Reactive High Power Impulse Magnetron Sputtering of Zinc

Home , Sputter deposition

REACTIVE HIGH POWER IMPULSE MAGNETRON SPUTTERING OF ZINC

OXIDE FOR THIN FILM TRANSISTOR APPLICATIONS

Dissertation

Submitted to

The School of Engineering of the

UNIVERSITY OF DAYTON

In Partial Fulfillment of the Requirements for

The Degree of

Doctor of Philosophy in Materials Engineering

Amber Nicole Reed

Dayton, Ohio

May 2015

REACTIVE HIGH POWER IMPULSE MAGNETRON SPUTTERING OF ZINC

OXIDE FOR THIN FILM TRANSISTOR APPLICATIONS

Name: Reed, Amber Nicole

APPROVED BY:

______Dr. Andrey A. Voevodin Dr. Charles Browning Advisory Committee Chairman Committee Member Adjunct Professor Department Chair Department of Chemical and Department of Chemical and Materials Engineering Materials Engineering

______Dr. Kevin Leedy Committee Member AFRL/RYDD

______Dr. Christopher Muratore Dr. Terry Murray Committee Member Committee Member Professor Professor Department of Chemical and Department of Chemical and Materials Engineering Materials Engineering

______John G. Weber, Ph.D. Eddy M. Rojas, Ph.D., M.A., P.E. Associate Dean Dean, School of Engineering School of Engineering

Amber Nicole Reed

2015

iii

ABSTRACT

REACTIVE HIGH POWER IMPULSE MAGNETRON SPUTTERING OF ZINC

OXIDE FOR THIN FILM TRANSISTOR APPLICATIONS

Name: Reed, Amber Nicole University of Dayton

Advisor: Dr. Andrey A. Voevodin

Zinc oxide (ZnO) is an emerging thin film transistor (TFT) material for transparent flexible displays and sensor technologies, where low temperature synthesis of highly crystallographically ordered films over large areas is critically needed. This study maps plasma assisted synthesis characteristics, establishes polycrystalline ZnO growth mechanisms and demonstrates for the first time low-temperature and scalable deposition of semiconducting grade ZnO channels for TFT applications using reactive high power impulse magnetron sputtering (HiPIMS). Plasma parameters, including target currents, ion species and their energies were measured at the substrate surface location with mass spectroscopy as a function of pressure and applied voltage during HiPIMS of Zn and ZnO targets in O2/Ar. The results were correlated to film microstructure development investigated with x-ray diffraction, atomic force microscopy, scanning electron microscopy and transmission electron microscopy which helped establish film nucleation

iv and growth mechanisms. Competition for nucleation by (100), (101) and (002) oriented crystallites was identified at the early stages of film growth, which can result in a layer of mixed crystal orientation at the substrate interface, a microstructural feature that is detrimental to TFT performance due to increased charge carrier scattering in back-gated

TFT devices. The study revealed that nucleation of both (100) and (101) orientations can be suppressed by increasing the plasma density while decreasing ion energy. After the initial nucleation layer, the microstructure evolves to strongly textured with the (002) crystal plane oriented parallel to the substrate surface. The degree of (002) alignment was pressure-dependent with lower deposition pressures resulting in films with (002)

o alignment less than 3.3 , a trend attributed to less energy attenuation of the low energy (2-

+ + + 6 eV) Ar , O , and O2 ions observed with mass spectrometry measurements. At

+ + + pressures of 7 mTorr and lower, a second population of ionized gas (Ar , O , and O2 ) species with energies up to 50 eV appeared. The presence of higher energy ions corresponded with a bimodal distribution of ZnO grain sizes, confirming that high energy bombardment has significant implications on microstructural uniformity during large area growth. Based on the established correlations between process parameters, plasma characteristics, film structure and growth mechanisms, optimum deposition conditions for

(002) oriented nanocrystalline ZnO synthesis at 150 ⁰C were identified and demonstrated for both silicon oxide wafers of up to 4 inch diameter and on flexible polymer (Kapton) substrates. The feasibility of the low temperature processing of ZnO films for TFT applications was verified by preliminary tests with back-gated device prototypes.

Directions of future research are outlined to further develop this low temperature growth method and apply results of this study for ZnO applications in semiconductor devices.

Dedicated to my dad.

ACKNOWLEDGEMENTS

I would like to thank C. Muratore, A.A. Voevodin and P.J. Shamberger for their guidance. I would also like to thank everyone who provided moral and technical support.

In particular, I would like to thank P.J. Shamberger and R.D. Naguy for the AFM imaging, J.J. Hu for the TEM analysis, A.J. Safriet and J.E. Bultman for design and fabrication work on many of the systems used in this study, J.E. Bultman for providing the SolidWorks schematics of the deposition chamber, and EQP, J.L. Jespersen and J.

Grant for assistance with the analysis of the XPS data. I’d also like to thank K.D. Leedy and G.H. Jessen for providing the PLD-grown ZnO films and technical discussion on

TFT devices. I’d also like to thank J.G. Jones, M.N. Pike, S.D. Apt, A.M. Campo, C.N.

Hunter, M. E. McConney, N.R. Glavin, A.R. Waite, and M.H. Check and the whole

Advanced Nanoscale Electronic Materials and Processing research team for various other help and support.

I am profoundly grateful for the support of my family. Additionally, I am thankful for my dogs, Icky, Ooey, Ollie and Lucy, who could always make me smile.

vii

TABLE OF CONTENTS

ABSTRACT ……………………………………………………………………………..iv

DEDICATION……………………………………………………………………………vi

ACKNOWLEDGEMENTS ……………………………………………………………..vii

LIST OF FIGURES ……………………………………………………………………...xi

LIST OF TABLES ……………………………………………………………………xxiii

LIST OF SYMBOLS/ABBREVIATIONS …………………………………...…….....xxiv

CHAPTER I: INTRODUCTION …………………………………………….…………...1

CHAPTER II: BACKGROUND………………………………………………………….4

2.1 Zinc Oxide as a Thin Film Transistor Channel……………………………..…4

2.2 Film Nucleation and Growth………………………………………………...11

2.3 Magnetron Sputtering and the Sputtering Discharge………………………...16

2.4 High Power Impulse Magnetron Sputtering…………………………………23

2.4.1 Plasma Dynamics during the HiPIMS Process…………………....24

2.4.2 Film Growth during HiPIMS…………………………...…………39

CHAPTER III: RESEARCH OBJECTIVES………………...…………………………..45

CHAPTER IV: EXPERIMENTAL METHODS……………………………...…………48

4.1 The Deposition System…………………………………..…………………..48

4.2 Process Characterization and Plasma Diagnostics……..…………………….50

4.2.1 Sputtering Yield Calculations……………………………………...50

viii

4.2.2 Energy-Resolved Mass Spectroscopy……………………………...52

4.2.3 Target Current Measurements………………………...……………54

4.3 Substrate Selection and Preparation……………………………..…………...54

4.4 Zinc Oxide Deposition…………………………………………..…………...55

4.5 Microstructural Characterization……………………………..……………...59

4.5.1 Thickness Measurements…………………...……………………...59

4.5.2 X-Ray Diffraction Analysis………………………………………..59

4.5.3 Electron Microscopy………………………………………...……..61

4.5.4 Atomic Force Microscopy………...……………………………….61

4.6 Electrical Characterization………..………………………………………….62

4.6.1 Device Fabrication…………………………………………...…….62

4.6.2 Preliminary Back-Gated TFT Device Testing ………...……….….63

CHAPTER V: RESULTS AND DISCUSSION…………………………………………64

5.1 Process and Plasma Characterization…………………..…………………….64

5.1.1 Reactive HiPIMS from a Zinc Target……………………...………64

5.1.2 Reactive HiPIMS from a Zinc Oxide Target………...…………….77

5.2 Film Growth and Microstructure…..………………………………………...89

5.2.1 Film Growth from a Zinc Target………………………..…………89

5.2.2 Film Growth from a Zinc Oxide Target ………...………………....93

5.2.3 Discussion of Film Growth Mechanisms and Plasma Conditions..104

5.2.4 Summary for Film Growth Study………………………………...107

5.3 Effect of Spatial Inhomogeneity of the Depositing Flux on Film

Microstructure...... 110

5.4 Electrical Characterization of Zinc Oxide TFT Channels………………….112

CHAPTER VI: CONCLUSIONS……...……………………………………………….115

CHAPTER VII: FUTURE WORK…………………………………….……………….119

REFERENCES………………………………………………………………………....121

APPENDICES………………………………………………………………………….134

A. Supplemental Ion Energy Distribution Data………………………………..134

B. Supplemental Microstructural Data………………………………………...136

C. Supplemental TFT Device Characterization Data………………………....145

D. X-Ray Photoelectron Spectroscopy………………………………………...156

LIST OF FIGURES

2.1 Schematic drawing of (a) ZnO FET on Si substrate for DC characterization and (b)

of microwave ZnO FET on PECVD SiO2/GaAs substrate. (c) Micrograph of the

microwave transistor. Reproduced with permission from [15]...... 5

2.2 Cross-sectional TEM of a PLD-deposited ZnO film over the TFT gate. Reproduced

with permission from [16]. (top left). Schematic diagram of TFT with a ZnO

channel (bottom left). Close up of the TFT channel showing the depletion zones at

the grain boundaries in the off-state (top right) and the collapse of the depletion

zones under the gate field in the on-state (bottom right)...... 6

2.3 Summary of FWHM of (002) rocking curve XRD peaks at different substrate

temperatures for ZnO films deposited with pulsed DC magnetron sputtering [52],

PLD[34-36], and RF sputtering [46,47]...... 8

2.4 AFM micrograph of surface of ZnO layer deposited using (a) bipolar pulsed DC

magnetron sputtering and (b) HiPIMS on glass substrates. (c) XRD of the ZnO

films in (a) and (b). Reproduced with permission from [8]...... 10

2.5 (a) SEM and (b) XRD of ZnO film deposited with HiPIMS on a sapphire substrate.

Reproduced with permission from [48]...... 11

2.6 Schematic of atomic processes in film nucleation and growth. Adapted from [58]. 12

2.7 Schematic diagram of the three modes of film nucleation and growth. Adapted from

[61]...... 13

2.8 Schematic of representation of dependence of film structure on argon pressure and

substrate temperature. Reproduced with permission from [62]...... 16

2.9 (a) Schematic of a DC powered glow discharge. (b) Voltages at different regions

between the cathode and anode in the glow discharge assuming the anode is held at

0 V. Adapted from [63]...... 18

2.10 (a) A typical magnetron gun used for sputtering. (b) Schematic of the electric and

magnetic fields for a magnetron gun with the particle velocity, plasma and plasma

sheath identified. (insert) The electron drift about a target and an image of a target

racetrack...... 19

2.11 Interactions of ions with the target surface. Adapted from [64]...... 20

2.12 Plots of Vp versus axial distance z from the cathode for (a) the ‘first peak’, (b) the

‘second peak’ and (c) the ‘off-phase’. (d) Plot of the time evolution of typical

plasma potential (Vp) measures at a set axial distance (z = 100 mm) from the target

surface and radial position(r= 45). Plots with ■ are for an average discharge power

of 650 W and gas pressure of 0.54 Pa, □ are for 650 W and 1.08 Pa and are for

950 W and 1.08 Pa. Reproduced with permission from [70]...... 26

2.13 (Top left) Schematic illustrating interactions of ions with the target surface.

Adapted from [64]. (Bottom left) OES spectra above and below self-sputtering

runaway. (Right) Illustration of gas-sputtering and self-sputtering. Reproduced

with permission from [78]...... 28

xii

2.14 Time-resolved target currents for (a) tungsten, (b) aluminum and (c) titanium with

different applied target voltages. Reproduced with permission from [76]...... 31

2.15 Pav = 4kW pulses, before and after the metallic to oxide transition for reactive

HiPIMS of TiO2. (a) Current pulse profile for Ipeak = 160 A; (b) Current pulse

profile for Ipeak = 400 A; (c) Voltage pulse profile for Ipeak = 400 A. Reproduced

with permission from [103]...... 33

2.16 Discharge current waveforms in the metal (Ar) and oxide mode of HiPIMS (Ar -

O2) for (a) Ti and (b) Al targets. Pulse frequency was 50 Hz, pulse length was 300

s and the average discharge power was varied between 100 and 400 V.

Reproduced with permission from [105]...... 34

2.17 Time averaged ion energy distribution from metal and oxide modes for (a) Ar+, (b)

Ti+, (c) O+ and (d) Ti2+. Reproduced with permission from [105]...... 36

2.18 Time averaged ion energy distribution from metal and oxide modes for (a) Ar+, (b)

Al+ and (c) O+. Reproduced with permission from [105]...... 36

- - - - 2.19 (a) Time-averaged negative ion energy distributions for TiO3 , TiO2 , TiO , O2 and

O- ions during reactive HiPIMS. (b) Ion energy distribution for the high energy O-

ions at different pressures. (c) Integrated intensity of the high energy O- population

as a function of the pressure-distance product (pd). Reproduced with permission

from [120]...... 38

2.20 Schematic of representation of dependence of film structure on energy deposited

per atom at substrate and substrate temperature. Reproduced with permission from

[124]...... 40

xiii

2.21 (a) XRD of TiO2 film deposited on unheated unbiased Si (100) substrates using

DCMS and HiPIMS with different peak powers. (b),(c). Energy distribution

functions for (b) Ti+ and (c) O+ ions for the DCMS and HiPIMS discharges.

Reproduced with permission from [96]...... 42

2.22 (a) Diagram of the trajectory of highly energetic O- ions sputtered from the target.

The black arrows indicate the trajectory from a new target surface with an

undefined erosion track. The grey arrows are the trajectory from an old target

surface with a well-define erosion track. (b) AFM micrographs of the surface of

TiO2 films grown by reactive HiPIMS of a new Ti target surface. (c) Grazing

incidence XRD of TiO2 films deposited with reactive HiPIMS of a new target.

Reproduced with permission from [125]...... 43

2.23 (a) Grazing incidence XRD and (b) O2/Ar and pressure dependent phase

information for TiO2 films deposited with HiPIMS. Reproduced with permission

from [102]...... 44

4.1 (a) Photograph of the vacuum chamber used in the experiments. Schematic of the (b)

interior and (c) exterior of the deposition chamber…………………...………….....49

4.2 Photographs of sputtering face of (a) ZnO target and (b) Zn target and the mounting

side of the (c) ZnO target and (d) Zn target...... 50

4.3 Screen-capture of the TRIM/SRIM software input menu...... 52

4. 4 Schematic of (a) exterior and (b) interior of the deposition chamber with the EQP

installed...... 53

xiv

4.5 Schematic diagram of the experimental geometry for synthesis of ZnO over a 4-inch

diameter area from (a) a 2-inch diameter ZnO target and (b) a 1.3-inch diameter Zn

target...... 58

4.6 Images of the Rigaku SmartLab used for X-ray Diffraction analysis of the films.

(Left) exterior image and (Right) interior view...... 60

4.7 (a) Schematic of the shadow mask used for device fabrication. (b) Schematic of the

geometry of the back-gated transistor devices used for electrical characterization. (c)

Micrograph of Al source and drain contacts on ZnO film...... 62

5.1 Calculated sputtering yields as a function of argon ion energy for (a) Zn and (b) ZnO

targets……………………………………………………………………………….65

5.2 Time-resolved current of zinc target at 10 mTorr with O2/Ar = 0.4 with different

pulsing parameters; (a) applied voltage, (b) pulse length, (c) pulse frequency...... 66

5.3 Time-resolved current of zinc target at (a) different total pressures with O2/Ar = 0.4

and (b) a fixed pressure of 20 mTorr and varied oxygen flow rates with a fixed Ar

flow rate of 25 sccm with 200 s pulses of 650 V at a frequency of 100 Hz...... 68

5.4 Area plot of the mass spectra of the plasma from reactive sputtering of a Zn target in

20 mTorr O2/Ar = 0.4 with different applied voltages with 200 s pulses at 100 Hz.

...... 70

+ + + + 5.5 IEDs for (a) Ar , (b) O , (c) O2+, (d) Zn and (e) ZnO measured in a reactive

HiPIMS plasma with a Zn target with 20 mTorr and Opp = 8 mTorr...... 71

+ + + + + 5.6 IEDs for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-

HiPIMS plasma with a Zn target with different total pressures and O2/Ar = 0.40…74

+ + + + + 5.7 IEDs for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-

HiPIMS plasma with a Zn target with 20 mTorr total pressure and different Opp. ... 76

5.8 Time-resolved current of ZnO target at 10 mTorr with O2/Ar = 0.4 with different

pulsing parameters; (a) applied voltage, (b) pulse length, (c) pulse frequency...... 79

5. 9 Time-resolved current of zinc oxide target with 650 V applied with 200 s pulses at

a frequency of 100 Hz with (a) and O2/Ar = 0.4 and the pressure varied and (b) with

the pressure held at 10 mTorr and the oxygen flow rate varied. In both graphs the

argon flow rate was set to 25 sccm...... 81

5.10 Plot of the mass spectra of the plasma from reactive sputtering of a ZnO target in 20

mTorr O2/Ar = 0.04 with different applied voltages with 200 s pulses at 100 Hz. 82

+ + + + + 5.11 The ion energy distributions for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO

measured in a reactive-HiPIMS plasma with a ZnO target with 20 mTorr and an Opp

= 0.8 mTorr...... 83

+ + + + + 5.12 IEDs for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-

HiPIMS plasma with a ZnO target with different total pressures and O2/Ar = 0.04. 85

+ + + + + 5.13 IEDS for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO with 1500 V 200 s 100 Hz

pulses at different total pressures with a fixed O2/Ar of 0.04...... 87

+ + + + + 5.14 IEDS for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-

HiPIMS plasma with a ZnO target with 20 mTorr total pressure and different Opp. 88

5.15 Wide angle XRD patterns for ZnO films deposited from a Zn target in 20 mTorr

with Opp = 8 mTorr with different (a) applied target voltages, (b) pulse lengths, and

xvi

5.16 Wide angle XRD patterns for ZnO films deposited from a Zn target with 500 V

applied in 200 s pulses at a frequency of 100 Hz, (a) different total pressures and

O2/Ar = 0.4 and (b) a total pressure of 20 mTorr and different O2/Ar...... 93

5.17 Wide angle XRD patterns for ZnO films deposited from a ZnO target in 20 mTorr

with Opp = 0.8 mTorr with different (a) applied target voltages, (b) pulse lengths, and

5.18 Wide angle X-Ray diffraction patterns for ZnO films deposited from a ZnO target

with different (a) total pressures and O2/Ar ratio of 0.04 and (b) a fixed total pressure

of 20 mTorr and different Opp...... 95

5.19 Wide angle XRD for ZnO films grown with (a) 200 s and (b) 50 s pulses at

different total pressures and O2/Ar = 0.04. (c) Ratio of the area of the (101) peak to

the total area of the (101) and (002) peaks from the wide angle XRD scan plotted

against pressure...... 97

5.20 Wide angle XRD pattern for films grown at 3 mTorr with (a) 200 s and (b) 50 s

pulse with the peak deconvoluted into its two possible components...... 98

5.21 Rocking curve XRD of the (002) peak for ZnO films grown with (a) 200 s and (b)

50 s pulses at different Ar/O2 pressures. The y-axis is plotted in log scale for

clarity. (c) Full width half maximum of the (002) rocking curve peak of HiPIMS-

grown ZnO...... 99

5.22 Cross-sectional (a) TEM with an insert of a close up of the ZnO-SiO2 interface and

(b) SEM of a film grown at 10 mTorr with a 200 s pulse length...... 100

5.23 Cross-sectional TEM of a ZnO film grown with a deposition pressure of 3 mTorr.

...... 101

xvii

5.24 Surface AFM images and schematics representation of grains of HiPIMS-grown

ZnO films grown with different pressures...... 103

5.25 Schematic diagram illustrating competition for nucleation by (100), (101) and (002)

oriented crystallites during early stages of ZnO growth…………………………..105

5.26 Summary of FWHM of (002) rocking curve XRD peaks at different substrate

temperatures for ZnO films deposited with pulsed DC magnetron sputtering [136],

PLD[34-36], and RF sputtering[46,47] with the data for films synthesized with

HiPIMS...... 108

5.27 (a) SEM micrograph of HiPIMS-grown ZnO on kapton. (b) Close-up of kapton-

ZnO interface...... 109

5.28 (a) Optical image of a 4-in. diameter sample ZnO sample deposited from reactive

HiPIMS of a Zn target with numbers indicating the locations for XRD (b) wide angle

and (c) rocking curve scans...... 111

5.29 (a) Optical image of a 4-in. diameter sample ZnO sample deposited from reactive

HiPIMS of a ZnO target with numbers indicating the locations for XRD

measurements. Wide angle XRD patterns for ZnO films grown at (b) 3 mTorr and

5.30 (a) Image of back-gated FETs with HiPIMS-deposited ZnO channels. (b) I-V

measurements for a FET with ZnO channel. (c) Close-up micrograph of FET source-

drain contacts after electrical measurements...... 113

+ + + + + A.1 The ion energy distributions for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO

measured in a reactive-HiPIMS plasma with a ZnO target with 650 V with different

pulse lengths at 100 Hz with 20 mTorr and an Opp = 0.8 mTorr.....………….…....134

xviii

+ + + + + A.2 The ion energy distributions for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO

measured in a reactive-HiPIMS plasma with a ZnO target with 650 V 200 s pulses

with different pulse frequencies at 20 mTorr and an Opp = 0.8 mTorr...... 135

B.1 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a Zn target in

20 mTorr with Opp = 8 mTorr with different (a) applied target voltages, (b) pulse

lengths, and (c) pulse frequency……...………………………………………...….136

B.2 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a Zn target

with different (a) total pressures and O2/Ar ratio of 0.4 and (b) a fixed total pressure

of 20 mTorr and different Opp…………………………………………………..…137

B.3 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a ZnO target

in 20 mTorr with Opp= 0.8 mTorr with different (a) applied target voltages, (b) pulse

lengths, and (c) pulse frequency...... 137

B.4 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a ZnO target

with different (a) total pressures and O2/Ar ratio of 0.04 and (b) a fixed total pressure

of 20 mTorr and different Opp...... 138

B.5 Additional (a) wide angle XRD and (b) (002) rocking curve measurements for ZnO

films grown with 1500 V, 200 s, 100 Hz with an O2/Ar = 0.04 and a target-

substrate distance of 10 cm...... 140

B.6 (a) Wide angle XRD and (b) (002) rocking curve measurements for ZnO films

grown with 1500 V, 200 s, 100 Hz with an O2/Ar = 0.04 and a target-substrate

distance of 5 cm...... 140

xix

B.7 Topography AFM scans for ZnO films grown with HiPIMS with 1500 V, 200 s

100 Hz pulses with different pressures and O2/Ar = 0.04. The substrate to target

distance was fixed at 5 cm...... 141

B.8 (a) Wide angle XRD and (b) (002) rocking curve measurements for ZnO films

grown with 1500 V, 200 s, 100 Hz with an O2/Ar = 0.04 on a Kapton substrate. 142

B.9 TEM of film cross-section for a ZnO film deposited from reactive HiPIMS of a ZnO

target with 1500 V 200 s 100 Hz at 10 mTorr (Opp= 0.4 mTorr) with 10 cm target-

substrate separation……………………………………………………………….. 142

B.10 TEM of cross-section of ZnO film deposited from reactive HiPIMS of a ZnO target

with 1500 V 200 s 100 Hz at 5 mTorr (Opp= 0.2 mTorr) with 10 cm target-substrate

separation...... 143

B.11 TEM of cross-section of film-substrate interface for ZnO deposited from reactive

HiPIMS of a ZnO target with 1500 V 200 s 100 Hz at 5 mTorr (Opp= 0.2 mTorr)

with 10 cm target-substrate separation...... 143

B.12 TEM of cross-section of film-substrate interface for ZnO deposited from reactive

HiPIMS of a ZnO target with 1500 V 200 s 100 Hz at 3 mTorr (Opp= 0.12 mTorr)

with 10 cm target-substrate separation...... 144

C.1 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr)…145

C.2 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). .. 146

C.3 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). .. 147

C.4 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). .. 148

C.5 Maximum drain current as a function of gate voltages for TFT devices from Fig. C.1

and Fig. C.2 with 50 m long channels...... 148

C.6 Maximum drain current as a function of gate voltages for TFT devices from Fig. C.2

and Fig. C.3 with 60 m long channels...... 149

C.7 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). 150

C.8 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). 151

C.9 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). 152

C.10 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr). 153

xxi

C.11 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr)..154

C.12 TFT drain current measurements as a function of source-drain potential for applied

gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a

ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr)..155

D.1 (a) Survey and (b) high resolution O 1s XPS of ZnO film measured in-situ and after

exposure to atmosphere……………...…………………………………………….157

D.2 High resolution XPS measurements of the O 1s peak for ZnO films synthesized by

reactive-HiPIMS of a ZnO target with different total pressures and O2/Ar = 0.04,

1500 V, 200 s and 100 Hz...... 158

xxii

LIST OF TABLES

Table 4.1 Parameters for ZnO films deposited by reactive-HiPIMS of a Zn target...... 56

Table 4.2 Parameters for HiPIMS-deposition of ZnO deposited by reactive-HiPIMS of a

ZnO target...... 56

Table 4.3 Deposition conditions for ZnO samples for optimized microstructure study. . 57

Table 4.4 Deposition conditions for ZnO growth on 4-in. diameter SiO2/Si substrates. . 58

