuously monitored
the resistance and microstructure of the sample. Each biasing step was held for 5 minutes. The
sample mounted inside scanning electron microscope (SEM) was checked after every current
step to observe any discernible microstructural change.
Figure 2.11. Schematic showing experimental set-up with temperature controlled stage.
Mechanical polishing followed by ion milling was employed to prepare the sample for in-
situ EBSD study to characterize microstructures. Sample was polished with both SiC sandpaper
up to a P2400 grit size, and with 3 µm diamond slurry. Final polishing was done on a vibratory
50
polisher for at least 2 hours with a chemical-mechanical polishing slurry consisting of 0.05 µm
alumina-silica. Additionally, etching was performed using micro-etchant solution consisting of
1.5 mL HF, 4 mL KNO3, and 94 mL of H2O for optical microscope imaging. Average Grain size
was calculated using ASTM E112–13, Standard Test Methods. We acquired EBSD data using
Helios 661 NanoLab FEI dual beam field emission scanning electron microscope (FE SEM)
equipped with Aztec EBSD Nanoanalysis software, ver. 4.2 (Oxford Instruments). Scanning was
conducted at a step size of 0.1 μm, and a minimum of 3 randomly selected scans were acquired
for each specimen. All EBSD analyses were performed using CHANNEL 5 software. Grain and
-lath boundaries were marked using a threshold of 10°. We also performed nanoindentation
experiment of the as-received and electric current processed specimens using Bruker
Nanoindenter with the Berkovich diamond tip [23] based on the Oliver and Pharr approach [24]
and ISO 14577 standard for measurements.
2.3.3. Results and Discussions
It is well known that during passage of electrical current through a conductive material, it
generates both resistive heating also known as Joule heating and electron wind force (EWF)
effect. As mentioned earlier, electrical current density effect is considered to have negative
impact on materials specifically above a threshold value (i.e., 106 A/cm2) [117, 118]. Thus,
during electrical current annealing one needs to take precaution regarding electrical current
density. In our present study, we keep the current density below this threshold value and actively
cool the sample to investigate the effects of EWF at low temperature. Recently this technique has
been shown to yield comparable microstructure and properties to that of conventional thermal
annealing at a very short duration and a relatively low temperature [34-36]. This has been
attributed largely to a non-thermal effect, rather than the thermal component (Joule heating).
51
EWF induced momentum transfer promotes intensive vacancy diffusion, atomic flux mobility at
the defects and grain boundaries [37,38]. This enhanced atomic mobility has the potential to
annihilate defects. Unlike previous sections, in the following section we will investigate EWF
effects on micro-scale sample to validate the scaling up feasibility of the low temperature EWF
processing.
In our previous section we have calculated EWF using Huntingdon-Grone model [93].
However, EWF can be also calculated form quantum dislocation theory [119]. Based on
quantum dislocation theory EWF develops due to the interactions between drifting free electrons
and defects such as dislocations. From quantum dislocation mechanics, the magnitude of the
EWF depends on the orientation between the Burgers vector and the direction of the current
vector. In addition, based on the principle of virtual work, the direction of the electron wind
force should be normal to the shear slip vector at the point of interest. Thus, considering specific
electrical resistivity due to dislocations, the electron wind force per unit length of dislocation can
be simplified as follows:
Fwind =ρd
Ndejne (2.3.1)
Where, j is the current density vector, ρd is electrical resistivity, Nd is the mobile dislocation
density, e is electron charge, and ne is electron concentration. Titanium has a density
of 4.50 g/cm3, and an atomic weight of 47.867 g/mol, electron density/concentration is 𝑛𝑒 =
2.265 × 1029 m−3, effective valence number Z*=4, electrical resistivity of 1.78μΩ.m [120], and
dislocation density is of the order of 1014 m-2 [121]. Thus, calculated EWF is 3.23N/m. On the
other hand, if we considers slip in the basal and prism systems, these systems have <a>
dislocations with a Burgers vector 𝒃 =1
3< 1120 >, and magnitude of |𝒃| = 𝑎, where a is the
52
lattice parameter for hcp Ti64, and c/a=1.589. Then we could estimate corresponding critical
shear drag force required per unit length of dislocation using the following equations:
𝐹𝐶𝑆𝐹 = 𝜏𝐶𝑅𝑆𝑆𝑏 (2.3.2)
Where, 𝜏𝐶𝑅𝑆𝑆 is critical resolved shear stress for Ti64 is 300~400MPa [122], and b is burger
vector. Thus, calculated required force to active dislocation is 0.118N/m, which is one order of
magnitude lower compared to EWF generated due to electron wind. Thus, we could conclude
that even in the absence of thermal field EWF alone can activate dislocations in the specimen.
Figure 2.12 represents the -lath microstructures of the as-built and electric current
processed specimens. The average calculated grain sizes were 143.9 ± 41.4 µm and 165.8 ± 33.4
µm, for the as-built and electric current processed specimens, respectively. EWF processed
sample shows 15% grain size increment compared to the as-received sample. In our present
study, we also performed EBSD to assess microstructural changes. To investigate the mechanical
properties, we performed a detail analysis of the Taylor factor/Schmid factors on specimens
Figure 2.12. Optical micrographs of Ti64 specimens in the (a) as-built, and (b) Low temperature
EWF processed conditions.
53
before and after low temperature processing. Among three primary slip systems, there is a
consensus that prismatic slip is easier to activate compared to basal and pyramidal slip
[25,27,28]. Pyramidal slip system is very resistant to slip, and its critical resolved shear stress
(CRSS) is reported to be about 7 times higher than that of basal slip system [25,29]. It can also
be argued that he <c + a> slips have Burgers vectors that are significantly longer than for the <a>
slips. Thus, in our present study we excluded pyramidal slip system analysis from our study.
Figure 2.13 shows calculated Schmid factor maps for basal slip system of pre- and post-
processed sample. It is apparent that both microstructures exhibit a basketweave -lath structure.
The widths of the -laths were determined to be 0.46 ± 0.41 µm and 0.53 ± 0.47 µm for the as-
received and EWF processed specimens, respectively. Interestingly, we notice same magnitude
of increment in -lath structure width and grain size i.e., 15%. The Schmid factor maps for basal
slip system show a distinctive change in Schmid factors from high in the as-built specimen to
low in electric current processed specimen. The corresponding Taylor factors are presented in
Figure 2.13c, and it exhibits higher values of Taylor factors for electric current processed
specimen compared to the as-built specimen. Figures 2.13d and 2.13e show TEM micrographs of
the microstructure of Ti64 alloy in the as-built condition and after electric current processing,
respectively. We have compared the same area of interest before and after the application of
electric current as shown in Figures. 2.13d and 2.13e by cyan color dotted circle. As-built
specimen does not contain any twins as shown in Figure. 2.13d. However, upon the application
of the electric current we noticed twin formation due to the electron wind force (EWF) effects as
shown by yellow color arrow in Figure. 2.13e. This twining formation might contribute to the
enhancement of the hardness of the sample after annealing.
54
Figure 2.13. Basal Schmid factor maps for the Ti64 specimens in the (a) as-built, and (b) electric
current processed sample, (c) Calculated Taylor factors; Twining during electrical annealing: (d)
as-built specimen and (e) after applying a current density of 5x103 A/cm2.
In this study, we have also analyzed residual strain in the pre-, and post processed sample
through measurement of local variations/spread in lattice orientations or Kernel Average
Misorientation (KAM) [53,54]. This analysis is done for every pixel in a kernel with a
predefined threshold value of 5°. The threshold is used to exclude well-defined grain boundaries
in the analysis. Thus, KAM study allows us to identify with high residual strain [53,54]. The
higher the residual strain, the higher the KAM value, and vice versa for lower residual strain.
Figure 2.14 clearly indicates significant reduction of residual strain in a post-processed sample.
55
It is well known that AM parts inherently possess significant residual stress due to the highly
non-equilibrium process. Hence, our study suggests that EWF processing has the ability to
eliminate this residual stress from AM parts.
Figure 2.14. KAM maps for the Ti64 specimens in the (a) as-built and (b) electric current
processed conditions. A threshold of 5° was used to exclude well-defined grain boundaries in the
analysis.
To assess the effectiveness of the low temperature processing we measured mechanical
properties of pre- and post-processed samples using nanoindentation test as shown in Figure
2.15. We notice 16% increment in hardness value i.e., from about 5 GPa to ~6 GPa after
processing. While harness value increased significantly, we did not notice any significant change
in elastic modulus calculated from Oliver and Pharr approach [24] as shown in Figure 2.15b.
Higher values of Taylor factors for electric current processed specimen compared to the as-built
specimen (as shown in Figure 2.13c) imply that the slip systems in the electric current processed
specimen provide more resistance to yielding than the slip systems in the as-built specimen,
which further reflects in their hardness value. Thus, we can postulate that the high EWF causes
local reorientation of the microstructure (-laths) and prior β grain that are harder or more
resistance to yield. This study corroborates the scaling up feasibility of EWF processing at low
56
temperature even without any loss in mechanical properties. In our present study, we have
annealed samples with a thickness of up to 100um. However, it is possible to scale up this low
temperature processing technique even for bulk-scale sample with a high-power supply capable
to deliver high output current, and we left this work for future study.
Figure 2.15. (a) Force-displacement plot obtained from nanoindentation experiment, (b)
calculated hardness and Young’s modulus.
2.3.4 Conclusion
In this study, we have presented a novel and energy efficient method of defects
elimination by electric current processing at low temperature without compromising mechanical
properties. Our study shows that electric current induced EWF can enhance mechanical
properties by reducing or eliminating defects from the sample. Investigation shows that under the
EWF defective atoms might have sufficient mobility to cause local reorientation, hence can
exhibit more resistance to deform. Due to this effect, higher external force is required to initiate
plastic deformation in the electric current processed specimen compared to the as-built specimen.
Additionally, study also shows that electric current processing reduces residual strain in the as-
57
built sample. All of these observations corroborate the scaling up feasibility of EWF processing
at low temperature without sacrificing mechanical properties.
58
Chapter 3
External Stimuli Effects on Two-dimensional (2D) Materials and Devices
Contents of this chapter are based on the following journal articles:
Zahabul Islam, Kehao Zhang, Joshua Robinson and Aman Haque, Quality enhancement of low
temperature metal organic chemical vapor deposited MoS2: an experimental and computational
investigation, Nanotechnology, Volume 30, Number 39, Pages 395402 (9pp), 2019.
Author of this dissertation designed the experiment, performed the sample transfer,
device fabrication, experimentation, data analysis and manuscript writing. Kehao Zhang,
Joshua Robinson synthesized materials and involved in manuscript writing. Aman Haque
guided on experiment design, and involved in data analysis as well as manuscript
preparation.
Zahabul Islam, Azim Kozhakhmetov, Joshua Robinson, Aman Haque, Enhancement of WSe2
FET Performance Using Low-Temperature Annealing. Journal of Electronic Materials 49, Pages
3770–3779 (2020).
Author of this dissertation designed the experiment, performed the sample transfer,
device fabrication, experimentation, data analysis and manuscript writing. Azim
Kozhakhmetov, Joshua Robinson synthesized materials and involved in manuscript
writing. Aman Haque guided on experiment design, and involved in data analysis as well
as manuscript preparation.
59
3.1 Low Temperature-Electron Wind Force Synergy in Molybdenum Disulfide
In this section, we will discuss low temperature processing of two-dimensional material
such as molybdenum disulfide (MoS2). Crystallinity plays important role in electronic quality of
chemical vapor deposited MoS2, which tends to deteriorate with decrease in deposition
temperature. Thermal annealing can improve the quality but requires very high temperatures. In
this study, we investigate an alternative low temperature (room temperature to 400 C) annealing
process that exploits the electron wind force (EWF) during the passage of current. Electrical
current density gives rise to atomic scale mechanical force whenever the electrons encounter
defects in the lattice or grain boundaries. After hypothesizing that this EWF can significantly
enhance defect mobility even without any temperature field, we demonstrate the process using
in-situ transmission electron microscope (in-situ TEM) and molecular dynamics (MD)
simulation. To validate the hypothesis, we choose monolayer metal organic chemical vapor
deposited MoS2 deposited at 400 C, and process at low temperature. Experimental results show
5 times enhancement in electrical conductivity, which is supported by the selected area electron
diffraction (SAED) patterns indicating significant grain growth. Discrete spots in SAED pattern
also indicate evolution of high crystallinity at low temperature. Computational investigation
shows that atomic mobility, defects healing as well as reorientation mechanism at the grain
boundaries.
3.1.1 Objective and Motivation
Two-dimensional (2D) transition metal dichalcogenides (TMDCs) such as MoS2 show
intrinsic band gap ranging from 1.2eV to 1.8eV depending on the number of layers [123]. MoS2
has attracted the materials community because of their exceptional electrical, optical, and
60
mechanical properties [124-126], thus making it desirable for field-effect transistors (FETs)
[127-129], flexible electronics [130-132] and sensors [133, 134] applications. Due to the limited
flake size and exfoliation-induced defects, exfoliated MoS2 layers remain concerns for scalability
[135-137]. Thus, a need for scalable production of large-area and highly crystalline MoS2 has
spurred advancement in chemical vapor deposition (CVD) processes [138-140]. It is well
established that quality of CVD grown 2D materials is a strong function of deposition
temperature. However, lower temperature deposition tends to decrease the materials quality. For
example, at low temperature multiple nucleation sites on supporting substrate could form
nanocrystalline MoS2 during the CVD growth, with randomly oriented grains connected by
grain boundaries (GBs) [141, 142]. These GBs in 2D materials have detrimental effects on
physical properties such as higher resistivity, lower carrier mobility, poor mechanical properties
and lower thermal conductivity of 2D materials [141, 143-147]. However, quality of low
temperature deposited 2D materials can still be improved with post-processing. Unfortunately,
for MoS2 the required thermal annealing temperatures may be in excess of deposition
temperature i.e., 800 C [148, 149]. From device fabrication process compatibility and residual
stress perspective, lower processing temperatures are desired.
The primary objective of this study is to minimize the defects and GBs in low
temperature CVD grown MoS2. Thermal annealing is a widely used technique to enhance defect
mobility to minimize their volume fraction in the solid, or to minimize the GB density (thus
increasing the grain size). Thermal annealing technique has been also employed for annealing of
2D materials. For example, post thermal annealing of MoS2 to enhance the crystallinity has been
studied by both ex-situ [150] and in-situ [151] experiments. However, reported thermal
annealing temperature was high, i.e., in the range of 700˚C-1000˚C, and time was in the order of
61
hours. A major drawback of, high temperature annealing is detrimental thermal stress in layered
materials. In addition, the required annealing time is also very long because highly uniform
temperature means random diffusion, where both lattice and defective regions are given equal
vibrational energy [152]. Thus in our present study, we propose low temperature electrical
annealing of MoS2. In our experiment, temperature was around 400˚C to avoid any undesirable
thermal stress, and time required for this annealing is in the order of minutes (i.e 5 minutes). We
proposed that defects in electrically conducting materials can be targeted at the GBs using a non-
thermal process, i.e., momentum transfer of electrons at defects. When we pass electrical current
through the sample electron scattering is massive at the defects, where the electrons lose their
momentum [58], developing atomic scale force known as the ‘electron wind force’ (EWF). Such
momentum transfer imparts high defect mobility and creates a ‘directional diffusion’ (compared
to the random diffusion in high temperature heating). Our hypothesis is that such directed, non-
thermal diffusion can be effectively exploited to control defect mobility and migration at low
temperature and high speed. Additionally, temperature from resistive Joule heating can be
helpful in enhancing defect mobility. In this study, a balance of Joule heating (i.e, below thermal
degradation failure) and EWF is achieved through careful consideration of specimen geometry
and thermal boundary conditions. Our proposed electrical annealing technique has been
successfully deployed for conductive thin films, additive manufactured materials and alloys
[116, 153, 154].
In this study, we investigate the proposed EWF induced low temperature annealing
process on 2D materials. Nominally, monolayer MoS2 specimens were synthesized at 400 C
using low temperature metal-organic chemical vapor deposition (MOCVD) technique. The
material was then transferred to a custom designed and fabricated micro-electro-mechanical
62
(MEMS) chip that can be mounted on a TEM specimen holder with electrical biasing capability.
The specimen resistance and selected area electron diffraction (SAED) patterns are recorded for
the input current density. Significant enhancement in electrical conductivity indicates the
effectiveness of the EWF induced low temperature processing. To obtain atomic-scale defect
mobility and migration mechanisms that lead to low temperature EWF induced annealing, we
develop MD simulation models using LAMMPS [75] simulation package.
3.1.2 Materials and Methods
3.1.2.1 MOCVD Synthesis
Low temperature MOCVD is used to synthesize MoS2, the reactor design and
experimental procedures are described elsewhere [155, 156]. To summarize, 4×10-4 sccm
molybdenum hexacarbonyl (Mo(CO)6) and 0.55 sccm diethyl sulfide (DES) were carried to the
hot wall MOCVD reactor by 50 sccm H2 and 565 sccm Ar. The growth is conducted at 400°C
for 1 hr on 300 nm SiO2 substrate that yielded nominally 1 layer of MoS2.
3.1.2.2 Device Fabrication and Sample Transfer
Wet transfer technique using PMMA and HF solution was employed to transfer MoS2
sample on the MEMS chip for electrical biasing. Details device design and fabrication details are
given elsewhere [91]. Figure 3.1 shows schematic of the transfer technique and experimental
setup. At first, MoS2/sapphire specimen was spin-coated with 950 K PMMA A2 at 1500 rpm.
After spin coating, MoS2/PMMA was released in 16% HF solution from sapphire utilizing the
sacrificial etching and surface tension to peel off the MoS2/PMMA film [157]. Then released
specimen was manipulated on the MEMS device for biasing purpose. Finally, acetone and
63
isopropyl alcohol (IPA) was used to dissolve PMMA film to obtain freestanding MoS2 sample.
The device is then wire bonded and mounted on an in-situ TEM holder for in-situ TEM study.
Figure 3.1. Schematic shows the transfer process of monolayer MoS2 on to a MEMS device and
subsequent experimental setup for in-situ TEM investigation [67].
3.1.2.3 Computational Details
We developed classical MD simulation model using Reactive Empirical Bond-Order
(REBO) potential [80] and implemented it in LAMMPS [75] package. Single layer
polycrystalline MoS2 were modeled with 10 randomly oriented grains with approximate grain
size of approximately 2.5 nm. We incorporate both Joule heating and EWF effects in our
simulation. We chose 0.5 fs simulation time step and maintained periodic boundary condition
along the length of the sample (i.e., x-direction) while keeping free surface boundary conditions
on transverse and film normal directions (i.e., y and z directions respectively). Energy
minimization was performed on the simulation cell using conjugate-gradient (CG) method
64
followed by NVT dynamics for several thousand steps in LAMMPS. Time integration was
carried out using the Verlet algorithm during the NVT dynamics. The EWF was applied on
individual atom. It is difficult for a large system to directly employ electron effects in classical
MD simulation as classical MD simulation solely depends on Newton’s equation of motion.
Thus in this present study, we indirectly quantify equivalent EWF as mentioned in section 2.1.2
on each atom using the Huntington-Grone [93] model. In our simulation, individual atom
experiences a total force that includes force from REBO potential and EWF, and can be written
as follows:
𝐹𝑇𝑜𝑡𝑎𝑙 = 𝐹𝑅𝐸𝐵𝑂 + 𝐹𝑤𝑖𝑛𝑑 (2)
To mimic the experimental condition i.e., direct current (dc) in the specimen, we impart the
electron wind force on individual atoms for 1.0 ns.
Figure 3.2. (a) Observation of almost one order of magnitude reduction in electrical resistance of
MoS2 specimens during EWF annealing, and (b) spatial temperature distribution for the highest
current density [67].
65
3.1.3 Results and Discussion
To investigate the effectiveness of the electrical current annealing induced quality
enhancement in low temperature MOCVD MoS2, we conducted in-situ TEM (electrical biasing
and resistance measuring) experiments. Experiments were performed inside a FEITM Talos
F200X scanning/transmission electron microscope (S/TEM) with a resolution of 1.2Å using field
emission gun and 200 kV acceleration voltage. In our experiment, we pass DC current through
the specimen in steps of 10µA until 0.1 mA limit is reached. We allow about 5 minutes between
each current density steps, which is remarkably shorter than conventional thermal annealing.
