University of Southern California Dissertations and Theses

Expanding sputtering synthesis domains for the discovery of novel film microstructures and morphologies

(USC Thesis Other)

Expanding sputtering synthesis domains for the discovery of novel film microstructures and morphologies

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Expanding Sputtering Synthesis Domains for the Discovery
of Novel Film Microstructures and Morphologies
by
Adie R. Alwen
Dr. Andrea M. Hodge, Advisor
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Ph.D. Degree
DOCTOR OF PHILOSOPHY
(MATERIALS SCIENCE)
August 2024
Copyright 2024 Adie Alwen

ii
Acknowledgements
First, I would like to thank my advisor, Professor Andrea Hodge, for all of her unconditional
support during my Ph.D. I would not have grown into the researcher and person I am today without
her guidance, knowledge, and constant drive to see me succeed. She has continually pushed me to
believe in myself and my work and has taught me so many things both inside and outside the lab.
I want to thank all the members of the Hodge Research Group: Danielle White, Kyle Russell,
Daniel Goodelman, T.J. Oros, Andre Bohn, Ikponmwosa Iyinbor, Ashley Maldonado Otero, Ariel
Capote, Karina Hemmendinger, Alina Garcia Taormina, Joel Bahena, and Chelsea Appleget. I
have learned so much from all of you and am truly grateful for being able to work with such good
friends for the past five years. Stopping by all of your desks to distract you with research questions
and random conversations will always be some of my favorite memories from my Ph.D.
I would like to thank the Core Center of Excellence in Nano Imaging (CNI) at USC, especially
Amir Avishai, Lucas Jordao, and John Curulli for all your assistance and time over these past five
years. I would also like to acknowledge the faculty and staff who have provided me with so much
support during my time at USC, including Monique Garcia, Diana Vuong, Andy Chen, Professor
Ravichandran, Professor Branicio, and so many others. Thank you, Professor Yu-Tsun Shao,
Professor Luis Villalobos, and Professor Richard Brutchey for serving on my dissertation
committee.
I would like to acknowledge and thank the funding sources for my research including the National
Science Foundation (Grant Numbers: OISE-2106597 and DMR-2227178) and the National
Defense Science and Engineering Graduate (NDSEG) Fellowship Program, which is offered by
the Department of Defense and is sponsored by the Air Force Research Laboratory, the Office of

iii
Naval Research, and the Army Research Office. Thanks to these sources I was able to further
develop my skills as a researcher and was afforded the amazing opportunity to participate in
international research endeavors at the Forschungszentrum Julich in the Institute for Energy and
Climate Research, Structure and Function of Materials (IEK-2) in Germany. I would also like to
thank the USC Graduate School for the Provost Fellowship.
I want to thank all of my distinguished collaborators, including Professor Ruth Schwaiger,
Professor Paulo Branicio, Dr. Nicolas Peters, Dr. Jin Wang, Dr. Mirko Zeigner, and Aoyan Liang,
this work would not have been possible without you. To Professor Schwaiger, Dr. Peters, and Dr.
Wang, thank you for being the most amazing hosts while I worked with you in Germany.
I would like to thank my friends. To my friends in LA, thank you for filling these last five years
with amazing memories and helping me make the most of every moment even while doing my
Ph.D. To my friends afar, thank you for constantly reaching out with support and for showing up
to help me celebrate every milestone along the way. To my roommate Zander Bashaw, thank you
for taking the leap of moving in with someone you didn’t know and being an amazing friend,
Wingspan opponent, and quarantine companion. To Nick Kellerman, thank you for always being
a shoulder to lean on, for coming back from New York and living a block away, and for being not
only an incredible friend for these past 15 years, but also a true part of my family.
I would like to thank Loren, Aliza, Rachel, G, Omi, and the rest of the Nizinski family. Thank you
for welcoming me with open arms into your life and for treating me like a son, brother, and
grandchild. Thank you for letting me live with you during the pandemic and for showing so much
love to me and my family. Thank you for being my new family and for always having an open
door if I ever needed anything.

iv
I would like to thank my family. I would never have had these opportunities without the support
and love I received from my parents and brother. Dad, thank you for always being there to answer
every random question I’ve called you with, for always pushing me to be the best person I can be,
and for giving me every opportunity in life. Mom, thank you for being my biggest cheerleader, for
instilling in me this competitive drive to succeed, and for always making me feel so loved. Sid,
thank you for making me take a few steps back from my day to day to laugh, be nostalgic, and
enjoy the things that still make me feel like a big kid.
Finally, I want to thank my girlfriend, Hannah. You have been with me every step of the way, even
as I was applying for Ph.D. programs back at UCLA. Thank you for listening to every presentation,
reading every paper and abstract, and helping me navigate every challenge I faced during my Ph.D.
Thank you for filling my life with so much love, for making me laugh to the point of tears, and for
always being there for me. I am so excited to explore the world and life with you (and Ash, and
Bowie). I love you soooo much.
~ Go Bruins!

v
Table of Contents
List of Tables ............................................................................................................................... viii
List of Figures................................................................................................................................ ix
Abstract........................................................................................................................................ xix
Chapter 1: Introduction................................................................................................................... 1
Chapter 2: Background ................................................................................................................... 3
2.1 Film and Coating Synthesis ............................................................................................. 3
2.1.1 Chemical Vapor Deposition and Atomic Layer Deposition..................................... 3
2.1.2 Electroless Plating and Electrodeposition................................................................. 5
2.1.3 Magnetron Sputtering ............................................................................................... 6
2.1.3.1 Sputtering Techniques....................................................................................... 8
2.1.3.2 Sputtering Apparatus Geometry ...................................................................... 11
2.2 Sputtering Microstructural and Morphological Domains .............................................. 12
2.2.1 Film Growth............................................................................................................ 12
2.2.2 Microstructure and Morphology............................................................................. 16
2.2.3 Nanostructures ........................................................................................................ 18
2.3 Nanotwins....................................................................................................................... 20
2.3.1 Twin Boundaries and Stacking Faults .................................................................... 20
2.3.2 Nanotwin Boundary Synthesis................................................................................ 22
2.3.3 Nanotwin Mechanical Properties............................................................................ 26
2.4 Combinatorial and High-Throughput Synthesis and Characterization .......................... 28
2.4.1 Combinatorial Synthesis......................................................................................... 29
2.4.1.1 Diffusion Couples............................................................................................ 29
2.4.1.2 Additive Manufacturing .................................................................................. 32
2.4.1.3 Compositionally Graded Sputtered Films ....................................................... 33
2.4.2 High-Throughput Characterization......................................................................... 36
2.4.2.1 Composition, Phase, Morphology, and Microstructure................................... 36
2.4.2.2 Electrical Properties......................................................................................... 39

vi
2.4.2.3 Mechanical Properties ..................................................................................... 40
2.5 Summary ........................................................................................................................ 45
Chapter 3: Experimental Methods................................................................................................ 46
3.1 Synthesis Methods.......................................................................................................... 46
3.1.1 Magnetron Sputtering ............................................................................................. 46
3.2 Characterization Techniques.......................................................................................... 48
3.2.1 Langmuir Probe Analysis ....................................................................................... 48
3.2.2 Four-Point Probe Analysis...................................................................................... 50
3.2.3 X-Ray Diffraction (XRD)....................................................................................... 51
3.2.4 Scanning Electron Microscopy (SEM)................................................................... 53
3.2.5 Focused Ion Beam (PFIB) ...................................................................................... 56
3.2.5 Transmission Electron Microscopy (TEM) ............................................................ 58
3.3 Mechanical Testing ........................................................................................................ 62
3.3.1 Nanoindentation...................................................................................................... 62
3.4 Heat Treatments ............................................................................................................. 64
3.4.1 Vacuum Furnace ..................................................................................................... 64
Chapter 4: Correlating sputtering target geometry and film growth............................................. 65
4.1 Introduction .................................................................................................................... 65
4.2 Methods.......................................................................................................................... 68
4.3 Results and Discussion................................................................................................... 70
4.3.1 Plasma analysis and film texturing ......................................................................... 70
4.3.2 Surface morphology characterization ..................................................................... 74
4.3.3 Orientation, grain size, and global morphology...................................................... 77
4.4 Conclusion...................................................................................................................... 81
Chapter 5: Combinatorial and high-throughput nanotwin investigation ...................................... 83
5.1 Introduction ......................................................................................................................... 83
5.2 Experimental Methods........................................................................................................ 86
5.3 Results and Discussion........................................................................................................ 90
5.3.1 Analysis of Material Libraries...................................................................................... 91
5.3.2 Analysis of NT formation............................................................................................. 94

vii
5.3.3 Revised Growth Twinning Model ................................................................................ 98
5.4 Conclusion......................................................................................................................... 102
5.5 Supplementary Materials................................................................................................... 104
Chapter 6: Exploring twinning and phase formation in CuNiAl alloys...................................... 107
6.1 Introduction ....................................................................................................................... 107
6.2 Experimental Methods ...................................................................................................... 109
6.3 Results and Discussion...................................................................................................... 111
6.3.1 Composition and NT Synthesis Domains................................................................... 111
6.3.2 XRD Analysis of Phase Evolution ............................................................................. 116
6.3.3 STEM Analysis of Microstructural Evolution............................................................ 120
6.4 Conclusion......................................................................................................................... 124
6.5 Supplementary Materials................................................................................................... 125
Chapter 7: Conclusions and Future Work................................................................................... 127
7.1 Conclusions.................................................................................................................. 127
7.2 Future Work ................................................................................................................. 129
References................................................................................................................................... 132
Appendix A: Summary of Sputtered Samples............................................................................ 150
Appendix B: High-Throughput Four-Point Probe Data ............................................................. 157
Appendix C: Validation of Growth Twinning Model................................................................. 158

viii
List of Tables
Table 1: Summary of the hollow and planar cathode sputtering parameters and measured
deposition conditions..................................................................................................................73
Table 2: Composition, nanotwin formation, grain width, and hardness data for selected
CuNi samples characterized via STEM, ImageJ, and nanoindentation ......................................98
Table 3: Measured as-sputtered twin spacings and calculated SFEs for the selected
compositions characterized via STEM in Figure 41c-f ............................................................115

ix
List of Figures
Figure 1: CVD schematic showing the flow of reactive gases, the chemical reaction
2AX+H2→2A+2HX, which deposits material A on the surface, and the continued gas flow
removing excess reactant and by-product HX away from the substrate [16]...............................4
Figure 2: Illustration of the (a) electroless plating and (b) electrodeposition solution-based
deposition techniques, highlighting the redox reactions and absence or presence of an
external power source. Adapted from [24] ...................................................................................6
Figure 3: Planar sputtering schematic depicting the flow of an inert gas (Ar) into a vacuum
chamber, the ionization process to form a plasma, and the ion bombardment on the target
surface that ejects sputtered atoms and secondary electrons........................................................8
Figure 4: Structure zone map for energetic deposition processes, where E*
is the normalized
energy from particle flux and T*
is the generalized temperature (homologous temperature
plus temperature shift caused by particle potential energy) [38]................................................10
Figure 5: Illustration of film growth modes depicting (a) Frank van der Merwe layer
growth, (b) Stranski- Krastanov mixed growth, and (c) Volmer-Weber island growth.
Adapted from [52].......................................................................................................................16
Figure 6: Structure zone evolution as a function of homologous temperature (Ts/Tm), where
Ts is the substrate temperature and Tm is the melting temperature. Adapted from [56].............17
Figure 7: Hall-Petch strengthening relationship for a nanoscale material, where d is the
feature or grain size, 𝜎0 is the yield strength of a single crystal and 𝑘𝐻 is the Hall-Petch
strengthening constant (which in this case equals the slope). The breakdown of the HallPetch relationship at small ( < 10nm) feature sizes is indicated by the dashed line ...................19

x
Figure 8: Twinning schematic depicting (a) coherent twin boundaries, (b) incoherent twin
boundaries, and (c) twin boundary energy, 𝛾, as a function of boundary coherency, 𝜑.
Adapted from [50].......................................................................................................................21
Figure 9: Illustration of stacking fault formation in an FCC metal, where (a) represents a
perfect FCC stacking sequence and (b) represents the formation of an HCP structure at a
stacking fault...............................................................................................................................22
Figure 10: Deformation twin nucleation and growth in an FCC crystal. (a) is a perfect FCC
lattice, (b) is the nucleation of a stacking fault due to stress induced planar slip, and (c) is
the growth of deformation twin boundary after consecutive slip of adjacent planes. The
arrows highlight sheer stress on each plane before slip occurs ..................................................23
Figure 11: Cross-sectional STEM image of growth nanotwin formation in a sputtered CuNi
film, where the film growth direction is indicated by the black arrow.......................................26
Figure 12: Stress-strain curves for coarse grained (CG), nanocrystalline (ICG), and
nanotwinned copper (A,B, and C). The average nanotwin spacings for samples A, B and C
are 96nm, 30nm, and 15nm, respectively [81]............................................................................28
Figure 13: Illustrations of (a) a diffusion couple comprised of materials A and B, (b)
composition change due to interdiffusion, and (c) the flux of atoms and vacancies along
the diffusion couple. Adapted from [50].....................................................................................31
Figure 14: Sputtered wedge-type structure for a ternary alloy, where materials A, B, and C
are deposited from sputtering sources spaced 120° apart [102] .................................................34
Figure 15: Co-sputtering schematic depicting the compositional gradient across a substrate
resulting from the variation in flux caused by distance from the sputtering sources. Each
square on the substrate represents a compositionally unique sample that is 5x5mm in size......35

xi
Figure 16: Computer Vision (CV) analysis of an SEM image of nanoparticles where (a) is
the original SEM image, (b) is the automatic identification of particles and background, (c)
is object detection, and (d) is the assignment of pixels to a given feature [108]........................39
Figure 17: Dog-bone synthesis process for high-throughput micro-tensile testing of
deposited films. In this example (a) the silicon wafer is thermally oxidized and (b and c)
spin coated with a lift-off system and photoresist that is (d) shaped into a dog-bone via
photolithography. (e) The excess resist is then removed and (f) the structures are coated
using magnetron sputtering. Finally, (g) the dog-bone patterned film is removed from the
resist by being immersed in acetone and (h) the specimen is micro-tensile tested. Adapted
from [130]...................................................................................................................................42
Figure 18: Top-surface characterization of TA15 alloy using (a) back scattered electron
imaging and (b and d) high-speed nanoindentation mapping of hardness and modulus. (c)
Shows change in mechanical properties over one line of indents [136].....................................44
Figure 19: Schematic of the hollow cathode (a) and planar cathode (b) sputtering target
geometries, highlighting substrate positioning, varied line of sight, and particle deposition.....47
Figure 20: Langmuir probe schematic illustrating the insertion of a probe (labeled P) into
the plasma, where Vdis is the discharge voltage used to create the plasma, Idis is the
discharge current of the plasma and Vp and Ip are the probe bias and measured current,
respectively. Vp and Ip are the values used to generate I-V curves [137]...................................48
Figure 21: Representative current-voltage, or I-V curve, that can be generated by the
Langmuir probe, highlighting the asymptotic electron and ion saturation currents. Adapted
from [138]...................................................................................................................................49

xii
Figure 22: Four-point probe schematic depicting the probe placement and uniform spacing
at the sample surface, the current applied to the outer probes (I), and the voltage
measurements at the inner probes (V). Adapted from [140] ......................................................51
Figure 23: Normalized XRD spectra of planar cathode (dashed lines) and hollow cathode
(solid lines) sputtered CuAl films. Films were deposited at a power density of 1.2 W cm-2
and at gas pressures ranging from 3 to 30 mTorr. The respective gas pressures are color
coded. Adapted from [141].........................................................................................................52
Figure 24: Interaction of the SEM electron beam with sample, highlighting the interaction
volume and production of secondary electrons (SE), backscattered electrons (BSE), and Xrays. Adapted from [143]............................................................................................................54
Figure 25: SEM image of the top surface morphology of a CuAl sample .................................55
Figure 26: Kikuchi patterns derived from backscattered electrons, which are used to map
grain orientation [144] ................................................................................................................56
Figure 27: Overview of the PFIB lift-out process, depicting (a) the deposition of a
protective cap, (b) the milling process to create the lamella, (c) the attachment of the
lamella to the micromanipulator, and (d) the attachment of the lamella to the TEM grid
[148]............................................................................................................................................58
Figure 28: Schematic of TEM operation in (A) selected area electron diffraction mode and
(B) imaging mode [149]..............................................................................................................60
Figure 29: Illustration of (a) the BF and ADF detector locations, (b) the intensities of the
transmitted and scattered electrons, and a comparison of the resulting (C) ADF STEM and
(D) BF STEM images [149] .......................................................................................................61

xiii
Figure 30: Overview of (a) the Berkovich indenter geometry and (b) the indent dimensions,
showing (c) a representative load-displacement curve that can be obtained with a
Berkovich indenter. Here A is the projected contact area, θ is the face angle, and hc is the
depth of contact measured from the apex of the indenter. Adapted from [133].........................63
Figure 31: (a) Plotted Langmuir probe measured ion densities plots for hollow and planar
cathode sputtering plasmas as a function of power density at varying Ar pressures. (b) Xray diffraction (XRD) patterns for the films sputtered at the low power density (0.4 W cm2
), which are indicated by the black oval on the left side of Figure 31a. (c) XRD patterns
for the films sputtered at the high power density (1.2 W cm-2
), which are indicated by the
black oval on the right side of Figure 31a. Hollow cathode samples and measurements are
represented by a solid line, the planar cathode samples and measurements have a dashed
line. Respective gas pressures are color coded and the main XRD intensity peaks are
labeled.........................................................................................................................................72
Figure 32: Scanning electron microscopy (SEM) micrographs of the as deposited top
surface morphologies at 0.4 W cm-2
power density for the hollow cathode (a-d) and planar
cathode (e-f) films sputtered at gas pressures between 3 to 30 mTorr. The hollow cathode
images have a solid border, the planar cathode images have a dashed border, and gas
pressures are color coded. Feature sizes for each film are shown to the right of the
corresponding SEM image..........................................................................................................75
Figure 33: Scanning electron microscopy (SEM) micrographs of the as deposited top
surface morphologies at 1.2 W cm-2
power density for the hollow cathode (a-d) and planar
cathode (e-f) films sputtered at gas pressures between 3 to 30 mTorr. The hollow cathode
images have a solid border, the planar cathode images have a dashed border, and gas

xiv
pressures are color coded. Feature sizes for each film are shown to the right of the
corresponding SEM image..........................................................................................................77
Figure 34: Electron backscatter diffraction (EBSD) grain orientation maps for planar
(dashed border) and hollow cathode (solid borders) sputtered films. The vertical column
(a-c) shows samples sputtered at the same power density (1.2 W cm-2
, green border) with
increasing ion densities. The horizontal row (c-e) displays samples sputtered with similar
ion densities at both the low (0.4 W cm-2
, gray border) and high (1.2 W cm-2
) power
densities. The inverse pole figure (IPF) IPF triangle defining grain orientations is to the
right of the EBSD scans..............................................................................................................79
Figure 35: Graphical summary of planar and hollow cathode top-surface and crosssectional morphologies with respect to ion density (x-axis) and homologous temperature
(T/Tm) (y-axis), where coating thickness is shown in the z-axis. The as-sputtered top
surface morphologies are outlined in red and the cross-sectional morphologies are outlined
in blue; all boxes are roughly 1.5 micron in length. Sample labels correspond to Table 1 ........81
Figure 36: Schematic of the combinatorial synthesis process. (a) Illustration of the cosputtering technique, where two materials (Cu and Ni) are simultaneously deposited to
form a compositional gradient. (b) Image showing a combinatorial array of co-sputtered
samples deposited on a 10 cm wafer, where each 5x5mm square represents a unique CuNi
sample .........................................................................................................................................87
Figure 37: Analysis of CuNi combinatorial samples. (a) Compositional maps obtained via
EDX for two sets of combinatorial wafers, with the first wafer (samples 1-169) containing
samples with compositions ranging from Cu – 6.8 at% Ni to Cu – 35.5 at% Ni and the
second wafer (samples 170-338) containing samples with compositions ranging from Cu –

xv
12.5 at% Ni to Cu – 58.8 at% Ni. (b) Corresponding hardness heat maps for the two wafers,
where hardness values were measured using high-throughput nanoindentation. (c) Selected
XRD patterns from the samples with yellow borders in 2a ........................................................93
Figure 38: Cross-sectional HAADF STEM micrographs for selected CuNi samples
highlighting the change in growth nanotwin formation as Ni concentration increases from
(a) 7.1 at % Ni to (f) 58.8 at % Ni...............................................................................................96
Figure 39: Nanotwin formation in CuNi alloys and quantitative assessment. (a) STEM
micrograph comparing nanotwinned (NT) and non-nanotwinned (Not NT) grains. A
nanotwinned grain is defined as a grain with the majority of its area containing twin
boundaries spaced ≤ 100nm apart. (b) Plotted comparison of the quantified percentage of
NT grains relative to the total observed grains, as a function of Ni concentration.....................97
Figure 40: Comparison of measured and predicted twin boundary spacings as a function of
stacking fault energy (SFE). The black data points represent the measured twin spacings,
while the predicted values for the Zhang model and updated model are shown by the red
and blue points, respectively [79].............................................................................................102
Figure S1: Convergence tests for stacking fault energy (SFE) as a function of (a) gliding
plane area and (b) simulation cell thickness for a Cu – 50 at% Ni alloy. Illustrations of the
atomic structure of the alloy system (c) before and (d) after generation of the stacking fault.
The transition from FCC to HCP structure upon faulting is marked, highlighting the two
layers of the HCP structure indicative of stacking fault formation ..........................................104
Figure S2: Top surface morphologies of selected as-sputtered CuNi alloys with varying Ni
contents including (a) Cu - 7.1 at% Ni, (b) Cu - 17.4 at% Ni, (c) Cu - 27.7 at% Ni, (d) Cu
- 33.8 at% Ni, (e) Cu - 48.1 at% Ni, and (f) Cu - 58.8 at% Ni .................................................105

xvi
Fig S3: Influence of Ni composition on mechanical and microstructural properties of CuNi
alloys. (a) Comparison of the change in hardness (GPa) and the percentage of nanotwinned
grains (% NT) as a function of the Ni concentration. (b) Compositional dependence of solid
solution (SS) and Hall-Petch strengthening (NT Hall-Petch and Grain Width Hall-Petch)
shown as a function of the Ni concentration.............................................................................106
Figure 41: Analysis of CuNiAl composition and NT formation. (a) Composition map for
the CuNiAl combinatorial array obtained via EDX, containing samples with compositions
ranging from 21.2 - 77.1 at% Cu, 13.4 - 51.2 at% Ni and 8.4 - 46.1 at% Al. The areas in
green, blue, and red indicate greater Cu, Ni, or Al content, respectively. (b) CuNiAl
compositions plotted on a ternary diagram highlighting the occupied composition space.
(c-f) Cross-sectional HAADF STEM micrographs and EDX maps for as-sputtered CuNiAl
alloys taken from the compositions noted with black circles in Figure 1b. The samples were
taken from Cu rich (1c), Ni rich (1d), Al rich (1e), and intermediate (1f) composition
regions.......................................................................................................................................114
Figure 42: High-throughput XRD analysis of the as-sputtered and annealed CuNiAl
combinatorial arrays. (a) and (b) depict an overview of the XRD patterns for the assputtered and annealed CuNiAl samples, respectively, where XRD patterns with high
background noise were removed to improve peak resolution. (c) and (d) highlight the two
types of XRD patterns (plotted on a log scale) that were observed in the combinatorial
samples: an FCC solid solution (2c) or a combination of an FCC and a B2 NiAl phase (2d).
In 2a-d, the FCC peaks are represented by the (*) and the B2 NiAl peaks are represented
by the (+)...................................................................................................................................117

xvii
Figure 43: Analysis of compositional effects on XRD diffraction patterns and CuNiAl
phase volume fractions. (a) FCC volume fraction heat map for the annealed CuNiAl
combinatorial array of samples, where FCC volume fraction was observed to vary from
0.06 to 0.96. A brighter green indicates a higher FCC volume fraction and a darker green
indicates a lower FCC volume fraction. (b-d) Selected XRD patterns plotted on a
normalized log scale from annealed CuNiAl samples with constant Al (43b), Cu (43c), and
Ni (43d) content. The arrows to the right of the selected diffractograms highlight the
changes in composition. In 43b-d, the FCC peaks are represented by the (*) and the B2
NiAl peaks are represented by the (+) ......................................................................................120
Figure 44: Cross-sectional HAADF STEM micrographs and Cu (green), Ni (blue), and Al
(red) EDX composition maps for four selected annealed CuNiAl compositions highlighting
changes in the cross-sectional microstructure and B2 NiAl phase formation. (a-c) Denote
the constant Ni set of samples in Figure 43d, with corresponding compositions of
Cu72.4Ni17.6Al10.0 (44a), Cu58.6Ni17.5Al23.9 (44b), and Cu45.1Ni16.5Al38.4 (44c) and (44d)
examines the role of varied Ni content on phase formation with a composition of
Cu53.0Ni35.5Al11.5........................................................................................................................123
Figure S4: Cross-sectional HAADF STEM micrographs and EDX maps for the
corresponding as-sputtered versions of the CuNiAl alloys examined in Figures 44a-c,
which had compositions of Cu72.4Ni17.6Al10.0 (S4a), Cu58.6Ni17.5Al23.9 (S4b), and
Cu45.1Ni16.5Al38.4 (S4c)...............................................................................................................126
Figure B1: Sheet resistance heatmaps for the two combinatorial arrays of CuNi samples
discussed in Chapter 5 and analyzed in Figure 37. The regions in dark red have lower sheet
resistivities and the regions in bright white have higher sheet resistivities..............................157

xviii
Figure B2: Sheet resistance heatmaps for the as-sputtered (a) and annealed (b) CuNiAl
combinatorial arrays discussed in Chapter 6 and analyzed in Figure 41. The regions in dark
red have lower sheet resistivities and the regions in bright white have higher sheet
resistivities. The missing points in the annealed map are due to samples that have deadhered or frayed due to the annealing process........................................................................157
Figure C1: Comparison of measured and predicted twin boundary spacings as a function
of stacking fault energy (SFE) for the CuNi alloys examined in Chapter 6 and for other
materials studied in literature. The black data points represent the measured twin spacings,
while the predicted values for the Zhang model and Alwen model are shown by the red and
blue points, respectively............................................................................................................158

xix
Abstract
The development of future technologies requires the discovery of new materials by exploring and
optimizing unique microstructural and morphological combinations within a vast compositional
space. However, navigating the nearly unlimited number of combinations to find promising
material systems is a major challenge. Thus, more efficient experimental techniques and
approaches are needed to expedite discovery when investigating unexplored synthesis domains.
This work seeks to accelerate material discovery by globally analyzing large material synthesis
spaces as a function of both changing processing conditions and composition using magnetron
sputtering. To analyze the role of varied processing conditions, film deposition in the planar and
hollow cathode was compared to examine how changes in deposition parameters influenced film
growth, microstructure, and morphology. With respect to composition, large compositional
domains were rapidly studied for changes in nanostructure formation using combinatorial cosputtering and high- and low-throughput characterization techniques. Overall, this work highlights
novel approaches to study composition-processing-property relationships in large synthesis
spaces, laying a foundation for accelerated material discovery.

1
Chapter 1: Introduction
The development of advanced technologies is dependent on the discovery and optimization
of new materials with novel properties. Research aimed at finding these materials holds immense
potential, given the nearly unlimited number of combinations of composition, microstructure, and
morphology. However, this vast synthesis space also poses a challenge to material discovery, as
only a small fraction of these combinations can be currently explored. Thus, to enable faster
identification of promising materials systems, more efficient experimental and computational
methodologies need to be leveraged. To date, current research has employed a number of
multidimensional approaches to more efficiently investigate material synthesis domains, including
mapping processing-microstructure relationships and screening large composition spaces using
combinatorial and high-throughput techniques. Still, further work is necessary, as these studies
only link a few variables and fail to understand more complex material systems. Therefore, this
dissertation seeks to enable expedited discovery of novel and optimized materials by utilizing
sputtering to globally and comprehensively investigate fundamental processing-compositionproperty relationships. Magnetron sputtering is a versatile physical vapor deposition technique that
is widely used for film synthesis that offers many key advantages in industry and research; notably,
the ability to manipulate composition, access nanoscale size effects, and engineer surface
microstructures and morphologies. With respect to synthesis and processing, planar and hollow
cathode sputtering are leveraged to investigate novel relationships between deposition conditions
and film growth, highlighting target geometry as a novel deposition variable that can be used to
tailor film morphology. To examine the role of composition, a combinatorial methodology
utilizing co-sputtering in conjunction with high- and low-throughput characterization techniques
was implemented to screen complex composition spaces for varied nanotwin formation and

2
elucidate relationships with compositionally dependent properties such as stacking fault energy.
Overall, by coupling analysis of sputtering processing variables with combinatorial and highthroughput experimentation, this study provides both a fundamental understanding and
experimental framework that can be used to accelerate material discovery in unexplored sputtering
domains.