Table B.1 Deposition rate, diffraction angle and grain size data for ZnO films grown by

reactive-HiPIMS of a Zn target with voltage and pulse parameters systematically

varied…………………………………………………………………………....138

Table B.2 Deposition rate, diffraction angle and grain size data for ZnO films grown by

reactive-HiPIMS of a Zn target with total pressure and oxygen partial pressure

systematically varied...... 138

Table B.3 Deposition rate, diffraction angle and grain size data for ZnO films grown by

reactive-HiPIMS of a ZnO target with voltage and pulse parameters

systematically varied...... 139

Table B.4 Deposition rate, diffraction angle and grain size data for ZnO films grown by

reactive-HiPIMS of a ZnO target with total pressure and oxygen partial pressure

systematically varied...... 139

xxiii

LIST OF SYMBOLS/ABBREVIATIONS

Å    Angstrom (10-10 meters)

    Probability that a Sputtered Atom will be Ionized

T Sticking Coefficient

    Probability that a Target Ion will Return to the Target

    XRD Peak FWHM

c-v Free Energy Associated with Formation of a new Surface

f Surface Free Energy for Film

i Surface Free Energy for Film-Substrate Interface

s Surface Free Energy of Substrate

ss Self-Sputtering Yield

SE Secondary Electron Yield

G Gibbs Free Energy

G* Activation Free Energy Barrier For Nucleation

Gv Volume Free Energy

 Voltage Equivalent of an Electron’s Energy

xxiv

iz Ionization Energy

 Mean Free Path

De Debye Length

I Ionization Mean Free Path

x-ray Wavelength of X-Ray Source

Fe Field Effect Mobility

s Microseconds (10-6 seconds)

 Resistivity

iz Ionization Cross-Section

    Mean Crystal Size

 Sheath Potential

g Larmor Frequency

A Amps

AFM Atomic Force Microscopy

Amu Atomic Mass Units

Ar Argon

B Magnetic Field

푐̅푒 Average Electron Speed

푐̅푖 Average Ion Speed

Cj Statistical Weight Factor

C/s, cps Counts per second

CVD Chemical Vapor Deposition

D Diffusivity

xxv

i+1 Dni Diffusion Coefficient for Single Atoms to Diffuse to a

Cluster of j Atoms

DC Direct Current

DCMS DC Magnetron Sputtering e Elementary Charge ( *10 C)

E Electric Field

* , E Average Energy Deposited to Substrate

EF Energy of Film at Substrate

EI Energy Incident Particles at Substrate

Ei Ion Energy ej Statistical Weight Factor

Ej Bonding Energy of Film Atoms to Film Atoms

EQP Electrostatic Quadrupole Probe

EPotential Potential Energy

ER Energy of Reflected Particles

ES Energy of Substrate eV Electron Volt

F Deposition Rate

FB Force from the Magnetic Field

FE Force from the Electric Field

FET Field Effect Transistor

FL Lorentz Force

FWHM Full Width at Half Maximum

HiPIMS High Power Impulse Magnetron Sputtering

xxvi

Hz Hertz i Number of Atoms in a Stable Cluster

IED Ion Energy Distribution

IEDF Ion Energy Distribution Function

Ii Ion Current

ISE Current from Secondary Electrons

IT, ITarget Target Current

IR Infrared

I-V Current-Voltage j Number of Atoms is a Cluster

Ji Flux of Accelerated Ions to Substrate

JMe Flux of Depositing Particles to Substrate

K Boltzmann’s Constant

K Dimensionless Shape Factor m Mass ms Milliseconds (10-3 seconds_ mTorr Millitorr (10-2 Torr)

No Number of Adsorption Sites no Number Density of a Critical Cluster n1 Number Density of Single Atom Clusters ne Number Density of Electrons ni Number Density of Ions nm Nanometers (10-9 meters) nx Number Density for Stable Clusters

O Oxygen

xxvii

OES Optical Emission Spectroscopy

Opp Oxygen Partial Pressure

Pa Pascals

Pav Average Power

PLD Pulsed Laser Deposition

PVD Physical Vapor Deposition q Charge

R Film Species Arrival Rate r* Critical Cluster Radius rc Cluster Radius

RC Rocking Curve

RF Radio Frequency rg Larmor Radius

RMS Root Mean Square s Sheath Thickness sccm Standard Cubic Centimeters per Minute

SEEC Secondary Electron Coefficient

SEM Scanning Electron Microscopy

SEM (C/s) Secondary Electron Multiplier (counts/second)

SRIM Stopping and Range of Ions in Mattter

SY Sputtering Yield

Um Rate at Which Single Atoms Join a Cluster t Time

Te Electron Temperature

xxviii

T* Homologous Temperature

TEM Transmission Electron Microscopy

TF Temperature of Film on Substrate

TFT Thin Film Transistor

Ti Titanium

TI Temperature of Incident Particles on Substrate

TM Melting Temperature of Film Material

TR Temperature of Reflected Particles

TRIM Transport of Ions in Matter

TS, Tsub Temperature of Substrate v Velocity vds Velocity Distribution Function

Vf Floating Potential

Vp Plasma Potential

Vo Applied Target Potential

W Tungten x Distance from Cathode

XPS X-Ray Photoelectron Spectroscopy

XRD X-Ray Diffraction

Z Fraction of Substrate Covered by Film

Zn Zinc

ZnO Zinc Oxide

xxix

CHAPTER I

INTRODUCTION

Oxide thin films are an integral material system for many nanoscale electronic applications. Depending on whether the oxide is insulating or semiconducting, the film can be used for field effect transistors (FET) channels or gate dielectrics. Many oxides can also be doped allowing their use as conducting contacts. The inherent wide band gap of oxides provides potential for transparent electronics. Another benefit of oxide materials is their tendency to have high melting points, high dielectric constants and high breakdown voltages making them robust materials for devices. Incorporation of oxide films into electronic devices requires tight control of the film structure and chemistry while still allowing scaling of the process for large area coverage. Additionally, the high melting temperature of these materials can create inherent processing challenges at low substrate temperatures. This study demonstrates the use of reactive high power impulse magnetron sputtering (HiPIMS) as a scalable low-temperature processing technique for nanoelectronic oxides, specifically semiconducting zinc oxide (ZnO) for thin film transistor (TFT) channels. This study also characterizes the effect of deposition parameters on plasma species and their energies and correlates those effects to the resulting film microstructure to establish mechanisms for film growth.

High power impulse magnetron sputtering is a physical vapor deposition (PVD) technique that has been the focus of considerable research in the past two decades [1-3].

HiPIMS processing possesses many of the benefits of dc magnetron sputtering, i.e. scalability, low-temperature processing, and control of film microstructure via tuning of deposition parameters, while achieving a large fraction of ionized sputtered material reaching the target. Achieving a large fraction of ionized sputter material at the substrate surface is difficult in dc magnetron sputtering because target heating effects limit the power density applied, limiting the plasma density and therefore the ionization of sputtered material. HiPIMS overcomes the power limitations experienced in DC magnetron sputtering though application of microsecond pulses with long off-times [4].

The small duty factor during HiPIMS, typically < 10%, allows peak power densities on the order of several kW∙cm-2 to be applied to the target while maintaining average power densities of a few W∙cm-2, values comparable to those during DC magnetron sputtering (DCMS). The high peak power during the pulse results in plasma densities on the order of 1015-1029 m-3 with electron densities in the ionization region in front of the target surface on the order of 1018-1019 m-3. These densities, which are several orders of magnitude larger than the densities typically observed in dc magnetron sputtering, result in a shortening of the ionization mean free path ( i ~ 1 cm for HiPIMS compared to i ~ 50 cm for dc magnetron sputtering) and increase the ionization rate of the material sputtered from the target [6]. The higher degree of ionization in the sputtered material promotes crystallization in the films, can reduce surface roughness and increase film density [7].

Most early investigations of HiPIMS focused on the growth of metal thin films

[1,2], nitrides [3], protective coatings [4] and for use as a substrate pretreatment to improve film-substrate adhesion [5]. Recently, however, HiPIMS has been demonstrated as a technique for the deposition of oxide films [6,7]. Oxide films deposited with

HiPIMS have been shown to have a lower surface roughness [7,8], higher film density

[8], lower electrical resistivity [7,9] and a higher refractive index [10,11] than films deposited with DC magnetron sputtering making HiPIMS a promising technique for the deposition of semiconducting channel materials for thin film transistors. This study demonstrates the feasibility of reactive-HiPIMS as a processing technique for the deposition of nanocrystalline ZnO TFT channels. To facilitate future incorporation of

HiPIMS in TFT fabrication, this study strives to strengthen the current understanding of

HiPIMS-growth of oxide films through characterization of the plasma and the resulting film microstructure. The effect of deposition parameters on film microstructure was studied by systematically varying the parameters and analyzing the microstructure of the resulting films. Parameters that had the greatest effect on film microstructure were identified and investigated in greater detail. The plasma was characterized and plasma phenomena correlated to the nucleation and evolution of film morphology.

The proceeding section will provide a brief background on ZnO, its processing, and incorporation into TFTs. To facilitate later discussion of HiPIMS processing of ZnO and the role of plasma conditions on film microstructure, a condensed review of film nucleation and growth, and plasma discharge physics will be given followed by a review of literature on plasma dynamics and film growth during HiPIMS.

CHAPTER II

BACKGROUND

2.1 Zinc Oxide as a Thin Film Transistor Channel

Zinc oxide is a semiconductor that was been the focus of considerable research

[12-14] for use in electronic applications due to its intriguing combination of high mobility, with field effect mobilities up to 110 cm2 V-1s-1 [15,16] (which approaches the mobilities of 200 cm2 V-1s-1 reported for single-crystalline Si [17] and surpasses the mobilities of less than 5 cm2 V-1s-1 reported for amorphous silicon and organic semiconductors [17]) and wide band gap (3.3-3.6 eV). The relatively high mobility of polycrystalline ZnO is attributed to ionic bonding in ZnO compared to the covalent bonding in Si-based semiconductors [18]. Overlap of oxygen conduction band 2s orbitals with neighboring Zn 4s orbitals provide paths for electrons regardless of distortions in the

Zn-O chemical bonds similar to that observed in amorphous oxide semiconductors (such as indium-gallium zinc oxide) [18]. The wide band gap results in higher breakdown voltages [12], the ability to sustain larger electric fields [12], lower noise generation [12], and the capability for high-temperature high-power operations [12]. The temperature stability and radiation hardness [19,20] of ZnO make it tolerant to extreme environments.

These properties make ZnO attractive as a channel material for a variety of optoelectronic and TFT devices used in active displays, solar cells, chem- and bio-sensors, radio frequency (RF) signal processing, power electronics and other applications [12-14].

Additionally, the wide band gap makes it optically transparent allowing for its incorporation into transparent electronics [12]. Extensive research has been done on ZnO as a transparent conductive oxide material [12-14,21-24] and on n-type or p-type doping of ZnO [15-17,19-21][25-27].

In TFT devices, the channel material can affect device performance by dictating the on/off ratio, peak current density, and saturation current. High performance TFTs, such as microwave FETs, require large current densities and high on/off ratios. The incorporation of ZnO channels in a microwave FET has been demonstrated by

Bayraktaroglu, Leedy and Neidhard [15] using ZnO films deposited with pulsed laser deposition (PLD) [28]. Schematic drawings of Bayraktaroglu et al.’s DC and microwave

FET devices are shown in Figure 2.1a and Fig. 2.1b, respectively. A micrograph of the microwave FET used for device characterization is shown in Figure 2.1c. Bayraktaroglu et al. reported field effect mobilities up to 110 cm2 V-1s-1 and on-off ratios as large as

1012. The current density of these FETs was dependent on deposition temperature and gate length.

Figure 2.1 Schematic drawing of (a) ZnO FET on Si substrate for DC characterization and (b) of microwave ZnO FET on PECVD SiO2/GaAs substrate. (c) Micrograph of the microwave transistor. Reproduced with permission from [15].

The high mobilities and on-off ratios in Bayraktaroglu et al.’s ZnO films are attributed to the films’ microstructure. Cross-sectional transmission electron microscopy

(TEM) of the ZnO film over the TFT gate (Figure 2.2) shows a continuous closely- packed nanocolumnar structure [16]. This morphology extends even over non-planar surfaces, such as the transition from SiO2 to the metal gate [16]. The high on/off ratios reported in these studies are a direct result of strong crystallinity and dense nanocolumnar morphology of the films, shown in Figure 2.2. When the TFT is in the off-state, the closely-packed vertical grain boundaries create back-to-back depletion zones that prevent horizontal current flow from the source to drain and minimize leakage currents. When a gate potential is applied during the on-state, the depletion zones collapse and current flow is possible. Additionally, control of the film microstructure is crucial for TFT applications because the conduction mechanism in ZnO is controlled by point defects in the film, most importantly interstitial zinc atoms and oxygen vacancies [19,23,29,30].

Figure 2.2 Cross-sectional TEM of a PLD-deposited ZnO film over the TFT gate. Reproduced with permission from [16]. (top left). Schematic diagram of TFT with a ZnO channel (bottom left). Close up of the TFT channel showing the depletion zones at the grain boundaries in the off-state (top right) and the collapse of the depletion zones under the gate field in the on-state (bottom right).

In addition to PLD [27,31-36], ZnO films have been synthesized using chemical vapor deposition (CVD) [37,38], atomic layer deposition [39], solution based hydrothermal deposition [40], filtered vacuum cathodic arc deposition [41-43], RF sputtering [30,44-47], DC and pulsed DC magnetron sputtering [22,25], HiPIMS [8,48] and epitaxial techniques [46,49-51]; methods that all result in wurtzite ZnO with a (002) preferred crystallographic orientation. The degree of alignment of the (002) orientation, presence of crystallites with other crystallographic orientations, grain size and surface roughness of the films, however, are affected by the energy of the film material arriving at the substrate. This energy is dependent on the processing technique and substrate temperature. Low energies (and low substrate temperatures) result in less surface mobility for the growing film, which can cause the formation of defects and small grains.

Charge carrier scattering at these defect sites and grain boundaries can reduce the film conductivity. The energy of the film species can be increased by depositing at higher substrate temperatures, higher powers during deposition or using techniques with inherently higher energy fluxes, like RF sputtering or HIPIMS.

The effect of substrate temperature and processing technique on film microstructure, specifically the degree of alignment of the (002) orientation relative to the substrate, for films synthesized with PVD techniques is illustrated in Figure 2.3. The graph shows the full width half maximum (FWHM) of the ZnO (002) rocking curve (RC) peak as a function of deposition temperatures. The smallest FWMH, corresponding to films with the highest degree of (002) alignment, occurs at high deposition temperatures and for films deposited with RF magnetron sputtering. Depositions at lower temperatures resulted in a broader distribution of alignment for the (002) crystallites. No data on the

7 degree of (002) alignment of HiPIMS-grown ZnO films was found in literature. Review of literature on PVD processing of ZnO found no reports for rocking measurements for films synthesized at substrate temperatures below 200 oC, a processing region critical

(highlighted by the blue circle in Fig. 2.3) for incorporation of ZnO films into flexible electronic applications and the focus of this study.

Figure 2.3 Summary of FWHM of (002) rocking curve XRD peaks at different substrate temperatures for ZnO films deposited with pulsed DC magnetron sputtering [52], PLD [34-36], and RF sputtering [46,47].

The intrinsic challenges to the scalability of PLD thin film growth to coat large area and complex-shaped substrates motivates investigation of other techniques for low temperature and scalable deposition of ZnO films that will potentially yield the material morphology and structure optimization identified with the PLD growth studies. To

8 achieve a high carrier mobility and large on/off ratios, PLD processing has shown a need for ZnO films to have a low surface roughness, dense morphology and a strong (002) orientation [15,16,28,52]. From this perspective, the energetic plasma fluxes during magnetron sputtering techniques have a great deal of potential for scalability of ZnO synthesis at low temperatures. However, initial investigations with conventional magnetron sputtering reported in literature yield ZnO films with low mobilities (FE ≤1.8 cm2V-1s-1 ) [26] and small on/off ratios, due to poor crystallinity, and an open columnar structure with a high degree of porosity. RF magnetron sputtering produces smooth crystalline films with highly aligned (002) crystals [46,47,53], however, the need for impedance matching of the substrate-plasma system make industrial-size scaling of this process difficult. The high ion fluxes and ion energies [1,2] associated with HiPIMS, combined with its compatibility with existing DC magnetron sputtering systems, make it an alternative technique for scalable growth of ZnO with controlled microstructures and crystallographic orientation.

Reactive HiPIMS of Zn targets in argon-oxygen background has previously been demonstrated to synthesize ZnO films on glass [8] and sapphire [48] substrates. The ZnO films grown on both substrate materials were nanocrystalline with small 1200-1300 nm2 size wurtzite grains and preferred c-axis orientation (Figs 2.4 and 2.5). These films had a surface roughness below 1.0 nm and lower porosity than ZnO films grown with dc bipolar magnetron sputtering [8]. The deposition rates of HiPIMS and DC sputtering were similar when HiPIMS was operated with short 10 s [8] and 50 s [48] pulses. The feasibility of reactive HiPIMS for the deposition of ZnO thin films for electronic applications was tested by Partridge, Mayes, McDougall, Bilek and McCulloch [48] by

9 measuring the films’ Hall mobility and the I-V characteristics of the ZnO film in a diode.

The as-deposited films had Hall mobilities of 7-8 cm2V-1s-1, values comparable to those reported for DC magnetron sputtering [54] but significantly lower than the Hall mobilities of >50 cm2V-1s-1 reported for ZnO films grown with PLD [16,28,52]. The lower Hall mobilities for the ZnO synthesized with HiPIMS compared to that of PLD- grown films likely results from increased carrier scattering due to the significantly smaller grains in the HiPIMS film.

Figure 2.4 AFM micrograph of surface of ZnO layer deposited using (a) bipolar pulsed DC magnetron sputtering and (b) HiPIMS on glass substrates. (c) XRD of the ZnO films in (a) and (b). Reproduced with permission from [8].

Figure 2.5 (a) SEM and (b) XRD of ZnO film deposited with HiPIMS on a sapphire substrate. Reproduced with permission from [48].

Neither Konstantinidis, Hemberg and Dauchot’s nor Partridge et al.’s studies report on the effect of deposition parameters on film microstructure. Additionally there are no reports on characterization of the HiPIMS plasma during the processing of ZnO.

Detailed characterization of the processing plasma during reactive-HiPIMS of ZnO and investigation of the effect of deposition conditions on film microstructure could facilitate the incorporation of HiPIMS as a processing technique for TFT channels.

2.2 Film Nucleation and Growth

During PVD synthesis of films, film species arrive in vapor phase at the substrate surface at a rate R, and then film nucleation and growth proceeds following the processes illustrated in Figure 2.6. Once on the substrate surface the films species are thermally accommodated by the substrate or re-evaporated. The condensation of “sticking’ coefficient”, which determines whether the species will re-evaporate, coalesce into a higher energy crystallite or contribute to film growth, can be expressed as

푇퐼−푇푅 퐸퐼−퐸푅 훼푇 = = (1) 푇퐹−푇푆 퐸퐼−퐸푆

where TI, TR, TF and TS are the temperatures of the incident particles, the re-evaporated particles, the film and the substrate, respectively, and EI, ER, EF and ES are their energies

[55]. If accommodation occurs, the film species diffuse on the substrate surface interacting with other film atoms and contributing to film growth. The diffusivity (D) of the film species of the substrate surface obeys an Arrhenius law [56] so diffusivity is higher at increased substrate temperatures. The rate at which deposited species are accommodated and add to film growth is the deposition rate (F).

Figure 2.6 Schematic of atomic processes in film nucleation and growth. Adapted from [58].

When the diffusion rate of the film species on the substrate is higher than the deposition rate, the adsorbed species have significant time to reach a minimal energy configuration. Under these conditions film nucleation and growth is near equilibrium and driven by thermodynamics [56]. Thermodynamically-driven nucleation and growth can be divided into three different modes based on the surface free energy of the film- substrate system [57]. The three modes are illustrated schematically in Figure 2.7. The first mode of growth (Fig 2.7a) is the Frank-Van der Merwe or 2-dimensional layer-by-

12 layer growth mode. In Frank-Van der Merwe growth, the total surface energy is lower for film-covered substrate than for bare substrate, this can be expressed as:

훾푓 + 훾푖 < 훾푠 (2) where f, i and s are the surface free energies for the film, film-substrate interface and the substrate, respectively. Frank-Van der Merwe growth occurs when the bonding between the substrate and film is strong enough to reduce i and satisfy equation (2)

[58]. This mode of growth is beneficial for many applications because it tends to result in highly uniform smooth films. If equation (2) is not satisfied for a film-substrate system then 3-dimensional island or Volmer-Weber growth (Fig. 2.7b) occurs. In

Volmer-Weber mode, growth begins with small clusters nucleated directly on the substrate surface. These clusters grow into islands and coalesce to form a continuous film. The third mode of growth, layer-island or Stanski-Krastanov growth (Fig. 2.7c), is a combination of the other two modes. In Stranski-Krastanov growth, the film initially grows layer by layer for one or more monolayers then a change in the free surface energy makes further layer growth energetically unfavorable resulting in a transition to 3D growth.

Figure 2.7 Schematic diagram of the three modes of film nucleation and growth. Adapted from [61].

In Volmer-Weber nucleation and growth, a high surface to volume ratio at the initial stages of film growth results in a minimum critical cluster size for film growth to proceed [59]. The critical cluster size is determined by considering the free energy of the cluster on the substrate surface. This free energy, G, can be expressed as

4 Δ퐺 = (휋푟 3)Δ퐺 + 4휋푟2Γ (3) 3 푐 푣 푐−푣 where rc is the cluster radius, Gv is the volume free energy and c-v, is the free energy associated with the formation of a new surface between the substrate and the condensate film material and always a positive value [59]. When equation (3) is plotted against the cluster radius, an activation free energy barrier to nucleation (G*) and a critical cluster size (r*) is observed. Clusters larger than r* can continue to grow, thereby lowering their free energy while clusters smaller than r* will dissolve. The critical cluster size can be expressed as:

−2Γ 푟∗ = 푐−푣 (4) . Δ퐺푣

The population density of clusters that meet the minimum cluster size (no) is determined by the bonding energy of the film atoms to film atoms (Ej) and the number of adsorption sites on the substrate (No). The Walton relation (Eqn. 5) considers the relationship between No, Ej and no and can be expressed as:

푛 푛 1 표 1 푗 훽퐸푗(푚) = ( ) ∑푚 퐶푗(푚)푒 where 훽 = (5) 푁0 푁표 푘푇 where n1 is the number density of single atoms, Cj is a statistical weight for each cluster and j is number of atoms in the cluster. The rate at which single atoms join stable clusters can be expressed as:

푑푛 푑푍 푥 = (1 − 푍)−푖훾 퐷푖+1 − 2푛 − 푈 (6) 푑푡 푖 푛1 푥 푑푡 푚

14 where nx is the population density of a stable cluster, Z is the fraction of the substrate

i+1 covered by stable clusters, i is the number of atoms in a stable cluster, Dn1 is the diffusion coefficient for single atoms to diffuse to a cluster of i atoms and Um is rate at which single atoms join mobile clusters [60].

Energy deposited into the growing film comes from either thermal effects (i.e. the substrate temperature) or kinetic energy from particles incident on the substrate. From the Arrhenius behavior of adatom diffusivity resulting in increased surface mobility at higher temperatures to lower sticking coefficients at higher temperatures promoting growth of crystallites with higher energy planes parallel to the substrate surface, substrate temperature during deposition plays a critical role in the development of film structure.

The dependence of film structure on substrate temperature has been illustrated in several structural zone models [61,62], including the Thornton diagram shown in Figure 2.8.

The substrate temperature axis in Figure 2.8 shows the homologous temperature (T*) which is the ratio of the substrate temperature (TS) to the melting point temperature of the film material (TM). Films grown at temperatures significantly lower than the material’s melting temperature have little adatom surface mobility during growth resulting in tapered crystallites and voids between grains. Higher substrate temperatures resulted in higher surface mobility allowing for grain boundary migration and denser films with higher quality crystals. Dense high quality crystalline films of high melting point materials, like many oxides, are therefore difficult to process using techniques where film nucleation and growth is thermodynamically driven, making processes that are kinetically driven beneficial.

Figure 2.8 Schematic of representation of dependence of film structure on argon pressure and substrate temperature. Reproduced with permission from [62].

During plasma-based deposition processes, such as DC magnetron sputtering and

HiPIMS, the deposition rate is typically higher than the diffusivity of the film species on the substrate surface and film nucleation and growth is driven by kinetic processes [56].

Kinetically-driven film growth can often lead to metastable structures. Film growth during these processes typically occurs under highly non-equilibrium conditions due to the considerable kinetic energy (typically tens of eV) and the multi-directional incidence of the sputtered atoms.

2.3 Magnetron Sputtering and the Sputtering Discharge

To facilitate discussion of the plasma dynamics during HiPIMS, and their subsequent effect on film growth, the basic architecture of a DC glow discharge, shown

16 schematically in Figure 2.9a, will be discussed in this section. A potential applied to the cathode using a voltage source results in a potential difference between the cathode and anode that drives a current from the cathode through the gas to the anode. The current through the gas results in dissociation of the gas atoms creating a quasi-neutral plasma with the number density of electrons (ne) being approximately equal to that of ions (ni).