Figure 3.2a shows a representative specimen resistance vs. current density data, which shows the
resistance as a function of current density. We noticed a dramatic reduction in initial resistance
from 1 MΩ to about 180 kΩ after the electrical annealing. Electro-thermal simulation results
predict the temperature profile in the specimen as shown in Figure 3.2b. In our present
simulation, we consider low thermal conductivity of MoS2 [158] to ensure maximum
temperature rise during the electrical annealing of MoS2. Literature on Raman thermometry
[159] measurement indicates that MoS2 temperature could reach as high as 380°C prior to their
breakdown at a drain current of 210uA/um. This reported current is more than twice in
magnitude higher compared to our biasing condition. Thus actual temperature rise in our sample
might be even lower than the predicted simulation results. Due to the nature of the freestanding
geometry, the specimen temperature is highest (around 400 C) at the vicinity of the center
region of the sample. Whereas massive silicon electrodes effectively constrain the specimen ends
to be at room temperature as shown in Figure 3.2. The arrows in Figure 3.2b indicates the
locations for the SAED patterns.
66
It is well known that thin film resistance is strongly influenced by surface and internal
(such as grain boundary) defects. For example, Figures 3.3a and 3.3b show the bright-field (BF)
and SAED images for the as-deposited specimen. After annealing at near room temperature with
EWF at 9x105 A/cm2 current density, the corresponding image is shown in Figure 3.3d. The
transformation of pre-annealed sample as shown in Figure 3.3c to apparent featureless
appearance in Figure 3.3d suggests a significant decrease in surface roughness, which may
contribute to the decreased surface scattering. This may have significant impact on the electrical
resistance as we have noticed in Figure 3.2a since the surface scattering of electrons plays a
dominant role [160] on electrical resistivity.
Figure 3.3. In-situ TEM electron wind force annealing results: (a) Bright-field image, (b) SAED
pattern of as-deposited nominally monolayer MoS2 specimen, and (c, d) The same location after
67
annealing at 9.5x105 A/cm2 current density [67].
Due to the high volume fraction of GBs pronounced scattering event takes place at the
GBs. This is in agreement with the SAED available in the literature [149]. Since the grain size
(~5nm) is close to the characteristic length scale of electron mean free path, a small increase in
grain size can result in very large decrease in resistance. Figure 3.3e shows a representative TEM
SAED pattern at low temperature locations (arrows in Figure 3.2b) in the specimens after EWF
annealing. Here, the initially diffused rings appear sharper after annealing. More remarkably, we
observe distinct hexagonal spots at a current density of 9.5x105 A/cm2, suggesting the existence
of large crystalline domains. This observation follows the scaling of EWF, which is the highest
at the grain boundaries. While the discrete spots in the SAED pattern are clear indication of
large-scale crystalline domains, the actual specimen morphology is difficult to achieve. Thus we
propose few possibilities such as (a) prominent secondary grain growth, where energetically
favorable grains may grow abnormally large at the expense of neighboring smaller grains [161]
and (b) electrical discontinuity in domains that lead to little annealing effect. Figure 3.3f
schematically exhibits the later scenario where some regions (labeled as region 2) may see lower
electron wind force as the current will flow mostly from region 1 to 3.
We performed a series of MD simulations to comprehend atomic scale mechanism of
defects annihilation. To investigate the quality enhancement mechanism such as GBs
reconstruction, we analyze simulation trajectories. Figures 3.4a-3.4c show such analysis, which
indicate dislocation migration between adjacent grains. Here, r1 and r2 indicates two rings at the
lower center of the sample to track the initial dislocation position as marked by () at the GB
during different simulation time. Figure 3.4a represents deep yellow colored grain (boundary
marked by black dotted line) and its three nearest neighbor grains. Initial dislocation position
68
adjacent to the ring r2 (as shown in Figure 3.4a) migrates to the next ring as shown in Figure
3.4b after 200ps. With the progression of simulation time dislocation marked in Figure 3.4c
further moves to the GB. This dislocation migration due to effect of EWF and Joule heating
during electrical current induced annealing could lead to the quality enhancement by rearranging
the atoms at the GBs. Figure 3.4d shows the transformation of a 6|8Mo ring defect to 6|6|4
defects due to the synergy of EWF and Joule heating. Thus one 6|8Mo ring defect could generate
two perfect 6 rings and one 4 ring during the electrical annealing as shown in Figures 3.4d and
3.4e.
Figure 3.4. (a)-(c) 6|8Mo type dislocation migration during the electrical annealing, (d)-(e)
Transformation of a 6|8 ring to 6|6|4 ring at the Grain boundary (GB) (Individual grains are
69
shown by different colors, smaller radius sphere indicates Sulphur atoms and larger sphere
indicates Molybdenum atoms) [67].
Figure 3.5 represents the annihilation mechanism of vacancy defects at the GBs of the
sample. Here, we mark r1 and r2 rings to track the defects during the annealing simulation.
Figure 3.5. Transformation of vacancy defects at the GB: (a) initial sample, (b) vacancy
transforms to 6|8 S defects, (c) 6|8 S transforms to 6|6|4 ring defects, (d) formation of 6|8 S due
to the dislocation motion, and (e) formation of 6|6|4 ring (Individual grains are shown by
different colors, smaller radius sphere indicates sulfur atoms and larger sphere indicates
molybdenum atoms) [67].
70
Figure 3.5a shows the vacancy defects at the GBs of the sample prior to the annealing simulation
marked by 0 ps. With the progression of simulation time, we notice vacancy defect transforms to
6|8S ring defects after 265ps as shown in Figure 3.5b. Figure 3.5c shows the intermediate
transition of 6|6|4 defects from 6|8S ring defects after a simulation time of 480 ps. These 6|6|4
defects further transform to 6|8 Mo ring defect after 600 ps as shown in Figure 3.5d. Thus,
dislocation migrates to one ring lower left (Figure 3.5b and 3.5d). After 615 ps of simulation
time, 6|8S ring defects transformed to 6|6|4 defects as shown in Figure 3.5e. This vacancy defect
annihilation at the GBs further facilitates the quality enhancement in the specimen.
3.1.4 Conclusion
In this section, we examined the effectiveness of the EWF induced quality enhancement
and defects mobility in 2D MoS2. Our study indicates that EWF annealing of MoS2 specimens
with defects (such as GBs in polycrystalline MoS2) can be achieved at low temperatures. This is
in stark contrast with the state of the art thermal annealing of 2D MoS2, where very high
temperatures (~800 C) are required. We synthesized monolayer MoS2 at low temperature (~400
C) to test our hypothesis. Then we performed in-situ TEM to observe and study the
enhancement in the quality of the specimen. The electrical resistance decreased by 5 times at a
dc current density of approximately 9.5x105 A/cm2. Such enhanced transport property is
supported by the EWF induced changes in SAED patterns in regions of specimens where the
temperature was constrained to the ambient. To gain insights on defects annihilation, we also
conducted MD simulation. Our simulation study captures the atomic scale defect migration and
grain reconstruction mechanisms that led to specimen quality enhancement. Simulation results
71
shows that during the proposed annealing, the grain boundary atoms could rearrange and re-
reorient due to impetus of EWF.
3.2 Low Temperature Processing of 2D Material based Thin Film Transistors
In this section, we will extend our proposed novel non-thermal annealing process for two-
dimensional (2D) materials based devices. Instead of high temperature, we exploit the electron
wind force (EWF) at near-room temperature conditions to process back-gated WSe2 field effect
transistor (FET). As mentioned in earlier section, EWF is an atomic scale mechanical force that
acts only in the defective regions, which is proposed to provide very high defect mobility. This
EWF processing is demonstrated on back-gated WSe2 transistors. EWF annealing was performed
by passing current through the device drain and source channel while actively removing the
Joule heating. We observed approximately one order of magnitude increase in the output current,
validating our hypothesis on the mobility imparted by the EWF to migrate and eliminate defects.
Our molecular dynamics (MD) simulation confirms the defects annihilation and local metallic
phase transformation, which further enhance device performance. Proposed technique will
potentially lead to time and cost-effective post-processing of two-dimensional materials based
devices.
3.2.1 Objective and Motivation
Two-dimensional transition metal dichalcogenides (2D TMDCs) [162-166] possess
extraordinary electrical, optical and mechanical properties [71, 124, 165, 167, 168] which make
them as a potential candidate materials for next generation nano-electronic applications. Unlike
graphene, single layer TMDCs exhibit indirect to direct band gap transition [169, 170]. This
tunable band gap properties of TMDCs offer new types of electronics such as field effect
72
transistor (FET), photo-detectors, and memories with selectable properties [127, 167, 171, 172].
Recently, WSe2 based electronic devices such as FET has attracted significant interest [173-175]
due to their p-type, n-type and ambipolar carrier transport properties via electrode engineering
and thickness modulation [176, 177].
Performances of the 2D electronics are a function of crystallinity of the TMDCs. Depending
upon the synthesis conditions different types of defects such as grain boundaries (GBs), ring
defects, dangling bonds, vacancy defects may appear in 2D materials [178-184]. These defects
potentially degrade the physical properties of 2D TMDCs such as carrier mobility, resistivity,
thermal conductivity and mechanical properties [141, 143-147, 185], which commute through
device performance. A conventional practice to control these defects and enhance device
performance is thermal annealing. However, thermal annealing requires very high temperature
(>800 ºC), which itself can deteriorate the material by creating more defects or stagnating the
existing ones due to the residual thermal stress between materials interfaces. Thermal annealing
is not suitable for low temperature materials as it is a slow process (at low temperature) based on
random diffusion. Instead of conventional thermal annealing [186], an electrothermal-annealing
(ETA) could have potential to enhance device performance after fabrication. ETA method uses
Joule heating as a driving force to eliminates defects and enhance device performance [187].
Recently, ETA techniques have been implemented on metal-oxide semiconductor field-effect
transistor (MOSFET) [187] and amorphous-oxide-semiconductor (AOS) thin-film transistors
[188] . ETA technique uses short-pulses (0.1-1 ms) at high voltages to repair devices. However,
their optimization efforts indicate that annealing time and voltage could be adjusted to achieve
desired outcome.
73
In this section, we propose an alternative, non-thermal route compared to conventional
thermal annealing. Instead of using heat energy to increase the atomic mobility, we specifically
target defective atoms with mechanical force know as electron wind force (EWF) by passing
current through the sample. While current passes through the sample, electrons are minimally
scattered by the lattice, whereas they transfer their momentum whenever they meet defects and
GBs [58] thus enhance atomic mobility. EWF is also coupled with resistive (Joule) heating,
which is well known in the electromigration literature [100] Thus, EWF is minimum inside the
lattice and maximum at the defects. Previous studies on blanket films of thin metal [116, 154,
189] [189] and 2D materials [190, 191] confirms the effectiveness of EWF, even in presence of
the Joule heating. To ensure the effectiveness of the low temperature processing we eliminates
the Joule heating using active cooling and yet demonstrate enhancement of crystallinity. Active
cooling decouples and eliminates the Joule heating to maintain the transistor surface at room
temperature. This is in contrast to the literature, where defect mobility is viewed as a
predominantly Joule heating effect. Major contribution of this study is the low temperature
processing of WSe2 FET to enhance the performance of as-fabricated device. Our present study
thus demonstrates EWF annealing on fully functional electronic devices and provides insights on
the processing parameters and their optimization for time and cost-effective performance
enhancement.
3.2.2 Materials and Methods
3.2.2.1 Metal-organic Chemical Vapor Deposition (MOCVD) Synthesis of Epitaxial WSe2
The MOCVD growth of WSe2 thin films is performed in a custom-built vertical cold-
wall CVD reactor as previously reported elsewhere[192]. The tungsten hexacarbonyl (W(CO)6)
74
(99.99 %, Sigma-Aldrich) and hydrogen selenide (H2Se (99.99 %, Matheson)) are used
precursors with high-purity H2 as the carrier gas. The epitaxial WSe2 was synthesized as
previously reported 3 step growth method (nucleation, ripening and lateral growth) at 800 °C and
700 Torr on c-plane sapphire substrates[192, 193]. Prior to the synthesis, substrates were cleaned
using ultrasonication in acetone (10 min) and isopropyl alcohol (10 min) followed by 20 min
immersion in commercial Piranha solution (Nanostrip, KMG Electronic Chemicals) at 90 °C and
deionized (DI) water rinse (10-14 times).
3.2.2.2 Device Fabrication and Characterization
The WSe2-based field effect transistors (FETs) were fabricated and characterized to study
the impact of the EWF annealing process on synthetic two-dimensional materials. The as-grown
monolayer WSe2 film was transferred to a 285 nm thermally grown SiO2 dielectric layer
supported by a rigid Si substrate. Deposited SiO2 serves as the global back-gate of the device.
Standard e-beam lithography (Vistec EBPG 5200) process was used to define the source and
drain electrodes. To isolate the WSe2 channels from blanket film Plasma-Therm Versalock 700
high-density inductively couple (ICP) plasma etch tool was used with a SF6/O2 30/10 sccm gas
mixture for 20 s. We used physical vapor deposited 40 nm Ni/ 30 nm Au metal stack as the
source and drain electrodes. All electrical measurements were carried out on a room temperature-
controlled stage in a high-vacuum (~10-5 Torr) CRX-VF Lake Shore probe station to minimize
threshold shifts due to moisture along with a Keysight B1500A Semiconductor Device Analyzer.
3.2.2.3 Experimental Setup, Modeling and Simulation
To study the effect of EWF we decoupled the Joule heating from the EWF and exploit the
latter to anneal the 2D WSe2 at near room temperature. In our experiments, the transistor was
75
placed on a temperature-controlled stage maintained at 296K. Due to the resolution limitation of
commercially available microscope, direct measurement of the temperature is difficult since the
dimension of the channel length and width are 1 μm and 5 μm respectively. Thus to assess
temperature rise in the sample during the electrical annealing, we performed multi-physics
modeling of Joule heating using COMSOL® software as shown in Figures 3.6c and 3.6d. To
mimic the exact experimental condition we consider interfacial thermal boundary conductance of
12 MW/m2.K [194] and 125 MW/m2.K [159] between WSe2-SiO2 and Si-SiO2 respectively. Our
electro-thermal simulation did not show any significant temperature rise up to 30 V annealing
voltage, which could be attributed to the low annealing current density (~3.2×107 A/m2) and
controlled stage temperature (i.e., 296 K). However, at high biasing condition sample
temperature may rise as high as 380 K as shown in Figure 3.6c due to the Joule heating effects at
50 V annealing voltage accompanied by moderately high current density (~3.2×108 A/m2). Our
calculated temperature rise is in a good agreement with recent study [159].
We have adopted molecular dynamics (MD) method to investigate the microstructural
changes during EWF annealing. Though length and time scales are different in MD simulation, it
will provide qualitative understanding of the defects annihilation mechanism and quality
enhancement of the WSe2 film and FET devices. Stillinger-Weber [81] potential is used to
simulate monolayer WSe2 polycrystalline films using LAMMPS [75] code. Polycrystalline WSe2
samples with 16 numbers of grains (Figure 3.9a) and a grain size of 8nm were modeled using
voronoi tessellation technique. The grains were rotated at 0°, 30°, 45° and 60° with respect to
film normal i.e., c-axis. We performed energy minimization using conjugate-gradient (CG)
method followed by isothermal–isobaric (NPT) dynamics runs for several thousand steps to
obtain residual stress-free structure on the as-deposited sample. Once we obtain equilibrated
76
structure, we apply wind force on each atom calculated from the Huntington-Grone [93] ballistic
model to mimic electrical annealing (EA) (as mentioned in section 3.1.2).
Figure 3.6. (a) Optical Microscope image of fabricated back-gated WSe2 FET transistor, (b)
schematic diagram of WSe2 transistor, (c) electro-thermal simulation model of WSe2 FET, and
(d) temperature profile across the cross-section of the sample obtained from model [68].
3.2.3 Results and Discussion
WSe2 FETs devices were annealed at different current density under vacuum condition.
During this annealing electrical annealing induces both Joule heating and EWF. However, due to
the controlled stage temperature Joule heating effects assumed to be insignificant. In addition to
this, transistor substrate itself acts as massive heat sink during the annealing process. The finite
element electro-thermal simulation also suggested negligible temperature (304 K) rise at 30 V
77
annealing voltage. Thus, we can hypothesize that single stimuli such as EWF can eliminate
defects even at low temperatures.
Output current as a function of drain voltage (i.e., Id vs Vd) is shown in Figure 3.7a which
reveals the effectiveness of the low temperature EWF annealing. In Figure 3.7a, we noticed an
increment in drain current by more than one order of magnitude after annealing at 20V as shown
by cyan color line. During 20 V EWF annealing we anneal the sample for 5 minutes and the
channel current reaches a maximum value of 0.2 μA. The annealing time is sufficiently longer
compared to the other study [188]. This relatively long duration annealing time makes EWF
method effective without introducing any significant temperature rise in the device.
Enhancement in drain output current during due to the EWF could be attributed to the
improvement of film quality as well as contact resistance at the interface of electrode and the
device layer [186, 190, 195]. We noticed almost one order of magnitude increment in output
current, which suggests significant defect annihilation in the WSe2 channel layer due to the low
temperature annealing. We performed annealing at different voltages namely 20 V, 30 V, 40 V
and 50 V to figure out optimum voltage. We found 30 V as an optimum voltage where annealing
effects is maximum. Above 30V annealing voltage, the device starts to degrade and then gets
damaged at 50 V, probably because the EWF itself starts to migrate atom from the cathode to
anode accompanied by Joule heating. Essentially, the EWF minimizes different type of defects
that can form in 2D materials during synthesis or device fabrication [196]. Due to the high
surface to bulk ratio in 2D materials these atomic scale defects can strongly influence physical
properties. Defects such as tungsten (W) and Selenium (Se) atoms vacancy might introduce
semiconducting to metallic behavior in WSe2 [197] . P-type semiconducting behavior has been
reported in MoS2 with Mo vacancy [198]. Additionally, electronic properties such as electron
78
mobility can be significantly limited by atomic scale defects such as vacancy, dislocations, and
grain boundaries [142].
Figure 3.7. (a) Output characteristics of WSe2 FET after annealing at different drain voltage, and
(b) Improvement in drain current after annealing while FET surface was maintained at 296K
[68].
Prior to the annealing, the device shows a maximum of 1.89×10-2 µA/µm drain current
achieved at 45 V gate voltage as shown by transfer characteristic curves in Figure 3.8, while the
drain voltage was kept constant at 5V.We notice significant improvement in drain current after
annealing at 20 V as indicated by pink color line in Figure 3.8a. Drain current reaches a
maximum plateau value of 5.47×10-2 µA/µm around 45 V gate biasing without any threshold
voltage shift (Figure 3.8a). Our optimization efforts shows a maximum drain current of 1.35×10-1
µA/µm (Figure 3.8b) at 30V annealing voltage, which is one order of magnitude higher
compared to the as-received device. This remarkable enhancement of device performance can be
attributed to the improvement of the channel quality. However, device starts degrading above 30
V annealing voltage and failed to operate at or above 50 V. Device degradation could be
79
attributed to the high channel current and hot electron (electrons with high kinetic energy) induce
traps in the channel layer [199]. These traps hinder the electron mobility which potentially
reduces the output current of the device at high annealing voltage. In our present study, all
measurements were performed three times and three sets of experiments were conducted to
ensure consistency. As mentioned earlier degradation at higher annealing voltage (i.e., >30 V)
could be attributed to the high current density and traps which are outcomes of combined
electrical and thermal field. Due to this detrimental thermo-migration effect, atomic flux mobility
increases, and defects/voids appears in the channel layer. We can express the atomic flux
mobility of atoms due to the EWF (JEWF) and temperature (JT) as follows [57]:
𝐽𝐸𝑊𝐹 =𝑁𝑒𝑍∗𝐷𝜌
𝑘𝑇𝑗; 𝐽𝑇 = −
𝑁𝐷𝑄𝜌
𝑘𝑇2 ∇𝑇 (3.2a)
Where, D is diffusivity, k is Boltzmann’s constant, T is absolute temperature, N is vacancy
concentration, Z* is effective charge number, j is current density, e is the elementary charge, and
Q is the heat of transport. Eq. (1) shows that atomic flux mobility is proportional to both current
density (j) and temperature gradient (∇𝑇). Thus, at higher annealing voltage high EWF in
conjunction with high thermal field can induce directed diffusion of atoms from cathode to anode
side of the transistor (shown later in Figure 3.10). This directed diffusion can lead to the defects
such as voids and cracks formation near the cathode side, and hillock at the anode side.
Eventually the device fails due to this defects formation and migration effect.