3
Chapter 2: Background
2.1 Film and Coating Synthesis
Films and coatings can be synthesized via a variety of deposition methods for a large range
of applications, such as, optics [1-3], thermal protection [4, 5], anti-corrosion [6-8],
semiconductors and electronic materials [9-11], additively manufactured and hierarchical
structures [12-14], and functional mechanical coatings [8, 15] to name a few. Some of the most
commonly utilized film synthesis methods are chemical vapor deposition (CVD), atomic layer
deposition (ALD), electroless/electroplating, and physical vapor deposition (PVD) techniques like
magnetron sputtering. Each deposition method can be leveraged to create functional coatings;
however, there are tradeoffs with respect to deposition material, coating coverage, deposition rates,
and film growth mechanisms. Given the direct relationship between material processing and the
resulting properties, the film synthesis techniques must be examined to enable further manipulation
and discovery of film processing-property relationships.
2.1.1 Chemical Vapor Deposition and Atomic Layer Deposition
CVD has been used for coating applications ranging from dielectric and conductive
coatings to oxidation and thermal barriers [16-18]. During CVD, a film is grown from a vapor via
chemical reactions near the substrate surface. The process is shown in Figure 1, where a reactive
gas is flowed into a chamber, the film grows as reactions deposit material on the substrate, and
finally, the gas flow removes by-products and excess reactants [16]. Since the process is driven by
chemical reactions at the sample surface, CVD can deposit uniform and fully dense coatings even
on substrates with complex topologies [12, 16]. Film morphology, microstructure, and growth
rates can be altered by adjusting deposition conditions such as the temperature, partial pressures,

4
and ion bombardment. This has resulted in several specialized CVD techniques such as Ultra-High
Vacuum CVD (UHVCVD), Plasma-Enhanced CVD (PECVD), Rapid Thermal CVD (RTCVD),
and Atomic Layer Deposition (ALD) [16, 19]. ALD is a variation of CVD which can control
coating thickness and conformality down to nanometer or even Angstrom precision on both planar
and complex three-dimensional substrates. By sequentially pulsing and purging reactants and byproducts to activate a surface and induce a chemical reaction, ALD can iteratively deposit material
one monolayer at a time. This enables the technique to control film growth rates down to the order
of 1 Å/cycle [20-22].
Figure 1: CVD schematic showing the flow of reactive gases, the chemical reaction
2AX+H2→2A+2HX, which deposits material A on the surface, and the continued gas flow
removing excess reactant and by-product HX away from the substrate [16].
Although CVD techniques enable the deposition of conformal coatings, they also have
disadvantages that restrict their applicability to other material studies, including a confined
material workspace and, in the case of ALD, low deposition rates (~1 nm/minute) which limits
studies to primarily investigate thin films (thicknesses < 1 micron) [21]. Film composition can be
changed by varying the vapor stoichiometry and partial pressures, but ultimately, CVD material
selection is dependent on the available chemical reactions during deposition. As a result, CVD is
primarily limited to depositing ceramic, oxide, and nitride films and is unable to deposit more

5
compositionally complex coatings like metallic alloys [16, 19]. As such, other coating techniques
have been investigated to access higher deposition rates and different compositional spaces with
metals and metallic alloys.
2.1.2 Electroless Plating and Electrodeposition
Electroless plating and electrodeposition are synthesis techniques that are readily used to
synthesize metal films [23-25]. During electroless plating, a substrate is submerged into a chemical
bath containing aqueous ions and a reducing or oxidizing agent and an autocatalytic reductionoxidation (redox) chemical reaction is utilized to deposit material [23]. Deposition automatically
occurs without an applied external bias as the potential between the ions and the redox agents
within the chemical bath cause an internal electric current [23]. Reducing agents are used for
cathodic processes to deposit single element metals or multielement alloys, while oxidizing agents
are used for anodic processes to from oxides and other non-metallic compounds [26].
Electrodeposition also submerges a substrate into a chemical bath, but, unlike electroless plating,
electrodeposition requires an external electric bias and a conductive substrate for the chemical
reactions and deposition to occur [27]. These two deposition processes are shown in Figure 2,
where the auto-catalytic redox deposition process for electroless plating is highlighted in Figure
2a and deposition with an applied external current is shown in Figure 2b [24]. Electrochemical
deposition techniques like electroless plating and electrodeposition offer a number of advantages
in film synthesis in addition to expanding the compositional workspace of CVD techniques.
Namely, they can achieve higher growth rates (>50 nm/min), are industrially scalable, allow
conformal deposition on complex substrates, and can influence coating microstructure and feature
sizes with their deposition parameters [24, 28, 29]. For electrodeposition, the applied external

6
current can be varied to increase or decrease the deposition rate, which alters the growth kinetics
and resulting grain size [30]. In electroless plating, microstructure, morphology, and other material
properties have been shown to be dependent on growth parameters such as the solution temperature
and concentration of aqueous ions and reduction or oxidizing agents [31, 32]. While both
techniques have been used to deposit single elements, alloys, ceramics, polymers, and other
compounds, the nature of the chemically driven deposition method limits the compositional range
of metallic coating materials to primarily single elements or binary alloy systems [33]. Thus, other
coating techniques like magnetron sputtering, which leverage the physical transport of atoms to a
surface, have been used to remove the compositional restrictions of methods that are dependent on
chemical reactions.
Figure 2: Illustration of the (a) electroless plating and (b) electrodeposition solution-based
deposition techniques, highlighting the redox reactions and absence or presence of an external
power source. Adapted from [24].
2.1.3 Magnetron Sputtering
Magnetron sputtering is a widely used and versatile PVD process due to its ability to readily
deposit films using nearly any material system, including single elements, ceramics, and complex
metallic alloys [34]. PVD encapsules a set of non-equilibrium deposition techniques that transfer

7
atoms from a solid or liquid material target to a substrate using physical mechanisms such as
evaporation or collisional impact. These techniques are referred to as non-equilibrium because the
resulting film can be comprised of metastable phases and/or solid solutions. Generally, these
coating methods operate in a vacuum environment to promote vapor transportation and are devoid
of chemical reactions, except in the case of some reactive PVD techniques [34, 35]. Figure 3
depicts the magnetron sputtering process, where a solid target is bombarded with high energy ions
from a plasma to eject target material and form a vapor that is deposited on a substrate [34]. This
process and its controllable deposition parameters are explained in further detail in Section 3.1. In
addition to its wide compositional workspace, sputtering offers a number of advantages during
deposition, including the ability to tailor microstructure, achieve compositional gradients, and
synthesize complex structures such as nanometallic multilayers [36-38]. However, unlike CVD
and electrodeposition techniques, sputtering faces challenges in synthesizing conformal and
uniform coatings on substrates with complex topologies due its momentum driven mechanism for
coating deposition. Sputtering was selected for use in this work because it enables both
compositional and processing condition-based exploration of film synthesis domains. Thus, the
following sections highlight sputtering techniques and relationships used to control and tailor film
deposition.

8
Figure 3: Planar sputtering schematic depicting the flow of an inert gas (Ar) into a vacuum
chamber, the ionization process to form a plasma, and the ion bombardment on the target surface
that ejects sputtered atoms and secondary electrons.
2.1.3.1 Sputtering Techniques
To improve uniformity and control of film properties during deposition, multiple types of
magnetron sputtering processes have been developed, including direct current (DC) sputtering,
radio frequency (RF) sputtering, ionized PVD (I-PVD), and high-power impulse magnetron
sputtering (HIPIMS) [3, 34, 39]. DC and RF sputtering refers to the type of power supply used to
bias the electrodes when creating the sputtering plasma. DC sputtering uses a DC power source,

9
which enables the technique to effectively deposit conductive materials, like metals, and can
achieve high deposition rates (> 100 nm/min) by keeping the cathode at a constant negative
potential to maximize ion bombardment [3]. In contrast, by using an alternating current (AC)
power supply, RF sputtering excels at depositing non-conductive materials, like ceramics and
dielectrics, by alternating positive and negative cathode voltages. This is at the cost of reduced
deposition rates (as low as 1 nm/min depending on the deposition conditions), as the periodically
positive cathode potential repels plasma ions from the target surface [3, 34, 40]. In addition to
altering deposition rates, changes in sputtering power affect the homologous temperature and
energy from ion and secondary electron bombardment on the substrate surface during deposition
[39, 41, 42]. Qualitative PVD structure zone maps have been developed to predict how changes in
these deposition properties affect the resulting film microstructure. For example, Figure 4 depicts
a structure zone map created by Anders that highlights a transition from columnar to equiaxed
grain growth with increasing homologous temperature and energized particle bombardment [38].

10
Figure 4: Structure zone map for energetic deposition processes, where E*
is the normalized
energy from particle flux and T*
is the generalized temperature (homologous temperature plus
temperature shift caused by particle potential energy) [38].
I-PVD and HIPIMS take advantage of these film growth relationships by intentionally
increasing ionized particle bombardment during deposition. An I-PVD sputtering process is
achieved when the deposition flux contains more ions than neutral atoms. To achieve this, I-PVD
techniques increase sputtering plasma densities and sputtered atom ionization by introducing
additional apparatus such as inductively coupled-plasma (ICP) sources or electron cyclotron
resonance discharges that create high-density plasmas in the region between the sputtering target
and the substrate [39]. HIPIMS is a specialized I-PVD process that utilizes high sputtering powers
to increase plasma densities and sputtered atom ionization during deposition. In conventional DC
magnetron sputtering, the maximum sputtering power is limited by the thermal load placed on the

11
target; to overcome this limitation, HIPIMS applies a short power duty-cycle (i.e. impulse) to
increase power without surpassing the max thermal load [39, 43].
2.1.3.2 Sputtering Apparatus Geometry
Film synthesis via magnetron sputtering can also be adapted by altering the deposition
configuration, either by changing the target geometry or introducing additional sputtering sources.
Two of the most commonly used target geometries are the hollow cathode (also known as inverted
cylindrical magnetron) and planar cathode; these configurations are shown and further discussed
in section 3.1.1 [44-46]. Altering the target geometry will directly affect the deposition line of
sight when coating a substrate, which can influence the resulting film growth, coating coverage,
and texture. Additionally, similar to I-PVD techniques, changing the target geometry can also
affect plasma densities and ion bombardment on the substrate during deposition. The hollow
cathode in particular has been observed to achieve greater plasma densities than the planar
configuration. This is due to the geometry and magnetic field promoting gas ionization by naturally
trapping electrons and extending their mean free path near the target surface. In fact, hollow
cathodes can achieve large enough plasma densities to be used as a separate ion source for I-PVD
techniques [34, 44].
Deposition with multiple sputtering sources is known as co-sputtering, which is a powerful
tool for studying complex coating structures, such as nanomultilayers, or exploring compositional
spaces with increased stoichiometric control [36, 47]. Given the additional deposition sources, cosputtering can manipulate film growth and microstructure using multiple sputtering techniques in
conjunction. For example, a study by Lelis et. al. investigated tailoring TiO2 film microstructure
using pulsed DC sputtering on one source and RF deposition on another. They showed that
combining the two approaches could alter nucleation and growth of TiO2 crystalline domains [48].

12
Compositionally, co-sputtering can enable rapid exploration of compositional space by depositing
films with compositional gradients [47]. This application of co-sputtering is the foundation of
many combinatorial and high-throughput materials research endeavors and will be discussed in
depth in section 2.4.1.3.
2.2 Sputtering Microstructural and Morphological Domains
The previous section provided an overview of the controllable parameters and deposition
geometries for magnetron sputtering. However, there remains the issue of understanding how
varying sputtering processing parameters and composition alters material properties. To discern
what affects material characteristics, the links between sputtering and microstructure must be
investigated. Thus, the following section examines sputtering microstructural and morphological
domains to understand how processing-microstructure relationships affect film growth,
morphology, and microstructure.
2.2.1 Film Growth
Sputtered film microstructure and morphology can first be understood by analyzing the
growth process, which includes nucleation and layer formation [49]. Nucleation encompasses the
transition of a particle from either a liquid or gaseous state into a solid. This typically occurs once
a particle is cooled below its melting temperature (Tm), where it then becomes thermodynamically
favorable to form a solid. For crystalline materials, grain nucleation is dictated by the
thermodynamic tradeoff between the decreasing volume free energy and increasing surface free
energy. This is highlighted below in Equation 1 [50].
∆𝐺𝑟 = −
4
3
𝜋𝑟
3∆𝐺𝑣 + 4𝜋𝑟
2𝛾𝑠
(1)

13
Where r is the radius of the nucleating grain, ∆𝐺𝑟
is the change in free energy associated with
nucleation, ∆𝐺𝑣 is the volume free energy, and 𝛾𝑠
is the surface free energy. A negative change in
free energy, ∆𝐺𝑟
, indicates an energetically favorable nucleation process. Thus, it can be seen that
due to the larger exponent associated with volume free energy, it typically is energetically
favorable to increase nucleating particle radii. However, at very small radii increased surface free
energy prevents nucleation by causing a positive change in Gibb’s free energy. The critical radii
length at which point it becomes thermodynamically favorable to nucleate can be determined by
taking the derivative of equation 1 and solving for the radius, r, which is shown below in Equation
2 [50].
𝑟
∗ =
2𝛾𝑠
∆𝐺𝑣
(2)
Here 𝑟
∗
is the critical radii length. Volume free energy, ∆𝐺𝑣, is derived below in Equation 3 by
calculating the difference in volume free energy between a solid and another phase, in this case a
liquid.
∆𝐺𝑣 = 𝐺𝑣
𝑙 − 𝐺𝑣
𝑠
∆𝐺𝑣 = (𝐻𝑙 − 𝑇𝑆𝑙) − (𝐻𝑠 − 𝑇𝑆𝑠)
𝐶𝑙𝑜𝑠𝑒 𝑡𝑜 𝑇 = 𝑇𝑚 → ∆𝐺𝑣 = 0 = 𝐿 − 𝑇𝑚∆𝑆
∆𝑆 =
𝐿
𝑇𝑚
𝑆𝑜, ∆𝐺𝑣 =
𝐿(𝑇𝑚−𝑇)
𝑇𝑚
=
𝐿∆𝑇
𝑇𝑚
(3)
Where 𝐺𝑣
𝑙
and 𝐺𝑣
𝑠
are the volume free energies for a liquid and solid, 𝐻𝑙and 𝐻𝑠 are the liquid and
solid enthalpies, 𝑆𝑙 and 𝑆𝑠 are the liquid and solid entropies, 𝐿 is the difference in enthalpy between
phases (𝐻𝑙 − 𝐻𝑠), ∆𝑆 is the difference in entropies (𝑆𝑙 − 𝑆𝑠), 𝑇𝑚 is the melting temperature, and
∆𝑇 is the undercooling below the melting temperature. Equation 4 substitutes this derivation of

14
volume free energy into the equation for critical radii, revealing a direct relationship between grain
size and undercooling [50].
𝑟
∗ =
2𝛾𝑠𝑇𝑚
𝐿∆𝑇
(4)
This explains why greater undercooling yields smaller grain sizes during nucleation. In the case of
sputtering, given the high melting temperature associated with metals and ceramics and the low
temperature vapor formation caused by plasma ablation, there are large undercooling values (thus
high cooling rates) that enable the technique to form nanocrystalline grains. In fact, if the critical
radii becomes less than the lattice parameter of a unit cell, then the deposition method can deposit
an amorphous phase [37, 50, 51]
As grains nucleate there are three modes of film growth that can occur. The first is island,
or Volmer-Weber, growth where the deposited material nucleates and then grows in the Z direction
(normal to the substrate) before completing a full monolayer. This growth mechanism generally
occurs when the deposited atoms form stronger bonds with each other than with the substrate. The
second mode is layer, or Frank-van der Merwe, growth where the atoms tend to condense and form
a full layer before growing in the Z direction. In contrast to island growth, the incoming atoms
form stronger bonds with the substrate than each other. Finally, the last mode is a combination of
the two growth mechanisms known as Stranski-Krastanov growth, where at first monolayers form
and then island growth becomes favorable; this mode can be observed due to binding energies
changing with increasing film thickness or because of certain lattice mismatches between the film
and substrate [52, 53]. The three growth modes are shown in Figure 5, where 5a is layer growth,
5b is mixed growth, and 5c is island growth [52].

15
Figure 5: Illustration of film growth modes depicting (a) Frank van der Merwe layer growth, (b)
Stranski- Krastanov mixed growth, and (c) Volmer-Weber island growth. Adapted from [52].
Finally, since PVD techniques like sputtering are non-equilibrium processes, there are
kinetic considerations that can affect film growth, specifically, the particle mobility during
deposition. Temperature, deposition energy, and gas pressure are key parameters than can affect
the distance a particle can travel on a substrate before nucleating to form a grain [34, 37, 49].
Increasing the homologous temperature (which is a normalization of the substrate temperature
divided by the melting temperature, T/Tm) increases the diffusion length scale and can enable a
transition from interplanar to bulk diffusion [37, 38, 49]. Similarly, increasing sputtered particle
energy or ion bombardment using techniques like HIPIMS or I-PVD increases adatom mobility
by providing more energy to move along the substrate/film surface [34, 38, 39]. Gas pressure is
correlated to particle energy as it dictates the mean free path of a sputtered atom travelling in the

16
sputtering chamber. A shorter mean free path indicates more particle collisions and therefore lower
energy when arriving at the substrate [37, 54].
2.2.2 Microstructure and Morphology
The correlations between deposition conditions, film growth modes, and particle mobility
allow sputtering to tailor coating properties. Sputtered film microstructural and morphological
domains have been linked to processing conditions using structure zone maps, such as the one seen
in Figure 4, which categorize films into four distinct structural regions, Zone 1, Zone T, Zone 2,
and Zone 3 [38, 49, 55, 56]. Zone 1 occurs at low homologous temperatures (T/Tm < 0.2) and
particles energies and is characterized by fibrous nanocrystalline columnar grains with voids
located at grain boundaries due to limited particle mobility. At higher homologous temperature
(0.2 < T/Tm < 0.4) or particle energies, increased diffusivity with limited grain boundary mobility
leads to the formation of Zone T structures. These are inhomogeneous films with slightly larger
V-shaped grains, weak texturing, and island growth that leads to increased surface area to volume
ratios. Homogeneous Zone 2 structures form at high particle energies and homologous
temperatures (T/Tm > 0.4), as increased bulk diffusion and adatom mobility enable large columnar
grains to grow perpendicular to the substrate with uniform texture. Finally, in Zone 3 there are
equiaxed three dimensional grains that form either due to bulk diffusion and recrystallization at
the highest homologous temperatures or as the result of impeded and blocked crystal growth during
deposition. Similar to a post annealing treatment, the former leads to larger grain sizes that are
more representative of bulk materials, while the latter yields nanocrystalline grains by periodically
interrupting Zone T columnar growth [37, 49, 56, 57]. The evolution of film structure and grain
size from Zone 1 to Zone 3 as a function of homologous temperature is depicted in Figure 6 [56].

17
Figure 6: Structure zone evolution as a function of homologous temperature (Ts/Tm), where Ts is
the substrate temperature and Tm is the melting temperature. Adapted from [56].
Given the directionality associated with ejecting a particle and depositing it on a surface,
sputtering can also affect film structure by varying the apparatus geometry used to deposit a film.
Work done by Greczynski et. al. demonstrated that sputtering texture and columnar growth are
influenced by the deposition angle-of-incidence with respect to the substrate [43]. Similarly,
Thornton noted that the hollow cathode target geometry can cause shadowing affects and lead to
different structure zone regions than a planar target [45, 58]. The impact of sputtered particle angle
of incidence is further highlighted in coatings on 3-dimensional substrates, where varying structure
zones, coating thicknesses, and film densities have been observed on different surfaces of the same
substrate [58, 59]. Although uniformity remains a challenge, deposition geometry offers another
route to influence microstructure and morphology. Collectively, these relationships between
deposition parameters, cathode geometries, and film growth enable sputtering to control film
characteristics and can be leveraged to further improve material properties by exploring
nanocrystalline microstructural domains (nanodomains) and accessing nanoscale size effects.

18
2.2.3 Nanostructures
Sputtering can enable the exploration of nanodomains by synthesizing coatings with
homogeneous or heterogeneous nanostructures, which have notable properties due to their
nanoscale feature sizes (on the order of 1-100 nm) and high interfacial densities [60, 61]. These
materials can have either uniform (homogeneous) or complex and variable (heterogeneous)
microstructures that offer novel combinations of strength and ductility. Regarding strength, grain
boundaries and other interfaces impede dislocation and crack propagation, which can result in
orders of magnitude increases in yield strength as compared to coarse grained bulk materials. This
strengthening mechanism can be quantified using the Hall-Petch relationship, which is seen below
in Equation 5 [61].
𝜎𝑦 = 𝜎0 + 𝑘𝐻𝑑
−1/2
(5)
Where 𝜎𝑦 is the total yield strength, 𝜎0 is the yield strength of a single crystal, 𝑘𝐻 is the Hall-Petch
strengthening constant, and d is the grain or feature size. Figure 7 depicts a graphical representation
of the increase in yield strength at decreasing features sizes for a nanoscale material. Additionally,
it can be seen in Figure 7, that beyond a critical a grain size (around 10nm), the Hall-Petch
relationship can breakdown and fail to represent the change in strength. At this point, yield strength
can begin to plateau or decrease as grain boundary sliding or other deformation mechanisms occur
[62].

19
Figure 7: Hall-Petch strengthening relationship for a nanoscale material, where d is the feature or
grain size, 𝜎0 is the yield strength of a single crystal and 𝑘𝐻 is the Hall-Petch strengthening constant
(which in this case equals the slope). The breakdown of the Hall-Petch relationship at small ( <
10nm) feature sizes is indicated by the dashed line.
In homogeneous nanostructured materials, increases in yield strength can lead to decreased
ductility and more brittle failure as plastic deformation is limited [63]. To prevent loss of ductility,
current research is investigating heterogeneous nanostructures, as they can combine multiple
microstructural components to overcome strength ductility trade-offs [64-66]. Sputtering can
synthesize a range of heterogeneous nanostructures, one of the most notable being nanotwins,
which have displayed increased strength, ductility, and thermal stability [67, 68]. The next section
will discuss nanotwin synthesis and the effects of nanotwins on sputtered film mechanical
properties.

20
2.3 Nanotwins
Nanotwins are controllable interfaces that have unique material properties derived from
twin boundaries with nanoscale spacing. They are readily found in FCC metals and other alloys
with low stacking fault energies (SFE’s) [66]. In sputtered films, these nanostructures form in
columnar grains and expand the microstructural complexity by increasing the number of interfaces
and the range of feature sizes. The nanoscale spacing can access Hall-Petch strengthening
mechanisms, while twin boundaries can alter material properties by lowering the grain boundary
energy [66]. To understand the total effects of nanotwins on material characteristics, the following
section discusses twin boundaries, stacking faults, and nanotwin formation and mechanical
properties.
2.3.1 Twin Boundaries and Stacking Faults
Twin boundaries are interfaces between two crystals whose atomic arrangements are
reflected over a mirror plane (known as the twin plane) due to a reversal in the atomic stacking
sequence [69]. Depending on the location of the twin plane, the resulting twin boundaries can
either be coherent or incoherent; a coherent twin boundary occurs when the atoms at the boundary
fit into both grains, while incoherent twin boundaries form when the plane of reflection is out of
phase with the twinning plane. This is seen in Figure 8, which compares a coherent (8a) and
incoherent (8b) twin boundary [50]. The change in stacking sequence is highlighted in 8a, where
going from top to bottom the crystal stacking sequences changes from ABC to CBA. Figure 8c
demonstrates that coherent twins can have significantly lower energies than incoherent twins and
other high-angle grain boundaries, which can increase their thermal stability, mechanical stability
and plasticity [66].

21
Figure 8: Twinning schematic depicting (a) coherent twin boundaries, (b) incoherent twin
boundaries, and (c) twin boundary energy, 𝛾, as a function of boundary coherency, 𝜑. Adapted
from [50].
Stacking faults are planar defects that can initiate twin boundary formation by causing a
change in the stacking sequence. Although stacking faults can form in other crystallographic
systems, this work will focus on FCC metals and alloys. In FCC structures, the stacking sequence
follows the ABC pattern referenced above. However, at a stacking fault, an atomic plane is either
added, removed, or shifted causing the local formation of an HCP structure, which is a two-layer
stacking sequence (i.e. ABA, BCB, or ACA) [50, 70]. An example of a stacking fault is shown in
Figure 9, where 9a represents a perfect FCC stacking sequence and 9b shows the formation of an
HCP structure. Twin boundaries form at stacking faults if the change in atomic stacking sequence
continues instead of returning to the original order. The increased energy per area, measured in mJ

22
m2
, associated with the formation of a stacking fault is known as the stacking fault energy (SFE).
Lower stacking fault energies (typically below 100 mJ m-2
) raise the probability of stacking fault
and twin boundary formation [71, 72]. However, it should be noted that SFE’s are intrinsic material
properties that are difficult to measure and vary unpredictably with changing composition, which
limits their use for identification of materials that can form twin boundaries.
Figure 9: Illustration of stacking fault formation in an FCC metal, where (a) represents a perfect
FCC stacking sequence and (b) represents the formation of an HCP structure at a stacking fault.
2.3.2 Nanotwin Boundary Synthesis
Twin boundaries can be created via annealing, deformation, and deposition (known as
annealing twins, deformation twins, and growth twins, respectively). During the annealing process,
annealing twins can form both coherent and incoherent twin boundaries along the diameter and
edge of a grain [73]. One proposed model for annealing twin formation considers the nucleation
process, where newly forming close-packed grains have similar mechanisms and energy penalties
for forming both the forward and reversed stacking sequences [74]. However, there is not an
overarching theory to completely describe annealing twin formation. For example, another work
found that low temperature annealing enabled twin formation before recovery and recrystallization

23
[75]. Deformation twins are achieved via stress induced plane sliding. As a material undergoes a
local shear stress, planes can shift to absorb energy by forming a twin boundary [76, 77]. There
are two stages of deformation twin formation; the first is nucleation of a twin boundary at a
dislocation and the second is twin growth [76]. Generally, twin growth occurs with repeated slip
along adjacent planes in the crystal, as shown in Figure 10. It is seen that the crystal progresses
from a perfect FCC structure (Figure 10a) to an FCC with a stacking fault (Figure 10b), and finally
to a deformation twin boundary (Figure 10c).
Figure 10: Deformation twin nucleation and growth in an FCC crystal. (a) is a perfect FCC lattice,
(b) is the nucleation of a stacking fault due to stress induced planar slip, and (c) is the growth of
deformation twin boundary after consecutive slip of adjacent planes. The arrows highlight sheer
stress on each plane before slip occurs.
Growth twins are unique when compared to both annealing and deformation twins because
they can form during material synthesis. With respect to PVD techniques like magnetron
sputtering, growth twins occur when the stacking sequence is interrupted by a stacking fault during
deposition [78]. Figure 11 shows an example of growth nanotwin formation in the columnar grains
of a CuNi film synthesized via magnetron sputtering. Other key aspects of growth twin formation
are the thermodynamic links between twin formation, stacking fault energy, and deposition
conditions that can be leveraged to alter twinning probability and spacing. Zhang et. al. proposed

24
a thermodynamic model that compares the Gibb’s free energies during nucleation for a perfect
FCC grain and a twinned grain and used it to derive the critical radii for FCC crystal and twin
boundary formation, which are seen below in Equations 6 and 7 [79]. It should be noted that the
nucleating grains in this model are cylindrically shaped.
𝑟𝑝𝑒𝑟𝑓
∗ =
𝛾
∆𝐺𝑣
(6)
𝑟𝑡𝑤𝑖𝑛
∗ =
𝛾
∆𝐺𝑣 −
𝛾𝑡
ℎ
(7)
Where 𝑟𝑝𝑒𝑟𝑓
∗
and 𝑟𝑡𝑤𝑖𝑛
∗
are the critical radii, 𝛾 is the surface energy (not stacking fault energy
although it also uses 𝛾), ∆𝐺𝑣 is the bulk free energy, 𝛾𝑡
is the twin boundary energy (which is
proportional to the stacking fault energy), and h is the height of the nuclei. For gas to solid
transformations, the bulk free energy can be linked to vapor pressure and deposition flux, which
can be substituted into the critical radii formulas to achieve Equations 8 and 9 seen below [79].
∆𝐺𝑣 =
𝑘𝑇
Ω
ln (
𝑃𝑣
𝑃𝑠
)
𝐽 =
𝑃𝑣
√2𝜋𝑚𝑘𝑇
𝑟𝑝𝑒𝑟𝑓
∗ =
𝛾
(
𝑘𝑇
Ω
ln (
𝐽√2𝜋𝑚𝑘𝑇
𝑃𝑠
)
(8)
𝑟𝑡𝑤𝑖𝑛
∗ =
𝛾
(
𝑘𝑇
Ω
ln(
𝐽√2𝜋𝑚𝑘𝑇
𝑃𝑠
)−
𝛾𝑡
ℎ
(9)
Here, k is Boltzman’s constant, T is the temperature, Ω is the atomic volume, 𝑃𝑣 is the vapor
pressure during deposition, 𝑃𝑠
is the solid vapor pressure (in this case the vapor pressure of the
material used for the sputtering target), and J is the deposition flux. The probability of twin
formation is proportional to the difference in critical radii; in the work by Zhang et. al. a radii
difference of 5% or less was said to promote nanotwin formation [79]. From these equations it can

25
be seen that growth twin formation can be promoted by varying sputtering parameters such as the
deposition flux or temperature or by depositing materials with low SFE’s ( < 100 mJ m-2
). Taking
this a step further, the critical free energies at the critical radii for perfect and twinned nucleating
grains can be used to estimate nucleation rates. Comparing the nucleation rates, the twin spacing
can be predicted by multiplying the ratio of perfect planes and twinned planes with the interplanar
spacing. The equations for the nucleation rates (Equations 10 and 11), the ratio of nucleation rates
(Equation 12), and the estimated twin boundary spacing (Equation 13) are seen below [79].
𝐼𝑝𝑒𝑟𝑓 = 𝜔0𝑒
−∆𝐺𝑝𝑒𝑟𝑓
∗
𝑘𝑇 (10)
𝐼𝑡𝑤𝑖𝑛 = 𝜔0𝑒
−∆𝐺𝑡𝑤𝑖𝑛
∗
𝑘𝑇 (11)
𝐼𝑝𝑒𝑟𝑓
𝐼𝑡𝑤𝑖𝑛
= exp [
−∆𝐺𝑝𝑒𝑟𝑓
∗
𝑘𝑇
+
∆𝐺𝑡𝑤𝑖𝑛
∗
𝑘𝑇 ] = exp[
𝜋𝛾
2ℎ𝑦𝑡
𝑘𝑇∆𝐺𝑣
(ℎ∆𝐺𝑣−𝛾𝑡
)
] (12)
𝜆 = 𝑑(
𝐼𝑝𝑒𝑟𝑓
𝐼𝑡𝑤𝑖𝑛
) (13)
𝐼𝑝𝑒𝑟𝑓 and 𝐼𝑡𝑤𝑖𝑛 are the nucleation rates of perfect and twin nuclei, 𝜔0 is a pre-exponential factor,
∆𝐺𝑝𝑒𝑟𝑓
∗
and ∆𝐺𝑡𝑤𝑖𝑛
∗
are the critical free energies for perfect and twin nuclei formation, d is the
interplanar spacing, and 𝜆 is the predicted twin boundary spacing. Given the dependence on bulk
free energy (which as seen above implies a dependence on deposition flux) and temperature, this
once again demonstrates the ability of magnetron sputtering to tailor twin properties, in this case,
nanotwin spacing. This has been leveraged in the work by Velasco et. al., which developed initial
nanotwin formation maps to predict nanotwin spacing as a function of SFE and sputtering
deposition conditions [80]. These nanotwinning maps and relationships enable sputtering
unprecedented control of a nanostructure that can greatly enhance material mechanical properties.

26
Figure 11: Cross-sectional STEM image of growth nanotwin formation in a sputtered CuNi film,
where the film growth direction is indicated by the black arrow.
2.3.3 Nanotwin Mechanical Properties
Several studies have demonstrated the ability of nanotwinned materials to improve
mechanical properties including strength, fracture toughness, and fatigue resistance, without
sacrificing ductility [81-83]. However, other research has observed low ductility in NT materials
with smaller (~ 5 nm) TB spacings [84]. Work by Shen et. al. analyzed the ability of

27
electrodeposited nanotwinned copper to overcome the strength-ductility tradeoff seen in
nanostructured materials. This is highlighted in Figure 12, which compares the stress-strain curves
for coarse grained (CG), nanocrystalline (IGC), and nanotwinned copper (A-C) [81]. For the
nanotwinned samples, the average twin spacings were approximately 96nm (A), 30nm (B), and
15nm (C). The decrease in twin spacing led to an increase in both yield strength and ductility and
all nanotwinned samples displayed greater yield strength than their coarse grained and
nanocrystalline counterparts. The increase in yield strength can be attributed to Hall-Petch
strengthening, while the interactions between dislocations and coherent twin boundaries improved
ductility [81, 82].
Similarly, Singh et. al. demonstrated that nanotwins can be introduced to nanocrystalline
materials to improve fracture toughness and fatigue resistance [83]. Typically, ultrafine-grained
materials have limited plasticity, which decreases the material’s ability to resist crack propagation.
In this study, due to the plasticity offered by the coherent twin boundaries, nanotwinned copper
fracture toughness increased nearly 50% over that of ultrafine grained copper and, additionally,
higher nanotwin densities slowed cyclic crack propagation during fatigue testing [83]. The
enhanced mechanical properties of nanotwins thus make it desirable to develop nanotwinned
microstructures in novel compositions to access unexplored material property spaces.