The electrons and ions formed from this dissociation, however, move through the chamber at different speeds due to the significantly smaller mass and higher temperatures of the electrons resulting in larger velocities compared to the heavier ions. The faster speed of the electrons results in higher electron loses to the cathode, anode and all the walls in the system creating a region near the electrodes and walls where the density of ions is greater than that of electrons. This results in a thin layer or sheath of positive charge between the system surfaces and the plasma bulk. The thickness of the plasma sheath (s) is determined by:

√2 2푉표 3⁄4 푠 = 휆퐷푒( ) (7) 3 T푒 where De is the electron Debye length, Vo is potential at the cathode (typically equal to the applied voltage) and Te is the electron temperature [63]. The potential in the sheath as a function of position is determined by

푥 Φ = −푉 ( )4⁄3 (8) 표 푠 with x being the distance from the cathode [63]. The potential profile for this system is shown in Figure 2.9b. The positive potential in the sheath increases through the thickness of the sheath as the electron density decreases. At the sheath edge the potential reaches its maximum value and remains constant through the rest of the plasma. This potential is referred to as the plasma potential (Vp).

Figure 2.9 (a) Schematic of a DC powered glow discharge. (b) Voltages at different regions between the cathode and anode in the glow discharge assuming the anode is held at 0 V. Adapted from [63].

In a magnetron sputtering discharge, the cathode is the sputtering target. The sputtering target is mounted to a magnetron gun (Figure 2.10a) which consists of two permanent magnets, a power connector and a cooling system. The permanent magnets in the magnetron gun create a magnetic field (B) shown in Figure 2.10b. When a voltage is applied to the target, there is an electric field (E) normal to the target face. The force from the two fields acting on a charged particle is called the Lorentz force and can be expressed as:

퐹퐿 = 퐹퐸 + 퐹퐵 = 푞푬 + 푞풗 × 푩 (9) where q is the charge, m is the mass and v is the velocity of the charged particle. The electric field component of eqn. (9) changes the magnitude of the particle’s velocity causing it to accelerate as a negatively charged particle moves away from the field source and decelerate as it moves toward the field source. Positively charged particles will decelerate while moving away from the target and accelerate while moving toward it.

The magnetic field component of the affects the direction of the charged particle’s motion causing the particle to gyrate of travel in a helix around the field lines with a radius r of and a frequency g. The radius of the particle gyration is called the Larmor radius and is expressed as:

푚푣 푟 = ⊥ (10). |푞|퐵 with 푣⊥ being the velocity component perpendicular to the magnetic field. The large mass of ions in the system results in a Larmor radius larger than the characteristic radius of the magnetron so ions do not gyrate around the target. The very small mass of electrons, however, results in a Larmor radius orders of magnitude smaller than the magnetron radius so the electrons become “trapped” by the magnetron’s magnetic field and travel around the target surface in a slanted helix orbit resulting in region with a high electron density. This region is often referred to as the target racetrack.

Figure 2.10 (a) A typical magnetron gun used for sputtering. (b) Schematic of the electric and magnetic fields for a magnetron gun with the particle velocity, plasma and plasma sheath identified. (insert) The electron drift about a target and an image of a target racetrack.

Although not appreciably affected by the magnetic field, gas ions can be accelerated across the target sheath to the target surface resulting in five possible phenomena, illustrated in Figure 2.11 [64]. The gas ion can be reflected, either with its charge or as a neutral, or it can become implanted in the target. The impact of the gas ion on the target surface may cause the ejection of an electron, called a secondary electron.

The number of secondary electrons ejected per incident ion is called the secondary electron yield (SE) and is determined by the secondary electron coefficient (SEEC). The ion impact can also result in structural rearrangement in the target material, which can vary from creation of vacancies and interstitials to creation of large lattice defects such as changes in the stoichiometry, electrical levels and electrical distribution of the target.

Lastly, impact from the gas ions may start a series of collisions between target ions leading to the ejection of one of these atoms resulting in “sputtering”. The number of target atoms ejected per colliding ion is referred to as the sputtering yield (SY). The sputtering yield is strongly dependent on the surface binding energy of the target [29] and the energy of the bombarding ions [29].

Figure 2.11 Interactions of ions with the target surface. Adapted from [64].

Sputtering in the presence of oxygen or other reactive gas is called reactive sputtering and results in compound films of the sputtering target and reactive gas. In reactive sputtering the reactive gas, is adsorbed on all surfaces of the deposition system.

The gas can also react with the target surface through two main mechanisms. The first mechanism is chemisorption where, in the case of sputtering in oxygen, oxygen forms a layer of target-oxide on the target surface. This formation of surface oxide is often called

‘target poisoning’. In addition to covering the target surface, reactive gas ions can also be implanted into the target by acceleration by the cathode sheath. This ion implantation is typically to a depth of several nanometers [65].

Coverage of the target with an oxide layer results in a decrease in the sputtering yield due to the coverage of the target by the compound material if the sputtering yield of the compound material is lower than that of the pure metal. Additionally, SEEC and resistivity of the target will be different for the compound material. The change in SEEC and resistivity will result in a decrease or increase in the current at the target and a rise in the hysteresis effect [66,67]. Another complicating factor in reactive sputtering is the introduction of instabilities in the discharge if the target-oxide material is electrically insulating. The portions of the target outside the racetrack are not sputtered and can develop a thick oxide layer that acts as a dielectric resulting in charging during the pulse and arcing [66,67]. Another potential problem is the disappearance of the anode as the oxide is deposited [67].

The anode region of the discharge usually consists of the substrate and its sheath.

The example given in Figure 2.9b assumes that the anode is held at zero. In this case the anode sheath potential is equal to the plasma potential. If the anode is biased, the sheath

21 potential will be the sum of the plasma potential and the bias potential. The anode relies on the glow for ions as very little ionization occurs in the anode sheath. A significant amount of the power input to the anode comes from fast electrons generated in the cathode sheath. In addition to bombardment by fast electrons, the anode is bombarded by ions, photons and slower electrons from the glow region. Ions enter the anode sheath with energies of about kTe/e (a requirement of the Bohn criteria) and accelerate through the sheath. Secondary electrons (originating from the cathode) are accelerated from the anode back to the glow.

If the anode in Figure 2.9a is electrically isolated (or an electrically isolated probe is placed in the plasma) it will initially be struck by electrons and ions with fluxes or current densities, je and ji, respectively. The fluxes can be expressed as:

푒푛 푐̅ 푗 푒 푒 (11) 푒 4

푒푛 푐̅ 푗 푖 푖 (12) 푖 4 where e is the elementary charge, 푐푒̅ is the average speed of electron and 푐i̅ is the average speed of the ions [63]. The flux of electrons will be much larger than the flux of ions due to the greater average speed of the electrons. This causes the substrate to develop a negative charge resulting in a negative potential with respect to the plasma. This negative potential is often referred to as the floating potential (Vf). The negative potential at the substrate repels electrons causing the flux of electrons to the substrate to decrease.

The negative potential also attracts positive ions. However, the flux of these ions to the substrate is limited by the ions’ random-arrival at the plasma-sheath interface, so the flux is not increased. The substrate will continue to gain a negative charge until the flux of electrons has been reduced sufficiently to be balanced by the flux of positive ions.

The difference between the plasma potential (Vp) and the floating potential (Vf) acts like a barrier to electrons. In order to pass through this barrier electrons need a potential energy equal to or greater than the potential difference in energy unit (PE ≥e(Vp-

Vf)). From the Maxwell-Boltzmann distribution, the fraction of electrons capable of

′ passing through the sheath (푛푒) are

푒(푉 −푉 ) ′ 푝 푓 푛푒 − = 푒 푘푇푒 (13) 푛푒 where k is the Boltzmann constant. Combining eqn (12) with the Bohm criterion, which requires ions entering the sheath to have energies of kTe/e and assuming charge flux

′ balance between 푛푒 and ni:

푛′ ̅푐̅̅̅ 푛 푐̅ 푒 푒′ = 푖 푖 (14) 4 4 with 푐̅̅푒̅̅′ and 푐̅푖 being the average speeds of the electrons capable of passing through the sheath and that of the ions, respectively [64], then the relationship between the Vp-Vf and the Te can be expressed as

푘푇푒 푚푖 푉푝 − 푉푓 = ln ( ) (15). 2푒 2.3푚푒

The difference between Vp and Vf will affect the energy delivered to the surface of a growing film during deposition, thereby affecting the nucleation and evolution of the film. Higher plasma potentials will result in higher energy ions arriving that the film surface.

2.4 High Power Impulse Magnetron Sputtering

Now that an understanding of the basics of film growth and magnetron discharges has been established, the next section will focus on target surface interactions and gas-

23 phase phenomena during reactive HiPIMS and how they affect the plasma dynamics and film growth. A brief review of current literature results will also be given.

2.4.1 Plasma Dynamics during the HiPIMS Process

In Section 2.3 the description of the plasma discharge there were no electric fields within the plasma bulk or glow region due to the constant plasma potential through the bulk of the plasma. However, anomalously large potential gradients (Vp ~ 22 V across

4 cm) in the plasma bulk have been reported for pulsed dc magnetron discharges [68,69].

The presence of similar gradients in the HiPIMS discharge could act as a barrier to ionized sputter material reaching the substrate surface [70]. The plasma potential Vp for a

HiPIMS discharges has been investigated by several groups using time-resolved

Langmuir probe measurements [71-73]. These studies showed only positive Vp during the entirety of the pulse with the peak potential value (~ 5 V) occurring at the end of the pulse ‘on-time’ [71,72]. A gas pressure-dependence of Vp was also observed with lower working gas pressures resulting in higher plasma potentials [71,72].

The poor temporal resolution (typically ~1 s) and potential for false readings for magnetized or pulsed plasmas associated with estimates of Vp using Langmuir probes, prompted Mishra, Kelly and Bradley to develop an emissive probe technique to obtain more accurate measurements [70]. In their study, Mishra et al. used an electron emitting probe to measure the temporal evolution of Vp along a line above the target racetrack of titanium target from the target to the substrate. Figure 2.12d shows a plot of the time evolution of a typical plasma potential in Mishra et al.’s study. Mishra’s group identified

24 three different times on the probe potential waveform in Figure 2.12d, which they labelled as the first peak, the second peak and the ‘off-phase’, to investigate in greater detail. The plasma potential at the three times identified in Figure 2.12d is plotted as a function of axial distance z from the target surface in Figure2.12a-c for different discharge power- gas pressures. Figure 2.12a shows that in the initial phase of the pulse large potentials ( -210 V) develop inside the magnetic trap then the potential levels out.

The presence of a ‘triple-layer-like’ feature near the cathode (z = 5 cm) could possibly support the field reversal in HiPIMS theoretically predicted by Brenning, Axnas, Lundin and Helmersson [74]. The potential profile of the second peak (Fig 2.12b) shows that as time progresses Vp becomes more positive until it reaches values higher than the ground potential outside the magnetic trap. From Mishra et al.’s study it was also determined that temporally and spatially Vp is strongly sensitive to changes in applied power and only mildly sensitive to changes in pressure [70].

Figure 2.12 Plots of Vp versus axial distance z from the cathode for (a) the ‘first peak’, (b) the ‘second peak’ and (c) the ‘off-phase’. (d) Plot of the time evolution of typical plasma potential (Vp) measures at a set axial distance (z = 100 mm) from the target surface and radial position(r= 45). Plots with ■ are for an average discharge power of 650 W and gas pressure of 0.54 Pa, □ are for 650 W and 1.08 Pa and are for 950 W and 1.08 Pa. Reproduced with permission from [70].

Another key difference between HiPIMS and pulsed DC magnetron sputtering is the high degree of ionization of the sputtered material. The prominent ionization mechanism during sputtering is electron impact ionization. In electron impact ionization an incident electron transfers to an atom’s valence electron energy equal to the ionization energy of the atom. The probability of electron impact ionization occurring is determined by the ionization cross-section (iz) which is a Thomson cross-section expressed as:

푒 2 1 1 1 휎푖푧 = 휋 ( ) ( − ) (16) 4휋휀표 휀 휀푖푧 휀

26 where  is the voltage equivalent of the electron’s energy, iz is the ionization energy of the atom and e is the elementary charge [63]. During HiPIMS, the power densities applied to the target are large enough to cause ionization of the sputtered atoms. The magnetic field structure of the magnetron determines where this ionization occurs, though it is typically within a few millimeters of the target surface [75]. Ionization of the sputtered material so close to the target surface introduces the possibility that the ionized sputtered material can contribute to further sputtering of the target, a process called ‘self- sputtering’ [76,77]. The extent of self-sputtering is determined by the self –sputtering yield which is largely dependent on the surface binding energy of the target material [78].

It has been demonstrated that self-sputtering of materials with low surface binding energies can operate in a pure ‘gasless’ mode once sputtering has been initialized [78].

Self-sputtering in HiPIMS was extensively studied by Anders et al. and the conditions required for sustained self-sputtering and a mode called runaway have been established

[76,78-81]. The right side of Figure 2.13 shows loops illustrating self-sputtering and gas- sputtering. The condition for sustained self-sputtering can be expressed as

Π = 훼훽훾푠푠 = 1 (17) where is the probability that a sputtered atom becomes ionized,  is the probability that the newly formed ion will return to the target and ss is the self-sputtering yield of the target material [78]. The condition for self-sputtering run away can be expressed as

Π = 훼훽훾푠푠 > 1 (18).

The bottom left of Figure 2.13 shows how the transition to self-sputtering runaway can be detected using optical emission spectroscopy (OES). Below runaway, the spectrum

27 mainly features emissions from excited argon neutral while after runaway argon ions begin to dominate the spectrum.

Figure 2.13 (Top left) Schematic illustrating interactions of ions with the target surface. Adapted from [64]. (Bottom left) OES spectra above and below self- sputtering runaway. (Right) Illustration of gas-sputtering and self-sputtering. Reproduced with permission from [78].

Another important phenomenon during HiPIMS is gas rarefaction or a reduction in the gas density in front of the target. Gas rarefaction has previously been observed in

DC magnetron sputtering discharges [82,83] and can be responsible for reduced deposition rates. There are two primary mechanisms for reduction in gas density in front of the target during sputtering. The first is due to heating of the gas directly in front of the target surface caused by the applied power. As the gas directly in front of the target is heated, it expands. Because the pressure of the gas is fixed as the gas volume is increased the density in the heated area is reduced. The gas density in front of the target can also be reduced as the material sputtered from the target acts as a piston, pushing the gas away from the target [84]. The higher power densities in HiPIMS could be expected

28 to increase rarefaction effects [85]. Rarefaction in the HiPIMS discharge has been shown to be time-dependent and dependent on pulse length with the effect being more pronounced for short pulses (< 50 s) [86]. Palmucci, Britun, Konstantinidis and

Snyders investigated rarefaction in HiPIMS using time-resolved measurements of the sputtered particles velocity distribution function (vds) parallel to the magnetron surface

[87]. At the beginning of the pulse gas rarefaction was weak and had little effect on the velocity distribution or current. The current increased during the pulse and a broadening of the vds occurred as gas rarefaction effects increased and the sputtered particles experienced fewer velocity-reducing collisions. When rarefaction surpassed a threshold, the mean free path () of the sputtered species was comparable or greater than the plasma volume the current and the vds leveled out. Palmucci et al. referred to the time interval at which this threshold occured as the “rarefaction window’ [87]. The rarefaction window ended with the termination of the pulse and the FWHM of the vds decreased over time as the background gas ‘cooled’ and the depleted region was refilled [87].

Because self-sputtering, gas rarefaction and gas refilling all play an important role in the time-evolution of the target current, the time-evolution of the target current can be used to study the plasma physics of HiPIMS processes [76,85,88-90]. The target current,

IT, can be expressed as:

퐼푇 = 퐼푖 + 퐼푆퐸 = [1 + 훾푆퐸(퐸푃표푡푒푛푡푖푎푙)]퐼푖 (19) where Ii is the ion current and SE(EPotential) is the secondary electron yield as a function of the potential energy. Anders, Andersson and Ehiasarian previously established that the current at the target has contributions from both secondary electrons and from ions [76].

The behavior of the current over time is material dependent because the self-sputter yield,

29 secondary electron coefficient and collision cross-sections that dictate the previously described phenomena are all material specific. Anders et al. investigated the time- evolution of the target current for several materials including tungsten, aluminum and titanium (Figure 2.14). The initial delay in current onset observed in Figure 2.14 was attributed to the time required for formation of initial “seed’ electrons and the time needed for secondary electrons to ionize the argon gas [85]. After current onset gas rarefaction, gas refilling and self-sputtering dictated the shape of the current profile.

Materials like titanium (Ti), which are capable of being ionized to higher ionization states[91], can exhibit runaway of self-sputtering evident by the steep increase in the target current observed in Fig. 2.14c. For materials like tungsten (W) and aluminum (Al), which have lower self-sputtering yields, rarefaction causes a reduction in current after the initial peak. A steady state is eventually reached as sputtering and gas-refilling counteract the rarefaction losses and the current plateaus for the duration of the pulse on- time.

Figure 2.14 Time-resolved target currents for (a) tungsten, (b) aluminum and (c) titanium with different applied target voltages. Reproduced with permission from [76].

Now that a very basic understanding of the HiPIMS discharge with non-reactive gas has been established, the current understanding of how the addition of a reactive gas, in this case oxygen, affects the discharge will be discussed. During HiPIMS the higher peak power densities can result in short periods of target heating which can cause the implanted gas ions to diffuse into the target [92], resulting in a third possible target- reactive-gas reaction mechanism during reactive HiPIMS, ion-bombardment-assisted diffusion.

Reports on stability and hysteresis effects during reactive HiPIMS in an argon- oxygen environment have to-date been ambiguous. Stable, hysteresis-free deposition has been demonstrated for the reactive HiPIMS of zirconium oxide [93], alumina [94], niobium oxide [95] and titanium dioxide (TiO2) [96,97]. Other groups, however, have

31 reported strong hysteresis and the need for feedback controls to achieve stable deposition

[7,98,99]. Hysteresis during deposition is affected by target size [100], deposition geometry [100], oxygen flow rate [98]and pulse parameters [99] .

During reactive HiPIMS of oxides, coverage of the target with an oxide layer results in a change in the secondary electron yield (SE) [101] and subsequently the current-voltage characteristics. The shape and peak value of the discharge current has been shown to change depending on the degree to which the target is covered with the oxide layer [96,102,103]. An example of the change in current-voltage characteristic during the transition from a metallic target to a poisoned target is shown in Figure 2.15.

During reactive HiPIMS of TiO2, a steep increase in the slope of the time-resolved discharge current (Fig. 2.14a and 2.14b) is observed for the transition from sputtering of a metallic target to sputtering of an oxide target [103]. This increase in current slope during reactive HiPIMS in the reactive mode has been reported by several groups

[102,104] and is contrary to what is observed for DCMS. Because the secondary electron coefficient (SEEC) for Ti is higher in the metal mode than in the oxide mode, a decrease in the current would be anticipated if HiPIMS followed only the same mechanisms as

DCMS.

Figure 2.15 Pav = 4kW pulses, before and after the metallic to oxide transition for reactive HiPIMS of TiO2. (a) Current pulse profile for Ipeak = 160 A; (b) Current pulse profile for Ipeak = 400 A; (c) Voltage pulse profile for Ipeak = 400 A. Reproduced with permission from [103].

The plasma current density of Ti and Al targets sputtered in both metallic and oxide mode with pulses > 100 s, recorded by Aiempanakit, Aijaz, Lundin, Helmersson and Kubart [105], are shown in Figure 2.16. For both the Ti and Al targets in metallic mode, the discharge current quickly reaches its peak value, after the peak current is reached the Ti target in metallic mode decreases to a low-current regime. For the Al target in metallic mode the decay to a lower current is more gradual. In the oxide mode, the Ti target has a slow initial current onset then the current rises steeply to a maximum value until pulse termination. The Al target in oxide mode behaves similarly, though the initial current onset is faster. Both the Ti and Al have peak currents at least a factor of two greater in the oxide mode than in the metallic mode. Aiempanakit et al. attributed the slow current onset for the Ti target to a slower formation of ions. Ions likely form

33 more slowly as a result of the lower SEEC of oxidized Ti compared to metallic Ti. In the case of Al, the oxide has a higher SEEC. After current onset the shape of the discharge current is determined predominantly by gas rarefaction and self-sputtering [76,106].

Aiempanakit et al. analyzed the time-evolution of the discharge voltage and current and determined that a strong gas depletion regime is not reached when operating in oxide mode. A limited amount of sputtering due to the reduced sputter yield of the oxidized metal results in less collisional heating of the discharge gas reducing Ar losses through ionization [105,106]. More current-voltage studies combined with ion energy distribution measurements show that self-sputtering by O+ ions on the oxidized target increases the concentration of secondary electrons resulting in the higher discharge currents observed in the oxide mode [105].

Figure 2.16 Discharge current waveforms in the metal (Ar) and oxide mode of HiPIMS (Ar - O2) for (a) Ti and (b) Al targets. Pulse frequency was 50 Hz, pulse length was 300 s and the average discharge power was varied between 100 and 400 V. Reproduced with permission from[105].

Understanding the plasma physics during HiPIMS and how plasma dynamics affect film growth has been the focus of multiple studies [10,71,86,107-114].

Development and transport of the plasma from the target occurs in several stages on a timescale similar to the duration of the pulse [107]. Gudmundsson, Alami and

Helmersson determined that diffusion of the plasma is dominated by elastic scattering based on observations that the diffusion rates of the plasma through the chamber are strongly dependent on pressure [71]. Two waves of plasma arriving at the substrate have been identified [10,86,112,113]. The first wave is dominated by the background gas species while the second wave is comprised mostly of the metal or target species. The diffusion of the discharge strongly affects the ion energy distribution functions (IEDF) of all the species [107]. IEDFs of singly and doubly charged metal and gas ions have been measured [107]. The sputtered metal atoms are ionized by electron collisions in the dense plasma region after the sheath. Because ionization occurs after the sheath, the ions do not experience acceleration through the sheath and initially have the same energy as before ionization [107]. The target ions lose energy through collisions with gas and target ions and atoms before reaching the substrate.

Time-averaged ion energy measurements of the positive ions during HiPIMS of a

Ti and Al targets in metal and oxide mode were measured by Aiempanakit et al. [105].

The ion energy distributions for Ar+, Ti+, O+ and Ti2+ are presented in Figure 2.17. The intensity for all the Ar+ and Ti+ decreases when in oxide mode and the shape of the Ti+ and Ti2+ energy distributions changes. The energy distributions for the Al target (Fig.

2.18) show little change in the Ar intensity between the metal and oxide modes. In Al+ energy distribution the higher energy tail decays more quickly in the oxide mode than in the metal mode. The O+ energy distribution for both the Ti and Al targets in oxide mode have a narrow low energy peak followed by a high energy tail.

Figure 2.17 Time averaged ion energy distribution from metal and oxide modes for (a) Ar+, (b) Ti+, (c) O+ and (d) Ti2+. Reproduced with permission from[105].

Figure 2.18 Time averaged ion energy distribution from metal and oxide modes for (a) Ar+, (b) Al+ and (c) O+. Reproduced with permission from [105].

In addition to the positive ions previously discussed, negative ions during sputtering have been reported during conventional reactive sputtering processes [115-

118]. Recent literature on reactive pulsed sputtering processes has shown bombardment by negative ions during film growth affect both crystal structure development and defect formation. Negative oxygen ions in particular have been demonstrated to have significant effects on film growth [119]. Energy-resolved mass spectrometry during the reactive HiPIMS of TiO2 showed the presence of negative ions [120]. The time-averaged energy distributions of those negative ion species are shown in Figure 2.19a. Though

- - - - - TiO3 , TiO2 , TiO and O2 ions were present, the O ions were most prevalent. As previously seen in reactive DC sputtering three energy distributions are present for the O- ions [121]. The first population consists of low energy O- ions with energies corresponding to a small fraction of the target material. These ions are likely formed in the cathode sheath via dissociative electron attachment:

- - e + O2  O + O (20) and then accelerated with a fraction of the target potential [119]. The next energy distribution consists of O- ions created by spontaneous or collision induced dissociation

- of O2 ions [121]:

- - O2  O + O (21) with energy equal to half the target potential. The third population consists of highly energetic O- ions with energies corresponding to the full target potential sputtered from the oxide layer of the target [116,117]. Bowes, Poolcharuansin and Bradley. found the integrated intensities of the high-energy O- ions to be a function of pressure and target- substrate distance [120]. The high energy O- distribution was particularly sensitive to

37 pressure changes as can be seen in Figure 2.19b. At 0.33 Pa (2.48 mTorr) the intensity was several orders of magnitude higher than at 2.00 Pa (15.00 mTorr). Figure 2.19c shows that the population of O- ions decreased exponentially with increased pressure and/or target-substrate distance.

- - Figure 2.19 (a) Time-averaged negative ion energy distributions for TiO3 , TiO2 , - - - TiO , O2 and O ions during reactive HiPIMS. (b) Ion energy distribution for the high energy O- ions at different pressures. (c) Integrated intensity of the high energy O- population as a function of the pressure-distance product (pd). Reproduced with permission from [120].

There are currently no reports on energy-resolved mass spectroscopy for reactive

HiPIMS-deposition of ZnO. The presence of highly energetic O- ions has been reported for DC sputtering ZnO [116]. The differences in oxygen content and c-axis orientation observed by Howell et al. during DC sputtering of ZnO films on substrates positioned off-axis from the target center and at non-parallel angles to the target [122] are attributed

38 to energetic O- bombardment. Higher quality stoichiometric films were obtained when the substrate was positioned closer to the target center [122].