Synthetic 2D TMDCs usually contains higher defect densities compare to the mechanically
exfoliated counterparts [31]. Different types of defects such as grain boundaries (GBs), metal as
well as chalcogen vacancies, and ring defects are commonly found in the synthetic materials,
80
Figure 3.8. (a) Transfer characteristics showing WSe2 transistors performance after annealing at
different voltage, and (b) maximum drain current obtained after annealing [68].
which ultimately degrade transport performance of the devices [183, 200-202]. Since defect-free
synthesis of TMDCs remains as challenging, we propose an alternative approach using external
stimuli to improve the film quality and transport performance [116, 154, 189, 190]. Direct
visualization of the defect annihilation or GBs reconstruction is difficult, even with an in-situ
atomic resolution microscope. Therefore, we take resort in MD simulation. To investigate the
effectiveness of EWF annealing we model a nanocrystalline WSe2 film with 16 grains and
average size of 8 nm. The initial equilibrated structure primarily contains 2H phase a shown in
Figure 3.9a), and after annealing it contains 2H, 1T and intermediate phase as well as
reconstructed grain boundaries of WSe2 (Figure 3.9b). At the two opposite ends of the sample,
we define electrode by fixing atoms as shown by the green and red bands in Figure 3.9b. The GB
defects could be annihilated via phase transformation, intermediate structure and twinning as
shown in Figure 3.9b. Similar types of 2H to 1T-phase transformation in monolayer MoS2 has
been observed in recent experiment [203]. Interestingly, our study also shows semiconducting to
metallic behavior of WSe2 during 2H to 1T phase transformation [184]. This local 1T phase
81
transformation might enhance the device performance due to the incorporation of metallic
behavior. Colored arrowhead in Figure 3.9b indicate possible electron flow paths without
significant scattering at the GBs. These defects annihilation at GBs might be the primary cause
for the higher output current as shown in Figure 3.7a.
Figure 3.9. Monolayer WSe2: (a) prior to the annealing, and (b) after annealing [68].
Our computational model also captured device degradation phenomenon beyond the
optimum value of the annealing voltage. This phenomenon could be attributed to the thermal and
electrical field induced failure, which is readily observed in metallic thin films [116, 204, 205].
At higher biasing voltage, both EWF and Joule heating are high enough to induce atomic
migration, which further degrades the WSe2 film. In order to investigate void formation and
migration in WSe2 FET, we apply one order of higher EWF in the WSe2 device layer and track
atomic motion as shown in Figure 3.10. Initial sample with anode and cathode marked by red
and blue color region is shown in Figure 3.10a. We observe void formation near the cathode area
of the sample after 5ps. As the simulation time progress, voids start to grow and we observe
82
stress relaxation at the vicinity of the voids. With the progression of simulation time, we notice
gradual increment in void size near the cathode area and mass accumulation at the anode area.
Figure 3.10. (a)-(f) Failure at high biasing condition due to the electrical and thermal field, and
(g)-(j) side view of the sample showing void creation at the cathode side (left) and mass
accumulation at the anode side (right) (Color bar in Figure 3.10 shows the atomic stress
distribution in the sample) [68].
Thus, device fails due to this high atomic mobility and thermal field generated by the electrical
current. Figures 3.10g-3.10j represent side view of the sample under high EWF. At the
beginning of the simulation there is no void or mass accumulation (Figure 3.10g) however, with
83
the progression of simulation time device degrades due to the voids enlargement (Figure 3.10h
and 3.10i) and mass accumulation in WSe2 sample as shown in Figure 3.10j.
In this section, we have employed electrical current annealing method to enhance the
WSe2 FET device performance at low temperature. We notice one order of magnitude output
current increment after annealing at an optimum annealing voltage (i.e 30 V). However, beyond
this optimum condition device performance starts to degrade, and device fails to operate after 50
V annealing voltage due to the high electro-thermal field. Present study provides insights on an
alternative, time and cost-efficient annealing technique using EWF, which can improve device
performance without introducing thermal stress, and can be applied to a wide range of devices
including flexible electronics.
3.2.4 Conclusion
We have adopted both experimental and computational approach to validate the
effectiveness of EWF annealing at low temperature. We chose WSe2 back-gated FETs, and our
investigation shows more than one order of magnitude drain current output. However,
enhancement of device performance is a non-linear function of annealing voltage, and device
performance degrades after an optimum annealing voltage. Experimental investigation indicates
device exhibits enhanced performance up to 30 V and fails to operate above 50 V annealing
voltage. Our computational study shows GB defects could be annihilated due to the atomic
mobility and phase transformation even at low temperature under the influence of EWF. Study
also shows that metallic 1T-phase transformation might enhance device performance. Simulation
results reveal the device degradation by capturing defects formation and migration, which
includes void creation near the cathode and hillock near the anode. In sum, low temperature
84
electrical annealing take less than 5 minutes and could be applied on flexible electronics or other
devices during its operation without introducing any thermal stress. In addition, our study shows
that EWF process is order of magnitude faster and energy efficient compared to the conventional
thermal annealing.
85
Chapter 4
Synergy of Stimuli on Operation and Degradation of Nanoscale Devices
Contents of this chapter are based on the following journal articles:
Zahabul Islam, Nicholas Glavin, and Aman Haque, Potential and Challenges of in-situ
Microscopy on Electronic Devices and Materials, IOP eBooks, Edited By S. J Pearton, 2020
(submitted)
Author of this dissertation designed the experiment, performed the sample and device
preparation, experimentation, data analysis and manuscript writing. Nicholas Glavin
guided on manuscript writing. Aman Haque guided on experiment design, and involved
in data analysis as well as manuscript preparation.
Zahabul Islam, Aman Haque, and Nicholas Glavin, Real-time visualization of GaN/AlGaN high
electron mobility transistor failure at off-state, Applied Physics Letters, Volume 113, 183102
(4pp), 2018.
Author of this dissertation designed the experiment, performed the sample and device
preparation, experimentation, data analysis and manuscript writing. Nicholas Glavin
guided on manuscript writing. Aman Haque guided on experiment design, and involved
in data analysis as well as manuscript preparation.
86
4.1 On-state Degradatation of High Electron Mobility Transistor
In this section, we have shifted our focus on thin film devices such as high electron
mobility transistor (HEMT), and studied their degradation behavior under external stimuli. Thin
film HEMT devices are in a plain strain condition due to the specimen geometry and boundary
condition. In addition to plane strain condition, external stimulus such as external biasing
develops thermal, electrical and mechanical stress fields in these HEMT devices. The unique
geometry and complex interaction of multiple domains make these thin film HEMT devices as
ideal candidates for stimuli-synergy study. Thus, in this section we have investigated thin film
HEMT device such as gallium nitride (GaN) based HEMT using in-situ transmission electron
microscopy (in-situ TEM) under external stimuli. Recently, AlGaN/GaN HEMTs have drawn
tremendous attention due to their superior breakdown voltage, energy bandgap, and electron
mobility. However, the realizations of potentials of these devices are impeded by several
reliability concerns. Thus, this chapter addresses reliability issues of these devices under external
stimuli using in-situ transmission electron microscopy (in-situ TEM) techniques.
4.1.1 Objective and Motivation
High electron mobility transistors (HEMT) are a new class of heterostructure field-effect
transistors primarily composed of III-V semiconductor groups. Confined carriers at the
heterojunction of HEMT act as two-dimensional electron gas (2DEG). Both Gallium arsenide
(GaAs), gallium nitride (GaN), and their compounds are widely used as HEMT. Owing to the
higher electron mobility of GaAs HEMT, their narrow bandgap limits their applications in power
electronics. On the other hand, GaN has wide bandgap and high breakdown voltage, which
87
makes them as a suitable candidate for high power and high-frequency applications. Figure 4.1
shows a comparison of properties among competing semiconductors [206].
Figure 4.1. Comparison of properties among competing semiconductors.
A comparison as shown in Figure 4.1 reveals that GaN has one order of magnitude higher
breakdown voltage, and both bandgap and electron saturation velocity of GaN is higher
compared to the competitor GaAs. Additionally, thermal conductivity of GaN is also higher
compared to other competing semiconductors. Thus superior material properties make GaN as an
excellent candidate for the next-generation high-frequency, high power applications. Literatures
reported that GaN exhibits more radiation-tolerance compared to SiC-based devices, which also
makes it an attractive choice for space applications [207-211]. Due to these extra-ordinary
properties, the scientific community has seen a surge in GaN-based HEMT applications in the
field of wireless, high voltage electronics, radar, automotive, space, to name a few.
88
Owing to their outstanding performance, long-term reliability issues under external
stimuli of GaN HEMT need to be addressed. Conventional approaches of reliability studies
involve electrical characterization data to detect failure signatures under external stimuli.
However, most of the reliability studies are performed outside the microscopes which includes
atomic force microscope (AFM), micro-Raman, cathodoluminescence, infrared,
thermoreflectance, scanning electron microscope (SEM) and transmission electron microscopes
(TEM) study. Thus, most of these studies are post-mortem in nature where failed specimens are
investigated to figure out the failure mechanism. Likewise, ex-situ techniques and post-failure
analysis is strenuous and not accurate, whereas in-situ microscopy could provide both real-time
visualization and quantitative capability to identify failure modes in electronic devices. However,
true potential of in-situ electron microscopes (EMs) study for device degradation under external
stimuli effects remains unrecognized until recently. In recent years, in-situ TEM study has been
employed to investigate GaN HEMT at different biasing conditions [28, 212, 213] to capture the
degradation mechanism in real-time.
It is well known that the degradation of GaN HEMT devices is a function of electrical,
mechanical and thermal fields. GaN HEMT can be operated in four different modes namely, low
current low field, high current low field, low current high field and high current high field.
Among these four operating conditions, both high current low field and high current high field
suffer from high electrical and high thermal fields also known as “on-state” mode, whereas low
current low field and low current high field are known as “off-state” mode and experience only
high electrical field. This chapter will discuss the effects of both “on-” and “off-” state
performance of GaN HEMT using in-situ TEM study. In the following sub-section, we will
89
briefly discuss the reliability issues of GaN HEMT under external stimuli at “on-state” modes
using in-situ TEM techniques.
4.1.2 Materials and Methods
To investigate the stimuli effects on degradation of HEMT deviecs, in our present study
we use a 6 W, 18 GHz, and 40 V rated depletion-mode GaN HEMTs on silicon-carbide
(Wolfspeed, CGHV1J006D) as shown in Figure 4.2. We use this HEMT die (Figure 4.2a) to
prepare electron transparent sample using FIB as shown in Figure 4.3. In-situ TEM experiments
are challenging due to the limited workspace of the TEM chamber, requirement of electron
transparent specimens and their transfer to the testing chip or specimen holder. For example,
preparation of an electron transparent (nominally 100nm thick) specimens involves Focused Ion
Beam (FIB) including series of steps of ion milling and cleaning without introducing significant
damage in the films.
Figure. 4.2. (a) Optical micrograph of GaN HEMT die, and (b) Low magnification SEM image
of the die.
90
Specimen preparation for in-situ TEM using FIB requires three important steps: (i) 100
nm thin electron transparent specimen preparation (Figure 4.3a-d), (ii) specimen transfer on
micro-electromechanical system (MEMS) device (Figure 4.3e) [212, 213], (iii) wire bonding of
(MEMS) device [91], and careful placement of chip on a TEM holder with biasing capability
(Figure 4.3f). The specimen preparation using FIB involves deposition of a protective layer to
define a thicker section in the GaN HEMT device and subsequent trench cutting (Figure 4.3a and
4.3b). In our study, a coupon was lifted-off using a controlled manipulator and placed on a TEM
grid (Figure 4.3c). At first, high current such as 21nA was used to lift out the specimen from the
bulk HEMT as shown in Figure 4.3b. However, thinning down to 100nm requires much smaller
current in the range of 0.79nA-34pA (values of currents are based on 4mm working distance).
The thickness of the specimen needs to be monitored at regular intervals during the milling and
Figure 4.3. Details of the GaN HEMT specimen preparation and transfer technique using FIB for
in-situ TEM reliability study.
91
cleaning process. In addition, both accelerating voltage and current need to be adjusted
depending on the specimen thickness during the cleaning. After electron transparent specimen
preparation, the final step requires a low accelerating voltage and current to transfer electron
transparent specimen from the TEM grid to the MEMS device (as shown in Figure 4.3e). Low
accelerating voltage was used to accomplish this goal without introducing any significant beam
damage and re-deposition on the specimen. Once the specimen is transferred on the MEMS
device, the specimen is ready for loading on the in-situ TEM holder (Figure 4.3f) and further
external stimuli study was conducted inside a transmission electron microscope (TEM) equipped
with EDS, EELS, SAED, Bright field (BF) and dark field (DF) imaging capability.
4.1.3 Results and Discussion
Electron microscopes (EMs) have been widely used to study the reliability of electronic
materials and devices due to their analytical as well as atomic resolution imaging capability.
Electron microscopes (EMs) provide high-resolution images that can be deployed to investigate
quality and degradation behavior of electronic materials and devices [214-216]. An EM equipped
with EDS, EELS, SAED, BF and DF makes it an excellent analytical tool for electronic materials
and device characterization. Thus, in this section we will employ in-situ TEM techniques to
investigate the effects of external stimuli i.e., voltage bias on GaN based HEMT under high
current high field state also known as “on-state” condition. During on-state mode operation
external stimuli such as electrical field and current is high enough to induce Joule heating, thus
the thermal field is unavoidable. Due to the high electrical and thermal fields, self-heating, as
well as hot electron effects, plays dominant role on device degradation [217, 218]. Figure 4.4
shows such failure mode obtained from in-situ TEM study. The threshold voltage of our device
is -3.1V, thus we apply 1V gate bias to make sure that the device is at on-state mode. A bright-
92
Figure 4.4. In-situ TEM reliability study of GaN HEMT: (a) electron transparent specimen prior
to the application of voltage stimulus, (b) device after failure, and (c) output characteristic curve
obtained from “on-state mode operation”.
field (BF) TEM image of GaN HEMT specimen prior to the application of external stimuli is
93
shown in Figure 4.4. Due to the specimen geometry and boundary conditions we noticed
significant amount of bend contour in the device layer, as well as in the substrate layer, (Figure
4.4a). For the purpose of reliability study under external stimuli, electrical connections were
made at the gate, drain and source pad during specimen preparation (Figure 4.3e). FIB deposited
conductive layer was used to accomplish this goal, which further introduced stress in the
specimen. Hence this specimen is in a plane strain condition and bend contour develops in the
specimen due to this residual stress. A failed device after on-state loading, and corresponding
output characteristic (i.e drain current vs drain voltage) is shown in Figure 4.4c. The device fails
at 10.8V drain voltage as shown in Figure 4.4c, and corresponding current density at this
breakdown point is in a good agreement with the reported value [219]. As mentioned earlier, on-
state biasing introduces sufficiently high current density and Joule heating, which further plays a
dominant role in the device degradation (Figure 4.4b). This high current density sublimates
buffer layers (GaN layer) as shown in Figure 4.4b. We have also noticed metal-semiconductor
diffusion near the drain electrode, which indicates thermal field is very high under on-state
biasing condition. Literatures have also reported similar types of failure mode [184, 220, 221]
during on-state biasing.
Figure 4.5a shows observed microstructural changes due to the thermal degradation in the
device layer. High thermal field during on-state operation is confirmed by the amorphization of
the passivation layer (Figure 4.5b) which requires very high temperature as the recrystallization
temperature of SiNx lies within the range of 1200-1400°C [222]. The diffused ring pattern
(Figure 4.5c) obtained from selected area electron diffraction (SAED) clearly indicates
amorphization and polycrystalline nature of the buffer layer. This structural change could be
partially attributed to the very high thermal field accompanied by high current density.
94
Figure 4.5. (a) Degradation of the passivation layer, (b) Evaporation of the buffer layer due to the
high thermal field, and (c) SAED indicating the transformation of GaN from crystalline to
amorphous state.
In our study, we have also noticed hot electron induced structural degradation in the
buffer layer as shown in Figure 4.6a. Under high electrical loading conditions, electrons could
achieve high kinetic energy (known as hot electron) which can develop traps and lattice defects
in the buffer layer. As the time progress, these defects could percolate and induce structural
defects as shown in Figure 4.6a, where we notice interface breaching at the GaN/SiC interface.
Weak bonding at the interface of GaN and SiC might make it vulnerable to the hot electron-
induced defects, thus interface easily separates under high external stimuli (Figure 4.6b).
Additionally, high thermal field can evaporate GaN, and small crystallites of gallium
nanoparticle might appear as shown in Figure 4.6c. Due to the lower binding energy of N atom
compared to Ga atom [223, 224], N atom can diffuse out of GaN and facilitates Ga nanoparticle
formation (Figure 4.6c).
95
Figure 4.6. GaN HEMT device degradation: (a) Hot electron induced failure at the source side,
(b) evaporation of the buffer layer and formation of small crystallites (nanoparticles) due to the
high thermal field, and (c) high-resolution image of a spherical crystallite.
In our present study, we have also captured failure incidence of GaN HEMT under
external stimuli as shown in Figure 4.7. This uniqueness of this study is to capture failure
initiation and propagation during real-time operation, thus it can pinpoint the exact failure
mechanism. Due to the high electro-thermal fields we have noticed metal pool formation near the
drain area during the on-state operation (Figure 4.7a). However, this metal pool rapidly diffuses
through the GaN buffer layer as shown in Figure 4.7b and 4.7c, and subsequent diffusion and
evaporation continue under high external field as shown in Figure 4.7d-4.7f. At the beginning,
diffusion was initiated in the buffer layer as shown by Figure 4.7g, and a rapid propagation of
diffusion through the substrate layer was observed after 12s as shown in Figure 4.7h and 4.7i. It
is obvious that the high thermal field originating from high drain current is responsible for the
device degradation during on-state operation. During on-state operation due to the minimum gate
leakage, no apparent degradation is observed at the gate (Figure 4.7) electrode. However, hot
96
electron ejected from the source during high power operation can breach the GaN/SiC interface
as represented in Figure 4.7i near the source side.
Figure 4.7. In-situ TEM reliability testing showing real-time device degradation.
97
To sum up, we have demonstrated a novel in-situ technique for the reliability study of
GaN-based HEMTs under external stimuli, namely drain current. We have observed high
electro-thermal fields induced degradation due to the high electric current. Study also indicates
hot-electron induced failure near the source side. However, no gate degradation was noticed due
to the minimum gate leakage under “on-state” operating condition.
4.1.4 Conclusion
In this present study we have studied complex material system such as HEMT devices
under external stimuli. In-situ TEM studies provide us quantitative as well as qualitative data at
the same time under external stimuli. Our “on-state” failure study shows that high electrical
current could induce high electro-thermal fields including Joule heating. Due to this high
electrical and thermal fields self-heating as well as hot electron effects are dominant and cause
the device failure during “on-state” operation. Though in-situ TEM provides useful information
on the degradation mechanism of HEMT, preparation of an integrated electron transparent GaN
HEMT specimen for in-situ TEM study is challenging and requires special attention. Lower
accelerating voltage and current need to be adopted to avoid any contamination and beam
damage during specimen preparation. During TEM experiment minimization of the e-beam
exposure is also required to enhance the accuracy of the measurements.
4.2 Off-state Failure of High Electron Mobility Transistor
In the previous section, we have studied degradation and failure phenomena in GaN HEMT
devices under external stimuli. It is well known that micro and nano-electronic devices are
complex functions of electrical, thermal and mechanical stresses as well as the quality of the
device materials and their interfaces. Unlike previous section, where we studied devices under
98
high external field with high current density, in this section we will demonstrated device
performance under high field but lower current density also known as “off-state” mode inside the
transmission electron microscope (TEM). Thus, in this present study we have studied reliability
of HEMTs under external stimuli using bright-field (BF), dark-field (DF), selected area electron
diffraction (SAED) and energy dispersive spectroscopy (EDS) techniques to characterize the
lattice defects, diffusion of the various elements. The ‘seeing while measuring’ approach
presented in this study can be useful in pinpointing the dominant failure mechanisms and their
fundamental origin under external stimuli.
4.2.1 Objective and Motivation
Recently, scientific community has seen a surge in Gallium nitride (GaN) based high
electron mobility transistors (HEMTs) for high-power and high-frequency applications [225].
Their reliability of HEMTs is influenced by external stimuli such as very high electrical, thermal
and mechanical stress fields during operation [219, 226]. As mentioned in earlier section,
HEMTs can degrade and fail at both ‘on’ (high temperature and electrical fields [227]) and ‘off’
(high electrical field [228, 229]) states. At high power conditions, thermally activated
mechanisms can lead to metal (ohmic contacts, gate and the interconnects) failure [230].