28
Figure 12: Stress-strain curves for coarse grained (CG), nanocrystalline (ICG), and nanotwinned
copper (A,B, and C). The average nanotwin spacings for samples A, B and C are 96nm, 30nm,
and 15nm, respectively [81].
2.4 Combinatorial and High-Throughput Synthesis and Characterization
Expanding current nanotwin synthesis domains requires efficient exploration of the vast
composition space offered by the periodic table. Typically, compositional research endeavors are
directed by theory and empirical data; however, in many research fields, discoveries and
breakthroughs have occurred as the result of chance and serendipitous observations during detailed
experiments. When exploring increasingly complex material systems, ranging from binary to high
entropy alloys (HEA’s), the number of possible compositional combinations to investigate
increases exponentially. Adding in other factors such as the deposition and processing conditions,
it is clear that, in addition to theoretically and computationally directed investigations, a rapid and
efficient experimental approach is necessary to increase the chances for nanotwin discovery. These
considerations are the basis for combinatorial and high-throughput materials research, a
methodology that’s been developed to rapidly analyze large composition spaces to identify unique

29
material characteristics [47, 85]. In practice, this process is comprised of two experimental phases:
(1) the synthesis of combinatorial samples and (2) the high-throughput characterization of sample
properties, which will be discussed in detail in the following sections.
2.4.1 Combinatorial Synthesis
Combinatorial materials synthesis techniques seek to enable the simultaneous and rapid
production of tens, hundreds, or even thousands of samples. Typically, this is achieved by creating
a material with a compositional gradient and subsequently dividing it into arrays of
compositionally unique samples. In this work, these arrays of specimen are referred to as
“combinatorial samples”. Some of the most common combinatorial synthesis methods are
diffusion couples, additive manufacturing (AM), and film deposition [47, 85, 86]. These
techniques either leverage diffusion or inhomogeneity during processing to achieve compositional
gradients. Each synthesis method can investigate distinct compositional ranges, microstructures,
and material feature sizes. The following section discusses the applications, advantages, and
limitations of these combinatorial synthesis techniques.
2.4.1.1 Diffusion Couples
Interdiffusion between two or more materials is a clear and efficient process for creating
compositional gradients [87]. A diffusion couple (also known as a diffusion multiple) is created
by joining wedges of distinct materials together to form an interface, as can be seen in Figure 13a
[50]. By heating the diffusion couples to a high temperature, interdiffusion of substitutional atoms
across the interface occurs following Fick’s 1st law of diffusion, which is seen below in Equation
14.

30
𝐽𝐴 = −𝐷𝐴
𝛿𝐶𝐴
𝑑𝑥
, 𝐽𝐵 = −𝐷𝐵
𝛿𝐶𝐵
𝑑𝑥
(14)
Where 𝐽𝐴 and 𝐽𝐵 are the flux of atoms for elements A and B, 𝐷𝐴 and 𝐷𝐵 are the intrinsic diffusion
coefficients for elements A and B, 𝐶𝐴 and 𝐶𝐵 are the concentrations of elements A and B, and x is
the position along the diffusion couple [50]. Fick’s 1st law dictates the flux of atoms for each
element as they move from areas of high concentration to regions of low concentration. So, a
compositional gradient will form as materials A and B diffuse into each other. It is important to
note that since the intrinsic diffusion coefficients, 𝐷𝐴 and 𝐷𝐵, of two elements are likely not equal,
the flux’s 𝐽𝐴 and 𝐽𝐵 will be different magnitudes, resulting in a greater net flow of atoms in one
direction. To equate flux across the interface, there is also a flux of atomic vacancies 𝐽𝑉. The
interdiffusion process is shown in Figures 13b and 13c, where the resulting compositional gradient
and the flux of atoms and vacancies across the diffusion couple interface are depicted, respectively
[50].

31
Figure 13: Illustrations of (a) a diffusion couple comprised of materials A and B, (b) composition
change due to interdiffusion, and (c) the flux of atoms and vacancies along the diffusion couple.
Adapted from [50].

32
Due to the thermodynamics driving interdiffusion, combinatorial samples synthesized
using diffusion couples are advantageous for high-throughput investigations of bulk material
properties at or near equilibrium [88]. Studies have analyzed a large range of compositional
combinations such as binary, ternary, and high entropy alloys, examining composition-structureproperty relationships for applications including constructing phase maps, discovering shape
memory alloys, and determining mechanical properties [87-91]. Additionally, diffusion couples
have been employed to analyze composition-microstructure relationships and for electronic
materials research [92, 93]. However, despite the versatility of diffusion couples in combinatorial
studies, other techniques such as AM and film synthesis have been selected over diffusion couples
because they offer the ability to analyze processing-property relationships and nanoscale size
effects in high-throughput studies.
2.4.1.2 Additive Manufacturing
AM techniques are typically used to create complex 3D structures using a layer-by-layer
or direct writing approach [94]. However, in combinatorial studies, AM is used to synthesize
samples with piecewise or continuous change in composition and microstructure. Laser additive
manufacturing techniques (LAM), such as laser powder deposition (LPD), laser metal deposition
(LMD), laser engineered net shaping (LENS) or selective laser melting (SLM), are commonly
used AM processes for combinatorial experiments [94-97]. In these techniques, a laser is used to
iteratively sinter a metal powder layer-by-layer to create a sample; combinatorial samples are
formed by incrementally varying the composition of the powder blend to create a compositional
gradient. Although most studies focus on metal systems, sintering based AM techniques can also
be used to investigate ceramic materials [98]. A key advantage for combinatorial experiments that

33
leverage AM is the ability to analyze both intrinsic material properties, such as composition,
modulus, and stacking fault energy (SFE), and processing dependent properties like microstructure
and morphology [95, 96]. Controllable processing parameters include the laser power, beam
diameter, scan velocity, and hatch spacing; these properties are utilized in combinatorial
experiments to directly alter thermomechanical history, cooling rates, and microstructure during
synthesis [94, 96]. AM processes can also access micro- and nano-scale size effects, however, they
have mostly been employed in combinatorial studies to analyze bulk properties for a range of
metallic systems including steels, HEA’s, and bulk metallic glasses [89, 95, 96, 99, 100].
Although AM is a versatile and tailorable technique, it faces challenges during processing
that limit the current range of combinatorial studies. These include material compatibility and
defect formation during synthesis [101]. The material selection space for AM is restricted, as some
materials are lost to vaporization during the sintering or melting process [102]. Defects, including
porosity, surface roughness, cracks, distortion and microstructural inhomogeneity or anisotropy
can form in AM samples, which limit the reproducibility and accuracy of property measurements.
Post processing techniques, such as annealing and hot isostatic pressing (HIP), have been applied
to eliminate defects, but they can also remove the desired processed microstructures [101]. To
readily enable analysis of material size effects in any composition space, in addition to processingcomposition-property relationships, combinatorial experiments can use compositionally graded
films and coatings.
2.4.1.3 Compositionally Graded Sputtered Films
Compositionally graded films for combinatorial experiments can be synthesized using both
CVD and PVD techniques; however, given the focus on expanding sputtering domains this work

34
will only discuss combinatorial studies using PVD, primarily magnetron sputtering [103].
Developing combinatorial samples using combinatorial magnetron co-sputtering is advantageous
as it can investigate both bulk and nanoscale material properties, alter film growth with processing
conditions, and has a nearly unlimited material workspace [37, 47, 57]. There are two main
combinatorial sputtering techniques used to create compositionally graded films: (1) depositing
wedge-type multilayers to be used as diffusion couples and (2) co-sputtering from multiple sources
without substrate rotation [47, 103]. Wedge multilayers are synthesized using controllable shutters
to sequentially deposit material from one sputtering source at a time. Single element sputtering
targets can be spaced 180° or 120° apart to deposit binary or ternary alloys, respectively, and
HEA’s and other more compositionally complex systems can be created by replacing the single
element sputtering targets with binary or ternary alloys [47, 103, 104]. An example of a ternary
wedge-type structure is shown in Figure 14 [103]. Following deposition of the wedge-type layers,
a post annealing treatment is performed to cause rapid diffusion and form the compositional
gradient.
Figure 14: Sputtered wedge-type structure for a ternary alloy, where materials A, B, and C are
deposited from sputtering sources spaced 120° apart [103].

35
Since sputtering excels at forming uniformly mixed solid-solutions during deposition, cosputtering and other PVD co-deposition techniques are used to form compositionally graded films
without a post-annealing process [51, 103]. By depositing onto large substrates from multiple
sputtering sources without rotation, compositional gradients are created as the sputtering flux is
not uniform over the entire surface [103]. This is shown in Figure 15, where the composition
gradient transitions from blue to orange, due to the change in flux with respect to changing distance
from the two sputtering sources. Combinatorial experiments with films and coatings have been
utilized for a range of studies including phase stability and microstructure investigations, shape
memory alloys identification, optical and electronic property optimization, and compositionmechanical property mapping [47, 104-107]. The next section discusses how high-throughput
characterization techniques are employed to efficiently characterize these properties.

36
Figure 15: Co-sputtering schematic depicting the compositional gradient across a substrate
resulting from the variation in flux caused by distance from the sputtering sources. Each square on
the substrate represents a compositionally unique sample that is 5x5mm in size.
2.4.2 High-Throughput Characterization
Material property spaces can be explored as a function of composition by pairing
combinatorial synthesis techniques with high-throughput and automated characterization methods
to efficiently develop extensive material property databases, known as material libraries (ML’s).
Here, a high-throughput characterization technique is defined as a testing method that can analyze
hundreds or even thousands of samples in a single day. This work has utilized high-throughput
characterization to investigate phase, composition, morphology, microstructure, and the electronic
and mechanical properties of combinatorial samples using techniques such as X-ray diffraction
(XRD), scanning electron microscopy (SEM), and nanoindentation. The following sections
discuss these techniques and other relevant high-throughput methodologies used to analyze
microstructural, electronic, and mechanical properties.
2.4.2.1 Composition, Phase, Morphology, and Microstructure
The compositional and structural properties of combinatorial samples can be ascertained
using high-throughput SEM, energy dispersive X-ray spectroscopy (EDS), XRD, synchrotron
XRD, electron backscatter diffraction (EBSD), and computational image analysis [108, 109]. The
application of these characterization techniques in high-throughput studies will be discussed here,
while operational theory will be explained in the experimental section. Compositional,
microstructural, and morphological data can be rapidly collected using SEM. Material properties
are determined by rastering an electron beam over a combinatorial sample and measuring the
generated secondary electrons, X-rays, and backscattered electrons with detectors in the SEM

37
[108, 110]. The secondary and backscattered electrons are used to image the top surface
morphology and microstructure, while the emitted X-rays are analyzed via EDS to identify
composition [108, 110]. This process is repeated on each combinatorial sample and can be used to
compositionally analyze or image hundreds or even thousands of samples in a single day. Although
image acquisition is easy to extrapolate for high-throughput experimentation, classification of
microstructural and morphological phenomena such as defects, precipitation, phase constitution,
and grain size typically require human analysis. This can limit the size of combinatorial studies as
even basic characterization can result in large processing times over the course of hundreds of
samples [108, 109]. Computer image analysis techniques and machine learning algorithms have
been developed to help ameliorate these limitations. For example, EBSD, which analyzes the
scattering and diffraction of backscattered electrons, can automatically calculate grain sizes by
measuring the distance between grain boundaries [111]. Another technique, known as Computer
Vision (CV), utilizes convolutional neural networks (CNN’s) to process pixel based information
from images to detect feature descriptors [109]. This has been applied in research for phase and
defect identification, particle size analysis, and morphology description [108, 109, 112, 113]. An
example of SEM image particle classification is shown in Figure 16, where CV breaks down the
foreground and background of the image and then identifies each particle [109]. Other machine
learning approaches can use this microstructural information in conjunction with material library
databases to elucidate processing-microstructure-property relationships [47, 85, 109, 114].
Material phase, crystal structure, lattice parameters, grain size, and texture offer insights
into material properties ranging from deformation mechanisms and ductility to electronic and
optical properties and can be characterized using high-throughput XRD, synchrotron XRD, or
EBSD [108]. XRD characterizes these properties by iteratively scanning combinatorial samples

38
with collimated X-rays and measuring the diffracted intensities as a function of the beam angle
[115]. Synchrotron XRD functions in a similar manner, calculating material properties by probing
a material with an X-ray beam. However, synchrotron XRD has two important distinctions from
conventional XRD. The first is that the synchrotron beam has a much greater brilliance (beam
intensity) and the second is that the beam has a much smaller radius. These two factors greatly
increase the scan resolution, rate of data collection, and interaction depth that is probed during
characterization, making it an excellent option for high-throughput studies. In addition to the
properties that can be determined via conventional XRD, the increased resolution enables
synchrotron XRD to determine other properties like elastic strain and residual stress [116]. The
main drawbacks for synchrotron analysis are cost and instrument availability. EBSD determines
phase and crystal structure properties by analyzing the electron scattering and diffraction that
occurs when probing the sample with the electron beam during SEM. Performing EBSD scans
over large areas (typically on of the order of hundreds of microns to millimeters) allows EBSD to
characterize the change in structural properties as a function of composition for both bulk and film
combinatorial samples [111].

39
Figure 16: Computer Vision (CV) analysis of an SEM image of nanoparticles where (a) is the
original SEM image, (b) is the automatic identification of particles and background, (c) is object
detection, and (d) is the assignment of pixels to a given feature [109].
2.4.2.2 Electrical Properties
Combinatorial and high-throughput experiments have measured electronic material
properties including the dielectric constant, piezoelectric or ferroelectric response, and sheet
resistance, to gain insight into charge storage, polarizability, lattice structure, and electron mobility
[85, 117]. Dielectric thin film combinatorial experiments have determined dielectric constants by
using a scanning-tip microwave near-field microscope (STMNM) to non-destructively determine
the electrical impedance [118, 119]. For ferroelectric and piezoelectric property measurements,
hysteresis loops are obtained for each sample using quantitative piezoresponse force microscopy
(PFM) [85, 117]. In these studies, changes in dielectric, piezoelectric, or ferroelectric response

40
were linked and mapped with respect to composition. Sheet resistance, which corresponds to the
resistance or resistivity per area of a thin film or semiconductor, is determined using a four-point
probe. In this technique, a current is applied to the sample and resistance or resistivity is measured
using the voltage drop between the probe tips/electrodes [120]. With regards to structural
materials, these electronic properties have been used to investigate phase transformations and
change in microstructure, as piezeo/ferroelectric responses and resistivity change with respect to
crystal structure, grain size, and morphology [121-123]. To more directly analyze change in
structural properties, combinatorial experiments also employ high-throughput mechanical testing.
2.4.2.3 Mechanical Properties
Identifying composition-processing-property relationships using high-throughput
mechanical testing is integral to the development of advanced structural materials. A number of
micro-mechanical testing techniques have been adapted for high-throughput studies including
wafer curvature measurements, micro-tensile testing, micro-pillar compression, and
nanoindentation [90, 124-127]. Wafer curvature tests are used in coating combinatorial
experiments to determine residual stresses resulting from adhesion between the film and substrate
[125]. These experiments leverage Stoney’s equation, which is highlighted below in Equation 15,
to calculate the residual stress using the difference in curvature between the film and substrate.
𝜎𝑆𝑡𝑜𝑛𝑒𝑦 =
−𝐸𝑠𝑡𝑠
2
6𝑡𝑐
(1−𝑣𝑠
)𝑅
(15)
Where 𝐸𝑠
is the elastic modulus, 𝑡𝑠
is the substrate thickness, 𝑡𝑐
is the coating thickness, 𝑣𝑠
is the
Poisson ratio of the substrate, and 𝑅 is the radius of curvature [128]. Typically, the radius of
curvature and residual stress can be determined via profilometry, a technique that uses a physical
stylus to measure the surface of a film or substrate. However, this technique must be applied to

41
each individual film and is therefore low-throughput. To enable high-throughput analysis of
residual stress, coatings are deposited onto an array of prefabricated cantilever beams and the
radius of curvature is taken from the coated beam using optical lasers/position sensors [125, 126,
129]. By iteratively measuring radius of curvature with the optical laser, this technique can rapidly
link composition with change in residual stress.
Micro-tensile testing is used to apply uniform strains over an entire specimen to produce
data that offers insight into the elastic and plastic properties of a sample, including the yield
strength, modulus, and hardening behavior [124]. Two methods that are used to quickly create
dog-bone structures for tensile testing out of bulk or deposited coatings are femtosecond laser
machining and photolithography masking [130, 131]. Femtosecond lasers enable high material
removal rates compared to other techniques like focus ion beam (FIB) milling and can produce
micro-tensile specimens out of bulk samples without causing significant damage to the
microstructure. This technique can be applied to most materials including metals, ceramics and
polymers to create dog-bone structures on the order of 10’s to 100’s of microns [130, 132]. For
sputtering and other combinatorial film synthesis process, photolithography is commonly used to
create micro-tensile testing specimens. Figure 17 highlights an example process where an array of
dog-bones is created via photolithography and subsequently coated via magnetron sputtering
before micro-tensile testing [131]. This technique can be applied to virtually any coating material
system and is used with both thick and thin films [130, 133].

42
Figure 17: Dog-bone synthesis process for high-throughput micro-tensile testing of deposited
films. In this example (a) the silicon wafer is thermally oxidized and (b and c) spin coated with a
lift-off system and photoresist that is (d) shaped into a dog-bone via photolithography. (e) The
excess resist is then removed and (f) the structures are coated using magnetron sputtering. Finally,
(g) the dog-bone patterned film is removed from the resist by being immersed in acetone and (h)
the specimen is micro-tensile tested. Adapted from [131].
Nanoindentation is another high-throughput technique to measure the mechanical
properties of both bulk materials and nanoscale thin films. This technique presses a hard tip into
the sample with a controlled loading rate, max load, and/or indentation depth to measure properties
including the hardness, modulus, and yield strength [124, 134]. These properties are calculated
using the indenter shape (typically a three-sided Berkovich tip with a radius ranging from 50-
150nm), the projected indent area, the applied load, and the slopes of the loading and unloading
segments from the load-displacement curve [134]. The procedure of calculating specific
mechanical values will be discussed further in the experimental section. Sample preparation for
nanoindentation is less time consuming and difficult as compared to other mechanical testing

43
processes and indentation rates can be as fast as 6 indents/second [135]. By iteratively performing
an array of indents on gradient samples, mechanical property maps linked to composition can be
generated. Additionally, fracture toughness and deformation behavior can be determined by
combining nanoindentation with FIB milling to perform arrays of micro-pillar compression tests
[124, 127, 136]. It should be noted that fracture toughness and elastic modulus can also be
determined by performing nanoindentation on the cantilever beams used for residual stress
characterization [126]. Similar to SEM and EBSD techniques, nanoindentation can be used to
“image” the sample surface by decreasing the indentation spacing and analyzing variations in
mechanical properties [135]. Representative hardness and modulus maps obtained using
nanoindentation arrays on a TA15 alloy are shown in Figure 18 and compared to a back scattered
electron (BSE) image; it can be seen that nanoindentation can accurately detect phase and
compositional change [137]. These high-throughput nanoindentation and micro-tensile
mechanical tests are important as analysis of mechanical properties gives insight into the effects
of composition, phase, morphology, and microstructural features such as nanotwinning on the
resulting properties of the material.

44
Figure 18: Top-surface characterization of TA15 alloy using (a) back scattered electron imaging
and (b and d) high-speed nanoindentation mapping of hardness and modulus. (c) Shows change in
mechanical properties over one line of indents [137].

45
2.5 Summary
Film synthesis is a multidisciplinary field with both industrial and research applications.
Magnetron sputtering offers many key advantages including a wide compositional workspace,
control over microstructure and morphology, the ability to alter deposition geometry, and access
to combinatorial experimentation. This versatility enables sputtering to effectively discover
materials with novel properties, synthesize advantageous nanostructures, and elucidate
composition-microstructure-property relationships. These relationships can then be leveraged to
tailor sputtered films for a given application. Increasing sputtering’s microstructural and
morphological synthesis spaces, for example, by achieving nanotwinning in novel compositions,
will further enable this technique to develop advanced materials. Thus, this research seeks to
expand the current capabilities of sputtering by studying both its processing and compositional
workspaces to enable it to access unexplored synthesis domains.

46
Chapter 3: Experimental Methods
The following sections provide an overview of the synthesis techniques used to create both
singular and combinatorial samples and the characterization methods employed to understand the
impact of deposition variables on sputtering plasma characteristics, microstructure, morphology,
and electronic or mechanical properties.
3.1 Synthesis Methods
3.1.1 Magnetron Sputtering
Magnetron sputtering is used to synthesize both thin and thick films and offers
compositional, morphological, and microstructural control [3, 37]. The sputtering techniques used
in this work (planar cathode, hollow cathode, and co-sputtering) are all variations of direct current
(DC) sputtering, leveraging the same core principles for deposition, which are depicted in Figure
3 from section 2.1.3. In this process, a gas (typically Argon) is flowed into a vacuum chamber and
ionized to create a plasma by electrons that are excited by an external DC bias. The negatively
charged plasma ions accelerate towards the positively charged cathode, where they bombard a
sacrificial “target” material and eject (known as sputtering) atoms and secondary electrons during
collision. This causes the formation of a vapor, which subsequently deposits material on a substrate
as it condenses [34, 35]. Controllable deposition parameters include the gas pressure, applied
power, deposition temperature, energy from ion bombardment, target composition, and the
working distance between the targets and the substrates, all of which can be varied to access the
different film growth regions discussed in Section 2.3 and shown in Figure 4 [37, 49]. Furthermore,
additional apparatus such as shutters, substrate masks, and biased or heated substrate stages can
be incorporated to increase control of film structure.

47
As was discussed in Section 2.1.3.2, sputtering can also influence film growth by changing
the apparatus geometry or by introducing additional sputtering sources to co-sputter. Figure 19
highlights the two cathode geometries used in this work, which are the hollow and planar cathode.
The hollow cathode (Figure 19a) has a 360° degree line of sight, while the planar cathode (Figure
19b) yields unidirectional deposition. Altering cathode geometry enables control of deposition
angle and plasma densities, which can both influence film growth during sputtering [43, 44]. Cosputtering, which was used to create combinatorial samples, is shown in section 2.4.1.3 in Figure
15. This technique can offer increased stoichiometric control by simultaneously sputtering from
two different target materials. Compositionally homogeneous or compositionally graded films can
be created using co-sputtering by depositing onto rotating or stationary substrates, respectively.
Figure 19: Schematic of the hollow cathode (a) and planar cathode (b) sputtering target
geometries, highlighting substrate positioning, varied line of sight, and particle deposition.

48
3.2 Characterization Techniques
3.2.1 Langmuir Probe Analysis
A cylindrical Langmuir probe was used to measure sputtering plasma characteristics
including ion and electron densities, fluxes, and electron temperature, which offer insight into the
role of charged particles on film growth during deposition. As can be seen in Figure 20, the probe
operates by biasing a Tungsten probe tip that is directly inserted into the plasma [138]. The term
“cylindrical” applies to the probe shape; a smaller probe tip its typically used to reduce plasma
perturbation during characterization. Additionally, probe size should be smaller than the mean free
paths of the charged particles to reduce the impact of neutral gas atoms on probe measurements
[139].
Figure 20: Langmuir probe schematic illustrating the insertion of a probe (labeled P) into the
plasma, where Vdis is the discharge voltage used to create the plasma, Idis is the discharge current
of the plasma and Vp and Ip are the probe bias and measured current, respectively. Vp and Ip are
the values used to generate I-V curves [138].
Depending on the bias, positive ions or negatively charged electrons will coalesce around
the probe tip to neutralize the local variation of charge within the plasma [139]. The probe
measures the current at each applied bias to generate a current-voltage, or I-V, curve, such as the

49
ones shown in Figure 21 [139]. Key features from the curves, such as the electron and ion
saturation currents or the slope of the transition region in between the two asymptotes are used to
calculate plasma characteristics. Equations 16 and 17, seen below, use these features to calculate
the electron temperature and ion density to understand the kinetic energy of electrons, measured
in electron volts (eV), and distribution of ions in a given volume [139, 140].
1
𝑇𝑒
=
𝛿ln (𝐼𝑝)
𝛿𝑉𝑝
(16)
𝑛𝑖 =
𝐼𝑖𝑠
0.6𝑒𝐴𝑝𝑟𝑜𝑏𝑒√
𝑘𝑇𝑒
𝑚𝑖
(17)
Where 𝑇𝑒
is the electron temperature, 𝐼𝑝 and 𝑉𝑝 are the probe current and bias, respectively, 𝑛𝑖
is
the ion density, 𝐼𝑖𝑠 is the ion saturation current, 𝐴𝑝𝑟𝑜𝑏𝑒 is the surface area of the probe tip, and 𝑚𝑖
is the ion mass. In this work, I-V curves were generated using a LabView program that
automatically recorded the current and voltage when using the probe to characterize the sputtering
plasma.
Figure 21: Representative current-voltage, or I-V curve, that can be generated by the Langmuir
probe, highlighting the asymptotic electron and ion saturation currents. Adapted from [139].

50
3.2.2 Four-Point Probe Analysis
The resistivity of sputtered films was measured using a four-point probe. Figure 22
illustrates the four-point probe contact with the sample surface and the uniform spacing between
the linearly arranged probe tips [141]. To determine sheet resistance and resistivity, the outer
probes apply a current (I) to the sample and the inner probes measure the resulting drop in potential
(V) [120]. The sheet resistance, Rsh, for a film can then be calculated using these values and
Equation 18, seen below [120].
𝑅𝑠ℎ =
𝜋
𝑙𝑛2
(
𝑉
𝐼
) (18)
Sheet resistance is measured in units of Ω/𝑠𝑞 (ohm per square), which are equivalent to the
standard ohm unit. The Ω/𝑠𝑞 notation is used to clarify that the observed value is not the bulk
resistance, since sheet resistance is dependent on the sample thickness. Equation 19 can be used to
determine the bulk resistivity, 𝜌, from the sheet resistance, where L, W, and t, are the length,
width, and thickness of the analyzed film [120].
𝜌 = 𝑅𝑠ℎ𝑡
𝑊
𝐿
(19)
Bulk resistivity is a material property that determines the increase in resistance over a give length
and is measured in units of ohm•m.

51
Figure 22: Four-point probe schematic depicting the probe placement and uniform spacing at the
sample surface, the current applied to the outer probes (I), and the voltage measurements at the
inner probes (V). Adapted from [141].
3.2.3 X-Ray Diffraction (XRD)
Crystal structure, phase, grain size, and orientation (texture) can be determined using
XRD, which is a nondestructive characterization technique that probes a material using X-rays.
X-rays represent a high energy sector of the electromagnetic wave spectrum and are generated
when electrons or other energized particles decelerate [115]. In the context of XRD, they are
created by bombarding a target material (typically Cu) with electrons that are emitted from an
electronically biased filament (typically W). When the incoming electrons hit the target material,
they can remove electrons from the inner “K” electron shell of the atom, which must then be
replaced by electrons from the outer rings. This process emits the high intensity characteristic Xrays that are used during XRD [115].

52
Figure 23: Normalized XRD spectra of planar cathode (dashed lines) and hollow cathode (solid
lines) sputtered CuAl films. Films were deposited at a power density of 1.2 W cm-2
and at gas
pressures ranging from 3 to 30 mTorr. The respective gas pressures are color coded. Adapted
from [142].
XRD spectra, such as the ones seen in Figure 23, measure the intensity of X-rays scattered
by the sample’s lattice planes as a function of 2𝜃, the angle between the diffracted and the
transmitted beam [142]. Scattering is governed by Bragg’s law (shown in Equation 20), which
determines if incident X-rays scattered by different atomic planes are in phase.
𝑛𝜆 = 2𝑑𝑠𝑖𝑛θ (20)
Here 𝜆 is the wavelength of the characteristic radiation (𝜆 =1.5406 Å for CuK𝛼 rays), d is the
interplanar spacing, θ is the angle between the incident X-ray and the atomic planes, and n is an

53
integer. From equation 21 it can be seen that varying atomic structure will shift the angle at which
scattered X-rays are in phase. However, some crystal structures cannot achieve in phase diffraction
at certain atomic planes because of the arrangement of atoms in their unit cells. This is predicted
using the structure factor equation, highlighted in Equation 21. The structure factor ascertains the
sum interaction of the X-ray with each atom in the unit cell.
𝐹ℎ𝑘𝑙 = ∑ 𝑓𝑛𝑒
𝑁 2𝜋𝑖(ℎ𝑢𝑛+𝑘𝑣𝑛+𝑙𝑤𝑛)
1
(21)
Where 𝐹ℎ𝑘𝑙 is the structure factor, N is the number of atoms in the unit cell, u,v, and w are the
coordinates of atoms in the unit cell, 𝑓𝑛 is the atomic scattering factor, and h,k, and l define the
atomic plane [115]. Thus, this enables XRD, to identify material structure by analyzing the atomic
planes that are present in the resulting spectra. Similarly, texturing is determined by comparing the
normalized intensities of the diffraction peaks; in a polycrystalline material, a higher diffraction
intensity on a given plane can indicate a greater fraction of grains with that plane orientation. In
Figure 23, it can be seen that deposition parameters like the sputtering geometry and gas pressure
can influence film texture.
3.2.4 Scanning Electron Microscopy (SEM)
SEM investigates material morphology, microstructure, composition, and texture by
rastering an electron beam over the sample surface. The electron beam is created using a field
emission gun (FEG) and focused on the sample by electromagnetic lenses and apertures [143]. As
shown in Figure 24, when the beam probes the sample, it can generate secondary electrons (SE),
backscattered electrons (BSE), and characteristic X-rays [144]. These byproducts are measured by
detectors in the SEM and this data is used to create the resulting images. Image resolution is

54
determined by the interaction volume, which is dependent on the accelerating voltage and current
of the electron beam.
Figure 24: Interaction of the SEM electron beam with sample, highlighting the interaction volume
and production of secondary electrons (SE), backscattered electrons (BSE), and X-rays. Adapted
from [144].
SE’s are produced due to inelastic collisions ejecting electrons from atoms at the top
surface of the sample. These electrons span a low energy range (approximately 1-50 eV) and are
used to image the top surface morphology and microstructure as seen in the example in Figure 25.
Image contrast is largely dependent on the sample topology, which influences the depth and angle
at which the incident beam interacts with the surface [35, 144].

55
Figure 25: SEM image of the top surface morphology of a CuAl sample.
BSE’s are electrons from the incident beam that are elastically scattered by the sample.
Since energy is conserved during the elastic collision, these electrons have energies on the same
order of magnitude as electrons produced by the FEG. Techniques such as electron back scatter
diffraction (EBSD) use these electrons to determine grain orientation and texture. As the BSE’s
travel out of the sample, they scatter elastically at atomic planes meeting the Bragg condition
referenced in section 3.2.3. This generates Kikuchi bands, such as the ones shown in Figure 26,
which can be collected and referenced spatially to determine the crystallographic orientation at
each point on the sample [145].

56
Figure 26: Kikuchi patterns derived from backscattered electrons, which are used to map grain
orientation [145].
Finally, similar to XRD, characteristic X-rays are produced during SEM as the incident
beam ejects electrons from inner electron shells of atoms in the sample. The energies of emitted
characteristic X-rays are material dependent and, as a result, they can be measured and used to
identify and map sample composition; this technique is known as energy dispersive X-ray
spectroscopy (EDS). EDS generates a spectra that analyzes the energy (x-axis) and intensity (yaxis) of characteristics X-rays to determine elemental compositions and atomic fractions in the
sample [35, 110].
3.2.5 Focused Ion Beam (PFIB)
FIB and plasma focused ion beam (PFIB) can be used to characterize sample morphology
and microstructure in a similar fashion to SEM’s. An ion beam is created via field emission and is
used to probe the sample, generating SE’s that can be collected by detectors to form an image. In
addition, the highly energetic ions in the beam can induce sputtering and thus can be used to mill

57
away material from the sample to expose cross-sections or features below the top surface [146].
The PFIB used in this work was a Thermo Scientific Helios G4 PFIB. Unlike traditional ion beams,
which utilize Ga+
ions, this system leverages a Xe plasma ion source, which has been shown to
enable higher milling rates with less surface damage [147]. Surface damage can be further reduced
by adding a gas injection system (GIS), which is used to deposit protective coatings (typically
made of C, Pt, or W) by decomposing the injected gas molecules with the ion beam [148].
Transmission electron microscopy (TEM) samples can be prepared by performing a “liftout” using a PFIB with a GIS and micromanipulators. This process is shown in Figure 27, where
a sample, known as a lamella, is milled from a specific region of the sample and welded to the
micromanipulators using the GIS, so that it can be transported and subsequently welded onto a
TEM grid [149]. Once on the grid, the lamella is further milled until it reaches a thickness that is
electron transparent (less than 100nm).