2.4.2 Film Growth during HiPIMS

Ionization of the sputtered material during HiPIMS allows for increased control of the average energy deposited to the film per atom. Enhanced-energy of the deposited material can reduce the surface free energy between the film-substrate interface resulting in smoother more uniform films. This reduction in film-substrate interface surface free energy can shift the growth from 3D nucleation to 2D growth by allowing eqn. (2) to be satisfied. The average energy deposited per atom, or E*, can be expressed as

퐽푖 < 퐸푑 >= 퐸푖( ) (22) 퐽푀푒 where Ei is the ion energy and Ji/JMe is the ratio of accelerated-ions to deposited-particle fluxes incident at the growing film surface [123]. The average energy deposited in the film per atom can be increased by increasing either component in eqn (22), however,

Petrov, Adibi, Greene, Hultman and Sundgren reported differences in film microstructure depending on whether Ei or Ji/JMe was increased [123]. In addition to reducing grain size and increasing nuclei density, ion bombardment of the growing films can promote specific crystalline orientations that allow the ion energy to be distributed over larger depths. These observations led Anders [124] to modify the structural zone diagram of

Movchan and Thornton to include the effects of energy on film microstructure. Figure

5.20 illustrates the effect of E* on film microstructure. Depositing flux with low energy results in porous tapered crystallites unless surface mobility is increased by elevated

39 substrate temperatures. As E* is increased the grain become denser and transitions from grains under tensile stress to compressively stressed grains.

Figure 2.20 Schematic of representation of dependence of film structure on energy deposited per atom at substrate and substrate temperature. Reproduced with permission from [124].

The effect of the ion and electron fluxes on film crystallinity and phase composition during reactive HiPIMS has been illustrated by several studies comparing

HiPIMS-grown TiO2 films to films grown with DC and pulsed DC magnetron sputtering.

Titanium oxide is an example of a material with applications that are very dependent on the microstructure of the film; anatase films have photocatalytic properties making it attractive for chemical-sensing and catalyst applications, while the higher refractive index and hardness of rutile TiO2 makes it useful for optical applications. In conventional

40 sputtering processes higher deposition temperatures are required to produce films with the rutile phase. Control of crystalline phase during HiPIMS growth has been repeatedly demonstrated for TiO2 [10,96,102,125,126] with the ratio of the anatase and rutile phases being dependent on the peak power [96], pressure [102], and substrate selection [10].

The effect of these deposition parameters on the energy of the fluxes to the substrate and how those fluxes affect film microstructure is discussed in detail in the proceeding paragraphs.

Aiempanakit et al. investigated the role of peak power on film growth during reactive HiPIMS of TiO2. The microstructure of TiO2 films reactively sputtered onto unheated unbiased Si (100) substrates using reactive HiPIMS DC magnetron sputtering was compared with that of films grown with DC magnetron sputtering (Fig. 2.21a) [96].

The films deposited with DCMS were amorphous. The films deposited with HiPIMS were crystalline with anatase (101) dominated orientation at low peak powers. Increasing the peak power resulted in a decrease in and anatase phase and an increase in the rutile

(110) orientation. The change in crystallinity and phase composition of the TiO2 films seen in Fig. 2.21a is attributed to differences in the energy of the particles arriving at the substrate. For the DCMS deposition the energy is insufficient to produce crystalline films. The HiPIMS-deposited films were all crystalline with the phase composition being dependent on the peak power. The shift toward rutile dominated films is attributed to the increase in energy for Ti+ and O+ ions seen in Figs 2.21b and 2.21c. Though not measured in Aiempanakit et al.’s study, O- ions are expected to also contribute to determining the TiO2 films’ microstructure [125].

Figure 2.21 (a) XRD of TiO2 film deposited on unheated unbiased Si (100) substrates using DCMS and HiPIMS with different peak powers. (b),(c). Energy distribution functions for (b) Ti+ and (c) O+ ions for the DCMS and HiPIMS discharges. Reproduced with permission from [96].

- The effect of O ion bombardment on the microstructure of TiO2 films has been studied by Amin, Kohl and Wuttig [125]. The microstructural inhomogeneities, seen in

Fig. 2.22b were attributed to a lateral distribution of the O- bombardment on the substrate. The distribution of the O- bombardment was determined by the O-‘s trajectory which was found to be dependent on the depth and shape of the racetrack [125]. For newer targets with undefined racetracks the O- ions are accelerated normally from the target face. The change in the O- ion trajectory as the target erosion track deepens is illustrated in Figure 2.22a. In the atomic force microscopy (AFM) micrograph in Fig

2.21b and the x-ray diffraction (XRD) in Fig 2.22c it can be seen that bombardment by O- resulted in a mixed distribution of grain sizes and a promotion of the rutile phase.

Atomic force microscopy (AFM) and XRD of films grown from an old target showed bimodal growth and promotion of the rutile phase at the sample center corresponding with the anticipated trajectory of the O- ions [125].

Figure 2.22 (a) Diagram of the trajectory of highly energetic O- ions sputtered from the target. The black arrows indicate the trajectory from a new target surface with an undefined erosion track. The grey arrows are the trajectory from an old target surface with a well-define erosion track. (b) AFM micrographs of the surface of TiO2 films grown by reactive HiPIMS of a new Ti target surface. (c) Grazing incidence XRD of TiO2 films deposited with reactive HiPIMS of a new target. Reproduced with permission from [125].

The effect of pressure and oxygen partial pressure on TiO2 microstructure is shown in the XRD in Figure 2.23 from Stranak et al.’s work [102]. Lower oxygen partial pressure resulted in -Ti films. Depositions below 2.00 Pa (15.00 mTorr) resulted in

43 rutile films when there was sufficiently high oxygen partial pressure for oxygen incorporation in the lattices. Based on Bowes et al.’s study [120] it can be assumed that these depositions had a higher incidence of O- bombardment, imparting more energy into the substrate and promoting the more energetic rutile phase. At higher pressures the TiO2 phase became dependent on the O2 partial pressure. Low oxygen partial pressures resulted in amorphous films, while high pressures with high oxygen partial pressure resulted in anatase films.

Figure 2.23 (a) Grazing incidence XRD and (b) O2/Ar and pressure dependent phase information for TiO2 films deposited with HiPIMS. Reproduced with permission from [102].

CHAPTER III

RESEARCH OBJECTIVES

Analysis of literature on ZnO TFT devices has shown the need for crystalline films with a high degree of (002) alignment, close-packed columnar grains and low surface roughness. For incorporation of ZnO TFTs into flexible electronics low- temperature (< 200 oC) scalable processing of films with the desired structure is needed, prompting the development of alternative techniques for synthesis of ZnO. The high fraction of ionized species at the substrate, with relative abundances and energies are dictated by deposition parameters, during HiPIMS growth make it a promising technique for low-temperature growth of ZnO with controlled microstructures and crystallographic orientations.

The overall goal of this study was to demonstrate the applicability of reactive

HiPIMS for scalable low-temperature synthesis of ZnO films for TFT applications. This study focused on growth in the temperature range applicable to processing of TFT on temperature sensitive flexible substrates, highlighted in Figure 2.3, and identified deposition conditions that produced ZnO films with highly aligned (002) oriented crystals, dense columnar grains, low surface roughness and minimal presence of mixed orientations at the film-substrate interface. Additionally this study established film growth mechanisms based on correlation of plasma parameters and the resultant film

45 structure to determine the processing conditions needed to synthesize ZnO films with the desired microstructure. Growth from a metallic Zn target and a fully oxidized ZnO target will be investigated to determine the effect of oxidation state of the target and high voltage-high current versus high voltage-low current conditions on the plasma and film growth. The goals of this study are achieved by the following objectives:

A. Establish correlations between deposition parameters (target voltage, pulse parameters, gas pressure and oxygen ratio) and plasma conditions (peak target current, ion species and energy at the substrate) for the HiPIMs discharge from a metallic Zn target and an compound ZnO target in O2-Ar gas

The reactive-HiPIMS process will be characterized during the synthesis of ZnO by first establishing a range of deposition parameters (i.e. total pressure, oxygen partial pressure, target potential, pulse length, pulse frequency) for stable operation. The stability of the process was determined using target current-voltage measurements, visual examination of the plasma and observing plasma-heating of the substrate. Once stable regimes had been identified, an electrostatic quadrupole mass spectrometer was used to identify the mass-to –charge ratio of the species in the flux arriving at the substrate surface. Energy-resolved mass spectroscopy was performed to measure the energies of the ions impinging on the substrate surface. These measurements were correlated to time-resolved target current measurements to determine the effects of gas volume phenomena, such as gas rarefaction and refilling, and target surface phenomena (self- sputtering and changes in SEEC due to oxide formation) on the fluxes to the substrate.

B. Find film growth mechanisms based on concentrations and energies of ionized species at the substrate surface

The effect of deposition conditions on film microstructure was determined by analyzing the crystallinity of the film using XRD and TEM and the morphology of the film surface using AFM. Wide angle XRD measurements were used to determine crystallographic orientations present in the film and for grain size calculations. Rocking curve measurements of the ZnO (002) diffraction peak were measured and the full width at half maximum of the RC peak was used to determine the degree of alignment of the

(002) oriented crystallites. Scanning electron microscopy (SEM) and TEM were used to analyze the film cross-section. The plasma characteristics were then correlated with resulting microstructures. Fundamental nucleation and growth models were considered to determine the role of process parameters on film microstructure and establish film growth mechanisms based on the concentrations and energies of ionized species at the substrate surface.

C. Determine the effect of the directionality of energetic sputtered particles on microstructural uniformity during large area growth

ZnO films were grown on large (4-inch diameter) wafers and the microstructure analyzed at different distances from the substrate center to determine the effect of the directionality of energetic ions on film growth.

D. Demonstrate reactive HiPIMS processing of ZnO TFT channels

HiPIMS-deposited ZnO channels will be incorporated into back-gated TFT devices and the I-V characteristics of the device measured to demonstrate feasibility of this processing technique.

CHAPTER IV

EXPERIMENTAL METHODS

4.1 The Deposition System

In this study all experiments were performed under ultra-high vacuum in the custom system shown in Figure 4.1. The system consisted of three separate zones, all kept under vacuum. The main chamber, the sphere to the left in Figure 4.1a, was the region where the film depositions and process characterization were performed. This chamber was pumped down to base pressures of 10-8 Torr or lower using a turbo molecular pump backed by a mechanical roughing pump. The pressure in this chamber was controlled with a butterfly valve and mass flow controllers tubed to argon, oxygen and nitrogen. To the right of the deposition chamber there was a load-lock chamber. The load-lock chamber allowed for increased throughput of samples, as the chamber was equipped with a sample rack that allowed up to five samples and a blank to be loaded at one time. The chamber’s small volume allowed the chamber to be quickly evacuated to pressures < 10-6 Torr. A port valve 90o from the gate valve to the deposition chamber led to a small analytical chamber for x-ray photoelectron spectroscopy (XPS) measurements.

Transfer forks allowed samples to be moved between the load-lock, deposition chamber and analytical chamber without breaking vacuum.

Figure 4.1 (a) Photograph of the vacuum chamber used in the experiments. Schematic of the (b) interior and (c) exterior of the deposition chamber.

The vacuum chamber was equipped with 1.3” and 2” unbalanced planar magnetron guns positioned with the target surface normal at an angle of 45o to the substrate normal. The distance from the target face to the substrate could be changed, without affecting the angle, by sliding the gun farther into the chamber. The plasma was driven with a Huettinger HMP-1200 HiPIMS power supply controlling the voltage applied to a 2-in. diameter ZnO (Fig. 4.2a) or a 1.3-in. diameter Zn target (Fig. 4.2b).

The HMP-1200 supply was capable of applying 200-2000 V pulses of 10-200 s at frequencies 10-250 Hz. The 0.25“ thick Zn target was magnetically mounted to the magnetron by an iron keeper mechanically bonded to the target back (Fig. 4.2c). The

0.125” thick ZnO target was bonded to a 0.125” thick copper backing plate (Fig. 4.2d).

An iron keeper was mechanically bonded to the copper backing plate and used to hold the target to the magnetron.

Figure 4.2 Photographs of sputtering face of (a) ZnO target and (b) Zn target and the mounting side of the (c) ZnO target and (d) Zn target.

The substrate holder in the deposition chamber rotated the samples promoting uniformity. The substrate holder allowed the substrates to be held at a floating bias, ground or a DC or pulsed DC bias. The minimum bias that the substrates could be held at was dependent on the pulse conditions during the deposition. A tungsten wire filament could be used to resistively heat the samples and the temperature measured with an IR inframeter.

4.2 Process Characterization and Plasma Diagnostics

4.2.1 Sputtering Yield Calculations

The effect of target oxidation and the energy of the argon ion bombarding the target on the sputtering yield was investigated using a Monte-Carlo program called

“Transport of Ions in Matter’ or TRIM. The TRIM program was contained in a larger set of programs called “Stopping and Range of Ions in Matter” or SRIM. The TRIM/SRIM

50 programs used a quantum mechanical treatment of ion-atom collisions to calculate interactions of energetic ions with atoms in an amorphous target.

While the TRIM/SRIM program was capable of calculating a range of target-ion interactions, for this study only sputtering yields were investigated. The set up screen for the TRIM/SRIM program is shown in Figure 4.3. For calculation of the sputtering yields,

“Monolayer Collision Steps/Calculation of Surface Sputtering” was selected from the drop-down menu of different types of TRIM calculations. These calculations require the ion to have a collision in each monolayer of the target and for each collision to be calculated without approximations. After selection of the type of TRIM calculation, information of the impinging ions and the target are input. The ion species is selected from the periodic table menu (the PT button in Figure 4.3). After the ion species is selected the program adds the atomic number and mass of the ion. For this study argon ions with energies between 0.30 keV and 1.5 keV were investigated. The chosen argon ion energies correspond to the applied target voltages of 300-1500 V used in this work.

The angle of incidence of the ion onto the target was set to 0o, making the argon ions strike the target normal to the target surface.

The other important inputs for the TRIM program were for the target. The program allowed the target to be composed of multiple layers. This was very useful for this study, where during the study of reactive sputtering of the Zn target a layer of ZnO formed on the target surface. For the studies involving the ZnO target, the target was considered to be one layer with a width of 0.125” (3.175 mm), the thickness of the target.

An estimated density of 4.283 g/cm3 and surface binding energies of 1.35 eV and 2.00 eV were used for the Zn and O, respectively. For the Zn target two target layers were used.

The bottom layer consisted of the Zn target and was assigned a width that matched the thickness of the target, 0.125” (3.175 mm). A density of 7.14 g/cm3 was used. The top- layer was assumed to be ZnO and estimated to have width of 1 m based on calculations of the thickness of the oxide layer formed during reactive sputtering [127].

Figure 4.3 Screen-capture of the TRIM/SRIM software input menu.

After the ion and target inputs were entered the TRIM calculations were run for

99,999 ions bombarding the target. Once the calculations were complete, the sputtering yield was plotted as a function of the final energy of the sputtered atom as is left the target.

4.2.2 Energy-Resolved Mass Spectroscopy

The mass and ion energy distributions (IEDs) of the flux arriving at the substrate were measured using an energy-resolved Hiden EQP 1000 electrostatic quadrupole

(EQP) mass spectrometer, shown schematically in Figure 4.4. Because system height

52 limitations of the vacuum chamber shown in Figure 4.1 prevented the EQP probe from being mounted with the detector head at the substrate position, the mass spectrometry and

IEDs were measured in the chamber shown in Figure 4.4. For these measurements the

Zn and ZnO targets were mounted on magnetron guns positioned 10 cm from the grounded orifice of the EQP. Figure 4.4b shows the arrangement of the chamber during the IED measurements. Mass spectrometry scans were taken from 0.40 amu to 95.00 amu with step sized of 0.50 amu while systematically varying the applied voltage, pulse parameters, total pressure and oxygen partial pressure. IEDs were measured for mass-

+ + + + charge ratios of 16 amu/e (O ), 32 amu/e (O2 ), 40 amu/e (Ar ), 65 amu/e (Zn ) and 80 amu/e (ZnO+) at the same conditions as the mass scans. For the IED measurements the scans went from 0 eV to 90 eV with steps of 0.50 eV. Both the mass scans and IEDs were taken with a dwell time of 300 ms so each data point for the standard pulse conditions (200 s; 100 Hz) was composed of 30 pulses.

Figure 4. 4 Schematic of (a) exterior and (b) interior of the deposition chamber with the EQP installed.

4.2.3 Target Current Measurements

The current at the Zn and ZnO targets was measured while systematically varying the applied voltage, pulse parameters, total pressure and oxygen partial pressure. The measurements were taken using a Tektronix TM502A current probe and recorded with a

Tektronix TDS 5400 oscilloscope. A Tektronix P5200 high voltage differential probe was used to trigger the current measurements on the oscilloscope; this ensured that the recorded currents could be time-resolved and that the evolution of the current could be correlated to the pulse initiation and termination.

4.3 Substrate Selection and Preparation For all the depositions except for the growth over large-area, the ZnO films were deposited onto conductive silicon wafers (= 1-10 cm) with a 100-300 nm thick thermal SiO2 layer. The SiO2/Si substrate system was chosen because the dielectric SiO2 on conductive Si facilitated the fabrication of back-gated devices. The silicon substrate was treated as the gate to fabricate back-gated devices with the SiO2 layer acting as the gate dielectric. Additionally, the amorphous SiO2 provided no template for ZnO crystal growth.

The substrates were cleaved into 1 cm2 pieces then cleaned using electronic-grade solvents following the procedure below:

1. Sonicated in acetone (15 minutes)

2. Rinsed with methanol

3. Sonicated in methanol (5 minutes)

4. Rinsed with ultra-pure H2O

5. Dried by blowing with lab nitrogen

The clean substrates were then individually mounted to a stainless steel disc using double-sided carbon tape. A small section of the substrates was masked with Kapton tape, allowing for thickness measurements and access to the underlying silicon substrate for back-gated device fabrication. The substrates were then placed in the load-lock shown in Figure 4.1a.

For the studies investigating the spatial uniformity of the depositing flux and its effect on film microstructure, ZnO films were deposited onto 4-inch diameter conductive silicon wafers with a 50 nm thick chemical vapor deposited (CVD) SiO2 layer. These substrates were rinsed with ultra-pure methanol and deionized water then blown with nitrogen before being loaded into the deposition chamber.

4.4 Zinc Oxide Deposition For the initial investigation of the effect of deposition parameters on film growth six sets of films were grown from each target. For each of the six sets of films one deposition parameter (applied voltage, pulse length, pulse frequency, total pressure and oxygen partial pressure) was changed while the other parameters were held constant. The conditions used for film growth from the Zn target are compiled in Table 4.1 while those for the ZnO target are shown in Table 4.2. During these depositions the target-substrate distance was fixed at 10 cm and the substrates were not heated. Films were initially grown for one hour. The deposition rate was calculated by

푡 퐹 = 푎푣푒푟푎푔푒 (23) 푑푒푝표푠푖푡푖표푛 푡푖푚푒

55 using the average thickness. All depositions were then repeated with the total deposition time for each sample adjusted to obtain films approximately 100 nm thick (+/-10%).

Table 4.1 Parameters for ZnO films deposited by reactive-HiPIMS of a Zn target.

Table 4.2 Parameters for HiPIMS-deposition of ZnO deposited by reactive-HiPIMS of a ZnO target.

After analysis of the microstructure of the films grown with the conditions outlined in Tables 4.1 and 4.2, more detailed studies into the nucleation and microstructural evolution of films deposited from the ZnO target were performed by growing four more sets of ZnO films over a wider pressure range with emphasis on growth with pressure lower than 10 mTorr. The conditions for this set of films are shown in Table 4.3. Films were grown with the substrate-target distance at the standard 10 cm and with it shortened to 5 cm to determine the effect of plasma density on film growth.

For this study the substrate temperature was held at 150 oC which was chosen for its applicability to synthesis of ZnO on flexible substrates. In addition to the SiO2/Si substrates, 100 nm of ZnO was deposited on Kapton substrates to demonstrate the feasibility of HiPIMS-processing on temperature-sensitive polymer substrates.

Table 4.3 Deposition conditions for ZnO samples for optimized microstructure study.

To investigate spatial inhomogeneity of the depositing flux and its subsequent effect on film growth and microstructure over large areas, ZnO films were grown on the

4-inch diameter samples. Due to the different sizes of the Zn and ZnO targets, 1.3-inch and 2.0-inch diameters, respectively, the geometries of these experiments were slightly different for these studies, as illustrated in Figure 4.5. The deposition conditions for this

57 experiment are listed in Table 4.4. Because the heating system was optimized for 1-inch diameter substrates, the samples were unheated during these depositions to avoid thermal gradients across the substrate surface. The target-substrate distance listed in Table 4.4 was measured from the center of the target face to the center of the substrate.

Figure 4.5 Schematic diagram of the experimental geometry for synthesis of ZnO over a 4-inch diameter area from (a) a 2-inch diameter ZnO target and (b) a 1.3- inch diameter Zn target.

Table 4.4 Deposition conditions for ZnO growth on 4-in. diameter SiO2/Si substrates.

4.5 Microstructural Characterization

4.5.1 Thickness Measurements

A small section of the substrate was masked during deposition, allowing for measurement of film thickness. Two methods, contact and optical, were employed to determine film thickness. For the contact measurements, a KLA Tencor P-15 stylus profilometer with a 2 m diameter probe tip was used to scan the film surface and measure the height difference between the film and exposed substrate. A Wyko white light profilometer was used for optical thickness measurements. The white light profilometer determined the film from the inference oscillations in the reflected light intensity.

4.5.2 X-Ray Diffraction Analysis

Film crystallinity and orientation were analyzed with x-ray diffraction using a

Rigaku Smartlab XRD system (Fig 4.6) with a Cu anode. Beam divergence was minimized using a 1 mm divergence slit and a 0.5o parallel slit analyzer. Parallel beam

2/ wide angle XRD scans were taken from 2 = 20-60o with 0.002o steps at scan rate of 0.2o/minute to determine the crystal orientations present in the films. The XRD patterns were fitted using a Pearson VII shape function to determine peak FWHM and integrated intensity. The crystal size in the films was determined using the Scherrer equation:

퐾휆 휏 = 푥−푟푎푦 (24) 훽푐표푠휃

59 where  is the mean size of the crystallite domain corresponding to the diffraction peak,

x-ray is the x-ray wavelength, K is a dimensionless shape factor (in this case 0.9), b is the line broadening or FWHM of the diffraction peak and  is the Bragg angle. For the

SMARTLAB system used in these experiments x-ray = 1.54 Å.

Figure 4.6 Images of the Rigaku SmartLab used for X-ray Diffraction analysis of the films. (Left) exterior image and (Right) interior view.

The degree of alignment of the crystal orientations identified in the wide angle

XRDs were measured using rocking curve (RC) measurements. The same optics and beam alignment used in the wide angle scans were used in the RC measurements. The x- ray source was scanned from 4o to 32o during these measurements while the detector angle was set to = 17.0o to 17.3o, angles corresponding to the maximum intensity of the

ZnO (002) peak (34.0o to 34.6). The FWHM of the RC scans were determined and used to determine the range of misalignment between the crystal plane normal and the substrate normal.

4.5.3 Electron Microscopy

Film cross-sections were imaged using a Sirion scanning electron microscope

(SEM) with a through-lens-detector, an acceleration voltage of 5.0 kV and working distances of 4.5-7.0 mm. The samples were cleaved by first scoring a line on the back side with a diamond scribe then snapping the substrate along the scored line. The scored- samples were adhered to an aluminum puck using double-sided carbon tape. The puck was then mounted into the SEM system with a 45o substrate holder so that the through- lens detector looked down on the cleaved surface of the sample.

A FEI Nova focused ion beam (FIB) microscope equipped with an Omniprobe manipulator was employed to prepare transmission electron microscope (TEM) specimens to analyze the cross-sectional microstructure of the films. The FIB microscope was operated using 5 kV electron beams and 30 kV Ga+ ion beams. High- resolution microstructures were observed using a FEI TitanTM 80-300 S/TEM, which was equipped with a Cs-corrector and was operated at 300kV. Energy-dispersive X-ray spectroscopy and electron energy loss spectroscopy were employed for chemistry measurements in the TEM.

4.5.4 Atomic Force Microscopy

Surface features and roughness were examined with atomic force microscopy in tapping mode with a scan rate of 0.5 Hz using an Asylum MFP-3D AFM with an Asylum

AC160TS Si cantilever. The cantilever had a resonant frequency of 300 kHz and a nominal tip radius of 9 ± 2 nm.

4.6 Electrical Characterization

4.6.1 Device Fabrication

Back-gated FET devices were created with HiPIMS-grown ZnO as the channel material using the silicon substrate as the gate and the thermal SiO2 layer as the gate- oxide. The SiO2 that had been covered by Kapton tape during deposition was removed using a diamond scribe to expose a section of the Si substrate. Aluminum source, drain and gate contacts (all 200 nm thick) were deposited using the shadow mask (shown in

Figure 4.7a) and a Denton electron beam evaporator system. The shadow mask had 56 devices on it with channel lengths ranging from 50-100 m and channel widths ranging from 500-1000 m.

Figure 4.7 (a) Schematic of the shadow mask used for device fabrication. (b) Schematic of the geometry of the back-gated transistor devices used for electrical characterization. (c) Micrograph of Al source and drain contacts on ZnO film.

4.6.2 Preliminary Back-Gated TFT Device Testing Preliminary electrical characterization was performed using a Keithley 4200

Semiconductor Characterization system. The I-V characteristics of the drain-source contacts were measured while sweeping the drain voltage from -4 V to 50 V and stepping the gate voltage from -40 V to 80 V.