Additionally, high temperatures also gives rise to thermo-mechanical stress, which is added to
the pre-existing residual stress generated during deposition [231]. Due to the piezoelectric nature
of GaN, another source of mechanical stress could be the inverse piezoelectric effect due to the
large values of the electrical field at off-state condition. Again, both high kinetic energy or hot
electrons [230] and diffusion causes [232] electrical and structural degradation of the device.
99
The motivation for this study comes from the opportunities for identifying external stimuli
effects on off-state degradation and failure [228]. In our present study, we apply large reverse
bias at the gate to operate the device at off-state mode. During off-state biasing thermally
activated mechanisms are known to be absent or insignificant due to the little or no current
output [233, 234]. Thus, moderate to large-scale defect forms [233, 235, 236] due to the inverse
piezo-electric stress. During off-state biasing condition degradation of AlGaN/GaN HEMTs is a
complex function of the gate bias (with respect to pinch-off), drain bias (with respect to
breakdown voltage) as well as the device geometry and more importantly, the crystallographic
quality of the device layers [237].
Typically, device degradation and failure mechanisms are inferred from the characterization
data obtained from time and stress-controlled tests followed by computational modeling and
microscopy. For example, electroluminescence [229, 232], Raman microscopy [226], and
transmission electron microscope (TEM) has been widely used to visualize the lattice defect and
diffusion induced damages [230, 235, 238, 239]. TEM offers the highest possible spatial
resolution in through-the-thickness imaging with additional features, such as selected area
electron diffraction (SAED), energy dispersive x-ray spectroscopy (EDS) and electron energy
loss spectroscopy (EELS) to identify crystallographic, chemical and electronic states and
damages. In the following sections we will demonstrate external stimuli effects on AlGaN/GaN
HEMTs at off-state using in-situ TEM study [240].
4.2.2 Materials and Methods
In our present study, we have used a 6W, 18GHz and 40V rated depletion mode GaN
HEMTs on silicon-carbide (Wolfspeed, CGHV1J006D). The specification of the device is as
100
follows: gate threshold voltage, saturated drain current, on-resistance and drain-source
breakdown voltages are -3V, 1.1A, 2.3 and 100 V respectively. In our present study, we used
both semiconductor parametric analyzer (Keithley 2400) and in-situ TEM to investigate the
xetrnal stimuli effects on device degradation. In-situ TEM study requires an electron transparent
sample. Figure 4.8a shows a scanning electron microscope (SEM) micrograph of such TEM-
ready specimen. The specimen preparation involves 100nm thin electron transparent sample
Figure 4.8. (a) The experimental setup showing a MEMS chip on a TEM specimen holder, (b)
The specimen integrated with the MEMS chip, (c) SEM image of the electron transparent GaN
HEMT specimen, and (d) transfer characteristic of the HEMT die, and (e) Off-state loading of
the 100 nm thick HEMT sample [241].
101
preparation using focused ion beam milling (FIB) and manipulation of this film to a micro-
electro-mechanical system (MEMS) chip equipped with mechanical actuators, heaters and
electrodes. Figure 4.8b shows a typical HEMT specimen integrated with the MEMS chip. The
custom-made chip was then mounted on to a commercially available TEM specimen holder with
electrical biasing capability, which is shown in Figure 4.8c. The in-situ TEM experiments were
performed in a 200 KV FEITM Talos F200X TEM with 1.2 Å resolutions.
4.2.3 Results and Discussion
Transfer characteristics as shown in Figure 4.8d suggests gate threshold voltage (Vth) is -
3.1V for depletion mode HEMTs. Thus to study off-state response, we kept the gate voltage
(VGS) below -3.1V i.e., at -5V. In this study, we performed the experiments in two phases, which
includes die-scale as well as 100 nm thin film of HEMTs characterization. These outcomes are
represented in Figures 4.9a and 4.9b respectively. A comparison between the die-level data and
Figure 4.9. Off-state characterization of (a) die-level transistor, and (b) 100 nm thick HEMT
during phase I loading [241].
electron transparent specimen data are shown in Figure 4.9. The higher drain current output of
100 nm thick specimens indicates higher leakage in the electron transparent sample compared to
102
the die. This can be realized by the difference in the boundary conditions and the higher surface
to ratio of thin films compared to the die. Surface defects in GaN HEMTs can contribute to
higher leakage current compared to the bulk. Due to the nature of our thin film specimen
geometry surface defects are unavoidable, and is expected to further increase the leakage.
Phase II experiments involve the loading of electron transparent HEMT specimens inside
the TEM until it fails under high negative stimuli applied at the gate electrode. Outcome of this
condition is shown in Figure 4.8e. In our study, the electron transparent HEMT specimens
survived up to 30V drain bias prior to the catastrophic failure. This value is in well agreement
with the 100 V breakdown rating of the die with gate width of about 200 μm.
Both electrical field and the inverse piezo-electric stress are prominent at the drain side of
the gate edge. Due to the presence of gate fieldplate and small reverse bias applied at the gate,
we anticipated degradation and failure mechanisms that are not common in the literature, such as
the source injection and the surface hopping conduction. Figure 4.10 shows source-injected
degradation that leads to the failure in the source-gate region. As shown in Figure 4.10a the
virgin specimen with pre-existing defects. Figure 4.10a identifies two hubs for defect nucleation
as marked with an oval and a rectangle at the drain and source side respectively. Figures 4.10b-
4.10d show the specimen during the various phases of loading. With the increment of drain bias,
we notice an appreciable amount of changes in the dislocations network and structure
(comparing the same region such as oval or rectangle area). Figure 4.10e shows the output
characteristics, with the locations of each of the TEM snapshots. These changes indicate that
with TEM it is possible to capture defect nucleation and propagation events under external
stimuli loading.
103
Figure 4.10. Bright field TEM images acquired at drain voltages: (a) 0V, (b) 7.2V (c) 11.6V and
(d) 23V, and (e) corresponding drain current vs. drain voltage data at gate voltage -5V for a 100
nm thick GaN HEMT specimen [241].
104
Source injection dominated failure has been shown in Figure 4.10d. Interestingly, the drain
side showed lattice degradation as well, whereas region underneath the gate remains mostly
intact. This phenomenon is attributed to the competing leakage mechanisms such as source-gate,
gate-drain and source-drain, which are influenced by the test (biasing) protocol as well as device
dimensions and quality of the materials and interfaces [242] Additionally, incorporation of a
field plate reduces the electrical field strength under the gate. Test conditions under high external
stimuli (in this case gate voltage below threshold value), show breakdown primarily from gate-
drain leakage. This is not the case represented in Figure 4.10 since VGS and Vth are about -5V
and -3V respectively. Thus, source injection governs the failure mode as confirmed by Figure
4.10e. This is also known as premature breakdown in the literature [242].
The true potential of in-situ TEM technique can be realized by predicting the failure
behavior for a given state of pre-existing defects as demonstrated by a separate set of experiment
in Figure 4.11. We notice dislocations network right under the source electrode marked in Figure
4.11a with arrow 1. Additionally, two more sets of potential defect structures were present
between the gate and the drain (marked with arrows 2 and 3). Due to this pre-existing defects we
assumed a punch-through [228] failure mode during off-state biasing, which actually took place
at premature drain voltage of 3V (the gate voltage fixed at -5V). Figure 4.11b manifests the
HEMT specimen right after the punch-through, where the leakage is also shown to activate the
dislocations in the locations marked by arrows 2 and 3, which has further moved under the gate.
This activated dislocation motion increases the source-drain leakage, and a high current density
The electron transparent thin GaN HEMT specimen yields very high current density in the
sample, which can melt the metal contact at the drain and subsequently damages the interface
with the AlGaN. Figure 4.11d shows the thin HEMT specimen at current density of 2000
105
mA/mm. The drain electrode forms a pool marked by arrow t1, which takes about 6.2 seconds to
diffuse completely through the GaN layer. This state is shown with arrow t2 in Figure 4.11b.
Figure 4.11. Real-time operation of GaN HEMT: (a) electron transparent HEMT specimen prior
to the loading, where arrows indicating pre-existing defects, (b) magnified view of the arrows
marked 2 and 3 at the on-set of source-drain leakage, (c) molten drain metal pool at this instant
the drain side at current density of 2000 mA/mm, (d) metal diffusion through the GaN layer, (e)
rapid breaching of the GaN-SiC interface, and (f) degradation of the SiC layer [241].
106
Since an interface offers the weakest resistance to diffusive transport of atoms, thus molten metal
rapidly penetrates the GaN-SiC interface. Figure 4.11e shows the moment when the interface is
breached, and the near-amorphous metal starts diffusing through the substrate SiC layer. The
substrate is also completely degraded by diffusion, albeit at a slower rate than the rate observed
for the previous phenomena.
Chemical elemental mapping capability of the TEM can provide us useful information on
Figure 4.12. EDS mapping of a failed HEMT specimen at the (a-d) gate and (e-h) drain areas. (i,
j) normalized weight percentage of the various elements in the gate and drain area respectively
[241].
107
multi-element diffusion processes. Energy dispersive (EDX) spectroscopy mapping is shown in
Figure 4.12 for both gate and drain regions. At the gate region (Figures 4.12a-4.12d), we noticed
diffusion of gallium and nitrogen through the gate. The gate region also shows oxygen, most
likely trapped during specimen handling in the ambient. Since the current density is lower during
off-state operation, temperature is also expected to be lower, which implies that the inverse
piezo-electric stress can enhance diffusion. Interestingly, we did not notice nitrogen diffusion at
the drain side (Figures 4.12e-4.12h). Figures 4.12j highlighted the compositional changes at the
gate and drain area respectively. This analysis will enable us to study the feasibility of this
technique to map the diffusion and predict the degradation in the HEMTs.
4.2.4 Conclusion
We summarize that degradation and failure mechanism in GaN-based HEMTs (and
microelectronic devices in general) devices are complicated phenomena. Our off-state failure
experiments demonstrate the feasibility of monitoring dimensional and microstructural aspects of
electronic devices and connect that to their performance and reliability. However, experimental
methods need to be improved for minimum contamination from the specimen preparation and
handling. Additionally, modified boundary conditions of an electron transparent thin specimen
require scaling to the standard device geometry through modeling and experimentation. Finally,
care must be taken to avoid the electron beam damage during the operation of the electronic
devices. These tasks are challenging, however the potential benefits of visualization of
operational and failure phenomena in real time can provide insights on device degradation
mechanism under external stimuli.
108
Chapter 5
Ion Irradiation and External Stimuli Effect at Nanoscale
Contents of this chapter are based on the following journal articles:
Zahabul Islam, Angela L. Paoletta, Anthony M. Monterrosa, Jennifer D. Schuler, Timothy J.
Rupert, Khalid Hattar, Nicholas Glavin, Aman Haque, Heavy ion irradiation effects on
GaN/AlGaN high electron mobility transistor failure at off-state, Microelectronics
Reliability,Volume 102, Pages 113493 (9pp), 2019.
Author of this dissertation designed the experiment, performed the sample preparation,
device fabrication, experimentation, data analysis and manuscript writing. Angela L.
Paoletta performed the die-level transistor characterization accompanied by Zahabul
Islam. Anthony M. Monterrosa, Jennifer D. Schuler and Khalid Hattar conducted
irradiation experiment at Sandia national Laboratory and involved in manuscript writing.
Timothy J. Rupert guided on manuscript writing. Aman Haque guided on experiment
design, and involved in data analysis as well as manuscript preparation.
109
5.1 Irradiation Damage and Degradation in Nanoscale Transistor
In this chapter, we will discuss the ion irradiation effects on AlGaN/GaN high electron
mobility electron transistors (HEMTs) using in-situ transmission electron microscopy (in-situ
TEM). The experiments were performed inside an electron microscope (EM) to visualize the
defects, microstructure and interfaces of ion irradiated HEMTs during operation. Experimental
results exhibits heavy Au4+ ion induced different types of defects such as vacancies, interstitials
and dislocations in the device layer. It is hypothesized that these defects act as carrier traps in the
device layer and the resulting charge accumulation lowers the breakdown voltage under external
stimuli. Sequential energy dispersive X-ray spectroscopy (EDS) mapping allows us to track
individual chemical elements during the experiment, which further suggests that electrical
degradation in the device layer may originate from oxygen and nitrogen vacancies.
5.1.1 Objective and Motivation
GaN based high-electron mobility transistors (HEMTs) are potential candidates for next
generation electronics such as power amplifiers, broadband communication and high-voltage
switches due to their high breakdown voltage and wide-bandgap [243, 244]. The high carrier
density and mobility of the two-dimensional (2D) electron gas (2DEG) channel yield low on-
resistance which further reduces switching losses when operated in switching mode power
converter [245]. In addition, the wide bandgap of GaN allows operation under high electrical and
temperature fields [246] which makes them an attractive choice for harsh environment
applications [247, 248].
Due to their size, weight and power effectiveness, GaN HEMTs are attractive for space
applications. However, high energy particles featuring the cosmic rays or solar flares in space,
110
with energies up to 100 MeV for the protons, and up to 10 GeV for heavy ions [249] can create
electron-hole pairs and displace atoms from their original lattice position, leaving vacancies,
interstitials and dislocations in the crystal [250]. Over time, these defects further accumulate and
interact with each other to generate stacking faults, dislocation loops, and vacancy/interstitial
clusters [251, 252]. Such microstructural degradation degrades device performance [253].
Literatures also suggest lower carrier density and mobility in the 2DEG sheet [254] and a
decrease of the Schottky barrier height at the gate [255].
Ionization effects in GaN are not severe due to the absence of gate dielectric in the
HEMT structure and the higher surface state density in GaN [256], which makes them superior
candidate for radiation tolerant electronics compared to silicon. Measurement of transport
properties of proton-irradiated GaN thin films and devices corroborates the radiation tolerant
properties of GaN compared to GaAs [247, 248, 257, 258].
However, literatures also report the evidence of a decrease in DC saturation current and
transconductance at a fluence of 1014 𝑐𝑚−2 protons or even at lower dose such as 1012 𝑐𝑚−2
[256, 259]. Thus exact mechanism(s) for irradiation-induced degradation and their signatures in
the transistor characteristics demand further investigation. The current trend in the literature is to
measure electrical properties using ex-situ techniques [260]. However, pinpointing the exact
mechanism using ex-situ techniques is strenuous, where in-situ microscopy can be of tremendous
help. Ex-situ studies are post-mortem in nature and can indicate the extent of radiation damage,
but it cannot capture their interaction with the defects generated due to the high electrical,
mechanical and thermal fields during HEMTs operation. This situation can be expected to be
exacerbated with the presence of radiation induced defects. Thus, we suggest that in-situ
microscopy could be helpful in resolving these concerns.
111
In the folowing section, we will demonstrate in-situ electron microscopy such as in-situ
transmission electron microscopy (in-situ TEM) technique for simultaneous quantitative and
qualitative Study of GaN HEMTs under external stimuli. In this study, we havecharaterized both
pristine and Au4+ irradiated GaN HEMTs die and electron transparent coupons using a
semiconductor parametric analyzer and TEM respectively. Our in-situ study not only provides
invaluable information on heavy ion induced defects geneartion in GaN HEMTs, but it also
predicts the performance and degradation mechanism of the device under heavy ion irradiation in
a harsh environment. Due to the nature of in-situ TEM study it allows us simultaneous
quantitative characterization and qualitative visualization during real time operation of HEMTs.
5.1.2 Materials and Methods
To study the effect external stimuli response on ion irradiated HEMTs’ electrical
performance and failure, we designed a series of experiments described as follows:
(a)At first, we characterized commercially available depletion mode GaN HEMTs
(Wolfspeed, CGHV1J006D rated at 6W, 18GHz and 40V) at die level using a Keithley 2400
semiconductor parametric analyzer at room temperature.
(b) In the next step, pristine HEMT dies were irradiated normal to the surface with 1.5
MeV Au+ ions using the 6 MV HVE Tandem accelerator at Sandia National Laboratories. The
ex-situ ion irradiation was performed on three dies with fluences approximately 6.5 x 1013, 6.5 x
1014, and 6.5 x 1015 ions/cm2. The ion energy was chosen based on a Stopping and Range of Ions
in Matter (SRIM) simulation [261] to have a relatively uniform damage profile in the device
layers. Irradiation damage level in term of dpa were 0.45, 4.5, and 45.0 dpa for each respective
112
fluence level. The irradiated dies are then electrically characterized using a Keithley 2400
semiconductor parametric analyzer to compare the die-level performance degradation.
(c) After characterization at die-level we prepared electron transparent (~100 nm thick)
HEMT specimens using focused ion beam (FIB) as described in earlier section [212, 240]. These
thin electron tarnsparent HEMT specimens were then in-situ ion-irradiated inside a TEM [262]
as mentioned in step (b) to obtain visual description of radiation induced defect generation.
However, in our present study we consider sample with highest dose for in-situ TEM study. Both
pre-irradiated and post-irradiated electron transparent HEMT specimens were characterized for
DC transfer and output characteristics inside a TEM. The specimens are electrically biased untill
they degrades.
During electron transparent specimens preparation, special care was taken to minimize
the FIB damage and to keep the three electrodes (i.e drain, gate and source) intact. The transfer
process of thin specimen from a custom TEM grid to a micro-electro-mechanical (MEMS)
device also requires low accelerating voltage i.e., 5kV and ion beam current exposure to avoid
any damage in the device layer. Figure 5.1b shows a mounted chip on a TEM specimen holder
with electrical biasing capability. Details of the sample preparation and transfer technique has
been described in the earlier section i.e., 4.1.2. Figure 5.1c shows the electron transparent thin
sample on the TEM grid prior to the mounting on the MEMS chip. We performed in-situ TEM
experiments in a 200 kV FEI Talos F200X S/TEM with 0.12nm resolution.
113
Figure 5.1. Experimental setup for in-situ TEM experiment of electron transparent HEMTs: (a)
GaN HEMT die, (b) a MEMS chip with the HEMT specimen mounted on in-situ TEM electrical
biasing holder, and (c) FIB lamella of the HEMT before mounting on to the MEMS chip [28].
High resolution transmission electron microscopy (HRTEM) allows us to estimate the
atomic-scale strain in the sample using geometric phase analysis technique (GPA) [263]. In the
GPA technique, phase image, 𝑃𝑔 can be expressed by the component of the displacement field,
𝒖(𝒓), in the direction of the reciprocal lattice vector g as described below:
𝑃𝑔(𝒓) = −2𝜋𝒈𝒖(𝒓) (5.1a)
Thus, a two-dimensional displacement field can be obtained using Eq. (5.1a) by choosing
two independent phase images (Pg1 and Pg2):
𝒖(𝑟) = −1
2𝜋[𝑃𝒈𝟏(𝒓)𝒂𝟏 + 𝑃𝒈𝟐(𝒓)𝒂𝟐] (5.1b)
(𝑢𝑥𝑢𝑦
) = −1
2𝜋(
𝑔1𝑥 𝑔1𝑦
𝑔2𝑥 𝑔2𝑥)
−1
(𝑃𝑔1
𝑃𝑔2) (5.1c)
Where 𝑃𝒈𝟏 and 𝑃𝒈𝟐 are two phase images, 𝒂𝟏and 𝒂𝟐 are lattice vectors in real space, 𝑢𝑥
and 𝑢𝑦 are displacement fields, 𝑔𝑥 and 𝑔𝑦 are two components of vector g in reciprocal space.
114
Once displacement field is known, atomic strain maps were obtained from elasticity theory using
the following equations:
𝜀𝑥𝑥 =𝜕𝑢𝑥
𝜕𝑥 (5.1d)
𝜀𝑥𝑦 =1
2(
𝜕𝑢𝑥
𝜕𝑦+
𝜕𝑢𝑦
𝜕𝑥) (5.1e)
𝜀𝑦𝑦 =𝜕𝑢𝑦
𝜕𝑦 (5.1f)
Where 𝜀𝑥𝑥 , 𝜀𝑥𝑦 and 𝜀𝑦𝑦 are normal strain in x direction, shear strain in x-y plane and normal
strain in y direction respectively.