58
Figure 27: Overview of the PFIB lift-out process, depicting (a) the deposition of a protective cap,
(b) the milling process to create the lamella, (c) the attachment of the lamella to the
micromanipulator, and (d) the attachment of the lamella to the TEM grid [149].
3.2.5 Transmission Electron Microscopy (TEM)
In TEM, a high energy electron beam is transmitted through an ultra-thin sample (less than
100nm thick) to obtain images with nanometer or even angstrom resolution. A FEG is used to
generate the electrons, which are subsequently aligned into a parallel beam by condenser lenses,
before passing through the sample. Images are obtained by projecting the transmitted electrons
onto a phosphor screen or charged coupled device (CCD) camera using another series of
electromagnetic lenses [150]. Figure 28 shows an overview of the two standard operational modes

59
for TEM: the selected area electron diffraction (SAED) mode and the imaging mode [150]. SAED
inserts a specific aperture to observe the formation of diffraction patterns, which can be used to
determine lattice spacing, phase, and plane orientation. In a single crystal, the diffraction pattern
forms distinct spots, while in a polycrystalline material, the diffraction patterns tend to form rings
since the beam interacts with multiple grains that can have different orientations [150]. To resolve
different material features, TEM can employ a range of imaging modes, two of the most common
being bright field (BF) and dark field (DF) TEM. The BF imaging mode uses an aperture to form
an image from only the direct beam electrons, while the DF mode shifts the aperture to collect
only specific scattered electrons with the phosphor screen or CCD camera. Varying the imaging
mode can improve resolution of defects, specific plane orientations, diffraction contrast, or
elemental/Z contrast [150].

60
Figure 28: Schematic of TEM operation in (A) selected area electron diffraction mode and (B)
imaging mode [150].
Scanning transmission electron microscopy (STEM) is a specific operational mode for the
TEM that realigns the incident beam so that it converges at the sample surface. Images are obtained
by rastering the converged beam across the specimen, similar to SEM, and measuring the
intensities of the transmitted or scattered electrons using a bright field or annular dark field (ADF)
detector, respectively. These two modes offer different contrasts as can be seen in Figure 29 [150].

61
High-angle annular dark field (HAADF) STEM is a specific alignment that moves the ADF
detector to collect high-angle incoherent scattered electrons. It has high elemental or “Z” contrast
since the angle of electron scattering is directly affected by the size of the atomic nuclei [150].
Figure 29: Illustration of (a) the BF and ADF detector locations, (b) the intensities of the
transmitted and scattered electrons, and a comparison of the resulting (C) ADF STEM and (D)
BF STEM images [150].

62
3.3 Mechanical Testing
3.3.1 Nanoindentation
Nanoindentation is a high-throughput method to derive the mechanical properties of a
material. This technique indents a sample with a probe tip that is typically 50 – 150 nm in diameter.
As a result, it has higher spatial resolution than other hardness testing techniques and can be used
to investigate non-bulk materials [151]. Controllable parameters during nanoindentation include
the maximum load, the loading rate, the shape of the indenter tip, the indent spacing, and the size
of the indentation matrix. The indentation matrix determines the number of indents used to
estimate mechanical properties for a single sample; for example, a 10 x 10 matrix yields 100 unique
indents. When indenting films and other nanoscale samples, nanoindentation must avoid substrate
affects by limiting the indentation depth. Generally, this is dictated by the 10% rule, which states
that the indentation depth must be less than 10% the sample thickness [134].
In this work, nanoindentation with a Berkovich indenter tip was utilized to determine
material hardness and modulus. Figure 30 shows an example of the Berkovich indenter geometry
(Figure 30a), a schematic of the indent dimensions (Figure 30b), and a representative loaddisplacement curve (Figure 30c) [134]. The load-displacement curve is generated by pressing the
indenter into the sample at a given load. The initial loading curve has both elastic and plastic
deformation components, while the unloading curve is primarily linked to the elastic recovery of
the material. To determine the mechanical properties, the contact area, A, must be derived using
the known indenter geometry and the contact depth, which can be calculated using Equation 22
[151].
ℎ𝑐 = ℎ𝑚𝑎𝑥 − 𝜀
𝑃𝑚𝑎𝑥
𝑑𝑃
𝑑ℎ
(22)

63
Where hc is the contact depth measured from the apex of the indenter, hmax is the maximum
indentation depth, 𝜀 is a factor dependent on the indenter geometry, Pmax is the max load, and 𝑑𝑃
𝑑ℎ
is the slope of the unloading curve. Once the contact area is determined, the reduced modulus, the
actual modulus, and hardness can be calculated using Equations 23, 24, and 25 [151].
𝐸𝑟 =
𝑑𝑃
𝑑ℎ
√𝜋
2√𝐴
(23)
𝐻 =
𝑃𝑚𝑎𝑥
𝐴
(24)
1
𝐸𝑟
=
1−𝑣
2
𝐸
+
1−𝑣𝑖
2
𝐸𝑖
(25)
Here Er is the reduced modulus, H is the hardness, 𝑣 is the Poisson ratio of the sample, E is the
modulus of the sample, 𝑣𝑖
is the Poisson ratio of the indenter, and Ei is the modulus of the indenter.
Since the reduced modulus accounts for elastic deformation from both the sample and the probe
tip, Equation 25 is used to remove the contribution of the indenter to distinguish the modulus of
the sample.
Figure 30: Overview of (a) the Berkovich indenter geometry and (b) the indent dimensions,
showing (c) a representative load-displacement curve that can be obtained with a Berkovich
indenter. Here A is the projected contact area, 𝜃 is the face angle, and hc is the depth of contact
measured from the apex of the indenter. Adapted from [134].

64
3.4 Heat Treatments
3.4.1 Vacuum Furnace
Heat treatments were leveraged to investigate the thermal evolution of the co-sputtered
CHT samples. These treatments enable microstructural and morphological transformations, by
increasing the diffusivity or inducing nucleation and grain growth, as the material attempts to
reach its equilibrium structure. The total time needed to reach an equilibrium state varies,
depending on the nucleation and diffusion rates, and thus, specific annealing times can be
selected to study intermediate phase and microstructural transformations during this process. In
the case of sputtered films, annealing is a valuable tool to study microstructural mechanisms
influencing the thermal evolution, since sputtering often yields a metastable microstructure in the
as-sputtered state. Combinatorial arrays were annealed on a high temperature quartz substrate in
a GSL 1100X tube furnace at a vacuum pressure of 7 x 10-4 Pa for 3 hours at 400 °C. After the
heat treatment, the samples were cooled to room temperature in the vacuum furnace and then
removed and stored in a desiccator. Subsequently, phase and microstructure were characterized
using XRD, STEM, and EDX maps.

65
Chapter 4: Correlating sputtering target geometry and film growth
This work analyzes the effects of varying processing parameters, target geometry and
plasma characteristics on film growth, microstructure, and morphology and can be found in a paper
titled “Correlation between plasma characteristics, morphology, and microstructure of sputtered
CuAl films with varied target geometry”, that has been published in the journal Materials Research
Express, 10, 016402, (2023) DOI: 10.1088/2053-1591-acb31a. The effect of target geometry on
coating microstructure and morphology is correlated to changes in deposition conditions, plasma
characteristics, and film growth during planar and hollow cathode sputtering. The sputtering
plasma properties for the two target geometries were characterized via Langmuir probe analysis
as a function of power density and Ar pressure to determine the evolution of ion density for each
configuration. Films were then synthesized at the low (0.4 W cm-2
) and high (1.2 W cm-2
) power
densities and characterized using X-ray diffraction, scanning electron microscopy, and electron
backscatter diffraction to link changes in texturing, morphology, and microstructure with
variations in ion density and sputtering deposition conditions caused by target geometry. It was
observed that varying target geometry led to an over threefold increase in deposition rate,
homologous temperature, and ion density, which altered the morphology and texture of the film
without significant changes to the grain size.
4.1 Introduction
Magnetron sputtering is a highly tailorable physical vapor deposition (PVD) technique that
enables deposition of a vast array of materials, where planar magnetron sputtering is the most
commonly used configuration as it allows for unidirectional deposition and high deposition rates

66
[34, 37]. Direct current (DC) magnetron sputtering can effectively deposit conductive materials
ranging from single elements to complex alloys, while radio frequency (RF) or reactive sputtering
techniques, which utilize high-frequency voltages or flow reactive gases, respectively, are
typically used to deposit oxide films, dielectrics, or other non-conductive materials [3, 34, 152].
Modified sputtering configurations, including high-power impulse magnetron sputtering
(HIPIMS) and ionized physical vapor deposition (I-PVD), have demonstrated the ability to change
deposition pathways and film growth mechanisms by ionizing sputtered atoms and increasing
overall ion bombardment on the substrate surface during deposition [34, 43, 153-156]. However,
studies using non-planar sputtering target geometries to alter film microstructures have been
limited. For example, a foundational comparison is still lacking between planar cathode and
hollow cathode (also known as inverted cylindrical magnetron (ICM)) sputtering, even though
they are two commonly utilized sputtering configurations that provide very distinct angles of
incidence during deposition. Understanding the effects of target geometry on plasma
characteristics and subsequent film growth presents a novel route to manipulate film morphology
and microstructure which provides access to unexplored microstructures and material properties
in sputtered films.
Planar cathodes produce plasma densities that are orders of magnitude lower than HIPIMS
and I-PVD, yet, planar sputtering can achieve uniform coatings on surfaces within its line-of-sight
by leveraging unidirectional deposition via techniques such as collimation and long throw
sputtering [34, 157, 158]. Qualitative tools for planar film deposition, such as structure zone maps,
have been developed to predict changes in grain size, microstructure, and morphology, which
enable the technique to tailor deposited film mechanical, optical, thermal, and electronic properties
[37, 38, 45, 159, 160]. In contrast, due to its target geometry, hollow cathode sputtering is affected

67
by additional factors such as greater plasma densities and a 360 line-of-sight, which in turn affect
the microstructure and morphology beyond current structure zone maps representations [37, 38,
44, 45, 161, 162]. Typically, ICMs and their plasma discharges have been used in research for
off-axis sputtering, where the substrate is outside of the cathode, to coat non-planar and large area
substrates ranging from optical lenses to microelectronics [46, 163-167]. To a lesser extent, some
studies have investigated ICM sputtering within the cathode volume, analyzing coating wires and
adhesion [45, 161, 168-170]. Furthermore, a few studies have explored this target geometry as an
alternate deposition configuration for coating substrates with convoluted complex topologies [161,
171, 172].
Recent work on 3D nano- and micro-lattices has demonstrated that coating techniques can
increase the material workspace for additively manufactured materials and improve functionality
by accessing novel material property spaces [12]. In general, sputter coated nano- and microlattices have mostly employed planar target geometries and focused on the coated lattice material
properties; some studies have also investigated coating uniformity and observed thickness
gradients resulting from the unidirectional deposition of the planar cathode [173-176]. GarciaTaormina et. al. demonstrated using a hollow cathode that changing target geometry could
potentially improve coating coverage on micro-lattice structures by increasing line-of-sight during
deposition [171]. Thus, in order to enable further exploration of deposition within complex target
geometries, such as the hollow cathode, the correlations between cathode geometry and film
growth warrant further research.
In this study, a comprehensive comparison between planar and hollow cathode sputtering
using Cu-Al targets, focusing on the influence of target geometry on plasma characteristics,
deposition conditions, and film microstructure is presented. Plasma conditions were characterized

68
using a Langmuir Probe at applied power densities ranging from 0.4 to 1.2 W cm-2
and Argon
pressures ranging from 3 to 30 mTorr. Coatings were then synthesized at the high and low power
densities to highlight the effects of changing ion density, deposition rate, and homologous
temperature on the film microstructure and morphology. The films were subsequently
characterized via X-ray diffraction (XRD), scanning electron microscopy (SEM), and electron
backscatter diffraction (EBSD) to observe variations in texture, feature size, and cross-sectional
microstructure. Ultimately, this work links target geometry as a variable to change the plasma
characteristics and angle of incidence, demonstrating a novel route to manipulate and expand
coating morphology and microstructure.
4.2 Methods
Films were deposited on (100) 25 mm x 25 mm Si substrates in a vacuum chamber via DC
hollow and planar cathode sputtering using 99.99% Cu-2wt.%Al targets. The target composition
was selected since sputtering with Cu and Cu-Al alloys is well documented under planar conditions
and these material systems can be easily machined for the hollow cathode geometry [37, 161, 177].
The hollow cathode target (Kurt J. Lesker Company) had an inner radius of 4.76 cm, outer radius
of 5.08 cm, and length of 16.51 cm, while the planar target was a flat disk with a radius of 3.81
cm. In the hollow cathode configuration, the substrate was inserted into the volume of the hollow
target (see Figure 19) and, as such, the sputtering working distance in both the planar and hollow
cathode equaled the cylindrical target radius of 4.76 cm. Prior to sputtering coatings, the currentvoltage or I-V characteristics of the sputtering discharge plasmas at the substrate surface were
obtained using a cylindrical Langmuir probe (radius 0.254 mm) with a W wire tip (length 12.7
mm), using a similar probe construction as seen in the work done by Fang and Marcus [178].

69
Plasma measurements were taken at deposition power densities of 0.4 W cm-2
, 0.6 W cm-2
, 0.8 W
cm-2
, 1.0 W cm-2
, and 1.2 W cm-2 with Ar pressures of 3, 6, 12, and 30 mTorr in the hollow cathode
and Ar pressures of 3 and 6 mTorr in the planar cathode. Power density was defined as the applied
sputtering power divided by the surface area of the sputtering target; the planar sputtering target
had a surface area of 46 cm2
and the hollow cathode had a surface area of 494 cm2
. The probe tip
was positioned at the same working distances as the substrates with respect to the hollow and
planar cathode target surfaces and was biased from -80V to 80V. The W filament probe tip was
replaced before the coating thickness reached 2 microns, which was approximately 1% of the tip
diameter, to prevent significant material build up from affecting probe accuracy. The measured IV curves were graphically utilized to determine the ion saturation current, ion flux, and electron
temperature. Ion densities, ni, were calculated using the following equation, which is derived from
the Bohm sheath theory. Iis is the saturation current, Te is the electron temperature, Aprobe is the
known area of the probe tip, mi is the ion mass, and k is Boltzman’s constant [139, 140, 179].
𝑛𝑖 =
𝐼𝑖𝑠
0.6∙𝑒∙𝐴𝑝𝑟𝑜𝑏𝑒√
𝑘∙𝑇𝑒
𝑚𝑖
(26)
Following plasma characterization, 1 µm thick films were synthesized at the low and high
power densities at each of the varying gas pressures. A summary of the synthesis conditions and
local plasma properties for the deposited coatings can be seen in Table 1. Film thickness was
determined by the deposition rate multiplied by the total deposition time. To ensure consistent film
thicknesses at the different sputtering parameters and target geometries, deposition time was
adjusted so that each resulting film was 1 µm thick. The deposition rates were measured using an
Ambios XP-2 profilometer and the homologous temperature (T/Tm) was measured using k-type

70
thermocouples (Thermoelectric). The deposited films were characterized using XRD, SEM, and
EBSD to study the film texturing, crystallinity, and surface morphology. A Rigaku Ultima-IV
diffractometer was used to obtain XRD spectra with 2 scans performed on the range of 30° to
110° at a rate of 1°/min and step size of 0.08° using Cu Kα radiation. The top surface and crosssectional surface morphologies were imaged using a Nova NanoSEM 450 Field Emission SEM.
Images were taken using the immersion lens at an accelerating voltage of 10kV and spot size of
4.0. From the SEM micrographs, the average feature sizes were obtained by averaging 200
measurements from each sample using Image-J software. EBSD was performed on film crosssections using a Helios G4 PFIB UXe DualBeam FIB/SEM and data was analyzed using the
Oxford Instruments Aztec Crystal software. The cross-sections were polished using the Xe ion
beam and subsequently analyzed with EBSD at an accelerating voltage of 15kV and current of 3.2
nA to examine grain sizes and orientations. Average grain widths for each cross-section were
determined by measuring 50 grains per sample using the Image-J software.
4.3 Results and Discussion
4.3.1 Plasma analysis and film texturing
A fundamental understanding of the influence of target geometry on sputtered film
microstructure and morphology should consider the plasma conditions, deposition pathways, and
film growth mechanisms of sputtered particles. Therefore, Langmuir probe analysis was conducted
to analyze the evolution of the plasma’s ion density with respect to power density and pressure to
investigate differences in ion production and bombardment for planer and ICM geometries. Figure
31 presents measured ion densities recorded over a range of sputtering power densities and

71
pressures as well as corresponding XRD patterns taken at the low and high power densities (0.4
W cm-2
and 1.2 W cm-2
). Power density, which is the applied power divided by the surface area of
the sputtering target, was chosen as a unifying parameter between the two configurations to
normalize differences in target surface area. Figure 31a shows that the measured ion density is
dependent on both power density and Ar pressure, increasing with applied power (as electron
density and energy increases to more frequently ionize neutral gas atoms) and decreasing with
increasing gas pressures (due to higher particle collision rates reducing electron energy and overall
ionization). These observations agree with a previous study conducted by Metwaly and Elbashar
that characterized a hollow cathode plasma and observed a decrease in electron temperature and
ion density at higher gas pressures as well as an increase in ion density with increasing current
densities [180]. Ion density has also been linked to pressure via Paschen’s Law, which explains
that the necessary potential to ionize a gas atom is directly proportional to the pressure and the
spacing of electrodes; thus, a greater potential is needed to ionize gas at a higher pressure [181].
Over the measured range of power densities and Ar pressures, the hollow cathode exhibited greater
ion densities and a steeper ion density slope than the planar cathode, which indicates that ion
density increases at a faster rate in the ICM; these differences are due to the changing cathode
geometry extending electron mean free paths [44, 162].

72
Figure 31: (a) Plotted Langmuir probe measured ion densities plots for hollow and planar cathode
sputtering plasmas as a function of power density at varying Ar pressures. (b) X-ray diffraction
(XRD) patterns for the films sputtered at the low power density (0.4 W cm-2
), which are indicated
by the black oval on the left side of Figure 31a. (c) XRD patterns for the films sputtered at the high
power density (1.2 W cm-2
), which are indicated by the black oval on the right side of Figure 31a.
Hollow cathode samples and measurements are represented by a solid line, the planar cathode
samples and measurements have a dashed line. Respective gas pressures are color coded and the
main XRD intensity peaks are labeled.
Given the correlations between power density, argon pressure, and ion density identified
by Langmuir probe measurements, samples were sputtered at each pressure at the low and high
power densities, as indicated by the circled regions in Figure 31a, and analyzed using XRD to
show changes in film texture. The corresponding XRD patterns for the planar and hollow cathode
sputtered films are depicted in Figures 31b (0.4 W cm-2
) and 31c (1.2 W cm-2
), where the sputtering
pressures are represented by the color of the XRD patterns. At both power densities, the planar
films have strong (111) texturing, while the ICM sputtered films display primarily a (111) texturing
with additional peaks. Table 1 provides a summary of the sputtering parameters and plasma

73
conditions for each film. From this table, it is seen at the same power density and pressure that
hollow cathode sputtering yields greater deposition rates, homologous temperatures, and ion
densities than the planar configuration. The presence of non (111) planes in the ICM sputtered
samples is likely due to the wider range of angles of incidence and increased sputtering rates
decreasing particle mobility and altering film growth directions during deposition, which can limit
the formation of the lowest energy (111) planes. Additionally, it is seen in the hollow cathode
samples that higher deposition rates in conjunction with lower ion densities may have further
increased the random texturing, yielding greater amounts of higher energy (non (111)) planes. For
example, in Figure 31c, the high-power density ICM sample sputtered at 30 mTorr had the largest
(220) peak. Altogether, Langmuir probe analysis highlighted that transitioning from a planar to a
hollow cathode target geometry alters the plasma ion density and the deposition environment
during sputtering, while XRD showed changes in texture due to varying particle mobility during
film growth.
Table 1: Summary of the hollow and planar cathode sputtering parameters and measured
deposition conditions.
Sample Power Density
(W cm
-2
)
Argon Pressure
(mTorr)
Sputtering Rate
(nm s
-1
)
Homologous Temperature
(T/Tm
)
Ion Density
(10
12
cm
-3
)
HC 1 0.4 3.00 1.10 0.10 0.70 ± 0.21
HC 2 0.4 6.00 0.77 0.13 0.70 ± 0.30
HC 3 0.4 12.00 0.86 0.15 0.61 ± 0.01
HC 4 0.4 30.00 1.02 0.12 0.53 ± 0.14
PLNR 1 0.4 3.00 0.29 0.04 0.40 ± 0.01
PLNR 2 0.4 6.00 0.28 0.05 0.29 ± 0.03
HC 5 1.2 3.00 2.25 0.20 2.02 ± 0.34
HC 6 1.2 6.00 2.12 0.23 2.31 ± 0.55
HC 7 1.2 12.00 2.61 0.23 1.39 ± 0.22
HC 8 1.2 30.00 3.18 0.22 1.11 ± 0.11
PLNR 3 1.2 3.00 0.82 0.06 0.66 ± 0.04
PLNR 4 1.2 6.00 0.72 0.07 0.43 ± 0.04

74
4.3.2 Surface morphology characterization
Figures 32 and 33 link deposition conditions to film microstructure and morphology in the
low and high-power density samples, respectively. The as-sputtered top surface morphologies and
measured feature sizes for the hollow and planar cathode samples sputtered over the range of Ar
pressures at the low power density (0.4 W cm-2
) are depicted in Figure 32. Comparing films
deposited by the two target geometries at the same Ar pressure (see Figures 32a vs. 32e and 32b
vs. 32f), the hollow cathode tends to yield more voided and porous top surface film morphology
with significantly larger feature sizes. For the planar films, increasing the gas pressure from 3 to
6 mTorr did not significantly change film morphology, but did reduce feature size by decreasing
the energy of the deposited flux and thus overall adatom mobility. In the ICM sputtered films,
changes in deposition parameters, such as Ar pressure, lead to greater variation in film morphology
and feature size. Voids and porosity in hollow cathode films sputtered within the target volume
has been reported in literature and can be attributed to higher deposition rates and a wider range
of angles of incidence with the surface compared to the planar cathode [161, 162]. However, given
the increased ion density in the ICM, it is apparent that additional factors influence the change in
film morphology, which can be observed when comparing films deposited at different Ar
pressures. For example, voids around crystallites, such as those present in Figures 32a-c, which
were sputtered at 3, 6, and 12 mTorr respectively, originate from adatom mobility and film growth
induced by ion bombardment. Island growth occurs as adatoms coalesce and this in turn leads to
void formation due to adatom-depleted regions [49, 182, 183]. Figure 32d highlights the inverse
of this relationship, forming smaller top surface features and adatom-depleted regions compared
to Figures 32a-c, which can be attributed to the increase in Ar pressure to 30 mTorr limiting ion
bombardment and adatom mobility by lowering the ion density.

75
Figure 32: Scanning electron microscopy (SEM) micrographs of the as deposited top surface
morphologies at 0.4 W cm-2
power density for the hollow cathode (a-d) and planar cathode (e-f)
films sputtered at gas pressures between 3 to 30 mTorr. The hollow cathode images have a solid
border, the planar cathode images have a dashed border, and gas pressures are color coded. Feature
sizes for each film are shown to the right of the corresponding SEM image.
Figure 33 presents the as-sputtered top surface morphologies and measured feature sizes
for the hollow and planar cathode samples sputtered over the range of Ar pressures at the high
power density (1.2 W cm-2
). Table 1 shows that the corresponding deposition rates more than
doubled for both configurations and the ion density increased at a faster rate in the ICM than the
planar cathode. Additionally, homologous temperature increased in the hollow cathode, but
remained relatively constant for the planar geometry. The planar films (Figures 33e and 33f)

76
display a similar morphology to the low power density planar samples (Figures 32e and 32f) with
an increase in feature size that can be attributed to greater particle energy during deposition at the
high power density. In contrast, a distinct change in morphology is observed when comparing the
low and high power density ICM sputtered films (Figures 32a-d and Figures 33a-d, respectively).
At the high power density, greater deposition rates caused feature size to decrease compared to the
low power density samples because of growing islands competing for surface area. Given the wide
range of angles of incidence during ICM sputtering, it would be expected that higher deposition
rates would also yield a more voided morphology; however, it was observed that the high power
density films in Figures 33a-d achieved a denser top surface morphology than the low power
density samples in Figures 32a-d. This is due to the greater homologous temperature and ion
density increasing deposition energy and particle mobility during film formation. The samples
sputtered at 3, 6, and 12 mTorr (Figures 33a-c) highlight this change in film growth, as increased
adatom mobility and surface diffusion allow the films to achieve denser top surface morphologies
by preventing the formation of adatom depleted zones. It should be noted that unlike the 3, 6, and
12 mTorr samples, the film sputtered at 30 mTorr (Figure 33d) shows residual voids on the top
surface due to the greater sputtering pressure limiting densification by reducing ion bombardment
and adatom mobility. In general, as power density increases over this range, the dominant growth
mechanisms in the planar cathode remain constant, while, due to its target geometry, ICM
sputtering transitions from adatom mobility and diffusion limited film growth to growth with
enough particle mobility to enable the formation of denser coating morphologies. This
demonstrates that the hollow cathode has competing factors influencing film formation including
higher deposition rates, wider range of angles of incidence, growing homologous temperatures,
and greater change in ion density.

77
Figure 33: Scanning electron microscopy (SEM) micrographs of the as deposited top surface
morphologies at 1.2 W cm-2
power density for the hollow cathode (a-d) and planar cathode (e-f)
films sputtered at gas pressures between 3 to 30 mTorr. The hollow cathode images have a solid
border, the planar cathode images have a dashed border, and gas pressures are color coded. Feature
sizes for each film are shown to the right of the corresponding SEM image.
4.3.3 Orientation, grain size, and global morphology
From the previous section, distinct relationships between target geometry and coating
morphology were established, which can be further evaluated by studying the film’s
microstructure. Figure 34 presents EBSD grain orientation maps for hollow cathode (34a, 34b,
and 34e) and planar cathode (34c, 34d) films sputtered at both the low (0.4 W cm-2
) and high power
(1.2 W cm-2
) densities as a function of ion density, where increasing ion density is denoted by the

78
arrow. The hollow cathode cross-sections in Figures 34a, 34b, and 34e revealed a columnar
microstructure, grain widths of roughly 70-80 nm, and a wide range of grain orientations. Similar
columnar microstructures and grain sizes were observed in the planar cathode samples in Figures
34c and 34d; however, the planar samples displayed a higher degree of (111) grain orientations.
The vertical column (Figures 34a-c) is used to compare the cross-sectional microstructures of
samples sputtered at the same power density (1.2 W cm-2
) with increasing ion density and
homologous temperature, while the horizontal row (Figures 34c-e) is used to compare crosssectional microstructure for samples sputtered at similar ion densities at the low and high power
densities. Although changes in target geometry and sputtering conditions led to over a threefold
increase in ion density, homologous temperature, and deposition rate, all grain orientation maps
revealed similar Zone T microstructures and grain sizes for all samples. This could be due to
limited change in bulk diffusion over the range of particle energies and homologous temperatures
observed in this study, which in turn would prevent the formation of Zone 2/3 microstructures as
shown in the Anders structure zone map predictions [38]. Additionally, for the samples with
higher ion density, the effects of increased energy from ion bombardment could be limited by the
greater deposition rates as the adatoms lose energy when colliding with newly sputtered particles.
This is supported by previous works, which have shown that the ion-to-neutral atom ratios can
affect microstructure, texturing, and morphology [43, 184]. However, regardless of changes in
bulk diffusion, a columnar structure similar to planar sputtered films was not predicted for the ICM
sputtered samples due to the ICM’s wide range of angles of incidence, as previous work has shown
that altering the deposition angle during unidirectional PVD can change the film growth direction
[43]. The observed columnar microstructure potentially indicates that target geometries that
deposit material equally from all directions with respect to the substrate, such as the hollow

79
cathode, can achieve similar microstructures to perpendicular unidirectional deposition. Thus,
target geometry could alter top surface morphology by varying adatom mobility without changing
bulk diffusion and the resulting film microstructure.
Figure 34: Electron backscatter diffraction (EBSD) grain orientation maps for planar (dashed
border) and hollow cathode (solid borders) sputtered films. The vertical column (a-c) shows
samples sputtered at the same power density (1.2 W cm-2
, green border) with increasing ion
densities. The horizontal row (c-e) displays samples sputtered with similar ion densities at both
the low (0.4 W cm-2
, gray border) and high (1.2 W cm-2
) power densities. The inverse pole figure
(IPF) IPF triangle defining grain orientations is to the right of the EBSD scans.
To highlight the differences in the evolution of hollow cathode film morphology as
compared to the planar geometry, a graphical summary of the top surface (red border) and crosssectional (blue border) morphologies as a function of ion density and homologous temperature is
shown in Figure 35. The planar film highlighted in the expanded circle in Figure 35 demonstrates
that unidirectional deposition can achieve a dense fully formed top surface layer even at low

80
homologous temperatures and ion densities. This is possible as film coarsening can still occur at
low fluxes and limited angular distributions during deposition. In contrast, due to the more
complex target geometry, the hollow cathode sputtered films show a wide range of top surface and
cross-sectional morphologies that are clearly influenced by the sputtering plasma and deposition
environment. The change in target geometry causes a competition between the increased angles of
incidence, homologous temperature, and ion bombardment on the particle mobility during film
growth. An increase in either homologous temperature or ion density results in film densification
and a less voided top surface morphology, as diffusion and adatom mobility increase. Thus, in
addition to traditional relationships observed in planar magnetron sputtering that can be used to
alter film properties like grain size and texturing, more complex target geometries, like the hollow
cathode, can also be used to influence film morphology by altering ion bombardment and angle of
incidence during deposition. Although the plasma properties within the hollow cathode are linked
to its deposition parameters, which prevent arbitrarily varying ion and electron densities like other
I-PVD techniques, sputtering within the hollow cathode volume can begin to bridge traditional
and ion-assisted sputtering to further manipulate coating morphology and microstructure.

81
Figure 35: Graphical summary of planar and hollow cathode top-surface and cross-sectional
morphologies with respect to ion density (x-axis) and homologous temperature (T/Tm) (y-axis),
where coating thickness is shown in the z-axis. The as-sputtered top surface morphologies are
outlined in red and the cross-sectional morphologies are outlined in blue; all boxes are roughly
1.5 micron in length. Sample labels correspond to Table 1.
4.4 Conclusion
Sputtering target geometry was linked to plasma characteristics and coating morphology
in order to provide a fundamental understanding of the impact of cathode shape on film growth.
Langmuir probe analysis of the sputtering plasmas revealed that the hollow cathode yields greater
ion densities than the planar configuration at the same deposition conditions. For both target
geometries, ion density decreased at higher gas pressures and increased at higher power densities.
In addition, it was shown that ion density changes at a greater rate with respect to power density
in the hollow cathode. For films synthesized at 0.4 W cm-2 and 1.2 W cm-2
, it can be seen that ICM
sputtered samples have larger feature sizes and greater fluctuation in film morphology and
microstructure with respect to the deposition conditions. This is caused by competing factors
influencing growth that are not apparent in planar deposition, such as different angles of incidence,

82
increased deposition rates, larger variation in homologous temperatures, and greater ion
bombardment. Due to changes in temperature, particle energy, and ion density, it is observed that
films sputtered inside the hollow cathode volume can transition from voided to dense top surface
morphologies as increased particle mobility aids uniform film growth and prevents adatom
depleted regions. Interestingly, cross-sectional EBSD characterization revealed that target
geometry can alter the top surface morphology but changes in microstructure and grain size were
limited. Overall, this study offers insight into the global impact of target geometry on film growth
in order to expand morphological and microstructural regimes, which can be used to guide
sputtering with non-planar systems, like the hollow cathode.