CHAPTER V

RESULTS AND DISCUSSION

5.1. Process and Plasma Characterization

5.1.1 Reactive HiPIMS from a Zinc Target

The sputtering yields calculated using the SRIM/TRIM program are shown in

Figure 5.1 plotted as a function of the Ar+ energy. It is important to note that Figure 5.1 only takes sputtering from argon gas ion into consideration. It neglects the sputtering yields of sputtering from oxygen ions or self-sputtering. The SRIM calculations were run assuming ZnO through the whole target thickness (for the ZnO target) and for a 1 m thick ZnO layer on a Zn target. The results were the same for both scenarios. The sputtering yield for Zn from the oxidized target was lower than that from the pure metal target. For 450 eV Ar+ the yield was 1.65 atoms/ion for the oxidized target compared to

4.84 atoms/ion for the pure metal. At 1500 eV the yields were 2.97 atoms/ion and 9.16 atoms/ions for the oxide and metal targets, respectively. The sputtering yield of the oxygen was lower than that of the Zn, 0.90 atoms/ion and 1.60 atoms/ion at 450 eV and

1500 eV, respectively.

Figure 5.1 Calculated sputtering yields as a function of argon ion energy for (a) Zn and (b) ZnO targets.

The time-resolved currents for the zinc target at different constant target voltages are compiled in Figure 5.2a. During the current measurements for Figure 5.2a, the pressure was held at 10 mTorr with an O2/Ar ratio of 0.40 and the voltage pulses were

200 s long at a frequency of 100 Hz. At applied voltages lower than 450 V, the plasma was unstable and negligible current ( < 1 mA) was measured. Above 450 V, the general character of the current shape was constant for the different target voltages. The shape of the current waveforms can be divided in several characteristic regions corresponding to plasma initiation, steady state, and pulse-off times. At pulse initiation, the current rises as argon ions bombard the layer of zinc oxide formed on the target surface causing the ejection of electrons, oxygen atoms/ions and zinc atoms/ions. The ejected electrons and ions contribute to the current resulting in an increase in current until a peak value is reached. As the near target density of sputtered atoms increases, gas rarefaction occurs and the flux of argon ions at the target surface is reduced [128,129], resulting in a decrease in the target current after the initial peak. The current decayed until a steady state current is reached. Once the pulse terminated the current quickly decayed to zero.

The magnitude of the peak and the steady state currents were voltage-dependent.

Increasing the target voltage from 450 V to 500 V resulted in the peak current increasing from 6.6 A to 15.5 A and the peak value was reached 24 s earlier. The increase in peak current was less dramatic when the voltage was increased above 500 V, however, the time required to reach the maximum current value continued to decrease at higher voltages.

Figure 5.2 Time-resolved current of zinc target at 10 mTorr with O2/Ar = 0.4 with different pulsing parameters; (a) applied voltage, (b) pulse length, (c) pulse frequency.

The time-dependent evolution of the current is further illustrated by looking at the shape of current profiles for different pulse lengths (Fig. 5.2b). With the conditions used in these measurements, 20 s was the shortest pulse length that resulted in a stable plasma, however, the pulse terminated before the peak current value is reached. When the pulse length was increased to 50 s, the pulse was sufficiently long for the peak target current to be reached and for rarefaction effects to begin to reduce the current. Pulse lengths of 50 s were too short for the steady state current to be reached. The peak current for the 50 s pulse (Ip = 24 A) was larger than the value for the other pulse lengths. The peak current decreased for longer pulse lengths. Longer pulse lengths result in less times for ZnO (which has a larger SE than Zn) to form on the target surface before being sputtered off by the next pulse. For the 50 s pulse there are 9.95 ms before the next pulse compared to 9.90 ms for the 100 s pulse and 9.80 ms for the 200 s pulse.

The third pulsing parameter investigated was the pulse frequency, which dictates the time between pulses. The time between pulses is particularly important during reactive sputtering because it affects how large of a poisoned layer can develop between pulses. The target currents for 200 s pulses of 500 V with different pulse frequencies are compiled in Figure 5.2c. At low pulse frequencies the plasma was unstable, apparent by observing the strobe-like pulsing of the plasma. The current profile for pulses with frequencies less than 40 Hz was unstable and the current profile shifted between beginning at pulse initiation and rising after a 50 s delay. The current profile at 50 Hz had a higher peak and steady state current. At higher pulse frequencies the general shape and time-evolution of the current profile remained the same, however, the peak current value decreased as the time between pulses increased. A shorter time between pulses

67 with higher pulse frequency allowed less time for oxide formation on the target surface resulting in a reduction in the current.

The effects of gas pressure and oxygen partial pressure (controlled by changing the oxygen flow rate) on target current are illustrated in Figure 5.3. For the current measurements shown in Figure 5.6a the flow rates for Ar and O2 were fixed at 25 sccm and 10 sccm, respectively while 650 V was applied to the target in 200 s pulses with a frequency of 100 Hz. The general shape of the current profile remained constant as the pressure was increased, however, the magnitude of the peak and steady state currents increased at higher pressures. This is likely because higher pressures resulted in more gas ions at the target surface contributing to sputtering and the target current. Higher pressures had no impact on the time required for the peak current to be reached.

Figure 5.3 Time-resolved current of zinc target at (a) different total pressures with O2/Ar = 0.4 and (b) a fixed pressure of 20 mTorr and varied oxygen flow rates with a fixed Ar flow rate of 25 sccm with 200 s pulses of 650 V at a frequency of 100 Hz.

The general shape of the current profile changed as the oxygen gas content was increased. When the process was run without the addition of oxygen, pure zinc metal

68 was sputtered. The target current in the pure argon was miniscule. When 1.0 sccm of oxygen was flowed into the chamber the plasma stabilized and a measurable target current was observed. The increase in target current with the addition of oxygen is likely due to the higher SE of ZnO (SE < 2.96) compared to that of Zn (SE < 1.10). The peak and steady state current values increased as more oxygen was added to the system. The initial current peak was more pronounced at the higher oxygen ratios.

The results for the mass spectrometry scans of the plasma at the substrate position with different applied voltages are compiled in Figure 5.4. For all measured voltages

+ + there were peaks at 16 amu/e, 32 amu/e and 40 amu/e, corresponding to O , O2 and Ar+ respectively. Although the peaks at 16 amu/e and 32 amu/e could contain contributions

2+ 2+ 2+ from O2 and Zn , respectively, it can be inferred from the low intensity of the Ar peak at 20 amu/e, values of <2500 C/s which is significantly lower than the expected value of 18% the intensity of the 40 amu/e peak for Ar+[130], that the concentrations of doubly charged ions will be small. At 600 V and above peaks at 64-67 amu/e emerge as the applied voltage was high enough for ionization of Zn atoms to occur. For most voltages (300-600 V) the highest intensity peak was for the molecular oxygen ion. The second most intense peak was for Zn+. The least intense ion was Ar+. The intensity of the mass peaks for the gas ions trended with the ion’s ionization energies. The ion with the least intense peak, Ar+ had the highest ionization energy (15.76 eV). The ionization energies for Zn+, O2+ and O+ were 9.39 eV, 12.1 eV and13.62 eV, respectively.

+ + Although Zn had the lowest ionization intensity, the intensity of the O2 was higher due to the larger concentration of gas and sputtered oxygen available. An increase of several

69 orders of magnitude was observed in the intensity of the Zn+ when the voltage was increased to 700 V.

Figure 5.4 Area plot of the mass spectra of the plasma from reactive sputtering of a Zn target in 20 mTorr O2/Ar = 0.4 with different applied voltages with 200 s pulses at 100 Hz.

+ + + + + The IEDs for Ar , O , O2 , Zn and ZnO for different applied target currents are compiled in Figure 5.5. For Ar+ at voltages of 450 V or lower the plasma was weak and there were few energetic ions. At 500 V, although there are low counts, ions begin to be evident. At 600 V Fig 5.5a shows a strong single peak with an average energy of 4 eV.

The IED for 700 V has a low energy peak at 3 eV and a second population with an average energy of 5.5 eV. The lower energy peak is attributed to ions that have experienced collisions with thermalized gas atoms and ions [107]. The intensity of the lower energy peak is lower at 700 V than at 600 V. The reduction in counts at 700 V is

70 likely a result of the thicker plasma sheath (s = 0.28 cm and 700 V vs. s = 0.26 cm at 600

V) increasing the probability of Ar+ experiencing energy-reducing collisions.

+ + + + Figure 5. 5 IEDs for (a) Ar , (b) O , (c) O2+, (d) Zn and (e) ZnO measured in a reactive-HiPIMS plasma with a Zn target with 20 mTorr and Opp = 8 mTorr.

The IED for O+ in Fig. 5.5b shows two peaks with average energies of 2 eV and 5 eV. At 500 V the IED had an intensity of 5.0  104 counts/second at 1 eV and 1.0  105 counts/second at 3 eV. The intensities of both populations increase with higher applied

+ + voltages. The IEDs for O2 are compiled in Figure 5.5c. For O2 the highest intensity was observed for 600 V while the lowest occurred at 700 V. At 500 V there was a

+ smaller population of O2 ions with energy of 2 eV and a larger population with 5 eV.

These energies correspond to the O+ energy distributions at 500 V. The lower ionization

+ + of O2 (12.1 eV) compared to that of O (13.62 eV) accounts for the order of magnitude

+ higher counts observed for the O2 . At an applied voltage of 600 V the intensity of the

+ + O2 IED was still an order of magnitude higher than that of the O . At 700 V, however,

+ 5 the intensity of the O2 IED decreased to 5.0  10 counts/second, values equal to those

+ + observed for the O IED at that voltage. The reduction in O2 at 700 V could be due to dissociation of the molecular oxygen.

The IEDs of the sputtered species are shown in Figure 5.5d-e. For both the Zn+ and ZnO+ the counts were low until the applied voltage reached 700 V. The IED for Zn+ shows a small population with average energy 2 eV and a much larger population with

3.5 eV energy. The ZnO+ IED also showed two average energies, one at 3 eV and another at 5 eV. The intensity of the Zn+ IED was two orders of magnitude larger than that of the ZnO+; this is understandable given that sputtering occurs primarily on an atomic level so there will be significantly more Zn atoms available for ionization than

ZnO clusters.

The effect of pressure on the individual IEDs is illustrated in Figure 5.6. The applied voltage was held at 600 V for all measurements. At 15 mTorr the IED for Ar+

(Fig. 5.6a) showed a broad distribution with counts around 1500 C/s and an average energy of 2 eV. At 20 mTorr the SEM counts increased to > 2  104 C/s showed two populations of ions with average energies of 1.5 eV and 4 eV at 20 mTorr. The energy distribution widened at 30 mTorr and increased in intensity. These results are as expected as higher pressures mean more gas atoms are present and increases the probability of ionizing collisions.

The IEDs for oxygen ions for different total pressures are compiled in Figures

5.6b and 5.6c. The O+ IED had an intensity < 100 C/s at 15 mTorr. When the pressure was increased to 20 mTorr the IED intensity increased and two populations of ion energies are observed. The lower energy peak at 1 eV had an intensity of 5  104 C/s while the peak at 4 eV had a higher intensity of 9  104 C/s. At 30 mTorr the IED intensity remained at 9  104 C/s though the energy shifted to and average value of 3 eV.

The lower energy peak at 1 eV is not present at 30 mTorr; however there is a shoulder on

+ 6 the IED at 6 eV. The intensity of the O2 IED (Fig. 5.6c) had an intensity of 10 C/s for

+ pressures above 15 mTorr. The higher intensity of the O2 IED compared to that of the

+ + O can be attributed to the lower ionization energy of O2 and the energy required to dissociate molecular oxygen. The shift toward lower energies observed in Fig. 5.6b and

Fig. 5.6c can be attributed to the shortening of the oxygen ions mean free paths at higher pressure. The mean free path () of a particle in plasma can be calculated from

1 휆 = 2 (25) 푛푔휋푎

where ng is the gas density, calculated from the pressure using the ideal gas law and a is the atomic radius of the particle. For the study, the mean free path of O+ at 20 mTorr was

1.56 cm. At 30 mTorr  is shortened to 1.04 cm. The shorter  means a higher probability that the ion will experience an energy reducing collision.

+ + + + + Figure 5.6 IEDs for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-HiPIMS plasma with a Zn target with different total pressures and O2/Ar = 0.40.

The IED for Zn+ and ZnO+ are shown in Figures 5.6d and 5.6e. It is important to note that the IEDs in Figure 5.6 were taken with 600 V pulses applied to the target. In the voltage study at 20 mTorr in Figure 5.5, measurable Zn+ and ZnO+ IEDs were not seen until the applied voltage reached 700 V. With an applied voltage of 600 V, Zn+

IEDs were measured when the total pressure was increased to 30 mTorr. The counts of the Zn+ IED were very low with a peak maximum value of 600 C/s at 2 eV and a lower peak of 300 C/s at 4 eV. The presence of Zn+ at higher pressures and lower voltages could be attributed to the higher pressure resulting in more sputtered Zn and more gas ions which increase the probability of ionizing collisions occurring. This results in the ionization of some Zn atoms but not at the same concentrations as the scenario where the higher applied voltage resulted in higher energy collisions.

+ + + + + The effect of oxygen partial pressure on the IEDs of Ar , O , O2 , Zn and ZnO are illustrated in Figure 5.7. All measurements were done with an applied voltage of 600

V and a total pressure of 20 mTorr. The argon flow rate was set to 25 sccm and the O2 flow rate varied from 0 sccm to 10 sccm. The Ar+ IED in pure argon had a low energy shoulder at 2 eV and a main peak at 4 eV. When oxygen was added to the system the Ar+

IED intensity, which is proportional to the total flux, dropped in counts. With 1 sccm of

5 oxygen (Opp = 0.8 mTorr) the counts/second decreased to 3.2  10 C/s and two most probable energies were measured at 4 eV and 6 eV. When the oxygen was increased to

4 Opp = 8.0 mTorr the intensity of the IED decreased to 1.8  10 C/s.

+ + + + + Figure 5.7 IEDs for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-HiPIMS plasma with a Zn target with 20 mTorr total pressure and different Opp.

+ + The IEDs for O and O2 decreased when the Opp was increased from 0.8 mTorr

+ + to 8.0 mTorr. The average energy for O IED was 4 eV while the O2 ions had a slightly most probable of 5 eV. The IEDs in pure argon still had a significant concentration of O+

4 + 3 (Intensity ~ 10 C/s) and a very small amount of O2 (Intensity ~10 C/s) from previous experiments even after being sputter-cleaned for 20 minutes in pure argon before this

+ study. No significant change in the concentration of the higher energy O2 population was observed when Opp was increased from 0.8 mTorr to 8.0 mTorr. The lower energy population, however, decreased from 2.03  106 C/s to 8.14  105 C/s. The IEDs for Zn+ and ZnO+ were in the range of the sensitivity range of the EQP detector (SEM ~ 100s

C/s). The Zn+ IED increased with the addition of oxygen while the ZnO+ IED decreased.

Negative oxygen ions reported during sputtering of ZnO [118,119,121] and

HiPIMS of other oxides [120] were not detected in this study or were below the detection limit of the instrumentation arrangements.

5.1.2 Reactive HiPIMS from a Zinc Oxide Target

The effect of pulse parameters and pressure on the target current was investigated for a zinc oxide in target in argon-oxygen gas mixture in a study similar to that performed for the zinc target. The target current for different applied voltages is compiled in Figure

5.8a. The target values for the ZnO ranged from 0.2 A to 3.5 A, values significantly lower than the 5 A to 25 A observed for the Zn target at similar applied target voltages.

The reduced target currents are partly due to the higher resistance of the ZnO target and elastomer bonding compared to that of the pure metal target causing a voltage drop through the bulk of the target-bonding system.

The plasma for the zinc oxide target was stable for a larger range of applied voltages; the minimum voltage for stable operation was the same; however the zinc oxide could handle applied voltages up to 1500 V before arcing began to occur. At all applied voltages, the ZnO target showed very different current evolution from the Zn target. At

400 V negligible current (< 1 mA) was measured and the plasma was weak and flickered.

Above 400 V, the current rose immediately after pulse initiation. After current onset, three distinct current-time behaviors were observed during the pulse depending on the applied voltage. Low applied voltages (≤ 600 V) resulted in a current that reached its peak value and plateaued shortly after pulse initiation. These observations suggest that at very low applied voltage there is a very small ion flux to the target. At intermediate voltage (650-1000 V) the current reached its peak value at the beginning of the pulse then decreased to a steady state value as rarefaction [131] of the process gas adjacent to the target occurred and the flux of available gas ions near the target surface is reduced. At higher voltages of > 1000 V, there is sufficient ionization of the sputtered atoms to counteract rarefaction and the target current increases for the entire duration of the pulse.

For applied voltages between 650 V and 1000 V, a subsequent rise in target current 100-300 s after termination of the applied pulse is seen. The emergence of this peak occurred sooner after pulse termination for higher applied voltages. This secondary pulse could be the result of ionized target material incident on the target [78].

The target currents recorded in Figure 5.8b show the evolution of target current for 650 V pulses applied with a frequency of 100 Hz with a background pressure of 10 mTorr and an O2/Ar ratio of 0.40. The lengths of the pulses in Fig 5.8b were varied between 20 s and 200 s. The general shape of the current decay after pulse termination

78 remained constant for the different pulse lengths. No second current peak was present for the 20 s pulse. For the 50 s pulse, the second current peak occurs 250 s after pulse termination while for the longer pulses there was a 300 s delay between pulse termination and the onset of the secondary current peak.

Figure 5.8 Time-resolved current of ZnO target at 10 mTorr with O2/Ar = 0.4 with different pulsing parameters; (a) applied voltage, (b) pulse length, (c) pulse frequency.

The effect of pulse frequency on the target current evolution is illustrated in

Figure 5.8c. While the general shape of the target current was unchanged by increasing the pulse frequency, the magnitude of the current increased as the pulse frequency was increased. This is likely due to the higher frequencies reducing the time between the

79 pulses. Less time between the pulses meant the oxygen flowing in the chamber had less time to chemisorb or react on the target surface before being cleaned away by subsequent pulses. The more efficient cleaning of the oxide layer on the target surface meant that the process experienced less of the reduction in deposition rate associated with a thick oxygen rich layer on the target surface.

The effect of total pressure on the target current evolution was similar to that observed for applied voltage. For all pressures investigated, the current rose immediately after pulse initiation. Like the case for varying the voltage, three current-time behaviors were observed depending on the total pressure. Low pressure (≤ 10 mTorr) resulted in a current that reached its peak value and plateaued shortly after pulse initiation, similar to the behavior observed for applied pulses, 600 V. At pressures greater than 10 mTorr the current reached its peak value at the beginning of the pulse then decreased to a steady state value as rarefaction [131] of the process gas adjacent to the target occurred and the flux of available gas ions near the target surface is reduced. As with the current profiles in Figure 5.8, a secondary current peak is seen after pulse termination. The occurrence of this second current peak was delayed with decreasing pressure. This secondary pulse could be the result of ionized target material incident on the target [78] and the pressure dependence of the time at which this pulse appears is likely determined by the mean free path of the material after liberation from the target.

Figure 5.9b shows the effect of oxygen flow rate (partial pressure) on the current profile. Sputtering in pure argon resulted in the highest target current, with the current reaching 1.1 A and remaining that value for the duration of the pulse. The addition of 1.0 sccm O2 to the gas background (Opp = 0.4 mTorr) resulted in the current decreasing to

0.87 A for the duration of the pulse. Increasing the oxygen gas content further caused the current to reduce to a peak value of 0.58-0.65 A followed by a slight decay to a lower steady state current. The current value was less sensitive to the addition of oxygen after the O2 flow rate was increased to 5.0 sccm. The secondary current peak after pulse termination was absent for the pure argon and 1.0 sccm O2 current profiles.

Figure 5. 9 Time-resolved current of zinc oxide target with 650 V applied with 200 s pulses at a frequency of 100 Hz with (a) and O2/Ar = 0.4 and the pressure varied and (b) with the pressure held at 10 mTorr and the oxygen flow rate varied. In both graphs the argon flow rate was set to 25 sccm.

The results for the mass spectrometry scan of the plasma at the substrate position with different voltages applied to a ZnO target are compiled in Figure 5.10. For this batch of experiments the maximum voltage used was 500 V due to arcing from the target to shield in the chamber. For all measured voltages there were the same peaks at 16 amu,

+ + + 32 amu and 40 amu, corresponding to O , O2 and Ar respectively, observed in the mass spectroscopy of the Zn target. The intensity of the peaks for the ZnO target were of the similar values to those measured for the Zn target except for the peaks corresponding to

+ Zn. The peak at 32 amu for O2 was the most intense peak for all the voltages measured.

Figure 5.10 Plot of the mass spectra of the plasma from reactive sputtering of a ZnO target in 20 mTorr O2/Ar = 0.04 with different applied voltages with 200 s pulses at 100 Hz.

+ + + + + The IEDs for Ar , O , O2 , Zn and ZnO for different applied voltages applied to a ZnO target are compiled in Figure 5.11. The Ar+ IED had a single broad peak corresponding to argon ions with an average energy of 3 eV. The concentration of these ions increased when the applied voltage was increases from 350 V to 400 V. Increasing the voltage from 400 V to 500 V had no significant impact on their concentration. The

Ar+ IED ended at 6 eV for all voltages.

The IED for O+ at different applied voltages is compiled in Fig. 5.11b. The IED showed O+ ions with energies ranging from 0 eV to 5.5 eV and a most probable energy of

3 eV. The concentration of the O+ ions increased with higher applied target voltages.

+ + The IEDs for O2 at different applied voltages are shown in Figure 5.11c. The O2 ions

82 also had energies from 0 eV to 6 eV and a most probable energy of 3 eV. There was little change in the concentration of these ions with increased applied voltage.

+ + + + Figure 5.11 The ion energy distributions for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO+ measured in a reactive-HiPIMS plasma with a ZnO target with 20 mTorr and an Opp = 0.8 mTorr.

The IEDs for the target species (Fig. 5.11d and Fig. 5.11e) show the emergence of

Zn+ and ZnO+ at lower applied voltages compared to the Zn target. When 400 V and 500

V were applied to the target there was a small population of Zn+ with a most probable energy of 1.5 eV. The concentration of these ions increased as higher voltages were applied. At 500 V there was another larger population with a most probable energy of

4.5 eV. Low concentrations (SEM ~ 100s C/s) of ZnO+ were measured at all applied voltages. These ions had an average energy of 3 eV.

The effect of pressure on ion energy at the substrate is illustrated in Figure 5.12.

The oxygen flow rate was held at 1 sccm for all the measurements in Fig. 5.13. For the

ZnO target at 600 V the plasma could be sustained a 10 mTorr, 5 mTorr lower than the lowest pressure that could sustain a stable plasma for the Zn target. The Ar+ IED (Fig.

5.12a) showed ions with energies ranging from 0 eV to 6 eV and a most probable energy of 3 eV. The concentration of Ar+ was highest at 10 mTorr with SEM = 1.8  104 C/s.

Increased pressure resulted in a reduction of the Ar+. 2  104 C/s showed two populations of ions with most probable energies of 1.5 eV and 4 eV at 20 mTorr. The intensity of the

+ IED for O2 (Fig. 5.12c) showed a pressure-dependent intensity and a most probable ion energy of 4 eV. The increase in intensity at higher pressures corresponds to an increase

+ in oxygen in the system. The intensity of the O2 IED was of the same order of magnitude of that for argon. The O+ IED had an order of magnitude lower intensity than

+ + + + the Ar and O2 . The IEDs for Zn (Fig. 5.12d.) and ZnO (Fig. 5.12e) displayed similar behavior to that observed in Figure 5.11. The low energy Zn+ population was present at all pressures with an intensity on the order of 103 counts. At 20 mTorr a higher energy

84 population emerged. These ions had a most probable energy of 4.5 eV. The intensity of the higher energy peak in the IED was 2.7  105 C/s.

+ + + + + Figure 5.12 IEDs for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-HiPIMS plasma with a ZnO target with different total pressures and O2/Ar = 0.04.

+ + + + + The IEDs for Ar , O , O2 , Zn and ZnO with 1500 V applied to the target are shown in Figure 5.13. The intensities for the gas ion species were 2-3 orders of magnitude higher than those of the target ions. At a pressure of 3 mTorr, the IED for argon had a most probable ion energy of 4 eV and a maximum ion energy of about 15 eV. Both the flux and its rate of decay decreased with increasing pressure. The next

+ most abundant species comprising the flux was O2 (Fig 5.13c) with a peak at ~3.5 eV and a higher energy tail. As with the oxygen, the flux decreased with increased pressure.

The O+ IED (Fig 5.13b) showed similar behavior though it had an order of magnitude lower intensity. The IEDs for the oxygen ions extended to approximately 20 eV compared to 15 eV for argon ions. The IED for Zn+ had a single peak centered at 3 eV and was weakly dependent on pressure in comparison to the gas ions; little change in intensity was observed until the pressure exceeded 30 mTorr then the counts sharply dropped to at 5.33 Pa and higher pressures. The Zn+ IED in Figure 5.12d shows less relative metal ionization in comparison to other ionized species than typically observed in

HiPIMS for metallic targets [96,105,132]. This lack of Zn ionization is likely due to the relatively lower peak currents during sputtering, Itarget > 3.5 A for the compound target compared to values >50 A reported in literature for metallic targets [76,103,107]. The flux of ZnO+ ions was lower than the detectable limits of the EQP.