5.1.3 Results and Discussion
5.1.3.1 Die-level Irradiation Effects: Figure 5.2a shows the schematic cross-sectional view of
the die and irradiation direction (downward arrowhead), whereas Figure 5.2b shows the SRIM
simulation results on the damage level in displacement per atom (i.e., dpa) units as function of
depth into the device. Bright field (BF) TEM images of a pristine and one ion irradiated HEMT
at 45 dpa as shown by Figure 5.2c and 5.2d respectively, indicate ion irradiation effect in the
device layer. Figures. 5.2c and 5.2d exhibit dislocation density comparison between 45dpa
irradiated sample and pristine counterparts to estimate the relative difference in dislocation
density. We noticed significantly higher (more than 100x) dislocation density in the irradiated
HEMTs compared to the pristine one, which is related with the degraded performance measured
by the parametric analyzer.
115
In our experiment, we first measured the performance of pristine and irradiated HEMTs
in their bulk form. Figure 5.3 shows both transfer and output characteristic curves of pre- and
post-irradiated GaN HEMT die. Transfer characteristics curve suggested that threshold voltage
(Vth) for a pristine HEMT is approximately -3.1V. However, for all irradiated HEMTs we do not
see any significant increase in output current compared to the pristine counterpart during the gate
voltage increment. This suggested that ion irradiation significantly damage 2DEG channel. In
our present study, we kept the gate voltage (Vg) at -5 V to ensure off-state condition. Figure 5.3b
shows output characteristics curve for all irradiated dies at off-state operating condition. Here,
Figure 5.2. (a) Downward arrowhead in the schematic diagram of the GaN HEMT showing
irradiation direction, (b) displacement per atom (dpa) profile as a function of depth for different
116
doses of irradiation, (c) TEM image of a pristine HEMT showing mostly bend contours, and (d)
TEM image of an irradiated HEMT at 45 dpa showing very high dislocation density [28].
four orders of magnitude reduction in output current (for the highest damage level i.e., 45 dpa)
indicates ion irradiation can create significant amount of defects/structural damage and thus
degrade device performance. This drastic reduction in output current for all irradiated devices is
a clear indication of defect introduction in the AlGaN layer as well as the GaN layer during the
Au4+ ion irradiation, which lead to lower carrier density and mobility in the 2DEG channel.
Figure 5.3. Die-level HEMTs specimens characterization curve as function of ion irradiation
damage in dpa: (a) transfer characteristics, and (b) output characteristics [28].
5.1.3.2 In-situ Ion Irradiation Effects (No Biasing): The objective of this set of experiments
was primarily to monitor and identify the irradiation-induced defects in the lattice and interfaces
of the HEMT system. Formation of defects due to the ion irradiation influences the device
performance, since the HEMTs devices experience a very large amount of thermal, electrical and
mechanical stresses during operation. Figure 5.4 shows TEM micrographs of a specimen before
and after exposure to 2.8 MeV Au4+ ion species for 60 minutes to a fluence of 4x1014 ions/cm2.
117
Micrograph shows shifts in sample contrast, particularly at the locations of pre-existing
dislocations marked by dotted rectangular region. The effect of ion bombardment is pronounced
right under the gate, where the mechanical stress is also the highest. During an irradiation
process, interstitials atoms might leave their regular lattice sites and occupy interstitials sites.
Figure 5.4. Electron transparent HEMT device: (a) before, and (b) after 2.8 MeV Au4+ ion
irradiation for 60 minutes to a fluence of at 4x1014 ions/cm2. The rectangular dashed box shows
contrast change due to point defect accumulation, while the arrows indicate dislocation activities
at the GaN-SiC interface [28].
These interstitials atoms might appear as dark contrast in a BF TEM image [34]. Thus, the
contrast change (dashed rectangular box in Figure 5.4b) is an outcome of increased number of
interstitials defects. We also noticed remarkable irradiation effects at the GaN-SiC interface.
Figure 5.4b shows how the initially sharp interface is deteriorated due to the introduction of
118
dislocations (indicated by arrows). However, these dislocations might be an outcome of the
radiation damage or resulted as a strain relief mechanism due to interface stress from the
radiation damage.
5.1.3.3 In-situ TEM Electrical Characterization: In addition to die-level characterization, we
also conduct experiments on the electron transparent specimens inside the TEM. We prepared
the specimens from the ex-situ irradiated dies because of the well-characterized nature of the
irradiation simulation and boundary conditions. The objective is to visualize the defect and
damage evolution processes under external stimulus such as electrical biasing. Major advantages
of in-situ analytical TEM study is simultaneous quantitative (device characterization) as well as
qualitative (microscopic visualization) characterization during real time operation. Electron
transparent specimens have different aspect ratio and boundary conditions compared to the die-
level specimens [241] thus their transfer and output characteristics are not numerically identical
to HEMT die. However, their characteristics functions follow similar trend, which suggests that
our in-situ TEM experiments can be useful for capturing the mechanics and physics of
degradation under external stimuli.
Figure 5.5a shows output characteristics of an irradiated sample during failure test. Data
shown here represents the transistor irradiated with highest damage (45 dpa). We observed that
drain current increases after each loading step up to a drain bias of 8V. However, after 8V drain
bias, the current shows large fluctuations prior to the failure at 10.2V bias. We also labeled 5
distinct data points (a, c, d, e, f) on Figure 5.5a, for which the corresponding bright-field (BF)
TEM images are presented in Figure 5.6. To identify the influence of irradiation on device
performance we have compared a pristine and an irradiated HEMT sample as shown in Figure
5.5b. We noticed more than two times higher operating drain voltage, and one order of
119
magnitude higher output current for pristine sample compared to the irradiated sample. This
degradation in device performance after irradiation is attributed to the defects formation during
an irradiation process [259, 260].
Figure 5.5. (a) Drain current vs. drain voltage plot of electron transparent GaN HEMT specimen
during off-state operation inside the TEM. The data labels correspond to the in-situ TEM images
in Figure 5.6., and (b) Comparison between pristine and irradiated conditions [28].
Figure 5.6 represents BF TEM images at different loading condition during the off-state
failure tests. These low magnification images (except Figure 5.6b) show numerous bend contours
as well as a number of defect clusters. Examples of these heavy Au4+ ion irradiation induced
defects such as dislocation is captured at very high magnification and is shown in Figure 5.6b.
Atomic resolution imaging enables us to observe individual dislocation indicated by red colored
perpendicular sign () in the sample. However, we did not notice any significant amount of
defects in pristine dies. Figure 5.6c shows the HEMT specimen at a drain voltage of 5V, which is
large enough to trigger very minute changes in the drain current (as shown by point c in Figure
5.5a). This drain voltage i.e., 5V eventually accelerates microstructural damage nucleating from
preexisting defects. For irradiated specimens, irradiation damage can trigger such nucleation.
120
Figure 5.6 also shows bend contours that arise from elastic bending during specimen preparation
(due to the mechanical constrains at the drain, gate and source electrode).
Figure 5.6. TEM BF images showing source, gate and drain at the same time. Drain voltage: (a)
Vd= 0V, (b) TEM image of screw dislocations in buffer layer, (c) Vd= 5V, (d) Vd= 7V, (e) Vd=
8.5V, and (f) Vd= 10.2V [28].
Low magnification micrographs as shown in Figures 5.6c-5.6f capture the microstructure
at the drain, gate and source at different drain bias as indicated on Figure 5.5a. Three arrows as
shown in Figure 5.6c (labeled 2, 3 and 4) identify the location of defect clusters that appear to
reproduce dislocations with current stressing. As we increased the drain bias, significant
microstructural changes are observed in the GaN layer. To identify damages those extended from
the arrows 2 and 4, we marked two dashed rectangular areas are shown in Figure 5.6e. Later on,
we noticed interfacial breaching in between these two locations. Figure 5.6f shows micrograph
of the sample just after the failure at 10.2V drain bias. Lower breakdown drain bias of the
irradiated sample could be attributed to defects generated at the interface of AlGaN layer as well
121
as GaN buffer layer during the ion irradiation. This scenario is exacerbated by the pre-existing
irradiation induced defects, and inverse-piezoelectric stress induced dislocations during the off-
state loading [233, 236]. These studies also indicate that the resulting mechanical stresses create
defects induced traps in the GaN layer through local charge accumulations. The contribution of
our study is therefore provide visual evidence of the lattice defects generation (Figure 5.2d and
5.6b) and interfacial (Figure 5.4b) defects, which lower the breakdown strength [264] and
facilitates failure through their percolation [265].
High-resolution in-situ TEM imaging of defects allows us to track individual defect and
thermo-mechanical strain in the sample during the real operation (as shown in Figure 5.7). For
example, Figure 5.7a exhibits breakdown of the buffer layer and generated dislocations at the
drain-gate channel. Figure 5.7b and 5.7c represent BF low magnification and high-resolution
TEM image of dislocations in the buffer layer respectively. These defects could act as surface
traps in the buffer layer, and percolation of these defects possibly lowers the breakdown voltage
due to the stress field around individual dislocation. To validate stress field associated with a
dislocation, in our present study we quantified atomic strain in the sample by employing GPA
technique as shown in Figure 5.7c. The atomic scale strain mapping process involves several
steps as follows: (a) Fast Fourier Transformation (FFT) of HRTEM image, (b) selection of two
diffraction spots along lattice direction, (c) reconstruction of image using inverse FFT (IFFT),
and (d) phase image, displacement field and strain field calculation using eqns. (5.1a), (5.1c) and
(5.1d-5.1f). Calculated normal strain, shear strain and corresponding lattice fringes of the sample
with dislocations (as marked by white upward arrow) are shown in Figures 5.7d-f. Dislocations
in the sample are discernable even by visual inspection as shown in Figure 5.7c and 5.7f. On the
other hand, atomic strain mapping as shown in Figure 5.7d and 5.7e allow us to identify
122
individual dislocation in the sample and associated strain field. In our present study, we notice
basal plane slip (i.e., [0001] plane) associated with [1120] prismatic dislocation. At the location
of dislocations, we notice the existence of both tensile and compressive strain field as shown in
Figure 5.7d and 5.7e, which further introduces stress field around a dislocation.
Figure 5.7. (a) TEM BF image at the drain side of drain-gate region, (b) Dislocations in the GaN
layer, (c) High-resolution TEM (HRTEM) image of dislocations in GaN layer, (d, e) atomic
strain mapping in the sample showing normal and shear strain field associated with individual
dislocations, and (f) simulated lattice fringes with the dislocations [28].
We also monitored real-time diffusion of chemical elements during the experiments using
EDS. Figure 5.8 shows such an experiment to capture chemical elements diffusion at different
drain bias. Figure 5.8 indicates nitrogen deficiency under the gate electrode. Literatures also
suggests that nitrogen vacancies are the primary defect forms in GaN [266] under ion irradiation
123
due to the lower displacement energy of nitrogen (N) atom compared to gallium (Ga) [267]
atom. This defect could act as a trap and reduce the conductivity in GaN [268], which in turn
reduces the carrier density and, increases carrier scattering in the 2DEG channel at the interface
of AlGaN/GaN. Thus, accelerated breakdown of an irradiated device could be attributed to these
nitrogen vacancy defects. The breakdown zone in the sample indicates nitrogen concentration
deficiency in GaN layer (as shown in Figure 5.8b). Figure 5.8c shows the quantitative data on
relative changes in the weight percentage of other elements in the region between the gate and
the drain (the dotted rectangular box in Figure 5.8a). We notice reduction of both Ga and N after
breakdown, which could be attributed to the diffusion of GaN buffer layer out of rectangular box
and possibly into the SiC substrate layer. The reduction of nitrogen is more prominent than
gallium. On contrary to this, weight percentage of gold increases after failure, which could be
attributed to the diffusion from the gate electrode to the buffer layer.
Figure 5.8. EDS mapping of GaN HEMT showing diffusion of chemical elements at different
drain bias: (a) Vd=0V, (b) onset of failure at Vd=10.2V, and (c) relative changes in diffusion of
chemical elements obtained from EDS at these two voltages [28].
124
To summarize, we demonstrated an in-situ TEM technique for visualizing nucleation of
ion irradiation induced defects as well as their proliferation during electrical stressing of GaN-
based HEMTs. Ion irradiation generates significant number of dislocations that contributed to
gate leakage, which further led to the failure.
5.1.4 Conclusion
In this study, we have investigated the effects of heavy ion irradiation on defects evolution
and performance of GaN HEMTs under external stimuli. Heavy ion (such as Au+) irradiation
generates lattice defects such as vacancies and dislocations, which degrade device performance
and accelerate permanent damage. We observed increased defect concentrations in the device
layer with the increment of irradiation dose. Study shows that defects could act as traps thus
degrade the carrier density and mobility in the 2DEG channel. Device degradation could be
attributed to the gate injection induced impact-ionization in the channel, which in turn dictates
the off-state failure. Oxide formation near the breakdown region of the channel layer suggests
that chemical oxidation also accelerate/partially contribute to the electrical degradation.
Continuation of this study in future will include (a) scaling the physics of degradation from
electron transparent device to die-level counterparts, and (b) investigation of the synergistic
relationship between defects and diffusion which leads to degradation and failure. On-state
device reliability is also required to comprehend fundamentals of the ion irradiation damage in
wide bandgap electronics.
5.2 Low Temperature Stimuli Synergy on Recovery of Irradiation Damage in Thin Films
In this section, we will explore the effectiveness of external stimuli such as electrical current
induced electron wind force (EWF) in annihilating defects originating from ion irradiation. In
125
this present study, self-ion irradiation to a dose of 5 x 1015 ions/cm2 (45 displacement per atom
(dpa)) was used to generate high density of displacement damage in a nanocrystalline gold
sample. Afterwards, an electron transparent specimen was prepared form bulk irradiated gold
sample to investigate external stimuli effects on ion irradiation defects using in-situ transmission
electron microscope (in-situ TEM). Unique geometry and boundary condition of electron
transparent specimen effectively decoupled Joule heating from the electron wind force (EWF),
thus allowing us to study the EWF effects on defects annihilation. Our study shows that EWF
can impart significant defect mobility even at low temperature, resulting in the migration and
elimination of defects in a few minutes. We propose that (EWF) interacts with defects to create
highly glissile Shockley partial dislocations, which makes the process fast and energy efficient.
This effect is even more pronounced for nanocrystalline materials with large fraction of grain
boundaries (GBs) and surface area, which subsequently act as active sinks for the migrating
defects.
5.2.1 Objective and Motivation
Irradiation is a process where highly energetic particles (ions, neutrons, electrons) collide
with atoms in materials, energizing them to be displaced from their original lattice positions,
thereby generating various types of defects [269]. For example, point defects, such as vacancies
and interstitials are prominent and they further agglomerate, migrate and form defects clusters
amorphous zone, dislocations loops, three-dimensional defects, ripples, adatoms, craters, to name
a few [270, 271]. These defects can significantly degrade materials properties due to the
embrittlement, swelling and hardening [272, 273]. The literatures have primarily focused on
radiation effects on microstructure and properties of materials using both experimental and
computational techniques [272, 274, 275]. This has led to foundational understanding of the
126
evolution and mechanics of defects. However, processes for defect annihilation are not studied,
other than high temperature annealing [276, 277].
In this study, we propose electron wind force (EWF) as a non-thermal route to actively
control defects and microstructure in an irradiated nanocrystalline materials. During EWF
annealing a highly localized mechanical force imparted to defects due to the collision and
momentum transferred to the defective atoms by the electrons [278]. EWF is strongly coupled
with Joule heating when electrical current is passed through metals [279]. Electro-thermal effects
have been demonstrated to control defects and microstructures [48, 280, 281] in metallic
materials. In our present study, our sample geometry allows us to decouple thermal contribution
from EWF. We achieve this by making a thin film specimen freestanding, so that the regions
near the edges are constrained to lower temperatures due to the missive silicon heatsinks attached
to it. Our hypothesis is that the high specificity of defects can make EWF a fast and low
temperature driver for defect annihilation. Nanocrystalline materials are attractive from this
perspective since they contain large volume fraction of grain boundaries (GBs), phase
boundaries and free surfaces that act as defects sinks [269, 271, 282].
5.2.2 Materials and Methods
In our present study, polycrystalline gold thick films were irradiated normal to the surface
with 1.5 MeV Au+ ions using the 6MV High Voltage Engineering (HVE) Tandem accelerator at
the Sandia National Laboratories. The fluence was approximately 6.5 x 1015 ions/cm2 and
corresponding displacement per atom (dpa) was 45 dpa as shown in Figure 5.9a. After ion
irradiation we used Focused Ion Beam (FIB) based lift out technique using FEI Helios Nanolab
660 dualbeam with a working distance of 4.0mm to prepare nominally 100nm thick sample as
127
mentioned in section 4.1.2. Then we transfer electron transparent specimens on a custom
designed MEMS device, and integrated it with in-situ TEM holder (Figure 5.9b). In-situ TEM
experiments were performed inside a 200 kV FEI Talos F200X S/TEM with 1.2 Å resolution.
We also model defect annihilation phenomena using molecular dynamics (MD) simulation to
qualitatively investigate the atomic scale processes behind defect annihilation. Both specimen
preparation [212] and simulation details can be found elsewhere [189].
5.2.3 Results and Discussion
We passed electrical current through the specimens while observing the low temperature
region inside the Transmission electron microscope (TEM). This is shown with the dotted box in
Figure 5.9c, where the massive heatsinks (electrodes) constrain the temperature to approximately
400K as indicated by our electro-thermal simulation. This temperature rise is sufficiently low
compared to the conventional thermal annealing temperature (~0.5Tm where Tm is homologous
temperature). We choose an individual grain to investigate specific types of defects and their
annihilation mechanism under EWF as shown by TEM bright field (BF) image in Figure 5.9d.
Irradiation induced dislocation lines (green color arrow), dislocation loops (cyan color arrow),
and vacancy cluster (blue color circle) in a specimen are shown in Figure 5.9d. During EWF
annealing we increased the electrical current density in finite steps of 0.5× 105𝐴/𝑐𝑚2 and with a
hold time of 5 minutes. A comparison between Figure 5.10d and 5.9e clearly shows the same
area transformed into a highly ordered crystalline area, corroborated by the selected area electron
diffraction (SAED) pattern in Figure 5.9f. BF images as shown in Figure 5.9e indicates
dislocations annihilation during EWF annealing.
128
Figure 5.9. (a) Displacement per atom (dpa) profile for irradiation dose of 6.5×1015 ion/cm2, (b)
micro-electro-mechanical system (MEMS) device mounted on in-situ TEM holder, (c)
temperature profile obtained from electro-thermal simulation, (d) TEM BF image showing
irradiation damage, (e) BF image showing dislocation annihilation at 9.5×105 A/cm2, and (f)
SAED pattern after EWF annealing.
Figure 5.10a shows a TEM Bright field (BF) image of the 45 dpa irradiated sample
before any EWF processing. It is evident that ion irradiation generates significant damage and
defects in the specimen. The defect types and their annihilation mechansim have been discussed
in details in the following sections from both experimental and computational view points. In
this section, we choose an individual grain to investigate specific types of defects and their
annihilation mechanism under EWF. The area was chosen close to the electrode area as shown
by black dotted box in Figure 5.9c. The simulated temperature in this area is approximately 400
129
K. This is sufficiently low compared to the conventional thermal annealing temperature (~0.5Tm
where Tm is homologous temperature). TEM dark field (DF) image as shown in Figure 5.10f
indicates complete annihilation of defects and transformation towards a highly ordered
crystalline area. To investigate the defects such as dislocation lines annihilation mechanism we
studied dislocation and defects evolution at different current density near the low temperature
region of the sample (Figure 5.9c). Figure 5.10a shows TEM BF image of high defect density
with dislocation lines (green color arrow), vacancy cluster (blue color circle), dislocation loops
(cyan color arrow) and grain boundary (light red color arrow). To unambiguously trace the
defects during EWF processing, we use weak beam dark field (WBDF) imaging technique by
activating diffraction vector = 220. Figure 5.10b-5.10f show such TEM WBDF images at
different current density and corresponding dislocations evolution. Figure 5.10b shows DF image
and dislocation lines before EWF processing. At a current density of 3 × 105𝐴/𝑐𝑚2 we notice
that dislocation lines interact with each other and start to migrate towards the GB as indicated by
cyan color arrow in Figure 5.10c. In Figure 5.10b, the pink color dotted circle encloses three
different types of dislocation lines which further interacts among them under EWF. It is well
known that GBs act as defects sink [271, 283] thus could annihilate dislocation lines. Such an
annihilation is shown in Figure 5.10d near the GB as shown by pink color dotted circle. As the
current density increases dislocation lines further moves towards the GB as indicated in Figure
5.10e. Complete annihilation of dislocation lines is observed at a current density of 9.8 ×
105𝐴/𝑐𝑚2 as shown in Figure 5.10f. Our results indicate annihilation of defects is a non-linear
function of current density, which agrees with recent study on grain growth [189]. Figure 5.10b-
5.10f is an indication of defects mobility at low temperature (~400 K) under the impetus of
EWF.