83
Chapter 5: Combinatorial and high-throughput nanotwin investigation
This section discusses rapid and efficient approaches to investigate sputtering composition
spaces and the corresponding work can be found in a paper titled “Combinatorial and highthroughput investigation of growth nanotwin formation”, that has been published in the journal
Acta Materialia, 270, 119839, (2024) DOI: 10.1016/j.actamat.2024.119839. In this work,
combinatorial and high-throughput experimental techniques are employed to examine the
synthesis of growth nanotwins in CuNi alloys. 338 unique CuNi samples were synthesized via cosputtering to create a material library encompassing composition, hardness, phase, and
crystallographic data. The material library data in conjunction with scanning transmission electron
microscopy was used to evaluate growth twinning over a wide compositional range (Cu – 6.8 to
58.8 at% Ni). A direct correlation between measured twin boundary spacings and the stacking fault
energies underscored limitations of the current growth twin model caused by an underestimation
of the free energy penalty for forming non-twinned grains. To address this, a refined model was
developed to accurately capture the variation in twin boundary spacing and formation across all
compositions. This model paves the way for high-throughput investigations into nanotwin
synthesis in various alloy systems.
5.1 Introduction
Nanoscale twin boundaries (TBs), referred to as nanotwins (NTs), are important
microstructural features that can be achieved via plastic deformation, annealing, and film synthesis
techniques [67, 74, 76, 78, 79, 82, 122, 185]. However, the latter (known as growth twinning) has
rarely been explored for NTs, despite offering a wide compositional workspace and the greatest

84
microstructural control. This is due to two main reasons: 1) the high experimental time costs to
evaluate the wide synthesis space, and 2) the finite number of known material stacking fault
energies (SFEs) [80, 186, 187]. SFEs are intrinsic material properties linked with growth TB
formation that are not well documented because they can vary unpredictably with changing
composition and are difficult and time intensive to measure (requiring detailed TEM
characterization of dislocations or XRD peak shifting analysis) [13, 70, 78, 80, 186-189].
Computational techniques including molecular statics (MS) and density functional theory (DFT)
have been employed as viable tools to estimate SFEs, but they still require experimental
verification [190-193]. One route to overcome the SFE bottleneck is by augmenting computational
approaches with combinatorial and high-throughput (CHT) experimental techniques to evaluate
growth twinning across large compositional spaces. Recently, CHT materials research has
emerged as a promising methodology to more efficiently discover and study materials, as it
examines entire composition domains instead of being restricted to discrete points [47, 85]. This
approach leverages compositional gradients and high-throughput characterization techniques to
generate large material property databases, known as material libraries, by analyzing hundreds or
even thousands of samples in a single experiment [47, 104, 106, 108]. Material libraries have been
used to elucidate trends and explain phenomena for a range of material characteristics such as
phase, crystal structure, and electronic and mechanical properties [85, 104, 117, 124]. For example,
Kube et al. used the data from their CHT study to identify new phase selection criteria with respect
to changing composition for high entropy alloys [104]. While the large datasets from CHT studies
have been successful in analyzing the aforementioned material characteristics, the impact and
applicability of CHT techniques can be broadened by expanding the use of material library data to
study microstructural and intrinsic material phenomena, such as growth NTs and SFEs.

85
In the case of growth NTs, studies have been limited by the small number of materials with
known SFEs, thus, CHT material libraries can provide large datasets that can be used to establish
a more direct understanding of the links between SFE, composition, and TB formation [80, 177,
194, 195]. Growth twinning typically occurs during non-equilibrium film synthesis processes and
it has been demonstrated that TB formation can be influenced by synthesis and material
parameters, including the deposition rate, temperature, and SFE [77, 78]. Given these
relationships, current research has utilized magnetron sputtering to investigate growth NTs
because it couples control of growth TB formation with a wide compositional workspace. This has
been used to examine NT formation in single-element and multi-element alloys such as Ag, Cu,
stainless steels, and HEAs [79, 185, 194, 196-199]. For the few materials with experimentally
characterized SFE’s, it has been observed that materials with lower SFEs (SFE ≤ 50 mJ m-2
) tend
to have greater rates of growth NT formation than higher SFE materials (SFE ≥ 120 mJ m-2
) [79,
80, 177, 200]. To capture this relationship, Zhang et al. developed a thermodynamic model that
utilizes known SFEs and deposition parameters to estimate the resulting growth TB spacings;
however, this model is restricted to order of magnitude approximations for low SFE materials due
to limited data, which hinders its ability to predict TB formation in other material systems [79,
200]. Therefore, growth twinning presents a model system to investigate and establish novel
approaches to analyze material library data, amplifying the impact of CHT research.
In this work, a high-throughput experimental approach to examine NT synthesis domains
is demonstrated using the CuNi alloy system. Combinatorial co-sputtering was employed to
synthesize 338 CuNi samples with compositions ranging from Cu – 6.8 at% Ni to Cu – 58.8 at%
Ni. Following synthesis, high-throughput characterization techniques, including X-ray diffraction
(XRD), scanning electron microscopy (SEM), and nanoindentation were used to generate material

86
libraries to analyze structural, morphological, and mechanical properties. Additionally, crosssectional microstructures of representative CuNi samples were examined via scanning
transmission electron microscopy (STEM) to elucidate relationships between composition and NT
formation. The observed microstructural trends were then compared with CuNi SFEs estimated
using MS calculations. Overall, the combination of CHT techniques and in-depth analysis in this
study provides new insights into growth NT formation leading to a revised growth twinning
thermodynamic model, ultimately demonstrating a novel path to analyze intrinsic and
microstructural material properties.
5.2 Experimental Methods
CuNi compositional libraries were synthesized via combinatorial co-sputtering, with two
sources used to deposit Cu (99.999%) and Ni (99.995%) from 5.08 cm diameter targets
(Plasmaterials) onto two stationary 10 cm Si (100) substrates at a base pressure of 5 x 10-4 mTorr,
sputtering pressure of 5 mTorr, a source to substrate working distance of 14cm, and cumulative
deposition rate of 1.2 nm s-1
. As shown in Figure 36, the material deposited onto each wafer was
divided using a mask into 169 5 x 5 mm alloy sections, resulting in 338 unique CuNi samples. The
average sample thickness was 2 𝜇m and thicknesses ranged from 1.1 𝜇m - 2.3 𝜇m depending on
location due to deposition on the stationary substrate. A total compositional range of Cu – 6.8 at%
Ni to Cu – 58.8 at% Ni was achieved by altering the sputtering power to the Cu and Ni targets
when depositing onto each substrate. Within a given square, composition was observed to vary by
up to +/- 1.7 at% Ni when measuring across a 4 mm distance. Characterization and analysis was
performed in the middle of each square (with a tolerance of 0.5 mm) to limit compositional
variation to less than ~0.2 at% Ni. For the first wafer, the Cu and Ni targets were sputtered at 500W

87
and 155W, respectively, while for the second wafer Cu was sputtered at 350W and Ni was
sputtered at 300W.
Figure 36: Schematic of the combinatorial synthesis process. (a) Illustration of the co-sputtering
technique, where two materials (Cu and Ni) are simultaneously deposited to form a compositional
gradient. (b) Image showing a combinatorial array of co-sputtered samples deposited on a 10 cm
wafer, where each 5x5mm square represents a unique CuNi sample.
High-throughput characterization techniques including energy dispersive X-ray
spectroscopy (EDX), SEM, XRD, and nanoindentation were performed to analyze the
composition, top surface morphology, crystal structure and phase, and mechanical properties of
each CuNi sample, respectively. Composition and top surface morphology were characterized

88
using the EDX and SEM capabilities of a Helios G4 PFIB UXe DualBeam FIB/SEM and Zeiss
Gemeni II SEM. SEM images were taken at a 10 kV accelerating voltage, 0.8 nA current, and 4.0
mm working distance in the Helios SEM and 5 kV accelerating voltage, 1.0 nA current, and 5.0
mm working distance in the Zeiss SEM. Both microscopes were equipped with the Oxford
Instruments Aztec software, which was used to acquire EDX spectra for each CuNi sample using
a 500,000 count limit, 5.5 mm working distance, and accelerating voltage and current of 20 kV,
0.8nA (Helios) or 15 kV, 1.0 nA (Zeiss). For analysis of the phase and crystal structure, XRD
spectra were collected with an Empyrean Panalytical X-ray diffractometer. To characterize
multiple samples in a single XRD session, an automated and programmable stage was used to
center the incident X-rays on each 5 x 5 mm sample (XY position) and a laser sensor was used to
identify and adjust the sample height (Z position). The incident X-rays were collimated to probe
an area of roughly 4 x 4 mm to ensure that only a single composition was analyzed during each
scan. Scans were performed using CuK𝛼 radiation with a 2𝜃 range from 30° - 110°, a step size of
0.026°, and rate of 0.3 seconds per step. The rate and step size were chosen to yield six or more
measurements above the full-width half maximum (FWHM) for each intensity peak. Hardness
values for 64 selected CuNi samples from each wafer were measured using a Hysitron
Triboindenter with a 100 nm Berkovich tip. A set of 12 indents, positioned 12.5 𝜇m apart was
performed in the center of each sample with a force-controlled, constant load rate, triangle load
function, and decreasing maximum load from 3000 𝜇N - 1000 𝜇N. The indentation parameters
were selected to ensure that the maximum indentation depth was 200nm or less than 10% the
average sample thickness. The equation to calculate hardness is provided below in Equation 27,
where H is the hardness, Pmax is the max load, and A is the contact area [151].
𝐻 =
𝑃𝑚𝑎𝑥
𝐴
(27)

89
To generate hardness heat maps, the hardness values from the selected samples were input into a
MATLAB code that used interpolation to estimate the hardness values for the remaining 105
samples.
Following high-throughput characterization, NT formation was analyzed in selected
samples via STEM with a FEI Talos F200C G2 TEM at a 200 KeV accelerating voltage. TEM
lamellae were prepared in the Helios G4 PFIB UXe DualBeam FIB/SEM using the plasma focused
ion beam (PFIB) lift-out technique [149]. ImageJ software was used to determine the average TB
spacing and percentage of NT grains observed in the sample cross-sections, where a NT grain was
defined as a grain with the majority of its area containing TBs spaced less than 100 nm apart. TB
spacing was calculated by measuring 100 TB per sample, while the percentage of NT grains was
determined by counting how many grains out of a set of 100 were nanotwinned. The
experimentally measured NT formation was compared with theoretically predicted TB spacings
calculated from SFEs obtained via MS calculations with the LAMMPS software. The MS SFE
calculations were performed over a compositional range from Cu – 0 at% Ni to Cu – 75 at% Ni.
First, pure face-centered cubic (FCC) Cu systems, oriented along the <112>, <111>, and <110>
crystallographic directions–corresponding to the x, y, and z directions– were constructed with a
simulation cell of dimensions 355 × 205 × 125 Å3
. This system comprised 80 × 80 × 40 lattice
cells along each direction. The dimensions were carefully chosen to minimize the influence of
local composition variations on the gliding plane and to lessen the free surface effects along the
<110> direction. Please see Figures S1(a) and (b) for the convergence tests of the gliding plane
area and thickness along the <110> direction. Alloy systems for each desired composition were
then created by randomly substituting certain amounts of Cu with Ni atoms, followed by energy
minimization to allow the system to reach equilibrium. Periodic boundary conditions were applied

90
in all three dimensions and a 10 Å vacuum separation layer was introduced at the top and bottom
to preclude periodic image interactions. Stacking faults were generated by displacing the upper
half of the simulation box along the <112> direction, which represents the most energetically
favored slip system for FCC metals. While the system was displaced, the atoms were allowed to
relax along the direction perpendicular to the stacking fault plane. Figures S1(c) and (d) provide a
visual representation of the local structure before and after the creation of the stacking fault. The
SFE, 𝛾𝑆𝐹, was calculated using Equation 28, where 𝐸𝑖 and 𝐸𝑓 are the system energies before and
after the creation of the stacking fault, and 𝐴 is the gliding plane area [193]. An Embedded Atom
Model (EAM) potential was used for Cu-Ni interactions, recognized for its satisfactory accuracy
in modeling the mechanical properties of alloy systems under various conditions, including
radiation damage, tensile and shock loading, and nanoindentation [201-204]. The LAMMPS
software was used to conduct all molecular statics calculations and OVITO was used in
visualization and local structural analysis [205, 206].
𝛾𝑆𝐹 =
𝐸𝑓−𝐸𝑖
𝐴
(28)
5.3 Results and Discussion
As described in the background, material library datasets can be leveraged to analyze
microstructural features and intrinsic material properties. Thus, to demonstrate a high-throughput
approach to investigate NT synthesis domains, the CuNi alloy system was selected as an ideal
candidate because Cu is a low SFE material (~ 40 mJ m-2
), Ni is a high SFE material (~120 mJ m2
), and adding Ni solute to Cu increases the SFE, unlike many alloys where SFE can vary
unpredictably with composition [207, 208]. Additionally, Cu and Ni tend to form solid solutions,
so NT formation should not be influenced by secondary phases and/or intermetallics, which could

91
complicate analysis of twin nucleation [69, 209]. As a result, this material system can be studied
and used to create foundational approaches to link CHT libraries with NTs and other material
phenomena.
5.3.1 Analysis of Material Libraries
In order to identify composition-property trends in the material libraries that could be
linked with NT formation, detailed property maps, shown in Figure 37, were created to summarize
the compositional, mechanical, and structural data collected via high-throughput EDX,
nanoindentation, and XRD. Figure 37a highlights the CuNi compositional gradients for the two
combinatorial wafers, spanning from Cu – 6.8 at% Ni to Cu – 58.8 at% Ni, where the areas in red
and blue indicate samples with greater Cu or Ni concentrations, respectively, and the yellow
borders identify selected samples that will be further discussed in Figure 37c. For quick
identification, the samples on each wafer were labeled numerically from left to right, with the first
wafer containing samples 1-169, which ranged from Cu – 6.8 at% Ni to Cu 35.5 at% Ni, while the
second wafer contained samples 170-338, which ranged from Cu – 12.5 at% Ni to Cu – 58.8 at%
Ni. The overlap in sample compositions from Cu – 12.5 at% Ni to Cu – 35.5 at% Ni was used to
check and verify material property measurements. It was observed that the Ni concentration of
neighboring samples varied by ± 1 to 5 at.% Ni depending on the distance from the Cu and Ni
sputtering targets, with the larger compositional changes observed in the middle of the wafer.
Complementing the compositional analysis, CuNi hardness values were then mapped for each
wafer, as seen in Figure 37b, and compared to the composition map in Figure 37a. From these
maps it can be seen that the CuNi alloy hardness increases with greater Ni content, likely due to
solid solution strengthening and Ni’s higher yield strength (𝜎𝑦𝑠,𝑁𝑖 = 138 MPa, 𝜎𝑦𝑠,𝐶𝑢 = 69 MPa)

92
[69, 210]. However, there were also variations in hardness that indicate other strengthening
mechanisms, such as Hall-Petch strengthening, are affecting the CuNi alloy mechanical properties
[211]. To further develop the CuNi material libraries, the structural properties of the CuNi alloys
were analyzed via SEM and XRD to determine how the top surface morphology, texture, or phase
could be affecting the resulting material properties. High-throughput SEM imaging revealed
minimal change in top surface morphology with respect to varying composition, as shown in
Figure S2 in the Supplementary Materials, which displays the top surface morphologies for six
compositionally unique representative CuNi samples. From the XRD analysis, it was determined
that all CuNi samples achieved a single-phase solid solution, with an FCC crystal structure and
strong (111) texturing. A small change in the ratio of (111) to (200) peak intensities was observed,
which was not dependent on composition or location within the array of combinatorial samples.
However, given the strong (111) texturing in all samples, this variation in (200) peak intensity is
not expected to significantly influence the CuNi material properties. An example of the XRD data
is highlighted in Figure 37c, which displays selected XRD diffractograms from the samples with
yellow borders mapped in Figure 37a; these samples were chosen as they represent the entire
composition range and are located at distinct positions on each wafer. Overall, the material library
data highlights that CuNi is an ideal system to investigate growth NT’s because there are no other
phases, morphologies, or textures present affecting the microstructure or material properties,
which enables a down selection from the entire compositional array to representative samples that
evaluate NT formation or the lack thereof.

93
Figure 37: Analysis of CuNi combinatorial samples. (a) Compositional maps obtained via EDX
for two sets of combinatorial wafers, with the first wafer (samples 1-169) containing samples with
compositions ranging from Cu – 6.8 at% Ni to Cu – 35.5 at% Ni and the second wafer (samples
170-338) containing samples with compositions ranging from Cu – 12.5 at% Ni to Cu – 58.8 at%
Ni. (b) Corresponding hardness heat maps for the two wafers, where hardness values were
measured using high-throughput nanoindentation. (c) Selected XRD patterns from the samples
with yellow borders in 2a.

94
5.3.2 Analysis of NT formation
To study growth NT formation as a function of composition, representative samples were
examined using STEM at approximately 10 at% Ni concentration intervals, with the resulting
analysis presented in Figures 38 and 39 and summarized in Table 2. Figure 38 shows the crosssectional STEM micrographs for the six selected CuNi samples, which had compositions of Cu –
7.1 at% Ni (38a), Cu – 17.4 at% Ni (38b), Cu – 27.7 at% Ni (38c), Cu – 33.8 at% Ni (38d), Cu –
48.1 at% Ni (38e) and Cu – 58.8 at% Ni (38f). From the STEM images, it can be seen that
increasing Ni content influences NT formation in grains, the columnar grain widths, and the overall
TB spacings. With respect to NT formation in grains, Figure 39 depicts examples of NT and nonNT grains (39a) and a plot of the decrease in the percentage of NT grains in each CuNi sample
from 84% NT (Cu – 7.1 at% Ni) to 52% NT (Cu – 58.8 at% Ni) as a function of increasing Ni
concentration (39b). A similar trend has been observed in previous studies, although at higher Ni
concentrations (~10 at% Ni or greater) NTs were not expected since greater Ni content increases
the SFE [79, 80, 177, 193, 200]. Table 2 shows a detailed summary of the measured percentages
of NT grains, columnar grain widths, TB spacings, and hardness values for the respective alloys,
and, from this table, it is observed that higher Ni content also yields a decrease in the columnar
grain width and an increase in the TB spacing. These varying microstructural features can directly
influence a wide range of material properties, such as the alloy’s resulting conductivity, thermal
stability, or hardness [78, 122, 185]. Specifically, in regard to the nanoindentation maps (Fig. 37b)
changes in NT formation and columnar grain width impact CuNi hardness by inversely affecting
Hall-Petch strengthening (see Table 2), where the hardness initially increases, due to the smaller
grain width, and then decreases due to the reduced NT formation. Further analysis can be found in

95
the Supplementary Materials section in Figure S3, noting that the overall changes in hardness can
be accounted for by the aforementioned mechanisms and compositional variations [210, 211].
From Table 2, it is noted that the percentage of NT grains and the TB spacing change as a
function of the Ni concentration. For TB spacing specifically, the minimal increase from 4.4 nm
(Cu – 7.1 at% Ni) to 16.9 nm (Cu – 58.8 at% Ni) largely deviates from the theoretical predictions
of the model developed by Zhang et al. where, due to the increase in SFE, the TB spacing would
be expected to increase orders of magnitude [79]. Thus, the presence of NTs in all samples
examined in this study (with concentrations as high as Cu – 58 at% Ni) highlights a disconnect in
the current model, limiting its ability to predict growth twinning over large composition and SFE
ranges.

96
Figure 38: Cross-sectional HAADF STEM micrographs for selected CuNi samples highlighting
the change in growth nanotwin formation as Ni concentration increases from (a) 7.1 at % Ni to (f)
58.8 at % Ni.

97
Figure 39: Nanotwin formation in CuNi alloys and quantitative assessment. (a) STEM micrograph
comparing nanotwinned (NT) and non-nanotwinned (Not NT) grains. A nanotwinned grain is
defined as a grain with the majority of its area containing twin boundaries spaced ≤ 100nm apart.
(b) Plotted comparison of the quantified percentage of NT grains relative to the total observed
grains, as a function of Ni concentration.

98
Table 2: Composition, nanotwin formation, grain width, and hardness data for selected CuNi
samples characterized via STEM, ImageJ, and nanoindentation.
Composition NT Percentage
(%)
Grain Width
(nm)
Twin Spacing
(nm)
Hardness
(GPa)
Cu - 7.1 at% Ni 84% 84.8 4.4 4.5
Cu - 17.4 at% Ni 76% 65.9 7.2 4.4
Cu - 27.7 at% Ni 71% 56.5 7.8 5.5
Cu - 33.8 at% Ni 71% 40.3 9.4 5.1
Cu - 48.1 at% Ni 70% 47.2 10.6 5.0
Cu - 58.8 at% Ni 52% 39.9 16.9 4.9
5.3.3 Revised Growth Twinning Model
Using the CuNi material library data presented herein, a revised thermodynamic model can
be developed to better understand growth twin nucleation during sputtering. First, NT formation
must be evaluated as a function of a generalized material parameter that influences TB spacing,
namely the SFE, which changes as a function of composition. The use of SFE to calculate growth
TB spacing is presented in the original thermodynamic model by Zhang et al., which is shown in
Equation 29 [79, 177].
𝜆 = [exp (
𝜋𝛾
2ℎ𝑦𝑡𝑤𝑖𝑛
𝑘𝑇Δ𝐺𝑣
(ℎΔ𝐺𝑣−𝛾𝑡𝑤𝑖𝑛)
)] ℎ (29)
Here, 𝜆 is the TB spacing, h is the height of the columnar grain (assumed to equal the (111)
interplanar spacing), ∆𝐺𝑣 is the bulk free energy per unit volume, k is Boltzman’s constant, T is
the temperature, 𝛾 is the surface energy, and 𝛾𝑡𝑤𝑖𝑛 is the twin boundary energy, which is
approximately equal to half the SFE (𝛾𝑡𝑤𝑖𝑛 ≈ 𝑆𝐹𝐸/2) [79, 177]. Surface energy, 𝛾, was assumed
to change with composition following the rule of mixtures. Thus, the surface energy will equal
(𝛾𝐶𝑢)x + (𝛾𝑁𝑖 )(1-x), where 𝛾𝐶𝑢and 𝛾𝑁𝑖 are the surface energies of pure Cu and Ni (1.185 J m-2
and
1.606 J m-2
respectively) and x is the atomic fraction of Cu in the alloy [212, 213]. The bulk free

99
energy per unit volume for a gas-solid transformation, ∆𝐺𝑣, is calculated using Equations 30 and
31, where Ω is the atomic volume, 𝑃𝑣 is the super saturated vapor pressure, 𝑃𝑠
is the vapor pressure
above the solid, m is the atomic mass of the deposited material, and J is the deposition flux [79].
Δ𝐺𝑣 =
𝑘𝑇
Ω
ln (
𝑃𝑣
𝑃𝑠
) (30)
𝐽 =
𝑃𝑣
√2𝜋𝑚𝑘𝑇
(31)
The growth twinning model determines TB spacing using the ratio of nucleation rates between
“perfect” and “twinned” columnar grains (𝐼𝑝𝑒𝑟𝑓𝑒𝑐𝑡 and 𝐼𝑡𝑤𝑖𝑛) highlighted below in Equation 32
[79].
ln (
𝐼𝑝𝑒𝑟𝑓𝑒𝑐𝑡
𝐼𝑡𝑤𝑖𝑛
) = −
∆𝐺𝑝𝑒𝑟𝑓𝑒𝑐𝑡
∗
𝑘𝑇
+
∆𝐺𝑡𝑤𝑖𝑛
∗
𝑘𝑇
(32)
∆𝐺𝑝𝑒𝑟𝑓𝑒𝑐𝑡
∗
and ∆𝐺𝑡𝑤𝑖𝑛
∗
are the critical free energies for the “perfect” and “twinned” nucleating
columnar grains, respectively. These variables are calculated using the change in total free energy
associated with forming each type of grain, which are shown in Equations 33 and 34, and are
related to the variables in Equation 32 through the critical radius [79].
∆𝐺𝑝𝑒𝑟𝑓𝑒𝑐𝑡 = 2𝜋𝑟ℎ𝛾 − 𝜋𝑟
2ℎ∆𝐺𝑣 (33)
∆𝐺𝑡𝑤𝑖𝑛 = 2𝜋𝑟ℎ𝛾 − 𝜋𝑟
2ℎ∆𝐺𝑣 + 𝜋𝑟
2𝛾𝑡𝑤𝑖𝑛 (34)
In the model, the shape of the columnar grain growth is assumed to be cylindrical, where Equation
34 includes an additional energy term for the TB at the top surface of the nucleating grain
(𝜋𝑟
2𝛾𝑡𝑤𝑖𝑛) [79]. This energy factor leads to the lower calculated rates for “twinned” grain
formation as compared to “perfect” grains. Specifically, it is assumed that there is no surface
energy penalty on the top surface of a “perfect” grain during nucleation, but this implicit
assumption leads to the model's exponential increase in predicted TB spacings. To account for this

100
energy contribution, a factor for the surface energy at the top of the “perfect” columnar grain, 𝛾𝑡𝑜𝑝,
must be added to Equation 33, which leads to Equation 35.
∆𝐺1 = 2𝜋𝑟ℎ𝛾 − 𝜋𝑟
2ℎ∆𝐺𝑣 + 𝜋𝑟
2𝛾𝑡𝑜𝑝 (35)
However, it is still more energetically favorable to form a “perfect” grain over a grain with a TB
defect, thus the following can be assumed:
𝛾𝑡𝑜𝑝 < 𝛾𝑡𝑤𝑖𝑛 (36)
and
𝑧 = 𝛾𝑡𝑤𝑖𝑛/𝛾𝑡𝑜𝑝 > 1 (37)
In Equation 37, because the TB energy is greater than the top surface energy of a “perfect” grain,
z must be greater than 1. 𝛾𝑡𝑜𝑝 is a distinct material value from 𝛾, since the latter is defined to be
the surface energy for the round side wall of the nucleating disc/cylinder that interacts with
adjacent grains, while the former is for the top surface, which interacts with the vacuum
environment during deposition. Using Equation 35 and the relationship between the TB and top
surface energies in Equation 37, an updated predictive model is derived below.
𝜆 = ℎ [exp (
𝜋𝛾
2ℎ𝑦𝑡𝑤𝑖𝑛
𝑘𝑇(Δ𝐺𝑣−
𝛾𝑡𝑜𝑝
ℎ
)(ℎΔ𝐺𝑣−𝛾𝑡𝑤𝑖𝑛)
)]
(1−1/𝑧)
(38)
A key difference between the updated model in Equation 38 and the original Zhang model is the
addition of the exponent term, (1 − 1/𝑧). Since z is greater than 1, the exponent term must have a
value between 0 and 1, which should reduce the rate at which TB spacing increases with respect
to increasing SFE. By identifying and linking composition, SFE, and growth twinning trends, the
model can be used to either validate experimentally determined and computationally predicted
SFEs or to directly estimate SFEs using measured TB spacings. To demonstrate the applicability
of the updated model, measured TB spacings were used in conjunction with MS estimated SFE’s

101
to determine the exponent term for the CuNi alloy system, which was calculated to be  0.25. See
supplementary materials for details regarding MS SFE calculations.
Figure 40 analyzes the fit of the updated and original models by comparing their predicted
TB spacings (calculated using the MS SFE values) with the experimentally observed values. TB
spacing is plotted as a function of the SFE in mJ m-2
, where the Zhang model is shown by the red
points, the updated model by the blue points, and the experimentally measured TB spacings by the
black points. Here it can be observed that the predicted TB spacings from the previous model
increase at an exponentially faster rate than the measured TB spacings from this study, while the
updated model is in agreement with the experimentally observed values. The improved fit of the
updated model supports its ability to directly predict and link growth TB formation with SFE and
corroborates the assumption that the top surface energy of a “perfect” columnar grain affects
growth twinning predictions. Thus, by having experimentally measured TB spacings and as few
as two SFE values, the updated model can be used to predict nanotwin formation over an entire
compositional space. The updated model was verified using previously published data on growth
twin formation from other material systems with low SFEs, including single elements, binary
alloys, and more complex engineering alloys, and a good fit was observed between the predicted
and experimental values [79, 177, 196]. This highlights an improved fundamental understanding
of growth twinning, which was obtained by using a CHT approach to globally evaluate growth NT
formation.

102
Figure 40: Comparison of measured and predicted twin boundary spacings as a function of
stacking fault energy (SFE). The black data points represent the measured twin spacings, while the
predicted values for the Zhang model and updated model are shown by the red and blue points,
respectively [79].
5.4 Conclusion
In this work, NT formation in CuNi alloys was investigated using high-throughput
experimental techniques in order to understand the fundamental relationships between growth
twinning and both intrinsic and extrinsic material properties. Over 300 unique CuNi samples were
synthesized via combinatorial sputtering. A comprehensive material library comprised of
composition, hardness, crystallographic, and phase data was compiled using high-throughput
characterization. This library highlighted the suitability of CuNi systems for probing growth NT
formation, evidenced by the consistent texture, phase, and morphology across samples. The
interrelationships among growth twinning, composition, and SFE, were examined by
characterizing representative samples using STEM and ImageJ. Contrary to the existing growth
twinning model, NT microstructures were observed in all compositional variations. Consequently,
a revised model was developed using the NT data from this study, which can predict NT formation
over entire composition spaces by accounting for all free energy contributions during nucleation.

103
In summary, this work delineates a novel approach for examining growth twinning and other
microstructural and intrinsic material phenomena using CHT techniques.

104
5.5 Supplementary Materials
Figure S1: Convergence tests for stacking fault energy (SFE) as a function of (a) gliding plane
area and (b) simulation cell thickness for a Cu – 50 at% Ni alloy. Illustrations of the atomic
structure of the alloy system (c) before and (d) after generation of the stacking fault. The transition
from FCC to HCP structure upon faulting is marked, highlighting the two layers of the HCP
structure indicative of stacking fault formation.

105
Figure S2: Top surface morphologies of selected as-sputtered CuNi alloys with varying Ni
contents including (a) Cu - 7.1 at% Ni, (b) Cu - 17.4 at% Ni, (c) Cu - 27.7 at% Ni, (d) Cu - 33.8
at% Ni, (e) Cu - 48.1 at% Ni, and (f) Cu - 58.8 at% Ni.
Hall-Petch strengthening calculations were made using the following equation where 𝜎0 is the flow
stress of a single crystal (𝜎0 = 20 𝑀𝑃𝑎 for Cu and Ni), 𝑘1 is a constant ( 𝑘1 = 0.14 𝑀𝑃𝑎 𝑚1/2
)
and d is the grain/feature size [214].
𝜎𝑦 = 𝜎0 + 𝑘1𝑑
−1/2
(39)
For columnar grain Hall-Petch strengthening, d equaled the measured grain widths, while for the
nanotwin Hall-Petch strengthening, d equaled the measure growth twin boundary spacing. An
order of magnitude estimation of solid solution strengthening was calculated with Fleischer’s
theory using the equation seen below [210]. Here 𝑘𝑠
is a constant (~138 MPa) and c is the Ni

106
concentration fraction. The solid solution strengthening coefficient, 𝑘𝑠
, was calculated using
known yield strengths for two CuNi alloys.
𝜎𝑦 = 𝑘𝑠𝑐
1/2
(40)
Fig S3. Influence of Ni composition on mechanical and microstructural properties of CuNi alloys.
(a) Comparison of the change in hardness (GPa) and the percentage of nanotwinned grains (% NT)
as a function of the Ni concentration. (b) Compositional dependence of solid solution (SS) and
Hall-Petch strengthening (NT Hall-Petch and Grain Width Hall-Petch) shown as a function of the
Ni concentration.