+ + + + + Figure 5.13 IEDS for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO with 1500 V 200 s 100 Hz pulses at different total pressures with a fixed O2/Ar of 0.04.

+ + + + + The effect of oxygen partial pressure on the IEDs of Ar , O , O2 , Zn and ZnO are illustrated in Figure 5.14. All measurements were done with an applied voltage of

600 V and a total pressure of 20 mTorr with the argon flow rate was set to 25 sccm and

87 the O2 flow rate varied from 1-10 sccm. The general shape of the IEDs was similar to that observed in Figure 5.11 and Figure 5.12. Increasing the amount of oxygen gas reduced the fraction of argon gas in the system. A decrease in intensity of the Ar+ IED, seen in Fig. 5.14a, accompanied this reduction. For the O+ IED, the highest intensity

+ + + occurred with 5 sccm O2. The intensity of the O2 , Zn and ZnO IEDs decreased with higher oxygen flow rates.

+ + + + + Figure 5.14 IEDS for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO measured in a reactive-HiPIMS plasma with a ZnO target with 20 mTorr total pressure and different Opp.

5.2 Film Growth and Microstructure

5.2.1 Film Growth from a Zinc Target

A range of operating parameters for film growth from the Zn target was determined using observations on plasma stability and changes in plasma intensity and emission color.

The deposition rate for the films grown in Table 4.1 were calculated from the measured film thickness and deposition time using eqn. (19) (the deposition rates can be found in the Appendix B). The applied voltage and pulsing parameters had a greater effect than the pressure and Opp on deposition rate. The deposition rate increased from

5.37 ± 0.33 nm/min at 450 V to 27.73 ± 0.30 nm/min at 600 V. The increase in deposition rate with higher applied voltages is much larger than the observed increase in sputtering yield observed in Fig. 5.1b. The calculated deposition rate was higher for films grown at higher pulse lengths, 4.10 ± 0.33 nm/min for 50 s pulses and 15.47 ±

0.30 nm/min for 200 s pulses. The deposition rate also increased with high pulse frequencies, 4.93 ± 0.73 nm/min for 50 Hz and 28.07 ± 0.70 nm/min for 250 Hz

Wide angle XRD patterns for ZnO films grown while systematically varying the pulsing parameters and pressure are compiled in Figure 5.15 and Figure 5.16, respectively. All the wide angle scans showed a peak at 2 = 34.3o± 0.3o corresponding to the ZnO (002) orientation. The intensity of the (002) diffraction peak and the presence of other peaks were dependent on the deposition conditions. The effect of the individual deposition parameters on the XRD patterns and corresponding film microstructures will be discussed in the proceeding paragraphs.

The effect of applied target voltage on film microstructure is illustrated in Figure

5.15a. Wide angle scans were analyzed for films grown at 350 V, 500 V and 600 V, applied voltages corresponding to three different behaviors of the target current from

Figure 5.15a. The film grown with 350 V showed a single low-intensity diffraction peak for at 2 = 34.5o. The intensity of the (002) peak increased an order of magnitude for the films grown with higher target voltages. The ZnO (002) diffraction peak remained the prominent feature for the films grown with applied voltages of 500 V and 600 V, but both films had other orientations present. The degree of (002) alignment, determined by the

FWHM in Figure B.1a (Appendix B), showed an increase in alignment of the (002) oriented crystals at higher applied voltages. The (100) and (101) orientations were both present in the film grown at 500 V. The film grown with 600 V had only the (002) and

(101) orientation. The FWHMs of the (002) peaks in Fig. 5.15 were for  in eqn (20) to calculate the average grain sizes which can be found in Appendix B. The grain sized increased from 11 nm to 19 nm when the applied voltage was increased from 350 V to

500 V. The grain size of the film grown with 600 V was determined to be 15 nm.

Correlation between the observations made in Figure 5.15a and the evolution of the target current in Figure 5.2a provided some insight into how the ion content in the plasma affects the film’s microstructural evolution. Films grown with 350 V had poor crystallinity (evident by the broad low intensity ZnO (002) diffraction peak) due to the low ( < 1 mA) target current and negligible flux of energetic ionized species to the substrate . At higher applied voltages, the current reached values between 5A and 17 A

+ + + and the substrate was bombarded with Ar , O and O2 with energies of 2-6 eV resulting in films with strong crystallinity and preferred (002) orientation (evident by the narrow

90 high intensity XRD peak). The films grown at voltages less than 600 V had a slight shift toward lower diffraction angles suggesting compressive stress in the films.

+ + Bombardment by the higher energy O and O2 that form at voltages of 600 V and larger (Fig. 5.5) caused a relaxation of the compressive stress in the films and the shift in diffraction angle is absent in the film grown at 600 V.

Figure 5.15 Wide angle XRD patterns for ZnO films deposited from a Zn target in 20 mTorr with Opp = 8 mTorr with different (a) applied target voltages, (b) pulse lengths, and (c) pulse frequency.

Figure 5.15b shows the wide angle XRD scan of films grown with 50 s, 100 s and 200 s pulses. The pulse lengths selected for film growth encompass the three target current regimes observed in Figure 5.2b: (1) where the pulse terminates immediately after

91 the peak current is reached; (2) where the pulse terminates after the current has decayed but before a steady state value is reached and (3) where the pulse terminates after the steady state has been reached. The ZnO (002) diffraction peak is the prominent feature in all the XRD patterns. For the 50 s film, the (002) peak is the only feature and the RC curve (Figure B.1b) showed random alignment of the (002) crystals. The wide angle

XRD patterns for film grown with different pulse frequencies are compiled in Figure

5.15c. No shift in the (002) diffraction peak was observed with changes in the pulse frequency, however, higher pulse frequencies resulted in an increase in the degree of alignment in the (002) oriented crystallites. The concentration of (101) and (100) orientations changed as higher pulse frequencies reduced the time between pulses for the deposited material to reach energetically favorable orientations before burial by the subsequent incoming fluxes.

Wide angle XRD scans of films grown with different total pressures and oxygen partial pressures are compiled in Figure 5.16. A pronounced peak was present at 2 =

36.6o, corresponding to the ZnO (101) orientation for the film grown at 10 mTorr. This peak was absent for the 20 mTorr film. The degree of (002) alignement in the films decreased with pressure, as evident by the broadening of the (002) RC FWHM (Figure

B.2a). The XRD of the films grown at different oxygen partial pressures (Fig. 5.16b) show pronounced (100) and (101) diffraction peaks when the oxygen partial pressure was high. The alignment of the (002) oriented crystallites (Figure B.2b), however, increased

+ with oxygen partial pressure and the accompanying higher concentrations of O2 incident on the substrate (Figure 5.7).

Figure 5.16 Wide angle XRD patterns for ZnO films deposited from a Zn target with 500 V applied in 200 s pulses at a frequency of 100 Hz, (a) different total pressures and O2/Ar = 0.4 and (b) a total pressure of 20 mTorr and different O2/Ar.

5.2.2 Film Growth from a Zinc Oxide Target

Based on observations of plasma stability and changes in the plasma for the ZnO target, a range of deposition parameters were selected at which to grow films that would be representative of growth in the different regimes of ion concentration and energies measured in section 5.1.2 . The deposition rates from the ZnO target were much lower than those observed for sputtering from a Zn target. At 400 V the deposition rate was less than 1.0 nm/min. At 800 V the rate was 1.8 ± 0.33 nm/min. Increasing the applied voltage to 1500 V resulted in a deposition rate of 10.7 ± 0.8 nm/min. An increase in deposition rate was observed when the pulse length and pulse frequency were increased; more details on the effect of the deposition parameters on deposition rate can be found in

Table B.3-4 in Appendix B .

The wide angle XRD patterns for the films grown with different applied voltages are shown in Figure 5.17a. Microstructural changes similar to those observed in the Zn target with increased voltage were seen. Films grown at low applied voltages had poor

93 crystallinity and randomly aligned (002) crystallites, evident by the a broad low intensity

ZnO (002) diffraction peak and (002) RC peak (Figure B.3), suggesting that the depositing flux had an insufficient fraction of ionized species to promote crystallization.

Increasing the applied voltage increased the ionized flux and its energy at the substrate promoting growth and alignment of the (002) orientation. The diffraction angle shifts observed in Figure 5.15 were not as pronounced for the films grown from the ZnO target.

Figure 5.17 Wide angle XRD patterns for ZnO films deposited from a ZnO target in 20 mTorr with Opp = 0.8 mTorr with different (a) applied target voltages, (b) pulse lengths, and (c) pulse frequency.

Increasing the pulse length yielded no change in the intensity or width of the

(002) diffraction peaks (Fig. 5.17b) or RC until 200 s. Growth with 200 s pulses

94 resulted in a decrease in the XRD peak intensity. Above a minimum pulse frequency neither the XRD pattern in Figure 5.17c nor the rocking curve in Figure B.3 showed any changes with increased pulse frequency.

The effect of total and oxygen partial pressure on film microstructure is illustrated in Figure 5.18 and Figure B.4. The intensity of the ZnO (002) diffraction peak was highest for the film grown with 10 mTorr. The intensity decreased when the pressure was increased to 20 mTorr but no further reduction was observed for higher pressure increases. The degree of alignment of the (002) crystallites decreased with pressure. No significant change was observed in the XRD pattern when films were grown with higher oxygen partial pressures, however, the alignment of the (002) crystals increased.

Figure 5.18 Wide angle X-Ray diffraction patterns for ZnO films deposited from a ZnO target with different (a) total pressures and O2/Ar ratio of 0.04 and (b) a fixed total pressure of 20 mTorr and different Opp.

After the initial study on the effect of deposition parameters on the intensity and width of the ZnO (002) diffraction peak and the presence of peaks corresponding to (101) and (100) oriented crystallites it was determined that higher applied voltages, lower total pressures and grounded substrates produced films with the highest degree of (002)

95 orientation. This prompted more detailed analysis of the growth of films deposited in these regimes.

Wide angle x-ray diffraction as a function of pressure for films grown with 200 s

(Figure 5.19a) and 50 s (Figure 5.19b) pulses showed two diffraction peaks, ZnO (002) at 2 = 34.3o and ZnO (101) at 2 = 36.0o. The wide angle scans were performed from

2= 25-80o, but the x-axis in Figure 5.19 is scaled to best show the region containing the

(100), (002) and (101) peaks. The (101) diffraction peak was two orders of magnitude less intense than the (002) peak for all the films. The (002) wide angle diffraction peak was broader for films grown with higher pressures. This result is most visible at pressures below 10 mTorr; the broadening of the (002) peak at 5 mTorr was less pronounced for the 50 s film compared to the 200 s. The vertical axis in Figure 5.19c represents integrated intensity of the (101) peak divided by the sum of these for the (101) and (002) peaks. Considering the 50 s and 200 s data, no significant relationship between deposition pressure and (101) texture appearance was observed. Reducing the substrate to target distance from 10 cm to 5 cm, however, decreased the fraction of (101) oriented grains. Looking at the films grown at 10 mTorr, an average diffraction intensity ratio I(101)/[I(101)+I(002)] of 0.004 was determined for the film grown at a distance of 5 cm compared to 0.047 at 10 cm. The XRD data for films grown with the 5 cm substrate- target distance can be found in Figure B.6 in Appendix B.

Figure 5.19 Wide angle XRD for ZnO films grown with (a) 200 s and (b) 50 s pulses at different total pressures and O2/Ar = 0.04. (c) Ratio of the area of the (101) peak to the total area of the (101) and (002) peaks from the wide angle XRD scan plotted against pressure.

Origin software was used to deconvolute the (002) XRD peak for the ZnO film grown at 3 mTorr from Figure 5.19. The two fitted peaks are shown in Figure 5.20.

Origin calculated the FWHM of the peak at 2 = 33.5o to be 0.84o and that of the peak at

2 = 34.2o to be 1.44o for the 200 s film. The (002) XRD peak for the film grown with

50 s pulses was deconvoluted into a peak at 2 = 33.6o with a FWHM = 1.02o and 2 =

34.4o with FWHM = 1.33o.

Figure 5.20 Wide angle XRD pattern for films grown at 3 mTorr with (a) 200 s and (b) 50 s pulse with the peak deconvoluted into its two possible components.

Figure 5.21 shows the rocking curve measurements for the films characterized in

Figure 5.19. For both the 200 s and 50 s films the rocking curves became broader with increased pressure. The vertical axis of Figure 5.21c shows the FWHM of the (002) rocking curves. The rocking curve FWHM for all the films increased with pressure. The narrowest FWHMs occurred when the pressure was below 10 mTorr, where FWHM appears to be most sensitive to pressure changes. The FWHM of the 200 s film at 3 mTorr is 6.58o, while the FWHM at 10 mTorr is 9.45o. For the 50 s pulse films the rocking curve (Fig. 5.21b) narrowed from a FWHM of 10.47o at 10 mTorr to a FWHM of

6.71o at 3 mTorr. No significant change in the FWHM was observed by shortening the substrate-target distance.

Figure 5.21 Rocking curve XRD of the (002) peak for ZnO films grown with (a) 200 s and (b) 50 s pulses at different Ar/O2 pressures. The y-axis is plotted in log scale for clarity. (c) Full width half maximum of the (002) rocking curve peak of HiPIMS-grown ZnO.

TEM and SEM micrographs of the cross-section of a representative film grown at

10 mTorr are shown in Figure 5.22. Fig. 5.22a shows a smooth interface between the

ZnO film and the SiO2 substrate. The initial ZnO growth at the substrate interface appears to be small v-shaped grains that become narrow columns after the first few tens of nanometers, as evident in the SEM micrograph in Fig 5.22b. The yellow-dash line shows the extent of the competitive growth layer and the green line and arrow represent the degree () of the c-axis orientation. Additional TEM micrographs can be found in

Figures B.9-12 in Appendix B.

Figure 5.22 Cross-sectional (a) TEM with an insert of a close up of the ZnO-SiO2 interface and (b) SEM of a film grown at 10 mTorr with a 200 s pulse length.

100

The TEM cross-section of the film grown at 3 mTorr is shown in the Figure 5.23.

The yellow lines drawn in Figure 5.23 highlight the lattice structures in the TEM micrograph. The micrograph shows the (002) crystal plane is parallel to the substrate in all grains. The lattice parallel to the substrate has a spacing of 2.607-2.610 Å, slightly larger than the standard value of 2.603 Å. The diagonal lattice, corresponding to diffraction from the (101) plane, has a spacing of 2.633 Å. Diffraction from the (100) crystal plane resulted in a diffraction pattern perpendicular to the substrate with a lattice spacing of 2.831 Å, larger than the standard value of 2.814 Å. The dotted white lines in

Figure 5.24 highlight grain boundaries in the film. The grain with the (101) lattice image is 6 nm across while the (100) grain is 17 nm.

Figure 5.23 Cross-sectional TEM of a ZnO film grown with a deposition pressure of 3 mTorr.

101

The AFM images of the ZnO film surfaces are presented in Figure 5.24 and show that the films were smooth with RMS roughnesses < 10 nm. The films grown at pressures 10 mTorr and higher have similar surface topography and surface roughnesses of 4.6-6.1 nm. The difference in surface roughness between the films is comparable to the uncertainty caused by AFM tip dulling during scans. The surface of the film grown at

3 mTorr has larger grains (~25 nm diameter) intermixed with the smaller grains and a lower surface roughness. AFM topography maps showed that the larger grains were 10-

15 nm taller than the smaller grains. Even with the larger grains intermixed amongst the small grained background, the RMS roughness of the 3 mTorr film was lower than the other films (RMS = 1.4 nm). ZnO films grown with 50 s pulses exhibited similar AFM topography (Figure B.7 in Appendix B) with bimodal grain sizes at low pressures (3 mTorr and 10 mTorr growth conditions).

102

Figure 5.24 Surface AFM images and schematics representation of grains of HiPIMS-grown ZnO films grown with different pressures.

103

5.2.3 Discussion of Film Growth Mechanisms and Plasma Conditions

The film microstructure at the substrate interface, investigated with cross- sectional TEM (Fig 5.22), showed a layer of mixed orientation which likely contributes to the low-intensity (101) diffraction in the wide angle XRD at 2 = 36o. This result is consistent with previous work on ZnO films where polycrystalline competitive growth regions near the substrate-film interface were also found [133,134]. The layer of mixed orientation at the substrate interface is the result of competition between nucleation by high surface energy (002) oriented crystals [135] and the lower energy (100) and (101) oriented crystals as illustrated in Figure 5.25. Because the competitive texture layer is located at the film-substrate interface, the location of the depletion zone in a back-gated

FET device designs, it could potentially play an exaggerated role in controlling transistor transport properties making the minimization of this layer crucial for TFT channel optimization. For growth from the ZnO target, this initial layer is comprised primarily of

(101) and (002) orientations, while growth from the Zn target showed some (100) oriented crystals. The higher concentration of (101) and the presence of (001) oriented crystals in the films deposited from the Zn target is attributed to the larger concentrations of energetic ionized gas species bombarding the substrate increasing the sticking coefficient of the lower energy crystallographic orientations compared to the ZnO target.

104

Figure 5.25 Schematic diagram illustrating competition for nucleation by (100), (101) and (002) oriented crystallites during early stages of ZnO growth.

From analysis of the XRD data in Figure 5.19c, no correlation between deposition pressure and the extent of the competitive texture layer growth can be inferred. However, decreasing the substrate to target distance appears to reduce the presence of the mixed orientation layer as evident from comparison of the XRD data for films grown with a target-substrate separation of 5 cm (compiled in Figure B.6 in Appendix B) compared to that for films grown with 10 cm separation. By shortening the distance between the substrate and target, the plasma density is increased and more energy is deposited to the substrate, lowering the sticking coefficient and suppressing nucleation by crystallites with lower surface energy. These conditions promote favorable growth of the low energy

(002) orientation, reducing the initial competitive layer of mixed (101) and (002) orientation. This is expected be beneficial for reducing charge carrier scattering in the channel at the channel-gate dielectric interface for ZnO TFT devices with back-gated architectures.

105

The XRD results in Figures 5.19 and 5.21 suggest the columns observed in the

SEM in Figure 5.22 are comprised of grains oriented with (002) parallel to the substrate interface. The degree of alignment of the (002) orientation in the films was dependent on deposition pressure with lower pressures resulting in higher alignment of the (002) orientation. The pressure- dependence of (002) alignment can be attributed to the differences in the concentration for the ionized species in Figure 5.12 and 5.13. The decrease in ion concentration is the result of energy losses of the ionized species at the substrate due to collisions in the plasma sheath. The sheath thickness, estimated to be ~

0.18 cm assuming a plasma density at the sheath edge of ~109 cm-3 and a floating potential of ~30 V, is the same order of magnitude as the estimated mean free path () of the ionized flux of target material to the substrate. Decreasing the pressure increases the mean free path and decreases the probability that the ionized species will experience energy-exchanging collisions resulting in a higher concentration of energetic ions reaching the substrate. This results in an increase in the ratio of ions to neutral (Ji/JMe) in eqn. (22) at the substrate surface which has been demonstrated by Petrov et. al to increase film crystallinity and promote crystallographic textures [123,136].

The narrowest RC FWHMs were obtained for films grown with pressures below

10 mTorr. However, the wide angle scans for these films showed broadening and shifting of the (002) peak. The (002) peak for the 3 mTorr and 5 mTorr Pa films can be deconvoluted into two peaks, a higher intensity narrow peak shifted to 33.5o and a lower intensity broader peak at 34.3o (33.6o and 34.4o for the 50 s film), consistent with the observed bimodal distribution of (002) oriented grains shown in Figure 5.24. When the

FWHM of the two deconvoluted peaks are used in the Scherrer equation (eqn. 24) the

106 crystal sizes are determined to be 10 nm and 6 nm for the 200 s film and 8 nm and 6 nm for the 50 s film. The cross-sectional TEM in Figure 5.23 shows that the bimodal distribution of grain sizes persists through the whole of the film to the substrate interface.

Measuring across the grains in Figure 5.23 the grains are determined to vary from about 5 nm to 15 nm. The bimodal distribution in grain sizes at pressures lower than 10 mTorr is

+ + attributed to bombardment by the higher energy O and O2 evident as the higher energy tail in Figures 5.12 and 5.13. Bombardment by highly energetic ions has been previously shown to result in compressive strain in the film and damage to the crystal lattice [123].

Because Ar+ has a larger momentum transfer cross-section than the oxygen ions, Ar+ is expected to experience a greater number of energy reducing collisions and not contribute to the formation of the bimodal grains. This is supported by the IEDs in Figure 5.13,

+ + + where the Ar IED terminates at lower energies than the O , and O2 IEDs.

5.2.4 Summary for Film Growth Study

o Reactive HiPIMS of Zn and ZnO target in O2-Ar gas at Tsub ≤ 150 C resulted in small-grained (6-20 nm), smooth (RMS surface roughness < 6 nm) highly (002) oriented crystalline films. The deposition parameters (pressure, oxygen partial pressure, pulse parameters, and target-substrate separation) allowed control of the film microstructure, specifically the degree of (002) alignment, through tailoring of the relative abundance

+ + + and energies of ionized species (O , O2 and Ar ) incident on the substrate and growing film surface. The extent of microstructural control available during HiPIMS synthesis of

ZnO is evident by the variation in the degree of alignment for (002) oriented crystals, determined by the FHMW of the (002) RC in Figure 5.26, of ZnO films grown at 150 oC.

107

High deposition pressures (20-50 mTorr) resulted in films with crystallites aligned with an angular distribution of 13.4o relative to the substrate normal, while, growth at lower pressure (≤10 mTorr) produced more highly aligned crystals with a narrower distribution of 4.54o.

Figure 5.26 Summary of FWHM of (002) rocking curve XRD peaks at different substrate temperatures for ZnO films deposited with pulsed DC magnetron sputtering [136], PLD[34-36], and RF sputtering[46,47] with the data for films synthesized with HiPIMS.

To demonstrate the feasibility of HiPIMS-processing of ZnO for flexible electronics, 100 nm thick ZnO was deposited on a Kapton substrate. The SEM micrograph in Figure 5.27 shows the edge of the ZnO-covered Kapton after the bending.

The ZnO grew on the Kapton with dense columnar grains (Fig. 5.27b) with a (002) crystallographic texture (XRD in Fig.B.8 in Appendix B). No delamination in the ZnO is

108 observed in Fig. 5.27a after bending the substrate. Some stress cracking is evident in the

ZnO film in Fig. 5.27a, however, the degree of substrate bending in this study exceeded that expected for flexible electronic applications.

Figure 5.27 (a) SEM micrograph of HiPIMS-grown ZnO on kapton. (b) Close-up of kapton-ZnO interface.

109

5.3 Effect of Spatial Inhomogeneity of the Depositing Flux on Film Microstructure

The rigidity of the EQP probe and limitations in ports to mount the probe greatly restricted the spatial resolution of the mass spectroscopy and ion energy measurements.

To investigate the spatial uniformity of the depositing flux and film uniformity over large areas, ZnO was deposited on 4-in. diameter SiO2/Si wafers like the one shown in Figure

5.28a. The dark cross on the wafer is Fig. 5.28a is where the substrate was masked with kapton during deposition. The height difference between the exposed substrate and the top of the deposited film was measure at multiple locations different distances from the samples center. The film thickness for films grown with the Zn target and the ZnO target was determined to vary less than 10 % (~12 nm) across the whole sample. The small variation in film thickness means that thickness affects can be neglected when considering microstructural differences in the XRD patterns in Fig 5.28 and Fig. 5.29.

Figure 5.28b shows the wide angle XRD patterns measured at different positions on a sample grown by reactive-HiPIMS of a Zn target; spot 1 was closest to the sample center while spot 4 was near the sample edge. The low intensity of the diffraction peak at spot one could partially be due to some of the x-ray beam being incident on the exposed substrate. The XRD pattern in Fig. 5.28b showed a broad two-component (002) diffraction peak, similar to that of the 3 mTorr sample in Fig 5.19. The intensity of the

(002) XRD peak increased and its FWHM decreased as the XRD was measures away from the samples center. The (002) peak also shifts to lower diffraction angles farther from the sample center suggesting compressive stress in the film in those regions. Higher concentrations of the (101) orientation were present closer to the sample edge. The RC

XRD measurements (Fig. 2.28c) of the film shows a higher degree of alignment of the

110

(002) orientation farther from the samples center. The angle of the target relative to the substrate and the shallow race track on the Zn target mean that the perimeter of the substrate is subjected to more energetic bombardment than the center.

Figure 5.28 (a) Optical image of a 4-in. diameter sample ZnO sample deposited from reactive HiPIMS of a Zn target with numbers indicating the locations for XRD (b) wide angle and (c) rocking curve scans.

The wide angle diffraction patterns for the samples grown from the ZnO target at

3 mTorr (Fig. 5.29b) and 10 mTorr (5.29c) show much smaller concentrations of the

(101) orientation across the whole film surface. The 3 mTorr film showed the broadening of the (002) peak observed in Fig. 2.28b, however, this effect was more pronounced at the sample center. The (002) intensity increased and the FWHM decreased for XRD patterns measured toward the sample edge. The microstructural differences at the sample center for the films grown from the Zn target versus the ZnO

111 target are due the different sizes of the targets, the Zn target had a 1.3” diameter while the

ZnO target had a 2” diameter, affecting where the most intense ion bombardment occurred. The XRD pattern for the film grown at 10 mTorr (Fig. 5.29c) showed microstructural uniformity across the sample surface.

Figure 5.29 (a) Optical image of a 4-in. diameter sample ZnO sample deposited from reactive HiPIMS of a ZnO target with numbers indicating the locations for XRD measurements. Wide angle XRD patterns for ZnO films grown at (b) 3 mTorr and (c) 10 mTorr.

5.4 Electrical Characterization of Zinc Oxide TFT Channels

A series of back-gated FET devices were created with the ZnO deposited using reactive-HiPIMS of a ZnO target. The channel material in Figure 5.30 was deposited with 50 s long pulses 1500V at a frequency of 100 Hz. The pressure was held at 20 mTorr with Opp = 0.8 mTorr. The target to substrate distance was 10 cm. The silver section in the upper right-hand corner of the sample in Fig. 5.30a is the gate contact.

112

Figure 5.30 (a) Image of back-gated FETs with HiPIMS-deposited ZnO channels. (b) I-V measurements for a FET with ZnO channel. (c) Close-up micrograph of FET source-drain contacts after electrical measurements.

Figure 5.30b shows a representative I-V plot for a FET device with a channel width of 750 m, a channel length of 60 m, a channel thickness of 100 nm and a gate dielectric thickness of 100 nm with 0-80 V gate voltages applied; additional I-V for TFT devices fabricated on this sample can be found in Figures. C.1-4 in Appendix C. Figure

5.30b shows that the FET device turned on at a gate voltage of 40 V. The channel resistance in this state is 740 kWhen in the on-state the drain current increased until it reached a saturation current. Typically in FETs the drain current levels off at the saturation current. Figure 5.30b, however, shows a slow decay in the current after

113 saturation due to leakage in the SiO2 gate dielectric. Increasing the gate voltage resulted in an increase in the saturation current. The lack of reproducibility in saturation current between TFT devices with similar geometries fabricated on the same ZnO samples is illustrated in Figure C.5-6. Additionally, there was damage to the source contact during the electrical measurement (Figure 5.30c). This damage prevented measuring the drain current as a function of gate voltage. This second electrical measurement is necessary for calculations of the FET mobility and on-off ratios. Electrical characterization for additional ZnO channels are compiled in Figures C.7-11.

The preliminary TFT device measurements performed in this study demonstrate the potential for incorporation of HiPIMS synthesized ZnO into TFT channels.

However, optimization of the gate dielectric and source-drain contacts is needed to continue with the device characterization and could be a whole separate study which will be discussed in the future work section.

114

CHAPTER VI

CONCLUSIONS

The overall goal of this study was to demonstrate low-temperature scalable deposition of ZnO films with controlled crystallinity and texture using reactive HiPIMS.

This study also sought to establish film growth mechanisms based on correlation between plasma parameters and resultant film structure to determine the processing conditions needed to synthesize ZnO films with optimal structures for potential applications in TFT semiconductor devices as determined from review of current literature.

Critical analysis of literature on ZnO for TFT device applications showed the need for crystalline structures with the (002) orientation aligned parallel to the substrate plane (i.e. the TFT gate in a back-gated device). To increase the on/off ratio, current density and saturation current smooth films with dense (002) oriented columnar structures are needed. From this perspective, several materials science challenges were identified for ZnO film growth due to i) processing limitations, where PLD and RF magnetron sputtering have inherent challenges to increasing scalability; ii) the need for high temperature growth, which limits substrate choice and prevents applications with temperature-sensitive substrates (e.g. flexible electronics); iii) a tendency toward competitive structure growth which affects charge carrier scattering in the depletion zone of back-gated TFTs; iv) difficulty in fabrication of top-gated devices due to localized

115 field effects resulting from the high surface roughness of the films. The literature study revealed that the increased energy of the depositing material and low energy ion bombardment during HiPIMS could potentially overcome these challenges but fundamental investigations of plasma properties and their correlation with film structure are absent and critically needed to enhance the current state of the art for ZnO-based devices via a scalable growth route such as magnetron sputtering.

HiPIMS plasma parameters were studied for ZnO and Zn sputtering by mapping the ion species and their energies across pressure and voltage processing variables.

Sputtering from both targets produced low energy (2-6 eV) ionized gas species (Ar+, O+,

+ 5 3 + O2 ) with of fluxes of < 10 C/s at the substrate surface and 10 C/s of Zn at voltages higher than 600 V. The pressure-dependence of the ionized gas species’ concentrations and appearance of a higher energy ion population (with energies up to 50 eV), however, behaved differently for the two targets. For the ZnO target, the ion concentration decreased with increased pressure and at pressures above 7 mTorr the higher energy ions were suppressed. For the Zn target, the low energy ion concentrations and presence of higher energy ions increased with pressure. The fluxes and relative abundances of ionized species identified in the plasma studies were then correlated with film structure, determined from XRD, AFM, TEM and SEM, and used to establish mechanisms for film growth and microstructural evolution.

For both Zn and ZnO targets in O2-Ar gas, initial growth at the substrate interface showed competition between the undesirable low surface energy (100), (101) planes and desirable high surface energy (002) orientations. The extent of the initial competitive growth could be minimized by increasing the average energy of incident species which

116 caused a reduction in the sticking coefficient of the sputtered material and suppressed lower surface energy (100) and (101) oriented crystallites in favor of (002) growth.

Deposition from the Zn target produced higher concentrations of the (101) and (100) crystal orientations than deposition from the ZnO target due to the higher peak current of the Zn target (5-25 A compared to 0.2-3.5 A for the ZnO target) resulting in a larger flux of energetic species. Additionally, the higher deposition rate from the Zn target allowed less time to nucleate the (002) orientation growth.

After (002) growth initiation, two important pressure-dependent microstructural trends were identified; a variation in the degree of (002) alignment with the surface plane and a bimodal size distribution of crystals. For ZnO target, lower pressures resulted in a higher degree of (002) alignment, which was linked to the longer mean free path of the plasma species and less energy attenuation of the energetic ions resulting in a larger flux of energetic ions to the substrate promoting the (002) texture. However, films grown at pressure lower than 7 mTorr had a bimodal distribution of (002) grain sizes which is attributed to the appearance of a high energy tail (with cutoff energies up to 50 eV) in the ionized gas species at such reduced pressures. For Zn target, the concentration and cutoff energy of the ionized gas species increased with increased pressure and the bimodal distribution of grain sizes persisted up to 10 mTorr. For Zn target, the higher ion energy bombardment also induced compressive stress and smaller strained crystallites.

This study also demonstrated large area deposition of ZnO using reactive

HiPIMS. For depositions across large area (4-inch diameter substrates) the effects of the high energy ion bombardment on film microstructure became less pronounced farther from the substrate center. Deposition from the ZnO target at pressures of 10 mTorr or

117 higher produced films with uniform microstructure and thickness variations less than

10% across the 4-inch diameter substrates.

From the plasma and microstructural studies it was determined that at Tsub = 150 oC reactive HiPIMS of a ZnO target with high applied voltages (-1500 V), moderate (10 mTorr) pressures and O2/Ar ratios of 0.04-0.20 produced (002) oriented nanocrystalline

ZnO films with the dense columnar microstructures and low surface roughness desired for TFT channels. Incorporation of HiPIMS-deposited ZnO films with these optimized microstructures into working back-gated FET devices was demonstrated, where the ZnO channels had resistances on the order of Ms and the devices switched on with applied gate voltages of 40 V to reduce channel resistances decreased to ks. The device results are preliminary, as the substrate was not optimized for the measurements of back-gated

TFT performance and further work with device architecture can reduce gate voltages for switching and minimize leakage currents.

These results have shown the applicability of HiPIMS to scalable production of highly oriented ZnO films at low substrate temperatures where the critical role of energetic ions in suppressing undesirable (101) texture and promoting highly aligned

(002) texture was identified. Additionally mechanisms for the optimum plasma control toward production of high quality ZnO films at temperatures as low as 150 oC for TFT applications requiring large area growth and temperature sensitive substrates, such as polymers for flexible device applications, have been established. Further investigation of the effect of the competitive growth layer at the gate dielectric interface on device performance and more detailed characterization and optimization of the devices is needed and will be discussed in more detail in the future work section in the proceeding section.

118

CHAPTER VII

FUTURE WORK

Although this study contained a detailed investigation of the plasma during reactive HiPIMS of Zn and ZnO targets important questions about the plasma dynamics remain and there are a plethora of studies that could still be performed. The next steps in the plasma characterization would be separation of the effects gas-phase phenomena (i.e. changes in SE and sputtering yield due to oxide formation on the target) and target- surface phenomena (i.e. changes in SE due to oxide formation) on the target current evolution, determination of the kinetic energies of the neutral species flux at the substrate. The growth mechanisms established in this study could be expanded through incorporation of quantitative data on the relative abundances and energies of neutral species, specifically Zn, O and ZnO, at the substrate surface with the information on ionized species from this study.

In this study the pulse frequency was varied from 50 Hz to 250 Hz, decreasing the time between pulses from 3.8 ms to 19.8 ms, meaning the difference in separation between pulses for the target currents in Fig. 5.2c and Fig. 5.8c changed on the order of thousands of microseconds. Since gas-refill phenomena occurs on a time scale of hundreds of microseconds, it cannot be determined whether the changes in target current as a function of pulse frequency were the result of surface phenomena, gas-phase

119 phenomena or both. Separation of the effects of target surface phenomena and gas-phase phenomena on the target current could be accomplished by varying the pulse frequency with smaller increments (~10 s of f = 1 Hz) in the frequency than used in this study.

Additionally incorporation of HiPIMS- deposited ZnO channels into TFT devices will require detailed contact studies, more in depth electrical characterization of the ZnO films and determination of device performance and reliability. In this study the Si wafers used as the TFT gate were not heavily doped and therefore were resistance (= 1-10 cm). Additionally the SiO2 gate dielectric was thick (100-300 nm) and large ( <40 V) gate voltages were needed to induce charge carriers in the channel. These high gate voltages resulted in leakage current and potentially breakdown of the ZnO channels.

Optimization of the gate and gate dielectric would allow for device operation at lower voltages. The three microstructural effects observed in this study could potentially strongly affect TFT performance. Detailed studies on the effect of these microstructural features on TFT device performance are still needed. Determination of how the competitive layer and (002) alignment affect FET mobility, current density and the device on/off ratio would allow for optimization of the TFT through control of the film microstructure. Further optimization of film stoichiometry and investigation of post- deposition treatments, such a annealing, could also yield improvements in device performance.

120

REFERENCES

[1] U. Helmersson, M. Lattemann, J. Alami, J. Bohlmark, A.P. Ehiasarian, and J.T. Gudmundsson, High power impulse magnetron sputtering discharges and thin film growth: A brief review, SVC, Society of Vacuum Coaters - 48th Annual Technical Conference, April 23, 2005 - April 28, 2005, Society of Vacuum Coaters, Denver, CO, United states, 2005, pp. 458-464.

[2] K. Sarakinos, J. Alami, and S. Konstantinidis, High power pulsed magnetron sputtering: A review on scientific and engineering state of the art, Surface and Coatings Technology, 204 (2010) 1661-1684.

[3] R. Machunze, A.P. Ehiasarian, F.D. Tichelaar, and G.C.A.M. Janssen, Stress and texture in HIPIMS TiN thin films, Thin Solid Films, 518 (2009) 1561-1565.

[4] K. Bobzin, N. Bagcivan, P. Immich, S. Bolz, J. Alami, and R. Cremer, Advantages of nanocomposite coatings deposited by high power pulse magnetron sputtering technology, Journal of Materials Processing Technology, 209 (2009) 165-170.

[5] A.P. Ehiasarian, J.G. Wen, and I. Petrov, Interface microstructure engineering by high power impulse magnetron sputtering for the enhancement of adhesion, Journal of Applied Physics, 101 (2007) 54301.

[6] V. Sittinger, O. Lenck, M. Vergohl, B. Szyszka, and G. Brauer, Applications of HIPIMS metal oxides, Thin Solid Films, 548 (2013) 18-26.

[7] V. Sittinger, F. Ruske, W. Werner, C. Jacobs, B. Szyszka, and D.J. Christie, High power pulsed magnetron sputtering of transparent conducting oxides, Thin Solid Films, 516 (2008) 5847-5859.

[8] S. Konstantinidis, A. Hemberg, J.P. Dauchot, and M. Hecq, Deposition of zinc oxide layers by high-power impulse magnetron sputtering, Journal of Vacuum Science and Technology B: Microelectronics and Nanometer Structures, 25 (2007) 19-21.

[9] F. Horstmann, V. Sittinger, and B. Szyszka, Heat treatable indium tin oxide films deposited with high power pulse magnetron sputtering, Thin Solid Films, 517 (2009) 3178-3182.

121

[10] S. Konstantinidis, J.P. Dauchot, and M. Hecq, Titanium oxide thin films deposited by high-power impulse magnetron sputtering, Thin Solid Films, 515 (2006) 1182- 1186.

[11] M. Hala, R. Vernhes, O. Zabeida, E. Bousser, J.E. Klemberg-Sapieha, R. Sargent, and L. Martinu, Growth and properties of high index Ta2O5 optical coatings prepared by HiPIMS and other methods, (2013).

[12] U. Ozgur, Y. Alivov, C. Liu, A. Teke, M.A. Reshchikov, S. Dogan, V. Avrutin, S.J. Cho, and H. Morkoc, A comprehensive review of ZnO materials and devices, Journal of Applied Physics, 98 (2005) 41301.

[13] U. Ozgur, D. Hofstetter, and H. Morkoc, ZnO devices and applications: A review of current status and future prospects, Institute of Electrical and Electronics Engineers Inc., 445 Hoes Lane / P.O. Box 1331, Piscataway, NJ 08855-1331, United States, 2010, pp. 1255-1268.

[14] C. Klingshirn, ZnO:material, physics and applications, ChemPhysChem, 8 (2007) 783-803.

[15] B. Bayraktaroglu, K. Leedy, and R. Neidhard, Microwave ZnO thin-film transistors, IEEE Electron Device Letters, 29 (2008) 1024-1026.

[16] B. Bayraktaroglu and K. Leedy, Ordered nanocrystalline ZnO films for high speed and transparent thin film transistors, 2011 11th IEEE International Conference on Nanotechnology, NANO 2011, August 15, 2011 - August 19, 2011, IEEE Computer Society, Portland, OR, United states, 2011, pp. 1450-1455.

[17] E. Fortunato, P. Barquinha, and R. Martins, Oxide semiconductor thin-film transistors: A review of recent advances, Advanced Materials, 24 (2012) 2945- 2986.

[18] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, and H. Hosono, Room- temperature fabrication of transparent flexible thin-film transistors using amorphous oxide semiconductors, Nature, 432 (2004) 488-492.

[19] D.C. Look, J.W. Hemsky, and J.R. Sizelove, Residual native shallow donor in ZnO, Physical Review Letters, 82 (1999) 2552-2555.

[20] D.C. Look, D.C. Reynolds, J.W. Hemsky, R.L. Jones, and J.R. Rizelove, Production and annealing of electron irradiation damage in ZnO, Applied Physics Letters, 75 (1999) 811.

[21] S.H. Jeong, B.N. Park, S.B. Lee, and J.H. Boo, Metal-doped ZnO thin films: Synthesis and characterizations, Surface and Coatings Technology, 201 (2007) 5318-5322.

122

[22] S. Lee, D. Cheon, W.J. Kim, K.J. Ahn, and W. Lee, Combined effect of the target composition and deposition temperature on the properties of ZnO:Ga transparent conductive oxide films in pulsed dc magnetron sputtering, Semiconductor Science and Technology, 26 (2011).

[23] D.C. Look, K.D. Leedy, L. Vines, B.G. Svensson, A. Zubiaga, F. Tuomisto, D.R. Doutt, and L.J. Brillson, Self-compensation in semiconductors: The Zn vacancy in Ga-doped ZnO, Physical Review B (Condensed Matter and Materials Physics), 84 (2011) 115202.

[24] C. Nahm, S. Shin, W. Lee, J.I. Kim, D.R. Jung, J. Kim, S. Nam, S. Byun, and B. Park, Electronic transport and carrier concentration in conductive ZnO:Ga thin films, Current Applied Physics, 13 (2013) 415-418.

[25] L. Sanghun, C. Dongkeun, K. Won-Jeong, H. Moon-Ho, and L. Woong, Ga- doped ZnO films deposited with varying sputtering powers and substrate temperatures by pulsed DC magnetron sputtering and their property improvement potentials, Applied Surface Science, 258 (2012) 6537-6544.

[26] K. Matsubara, P. Fons, K. Iwata, A. Yamada, and S. Niki, Room-temperature deposition of Al-doped ZnO films by oxygen radical-assisted pulsed laser deposition, Thin Solid Films, 422 (2002) 176-179.

[27] A.V. Singh, R.M. Mehra, N. Buthrath, A. Wakahara, and A. Yoshida, Highly conductive and transparent aluminum-doped zinc oxide thin films prepared by pulsed laser deposition in oxygen ambient, Journal of Applied Physics, 90 (2001) 5661.

[28] B. Bayraktaroglu and K. Leedy, Pulsed laser deposited ZnO for thin film transistor applications, ZnO Based Thin Films, Nano-Wires, and Nano-Belts for Photonic and Electronic Devices and Sensors - 214th ECS Meeting, October 12, 2008 - October 17, 2008, Electrochemical Society Inc., Honolulu, HI, United states, 2008, pp. 61-73.

[29] S. Lany, J. Osorio-Guillen, and A. Zunger, Origins of the doping asymmetry in oxides: hole doping in NiO versus electron doping in ZnO, Physical Review B (Condensed Matter and Materials Physics), 75 (2007) 241203-1.

[30] F.R. Blom, F.C.M. Van de Pol, G. Bauhuis, and T.J.A. Popma, RF planar magnetron sputtered ZnO films. II. Electrical properties, Thin Solid Films, 204 (1991) 365-376.

[31] C.F. Yu, C.W. Sung, S.H. Chen, and S.J. Sun, Relationship between the photoluminescence and conductivity of undoped ZnO thin films grown with various oxygen pressures, Applied Surface Science, 256 (2009) 792-796.

123

[32] S. Amirhaghi, V. Craciun, D. Craciun, J. Elders, and I.W. Boyd, Low temperature growth of highly transparent c-axis oriented ZnO thin films by pulsed laser deposition, Symposium D: Integrated Processing for Micro and Optoelectronics of the 1993 E-MRS Spring Meeting Conference, 4-7 May 1993, Netherlands, 1994, pp. 321-326.

[33] V. Craciun, J. Elders, J.G.E. Gardeniers, and I.W. Boyd, Characteristics of high quality ZnO thin films deposited by pulsed laser deposition, Applied Physics Letters, 65 (1994) 2963-2965.

[34] V. Craciun, J. Elders, J.G.E. Gardeniers, J. Geretovsky, and I.W. Boyd, Growth of ZnO thin films on GaAs by pulsed laser deposition, Thin Solid Films, 259 (1995) 1-4.

[35] L. Han, F. Mei, C. Liu, C. Pedro, and E. Alves, Comparison of ZnO thin films grown by pulsed laser deposition on sapphire and Si substrates, Physica E: Low- Dimensional Systems and Nanostructures, 40 (2008) 699-704.

[36] S.L. King, J.G.E. Gardeniers, and I.W. Boyd, Pulsed-laser deposited ZnO for device applications, Applied Surface Science, 96-98 (1996) 811-818.

[37] M. Shimizu, T. Shiosaki, and A. Kawabata, Growth of c-axis oriented ZnO thin films with high deposition rate on silicon by CVD method, Journal of Crystal Growth, 57 (1982) 94-100.

[38] K. Haga, M. Kamidaira, Y. Kashiwaba, T. Sekiguchi, and H. Watanabe, ZnO thin films prepared by remote plasma-enhanced CVD method, II-VI Compounds 1999. Ninth International Conference, 1-5 Nov. 1999, Elsevier, Netherlands, 2000, pp. 77-80.

[39] W.J. Maeng, S.-J. Kim, J.-S. Park, K.-B. Chung, and H. Kim. Low temperature atomic layer deposited Al-doped ZnO thin films and associated semiconducting properties. J.Vac.Sci Technology B 30, 031210-1-031210-6. 2012.

[40] J.L. Wang, P.Y. Yang, T.Y. Hsieh, C.C. Hwang, D.C. Shye, and I.C. Lee, The effects of oxygen annealing on the electrical characteristics of hydrothermally grown zinc oxide thin-film transistors, Solid-State Electronics, 77 (2012) 72-76.

[41] S. Elzwawi, D.H.S. Kim, R. Heinhold, M. Lynam, G. Turner, J.G. Partridge, D.G. McCulloch, R.J. Reeves, and M.W. Allen, Device quality ZnO grown using a filtered cathodic vacuum arc, Physica B: Condensed Matter, 407 (2012) 2903- 2906.

[42] R.J. Mendelsberg, S.H.N. Lim, D.J. Milliron, and A. Anders, High rate deposition of high quality ZnO:Al by filtered cathodic arc, 2010 MRS Fall Meeting, November 29, 2010 - December 3, 2010, Materials Research Society, Boston, MA, United states, 2012, pp. 45-51.

124

[43] X.L. Xu, S.P. Lau, and B.K. Tay, Structural and optical properties of ZnO thin films produced by filtered cathodic vacuum arc, 28th International Conference on Metallurgical Coatings and Thin Films, 30 April-4 May 2001, Elsevier, Switzerland, 2001, pp. 244-249.

[44] F.C.M. Van de Pol, F.R. Blom, and T.J.A. Popma, AF planar magnetron sputtered ZnO films. I. Structural properties, Thin Solid Films, 204 (1991) 349-364.

[45] R. Menon, K. Sreenivas, and V. Gupta, Influence of stress on the structural and dielectric properties of RF magnetron sputtered zinc oxide thin film, Journal of Applied Physics, 103 (2008) 094903-1.

[46] K.K. Kim, J.H. Song, H.J. Jung, W.K. Choi, S.J. Park, J.H. Song, and J.Y. Lee, Photoluminescence and heteroepitaxy of ZnO on sapphire substrate (0001) grown by rf magnetron sputtering, Journal of Vacuum Science and Technology A: Vacuum, Surfaces and Films, 18 (2000) 2864-2868.

[47] J. Sang-Hun, K. Bong-Soo, and L. Byung-Teak, Photoluminescence dependence of ZnO films grown on Si(100) by radio-frequency magnetron sputtering on the growth ambient, Applied Physics Letters, 82 (2003) 2625-2627.

[48] J.G. Partridge, E.L.H. Mayes, N.L. McDougall, M.M.M. Bilek, and D.G. McCulloch, Characterization and device applications of ZnO films deposited by high power impulse magnetron sputtering (HiPIMS), Journal of Physics D: Applied Physics, 46 (2013) 165105.

[49] S. Choopun, R.D. Vispute, W. Noch, A. Balsamo, R.P. Sharma, T. Venkatesan, A. Iliadis, and D.C. Look, Oxygen pressure-tuned epitaxy and optoelectronic properties of laser-deposited ZnO films on sapphire, Applied Physics Letters, 75 (1999) 3947-3949.

[50] J.S. Park, S.K. Hong, T. Minegishi, I.H. Im, S.H. Park, T. Hanada, J.H. Chang, M.W. Cho, and T. Yao, The high quality ZnO growth on c-Al2O3 substrate with Cr2O3 buffer layer using plasma-assisted molecular beam epitaxy, Applied Surface Science, 254 (2008) 7786-7789.

[51] R.D. Vispute, V. Talyansky, S. Choopun, R.P. Sharma, T. Venkatesan, M. He, X. Tang, J.B. Halpern, M.G. Spencer, Y.X. Li, L.G. Salamanca-Riba, A.A. Iliadis, and K.A. Jones, Heteroepitaxy of ZnO on GaN and its implications for fabrication of hybrid optoelectronic devices, Applied Physics Letters, 73 (1998) 348-350.

[52] B. Bayraktaroglu, K. Leedy, and R. Neidhard, High frequency ZnO thin film transistors, 66th DRC Device Research Conference Digest, DRC 2008, June 23, 2008 - June 25, 2008, Institute of Electrical and Electronics Engineers Inc., Santa Barbara, CA, United states, 2008, pp. 237-238.

125

[53] K.K. Kim, J.H. Song, H.J. Jung, W.K. Choi, S.J. Park, and J.H. Song, Grain size effects on the photoluminescence of ZnO/-Al2O3 grown by radio-frequency magnetron sputtering, Journal of Applied Physics, 87 (2000) 3573.

[54] P.F. Carcia, R.S. McLean, M.H. Reilly, and J. Nunes, Transparent ZnO thin-film transistor fabricated by rf magnetron sputtering, Applied Physics Letters, 82 (2003) 1117-1119.

[55] K.L. Chopra, Thin film phenomena, McGraw-Hill, 1969.

[56] J.V. Barth, G. Costantini, and K. Kern, Engineering atomic and molecular nanostructures at surfaces, Nature, 437 (2005) 671-679.

[57] G.D.T.S.a.M.H. J.A. Venables, Nucleation and Growth of Thin Films, 47 (1984).

[58] D.A. Smith, Thin-Film Deposition: Principles and Practices, Thin-Film Deposition: Principles and Practices, McGraw-Hill, Inc., 1995.

[59] J.E. Greene, Handbook of Deposition Technologies for Films and Coatings, Handbook of Deposition Technologies for Films and Coatings, Noyes Publications.

[60] J.A. Venables, Rate Equations Approaches to Thin Film Nucleation Kinetics, Rate Equations Approaches to Thin Film Nucleation Kinetics, 1972.

[61] B.A. Movchan and A.V. Demchishin, Investigation of the structure and properties of thick vacuum- deposited films of nickel, titanium, tungsten, alumina and zirconium dioxide, Fizika Metallov i Metallovedenie, 28 (1969) 653-660.