130
Figure 5.10. (a) Irradiated sample before processing, (b) dark field image of pre-processed
sample showing dislocation lines, (c) dislocation lines interaction during EWF processing at
3.5×105 A/cm2, (d) partial annihilation (pink circle) of dislocation lines, and migration towards
grain boundary at 7×105 A/cm2, (e) migration of dislocation lines towards GB and partial
annihilation of dislocation lines at 9×105 A/cm2, and (f) complete annihilation of dislocations and
defects at 9.8×105 A/cm2.
To investigate how the EWF interacts with the defects and migrate them toward the sinks, we
performed MD simulation. Extensive computational experimentation shows that the fundamental
mechanism is the splitting of defects into Shockley partial dislocation. Figures. 5.11a-5.11c show
this phenomena for dislocations and vacancy clusters, while Figures 5.11d-5.11e are for stacking
fault tetrahedra (SFT). These are highly mobile (through gliding motion) defects [284] that can
131
be swept by the EWF towards the defect sinks. Our experiments took less than 5 minutes to
achieve a defect free single crystalline area where the temperature was constrained below
~400K.
Figure 5.11. Computational results on (a)-(c): Annihilation of dislocations and vacancy clusters,
and (d)-(f): SFT annihilation under EWF.
In low stacking fault energy materials, irradiation generates SFTs that are very difficult to
remove below 873K [285]. Figure 5.11d-5.11f (computational) and Figure 5.12 (experimental)
show such annihilation of SFT during EWF annealing. Figure 5.12a manifests high-resolution
TEM (HRTEM) BF image with high density of SFTs (cyan color arrow and yellow color dotted
triangle). Figure 5.12b represents an atomic resolution BF image of a single SFT defect before
EWF processing. Surface induced SFT annihilation under the effect of EWF is shown in Figure
5.12c. Such annihilations of SFT are marked by cyan color dotted arrow in Figure 5.12c. At
higher magnitude of EWF SFT could further interact with GBs as shown in Figure 5.12d, and
subsequently absorbed by the GBs. Complete defects annihilation is observed at a current density
of 9.5×105 A/cm2 and corresponding BF images is shown in Figure 5.12e and 5.12f. MD
132
simulation trajectory of SFT annihilation under EWF are shown in Figure 5.11d-5.11f, which
indicate SFTs are composed of stair-rod dislocation, and disintegrates to Shockley partial
dislocation during EWF annealing. This Shockley partial dislocation is further annihilated by
GBs as shown in Figure 5.12e. Dislocation density calculation shows that stair-rod dislocations
diminishes after 60ps simulation time as represented by Figure 5.12f.
Figure 5.12. HRTEM images of low temperature processing of irradiated materials: (a) defects in
the irradiated sample before processing, (b) HRTEM image of SFT, (c) surface induced
annihilation of SFT at a current density of 7×105 A/cm2, (d) SFT-GB interaction at 9×105 A/cm2,
(e) after EWF annealing at 9.5×105 A/cm2, and (f) HRTEM image after annealing at 9.5×105
A/cm2.
133
Our in-situ TEM study shows that EWF can introduce significant defect mobility such as
dislocations and SFT mobility even at room temperature. This non-thermal process is explained
from the interaction of defects with the EW, producing very large density of glissile and reactive
(with other defects) Shockley partial dislocations. An important feature of this process is targeted
defects momentum transfer at the non-defective regions, therefore the electrical energy input is
specifically targeted towards the defective regions, resulting in highly efficient ‘just in location’
annihilation of defects.
5.2.4 Conclusion
We studied the effects of external stimuli such as EWF on ion irradiation induced defects
annihilation in polycrystalline gold specimens at low temperature. Using heavy-ion irradiation,
we prepared specimens with significantly high density of dislocation lines, loops, SFTs and
vacancy clusters. We then passed electrical current through the specimens to investigate the
effectiveness of the EWF effect on defect annihilation by decoupling (exploiting the specimen
geometry and boundary conditions) it from the Joule heating. In-situ TEM evidence shows fast
(less than 5 minutes) and efficient transformation to defect-free crystalline regions. The
experimental and computational results support our hypothesis that EWF targets defects in a
highly localized manner to create highly mobile Shockley partial dislocations. The reactivity and
mobility of this species of defects generated by EWF lead to the fast and efficient defect
annihilation. In comparison, thermal annealing takes high temperature and longer time because
all the atoms need to be heated and their mobility is governed by random diffusion.
134
References
1. Chawla, M.A.M.a.K.K., Mechanical Behavior of Materials.
2. Arzt, E., Overview no. 130 - Size effects in materials due to microstructural and
dimensional constraints: A comparative review. Acta Materialia, 1998. 46(16): p. 5611-
5626.
3. Chokshi, A.H., et al., On the Validity of the Hall-Petch Relationship in Nanocrystalline
Materials. Scripta Metallurgica, 1989. 23(10): p. 1679-1683.
4. Kumar, K.S., H. Van Swygenhoven, and S. Suresh, Mechanical behavior of
nanocrystalline metals and alloys. Acta Materialia, 2003. 51(19): p. 5743-5774.
5. Greer, J.R. and J.T.M. De Hosson, Plasticity in small-sized metallic systems: Intrinsic
versus extrinsic size effect. Progress in Materials Science, 2011. 56(6): p. 654-724.
6. Hahn, E.N. and M.A. Meyers, Grain-size dependent mechanical behavior of
nanocrystalline metals. Materials Science and Engineering: A, 2015. 646: p. 101-134.
7. Takata, N., S.-H. Lee, and N. Tsuji, Ultrafine grained copper alloy sheets having both
high strength and high electric conductivity. Materials Letters, 2009. 63(21): p. 1757-
1760.
8. Bo, F., L. Zhixin, and Z. Xing, Effect of grain-boundary scattering on the thermal
conductivity of nanocrystalline metallic films. Journal of Physics D: Applied Physics,
2009. 42(5): p. 055311.
9. Xu, X.F., et al., Length-dependent thermal conductivity in suspended single-layer
graphene. Nature Communications, 2014. 5.
10. Wang, Z., et al., Thermal conductivity of nanocrystalline silicon: importance of grain size
and frequency-dependent mean free paths. Nano Lett, 2011. 11(6): p. 2206-13.
11. Tschöpe, A., Grain size-dependent electrical conductivity of polycrystalline cerium oxide
II: Space charge model. Solid State Ionics, 2001. 139(3): p. 267-280.
12. Lee, K.-R., J.-H. Lee, and H.-I. Yoo, Grain size effect on the electrical properties of
nanocrystalline ceria. Journal of the European Ceramic Society, 2014. 34(10): p. 2363-
2370.
13. du Plessis, A., I. Yadroitsava, and I. Yadroitsev, Effects of defects on mechanical
properties in metal additive manufacturing: A review focusing on X-ray tomography
insights. Materials & Design, 2020. 187: p. 108385.
14. Reznik, A., C. Uzan-Saguy, and R. Kalish, Effects of point defects on the electrical
properties of doped diamond. Diamond and Related Materials, 2000. 9(3): p. 1051-1056.
135
15. Guo, Z., D. Zhang, and X.-G. Gong, Thermal conductivity of graphene nanoribbons.
Applied Physics Letters, 2009. 95(16): p. 163103.
16. Ma, T., et al., Tailoring the thermal and electrical transport properties of graphene films
by grain size engineering. Nature Communications, 2017. 8(1): p. 14486.
17. Min, S.Y., et al., Two-dimensional sheet resistance model for polycrystalline graphene
with overlapped grain boundaries. FlatChem, 2018. 7: p. 19-25.
18. Isacsson, A., et al., Scaling properties of polycrystalline graphene: a review. 2D
Materials, 2016. 4(1): p. 012002.
19. Jiang, H., et al., Two-dimensional materials: From mechanical properties to flexible
mechanical sensors. InfoMat, 2019. n/a(n/a).
20. Barmak, K., et al., Surface and grain boundary scattering in nanometric Cu thin films: A
quantitative analysis including twin boundaries. Journal of Vacuum Science &
Technology A, 2014. 32(6): p. 061503.
21. Dong, H., B. Wen, and R. Melnik, Relative importance of grain boundaries and size
effects in thermal conductivity of nanocrystalline materials. Scientific Reports, 2014. 4:
p. 7037.
22. Kitagawa, K., et al., Ion-irradiation experiment for the fundamental studies of damage
evolution of fusion materials. Journal of Nuclear Materials, 1985. 133-134: p. 395-399.
23. Zinkle, S.J., 1.03 - Radiation-Induced Effects on Microstructure**Prepared for the Oak
Ridge National Laboratory under Contract No. DE-AC05-000R22725, in Comprehensive
Nuclear Materials, R.J.M. Konings, Editor. 2012, Elsevier: Oxford. p. 65-98.
24. Lewis, M.H., The defect structure and mechanical properties of spinel single crystals.
The Philosophical Magazine: A Journal of Theoretical Experimental and Applied
Physics, 1968. 17(147): p. 481-499.
25. Niklas, A., et al., Effect of Microstructure and Casting Defects on the Mechanical
Properties of Secondary AlSi10MnMg(Fe) Test Parts Manufactured by Vacuum Assisted
high Pressure Die Casting Technology. Materials Today: Proceedings, 2015. 2(10, Part
A): p. 4931-4938.
26. Islam, M.Z., M. Mahboob, and R.L. Lowe, Mechanical properties of defective carbon
nanotube/polyethylene nanocomposites: A molecular dynamics simulation study. Polymer
Composites, 2016. 37(1): p. 305-314.
27. Onga, S., et al., Effects of Crystalline Defects on Electrical Properties in Silicon Films on
Sapphire. Japanese Journal of Applied Physics, 1976. 15(S1): p. 225.
28. Islam, Z., et al., Heavy ion irradiation effects on GaN/AlGaN high electron mobility
transistor failure at off-state. Microelectronics Reliability, 2019. 102: p. 113493.
136
29. Zahabul Islam, M., et al., Characterization of the thermal expansion properties of
graphene using molecular dynamics simulations. Journal of Physics D: Applied Physics,
2013. 46(43): p. 435302.
30. Borca-Tasciuc, T., et al., Thermal conductivity of symmetrically strained Si/Ge
superlattices. Superlattices and Microstructures, 2000. 28(3): p. 199-206.
31. Yasmin, A. and I.M. Daniel, Mechanical and thermal properties of graphite
platelet/epoxy composites. Polymer, 2004. 45(24): p. 8211-8219.
32. Katenkamp, U., H. Karge, and R. Prager, Radiation defects and optical properties of ion
implanted silicon dioxide. Radiation Effects, 1980. 48(1-4): p. 31-34.
33. Li, S.X., et al., Effect of native defects on optical properties of InxGa1−xN alloys.
Applied Physics Letters, 2005. 87(16): p. 161905.
34. Kityk, I.V., et al., Influence of Non-Stoichiometric Defects on Optical Properties in
LiNbO3. Crystal Research and Technology, 2001. 36(6): p. 577-588.
35. Wu, Z., et al., Effects of the microstructure of copper through-silicon vias on their
thermally induced linear elastic mechanical behavior. Electronic Materials Letters, 2014.
10(1): p. 281-292.
36. Collins, P.C., et al., Microstructural Control of Additively Manufactured Metallic
Materials. Annual Review of Materials Research, 2016. 46(1): p. 63-91.
37. Murr, L.E., et al., Fabrication of Metal and Alloy Components by Additive
Manufacturing: Examples of 3D Materials Science. Journal of Materials Research and
Technology, 2012. 1(1): p. 42-54.
38. Olakanmi, E.O., R.F. Cochrane, and K.W. Dalgarno, A review on selective laser
sintering/melting (SLS/SLM) of aluminium alloy powders: Processing, microstructure,
and properties. Progress in Materials Science, 2015. 74: p. 401-477.
39. Hong, J., et al., Exploring atomic defects in molybdenum disulphide monolayers. Nat
Commun, 2015. 6.
40. Najmaei, S., et al., Vapour phase growth and grain boundary structure of molybdenum
disulphide atomic layers. Nat Mater, 2013. 12(8): p. 754-759.
41. van der Zande, A.M., et al., Grains and grain boundaries in highly crystalline monolayer
molybdenum disulphide. Nature materials, 2013.
42. Masafumi Noda, K.F., Yoshihito Kawamura and Tsuyoshi Mayama, Thermal Stability
and Mechanical Properties of Extruded Mg-Zn-Y Alloys with a Long-Period Stacking
Order Phase and Plastic Deformation, in New Features on Magnesium Alloys. 2012.
43. G. F. Carter, G.F.C.a.D.E.P., Materials Science and Engineering.
137
44. Dhaliwal, E.A.S.W.a.E.H.S., A Textbook of Engineering Material and Metallurgy.
45. RESEARCH, N.I.O.I., The Complete Technology Book on Hot Rolling of Steel
46. Kim, M.-J., et al., Electric current-induced annealing during uniaxial tension of
aluminum alloy. Scripta Materialia, 2014. 75: p. 58-61.
47. Sheng, Y., et al., Application of High-Density Electropulsing to Improve the Performance
of Metallic Materials: Mechanisms, Microstructure and Properties. Materials (Basel,
Switzerland), 2018. 11(2): p. 185.
48. Song, H., et al., Self-healing of damage inside metals triggered by electropulsing stimuli.
Scientific Reports, 2017. 7(1): p. 7097.
49. Huang, J., et al., Electropulsing-Induced α to β Phase Transformation of Ti–6Al–4V.
Journal of Manufacturing Science and Engineering, 2019. 141(11).
50. Zhao, J., et al., Multiscale modeling of localized resistive heating in nanocrystalline
metals subjected to electropulsing. Journal of Applied Physics, 2017. 122(8): p. 085101.
51. Asoka‐Kumar, P., et al., Detection of current‐induced vacancies in thin aluminum–
copper lines using positrons. Applied Physics Letters, 1996. 68(3): p. 406-408.
52. Basaran, C. and M. Lin, Damage mechanics of electromigration induced failure.
Mechanics of Materials, 2008. 40(1): p. 66-79.
53. Wu, X., et al., Mater. Des., 2004. 25(2): p. 137.
54. Tan, C.M., et al., Development of Physics-Based Modeling for ULSI Interconnections
Failure Mechanisms: Electromigration and Stress-Induced Voiding, in Applications of
Finite Element Methods for Reliability Studies on ULSI Interconnections, C.M. Tan, et
al., Editors. 2011, Springer London: London. p. 5-38.
55. Zhang, J. and J. Zhang, Thermal-Mechanical Stress Enhanced Atomic Migration at the
Solder/Under Bump Metallization Interface with Current Loading. Advanced Science
Letters, 2012. 5(2): p. 667-671.
56. Kirchheim, R., Stress and electromigration in Al-lines of integrated circuits. Acta
Metallurgica et Materialia, 1992. 40(2): p. 309-323.
57. de Orio, R.L., H. Ceric, and S. Selberherr, Physically based models of electromigration:
From Black's equation to modern TCAD models. Microelectronics Reliability, 2010.
50(6): p. 775-789.
58. Conrad, H., et al., Electroplasticity—the effect of electricity on the mechanical properties
of metals. JOM, 1990. 42(9): p. 28-33.
138
59. Aaboud, M., et al., Search for anomalous electroweak production of WW/WZ in
association with a high-mass dijet system in pp collisions at root S=8 TeV with the
ATLAS detector. Physical Review D, 2017. 95(3).
60. Biehl, V., et al., Evaluation of the haemocompatibility of titanium based biomaterials.
Biomolecular Engineering, 2002. 19(2-6): p. 97-101.
61. Huang, R.Z., et al., Energy and emissions saving potential of additive manufacturing: the
case of lightweight aircraft components. Journal of Cleaner Production, 2016. 135: p.
1559-1570.
62. Novoselov, K.S., et al., Electric field effect in atomically thin carbon films. Science,
2004. 306(5696): p. 666-669.
63. Feng, W., W. Luo, and Y.Y. Feng, Photo-responsive carbon nanomaterials
functionalized by azobenzene moieties: structures, properties and application. Nanoscale,
2012. 4(20): p. 6118-6134.
64. Sun, Y.F., et al., Ultrathin Two-Dimensional Inorganic Materials: New Opportunities for
Solid State Nanochemistry. Accounts of Chemical Research, 2015. 48(1): p. 3-12.
65. Li, S.L., et al., Charge transport and mobility engineering in two-dimensional transition
metal chalcogenide semiconductors. Chemical Society Reviews, 2016. 45(1): p. 118-151.
66. Ferrari, A.C., et al., Science and technology roadmap for graphene, related two-
dimensional crystals, and hybrid systems. Nanoscale, 2015. 7(11): p. 4598-4810.
67. Islam, Z., et al., Quality enhancement of low temperature metal organic chemical vapor
deposited MoS2: an experimental and computational investigation. Nanotechnology,
2019. 30(39): p. 395402.
68. Islam, Z., et al., Enhancement of WSe2 FET Performance Using Low-Temperature
Annealing. Journal of Electronic Materials, 2020. 49(6): p. 3770-3779.
69. Oughaddou, H., et al., Silicene, a promising new 2D material. Progress in Surface
Science, 2015. 90(1): p. 46-83.
70. Watts, M.C., et al., Production of phosphorene nanoribbons. Nature, 2019. 568(7751): p.
216-220.
71. Fiori, G., et al., Electronics based on two-dimensional materials. Nature
Nanotechnology, 2014. 9(10): p. 768-779.
72. Zhu, A.Y. and E. Cubukcu, Graphene nanophotonic sensors. 2d Materials, 2015. 2(3).
73. Deng, D.H., et al., Catalysis with two-dimensional materials and their heterostructures.
Nature Nanotechnology, 2016. 11(3): p. 218-230.
139
74. Bonaccorso, F., et al., Graphene, related two-dimensional crystals, and hybrid systems
for energy conversion and storage. Science, 2015. 347(6217).
75. Plimpton, S., Fast Parallel Algorithms for Short-Range Molecular-Dynamics. Journal of
Computational Physics, 1995. 117(1): p. 1-19.
76. Shingubara, S., I. Utsunomiya, and T. Fujii, Molecular dynamics simulation of void
electromigration under a high-density electric current stress in an aluminum
interconnection. Electronics and Communications in Japan (Part II: Electronics), 1995.
78(12): p. 82-95.
77. Saitoh, K. and H. Kitagawa, Molecular dynamics study of surface effects on atomic
migration near aluminum grain boundary. Computational Materials Science, 1999.
14(1): p. 13-18.
78. Mendelev, M.I. and G.J. Ackland, Development of an interatomic potential for the
simulation of phase transformations in zirconium. Philosophical Magazine Letters, 2007.
87(5): p. 349-359.
79. Zhou, X.W., R.A. Johnson, and H.N.G. Wadley, Misfit-energy-increasing dislocations in
vapor-deposited CoFe/NiFe multilayers. Physical Review B, 2004. 69(14).
80. Liang, T., S.R. Phillpot, and S.B. Sinnott, Parametrization of a reactive many-body
potential for Mo-S systems (vol 79, 245110, 2009). Physical Review B, 2012. 85(19).
81. Mobaraki, A., et al., Validation of inter-atomic potential for WS2 and WSe2 crystals
through assessment of thermal transport properties. Computational Materials Science,
2018. 144: p. 92-98.
82. Ostadhossein, A., et al., ReaxFF Reactive Force-Field Study of Molybdenum Disulfide
(MoS2). The Journal of Physical Chemistry Letters, 2017. 8(3): p. 631-640.
83. Hasanian, M., et al., Hydrogenation and defect formation control the strength and
ductility of MoS2 nanosheets: Reactive molecular dynamics simulation. Extreme
Mechanics Letters, 2018. 22: p. 157-164.
84. Thouless, M.D., Stress evolution during electromigration in a bamboo structure. Scripta
Materialia, 1996. 34(12): p. 1825-1831.
85. Nguyen-Tran, H.-D., et al., A review of electrically-assisted manufacturing. International
Journal of Precision Engineering and Manufacturing-Green Technology, 2015. 2(4): p.
365-376.
86. Zheng, H., et al., Discrete plasticity in sub-10-nm-sized gold crystals. Nature
Communications, 2011. 1: p. 144.
87. Latapie, A. and D. Farkas, Effect of grain size on the elastic properties of nanocrystalline
alpha-iron. Scripta Materialia, 2003. 48(5): p. 611-615.