107
Chapter 6: Exploring twinning and phase formation in CuNiAl alloys
In this chapter, growth nanotwin formation was explored in a more complex ternary alloy
composition space and the corresponding study is currently in preparation for publication. A
combinatorial and high-throughput approach was leveraged to investigate nanotwin behavior in
the ternary CuNiAl alloy system. Combinatorial co-sputtering was used to synthesize 169 unique
CuNiAl alloys, which were characterized both as-sputtered and annealed to elucidate relationships
between composition, nanotwin formation, and phase evolution. Compositional effects on phase
formation were investigated using high-throughput X-ray diffraction, while scanning transmission
electron microscopy was used to identify nanotwin compositional boundaries and isolate the roles
of varied composition and nanotwin formation on microstructural evolution. Overall, it was
determined that Al content was the primary variable influencing thermal evolution in the
nanotwinned CuNiAl alloys, as it altered the thermodynamic driving forces by changing
composition and reducing the as-sputtered twin boundary spacing.
6.1 Introduction
The development and design of novel nanotwinned (NT) materials has garnered significant
interest as NT microstructures exhibit notable materials properties attributed to their nanoscale
features and increased thermal stability compared to their nanocrystalline or ultra-fined grained
counterparts [82, 122, 185, 215]. The enhanced properties are largely due to the presence of low
energy twin boundaries (TBs), which can both physically impede grain boundary mobility and
reduce the energy penalty that drives grain growth and recrystallization [198, 215-219]. For
example, in NT Cu when annealed at 800 °C for 1 hour, it was shown that the average TB spacing

108
only increased from 5 to 20 nm, while the same annealing treatment caused an order of magnitude
increase in the Cu columnar grain size [215, 220]. To date, most research has focused on
microstructural approaches to manipulate NT material properties including varying TB spacing
(λ), inducing nanoprecipitate formation, dispersing nanoparticles, and altering texture, while
investigations into compositional relationships have been minimal [67, 177, 198, 221-224]. This
is due to the fact that (1) only a finite number of materials have been shown to achieve NT
microstructures and (2) there is a high experimental cost of examining a wide synthesis space [47,
77, 80, 225]. As a result, NT research has largely been limited to investigating single element or
binary alloys [177, 226]. However, by implementing combinatorial and high-throughput
experimental (CHT) techniques, where compositional gradients can be used to synthesize large
arrays of samples in a single experiment, it is possible to efficiently investigate NT formation and
microstructural evolution over large compositional spaces [47, 104, 106, 226].
Recent work has demonstrated that combinatorial co-sputtering can leverage growth
twinning to explore NT synthesis domains [226]. This approach was utilized to study the CuNi
alloy system and overcome limitations in previous NT research by developing an updated
thermodynamic model that links composition, stacking fault energy (SFE), and TB formation [72,
78-80, 186, 226]. With this improved understanding, NT formation can now be investigated in
more complex compositional spaces, such as ternary and high entropy alloys, where the presence
of secondary phases and multiple alloying elements will influence NT material behavior and
thermal evolution [47, 69, 185, 199, 224]. For example, composition has been used in
nanocrystalline materials to improve thermal stability; Darling et al. demonstrated that Fe alloyed
with 1 at% Zr remained nanocrystalline when annealed for 1 hour at 1400 °C, while in pure Fe the
grain size coarsened to greater than 1 micron when annealed for 1 hour at 700 °C [227]. However,

109
since there has been limited research into NT formation in more complex alloy systems, the role
of individual elements on the development of NTs and secondary phases is not understood. Thus,
combinatorial sputtering can leverage heat treatments to comprehensively investigate complex NT
synthesis domains and elucidate the fundamental relationships between composition, NT
formation, and microstructural evolution.
In this work, a CHT experimental approach is employed to evaluate NT formation and
microstructural evolution in CuNiAl alloys. CuNiAl was selected as a model complex material
system, since the constituent elements are not fully miscible and NT formation has been studied
in the corresponding binary alloys, providing a solid foundation for understanding the role of each
alloying element [177, 226, 228]. Combinatorial arrays of samples were synthesized via cosputtering with alloy concentrations that ranged from 21.2 - 77.1 at% Cu, 13.4 - 51.2 at% Ni and
8.4 - 46.1 at% Al. Following synthesis, the CuNiAl alloys were analyzed both as-sputtered and
annealed using high-throughput X-ray diffraction (XRD) to examine changes in crystal structure
and identify trends in phase formation. Composition was linked with secondary phase formation
and NT microstructural evolution by evaluating representative as-sputtered and annealed samples
using scanning transmission electron microscopy (STEM). Ultimately, by leveraging a CHT
approach to study relationships between composition and NT behavior, this work demonstrates a
novel route to explore NT materials in complex compositional domains.
6.2 Experimental Methods
The arrays of combinatorial CuNiAl samples were synthesized using co-sputtering. Cu
(99.999%), Ni (99.995%), and Al (99.999%) were deposited from three 5.08 cm diameter
sputtering targets (Plasmaterials) onto two stationary 10 cm high-temperature quartz glass

110
substrates (McMaster-Carr) and a 10 cm Si (100) substrate at a base pressure of 1.0 x 10-3 mTorr,
Ar pressure of 5 mTorr, a 14 cm working distance, and a total deposition rate of 1.2 nm s-1
. All
substrates were sputtered under identical deposition conditions to yield three samples with the
same composition gradient. For each substrate, the compositional gradient was separated using a
mask that divided the sample into 169 unique 5 x 5 mm squares with thicknesses ranging from 400
nm to 1000 nm and an average thickness of 750 nm. The maximum concentration for each
sputtered element within the combinatorial array was 77.1 at% Cu, 51.2 at% Ni and 46.1 at% Al.
Following synthesis, analysis of each combinatorial sample was performed in the center of
each square to minimize compositional variation, similar to previous work [226]. Both arrays of
samples on the high-temperature quartz substrates were characterized using high-throughput XRD.
One quartz substrate was analyzed as-sputtered, while the other was first heat treated at 400 °C for
3 hours in a GSL1100X tube furnace (MTI Corporation) at 3.8 x 10-3 mTorr before
characterization. Automated XRD analysis of phase and crystal structure was performed using an
Malvern Panalytical Empyrean X-ray diffractometer. The incident X-rays were collimated to
probe an area of roughly 4 x 4 mm and the incident X-rays were centered on each square using the
programmable XY stage. Additionally, the Z height for each scan was corrected using a laser
sensor. All measurements were performed on a 2𝜃 range from 30° - 125° using CuK𝛼 radiation, a
step size of 0.026°, and scan duration of 0.3 seconds per step. The measuring conditions were
selected to resolve six or more points above the full-width half maximum for each peak in the
XRD spectra. The XRD analysis was used to determine phase volume fractions in each sample
following the direct comparison method detailed in the work by Cullity [229]. Composition
analysis was conducted on the array of combinatorial samples on the Si (100) substrate using the
energy dispersive X-ray spectroscopy (EDX) capabilities of a Helios G4 PFIB UXe DualBeam

111
FIB/SEM. EDX spectra were collected using the Oxford Instruments Aztec software, with a 5.5
mm working distance, 20 kV accelerating voltage, 0.8 nA current, and 500,000 count limit.
NT and phase formation was then investigated in selected samples from the as-sputtered
and heat treated high-temperature quartz substrates using the STEM and EDX capabilities of a FEI
Talos F200C G2 TEM and a Thermo Fisher Spectra 200 S/TEM. The Helios G4 PFIB UXe
DualBeam FIB/SEM was used to prepare TEM lamellae following the plasma focused ion beam
(PFIB) lift-out technique [149]. EDX maps were generated for a 500 x 500 nm area by collecting
at least 4 million counts per sample and using an 849 pm pixel size. TB spacing and
precipitate/phase formation in the imaged cross-sections was measured using ImageJ. CuNiAl SFE
values were calculated using the average TB spacings and the updated growth twinning model
found elsewhere [226].
6.3 Results and Discussion
6.3.1 Composition and NT Synthesis Domains
To date, NT research has primarily focused on single element or binary alloys, where TB
formation typically occurs in a single-phase material and varies at most as a function of one
alloying element. For example, in binary Cu alloys, varying Al and Ni content can promote or
inhibit TB formation, respectively, by changing the SFE [177, 226]. Although NT formation has
been observed in more compositionally complex systems like Inconel alloys, and even in a CHT
study on NiMoW [106], the role of each element on NT formation and microstructural evolution
has not been explored [230, 231]. Thus, the ternary CuNiAl system can be studied to elucidate
novel relationships between composition, secondary phase formation, and NT behavior. In order
to explore these relationships, NT formation must first be investigated and confirmed across the

112
synthesized composition domain. Figure 41 highlights CuNiAl high-throughput EDX material
library data and representative STEM images to develop detailed maps examining NT formation
in the composition space. Figure 41a depicts the composition gradient within the array of
combinatorial CuNiAl samples, where the map areas in green, blue, and red indicate greater Cu,
Ni, and Al content, respectively. Each square on the substrate was given a numeric label from 1 to
169 for quick identification and the composition across the array ranged from 21.2 - 77.1 at% Cu,
13.4 - 51.2 at% Ni and 8.4 - 46.1 at% Al. Figure 41b maps the total composition space that was
investigated by plotting the 169 squares on a ternary diagram. This compositional range was
selected based on prior NT studies of CuNi and CuAl, where similar variations in Ni and Al content
were linked with changes in SFE and NT spacing (ranging from 6 to 100 mJ m-2
and 1 to 35 nm)
[80, 177, 207, 226]. However, with more compositionally complex systems, like ternary CuNiAl
alloys, the compositional effects on SFE evolution and NT formation are unknown. Figures 41c-f
highlight cross-sectional STEM micrographs and corresponding EDX composition maps for four
selected compositions, taken from the circles with the black borders in Figure 41b. The
representative samples were chosen from the regions with the greatest Cu (Figure 41c), Ni (Figure
41d), and Al content (Figure 41e), as well as an intermediate composition (Figure 41f) to globally
evaluate NT formation across the combinatorial array. The STEM micrographs and EDX maps
revealed that each as-sputtered sample formed a solid solution and NT columnar microstructure,
with average TB spacings of 7.4 nm (Cu rich), 2.5 nm (Ni rich), 1.1 nm (Al rich), and 2.8 nm
(intermediate). Unlike conventional experiments, CHT studies can leverage specific compositions
to gain insights into entire composition spaces using the known relationships between samples
synthesized in the same combinatorial array. Therefore, the analysis of NT formation at the
CuNiAl compositional extremes in Figures 41c-e, can be used to predict NT formation in the

113
intermediate compositions. Since highly NT columnar microstructures were observed in all three
samples, this qualitatively indicates that highly NT microstructures are expected form across the
entire composition space. To quantitatively support this observation, the compositional boundaries
for NT formation, which are defined by CuNiAl alloy SFEs, can be determined with the updated
growth twinning model shown in Equation 41 [226].
𝜆 = ℎ [exp (
𝜋𝛾
2ℎ𝑦𝑡𝑤𝑖𝑛
𝑘𝑇(Δ𝐺𝑣−
𝛾𝑡𝑜𝑝
ℎ
)(ℎΔ𝐺𝑣−𝛾𝑡𝑤𝑖𝑛)
)]
(1−1/𝑧)
(41)
Here 𝜆 is the average TB spacing, h is the height of the columnar grain (assumed to equal the (111)
interplanar spacing), 𝛾 is the surface energy, 𝛾𝑡𝑤𝑖𝑛 is the twin boundary energy (𝛾𝑡𝑤𝑖𝑛 ≈ 𝑆𝐹𝐸/2),
𝛾𝑡𝑜𝑝 is the surface energy at the top surface of a non-twinned columnar grain, ∆𝐺𝑣 is the bulk free
energy per unit volume, k is Boltzman’s constant, T is the temperature, and 𝑧 = 𝛾𝑡𝑤𝑖𝑛/𝛾𝑡𝑜𝑝
(approximately 0.25) [226]. Using this equation, SFEs for the representative CuNiAl alloys can be
calculated with the measured TB spacings [226]. These values are summarized in Table 3, where
it is observed that SFE ranged from a low of 52 mJ m-2
(Al rich) to a high of 102 mJ m-2
(Cu rich).
In previous studies on single element, binary, and even more complex alloys including, Ag, Cu,
CuNi, CuAl, and 330 stainless steels, highly NT microstructures have been observed over a SFE
range from ~10 to 110 mJ m-2
, while in higher SFE materials NT formation is not generally
expected [79, 80, 185, 217, 226, 232]. Thus, in this study all CuNiAl compositions are expected
to be highly NT from the calculated SFEs values, demonstrated by the intermediate composition
shown in Figure 41f, which has both a SFE and TB spacing that falls within the range set by the
compositional extremes.

114
Figure 41: Analysis of CuNiAl composition and NT formation. (a) Composition map for the
CuNiAl combinatorial array obtained via EDX, containing samples with compositions ranging
from 21.2 - 77.1 at% Cu, 13.4 - 51.2 at% Ni and 8.4 - 46.1 at% Al. The areas in green, blue, and
red indicate greater Cu, Ni, or Al content, respectively. (b) CuNiAl compositions plotted on a
ternary diagram highlighting the occupied composition space. (c-f) Cross-sectional HAADF
STEM micrographs and EDX maps for as-sputtered CuNiAl alloys taken from the compositions
noted with black circles in Figure 41b. The samples were taken from Cu rich (41c), Ni rich (41d),
Al rich (41e), and intermediate (41f) composition regions.
In addition to identifying SFE and NT formation across large compositional spaces, the
combinatorial array of CuNiAl alloys can be used to directly investigate the effects of two or more
alloying elements on growth NT formation. For example, in the representative samples at the
compositional extremes in Figures 41c-e, the Al rich square exhibited the smallest average TB
spacing despite having the largest single element SFE (Al SFE = 166 mJ m-2
) [232]. Typically,

115
higher SFEs would inhibit growth twinning, however, the reduced TB spacing at higher Al
concentrations is actually consistent with composition-NT formation trends observed in the binary
CuAl alloy system [177]. Interestingly, the largest average TB spacing was observed in the Cu rich
alloy and not the Ni rich sample. This is contrary to expectations since pure Cu (SFE ~40 mJ m-2
)
has a lower SFE than pure Ni (SFE ~125 mJ m-2
) and greater Ni content increases the SFE and
average TB spacing in binary CuNi alloys [226, 232]. This deviation from the binary system
underscores the importance of investigating the effects of compositional complexity on NT
formation. In this case, the reduced TB spacing in the Ni rich sample could be attributed to the
greater Al content increasing NT formation, similar to the Al rich sample. Overall, this highlights
that CHT techniques can investigate compositional effects on NT behavior in more complex alloy
systems by identifying the compositional boundaries for NT formation.
Table 3: Measured as-sputtered twin spacings and calculated SFEs for the selected compositions
characterized via STEM in Figure 41c-f.
Sample Name As-Sputtered Twin Spacing (nm) Estimated SFE (mJ m-2
)
Cu Rich
Cu72.4Ni17.6Al10.0
7.4 +/- 6.0 102
Ni Rich
Cu35.6Ni49.2Al15.2
2.5 +/- 1.7 67
Al Rich
Cu45.0Ni16.5Al38.5
1.1 +/- 0.6 52
Intermediate
Cu49.9Ni29.7Al20.4
2.8 +/- 1.8 74

116
6.3.2 XRD Analysis of Phase Evolution
To further evaluate individual compositional effects on NT evolution, relationships
between composition, NT formation, and secondary phase evolution were investigated using highthroughput XRD material library data for both as-sputtered and annealed combinatorial samples.
The annealed combinatorial array of samples was used to provide greater resolution when studying
these relationships, as the STEM analysis in Figure 41 highlighted that the cross-sectional
microstructures of the as-sputtered samples were predominantly highly NT solid solutions. An
annealing treatment of 400 °C for 3 hours was chosen based on a previous study to examine the
phase evolution at an intermediate step before reaching equilibrium [198]. Figure 42 presents an
overview of the XRD data for both the as-sputtered and annealed combinatorial arrays, with the
data for each specific square available in a repository at the following source [233]. Figures 42a
and 42b show the normalized XRD patterns for the 169 as-sputtered and annealed samples,
respectively, where high background noise diffractograms were removed to enhance peak
visibility. Both combinatorial arrays displayed FCC and B2 NiAl phases, although greater
variation in XRD patterns was observed in the annealed samples. This is expected, as sputtering
often promotes the formation of a single phase solid solution, while annealing allows the material
to approach equilibrium phases [234]. Figures 42c and 42d highlight the two types of XRD patterns
that were observed, which were either a pure single phase FCC solid solution (42c) or a
combination of the FCC solid solution with a B2 NiAl phase (42d). All FCC and B2 NiAl XRD
patterns displayed strong (111) and (110) texturing, respectively, and the ratio of these two peak
intensities was observed to vary with changing composition, indicating a change in the phase
volume fractions. NiAl phase formation upon annealing is expected around this compositional
range since the ratio of Al to Ni generally aligns with the binary and ternary phase diagrams and

117
NiAl is a phase with high thermodynamic stability [228, 235, 236]. However, the correlation
between secondary phase formation and composition remains unclear, as composition will affect
both the change in Gibb’s free energy for phase formation as well as the presence of nanofeatures,
such as NTs, which can affect the phase transformation pathways [221, 237-239]. For the assputtered array of combinatorial samples in Figure 42a, the majority of XRD diffractograms
showed the single-phase FCC solid solution, while for the annealed CuNiAl alloys, a larger
fraction of samples exhibited the dual FCC and B2 NiAl XRD patterns. This global XRD
evaluation highlights that there are varying degrees of phase evolution across the NT CuNiAl
composition space.

118
Figure 42: High-throughput XRD analysis of the as-sputtered and annealed CuNiAl combinatorial
arrays. (a) and (b) depict an overview of the XRD patterns for the as-sputtered and annealed
CuNiAl samples, respectively, where XRD patterns with high background noise were removed to
improve peak resolution. (c) and (d) highlight the two types of XRD patterns (plotted on a log
scale) that were observed in the combinatorial samples: an FCC solid solution (42c) or a
combination of an FCC and a B2 NiAl phase (42d). In 2a-d, the FCC peaks are represented by the
(*) and the B2 NiAl peaks are represented by the (+).
In order to further link composition and phase formation in the annealed CuNiAl
combinatorial samples, the FCC and B2 NiAl (111) and (110) diffraction peaks were used to
calculate the corresponding phase volume fractions. Figure 43a maps the change in FCC volume
fraction across the combinatorial array, which ranged from a high of 0.94 (shown in bright green)
to a low of 0.06 (shown in dark green). When compared with the composition map in Figure 41a,
it can be observed that there is a greater presence of B2 NiAl phase in the annealed samples with
higher Al concentrations. This is likely due to the fact that these samples contain larger
concentrations of both Al and Ni, which can increase the rate of B2 NiAl phase formation [236].
The change in volume fraction was also studied as a function of constant Al, Cu, or Ni content to
further analyze the effects of individual elements on varying phase formation, highlighted by the
squares in Figure 43a with red (Al), blue (Cu), and black borders (Ni). The XRD patterns from
these selected samples are shown in Figures 43b, 43c and 43d, where the varying peak intensities
and diffraction patterns were compared and linked with changing phase formation. Figure 43b
depicts the selected diffractograms from the constant Al set of samples, where it is observed that
the XRD patterns do not significantly change as a function of Ni or Cu concentration and,
correspondingly, the FCC volume fraction only varies from 0.09 to 0.16. This result is in contrast
with studies on Cu and NiAl precipitate formation in steels, where it was demonstrated that the
nucleation of these two phases was linked with Cu concentration [240, 241]. For the constant Cu
samples in Figure 43c, the four compositions with greater Ni content displayed an FCC XRD

119
pattern with similar relative peak intensities, resulting in minimal change in the FCC volume
fraction (ranging from 0.87 to 0.82). However, the XRD pattern for the fifth sample, which had
the greatest Al concentration, exhibited both the FCC phase and a B2 NiAl (110) peak, causing
the FCC volume fraction to decrease to 0.50. This highlights that increasing Al aluminum content
plays a more significant role than Ni in promoting B2 NiAl phase formation in the NT alloys.
The relationship between Al content and B2 NiAl phase formation is most clearly observed
in the constant Ni set of samples, shown in Figure 43d, where the normalized intensity of the B2
NiAl (110) peak increases as a function of increasing Al concentration. As a result, the FCC
volume fraction in the constant Ni set of CuNiAl samples (Figure 43d) decreases from 0.93 to
0.16. The change in volume fraction indicates that B2 NiAl phase formation in the CuNiAl alloys
is primarily dependent on Al concentration. The singular dependence on Al concentration is
interesting, as previous research has demonstrated that both Al and Ni content directly influence
the thermodynamics driving phase formation in the binary alloy system [242]. This suggests that
variations in both composition and microstructural features, such as NT formation, affect the phase
evolution of B2 NiAl in the CuNiAl alloys; however, the role of each variable remains unclear
[243]. Thus, in the following section, STEM analysis of the annealed CuNiAl combinatorial
squares is used to identify the key microstructural mechanisms driving phase evolution by
deconvoluting the effects of composition and the initial NT microstructure.

120
Figure 43: Analysis of compositional effects on XRD diffraction patterns and CuNiAl phase
volume fractions. (a) FCC volume fraction heat map for the annealed CuNiAl combinatorial array
of samples, where FCC volume fraction was observed to vary from 0.06 to 0.96. A brighter green
indicates a higher FCC volume fraction and a darker green indicates a lower FCC volume fraction.
(b-d) Selected XRD patterns plotted on a normalized log scale from annealed CuNiAl samples
with constant Al (43b), Cu (43c), and Ni (43d) content. The arrows to the right of the selected
diffractograms highlight the changes in composition. In 43b-d, the FCC peaks are represented by
the (*) and the B2 NiAl peaks are represented by the (+).
6.3.3 STEM Analysis of Microstructural Evolution
Figure 44 depicts cross-sectional STEM micrographs and EDX maps from four selected
annealed CuNiAl alloys, which were used to analyze the roles of composition and the initial NT
microstructure on phase evolution. The cross-sections in Figures 44a-c were taken from the
constant Ni set of samples in Figure 43, where B2 NiAl phase formation was observed to increase

121
as a function of Al content. To link microstructural evolution with initial NT formation, the TB
spacings for the corresponding as-sputtered versions of selected samples were also determined (see
supplementary materials and Table 3). In the lowest Al content sample in Figure 44a (10.0 at%
Al), the majority of the as-sputtered columnar NT microstructure is retained in the annealed crosssection, with only small B2 NiAl precipitate formation and negligible change from the initial
average TB spacing of 7.4 +/- 6.0 nm. For the sample with intermediate Al concentration (23.9
at% Al) shown in Figure 44b, there is also minimal change from the as-sputtered TB spacing of
1.6 +/- 1.2 nm, however, there is larger B2 NiAl precipitate formation. The negligible change in
TB spacing in Figure 44b is unexpected since higher initial NT densities have been shown to
reduce the thermal stability [198, 216, 220, 239]. The largest microstructural transformation is
observed in the sample with the greatest Al content (38.5 at % Al) shown in Figure 44c, where the
as-sputtered columnar NT grains, which had the smallest initial TB spacing of 1.1 +/- 0.6 nm, have
been completely replaced with a microstructure comprised of a B2 NiAl matrix and Cu
precipitates. The increase in B2 NiAl phase formation aligns with the XRD data in Figure 43;
however, the observations in Figure 44c indicate that the FCC volume fraction in some of the
samples could be attributed to Cu precipitate formation and not from the initial NT columnar
microstructure. The STEM and EDX analysis also highlight that both composition and the initial
NT microstructure influence the thermally driven phase and microstructural evolution. With
respect to initial NT formation, the varying TB spacing could be altering transformation pathways
by increasing the interfacial free energy and thereby the driving forces for thermal processes like
grain growth and recrystallization [216, 244]. For example, Bahena et al. demonstrated that
reducing the initial TB spacing from 18 nm to 5 nm in Cu 2 at% Al alloys increased the driving
force from ~1333 kJ m-3
to ~4800 kJ m-3
, leading to greater abnormal grain growth [216, 244]. For

122
the CuNiAl alloys in this study, similar calculations were performed, and the calculated driving
forces were approximately 13784 kJ m-3 (Figure 44a), 40000 kJ m-3 (Figure 44b), and 47272 kJ m3
(Figure 44c), where the threefold increase in driving force between the low and intermediate Al
content samples, could explain the increase in precipitate formation. However, interfacial free
energy does not fully describe phase evolution in the CuNiAl alloys since there is a much smaller
difference in driving force between the intermediate and high Al content samples. This indicates
that, in addition to influencing NT formation, composition is also directly affecting thermal
evolution by changing either (1) the diffusivity and transformation kinetics, or (2) the
thermodynamics driving phase formation. Regarding the transformation kinetics, greater diffusion
rates can enable faster precipitate coarsening and secondary phase formation and it has been
demonstrated that varying composition can increase diffusivity [50]. For example, CuAl alloys
with greater Al content have exhibited increased rates of Al and Cu self-diffusion [245]. However,
since the CuNiAl alloys in this study were annealed at 400 °C, the self-diffusion coefficients are
predicted to be extremely low (NiAl and CuAl binary alloy self-diffusion constants are on the
order of 10-24 cm2
s
-1
at 400 °C) and thus, this mechanism is not expected to dictate thermal
evolution [245, 246]. Therefore, composition is influencing phase evolution by altering the
thermodynamic driving forces, explaining the combined effects of composition and
microstructure. To determine how increased compositional complexity is affecting phase
formation, thermodynamic trends in binary NiAl alloys can be compared with the ternary CuNiAl
system. In NiAl alloys, Al content has been shown to influence B2 NiAl phase formation by
altering the enthalpy of formation, reaching a maximum driving force at approximately equal
concentrations of Ni and Al [237]. For the CuNiAl alloys in this study, introducing Cu appears to
change the relationships between composition and the thermodynamic driving force, as the greatest

123
B2 NiAl formation is observed in the sample with the highest Al concentration (Figure 44c),
instead of the sample with equal Al and Ni content (Figure 44b).
Figure 44: Cross-sectional HAADF STEM micrographs and Cu (green), Ni (blue), and Al (red)
EDX composition maps for four selected annealed CuNiAl compositions highlighting changes in
the cross-sectional microstructure and B2 NiAl phase formation. (a-c) Denote the constant Ni set
of samples in Figure 43d, with corresponding compositions of Cu72.4Ni17.6Al10.0 (44a),
Cu58.6Ni17.5Al23.9 (44b), and Cu45.1Ni16.5Al38.4 (44c) and (44d) examines the role of varied Ni
content on phase formation with a composition of Cu53.0Ni35.5Al11.5.
To further investigate the relationships between composition, initial NT formation, and
thermal evolution, Figure 44d examines the cross-sectional microstructure of a CuNiAl sample

124
with similar Al content and FCC volume fractions (Cu53.0Ni35.5Al11.6, 0.87) as the sample in Figure
44a (Cu72.4Ni17.6Al10.0, 0.93), but with a varied Ni concentration. Comparing the STEM
micrographs and EDX maps it is observed that despite a twofold increase in Ni content, the sample
in Figure 44d has similar B2 NiAl precipitate formation and annealed TB spacings as the sample
in Figure 44a (3.7 +/- 2.2 nm in Figure 44d and 4.7 +/- 3.8 nm in Figure 44a). In contrast, doubling
the Al content in Figure 44b resulted in significantly larger B2 NiAl precipitate formation. The
limited effect of Ni content on phase evolution in the NT CuNiAl system could be attributed to its
miscibility in Cu alloys promoting the increased thermodynamic stability of an FCC phase [209,
228]. For example, Wang et al. demonstrated using CALPHAD that Ni rich CuNiAl alloys (~ 80
- 90 at% Ni) exhibited an equilibrium FCC phase from 500 – 1400 °C [228]. Additionally,
increasing Ni content typically yields larger NT spacing, reducing the thermodynamic driving
force from the microstructural feature [226]. Overall, this highlights that Al content is the primary
compositional variable influencing thermal evolution in the NT CuNiAl alloys.
6.4 Conclusion
NT formation was investigated in the CuNiAl alloy system using a combinatorial and highthroughput approach, in order to elucidate relationships between composition , NT formation, and
microstructural evolution in alloys with three or more elements. 169 unique CuNiAl compositions
were analyzed both as-sputtered and annealed via high-throughput XRD and STEM. STEM
analysis of the compositional extremes was used to establish the compositional boundaries for NT
formation and also to examine the effects of individual elements, where it was observed that the
presence of a third alloying element altered NT formation compared to the binary alloy systems.
After identifying the NT compositional domains, phase evolution in the CuNiAl alloys was

125
analyzed using the high-throughput XRD data. It was shown that the as-sputtered samples were
more likely to yield an FCC solid solution, while the annealed CuNiAl combinatorial array
displayed both FCC and B2 NiAl diffraction patterns. The relative intensities of the diffraction
patterns were used to calculate phase volume fractions and it was determined that B2 NiAl phase
formation after annealing was primarily dependent on Al concentration. To deconvolute the roles
of composition and initial NT formation on microstructural and phase evolution, selected annealed
CuNiAl compositions were investigated using STEM. This highlighted that greater Al content
increased the thermodynamic driving forces for phase formation by changing both the composition
as well as promoting higher initial NT densities. Ultimately, this work demonstrates the ability of
a CHT approach to investigate and develop a fundamental understanding of NT behavior in more
complex compositional domains.
6.5 Supplementary Materials
Figure S4 depicts STEM micrographs of the as-sputtered versions of the CuNiAl alloys analyzed
in Figures 44a-c. The averaged TB spacings for the as-sputtered samples were 7.4 +/- 6.0 nm
(S4a), 1.6 +/- 1.2 nm (S4b), and 1.1 +/- 0.6 nm (S4c).

126
Figure S4: Cross-sectional HAADF STEM micrographs and EDX maps for the corresponding
as-sputtered versions of the CuNiAl alloys examined in Figures 44a-c, which had compositions
of Cu72.4Ni17.6Al10.0 (S4a), Cu58.6Ni17.5Al23.9 (S4b), and Cu45.1Ni16.5Al38.4 (S4c)

127
Chapter 7: Conclusions and Future Work
7.1 Conclusions
There is great potential for the development of new materials with enhanced properties,
given the nearly infinite number of combinations of composition, microstructure, and morphology.
However, navigating large synthesis spaces is a major challenge for material discovery, as only a
limited fraction of these combinations can be explored. Thus, Chapters 4 through 6 highlight novel
approaches that can be used to more effectively discover and optimize new materials. In these
investigations, magnetron sputtering is leveraged to demonstrate how processing parameters and
composition can be varied to efficiently explore large synthesis domains.
In Chapter 4, planar and hollow cathode sputtering were compared to understand how
variations in the deposition environment could affect film growth, microstructure, and
morphology. It was observed that at the same sputtering parameters the hollow cathode target
geometry yielded higher ion densities, homologous temperatures, and deposition rates than the
planar configuration. By examining the extremes of the processing environment, this study was
able to globally map changes in sputtered film morphology. In the planar cathode, it was observed
that film morphology was primarily dependent on the unidirectional deposition angle, while in the
hollow cathode, the increased line of sight enabled film growth to be influenced by other deposition
variables, such as ion density and homologous temperature. This highlighted that target geometry
can be used as a novel processing variable to access unique combinations of microstructure and
morphology.
To examine the role of composition in sputtering synthesis domains, compositionmicrostructure relationships were studied in Chapter 5 using combinatorial co-sputtering. Growth

128
nanotwinning specifically, was investigated in the CuNi alloy system to elucidate relationships
between composition, stacking fault energy, and twin boundary formation. Nanotwin formation
was analyzed across the composition space using over 300 unique CuNi samples and scanning
transmission electron microscopy was utilized in conjunction with high-throughput
characterization techniques to create comprehensive material libraries with composition, phase,
mechanical property, and nanotwin formation data. Using the material library data, this study was
able to demonstrate a novel approach to efficiently investigate nanotwin formation in large
compositional spaces and develop a more accurate thermodynamic model linking composition,
stacking fault energy, and growth twin formation.
The updated thermodynamic model was used in the work in Chapter 6 to investigate growth
nanotwin formation and behavior in more complex composition domains. In order to study the
relationships between composition, secondary phases, nanotwin formation, and microstructural
evolution, the combinatorial co-sputtering approach was expanded to synthesize the CuNiAl alloy
system, which was analyzed both as-sputtered and annealed. Scanning transmission electron
microscopy and high-throughput X-ray diffraction were used to identify the compositional
boundaries for nanotwin formation as well as determine changes in phase volume fraction and
cross-sectional microstructure. It was demonstrated that Al content was the dominant variable
influencing both nanotwin formation and microstructural evolution. Ultimately, this work
highlights a novel approach to study nanotwin formation and behavior in alloys with three or more
elements and elucidate the roles of individual elements on phase evolution.