[62] J.A. Thornton, Influence of apparatus geometry and deposition conditions on the structure and topography of thick sputtered coatings, Conference on Structure/Property Relationships in Thick Films and Bulk Coatings, 28-30 Jan. 1974, USA, 1974, pp. 666-670.

[63] M.A. Lieberman and A.J. Lichtenberg, Principles of Plasma Discharges and Meterials Processing, John Wiley and Sons, Inc., 1994.

[64] B. Chapman, Glow Discharge Processes: Sputtering and Plasma Etching, Wiley- Interscience, 1980.

[65] D. Guttler, B. Abendroth, R. Grotzschel, W. Moller, and D. Depla, Mechanisms of target poisoning during magnetron sputtering as investigated by real-time in situ analysis and collisional computer simulation, Applied Physics Letters, 85 (2004) 6134-6136.

126

[66] J. Musil, P. Baroch, J. Vlcek, K.H. Nam, and J.G. Han, Reactive magnetron sputtering of thin films: present status and trends, Thin Solid Films, 475 (2005) 208-218.

[67] I. Safi, Recent aspects concerning DC reactive magnetron sputtering of thin films: a review, Surface and Coatings Technology, 127 (2000) 203-218.

[68] J.W. Bradley, S.K. Karkari, and A. Vetushka, A study of the transient plasma potential in a pulsed bi-polar dc magnetron discharge, Plasma Sources Science and Technology, 13 (2004) 189-198.

[69] A. Vetushka, S.K. Karkari, and J.W. Bradley, Two-dimensional spatial survey of the plasma potential and electric field in a pulsed bipolar magnetron discharge, Journal of Vacuum Science and Technology A: Vacuum, Surfaces and Films, 22 (2004) 2459-2468.

[70] A. Mishra, P.J. Kelly, and J.W. Bradley, The evolution of the plasma potential in a HiPIMS discharge and its relationship to deposition rate, Plasma Sources Science and Technology, 19 (2010) 045014.

[71] J.T. Gudmundsson, J. Alami, and U. Helmersson, Evolution of the electron energy distribution and plasma parameters in a pulsed magnetron discharge, Applied Physics Letters, 78 (2001) 3427.

[72] J.T. Gudmundsson, J. Alami, and U. Helmersson, Spatial and temporal behavior of the plasma parameters in a pulsed magnetron discharge, Surface & Coatings Technology, 161 (2002) 249-256.

[73] A.D. Pajdarova, J. Vlcek, P. Kudlacek, and J. Luka, Electron energy distributions and plasma parameters in high-power pulsed magnetron sputtering discharges, Plasma Sources Science and Technology, 18 (2009).

[74] N. Brenning, Axnas, M.A. Raadu, D. Lundin, and U. Helmerson, A bulk plasma model for dc and HiPIMS magnetrons, Plasma Sources, Science and Technology, 17 (2008) 045009.

[75] J. Bohlmark, J. Alami, C. Christou, A.P. Ehiasarian, and U. Helmersson, Ionization of sputtered metals in high power pulsed magnetron sputtering, Journal of Vacuum Science & Technology A (Vacuum, Surfaces, and Films), 23 (2005) 18-22.

[76] A. Anders, J. Andersson, and A. Ehiasarian, High power impulse magnetron sputtering: Current-voltage-time characteristics indicate the onset of sustained self-sputtering, Journal of Applied Physics, 102 (2007) 113303-113311.

127

[77] N. Hosokawa, T. Tsukada, and T. Misumi, Self-sputtering phenomena in high- rate coaxial cylindrical magnetron sputtering, Journal of Vacuum Science and Technology, 14 (1977) 143-146.

[78] A. Anders, J. Capek, M. Hala, and L. Martinu, The recycling trap: A generalized explanation of discharge runaway in high-power impulse magnetron sputtering, Journal of Physics D: Applied Physics, 45 (2012) 012003.

[79] A. Anders and G.Y. Yushkov, Plasma "anti-assistance" and "self-assistance" to high power impulse magnetron sputtering, Journal of Applied Physics, 105 (2009).

[80] A. Anders, P. Ni, and A. Rauch, Drifting localization of ionization runaway: Unraveling the nature of anomalous transport in high power impulse magnetron sputtering, Journal of Applied Physics, 111 (2012).

[81] J. Andersson and A. Anders, Self-sputtering far above the runaway threshold: An extraordinary metal-ion generator, Physical Review Letters, 102 (2009) 045003.

[82] T.P. Drusedau, Investigation of the throw distance of sputtered atoms under consideration of gas rarefaction, Surface and Coatings Technology, 174-175 (2003) 470-474.

[83] S.M. Rossnagel, Gas density reduction effects in magnetrons, Journal of Vacuum Science and Technology A, 6 (1988) 19-24.

[84] D.W. Hoffman, A sputtering wind, Journal of Vacuum Science and Technology A, 3 (1985) 561-566.

[85] A. Anders, Discharge physics of high power impulse magnetron sputtering, Surface and Coatings Technology, 205 (2011) S1-S9.

[86] K. Macak, V. Kouznetsov, J. Schneider, U. Helmersson, and I. Petrov, Ionized sputter deposition using an extremely high plasma density pulsed magnetron discharge, 46th National Symposium of the American Vacuum Society. Vacuum, Surfaces, and Films, 25-29 Oct. 1999, AIP for American Vacuum Soc, USA, 2000, pp. 1533-1537.

[87] M. Palmucci, N. Britun, S. Konstantinidis, and R. Snyders, Rarefaction windows in a high-power impulse magnetron sputtering plasma, Journal of Applied Physics, 114 (2013).

[88] K. Yukimura, R. Mieda, H. Tamagaki, and T. Okimoto, Electrical characteristics of arc-free high-power pulsed sputtering glow plasma, Surface and Coatings Technology, 202 (2008) 5246-5250.

128

[89] K. Yukimura, R. Mieda, K. Azuma, H. Tamagaki, and T. Okimoto, Voltage- current characteristics of a high-power pulsed sputtering (HPPS) glow discharge and plasma density estimation, Nuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and Atoms, 267 (2009) 1692-1695.

[90] D. Lundin, N. Brenning, D. Jadernas, P. Larsson, E. Wallin, M. Lattemann, M.A. Raadu, and U. Helmersson, Transition between the discharge regimes of high power impulse magnetron sputtering and conventional direct current magnetron sputtering, Plasma Sources Science and Technology, 18 (2009) 045008.

[91] J. Andersson, A.P. Ehiasarian, and A. Anders, Observation of Ti4+ ions in a high power impulse magnetron sputtering plasma, Applied Physics Letters, 93 (2008) 071504.

[92] M. Audronis, G. Abrasonis, F. Munnik, R. Heller, P. Chapon, and V. Bellido- Gonzalez, Diffusive racetrack oxidation in a Ti sputter target by reactive high power impulse magnetron sputtering, Journal of Physics D: Applied Physics, 45 (2012) 375203.

[93] K. Sarakinos, J. Alami, C. Klever, and M. Wuttig, Process stabilization and enhancement of deposition rate during reactive high power pulsed magnetron sputtering of zirconium oxide, Surface and Coatings Technology, 202 (2008) 5033-5035.

[94] E. Wallin and U. Helmersson, Hysteresis-free reactive high power impulse magnetron sputtering, Thin Solid Films, 516 (2008) 6398-6401.

[95] M. Hala, J. Capek, O. Zabeida, J.E. Klemberg-Sapieha, and L. Martinu, Hysteresis-free deposition of niobium oxide films BY HiPIMS using different pulse management strategies, Journal of Physics D: Applied Physics, 45 (2012) 055204.

[96] M. Aiempanakit, U. Helmersson, A. Aijaz, P. Larsson, R. Magnusson, J. Jensen, and T. Kubart, Effect of peak power in reactive high power impulse magnetron sputtering of titanium dioxide, Surface & Coatings Technology, 205 (2011) 4828-4831.

[97] T. Kubart, M. Aiempanakit, J. Andersson, T. Nyberg, S. Berg, and U. Helmersson, Studies of hysteresis effect in reactive HiPIMS deposition of oxides, Surface and Coatings Technology, 205 (2011) S303-S306.

[98] M. Audronis, V. Bellido-Gonzalez, and B. Daniel, Control of reactive high power impulse magnetron sputtering processes, Surface and Coatings Technology, 204 (2010) 2159-2164.

129

[99] M. Audronis and V. Bellido-Gonzalez, Hysteresis behaviour of reactive high power impulse magnetron sputtering, Thin Solid Films, 518 (2010) 1962-1965.

[100]A. Surpi, T. Kubart, D. Giordani, M. Tosello, G. Mattei, M. Colasuonno, and A. Patelli, HiPIMS deposition of TiOx in an industrial-scale apparatus: Effects of target size and deposition geometry on hysteresis, Surface and Coatings Technology, 235 (2013) 714-719.

[101]D. Depla, S. Heirwegh, S. Mahieu, J. Haemers, and R. De Gryse, Understanding the discharge voltage behavior during reactive sputtering of oxides, Journal of Applied Physics, 101 (2007) 13301.

[102]V. Stranak, M. Quaas, H. Wulff, Z. Hubicka, S. Wrehde, M. Tichy, and R. Hippler, Formation of TiOx films produced by high-power pulsed magnetron sputtering, Journal of Physics D: Applied Physics, 41 (2008).

[103]C. Nouvellon, M. Michiels, J.P. Dauchot, C. Archambeau, F. Laffineur, E. Silberberg, S. Delvaux, R. Cloots, S. Konstantinidis, and R. Snyders, Deposition of titanium oxide films by reactive High Power Impulse Magnetron Sputtering (HiPIMS): Influence of the peak current value on the transition from metallic to poisoned regimes, Surface and Coatings Technology, 206 (2012) 3542-3549.

[104]F. Magnus, B. Agnarsson, A.S. Ingason, K. Leosson, S. Olafsson, and J.T. Gudmundsson, Growth of TiO2 thin films on Si(001) and SiO2 by reactive high power impulse magnetron sputtering, 2011 MRS Spring Meeting, April 25, 2011 - April 29, 2011, Materials Research Society, San Francisco, CA, United states, 2012, pp. 39-44.

[105]M. Aiempanakit, A. Aijaz, D. Lundin, U. Helmersson, and T. Kubart, Understanding the discharge current behavior in reactive high power impulse magnetron sputtering of oxides, Journal of Applied Physics, 113 (2013) 133302.

[106]C. Huo, M.A. Raadu, D. Lundin, J.T. Gudmundsson, A. Anders, and N. Brenning, Gas rarefaction and the time evolution of long high-power impulse magnetron sputtering pulses, Plasma Sources Science and Technology, 21 (2012).

[107]A. Hecimovic, K. Burcalova, and A.P. Ehiasarian, Origins of ion energy distribution function (IEDF) in high power impulse magnetron sputtering (HIPIMS) plasma discharge, Journal of Physics D: Applied Physics, 41 (2008) 095203-095213.

[108]A. Hecimovic and A.P. Ehiasarian, Time evolution of ion energies in HIPIMS of chromium plasma discharge, Journal of Physics D: Applied Physics, 42 (2009) 135209-135218.

[109]J. Bohlmark, M. Lattemann, J.T. Gudmundsson, A.P. Ehiasarian, Y. Aranda Gonzalvo, N. Brenning, and U. Helmersson, The ion energy distributions and ion

130

flux composition from a high power impulse magnetron sputtering discharge, Thin Solid Films, 515 (2006) 1522-1526.

[110]J.T. Gudmundsson, Ionization mechanism in the high power impulse magnetron sputtering (HiPIMS) discharge, 17th International Vacuum Congress (IVC-17), 13th International Conference on Surface Science, (ICSS-13) and International Conference on Nanoscience and Technology (ICN+T 2007), 2-6 July 2007, IOP Publishing Ltd., UK, 2008, p. 082013.

[111]J.T. Gudmundsson, P. Sigurjonsson, P. Larsson, D. Lundin, and U. Helmersson, On the electron energy in the high power impulse magnetron sputtering discharge, Journal of Applied Physics, 105 (2009) 123302.

[112]J. Bohlmark, J. Alami, C. Christou, A.P. Ehiasarian, and U. Helmersson, Ionization of sputtered metals in high power pulsed magnetron sputtering, Journal of Vacuum Science and Technology A: Vacuum, Surfaces and Films, 23 (2005) 18-22.

[113]A. Vetushka and A.P. Ehiasarian, Plasma dynamic in chromium and titanium HIPIMS discharges, Journal of Physics D: Applied Physics, 41 (2008) 015204.

[114]P. Poolcharuansin and J.W. Bradley, Short- and long-term plasma phenomena in a HiPIMS discharge, Plasma Sources, Science and Technology, 19 (2010) 025010.

[115]S. Mahieu and D. Depla, Correlation between electron and negative O- ion emission during reactive sputtering of oxides, Applied Physics Letters, 90 (2007) 121117-1.

[116]K. Tominaga, T. Murayama, Y. Sato, and I. Mori, Energetic oxygen particles in the reactive sputtering of Zn targets in Ar/O2 atmospheres, 14th International Vacuum Congress (IVC-14). 10th International Conference on Solid Surfaces (ICSS-10). 5th International Conference on Nanometre-scale Science and Technology (NANO-5). 10th International Conference on Quantitative Surface Analysis (QSA-10), 31 Aug.-4 Sept. 1998, Elsevier, Switzerland, 1999, pp. 81-84.

[117]M. Zeuner, H. Neumann, J. Zalman, and H. Biederman, Sputter process diagnostics by negative ions, Journal of Applied Physics, 83 (1998) 5083-5086.

[118]P. Pokorny, M. Misina, J. Bulir, J. Lancok, P. Fitl, J. Musil, and M. Novotny, Investigation of the Negative Ions in Ar/O2 Plasma of Magnetron Sputtering Discharge with Al:Zn Target by Ion Mass Spectrometry, Plasma Processes and Polymers, 8 (2011) 459-464.

[119]S. Mraz and J.M. Schneider, Influence of the negative oxygen ions on the structure evolution of transition metal oxide thin films, Journal of Applied Physics, 100 (2006).

131

[120]M. Bowes, P. Poolcharuansin, and J.W. Bradley, Negative ion Energy Distributions in Reactive HiPIMS, Journal of Physics D: Applied Physics, 46 (2013) 045204.

[121]S. Mraz and J.M. Schneider, Energy distribution of O- ions during reactive magnetron sputtering, Applied Physics Letters, 89 (2006).

[122]M.D. Howell, S. Akamine, L.J. LaComb, Jr., B. Hadimioglu, T.R. Albrecht, B.T. Khuri-Yakub, L.C. Goddard, and T.E. Carver, Dependence of orientation and electrical properties in sputtered ZnO upon deposition conditions and substrate, IEEE 1988 Ultrasonics Symposium. Proceedings (IEEE Cat. No.88CH2578-3), 2- 5 Oct. 1988, IEEE, New York, NJ, USA, 1988, pp. 677-680.

[123]I. Petrov, F. Adibi, J.E. Greene, L. Hultman, and J.E. Sundgren, Average energy deposited per atom: A universal parameter for describing ion-assisted film growth?, Applied Physics Letters, 63 (1993) 36-38.

[124]A. Anders. A strucutre zone diagram including plasma-based deposition and ion etching. Thin Solid Films 518, 4087-4090. 2010.

[125]A. Amin, D. Kohl, and M. Wuttig, The role of energetic ion bombardment during growth of TiO2 thin films by reactive sputtering, Journal of Physics D: Applied Physics, 43 (2010).

[126]K. Sarakinos, J. Alami, and M. Wuttig, Process characteristics and film properties upon growth of TiOx films by high power pulsed magnetron sputtering, Journal of Physics D (Applied Physics), 40 (2007) 2108-2114.

[127]Schneider, Handbook of Thin Film Processing Technology, CRC Press, 1998.

[128]A.P. Ehiasarian, R. New, W.D. Munz, L. Hultman, U. Helmersson, and V. Kouznetsov, Influence of high power densities on the composition of pulsed magnetron plasmas, Vacuum, 65 (2002) 147-154.

[129]J. Alami, K. Sarakinos, G. Mark, and M. Wuttig, On the deposition rate in a high power pulsed magnetron sputtering discharge, Applied Physics Letters, 89 (2006) 154104.

[130]NIST Mass Spectrometry Data CenterCollection. NIST Chemistry WebBook: Argon Mass Spectrum (Electron Ionization). NIST . 2014. NIST Mass Spectrometry Data Center Collection.

[131]D. Horwat and A. Anders, Compression and strong rarefaction in high power impulse magnetron sputtering discharges, Journal of Applied Physics, 108 (2010) 123306-123312.

132

[132]P.Y. Jouan, L. Le Brizoual, M. Ganciu, C. Cardinaud, S. Tricot, and M.A. Djouadi, HiPIMS ion energy distribution measurements in reactive mode, Institute of Electrical and Electronics Engineers Inc., 445 Hoes Lane / P.O. Box 1331, Piscataway, NJ 08855-1331, United States, 2010, pp. 3089-3094.

[133]J. Hupkes, J.I. Owen, S.E. Pust, and E. Bunte, Chemical etching of zinc oxide for thin-film silicon solar cells, ChemPhysChem, 13 (2012) 66-73.

[134]J.W. Shin, J.Y. Lee, T.W. Kim, Y.S. No, W.J. Cho, and W.K. Choi, Growth mechanisms of thin-film columnar structures in zinc oxide on p-type silicon substrates, Applied Physics Letters, 88 (2006).

[135]J.I. Sohn, W.K. Hong, S. Lee, S. Lee, J. Ku, Y.J. Park, J. Hong, S. Hwang, K.H. Park, J.H. Warner, S. Cha, and J.M. Kim, Surface energy-mediated construction of anisotropic semiconductor wires with selective crystallographic polarity, Sci. Rep., 4 (2014).

[136]I. Petrov, F. Adibi, J.E. Greene, W.D. Sproul, and W.D. Munz, Use of an externally applied axial magnetic field to control ion/neutral flux ratios incident at the substrate during magnetron sputter deposition, Journal of Vacuum Science & Technology A (Vacuum, Surfaces, and Films), 10 (/9) 3283-3287.

[137]M. Chen, X. Wang, Y.H. Yu, Z.L. Pei, X.D. Bai, C. Sun, R.F. Huang, and L.S. Wen, X-ray photoelectron spectroscopy and auger electron spectroscopy studies of Al-doped ZnO films, Applied Surface Science, 158 (2000) 134-140.

133

APPENDIX A

SUPPLEMENTAL ION ENERGY DISTRIBUTION DATA

This appendix contains ion energy measurements as a function of pulse length and parameter. Because the dwell time for the IED measurements was 300 ms and each measurement contained

+ + + + Figure A.1 The ion energy distributions for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO+ measured in a reactive-HiPIMS plasma with a ZnO target with 650 V with different pulse lengths at 100 Hz with 20 mTorr and an Opp = 0.8 mTorr.

134

+ + + + Figure A.2 The ion energy distributions for (a) Ar , (b) O , (c) O2 , (d) Zn and (e) ZnO+ measured in a reactive-HiPIMS plasma with a ZnO target with 650 V 200 s pulses with different pulse frequencies at 20 mTorr and an Opp = 0.8 mTorr.

135

APPENDIX B

SUPPLEMENTAL MICROSTRUCTURAL DATA

Figure B.1 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a Zn target in 20 mTorr with Opp = 8 mTorr with different (a) applied target voltages, (b) pulse lengths, and (c) pulse frequency.

136

Figure B.2 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a Zn target with different (a) total pressures and O2/Ar ratio of 0.4 and (b) a fixed total pressure of 20 mTorr and different Opp.

Figure B.3 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a ZnO target in 20 mTorr with Opp= 0.8 mTorr with different (a) applied target voltages, (b) pulse lengths, and (c) pulse frequency.

137

Figure B.4 ZnO (002) rocking curve XRD patterns for ZnO films deposited from a ZnO target with different (a) total pressures and O2/Ar ratio of 0.04 and (b) a fixed total pressure of 20 mTorr and different Opp.

Table B.1 Deposition rate, diffraction angle and grain size data for ZnO films grown by reactive-HiPIMS of a Zn target with voltage and pulse parameters systematically varied.

Table B.2 Deposition rate, diffraction angle and grain size data for ZnO films grown by reactive-HiPIMS of a Zn target with total pressure and oxygen partial pressure systematically varied.

138

Table B.3 Deposition rate, diffraction angle and grain size data for ZnO films grown by reactive-HiPIMS of a ZnO target with voltage and pulse parameters systematically varied.

Table B.4 Deposition rate, diffraction angle and grain size data for ZnO films grown by reactive-HiPIMS of a ZnO target with total pressure and oxygen partial pressure systematically varied.

139

Figure B.5 Additional (a) wide angle XRD and (b) (002) rocking curve measurements for ZnO films grown with 1500 V, 200 s, 100 Hz with an O2/Ar = 0.04 and a target-substrate distance of 10 cm.

Figure B.6 (a) Wide angle XRD and (b) (002) rocking curve measurements for ZnO films grown with 1500 V, 200 s, 100 Hz with an O2/Ar = 0.04 and a target-substrate distance of 5 cm.

140

Figure B.7 Topography AFM scans for ZnO films grown with HiPIMS with 1500 V, 200 s 100 Hz pulses with different pressures and O2/Ar = 0.04. The substrate to target distance was fixed at 5 cm.

141

Figure B.8 (a) Wide angle XRD and (b) (002) rocking curve measurements for ZnO films grown with 1500 V, 200 s, 100 Hz with an O2/Ar = 0.04 on a Kapton substrate.

Figure B.9 TEM of film cross-section for a ZnO film deposited from reactive HiPIMS of a ZnO target with 1500 V 200 s 100 Hz at 10 mTorr (Opp= 0.4 mTorr) with 10 cm target-substrate separation.

142

Figure B.10 TEM of cross-section of ZnO film deposited from reactive HiPIMS of a ZnO target with 1500 V 200 s 100 Hz at 5 mTorr (Opp= 0.2 mTorr) with 10 cm target-substrate separation.

Figure B.11 TEM of cross-section of film-substrate interface for ZnO deposited from reactive HiPIMS of a ZnO target with 1500 V 200 s 100 Hz at 5 mTorr (Opp= 0.2 mTorr) with 10 cm target-substrate separation.

143

Figure B.12 TEM of cross-section of film-substrate interface for ZnO deposited from reactive HiPIMS of a ZnO target with 1500 V 200 s 100 Hz at 3 mTorr (Opp= 0.12 mTorr) with 10 cm target-substrate separation.

144

APPENDIX C

SUPPLEMENTAL TFT DEVICE CHARACTERIZATION DATA

Figure C.1 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

145

Figure C.2 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

146

Figure C.3 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

147

Figure C.4 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 50 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

Figure C.5 Maximum drain current as a function of gate voltages for TFT devices from Fig. C.1 and Fig. C.2 with 50 m long channels.

148

Figure C.6 Maximum drain current as a function of gate voltages for TFT devices from Fig. C.2 and Fig. C.3 with 60 m long channels.

149

Figure C.7 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

150

Figure C.8 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

151

Figure C.9 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

152

Figure C.10 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

153

Figure C.11 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

154

Figure C.12 TFT drain current measurements as a function of source-drain potential for applied gate voltages of 0-80 V. The TFT channel was ZnO synthesized by HiPIMS of a ZnO target with 1500 V 200 s 100 Hz pulses with 20 mTorr (Opp = 0.8 mTorr).

155

APPENDIX D

X-RAY PHOTOELECTRON SPECTROSCOPY

The chemical bonding of the films was analyzed with x-ray photoelectric spectroscopy (XPS), using the in-situ XPS system. The load-lock seen in Fig 4.1 allowed the films to be transferred from the deposition chamber directly to the XPS analytical chamber without leaving vacuum. This eliminated contamination from the outside environment and assured that the chemistry measured in the XPS was a result of the deposition and not outside influences. Survey scans from 0 eV to 1100 eV were taken for all the films. High resolution scans were taken of the O 1s, and Zn 2p3/2 peaks at binding energies 531 eV and 1021.5 eV, respectively. Analysis of the XPS data was performed using Casa software. Sensitivity factors of 9.69 for the Zn 2p3/2, 2.49 for the O 1s were used to correct for the orbitals’ different X-ray adsorption cross-sections.

Figure D.1 shows the benefits of the in-situ XPS analysis. The survey scan for the film measured in-situ (top line in Fig. D.1a) shows a distinct absence of the carbon peak.

A peak for carbon is seen in the film after exposure to atmosphere. Other differences can be seen in the high resolution scan of the oxygen 1s peak, shown in Fig D.1b. In both the in-situ and exposed analyses, the O 1s peak in Figure D.1b can be fitted to 3 separate peaks centered at 530.1 eV, 531.6 eV and 532.5 eV. Chen et al attributed these peaks (in order of increasing binding energy) to oxygen in Zn-O bonds, O2- in oxygen deficient regions and chemisorbed oxygen or OH on the film surface [137]. From Figure D.1b it

156 can be seen that the contribution from chemisorbed oxygen and OH (the peak at 532.5 eV) is increased after exposure to atmosphere.

Figure D.1 (a) Survey and (b) high resolution O 1s XPS of ZnO film measured in- situ and after exposure to atmosphere.

157

Figure D. 2 High resolution XPS measurements of the O 1s peak for ZnO films synthesized by reactive-HiPIMS of a ZnO target with different total pressures and O2/Ar = 0.04, 1500 V, 200 s and 100 Hz.

158