140
88. Ruestes, C.J., et al., Grain size effects in the deformation of [0001] textured
nanocrystalline Zr. Scripta Materialia, 2014. 71: p. 9-12.
89. Schiotz, J., F.D. Di Tolla, and K.W. Jacobsen, Softening of nanocrystalline metals at very
small grain sizes. Nature, 1998. 391(6667): p. 561-563.
90. Shingubara, S., I. Utsunomiya, and T. Fujii, Molecular dynamics simulation of void
electromigration under a high-density electric current stress in an aluminum
interconnection. Electronics and Communications in Japan Part Ii-Electronics, 1995.
78(12): p. 82-95.
91. Haque, M.A. and M.T.A. Saif, Deformation mechanisms in free-standing nanoscale thin
films: A quantitative in situ transmission electron microscope study. Proceedings of the
National Academy of Sciences of the United States of America, 2004. 101(17): p. 6335-
6340.
92. Islam, Z., B. Wang, and A. Haque, Current density effects on the microstructure of
zirconium thin films. Scripta Materialia, 2018. 144: p. 18-21.
93. Huntington, H.B. and A.R. Grone, Current-induced marker motion in gold wires. Journal
of Physics and Chemistry of Solids, 1961. 20(1): p. 76-87.
94. Sorbello, R.S., A. Lodder, and S.J. Hoving, Finite-Cluster Description of
Electromigration. Physical Review B, 1982. 25(10): p. 6178-6187.
95. Dekker, J.P., A. Lodder, and J. vanEk, Theory for the electromigration wind force in
dilute alloys. Physical Review B, 1997. 56(19): p. 12167-12177.
96. Desai, P.D., H.M. James, and C.Y. Ho, Electrical-Resistivity of Vanadium and
Zirconium. Journal of Physical and Chemical Reference Data, 1984. 13(4): p. 1097-1130.
97. Conrad, H., Electroplasticity in metals and ceramics. Materials Science and Engineering
a-Structural Materials Properties Microstructure and Processing, 2000. 287(2): p. 276-
287.
98. Wesley A. Salandro, J.J.J., Cristina Bunget, Laine Mears, and John T. Roth, The Effect of
Electric Current on Metals. Electrically Assisted Forming, 2014: p. 37-54.
99. Du, S. and Y. Li, Effect of Annealing on Microstructure and Mechanical Properties of
Magnetron Sputtered Cu Thin Films. Advances in Materials Science and Engineering,
2015. 2015: p. 8.
100. Lienig, J. and M. Thiele, Fundamentals of Electromigration, in Fundamentals of
Electromigration-Aware Integrated Circuit Design, J. Lienig and M. Thiele, Editors.
2018, Springer International Publishing: Cham. p. 13-60.
101. Obodo, J., et al., Current-induced changes of migration energy barriers in graphene and
carbon nanotubes. Nanoscale, 2016. 8(19): p. 10310-10315.
141
102. Simoes, S., et al., In situ TEM study of grain growth in nanocrystalline copper thin films.
Nanotechnology, 2010. 21(14).
103. Kumar, S., T. Alam, and A. Haque, Quantitative in-situ TEM study of stress-assisted
grain growth. MRS Communications, 2013. 3(02): p. 101-105.
104. Seita, M., et al., Direct evidence for stress-induced texture evolution and grain growth of
silver thin films upon thermal treatment and self-ion bombardment. Acta Materialia,
2010. 58(19): p. 6513-6525.
105. Gianola, D.S., et al., Stress-assisted discontinuous grain growth and its effect on the
deformation behavior of nanocrystalline aluminum thin films. Acta Materialia, 2006.
54(8): p. 2253-2263.
106. Shu, J., et al., Catalytic Palladium-Based Membrane Reactors - a Review. Canadian
Journal of Chemical Engineering, 1991. 69(5): p. 1036-1060.
107. Mroczkowski, R.S., Electronic connector handbook : theory and applications. Electronic
packaging and interconnection series. 1998: New York : McGraw-Hill, ©1998.
108. Islam, Z., H. Gao, and A. Haque, Synergy of elastic strain energy and electron wind force
on thin film grain growth at room temperature. Materials Characterization, 2019. 152: p.
85-93.
109. Stukowski, A., Visualization and analysis of atomistic simulation data with OVITO-the
Open Visualization Tool. Modelling and Simulation in Materials Science and
Engineering, 2010. 18(1).
110. Islam, Z., B. Wang, and A. Haque, Current density effects on the microstructure of
zirconium thin films. Scripta Materialia, 2018. 144(Supplement C): p. 18-21.
111. Hariharan, K., et al., Decoupling thermal and electrical effect in an electrically assisted
uniaxial tensile test using finite element analysis. Metallurgical and Materials
Transactions A, 2015. 46(7): p. 3043-3051.
112. Zhang, D., et al., Dynamic Electropulsing Induced Phase Transformations and Their
Effects on Single Point Diamond Turning of AZ91 Alloy. Journal of Surface Engineered
Materials and Advanced Technology, 2012. Vol.02No.01: p. 6.
113. Fitzpatrick, M.E., M.T. Hutchings, and P.J. Withers, Separation of macroscopic, elastic
mismatch and thermal expansion misfit stresses in metal matrix composite quenched
plates from neutron diffraction measurements. Acta Materialia, 1997. 45(12): p. 4867-
4876.
114. Ma, E., Watching the nanograins roll. Science, 2004. 305(5684): p. 623-624.
115. Monk, J. and D. Farkas, Strain-induced grain growth and rotation in nickel nanowires.
Physical Review B, 2007. 75(4).
142
116. Islam, Z., B.M. Wang, and A. Haque, Current density effects on the microstructure of
zirconium thin films. Scripta Materialia, 2018. 144: p. 18-21.
117. Sarychev, M.E., et al., General model for mechanical stress evolution during
electromigration. Journal of Applied Physics, 1999. 86(6): p. 3068-3075.
118. Korhonen, M.A., P. Børgesen, and C.-Y. Li, Mechanisms of Stress-Induced and
Electromigration-Induced Damage in Passivated Narrow Metallizations on Rigid
Substrates. MRS Bulletin, 2013. 17(7): p. 61-69.
119. Conrad, H., Thermally activated plastic flow of metals and ceramics with an electric field
or current. Materials Science and Engineering: A, 2002. 322(1): p. 100-107.
120. Bolzoni, L., E.M. Ruiz-Navas, and E. Gordo, Flexural properties, thermal conductivity
and electrical resistivity of prealloyed and master alloy addition powder metallurgy Ti–
6Al–4V. Materials & Design (1980-2015), 2013. 52: p. 888-895.
121. Gorsse, S., et al., Additive manufacturing of metals: a brief review of the characteristic
microstructures and properties of steels, Ti-6Al-4V and high-entropy alloys. Science and
Technology of Advanced Materials, 2017. 18(1): p. 584-610.
122. Li, H., et al., Methodology for estimating the critical resolved shear stress ratios of α-
phase Ti using EBSD-based trace analysis. Acta Materialia, 2013. 61(20): p. 7555-7567.
123. Mak, K.F., et al., Atomically Thin MoS2: A New Direct-Gap Semiconductor. Physical
Review Letters, 2010. 105(13).
124. Wang, Q.H., et al., Electronics and optoelectronics of two-dimensional transition metal
dichalcogenides. Nature Nanotechnology, 2012. 7(11): p. 699-712.
125. Xu, M.S., et al., Graphene-Like Two-Dimensional Materials. Chemical Reviews, 2013.
113(5): p. 3766-3798.
126. Novoselov, K.S., et al., 2D materials and van der Waals heterostructures. Science, 2016.
353(6298).
127. Radisavljevic, B., et al., Single-layer MoS2 transistors. Nature Nanotechnology, 2011.
6(3): p. 147-150.
128. Tong, X., et al., Advances in MoS2-Based Field Effect Transistors (FETs). Nano-Micro
Letters, 2015. 7(3): p. 203-218.
129. Nam, H., et al., Multiple MoS2 Transistors for Sensing Molecule Interaction Kinetics.
Scientific Reports, 2015. 5.
130. Cheng, R., et al., Few-layer molybdenum disulfide transistors and circuits for high-speed
flexible electronics. Nature Communications, 2014. 5.
143
131. Pu, J., et al., Highly Flexible MoS2 Thin-Film Transistors with Ion Gel Dielectrics. Nano
Letters, 2012. 12(8): p. 4013-4017.
132. Chang, H.Y., et al., High-Performance, Highly Bendable MoS2 Transistors with High-K
Dielectrics for Flexible Low-Power Systems. Acs Nano, 2013. 7(6): p. 5446-5452.
133. Li, N., et al., Ultra-High Sensitivity Humidity Sensor Based on MoS2/Ag Composite
Films. Ieee Electron Device Letters, 2017. 38(6): p. 806-809.
134. Yan, L.R., et al., MoS2-DNA and MoS2 based sensors. Rsc Advances, 2017. 7(38): p.
23573-23582.
135. Baugher, B.W.H., et al., Intrinsic Electronic Transport Properties of High-Quality
Monolayer and Bilayer MoS2. Nano Letters, 2013. 13(9): p. 4212-4216.
136. Choi, W., et al., Recent development of two-dimensional transition metal dichalcogenides
and their applications. Materials Today, 2017. 20(3): p. 116-130.
137. Hong, J.H., et al., Exploring atomic defects in molybdenum disulphide monolayers.
Nature Communications, 2015. 6.
138. Lin, Y.C., et al., Realizing Large-Scale, Electronic-Grade Two-Dimensional
Semiconductors. Acs Nano, 2018. 12(2): p. 965-975.
139. Nie, C.B., et al., Ultrafast growth of large-area monolayer MoS2. film via gold foil
assistant CVD for a. highly. sensitive photodetector. Nanotechnology, 2017. 28(27).
140. Dumitru Dumcenco, D.O., Kolyo Marinov, Predrag Lazić, Marco Gibertini, Nicola
Marzari, Oriol Lopez Sanchez, Yen-Cheng Kung, Daria Krasnozhon, Ming-Wei Chen,
Simone Bertolazzi, Philippe Gillet, Anna Fontcuberta i Morral, Aleksandra Radenovic,
and Andras Kis, Large-Area Epitaxial Monolayer MoS2. ACS Nano, 2015. 9(4): p. 4611-
4620.
141. Liu, X.L., et al., Point Defects and Grain Boundaries in Rotationally Commensurate
MoS2 on Epitaxial Graphene. Journal of Physical Chemistry C, 2016. 120(37): p. 20798-
20805.
142. Zhou, W., et al., Intrinsic Structural Defects in Monolayer Molybdenum Disulfide. Nano
Letters, 2013. 13(6): p. 2615-2622.
143. Huang, P.Y., et al., Grains and grain boundaries in single-layer graphene atomic
patchwork quilts. Nature, 2011. 469(7330): p. 389-+.
144. Yu, Q.K., et al., Control and characterization of individual grains and grain boundaries
in graphene grown by chemical vapour deposition. Nature Materials, 2011. 10(6): p. 443-
449.
144
145. Najmaei, S., et al., Electrical Transport Properties of Polycrystalline Monolayer
Molybdenum Disulfide. Acs Nano, 2014. 8(8): p. 7930-7937.
146. Cao, A.J. and J.M. Qu, Atomistic simulation study of brittle failure in nanocrystalline
graphene under uniaxial tension. Applied Physics Letters, 2013. 102(7).
147. Yasaei, P., et al., Bimodal Phonon Scattering in Graphene Grain Boundaries. Nano
Letters, 2015. 15(7): p. 4532-4540.
148. Jurca, T., et al., Low-Temperature Atomic Layer Deposition of MoS2 Films. Angewandte
Chemie, 2017. 129(18): p. 5073-5077.
149. Fei, L., et al., Direct TEM observations of growth mechanisms of two-dimensional MoS2
flakes. Nature Communications, 2016. 7: p. 12206.
150. Kong, D.S., et al., Synthesis of MoS2 and MoSe2 Films with Vertically Aligned Layers.
Nano Letters, 2013. 13(3): p. 1341-1347.
151. Zink, N., et al., In situ heating TEM study of onion-like WS2 and MoS2 nanostructures
obtained via MOCVD. Chemistry of Materials, 2008. 20(1): p. 65-71.
152. Humphreys F.J., H.M., Recrystallization and Related Annealing Phenomena. second
edition ed. 2012: Elsevier Science. 658.
153. Liang, C.L. and K.L. Lin, The microstructure and property variations of metals induced
by electric current treatment: A review. Materials Characterization, 2018. 145: p. 545-
555.
154. Waryoba, D., et al., Low temperature annealing of metals with electrical wind force
effects. Journal of Materials Science & Technology, 2019. 35(4): p. 465-472.
155. Lo, C.L., et al., Large-Area, Single-Layer Molybdenum Disulfide Synthesized at BEOL
Compatible Temperature as Cu Diffusion Barrier. Ieee Electron Device Letters, 2018.
39(6): p. 873-876.
156. Zhang, K., et al., Considerations for Utilizing Sodium Chloride in Epitaxial Molybdenum
Disulfide. ACS Applied Materials & Interfaces, 2018. 10(47): p. 40831-40837.
157. Wang, B.M., et al., Role of sulphur atoms on stress relaxation and crack propagation in
monolayer MoS2. Nanotechnology, 2017. 28(36).
158. Sledzinska, M., et al., Thermal conductivity of MoS2 polycrystalline nanomembranes. 2d
Materials, 2016. 3(3).
159. Yalon, E., et al., Energy Dissipation in Monolayer MoS2 Electronics. Nano Letters, 2017.
17(6): p. 3429-3433.
145
160. Dutta, S., et al., Finite size effects in highly scaled ruthenium interconnects. IEEE
Electron Device Letters, 2018. 39(2): p. 268-271.
161. Thompson, C. and H.I. Smith, Surface‐energy‐driven secondary grain growth in ultrathin
(< 100 nm) films of silicon. Applied Physics Letters, 1984. 44(6): p. 603-605.
162. Chhowalla, M., et al., The chemistry of two-dimensional layered transition metal
dichalcogenide nanosheets. Nature Chemistry, 2013. 5(4): p. 263-275.
163. Huang, X., Z.Y. Zeng, and H. Zhang, Metal dichalcogenide nanosheets: preparation,
properties and applications. Chemical Society Reviews, 2013. 42(5): p. 1934-1946.
164. Eichfeld, S.M., et al., Highly Scalable, Atomically Thin WSe2 Grown via Metal-Organic
Chemical Vapor Deposition. Acs Nano, 2015. 9(2): p. 2080-2087.
165. Luo, K., et al., Ab-Initio Simulations of Monolayer InSe and MoS2 Strain Effect: From
Electron Mobility to Photoelectric Effect. Journal of Electronic Materials, 2019.
166. Gong, Y.Y., et al., Thin Film Transistors Using Wafer-Scale Low-Temperature MOCVD
WSe2. Journal of Electronic Materials, 2016. 45(12): p. 6280-6284.
167. Hsu, A., et al., Large-Area 2-D Electronics: Materials, Technology, and Devices.
Proceedings of the Ieee, 2013. 101(7): p. 1638-1652.
168. Das, S. and J. Appenzeller, Where Does the Current Flow in Two-Dimensional Layered
Systems? Nano Letters, 2013. 13(7): p. 3396-3402.
169. Zhao, W.J., et al., Origin of Indirect Optical Transitions in Few-Layer MoS2, WS2, and
WSe2. Nano Letters, 2013. 13(11): p. 5627-5634.
170. Zhang, Y., et al., Direct observation of the transition from indirect to direct bandgap in
atomically thin epitaxial MoSe2. Nature Nanotechnology, 2014. 9(2): p. 111-115.
171. Wang, H., et al., Integrated Circuits Based on Bilayer MoS2 Transistors. Nano Letters,
2012. 12(9): p. 4674-4680.
172. Lopez-Sanchez, O., et al., Ultrasensitive photodetectors based on monolayer MoS2.
Nature Nanotechnology, 2013. 8(7): p. 497-501.
173. Hao, K., et al., Direct measurement of exciton valley coherence in monolayer WSe2.
Nature Physics, 2016. 12(7): p. 677-+.
174. Jones, A.M., et al., Spin-layer locking effects in optical orientation of exciton spin in
bilayer WSe2. Nature Physics, 2014. 10(2): p. 130-134.
175. Eftekhari, A., Tungsten dichalcogenides (WS2, WSe2, and WTe2): materials chemistry
and applications. Journal of Materials Chemistry A, 2017. 5(35): p. 18299-18325.
146
176. Chuang, H.J., et al., High Mobility WSe2 p- and n-Type Field-Effect Transistors
Contacted by Highly Doped Graphene for Low-Resistance Contacts. Nano Letters, 2014.
14(6): p. 3594-3601.
177. Liu, W., et al., Role of Metal Contacts in Designing High-Performance Monolayer n-
Type WSe2 Field Effect Transistors. Nano Letters, 2013. 13(5): p. 1983-1990.
178. Zeng, C., et al., Investigation of abrupt degradation of drain current caused by under-
gate crack in AlGaN/GaN high electron mobility transistors during high temperature
operation stress. Journal of Applied Physics, 2015. 118(12).
179. Ahn, G.H., et al., Strain-engineered growth of two-dimensional materials. Nature
Communications, 2017. 8.
180. Zhou, H.L., et al., Large Area Growth and Electrical Properties of p-Type WSe2 Atomic
Layers. Nano Letters, 2015. 15(1): p. 709-713.
181. Ye, H., et al., Toward a Mechanistic Understanding of Vertical Growth of van der Waals
Stacked 2D Materials: A Multiscale Model and Experiments. Acs Nano, 2017. 11(12): p.
12780-12788.
182. Walsh, L.A. and C.L. Hinkle, van der Waals epitaxy: 2D materials and topological
insulators. Applied Materials Today, 2017. 9: p. 504-515.
183. van der Zande, A.M., et al., Grains and grain boundaries in highly crystalline monolayer
molybdenum disulphide. Nature Materials, 2013. 12(6): p. 554-561.
184. Lei, B., et al., Direct Observation of Semiconductor-Metal Phase Transition in Bilayer
Tungsten Diselenide Induced by Potassium Surface Functionalization. Acs Nano, 2018.
12(2): p. 2070-+.
185. Chen, Y., et al., Tuning Electronic Structure of Single Layer MoS2 through Defect and
Interface Engineering. Acs Nano, 2018. 12(3): p. 2569-2579.
186. Namgung, S.D., et al., Influence of post-annealing on the off current of MoS2 field-effect
transistors. Nanoscale Research Letters, 2015. 10: p. 1-6.
187. Park, J.Y., et al., Self-Curable Gate-All-Around MOSFETs Using Electrical Annealing to
Repair Degradation Induced From Hot-Carrier Injection. Ieee Transactions on Electron
Devices, 2016. 63(3): p. 910-915.
188. Kim, C.K., et al., Electrothermal Annealing (ETA) Method to Enhance the Electrical
Performance of Amorphous-Oxide-Semiconductor (AOS) Thin-Film Transistors (TFTs).
Acs Applied Materials & Interfaces, 2016. 8(36): p. 23820-23826.
189. Islam, Z., H.J. Gao, and A. Haque, Synergy of elastic strain energy and electron wind
force on thin film grain growth at room temperature. Materials Characterization, 2019.
152: p. 85-93.
147
190. Islam, Z., et al., Quality enhancement of low temperature metal organic chemical vapor
deposited MoS2: an experimental and computational investigation. Nanotechnology,
2019. 30(39).
191. Prasad, N., et al., Current induced annealing and electrical characterization of single
layer graphene grown by chemical vapor deposition for future interconnects in VLSI
circuits. Applied Physics Letters, 2014. 105(11).
192. Zhang, X., et al., Diffusion-Controlled Epitaxy of Large Area Coalesced WSe2
Monolayers on Sapphire. Nano Letters, 2018. 18(2): p. 1049-1056.
193. Kozhakhmetov, A., et al., Scalable BEOL compatible 2D tungsten diselenide. 2d
Materials, 2020. 7(1).
194. Behranginia, A., et al., Power Dissipation of WSe2 Field-Effect Transistors Probed by
Low-Frequency Raman Thermometry. Acs Applied Materials & Interfaces, 2018. 10(29):
p. 24892-24898.
195. Kim, Y.J., et al., Contact Resistance Reduction of WS2 FETs Using High-Pressure
Hydrogen Annealing. Ieee Journal of the Electron Devices Society, 2018. 6(1): p. 164-
168.
196. Komsa, H.P., et al., Two-Dimensional Transition Metal Dichalcogenides under Electron
Irradiation: Defect Production and Doping. Physical Review Letters, 2012. 109(3).
197. Yang, D.X., et al., Electronic and Magnetic Properties of Defected Monolayer WSe2 with
Vacancies. Nanoscale Research Letters, 2019. 14.
198. Cao, D., et al., First-principles study of the origin of magnetism induced by intrinsic
defects in monolayer MoS2. Applied Surface Science, 2016. 361: p. 199-205.
199. Tsuchiya, T., Trapped-Electron and Generated Interface-Trap Effects in Hot-Electron-
Induced Mosfet Degradation. Ieee Transactions on Electron Devices, 1987. 34(11): p.
2291-2296.
200. Rasool, H.I., C. Ophus, and A. Zettl, Atomic Defects in Two Dimensional Materials.
Advanced Materials, 2015. 27(38): p. 5771-5777.
201. Hong, J.H., et al., Atomic Defects in Two-Dimensional Materials: From Single-Atom
Spectroscopy to Functionalities in Opto-/Electronics, Nanomagnetism, and Catalysis.
Advanced Materials, 2017. 29(14).
202. Jain, S.K., V. Juricic, and G.T. Barkema, Boundaries determine the formation energies of
lattice defects in two-dimensional buckled materials. Physical Review B, 2016. 94(2).
203. Lin, Y.C., et al., Atomic mechanism of the semiconducting-to-metallic phase transition in
single-layered MoS2. Nature Nanotechnology, 2014. 9(5): p. 391-396.
148
204. Hummel, R.E., Electromigration and Related Failure Mechanisms in Integrated-Circuit
Interconnects. International Materials Reviews, 1994. 39(3): p. 97-111.
205. Arzt, E., et al., Electromigration Failure by Shape Change of Voids in Bamboo Lines.
Journal of Applied Physics, 1994. 76(3): p. 1563-1571.
206. Mishra, U.K., P. Parikh, and Y.F. Wu, AlGaN/GaN HEMTs - An overview of device
operation and applications. Proceedings of the Ieee, 2002. 90(6): p. 1022-1031.
207. Chow, T.P. and R. Tyagi, Wide Bandgap Compound Semiconductors for Superior High-
Voltage Unipolar Power Devices. Ieee Transactions on Electron Devices, 1994. 41(8): p.
1481-1483.
208. Wu, Y., et al., 30-W/mm GaN HEMTs by field plate optimization. IEEE Electron Device
Letters, 2004. 25(3): p. 117-119.
209. Kanamura, M., T. Kikkawa, and K. Joshin. A 100-W high-gain AlGaN/GaN HEMT
power amplifier on a conductive n-SiC substrate for wireless base station applications. in
IEDM Technical Digest. IEEE International Electron Devices Meeting, 2004. 2004.
210. Wu, Y., et al. High-gain microwave GaN HEMTs with source-terminated field-plates. in
IEDM Technical Digest. IEEE International Electron Devices Meeting, 2004. 2004.
211. Joshin, K., et al. A 174 W high-efficiency GaN HEMT power amplifier for W-CDMA base
station applications. in IEEE International Electron Devices Meeting 2003. 2003.
212. Islam, Z., A. Haque, and N. Glavin, Real-time visualization of GaN/AlGaN high electron
mobility transistor failure at off-state. Applied Physics Letters, 2018. 113(18).
213. Wang, B.M., et al., In situ transmission electron microscopy of transistor operation and
failure. Nanotechnology, 2018. 29(31).
214. Lee, H.P., et al., Investigation of AlGaN/GaN high electron mobility transistor structures
on 200-mm silicon (111) substrates employing different buffer layer configurations.
Scientific Reports, 2016. 6: p. 37588.
215. Syaranamual, G.J., et al., Role of two-dimensional electron gas (2DEG) in AlGaN/GaN
high electron mobility transistor (HEMT) ON-state degradation. Microelectronics
Reliability, 2016. 64: p. 589-593.
216. Makaram, P., et al., Evolution of structural defects associated with electrical degradation
in AlGaN/GaN high electron mobility transistors. Applied Physics Letters, 2010. 96(23).
217. Cheney, D.J., et al., Reliability studies of AlGaN/GaN high electron mobility transistors.
Semiconductor Science and Technology, 2013. 28(7): p. 074019.
218. Marcon, D., et al., Reliability Analysis of Permanent Degradations on AlGaN/GaN
HEMTs. IEEE Transactions on Electron Devices, 2013. 60(10): p. 3132-3141.
149
219. del Alamo, J.A. and J. Joh, GaN HEMT reliability. Microelectronics Reliability, 2009.
49(9): p. 1200-1206.
220. Cheney, D.J., et al., Degradation Mechanisms for GaN and GaAs High Speed
Transistors. Materials, 2012. 5(12): p. 2498-2520.
221. Piazza, M., et al., Degradation of TiAlNiAu as ohmic contact metal for GaN HEMTs.
Microelectronics Reliability, 2009. 49(9-11): p. 1222-1225.
222. Vassiliou, B. and F.G. Wilde, A Hexagonal Form of Silicon Nitride. Nature, 1957.
179(4556): p. 435-436.
223. Ganchenkova, M.G. and R.M. Nieminen, Nitrogen Vacancies as Major Point Defects in
Gallium Nitride. Physical Review Letters, 2006. 96(19): p. 196402.
224. Usman, M., et al., Electrical and structural characterization of ion implanted GaN.
Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with
Materials and Atoms, 2009. 267(8): p. 1561-1563.
225. Amano, H., et al., The 2018 GaN power electronics roadmap. Journal of Physics D:
Applied Physics, 2018. 51(16): p. 163001.
226. Bagnall, K.R., et al., Simultaneous measurement of temperature, stress, and electric field
in GaN HEMTs with micro-Raman spectroscopy. Review of Scientific Instruments, 2017.
88(11).
227. Wu, Y., C.Y. Chen, and J.A. del Alamo, Electrical and structural degradation of GaN
high electron mobility transistors under high-power and high-temperature Direct
Current stress. Journal of Applied Physics, 2015. 117(2): p. 025707.
228. Meneghini, M., et al., OFF-State Degradation of AlGaN/GaN Power HEMTs:
Experimental Demonstration of Time-Dependent Drain-Source Breakdown. Ieee
Transactions on Electron Devices, 2014. 61(6): p. 1987-1992.
229. Meneghesso, G., et al., Degradation of AlGaN/GaN HEMT devices: Role of reverse-bias
and hot electron stress. Microelectronic Engineering, 2013. 109: p. 257-261.
230. Cheney, D.J., et al., Reliability studies of AlGaN/GaN high electron mobility transistors.
Semiconductor Science and Technology, 2013. 28(7): p. 074019.
231. Choi, S., et al., The impact of mechanical stress on the degradation of AlGaN/GaN high
electron mobility transistors. Journal of Applied Physics, 2013. 114(16): p. 164501.
232. Kuball, M., et al., AlGaN/GaN HEMT device reliability and degradation evolution:
Importance of diffusion processes. Microelectronics Reliability, 2011. 51(2): p. 195-200.
150
233. Chang, C.Y., et al., Electric-Field-Driven Degradation in off-State Step-Stressed
AlGaN/GaN High-Electron Mobility Transistors. IEEE Transactions on Device and
Materials Reliability, 2011. 11(1): p. 187-193.
234. Douglas, E.A., et al., AlGaN/GaN High Electron Mobility Transistor degradation under
on- and off-state stress. Microelectronics Reliability, 2011. 51(2): p. 207-211.
235. Chowdhury, U., et al., TEM Observation of Crack- and Pit-Shaped Defects in
Electrically Degraded GaN HEMTs. IEEE Electron Device Letters, 2008. 29(10): p.
1098-1100.
236. Cullen, D.A., et al., Electroluminescence and Transmission Electron Microscopy
Characterization of Reverse-Biased AlGaN/GaN Devices. IEEE Transactions on Device
and Materials Reliability, 2013. 13(1): p. 126-135.
237. Xu, W., et al., Off-state breakdown and leakage current transport analysis of
AlGaN/GaN high electron mobility transistors. Microelectronics Reliability, 2014.
54(11): p. 2406-2409.
238. Marcon, D., et al., Reliability analysis of permanent degradations on AlGaN/GaN
HEMTs. IEEE Transactions on Electron Devices, 2013. 60(10): p. 3132-3141.
239. Smith, D.J., et al., Applications of TEM imaging, analysis and electron holography to III-
nitride HEMT devices. Microelectronics Reliability, 2010. 50(9): p. 1514-1519.
240. Baoming, W., et al., In-situ Transmission Electron Microscopy of Transistor Operation
and Failure. Nanotechnology, 2018.
241. Islam, Z., A. Haque, and N. Glavin, Real-time visualization of GaN/AlGaN high electron
mobility transistor failure at off-state. Applied Physics Letters, 2018. 113(18): p. 183102.
242. Maojun Wang and Kevin J. Chen, Off-State Breakdown Characterization in AlGaN/GaN
HEMT Using Drain Injection Technique. IEEE TRANSACTIONS ON ELECTRON
DEVICES, 2010. 57(7): p. 1492 - 1496.
243. Wu, Y.F., et al., Very high breakdown voltage and large transconductance realized on
GaN heterojunction field effect transistors. Applied Physics Letters, 1996. 69(10): p.
1438-1440.
244. Khan, M.A., et al., GaN based heterostructure for high power devices. Solid-State
Electronics, 1997. 41(10): p. 1555-1559.
245. Moens, P., et al., Impact of buffer leakage on intrinsic reliability of 650V AlGaN/GaN
HEMTs. 2015 Ieee International Electron Devices Meeting (Iedm), 2015.
246. Maier, D., et al., Testing the Temperature Limits of GaN-Based HEMT Devices. Ieee
Transactions on Device and Materials Reliability, 2010. 10(4): p. 427-436.
151
247. Khanna, S.M., et al., 2 MeV proton radiation damage studies of gallium nitride films
through low temperature photoluminescence spectroscopy measurements. Ieee
Transactions on Nuclear Science, 2000. 47(6): p. 2322-2328.
248. Gaudreau, F., et al., Transport properties of proton-irradiated gallium nitride-based two-
dimensional electron-gas system. Ieee Transactions on Nuclear Science, 2002. 49(6): p.
2702-2707.
249. Lei, Z.F., et al., Degradation mechanisms of AlGaN/GaN HEMTs under 800MeV Bi ions
irradiation. Microelectronics Reliability, 2018. 80: p. 312-316.
250. Galloway, K.F. and R.D. Schrimpf, MOS device degradation due to total dose ionizing
radiation in the natural space environment: A review. Microelectronics Journal, 1990.
21(2): p. 67-81.
251. Arakawa, K., et al., Observation of the one-dimensional diffusion of nanometer-sized
dislocation loops. Science, 2007. 318(5852): p. 956-959.
252. Dai, C., et al., Atomistic simulations of Ni segregation to irradiation induced dislocation
loops in Zr-Ni alloys. Acta Materialia, 2017. 140: p. 56-66.
253. Pearton, S., et al., Ionizing radiation damage effects on GaN devices. ECS Journal of
solid state science and technology, 2016. 5(2): p. Q35-Q60.
254. Sonia, G., et al., 2 MeV ion irradiation effects on AlGaN/GaN HFET devices. Solid-State
Electronics, 2008. 52(7): p. 1011-1017.
255. Petrosky, J.C., et al., Trap Assisted Tunneling Induced Currents in Neutron Irradiated
AlGaN/GaN HFETs. Ieee Transactions on Nuclear Science, 2009. 56(5): p. 2905-2909.
256. Xinwen, H., et al., The energy dependence of proton-induced degradation in AlGaN/GaN
high electron mobility transistors. IEEE Transactions on Nuclear Science, 2004. 51(2): p.
293-297.
257. Karmarkar, A.P., et al., Proton irradiation effects on GaN-based high electron-mobility
transistors with Si-doped AlxGa1-xN and thick GaN cap layers. Ieee Transactions on
Nuclear Science, 2004. 51(6): p. 3801-3806.
258. White, B.D., et al., Electrical, spectral, and chemical properties of 1.8 MeV proton
irradiated AlGaN/GaN HEMT structures as a function of proton fluence. Ieee
Transactions on Nuclear Science, 2003. 50(6): p. 1934-1941.
259. Cai, S.J., et al., Annealing behavior of a proton irradiated AlxGa1-xN/GaN high electron
mobility transistor grown by MBE. Ieee Transactions on Electron Devices, 2000. 47(2):
p. 304-307.
260. Paine, B.M., et al., Lifetesting GaN HEMTs with multiple degradation mechanisms. IEEE
Transactions on Device and Materials Reliability, 2015. 15(4): p. 486-494.
152
261. Ziegler, J.F., M.D. Ziegler, and J.P. Biersack, SRIM – The stopping and range of ions in
matter (2010). Nuclear Instruments and Methods in Physics Research Section B: Beam
Interactions with Materials and Atoms, 2010. 268(11–12): p. 1818-1823.
262. Hattar, K., D.C. Bufford, and D.L. Buller, Concurrent in situ ion irradiation transmission
electron microscope. Nuclear Instruments and Methods in Physics Research Section B:
Beam Interactions with Materials and Atoms, 2014. 338: p. 56-65.
263. Hytch, M.J., E. Snoeck, and R. Kilaas, Quantitative measurement of displacement and
strain fields from HREM micrographs. Ultramicroscopy, 1998. 74(3): p. 131-146.
264. Zhang, A.P., et al., Microwave power SiC MESFETs and GaN HEMTs. Ieee Lester
Eastman Conference on High Performance Devices, Proceedings, 2002: p. 181-185.
265. Whiting, P.G., et al., Nanocrack formation in AlGaN/GaN high electron mobility
transistors utilizing Ti/Al/Ni/Au ohmic contacts. Microelectronics Reliability, 2017. 70: p.
41-48.
266. Ganchenkova, M.G. and R.M. Nieminen, Nitrogen vacancies as major point defects in
gallium nitride. Physical Review Letters, 2006. 96(19).
267. Usman, M., et al., Electrical and structural characterization of ion implanted GaN.
Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with
Materials and Atoms, 2009. 267(8-9): p. 1561-1563.
268. Usman, M., A. Hallen, and A. Nazir, Ion implantation induced nitrogen defects in GaN.
Journal of Physics D-Applied Physics, 2015. 48(45).
269. Zhang, X.H., et al., Radiation damage in nanostructured materials. Progress in Materials
Science, 2018. 96: p. 217-321.
270. Robach, J.S., et al., Dynamic observations and atomistic simulations of dislocation-defect
interactions in rapidly quenched copper and gold. Acta Materialia, 2006. 54(6): p. 1679-
1690.
271. Barr, C.M., et al., Examining the influence of grain size on radiation tolerance in the
nanocrystalline regime. Applied Physics Letters, 2018. 112(18).
272. Wirth, B.D., et al., Mechanical property degradation in irradiated materials: A
multiscale modeling approach. Nuclear Instruments & Methods in Physics Research
Section B-Beam Interactions with Materials and Atoms, 2001. 180: p. 23-31.
273. Khan, M.A., et al., Carbon ion beam irradiation-induced effects on structural,
morphological and optical properties of boron nitride nanowires (vol 189, pg 28, 2017).
Materials Letters, 2017. 195: p. 246-246.
274. Chen, L., et al., Irradiation effect on mechanical properties of tungsten from molecular
dynamic simulation. Materials Letters, 2019. 241: p. 27-30.
153
275. Wei, B.X., et al., Microstructure evolution of nonstoichiometric ZrC0.6 with ordered
carbon vacancies under ion irradiation. Materials Letters, 2018. 228: p. 254-257.
276. Ferroni, F., et al., High temperature annealing of ion irradiated tungsten. Acta
Materialia, 2015. 90: p. 380-393.
277. Zibrov, M., et al., High temperature recovery of radiation defects in tungsten and its
effect on deuterium retention. Nuclear Materials and Energy, 2020: p. 100747.
278. Huntington, H.B., Effect of Driving Forces on Atom Motion. Thin Solid Films, 1975.
25(2): p. 265-280.
279. Conrad, H., et al., Electroplasticity—the effect of electricity on the mechanical properties
of metals. Jom, 1990. 42(9): p. 28-33.
280. Sheng, Y., et al., Application of high-density electropulsing to improve the performance
of metallic materials: Mechanisms, microstructure and properties. Materials, 2018.
11(2): p. 185.
281. Liang, C.-L. and K.-L. Lin, The microstructure and property variations of metals induced
by electric current treatment: A review. Materials Characterization, 2018. 145: p. 545-
555.
282. Was, G.S., Fundamentals of Radiation Materials Science. 2017: Springer-Verlag New
York.
283. El-Atwani, O., et al., Grain size threshold for enhanced irradiation resistance in
nanocrystalline and ultrafine tungsten. Materials Research Letters, 2017. 5(5): p. 343-
349.
284. Cai, W., et al., Dislocation core effects on mobility. Dislocations in solids, 2004. 12: p. 1-
80.
285. Singh, B.N., et al., Review: Evolution of stacking fault tetrahedra and its role in defect
accumulation under cascade damage conditions. Journal of Nuclear Materials, 2004.
328(2): p. 77-87.
Md Zahabul Islam
Phone: (814) 321 3140 E-Mail: zahabul1@gmail.com, mqi5074@psu.edu
Education
• Ph.D.: Department of Mechanical Engineering, The Pennsylvania State University, University Park,
PA.(USA), August 2020
• M. S.: Department of Mechanical Engineering, Bangladesh University of Engineering and
Technology (BUET), Dhaka-1000 (Bangladesh), July 2013
• B. S.: Department of Mechanical Engineering, Bangladesh University of Engineering and
Technology (BUET), Dhaka-1000 (Bangladesh), February 2011
Selected Publications (Asterisk denotes undergraduate mentee)
Book Chapter 1. Potential and Challenges of in-situ Microscopy on Electronic Devices and Materials, Zahabul Islam,
Aman Haque, Wide Bandgap Semiconductor-Based Electronics, IOPscience (2020)
Journal Publications
1. Enhancement of WSe2 FET device performance using low temperature annealing, Zahabul Islam, Azimkhan Kozhakhmetov, Joshua Robinson, Aman Haque, Journal of Electronic Materials, volume 49, pages 3770–3779 (2020)
2. β-Ga2O3 Schottky diode failure under forward biasing condition: In-situ TEM study, Zahabul Islam, Minghan Xian, Aman Haque, Fan Ren, Marko Tadjer, Nicholas Glavin, S.J. Pearton, IEEE
Transactions on Electron Devices, doi: 10.1109/TED.2020.3000441 (2020)
3. Heavy ion irradiation effects on GaN/AlGaN high electron mobility transistor failure, Zahabul Islam,
Angela L. Paoletta⁕, Anthony M. Monterrosa, Jennifer D. Schuler, Timothy J. Rupert, Khalid Hattar,
Nicholas Glavin, Aman Haque, Microelectron. Reliab.,Vol. 102, pp. 113493 (2019)
4. Quality enhancement of nanocrystalline MoS2 via electrical annealing: an experimental and
computational investigation, Zahabul Islam, Kehao Zhang, Joshua Robinson, Aman Haque,
Nanotechnology, Vol. 30 (39) pp. 395402 (2019)
5. Synergy of elastic strain energy and electron wind force on thin film grain growth at room
temperature, Zahabul Islam, Huajian Gao, Aman Haque, Mater. Charact., vol. 152, pp. 85-93 (2019)
6. Real-time visualization of GaN/AlGaN high electron mobility transistor failure at off-state, Zahabul
Islam, Aman Haque, Nicholas Glavin, Appl. Phys. Lett., Vol. 113, pp.183102 (2018)
7. Departing from the mutual exclusiveness of strength and ductility in nanocrystalline metals with
vacancy induced plasticity, Zahabul Islam, Baoming Wang, Khalid Hattar, Huajian Gao, Aman
Haque, Scripta Materialia, Vol. 157, pp. 39-43 (2018)
8. Current density effects on the microstructure of zirconium thin films, Zahabul Islam, Baoming Wang,
Aman Haque, Scripta Materialia, vol. 144, pp.18-21 (2018) Awards/Honor
• Kulakowski Travel Award (Dept. of Mech. And Nuc. Engg.) for Attending MRS Fall Meeting, 2017
• First Annual Student Paper Competition Finalist (Applied Mechanics Division) IMECE-ASME, 2019
Google website: https://sites.google.com/site/mzahabulislam/
top related