129
7.2 Future Work
The studies detailed in Chapters 4 through 6 demonstrate multiple approaches to rapidly
investigate entire synthesis spaces as well as locally study specific material phenomena. However,
these techniques still need to be optimized to further enable material discovery and there are many
unanswered scientific questions still remaining. Three areas of considerable interest are:
investigating the effects of sputtering parameters when depositing onto more complex topologies,
understanding compositional effects on SFE in quaternary, high entropy, and engineering alloys,
and leveraging the data from combinatorial and high-throughput investigations to develop machine
learning models that can further expedite investigations.
The increased line of sight during deposition with the hollow cathode has been used to
improve coating coverage and uniformity when depositing onto complex topology substrates like
nano- and micro-lattices [59]. Coatings are used on these architected materials to access larger
material workspaces and compositions with favorable material properties [12]. Additionally, the
microstructure of the deposited coating can be manipulated to improve material properties by
adding another layer of structural hierarchy. However, film growth when depositing onto more
complex topologies is not well understood; for example, it has been demonstrated that during
sputtering different microstructures can form on various surfaces of the same substrate [58]. Thus,
a fundamental understanding of deposition onto complex topology substrates is needed to improve
control of coating microstructure and morphology on these materials. Further investigations could
address this challenge by mapping change in film uniformity, microstructure, and morphology on
complex substrates as a function of sputtering processing parameters, including deposition
geometry, plasma ion density, and homologous temperature, to identify routes to optimized
synthesis conditions.

130
To enable further investigations into nanotwin formation and behavior in more complex
compositional spaces, it is necessary to develop a fundamental understanding of how stacking fault
energy evolves in alloys with three or more elements. Future research can elucidate relationships
between composition and stacking fault energy by creating stacking fault energy databases using
either computational or experimental approaches. Computationally, stacking fault energies can be
calculated using molecular static and density functional theory simulations, while, experimentally,
they can be determined using measured growth twin boundary spacings and the model developed
in Chapter 5 [193, 226]. To study the role of specific elements, complex engineering alloys like
Inconel can be broken down into their binary and ternary constituents, such as NiFe, NiCr, and
NiFeCr. Composition-stacking fault energy trends in these alloy systems can be used to identify
elements that could promote or limit nanotwin formation. Additionally, combinatorial and highthroughput techniques can be used to examine the binary and ternary constituents of the high
entropy or engineering alloys for varied phase formation and microstructural evolution to
understand how these features can affect nanotwin behavior.
Finally, the data from combinatorial and high-throughput growth nanotwin investigations
can be used to train machine learning models, where inputs such as composition, deposition rate,
hardness, sheet resistivity, and stacking fault energy, can be linked with nanotwin formation and
twin boundary spacing. After ingesting the training data, these models can be used to more
efficiently identify other material systems that can be investigated for growth nanotwin formation.
This would greatly reduce the number of time intensive stacking fault energy simulations and
cross-sectional transmission electron microscopy investigations that are currently needed to study
nanotwin formation in unexplored composition spaces. There are many other areas to explore, but
these present a path forward to expand analysis into more complex synthesis domains.

131
Overall, the studies in this dissertation present novel approaches to rapidly investigate
synthesis spaces as a function of processing parameters and composition. Sputtering target
geometry was identified as a unique variable that could alter the deposition environment and film
growth mechanisms and co-sputtering was used to scan entire composition spaces and develop a
deeper understanding of the relationships between composition, stacking fault energy and
nanotwin formation. These studies have explored exciting scientific challenges and provide a clear
foundation to enable more efficient material discovery.

132
References
1. Kaiser, N., Review of the fundamentals of thin-film growth. Applied optics, 2002. 41(16):
p. 3053-3060.
2. Appleget, C.D. and A.M. Hodge, Synthesis and characterization of optically transparent
ceramic crystalline/amorphous and amorphous/amorphous multilayers. Scripta
Materialia, 2020. 187: p. 157-162.
3. Kelly, P.J. and R.D. Arnell, Magnetron sputtering: a review of recent developments and
applications. Vacuum, 2000. 56(3): p. 159-172.
4. Clarke, D.R., Materials selection guidelines for low thermal conductivity thermal barrier
coatings. Surface and Coatings Technology, 2003. 163: p. 67-74.
5. Vassen, R., A. Stuke, and D. Stöver, Recent developments in the field of thermal barrier
coatings. Journal of thermal spray technology, 2009. 18: p. 181-186.
6. Cui, G., et al., A comprehensive review on smart anti-corrosive coatings. Progress in
organic coatings, 2020. 148: p. 105821.
7. Khamseh, S., et al., Magnetron-sputtered copper/diamond-like carbon composite thin
films with super anti-corrosion properties. Surface and Coatings Technology, 2018. 333:
p. 148-157.
8. Fotovvati, B., N. Namdari, and A. Dehghanghadikolaei, On coating techniques for
surface protection: A review. Journal of Manufacturing and Materials processing, 2019.
3(1): p. 28.
9. Licari, J.J., Coating materials for electronic applications: polymers, processing,
reliability, testing. 2003.
10. Ojo, A.A. and I.M. Dharmadasa, Electroplating of semiconductor materials for
applications in large area electronics: A review. Coatings, 2018. 8(8): p. 262.
11. Smentkowski, V.S., Trends in sputtering. Progress in Surface Science, 2000. 64(1-2): p.
1-58.
12. Garcia-Taormina, A.R., et al., A review of coated nano-and micro-lattice materials.
Journal of Materials Research, 2021. 36: p. 3607-3627.
13. Meza, L.R., S. Das, and J.R. Greer, Strong, lightweight, and recoverable threedimensional ceramic nanolattices. Science, 2014. 345(6202): p. 1322-1326.
14. Xiong, J., et al., Advanced micro‐lattice materials. Advanced Engineering Materials,
2015. 17(9): p. 1253-1264.

133
15. Larhlimi, H., et al., Magnetron sputtered titanium carbide-based coatings: A review of
science and technology. Vacuum, 2021: p. 110853.
16. Carlsson, J.-O. and P.M. Martin, Chemical vapor deposition, in Handbook of Deposition
Technologies for films and coatings. 2010, Elsevier. p. 314-363.
17. Cheng, L.-f., et al., Preparation of an oxidation protection coating for C/C composites by
low pressure chemical vapor deposition. Carbon, 2000. 38(10): p. 1493-1498.
18. Garcia, J.V. and T. Goto, Thermal barrier coatings produced by chemical vapor
deposition. Science and Technology of Advanced Materials, 2003. 4(4): p. 397-402.
19. Vossen, J.L., W. Kern, and W. Kern, Thin film processes II. Vol. 2. 1991: Gulf
Professional Publishing.
20. Dendooven, J. and C. Detavernier, Basics of atomic layer deposition: growth
characteristics and conformality. Atomic layer deposition in energy conversion
applications, 2017: p. 1-40.
21. Johnson, R.W., A. Hultqvist, and S.F. Bent, A brief review of atomic layer deposition:
from fundamentals to applications. Materials today, 2014. 17(5): p. 236-246.
22. Parsons, G.N., S.M. George, and M. Knez, Progress and future directions for atomic
layer deposition and ALD-based chemistry. Mrs Bulletin, 2011. 36(11): p. 865-871.
23. Ohno, I., Electrochemistry of electroless plating. Materials Science and Engineering: A,
1991. 146(1-2): p. 33-49.
24. Muench, F., Electroless plating of metal nanomaterials. ChemElectroChem, 2021. 8(16):
p. 2993-3012.
25. Loto, C., Electroless nickel plating–a review. 2016, Springer.
26. Djokić, S.S. and P.L. Cavallotti, Electroless deposition: theory and applications.
Electrodeposition: Theory and Practice, 2010: p. 251-289.
27. Giurlani, W., et al., Electroplating for decorative applications: Recent trends in research
and development. Coatings, 2018. 8(8): p. 260.
28. Gurrappa, I. and L. Binder, Electrodeposition of nanostructured coatings and their
characterization—a review. Science and Technology of Advanced Materials, 2008.
29. Zhang, B., Amorphous and nano alloys electroless depositions: technology, composition,
structure and theory. 2015: Elsevier.
30. Aliofkhazraei, M., et al., Development of electrodeposited multilayer coatings: A review
of fabrication, microstructure, properties and applications. Applied Surface Science
Advances, 2021. 6: p. 100141.

134
31. Tarozait, R., et al., Composition, microstructure and magnetic properties of electrolessplated thin Co–P films. Surface and Coatings Technology, 1999. 115(1): p. 57-65.
32. Ambat, R. and W. Zhou, Electroless nickel-plating on AZ91D magnesium alloy: effect of
substrate microstructure and plating parameters. Surface and Coatings Technology,
2004. 179(2-3): p. 124-134.
33. Brenner, A., Electrodeposition of alloys: principles and practice. 2013: Elsevier.
34. Gudmundsson, J.T., Physics and technology of magnetron sputtering discharges. Plasma
Sources Science and Technology, 2020. 29(11): p. 113001.
35. Ohring, M., Materials Science of Thin Films: Depositon and Structure. 2001: Elsevier.
36. Appleget, C.D., J.S. Riano, and A.M. Hodge, An Overview of Nano Multilayers as Model
Systems for Developing Nanoscale Microstructures. Materials, 2022. 15(1): p. 382.
37. Thornton, J.A., The microstructure of sputter‐deposited coatings. Journal of Vacuum
Science & Technology A: Vacuum, Surfaces, and Films, 1986. 4(6): p. 3059-3065.
38. Anders, A., A structure zone diagram including plasma-based deposition and ion
etching. Thin Solid Films, 2010. 518(15): p. 4087-4090.
39. Helmersson, U., et al., Ionized physical vapor deposition (IPVD): A review of technology
and applications. Thin solid films, 2006. 513(1-2): p. 1-24.
40. d’Heurle, F., Aluminum films deposited by rf sputtering. Metallurgical and Materials
Transactions B, 1970. 1: p. 725-732.
41. Vossen, J., Control of film properties by rf-sputtering techniques. Journal of Vacuum
Science and Technology, 1971. 8(5): p. S12-S30.
42. Solovyev, A., et al., Comparative study of Cu films prepared by DC, high-power pulsed
and burst magnetron sputtering. J. Electron. Mater., 2016. 45: p. 4052.
43. Greczynski, G., J. Jensen, and L. Hultman, Mitigating the geometrical limitations of
conventional sputtering by controlling the ion-to-neutral ratio during high power pulsed
magnetron sputtering. Thin Solid Films, 2011. 519(19): p. 6354-6361.
44. Muhl, S. and A. Pérez, The use of hollow cathodes in deposition processes: A critical
review. Thin Solid Films, 2015. 579: p. 174-198.
45. Thornton, J.A., Influence of apparatus geometry and deposition conditions on the
structure and topography of thick sputtered coatings. Journal of Vacuum Science and
Technology, 1974. 11(4): p. 666-670.

135
46. Glocker, D.A., M.M. Romach, and V.W. Lindberg, Recent developments in inverted
cylindrical magnetron sputtering. Surface and Coatings Technology, 2001. 146: p. 457-
462.
47. Ludwig, A., Discovery of new materials using combinatorial synthesis and highthroughput characterization of thin-film materials libraries combined with computational
methods. NPJ Computational Materials, 2019. 5(1): p. 70.
48. Lelis, M., et al., Tailoring of TiO2 film microstructure by pulsed-DC and RF magnetron
co-sputtering. Surface and Coatings Technology, 2019. 377: p. 124906.
49. Petrov, I., et al., Microstructural evolution during film growth. Journal of Vacuum
Science & Technology A: Vacuum, Surfaces, and Films, 2003. 21(5): p. S117-S128.
50. Porter, D.A. and K.E. Easterling, Phase transformations in metals and alloys (revised
reprint). 2009: CRC press.
51. Musil, J. and J. Vlček, Magnetron sputtering of alloy and alloy-based films. Thin solid
films, 1999. 343: p. 47-50.
52. Venables, J., G. Spiller, and M. Hanbucken, Nucleation and growth of thin films. Reports
on progress in physics, 1984. 47(4): p. 399.
53. Lozovoy, K.A., et al., Kinetics of epitaxial formation of nanostructures by Frank–van der
Merwe, Volmer–Weber and Stranski–Krastanow growth modes. Surface and Coatings
Technology, 2020. 384: p. 125289.
54. Myers, A., et al., Monte Carlo simulations of magnetron sputtering particle transport.
Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films, 1991. 9(3):
p. 614-618.
55. Messier, R., A. Giri, and R. Roy, Revised structure zone model for thin film physical
structure. Journal of Vacuum Science & Technology A: Vacuum, Surfaces, and Films,
1984. 2(2): p. 500-503.
56. Barna, P. and M. Adamik, Fundamental structure forming phenomena of polycrystalline
films and the structure zone models. Thin solid films, 1998. 317(1-2): p. 27-33.
57. Wolff-Goodrich, S., et al., Combinatorial exploration of B2/L21 precipitation
strengthened AlCrFeNiTi compositionally complex alloys. Journal of Alloys and
Compounds, 2021. 853: p. 156111.
58. Thornton, J.A. and V.L. Hedgcoth, Tubular hollow cathode sputtering onto substrates of
complex shape. Journal of Vacuum Science and Technology, 1975. 12(1): p. 93-97.
59. Garcia-Taormina, A.R., et al., Coatings for Core–Shell Composite Micro‐Lattice
Structures: Varying Sputtering Parameters. Advanced Engineering Materials, 2022.
24(4): p. 2101264.

136
60. Gleiter, H., Nanostructured materials: state of the art and perspectives. Nanostructured
materials, 1995. 6(1-4): p. 3-14.
61. Suryanarayana, C., Nanocrystalline materials. International materials reviews, 1995.
40(2): p. 41-64.
62. Cordero, Z.C., B.E. Knight, and C.A. Schuh, Six decades of the Hall–Petch effect–a
survey of grain-size strengthening studies on pure metals. International Materials
Reviews, 2016. 61(8): p. 495-512.
63. Malygin, G., Plasticity and strength of micro-and nanocrystalline materials. Physics of
the Solid State, 2007. 49: p. 1013-1033.
64. Wang, J., et al., Enhanced corrosion resistance of CoCrFeMnNi high entropy alloy using
heterogeneous structure design. Corrosion Science, 2022. 209: p. 110761.
65. Xiong, T., et al., Enhancing strength and thermal stability of TWIP steels with a
heterogeneous structure. Materials Science and Engineering: A, 2018. 720: p. 231-237.
66. Ma, Y., et al., A review on heterogeneous nanostructures: a strategy for superior
mechanical properties in metals. Metals, 2019. 9(5): p. 598.
67. Edwards, T.E.J., et al., Thermally stable nanotwins: new heights for Cu mechanics.
Advanced science, 2022. 9(34): p. 2203544.
68. Sansoz, F., et al., Strengthening and plasticity in nanotwinned metals. Mrs Bulletin, 2016.
41(4): p. 292-297.
69. Callister, W.D. and D.G. Rethwisch, Materials Science and Engineering: An
Introduction. 9 ed. 2014, Hoboken, New Jersey: Wiley.
70. Christian, J. and V. Vitek, Dislocations and stacking faults. Reports on Progress in
Physics, 1970. 33(1): p. 307.
71. Hull, D. and D.J. Bacon, Introduction to dislocations. Vol. 37. 2011: Elsevier.
72. Li, W., et al., Generalized stacking fault energies of alloys. Journal of Physics:
Condensed Matter, 2014. 26(26): p. 265005.
73. Humphreys, F.J. and M. Hatherly, Recrystallization and related annealing phenomena.
2012: elsevier.
74. Gleiter, H., The formation of annealing twins. Acta metallurgica, 1969. 17(12): p. 1421-
1428.
75. Roy, B. and J. Das, Strengthening face centered cubic crystals by annealing induced
nano-twins. Scientific Reports, 2017. 7(1): p. 1-8.

137
76. Mahajan, S. and D. Williams, Deformation twinning in metals and alloys. International
Metallurgical Reviews, 1973. 18(2): p. 43-61.
77. Uttam, P., et al., Nanotwinning: Generation, properties, and application. Materials &
Design, 2020. 192: p. 108752.
78. Beyerlein, I.J., X. Zhang, and A. Misra, Growth twins and deformation twins in metals.
Annual Review of Materials Research, 2014. 44: p. 329-363.
79. Zhang, X., et al., Enhanced hardening in Cu/330 stainless steel multilayers by nanoscale
twinning. Acta materialia, 2004. 52(4): p. 995-1002.
80. Velasco, L. and A.M. Hodge, The mobility of growth twins synthesized by sputtering:
Tailoring the twin thickness. Acta Materialia, 2016. 109: p. 142-150.
81. Shen, Y., et al., Tensile properties of copper with nano-scale twins. Scripta Materialia,
2005. 52(10): p. 989-994.
82. Lu, K., L. Lu, and S. Suresh, Strengthening materials by engineering coherent internal
boundaries at the nanoscale. Science, 2009. 324(5925): p. 349-352.
83. Singh, A., et al., Fracture toughness and fatigue crack growth characteristics of
nanotwinned copper. Acta Materialia, 2011. 59(6): p. 2437-2446.
84. Heckman, N.M., et al., Microstructural deformation in fatigued nanotwinned copper
alloys. Acta Materialia, 2018. 144: p. 138-144.
85. Potyrailo, R., et al., Combinatorial and high-throughput screening of materials libraries:
review of state of the art. ACS combinatorial science, 2011. 13(6): p. 579-633.
86. Xiang, X.-D., et al., A combinatorial approach to materials discovery. Science, 1995.
268(5218): p. 1738-1740.
87. Zhao, J.-C., A combinatorial approach for efficient mapping of phase diagrams and
properties. Journal of Materials Research, 2001. 16(6): p. 1565-1578.
88. Zhao, J.-C., A combinatorial approach for structural materials. Advanced Engineering
Materials, 2001. 3.
89. Li, Z., et al., Combinatorial metallurgical synthesis and processing of high-entropy
alloys. Journal of Materials Research, 2018. 33(19): p. 3156-3169.
90. Zhao, J.-C., Combinatorial approaches as effective tools in the study of phase diagrams
and composition–structure–property relationships. Progress in materials science, 2006.
51(5): p. 557-631.

138
91. Shastry, V., et al., Combining indentation and diffusion couple techniques for
combinatorial discovery of high temperature shape memory alloys. Acta materialia,
2013. 61(15): p. 5735-5742.
92. Di, W., et al., Effect of Fe content on microstructures and properties of Ti6Al4V alloy
with combinatorial approach. Transactions of Nonferrous Metals Society of China, 2018.
28(9): p. 1714-1723.
93. Zhang, J., et al., Novel Ge-Sb-Te thermoelectric materials: A demonstration for an
efficient diffusion couple technique in expediently exploiting new thermoelectric
materials. Ceramics International, 2019. 45(13): p. 16039-16045.
94. Bikas, H., P. Stavropoulos, and G. Chryssolouris, Additive manufacturing methods and
modelling approaches: a critical review. The International Journal of Advanced
Manufacturing Technology, 2016. 83: p. 389-405.
95. Welk, B.A., M.A. Gibson, and H.L. Fraser, A combinatorial approach to the
investigation of metal systems that form both bulk metallic glasses and high entropy
alloys. Jom, 2016. 68: p. 1021-1026.
96. Knoll, H., et al., Combinatorial alloy design by laser additive manufacturing. steel
research international, 2017. 88(8): p. 1600416.
97. Tsai, P. and K.M. Flores, A combinatorial strategy for metallic glass design via laser
deposition. Intermetallics, 2014. 55: p. 162-166.
98. Travitzky, N., et al., Additive manufacturing of ceramic‐based materials. Advanced
engineering materials, 2014. 16(6): p. 729-754.
99. Zhao, Y., et al., Compositionally graded CoCrFeNiTix high-entropy alloys manufactured
by laser powder bed fusion: A combinatorial assessment. Journal of Alloys and
Compounds, 2021. 883: p. 160825.
100. Tsai, P. and K.M. Flores, High-throughput discovery and characterization of
multicomponent bulk metallic glass alloys. Acta Materialia, 2016. 120: p. 426-434.
101. Srivastava, M., et al., A review of various materials for additive manufacturing: Recent
trends and processing issues. journal of materials research and technology, 2022. 21: p.
2612-2641.
102. Mukherjee, T., et al., Printability of alloys for additive manufacturing. Scientific reports,
2016. 6(1): p. 19717.
103. McGinn, P.J., Thin-film processing routes for combinatorial materials investigations—a
review. ACS Combinatorial Science, 2019. 21(7): p. 501-515.

139
104. Kube, S.A., et al., Phase selection motifs in High Entropy Alloys revealed through
combinatorial methods: Large atomic size difference favors BCC over FCC. Acta
Materialia, 2019. 166: p. 677-686.
105. Mihai, C., et al., Structural and optical properties of amorphous Si–Ge–Te thin films
prepared by combinatorial sputtering. Scientific reports, 2021. 11(1): p. 1-15.
106. Kim, K., et al., Mechanical, electrical properties and microstructures of combinatorial
Ni-Mo-W alloy films. Journal of Alloys and Compounds, 2022. 919: p. 165808.
107. Mao, F., et al., Combinatorial study of gradient Ag–Al thin films: microstructure, phase
formation, mechanical and electrical properties. ACS Applied Materials & Interfaces,
2016. 8(44): p. 30635-30643.
108. Miracle, D.B., et al., Emerging capabilities for the high-throughput characterization of
structural materials. Annual Review of Materials Research, 2021. 51: p. 131-164.
109. Holm, E.A., et al., Overview: Computer vision and machine learning for microstructural
characterization and analysis. Metallurgical and Materials Transactions A, 2020. 51: p.
5985-5999.
110. Shindo, D., et al., Energy dispersive x-ray spectroscopy. Analytical electron microscopy
for materials science, 2002: p. 81-102.
111. Rajan, K., Combinatorial Materials Science and EBSD: A High Throughput
Experimentation Tool. Electron Backscatter Diffraction in Materials Science, 2009: p.
189-199.
112. Dey, B., et al. Deep learning-based defect classification and detection in SEM images. in
Metrology, Inspection, and Process Control XXXVI. 2022. SPIE.
113. Kim, H., J. Han, and T.Y.-J. Han, Machine vision-driven automatic recognition of
particle size and morphology in SEM images. Nanoscale, 2020. 12(37): p. 19461-19469.
114. Gao, C., et al., Innovative materials science via machine learning. Advanced Functional
Materials, 2022. 32(1): p. 2108044.
115. Cullity, B. and S.R. Stock, Elements of x-ray diffraction. 2014, Pearson: Essex.
116. Withers, P.J., Synchrotron X‐ray Diffraction. Practical residual stress measurement
methods, 2013: p. 163-194.
117. Green, M.L., I. Takeuchi, and J.R. Hattrick-Simpers, Applications of high throughput
(combinatorial) methodologies to electronic, magnetic, optical, and energy-related
materials. Journal of Applied Physics, 2013. 113(23): p. 9_1.

140
118. Chang, H., et al., Combinatorial synthesis and high throughput evaluation of
ferroelectric/dielectric thin-film libraries for microwave applications. Applied Physics
Letters, 1998. 72(17): p. 2185-2187.
119. Okazaki, N., et al., Development of scanning microwave microscope with a lumpedconstant resonator probe for high-throughput characterization of combinatorial
dielectric materials. Applied surface science, 2002. 189(3-4): p. 222-226.
120. Miccoli, I., et al., The 100th anniversary of the four-point probe technique: the role of
probe geometries in isotropic and anisotropic systems. Journal of Physics: Condensed
Matter, 2015. 27(22): p. 223201.
121. Al Hasan, N.M., et al., Combinatorial Exploration and Mapping of Phase
Transformation in a Ni–Ti–Co Thin Film Library. ACS combinatorial science, 2020.
22(11): p. 641-648.
122. Anderoglu, O., et al., Significant enhancement of the strength-to-resistivity ratio by
nanotwins in epitaxial Cu films. Journal of Applied Physics, 2009. 106(2): p. 024313.
123. Liu, X.-D., et al., A general model of dielectric constant for porous materials. Applied
Physics Letters, 2016. 108(10): p. 102902.
124. Zhang, X. and Y. Xiang, Combinatorial approaches for high-throughput
characterization of mechanical properties. Journal of Materiomics, 2017. 3(3): p. 209-
220.
125. Woo, N.C., B.G. Ng, and R.B. van Dover, High-throughput combinatorial study of local
stress in thin film composition spreads. Review of Scientific Instruments, 2007. 78(7): p.
072208.
126. Sasangka, W.A., et al., Characterization of the Young’s modulus, residual stress and
fracture strength of Cu–Sn–In thin films using combinatorial deposition and microcantilevers. Journal of Micromechanics and Microengineering, 2015. 25(3): p. 035023.
127. Xiao, Y., et al., Combinatorial investigation of Al–Cu intermetallics using small-scale
mechanical testing. Journal of Alloys and Compounds, 2020. 822: p. 153536.
128. Schajer, G.S., Stress determination for coatings. Surface Engineering, ASM, 1994.
129. Ludwig, A., et al., MEMS tools for combinatorial materials processing and highthroughput characterization. Measurement Science and Technology, 2004. 16(1): p. 111.
130. Magagnosc, D.J., et al. Femtosecond laser machining of micro-tensile specimens for high
throughput mechanical testing. in Micro and Nanomechanics, Volume 5: Proceedings of
the 2017 Annual Conference on Experimental and Applied Mechanics. 2018. Springer.

141
131. Oellers, T., et al., Thin-film microtensile-test structures for high-throughput
characterization of mechanical properties. ACS combinatorial science, 2020. 22(3): p.
142-149.
132. Slaughter, S.K., et al. High throughput femtosecond-laser machining of micro-tension
specimens. in TMS 2015 144 th Annual Meeting & Exhibition: Supplemental
Proceedings. 2016. Springer.
133. Gaspar, J., et al. High-throughput wafer-scale microtensile testing of thin films. in 2008
IEEE 21st International Conference on Micro Electro Mechanical Systems. 2008. IEEE.
134. Fischer-Cripps, A.C., Critical review of analysis and interpretation of nanoindentation
test data. Surface and coatings technology, 2006. 200(14-15): p. 4153-4165.
135. Hintsala, E.D., U. Hangen, and D.D. Stauffer, High-throughput nanoindentation for
statistical and spatial property determination. Jom, 2018. 70(4): p. 494-503.
136. Wang, C., et al., High throughput analysis of solute effects on the mechanical behavior
and slip activity of beta titanium alloys. Materials & Design, 2018. 137: p. 371-383.
137. Liu, Z., et al., High-speed nanoindentation mapping of a near-alpha titanium alloy made
by additive manufacturing. Journal of Materials Research, 2021. 36: p. 2223-2234.
138. Conde, L., An introduction to Langmuir probe diagnostics of plasmas. Madrid: Dept.
Física. ETSI Aeronáut ngenieros Aeronáuticos Universidad Politécnica de Madrid, 2011:
p. 1-28.
139. Merlino, R.L., Understanding Langmuir probe current-voltage characteristics. American
Journal of Physics, 2007. 75(12): p. 1078-1085.
140. Cherrington, B., The use of electrostatic probes for plasma diagnostics—A review.
Plasma chemistry and plasma processing, 1982. 2: p. 113-140.
141. Petersen, D.H., et al., Review of electrical characterization of ultra-shallow junctions
with micro four-point probes. Journal of Vacuum Science & Technology B,
Nanotechnology and Microelectronics: Materials, Processing, Measurement, and
Phenomena, 2010. 28(1): p. C1C27-C1C33.
142. Alwen, A. and A.M. Hodge, Correlation between plasma characteristics, morphology,
and microstructure of sputtered CuAl alloy films with varied target geometry. Materials
Research Express, 2023.
143. Leng, Y., Materials characterization: introduction to microscopic and spectroscopic
methods. 2009: John Wiley & Sons.
144. Inkson, B.J., Scanning electron microscopy (SEM) and transmission electron microscopy
(TEM) for materials characterization, in Materials characterization using nondestructive
evaluation (NDE) methods. 2016, Elsevier. p. 17-43.

142
145. Schwartz, A.J., et al., Electron backscatter diffraction in materials science. Vol. 2. 2009:
Springer.
146. Volkert, C.A. and A.M. Minor, Focused ion beam microscopy and micromachining.
MRS bulletin, 2007. 32(5): p. 389-399.
147. Ernst, A., M. Wei, and M. Aindow, A comparison of Ga FIB and Xe-plasma FIB of
complex Al alloys. Microscopy and Microanalysis, 2017. 23(S1): p. 288-289.
148. Gierak, J., Focused ion beam technology and ultimate applications. Semiconductor
science and technology, 2009. 24(4): p. 043001.
149. Tomus, D. and H.P. Ng, In situ lift-out dedicated techniques using FIB–SEM system for
TEM specimen preparation. Micron, 2013. 44: p. 115-119.
150. Williams, D.B. and C.B. Carter, Transmission Electron Microscopy. 2nd ed. 2009, New
York, New York: Springer.
151. Pharr, G. and W. Oliver, Measurement of thin film mechanical properties using
nanoindentation. Mrs Bulletin, 1992. 17(7): p. 28-33.
152. Kana, J.K., et al., Thermochromic nanocrystalline Au–VO2 composite thin films prepared
by radiofrequency inverted cylindrical magnetron sputtering. Thin Solid Films, 2010.
518(6): p. 1641-1647.
153. Manova, D., J.W. Gerlach, and S. Mändl, Thin film deposition using energetic ions.
Materials, 2010. 3(8): p. 4109-4141.
154. Rossnagel, S. and J. Cuomo, Film modification by low energy ion bombardment during
deposition. Thin Solid Films, 1989. 171(1): p. 143-156.
155. Harper, J.M., et al., Modification of thin film properties by ion bombardment during
deposition. Nucl. Instrum. Meth. B, 1985. 7: p. 886-892.
156. Mbam, S.O., et al., Thin-film coating; historical evolution, conventional deposition
technologies, stress-state micro/nano-level measurement/models and prospects
projection: a critical review. Materials Research Express, 2019. 6(12): p. 122001.
157. Rossnagel, S.M., Magnetron sputtering. J. Vac. Sci. Technol. A, 2020. 38: p. 060805.
158. Motegi, N., et al., Long-throw low-pressure sputtering technology for very large-scale
integrated devices. J. Vac. Sci. Technol. B, 1995. 13: p. 1906-1909.
159. Patel, N.P. and K.V. Chauhan, Effect of sputtering power and substrate temperature on
structural, optical, wettability and anti-icing characteristics of aluminium doped zinc
oxide. Materials Research Express, 2022. 9(7): p. 076402.

143
160. Solis-Pomar, F., et al., Study of the structural properties of nanostructured PbS thin films
deposited by RF sputtering at room temperature. Materials Research Express, 2018.
5(10): p. 106403.
161. Thornton, J. and V. Hedgcoth, Tubular hollow cathode sputtering onto substrates of
complex shape. J. Vac. Sci. Technol. A, 1974. 12(1): p. 93-97.
162. Rane, R., et al., Comparative study of discharge characteristics and associated film
growth for post-cathode and inverted cylindrical magnetron sputtering. Pramana - J.
Phys., 2019. 92(4): p. 1-9.
163. Glocker, D., et al., Inverted Cylindrical Magnetron Sputtering: Advantages of off‐axis
alignment of substrates on the deposition of optical coatings. Vak. Forsch. Prax., 2014.
26(4): p. 18-23.
164. Klawuhn, E., et al., Ionized physical-vapor deposition using a hollow-cathode magnetron
source for advanced metallization. J. Vac. Sci. Technol. A, 2000. 18(4): p. 1546-1549.
165. Koch, H., et al., Hollow cathode discharge sputtering device for uniform large area thin
film deposition. J. Vac. Sci. Technol. A, 1991. 9(4): p. 2374-2377.
166. Poluektov, N., et al., Plasma parameters of the hollow cathode magnetron inside and
downstream. Plasma Sources Sci. T., 2015. 24(3): p. 035009.
167. Abe, S., et al., Comparison of plasma characteristics of high-power pulsed sputtering
glow discharge and hollow-cathode discharge. Jpn. J. of Appl. Phys., 2020. 60(1): p.
015501.
168. Esparza-Contro, C., et al., Microstructures of titanium oxide thin films grown
continuously on stainless steel wires by PVD in an inverted cylindrical magnetron:
Towards an industrial process. Surf. Coat. Tech., 2020. 389: p. 125643.
169. Todoran, A., et al., Control of particle flux and energy on substrate in an inverted
cylindrical magnetron for plasma PVD. Plasma Sources Sci. T., 2014. 23: p. 065039.
170. Kaneko, T. and O. Nittono, Improved design of inverted magnetrons used for deposition
of thin films on wires. Surf. Coat. Tech., 1997. 90(3): p. 268.
171. Garcia-Taormina, A.R., et al., Coatings for Core‐shell Composite Micro‐lattice
Structures: Varying Sputtering Parameters. Adv. Eng. Mater., 2021. 24(4): p. 2101264.
172. Eberhardt, M.G., A.M. Hodge, and P.S. Branicio, Atomistic modeling of physical vapor
deposition on complex topology substrates. Comput. Mater. Sci, 2022. 203: p. 111111.
173. Montemayor, L. and J. Greer, Mechanical response of hollow metallic nanolattices:
combining structural and material size effects. J. Appl. Mech., 2015. 82(7).

144
174. Juarez, T., et al., Evaluating sputter deposited metal coatings on 3D printed polymer
micro-truss structures. Mater. Design, 2018. 140: p. 442-450.
175. Gao, L., et al., High‐entropy alloy (HEA)‐coated nanolattice structures and their
mechanical properties. Adv. Eng. Mater., 2018. 20(1): p. 1700625.
176. Zhang, X., et al., Three-dimensional high-entropy alloy–polymer composite nanolattices
that overcome the strength–recoverability trade-off. Nano Lett., 2018. 18(7): p. 4247-
4256.
177. Velasco, L., M.N. Polyakov, and A.M. Hodge, Influence of stacking fault energy on twin
spacing of Cu and Cu–Al alloys. Scripta Mater, 2014. 83: p. 33-36.
178. Fang, D. and R.K. Marcus, Use of a cylindrical Langmuir probe for the characterization
of charged particle populations in a planar, diode glow discharge device. Spectrochim
Acta B, 1990. 45(9): p. 1053-1074.
179. Navid, A. and A. Hodge, Nanostructured alpha and beta tantalum formation—
Relationship between plasma parameters and microstructure. Materials Science and
Engineering: A, 2012. 536: p. 49-56.
180. Metwaly, K.H. and Y.H. Elbashar, Probe measurements of a plasma hollow cathode
discharge used as a sputter source for copper thin film. J. Opt-Uk, 2022. 51(2): p. 241-
245.
181. Keidar, M. and I. Beilis, Plasma Engineering: Applications from Aerospace to Bio and
Nanotechnology. 2013: Academic Press.
182. Marinov, M., Effect of ion bombardment on the initial stages of thin film growth. Thin
Solid Films, 1977. 46(3): p. 267-274.
183. Michely, T. and G. Comsa, Generation and nucleation of adatoms during ion
bombardment of Pt (111). Phys. Rev. B, 1991. 44(15): p. 8411-8414.
184. Ensinger, W., Low energy ion assist during deposition—an effective tool for controlling
thin film microstructure. Nucl. Instrum. Meth. B, 1997. 127: p. 796-808.
185. Lu, L., et al., Ultrahigh strength and high electrical conductivity in copper. Science,
2004. 304(5669): p. 422-426.
186. Gallagher, P., The influence of alloying, temperature, and related effects on the stacking
fault energy. Metall. Trans., 1970. 1: p. 2429-2461.
187. Smallman, R. and P. Dobson, Stacking fault energy measurement from diffusion. Metall.
Trans., 1970. 1: p. 2383-2389.
188. Ferreira, P. and P. Müllner, A thermodynamic model for the stacking-fault energy. Acta
Mater., 1998. 46(13): p. 4479-4484.

145
189. Castañeda, J.A., et al., Stacking fault energy determination in fe-mn-al-c austenitic steels
by x-ray diffraction. Metals, 2021. 11(11): p. 1701.
190. Arora, G. and D.S. Aidhy, Machine learning enabled prediction of stacking fault
energies in concentrated alloys. Metals, 2020. 10(8): p. 1072.
191. Sun, X., et al., Can experiment determine the stacking fault energy of metastable alloys?
Mater. Des., 2021. 199: p. 109396.
192. Su, Y., S. Xu, and I.J. Beyerlein, Density functional theory calculations of generalized
stacking fault energy surfaces for eight face-centered cubic transition metals. J. Appl.
Phys., 2019. 126(10): p. 105112.
193. Janani, R.D., et al., Effect of composition on the stacking fault energy of copper-nickel
alloys using molecular dynamics simulations. Mater. Today-Proc., 2021. 39: p. 1796-
1800.
194. Bufford, D., H. Wang, and X. Zhang, High strength, epitaxial nanotwinned Ag films.
Acta Mater., 2011. 59(1): p. 93-101.
195. Valentino, G.M., et al., Nanotwin formation in Ni–Mo–W alloys deposited by dc
magnetron sputtering. Scr. Mater., 2020. 186: p. 247-252.
196. Ott, R.T., et al., Optimization of strength and ductility in nanotwinned ultra-fine grained
Ag: Twin density and grain orientations. Acta Mater., 2015. 96: p. 378-389.
197. Wang, Z., et al., High hardness and fatigue resistance of CoCrFeMnNi high entropy
alloy films with ultrahigh-density nanotwins. Int. J. Plasticity, 2020. 131: p. 102726.
198. Zhao, Y., et al., Thermal stability of highly nanotwinned copper: The role of grain
boundaries and texture. J. Mater. Res., 2012. 27(24): p. 3049-3057.
199. Furnish, T. and A. Hodge, On the mechanical performance and deformation of
nanotwinned Ag. APL Mater., 2014. 2(4).
200. Yu, K., et al., Basic criteria for formation of growth twins in high stacking fault energy
metals. Appl. Phys. Lett., 2013. 103(18): p. 181903.
201. Deluigi, O.R., et al., Simulations of primary damage in a High Entropy Alloy: Probing
enhanced radiation resistance. Acta Mater., 2021. 213: p. 116951.
202. Gao, T., et al., Molecular dynamics simulations of tensile response for FeNiCrCoCu
high-entropy alloy with voids. Int. J. Mech. Sci., 2023. 237: p. 107800.
203. Xie, H., et al., Amorphization transformation in high-entropy alloy FeNiCrCoCu under
shock compression. J. Mat. Sci. Technol., 2024. 175: p. 72-79.

146
204. Doan, D.-Q., Effects of crystal orientation and twin boundary distance on mechanical
properties of FeNiCrCoCu high-entropy alloy under nanoindentation. Mater. Chem.
Phys., 2022. 291: p. 126725.
205. Thompson, A.P., et al., LAMMPS-a flexible simulation tool for particle-based materials
modeling at the atomic, meso, and continuum scales. Comput. Phys. Commun., 2022.
271: p. 108171.
206. Stukowski, A., Visualization and analysis of atomistic simulation data with OVITO–the
Open Visualization Tool. Model. Simul. Mat. Sci. Eng., 2009. 18(1): p. 015012.
207. Wang, Z., D. Han, and X. Li, Competitive effect of stacking fault energy and short-range
clustering on the plastic deformation behavior of Cu-Ni alloys. Mater. Sci. Eng. A, 2017.
679: p. 484-492.
208. Carter, C. and S. Holmes, The stacking-fault energy of nickel. Philos. Mag., 1977. 35(5):
p. 1161-1172.
209. Massalski, T., et al., Binary alloy phase diagrams. Vol. 2. 1990, Materials Park, OH:
ASM International.
210. Soboyejo, W., Mechanical properties of engineered materials. 2002: CRC press.
211. Weertman, J., Hall-Petch strengthening in nanocrystalline metals. Mater. Sci. Eng. A,
1993. 166(1-2): p. 161-167.
212. Lee, B.-J., J.-H. Shim, and M. Baskes, Semiempirical atomic potentials for the fcc metals
Cu, Ag, Au, Ni, Pd, Pt, Al, and Pb based on first and second nearest-neighbor modified
embedded atom method. Phys. Rev. B, 2003. 68(14): p. 144112.
213. Udler, D. and D. Seidman, Grain boundary and surface energies of fcc metals. Phys.
Rev. B, 1996. 54(16): p. R11133.
214. Hansen, N., Hall–Petch relation and boundary strengthening. Scr. Mater., 2004. 51(8): p.
801-806.
215. Zhang, X. and A. Misra, Superior thermal stability of coherent twin boundaries in
nanotwinned metals. Scripta Materialia, 2012. 66(11): p. 860-865.
216. Bahena, J.A., et al., Grain boundary evolution of highly nanotwinned alloys: Effect of
initial twinned microstructure. Scripta Materialia, 2021. 190: p. 27-31.
217. Bufford, D., H. Wang, and X. Zhang, Thermal stability of twins and strengthening
mechanisms in differently oriented epitaxial nanotwinned Ag films. Journal of Materials
Research, 2013. 28(13): p. 1729-1739.

147
218. Duan, F., et al., Hardness-thermal stability synergy in nanograined Ni and Ni alloys:
Superposition of nanotwin and low-energy columnar boundary. Journal of Materials
Science & Technology, 2023. 137: p. 123-131.
219. Andrievski, R., Review of thermal stability of nanomaterials. Journal of materials
science, 2014. 49: p. 1449-1460.
220. Anderoglu, O., et al., Thermal stability of sputtered Cu films with nanoscale growth
twins. Journal of Applied Physics, 2008. 103(9).
221. Lu, W., et al., Interfacial nanophases stabilize nanotwins in high-entropy alloys. Acta
Materialia, 2020. 185: p. 218-232.
222. Sikdar, K., et al., Enhanced thermal stability of nanocrystalline Cu-Al alloy by nanotwin
and nanoprecipitate. Journal of Alloys and Compounds, 2022. 922: p. 166273.
223. He, M.-R., et al., In situ TEM study of the thermal stability of nanotwinned Ni-Mo-W
alloys. Materials Research Letters, 2023. 11(10): p. 879-887.
224. Kurz, S., A. Leineweber, and E.-J. Mittemeijer, Anomalously high density and thermal
stability of nanotwins in Ni (W) thin films: Quantitative analysis by x-ray diffraction.
Journal of Materials Research, 2014. 29(15): p. 1642-1655.
225. Sun, L., X. He, and J. Lu, Nanotwinned and hierarchical nanotwinned metals: A review
of experimental, computational and theoretical efforts. npj Computational Materials,
2018. 4(1): p. 6.
226. Alwen, A., et al., Combinatorial and high-throughput investigation of growth nanotwin
formation. Acta Materialia, 2024: p. 119839.
227. Darling, K., et al., Thermal stability of nanocrystalline Fe–Zr alloys. Materials Science
and Engineering: A, 2010. 527(15): p. 3572-3580.
228. Wang, W., et al., Thermodynamic constitution of the Al–Cu–Ni system modeled by
CALPHAD and ab initio methodology for designing high entropy alloys. Calphad, 2019.
65: p. 346-369.
229. Cullity, B.D., Elements of X-Ray Diffraction. Second ed. 1978, Boston: Addison-Wesley
Publishing Company.
230. Bahena, J.A., et al., Development of a heterogeneous nanostructure through abnormal
recrystallization of a nanotwinned Ni superalloy. Acta Materialia, 2020. 195: p. 132-140.
231. Goodelman, D.C. and A.M. Hodge, Distribution of nanodomains in heterogeneous Nisuperalloys: Effect on microstructure and mechanical deformation. Acta Materialia,
2023. 252: p. 118940.

148
232. Velasco, L. and A.M. Hodge, Growth twins in high stacking fault energy metals:
Microstructure, texture and twinning. Materials Science and Engineering: A, 2017. 687:
p. 93-98.
233. Alwen, A., et al., Combinatorial CuNiAl As-Sputtered and Annealed XRD Data. 2024:
Materials Data Facility.
234. Cantor, B. and R. Cahn, Metastable alloy phases by co-sputtering. Acta Metallurgica,
1976. 24(9): p. 845-852.
235. Noebe, R.D., R.R. Bowman, and M.V. Nathal, Physical and mechanical properties of the
B2 compound NiAl. International Materials Reviews, 1993. 38(4): p. 193-232.
236. Camagu, S.T., et al., Investigation into the thermal behaviour of the B2–NiAl
intermetallic alloy produced by compaction and sintering of the elemental Ni and Al
powders. Vacuum, 2019. 169: p. 108919.
237. Pretorius, R., et al., Compound phase formation in thin film structures. Critical reviews in
solid state and materials sciences, 1999. 24(1): p. 1-62.
238. Emigh, M.G., et al., Influence of a nanotwinned, nanocrystalline microstructure on aging
of a Ni-25Mo-8Cr superalloy. Acta Materialia, 2018. 156: p. 411-419.
239. Tang, Y., et al., Grain size effect on precipitation behavior of nanostructured Inconel
718. Journal of Materials Science & Technology, 2024.
240. Shen, Q., et al., Effect of Cu content on the precipitation behavior of Cu-rich and NiAl
phases in steel. Materials Characterization, 2022. 187: p. 111849.
241. Yang, X., et al., The co-precipitation evolution of NiAl and Cu nanoparticles and its
influence on strengthening and toughening mechanisms in low-carbon ultra-high
strength martensite seamless tube steel. International Journal of Plasticity, 2023. 166: p.
103654.
242. Nash, P. and O. Kleppa, Composition dependence of the enthalpies of formation of NiAl.
Journal of Alloys and compounds, 2001. 321(2): p. 228-231.
243. Wu, Q., et al., Study on behavior of NiAl coating with different Ni/Al ratios. Vacuum,
2013. 93: p. 37-44.
244. Ellis, E.A., et al., Driving forces for texture transformation in thin Ag films. Acta
Materialia, 2016. 105: p. 495-504.
245. Oikawa, H. and S. Karashima, On the self-diffusion coefficients of aluminum in copper
(rich)-aluminum solid solutions. Transactions of the Japan Institute of Metals, 1970.
11(6): p. 431-433.

149
246. Frank, S., et al., Ni tracer diffusion in the B2-compound NiAl: influence of temperature
and composition. Acta materialia, 2001. 49(8): p. 1399-1411.

150
Appendix A: Summary of Sputtered Samples
This appendix contains tables of the sputtered films created during the course of this dissertation
along with corresponding information on the synthesis conditions and notes of the
characterization techniques employed.
Legend
SEM = Scanning Electron Microscopy, EDS = Energy Dispersive X-Ray Spectroscopy,
XRD= X-Ray Diffraction, FPP = Four-Point Probe, TEM = Transmission Electron Microscopy,
NI = Nanoindentation, HT = Heat Treatment, CHT = Combinatorial and High-Throughput,
RS = Residual Stress, MT = Microtome, FZJ = Forschungszentrum Julich (Germany),
KIT = Karlsruhe Institute of Technology
Planar Cathode (PLNR)
Sample Name
(Composition)
Working
Pressure
(mTorr)
Sputtering
Power
(W)
Target
Diameter
(in)
Working
Distance
(in)
Deposition
Rate
(nm/s)
Substrate Thickness
(nm) Notes
PLNR 2021 –
CuAl – 001
(Cu 2 at% Al)
3 18 3 1.875 0.29 Si (100)
1” 1000
XRD,
SEM,
EBSD,
RS
PLNR 2021 –
CuAl – 002
(Cu 2 at% Al)
6 18 3 1.875 0.28
Si (100)
1” 1000
XRD,
SEM,
RS
PLNR 2021 –
CuAl – 003
(Cu 2 at% Al)
3 55 3 1.875 0.81 Si (100)
1” 1000
XRD,
SEM,
RS
PLNR 2021 –
CuAl – 004
(Cu 2 at% Al)
6 55 6 1.875 0.72 Si (100)
1” 1000
XRD,
SEM,
RS

151
PLNR 2021 –
CuAl – 005
(Cu 2 at% Al)
3 55 3 1.875 0.81 Si (100)
1” 1000
XRD,
SEM,
EBSD,
RS
Hollow Cathode (HC)
Sample Name
(Composition)
Working
Pressure
(mTorr)
Sputtering
Power
(W)
Target
Size
(in)
Working
Distance
(in)
Deposition
Rate
(nm/s)
Substrate Thickness
(nm) Notes
HC 2021 –
CuAl – 001
(Cu 2 at% Al)
Aligned Parallel
to HC
3 200
Radius:
1.875
Length:
6.500
1.875 0.58 Si (100)
1” 1000
XRD,
SEM,
RS,
HC 2021 –
CuAl – 002
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
3 200
Radius:
1.875
Length:
6.500
1.875 1.10 Si (100)
1” 1000
XRD,
SEM,
RS,
NI
HC 2021 –
CuAl – 003
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
6 200
Radius:
1.875
Length:
6.500
1.875 0.77 Si (100)
1” 1000
XRD,
SEM,
RS,
NI
HC 2021 –
CuAl – 004
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
30 200
Radius:
1.875
Length:
6.500
1.875 1.02 Si (100)
1” 1000
XRD,
SEM,
RS,
NI,
EBSD
HC 2021 –
CuAl – 005
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
12 200
Radius:
1.875
Length:
6.500
1.875 0.86 Si (100)
1” 1000
XRD,
SEM,
RS,
NI
HC 2021 –
CuAl – 006
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
3 600
Radius:
1.875
Length:
6.500
1.875 2.25 Si (100)
1” 1000
XRD,
SEM,
RS,
NI,
EBSD

152
HC 2021 –
CuAl – 007
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
6 600
Radius:
1.875
Length:
6.500
1.875 2.12 Si (100)
1” 1000
XRD,
SEM,
RS,
NI,
EBSD
HC 2021 –
CuAl – 008
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
12 600
Radius:
1.875
Length:
6.500
1.875 2.61 Si (100)
1” 1000
XRD,
SEM,
RS,
NI,
EBSD
HC 2021 –
CuAl – 009
(Cu 2 at% Al)
Aligned
Perpendicular
to HC
30 600
Radius:
1.875
Length:
6.500
1.875 3.18 Si (100)
1” 1000
XRD,
SEM,
RS,
NI,
EBSD
HC 2021 –
CuAl – 010
(Cu 2 at% Al)
Aligned Parallel
to HC
3 700
Radius:
1.875
Length:
6.500
1.875 3.04
Corning
Eagle
Glass
1”
1000
XRD
Rotated
1.1
RPM
HC 2021 –
Cu6Al – 001
(Cu 6 at% Al)
Aligned Parallel
to HC
3 700
Radius:
1.875
Length:
6.500
1.875 2.80
Corning
Eagle
Glass
1”
1000
XRD
Rotated
1.1
RPM
HC 2021 –
Cu6Al – 002
(Cu 6 at% Al)
Aligned Parallel
to HC
3 700
Radius:
1.875
Length:
6.500
1.875 2.80 Si (100)
1” 1000
XRD
Rotated
1.1
RPM
HC Deposition on IP-DIP Tetrahedral 3D Printed Lattices
HC IP-Dip –
Cu2Al – 001
(Cu 2 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 3.04
5𝜇m and 10
𝜇m strut
length
lattices
200
SEM
Rotated
1.1
RPM
HC IP-Dip –
Cu2Al – 002
(Cu 2 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 3.04
5𝜇m and 10
𝜇m strut
length
lattices
200
SEM
Rotated
1.1
RPM

153
HC IP-Dip –
Cu2Al – 003
(Cu 2 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 3.04
5𝜇m and 10
𝜇m strut
length
lattices
200 SEM
HC IP-Dip –
Cu6Al – 001
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
5𝜇m and 10
𝜇m strut
length
lattices
200
SEM
Rotated
1.1
RPM
HC IP-Dip –
Cu6Al – 002
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
5𝜇m and 10
𝜇m strut
length
lattices
1000
SEM
Rotated
10 RPM
HC IP-Dip –
Cu6Al – 003
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
5𝜇m and 10
𝜇m strut
length
lattices
1000
SEM,
MT
Rotated
10 RPM
HC Deposition on IP-DIP Four Geometry 3D Printed Lattices
(Tetrahedral, Simple Cubic, Braced Cubic, Hexagonal)
HC IP-Dip – 4
Geometries
Cu6Al – 001
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
Four
Lattice
Geometries
1000
SEM
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Cu6Al – 002
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
Four
Lattice
Geometries
500
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Cu6Al – 003
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
Four
Lattice
Geometries
1000
Rotated
10 RPM
At USC
HC IP-Dip – 4
Geometries
Cu6Al – 004
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
Four
Lattice
Geometries
1500
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Cu6Al – 005
(Cu 6 at% Al)
3 700
Radius:
1.875
Length:
6.500
1.875 2.81
Four
Lattice
Geometries
1500
Rotated
10 RPM
At USC
HC IP-Dip – 4
Geometries
Ti6Al4V – 001
3 700
Radius:
1.875
Length:
6.500
1.875 0.86
Four
Lattice
Geometries
500
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Ti6Al4V – 002
3 700
Radius:
1.875
Length:
6.500
1.875 0.86
Four
Lattice
Geometries
500
Rotated
10 RPM
At USC

154
HC IP-Dip – 4
Geometries
Ti6Al4V – 003
3 700
Radius:
1.875
Length:
6.500
1.875 0.86
Four
Lattice
Geometries
1000
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Ti6Al4V – 004
3 700
Radius:
1.875
Length:
6.500
1.875 0.86
Four
Lattice
Geometries
1000
Rotated
10 RPM
At USC
HC IP-Dip – 4
Geometries
Ti6Al4V – 005
3 700
Radius:
1.875
Length:
6.500
1.875 0.86
Four
Lattice
Geometries
1500
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Ti6Al4V – 006
3 700
Radius:
1.875
Length:
6.500
1.875 0.86
Four
Lattice
Geometries
1500
Rotated
10 RPM
At USC
HC IP-Dip – 4
Geometries
Al 6061 – 001
3 700
Radius:
1.875
Length:
6.500
1.875 2.28
Four
Lattice
Geometries
500
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Al 6061 – 002
3 700
Radius:
1.875
Length:
6.500
1.875 2.28
Four
Lattice
Geometries
500
Rotated
10 RPM
At USC
HC IP-Dip – 4
Geometries
Al 6061 – 003
3 700
Radius:
1.875
Length:
6.500
1.875 2.28
Four
Lattice
Geometries
1000
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Al 6061 – 004
3 700
Radius:
1.875
Length:
6.500
1.875 2.28
Four
Lattice
Geometries
1000
Rotated
10 RPM
At USC
HC IP-Dip – 4
Geometries
Al 6061 – 005
3 700
Radius:
1.875
Length:
6.500
1.875 2.28
Four
Lattice
Geometries
1500
Rotated
10 RPM
Sent to
KIT
HC IP-Dip – 4
Geometries
Al 6061 – 006
3 700
Radius:
1.875
Length:
6.500
1.875 2.28
Four
Lattice
Geometries
1500
Rotated
10 RPM
At USC
Combinatorial and High-Throughput Samples (CHT)
CuAl 2022 001
(CHT Gradient) 5
Al: 470
Cu: 500
Al: 2
Cu: 2 5.511 1.60 Si (100)
4”
1000 SEM,
EDS
CuAl 2022 002
(CHT Gradient) 5
Al: 470
Cu: 500
Al: 2
Cu: 2 5.511 1.60 Si (100)
4”
1000 SEM,
EDS

155
CuAl 2022 003
(CHT Gradient)
5
Al: 155
Cu: 500
Al: 2
Cu: 2 5.511 1.20 Si (100) 4”
2000
SEM,
EDS,
XRD,
TEM
CuNi 2022 001
(CHT Gradient)
5
Ni: 155
Cu: 500
Ni: 2
Cu: 2 5.511 1.20 Si (100) 4”
1000 SEM,
EDS,
CuNi 2022 002
(CHT Gradient)
5
Ni: 155
Cu: 500
Ni: 2
Cu: 2 5.511 1.20 Si (100) 4”
2000
SEM,
EDS,
XRD,
FPP,
TEM,
NI
CuNi 2022 003
(CHT Gradient)
5
Ni: 300
Cu: 350
Ni: 2
Cu: 2 5.511 1.20 Si (100) 4”
2000
SEM,
EDS,
XRD,
FPP,
TEM,
NI
CuNiAl 2022
001
(CHT Gradient)
5
Ni: 105
Cu: 415
Al: 105
Ni: 2
Cu: 2
Al: 2
5.511 1.20 Si (100) 4”
2000
SEM,
EDS,
XRD,
FPP,
TEM,
CuNiAl 2022
002
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20 Si (100) 4”
2000 EDS
CuNiAl 003
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20
High
Temp.
Quartz
2000 SEM,
EDS
CuNiAl 004
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20
High
Temp.
Quartz
2000 SEM,
EDS
CuNiAl 005
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20
High
Temp.
Quartz
1000 SEM,
EDS
CuNiAl 006
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20
High
Temp.
Quartz
1000
SEM,
EDS,
Heated
Stage
(100°C)
HT
(500°C
,
3 hours)

156
CuNiAl 007
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20
High
Temp.
Quartz
1000
SEM,
XRD,
FPP,
TEM,
NI
Heated
Stage
(100°C)
HT
(400°C
,
3 hours)
CuNiAl 008
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20
High
Temp.
Quartz
1000
SEM,
XRD,
FPP,
TEM,
NI
Heated
Stage
(100°C)
CuNiAl 009
(CHT Gradient)
5
Ni: 225
Cu: 275
Al: 250
Ni: 2
Cu: 2
Al: 2
5.511 1.20 Si (100) 4”
1000
EDS
Heated
Stage
(100°C)
NiAl Single
Layer 001
(CHT Gradient)
5
Ni: 105
Al: 105
Ni: 2
Al: 2 5.511 0.22
High
Temp.
Quartz
2000
None,
Rough
Surface
NiAl Single
Layer 002
(CHT Gradient)
5
Ni: 220
Al: 300
Ni: 2
Al: 2 5.511 0.66
High
Temp.
Quartz
2000
Heated
Stage
(100°C)
Sent to
FZJ
NiAl Single
Layer 003
(CHT Gradient)
5
Ni: 220
Al: 300
Ni: 2
Al: 2 5.511 0.66 Si (100) 4”
2000
Heated
Stage
(100°C)
Sent to
FZJ
NiAl/Al Bilayer
001
(CHT Gradient)
5
Ni: 105
Al: 105
Al: 105
Ni: 2
Al: 2
Al: 2
5.511 NiAl: 0.22
Al: 0.27
High
Temp.
Quartz
2000
None,
Rough
Surface
NiAl/Al Bilayer
002
(CHT Gradient)
5
Ni: 220
Al: 300
Al: 300
Ni: 2
Al: 2
Al: 2
5.511 NiAl: 0.66
Al: 0.74
High
Temp.
Quartz
2000
Heated
Stage
(100°C)
Sent to
FZJ
NiAl/Al
Multilayer 001
100 nm Layers
(CHT Gradient)
5
Ni: 220
Al: 300
Al: 300
Ni: 2
Al: 2
Al: 2
5.511 NiAl: 0.66
Al: 0.74
High
Temp.
Quartz
2000
Heated
Stage
(100°C)
Sent to
FZJ

157
Appendix B: High-Throughput Four-Point Probe Data
Figure B1: Sheet resistance heatmaps for the two combinatorial arrays of CuNi samples discussed
in Chapter 5 and analyzed in Figure 37. The regions in dark red have lower sheet resistivities and
the regions in bright white have higher sheet resistivities.
Figure B2: Sheet resistance heatmaps for the as-sputtered (a) and annealed (b) CuNiAl
combinatorial arrays discussed in Chapter 6 and analyzed in Figure 41. The regions in dark red
have lower sheet resistivities and the regions in bright white have higher sheet resistivities. The
missing points in the annealed map are due to samples that have de-adhered or frayed due to the
annealing process.

158
Appendix C: Validation of Growth Twinning Model
Figure C1: Comparison of measured and predicted twin boundary spacings as a function of
stacking fault energy (SFE) for the CuNi alloys examined in Chapter 6 and for other materials
studied in literature. The black data points represent the measured twin spacings, while the
predicted values for the Zhang model and Alwen model are shown by the red and blue points,
respectively [79, 177, 196].

Abstract (if available)

Abstract The development of future technologies requires the discovery of new materials by exploring and optimizing unique microstructural and morphological combinations within a vast compositional space. However, navigating the nearly unlimited number of combinations to find promising material systems is a major challenge. Thus, more efficient experimental techniques and approaches are needed to expedite discovery when investigating unexplored synthesis domains. This work seeks to accelerate material discovery by globally analyzing large material synthesis spaces as a function of both changing processing conditions and composition using magnetron sputtering. To analyze the role of varied processing conditions, film deposition in the planar and hollow cathode was compared to examine how changes in deposition parameters influenced film growth, microstructure, and morphology. With respect to composition, large compositional domains were rapidly studied for changes in nanostructure formation using combinatorial co-sputtering and high- and low-throughput characterization techniques. Overall, this work highlights novel approaches to study composition-processing-property relationships in large synthesis spaces, laying a foundation for accelerated material discovery.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Tailoring compositional and microstructural complexity in nanostructured alloys

PDF

A comparative study of plasma conditions, microstructure and residual stress in sputtered thin films

PDF

Expanding the synthesis space of 3D nano- and micro-architected lattice materials

PDF

Corrosion studies of engineered interfaces

PDF

Understanding the formation and evolution of boundaries and interfaces in nanostructured metallic alloys

PDF

Water desalination: real-time membrane characterization for performance prediction and system analysis for energetic enhancement

PDF

Development and characterization of hierarchical cellular structures

PDF

Microstructural evolution and mechanical properties of a copper-zirconium alloy processed by severe plastic deformation

PDF

Mechanical behavior and deformation of nanotwinned metallic alloys

PDF

Mechanical behavior and microstructure optimization of ultrafine-grained aluminum alloys and nanocomposites

PDF

Inorganic/organic hybrid materials and grafted coatings via vapor phase deposition processes

PDF

Development of organic-inorganic optical microcavities for studying polymer thin films

PDF

Synthesis and mechanical behavior of highly nanotwinned metals

PDF

Modeling of bicontinuous nanoporous materials and investigation of their mechanical and hydraulic properties

PDF

A comprehensive study of twinning phenomena in low and high stacking fault energy metals

PDF

Tailoring asymmetry in light-matter interaction with atomic-scale lattice modulations

PDF

Real time surface analysis of complex oxide thin films during pulsed laser deposition

PDF

Development and characterization of transparent metal/ceramic and ceramic/ceramic nanomultilayers

PDF

Morphological and microstructural models of typical development in the hippocampus and white matter

PDF

Investigation of the influence of severe plastic deformation on the microstructure and mechanical properties of Al-7136 alloy

Asset Metadata

Creator Alwen, Adie Randall (author)

Core Title Expanding sputtering synthesis domains for the discovery of novel film microstructures and morphologies

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Materials Science

Degree Conferral Date 2024-08

Publication Date 06/20/2024

Defense Date 06/06/2024

Publisher Los Angeles, California (original), University of Southern California (original), University of Southern California. Libraries (digital)

Tag combinatorial,high-throughput,microstructure,morphology,nanotwins,OAI-PMH Harvest,sputtering

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Hodge, Andrea (committee chair), Brutchey, Richard (committee member), Shao, Yu-Tsun (committee member), Villalobos, Luis (committee member)

Creator Email adierandallalwen@gmail.com,alwen@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113996WT2

Unique identifier UC113996WT2

Identifier etd-AlwenAdieR-13123.pdf (filename)

Legacy Identifier etd-AlwenAdieR-13123

Document Type Dissertation

Format theses (aat)

Rights Alwen, Adie Randall

Internet Media Type application/pdf

Type texts

Source 20240620-usctheses-batch-1172 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu