Growth, Optimization, and Characterization of Transition Metal

GROWTH, OPTIMIZATION, AND CHARACTERIZATION OF TRANSITION

METAL NITRIDES AND TRANSITION METAL OXIDES FOR ELECTRONIC

AND OPTICAL APPLICATIONS

Thesis

Submitted to

The School of Engineering of the

UNIVERSITY OF DAYTON

In Partial Fulfillment of the Requirements for

The Degree of

Master of Science in Electro-Optics

Zachary Biegler

UNIVERSITY OF DAYTON

Dayton, Ohio

December 2019

GROWTH, OPTIMIZATION, AND CHARACTERIZATION OF TRANSITION

METAL NITRIDES AND TRANSITION METAL OXIDES FOR ELECTRONIC AND

OPTICAL APPLICATIONS

Name: Biegler, Zachary Jay

APPROVED BY:

Andrew Sarangan, Ph.D., P.E. Amber Reed, Ph.D. Advisory Committee Chairman Committee Member Professor Materials Engineer Department of Electro-Optics and Photonics AFRL/RXAN

Partha Banerjee, Ph.D. Committee Member Professor Department of Electro-Optics and Photonics

Robert J. Wilkens, Ph.D., P.E. Eddy M. Rojas, Ph.D., M.A., P.E. Associate Dean for Research and Innovation Dean, School of Engineering Professor School of Engineering

ABSTRACT

GROWTH, OPTIMIZATION, AND CHARACTERIZATION OF TRANSITION

METAL NITRIDES AND TRANSITION METAL OXIDES FOR ELECTRONIC AND

OPTICAL APPLICATIONS

Name: Biegler, Zachary J. University of Dayton

Advisor: Dr. Andrew Sarangan

The next generation of electronic and optical devices require high quality, crystalline materials in order to obtain relevant properties for novel devices. Two classes of materials offer unique material properties that can satisfy the requirements for next generation devices. These two classes of materials are the transition metal nitrides

(TMNs) and transition metal oxides (TMOs). These materials offer electronic properties that range from conductive, metallic, materials to semiconducting and insulating materials. However, for many optical and electronic applications, the band structure and crystalline symmetries must be preserved. This work examines the growth and characterization of the TMN materials AlN and ScN as well as the TMO materials VO2 and TiO2. In all these materials, the crystalline structure plays and extremely important role in the desired properties. In addition, incorporation of other impurities can detrimentally impact the functionality of these film materials. In order to minimize the impurity incorporation and maintain the crystalline structure, the growth of AlN, ScN,

VO2, and TiO2 films by various deposition techniques were examined and optimized.

This allowed growth of high quality TMN and TMO materials that resulted in

iii characterization and optimization of the relevant optical, electronic, and structural properties and, somewhat, the degree to which these properties could be tuned through growth conditions.

DEDICATION

This work is dedicated to my parents and siblings.

ACKNOWLEDGMENTS

This work would not have been possible without the help and guidance from so many different people. My advisors Dr. Andrew Sarangan and Dr. Amber Reed both spent many hours helping me get into a field of study of which I was not familiar. Words cannot express how much their guidance and patience means to me. I also would like to thank Dr. Kurt Eyink, Dr. Tyson Back, Dr. John Centar, and Dr. David Look for all of their help taking data and discussing results obtained through the characterization of these films. Dr. Dean Brown provided COMSOL simulations of the magnetic fields in the unbalanced magnetron system. In addition, none of this work could have been possible without the help of Hadley Smith and Rachel Adams who helped with sample growth and

XRD characterization of some of the TMN materials. Dr. Pengfei Guo was instrumental in the growth of VO2 by thermal oxidation. Last, but certainly not least, I want to thank

Madelyn Hill for all of the AFM surface analysis that was performed in this work as well as Dr. Albert Hilton for the piezoelectric force microscopy measurements.

Additionally, none of the transition metal nitride work could have commenced without the support of AFOSR under the award number FA9550-17RYCOR490.

TABLE OF CONTENTS

ABSTRACT ...... iii

DEDICATION ...... v

ACKNOWLEDGMENTS ...... vi

LIST OF FIGURES ...... x

LIST OF ABBREVIATIONS AND NOTATIONS ...... xv

CHAPTER 1 INTRODUCTION TO MATERIAL SYSTEMS ...... 1

1.1 Transition Metal Nitrides ...... 2

1.1.1 Titanium Nitride...... 3

1.1.2 Scandium Nitride ...... 4

1.1.3 Aluminum Nitride ...... 5

1.2 Transition Metal Oxides ...... 9

1.2.1 Vanadium Oxide ...... 9

1.2.2 Titanium Oxide ...... 12

CHAPTER 2 GROWTH SYSTEMS ...... 15

2.1 DC Sputtering System ...... 15

2.1.1 Sputter Deposition ...... 15

2.1.2 Reactive Sputtering ...... 17

2.1.3 Magnetron Sputtering ...... 18

2.1.4 Sputtering System ...... 21

2.2 Controllably-Unbalanced DC Reactive Magnetron Sputtering ...... 22

2.2.1 Unbalanced Magnetron Sputtering ...... 23

vii

2.2.2 Controllably-Unbalanced Sputtering System ...... 26

2.3 Ion Assisted Evaporation System ...... 30

2.3.1 Thermal Evaporation ...... 31

2.3.2 Ion Assisted Evaporation ...... 33

2.3.3 IAD System ...... 34

CHAPTER 3 CHARACTERIZATION TECHNIQUES ...... 37

3.1 X-Ray Diffractometry ...... 37

3.1.1 Introduction to XRD ...... 38

3.1.2 Grazing Incident X-Ray Diffraction ...... 44

3.1.3 Coupled Scans ...... 46

3.1.3.1 Symmetric Coupled Scans ...... 46

3.1.3.2 Asymmetric Coupled Scans ...... 52

3.1.4 Rocking Curves ...... 54

3.1.5 Pole Figures ...... 56

3.2 X-Ray Photoelectron Spectroscopy ...... 59

3.3 Secondary Ion Mass Spectrometry ...... 62

3.4 Spectroscopic Ellipsometry ...... 63

3.5 Hall Effect and Transport ...... 68

CHAPTER 4 SCANDIUM NITRIDE GROWTH AND CHARACTERIZATION ...... 72

4.1 Nitrogen Fraction ...... 76

4.2 Magnetron Power ...... 79

4.3 Substrate Temperature...... 81

4.4 Final Optimization...... 84

viii

4.5 ScN Optical and Electronic Properties ...... 91

CHAPTER 5 ALUMINUM NITRIDE GROWTH AND CHARACTERIZATION ...... 99

5.1 Nitrogen Fraction ...... 99

5.2 Substrate Temperature...... 104

5.3 Coil Current ...... 108

5.4 Magnetron Power ...... 112

5.5 Final Optimization...... 116

5.6 Piezoelectric Properties ...... 119

5.6.1 TiN on Al2O3...... 119

5.6.2 AlN on TiN on Al2O3 ...... 121

CHAPTER 6 TITANIUM OXIDE GROWTH AND CHARACTERIZATION ...... 128

CHAPTER 7 VANADIUM OXIDE GROWTH AND CHARACTERIZATION ...... 131

7.1 Deposition of VO2 on Al2O3 ...... 131

7.2 Deposition of VO2on TiO2 ...... 134

7.3 VO2 Films by Thermal Oxidation ...... 137

CONCLUSIONS AND FUTURE WORK ...... 151

REFERENCES ...... 152

LIST OF FIGURES

Figure 1.1: Rock-salt crystal structure...... 3

Figure 1.2: Wurtzite crystal structure ...... 6

Figure 1.3: Comparison of piezoelectric material maximum use temperatures ...... 7

Figure 1.4: Al1-xScxN piezoelectric coefficients with varying Sc content ...... 8

Figure 1.5: VO2 high temperature rutile phase ...... 10

Figure 1.6: VO2 room temperature monoclinic phase ...... 11

Figure 1.7: TiO2 rutile and anatase structures ...... 13

Figure 2.1: Sputter deposition ...... 17

Figure 2.2: Magnetron sputtering field lines...... 18

Figure 2.3: Simulated electron trajectory in magnetron field ...... 19

Figure 2.4: Magnetron plasma at target ...... 20

Figure 2.5: Magnetron sputter target ...... 20

Figure 2.6: Denton Vacuum Explorer 14 sputter system ...... 22

Figure 2.7: Balanced and unbalanced field lines ...... 23

Figure 2.8: Unbalanced magnetron field lines ...... 24

Figure 2.9: Simulated electron trajectories in an unbalanced system ...... 24

Figure 2.10: Plasma in an unbalanced system ...... 25

Figure 2.11: Controllably-unbalanced reactive magnetron sputtering system ...... 26

Figure 2.12: Substrate heating calibration ...... 27

Figure 2.13: COMSOL 퐵 Simulation ...... 29

Figure 2.14: Plasma comparison of balanced and unbalanced configuration ...... 30

Figure 2.15: Thermal evaporation...... 31

Figure 2.16: Electrom beam evaporation ...... 32

Figure 2.17: Ion assisted evaporation ...... 34

Figure 2.18: Ion assisted evaporation system ...... 35

Figure 3.1: Bragg diffraction ...... 39

Figure 3.2: Powder XRD system diagram ...... 41

Figure 3.3: Powder diffraction scan ...... 42

Figure 3.4: Powder diffraction – differing planes ...... 43

Figure 3.5: Example of a GIXRD scan ...... 45

Figure 3.6: Reciprocal space map of Al2O3 ...... 47

Figure 3.7: Reciprocal space map of Al2O3 – symmetric axis ...... 48

Figure 3.8: Thin TiN coupled scan ...... 49

Figure 3.9: Thick TiN coupled scan ...... 52

Figure 3.10: Reciprocal space map of Al2O3 – asymmetric axis ...... 53

Figure 3.11: Rocking curve FWHM comparison ...... 55

Figure 3.12: Purely polycrystalline pole figure ...... 57

Figure 3.13: Single crystal pole figure ...... 58

Figure 3.14: CasaXPS peak fitting...... 60

Figure 3.15: Optical propagation in a film...... 65

Figure 3.16: Hall Effect configuration ...... 70

Figure 4.1: ScN hexagonal structure ...... 73

Figure 4.2: In plane rotation...... 74

Figure 4.3: ScN (111) pole figure ...... 75

Figure 4.4: Al2O3 (1-12) pole figure ...... 76

Figure 4.5: ScN nitrogen fraction surveys ...... 78

Figure 4.6: ScN nitrogen fractions – zoomed in ...... 79

Figure 4.7: ScN original power series surveys ...... 80

Figure 4.8: ScN power series – zoomed in ...... 81

Figure 4.9: ScN temperature series surveys ...... 82

Figure 4.10: ScN temperature series detailed scans...... 83

Figure 4.11: ScN nitrogen fraction at 860°C substrate temperature ...... 84

Figure 4.12: Power series optimization at high temp – original set...... 85

Figure 4.13: ScN power series optimization at higher temperature ...... 86

Figure 4.14: Surface AFM measurements of ScN power series ...... 88

Figure 4.15: Rocking curve of 75W ScN sample ...... 89

Figure 4.16: ScN power series rocking curve FWHMs ...... 90

Figure 4.17: ScN full range optical constants ...... 92

Figure 4.18: ScN bandgap examination of power series ...... 93

Figure 4.19: ScN room temperature Hall Effect results ...... 94

Figure 4.20: ScN SIMS analysis ...... 95

Figure 4.21: ScN power series planar spacings ...... 96

Figure 5.1: AlN nitrogen fraction coupled scans ...... 101

Figure 5.2: AlN overlaid nitrogen fraction coupled scans ...... 102

Figure 5.3: AlN fN2 AFM analysis ...... 103

Figure 5.4: AlN substrate temperature stacked coupled scans ...... 105

Figure 5.5: AlN overlaid substrate temperature optimization ...... 106

xii

Figure 5.6: AlN substrate temperature series AFM ...... 107

Figure 5.7: AlN stacked external coil current coupled scans...... 109

Figure 5.8: AlN overlaid coil current coupled scans ...... 110

Figure 5.9: AlN coil current series AFM measurements ...... 111

Figure 5.10: AlN power series stacked coupled scans ...... 113

Figure 5.11: AlN power series overlaid coupled scans...... 113

Figure 5.12: AlN power series AFM ...... 115

Figure 5.13: AlN thin film power series coupled scans ...... 117

Figure 5.14: Thin AlN power series surface morphology ...... 118

Figure 5.15: TiN on Al2O3 example ...... 120

Figure 5.16: TiN surface morphology ...... 121

Figure 5.17: AlN/TiN/Al2O3 power series stacked coupled scans ...... 122

Figure 5.18: AlN/TiN stacks power series surface morphology...... 123

Figure 5.19: AlN/TiN stacks fN2 coupled scans ...... 124

Figure 5.20: AlN/TiN stacks fN2 surface morphology ...... 125

Figure 5.21: AlN power series piezoelectric coefficients ...... 126

Figure 5.22: AlN fN2 piezoelectric coefficients ...... 126

Figure 6.1: TIO2 on Al2O3 coupled scans ...... 129

Figure 6.2: TiO2 optical properties ...... 130

Figure 7.1: Resistivity switiching of VO2 on Al2O3 ...... 132

Figure 7.2: Room temperature VO2 optical properties ...... 133

Figure 7.3: High temperature (80°C) VO2 optical properties ...... 134

Figure 7.4: VO2/TiO2/Al2O3 resistivity curve ...... 135

xiii

Figure 7.5: Comparison of VO2 films grown on Al2O3 and TiO2 ...... 136

Figure 7.6: Thermally annealed VO2 sheet resistance curves ...... 138

Figure 7.7: VO2 GIXRD scans...... 140

Figure 7.8: VO2 symmetric coupled scans ...... 141

Figure 7.9: XPS of V peak in VO2 films ...... 142

Figure 7.10: Optimized thermally annealed VO2 resistance switching ...... 144

Figure 7.11: VO2 optimum and near optimum resistivity switching ...... 145

Figure 7.12: Optimized thermally annealed VO2 coupled scan ...... 146

Figure 7.13: Optimized thermally annealed VO2 XPS spectrum ...... 147

Figure 7.14: VO2 25°C optical properties ...... 148

Figure 7.15: VO2 80°C optical properties ...... 149

xiv

LIST OF ABBREVIATIONS AND NOTATIONS

AFM Atomic Force Microscopy

Al2O3 Sapphire (Corundum)

BE Binding Energy dhkl Atomic planer spacing of (hkl) family of planes

FCC Face Centered Cubic fN2 Nitrogen gas fraction

FWHM Full Width at Half Maximum

GIXRD Grazing Incidence X-Ray Diffraction

ħ Reduced Plank’s Constant (1.0546 *10-34 J∙s)

HRXRD High Resolution X-Ray Diffractometry

IAD Ion Assisted Deposition

IC Magnetic coil current

IH Heater current

KE Kinetic Energy

MIT Metal-Insulator Transition e Charge on the Electron (1.602*10-19 C)

QCM Quartz crystal microbalance

RMS Root Mean Square

RSM Reciprocal Space Map

SE Spectroscopic Ellipsometry

SIMS Secondary Ion Mass Spectrometrys

TE Transverse Electric

TM Transverse Magnetic

TMNs Transition Metal Nitrides

TMO Transition Metal Oxides

UHV Ultra-High Vacuum

VASE Variable Angle Spectroscopic Ellipsometry

Wt Magnetron Powers

XPS X-Ray Photoelectron Spectroscopy

XRD X-Ray Diffractometry

xvi

CHAPTER 1

INTRODUCTION TO MATERIAL SYSTEMS

The materials examined in this work consist of a transition metal element bound to either nitrogen or oxygen. These materials find use in optical and electronic applications.

However, in order understand the experiments in this work, one must first understand the necessary properties of each material examined and the specific application examined for each material.

The use of a material in optoelectronic applications often requires one to preserve the band structure of said material. This means that one must grow materials with crystalline symmetries intact. As such, growth of some optoelectronic devices necessitates epitaxy, which is the process of growing a film that matches its crystalline lattice structure to the substrate (the material one is growing on). In other words, epitaxy is the process of using the substrate material as a pattern for the growth of a film material. Not only does the shape of the substrate material matter for film growth, but the size of the crystal also plays an important role. If the crystal structure of the substrate material does not match well to the film material, it is difficult to grow films without induced dislocations and defects.52 These dislocations and defects will often destroy the electronic properties of the material. As such, single-crystal, epitaxial materials are desirable for integration into electronic devices. So, one must not only examine the electrical and optical properties of the material, but also must examine the crystallinity of the material. In addition, it is necessary to understand what materials will lattice match well to the desired film materials prior to engaging in growth of said films.

Sometimes the desired application requires the use of a material property that is observed in the crystalline material, but not an amorphous film of similar stoichiometry.

One example would be birefringence. Quartz is chemically identical to silica, both are

SiO2, but only in crystalline quartz does SiO2 show any form of birefringence. In this case, it one might not require quite as low of defects as one would for an electronic device, but the crystalline structure is still important. Another example would be taking advantage of a crystalline, structural phase transition that a material undergoes. In this case, the crystal does not need to be a perfect single crystal, but a high degree of crystallinity and a single phase are required in order to have the material function as intended.

For these two applications (optic and electronic) different materials, techniques, characterization, and requirements exist. As such, one must examine the material of interest and decide what properties are most important for the application. Then one must examine methods to test the desired properties in the experimental materials.

However, this process cannot happen without first understanding what potentials exist in the materials of interest.

1.1 Transition Metal Nitrides

Transition metal nitrides (TMNs) are materials that are made up of a transition metal bonded to nitrogen. These materials are often refractory, meaning they possess high melting points and corrosion resistance.15 Uses for these materials include superconductors,55 plasmonics,12, 15, 40 as well as wear and corrosive resistant coatings.59

In addition, many have suggested that it is possible to change the optical and electronic properties of TMN materials through the use of doping and changes in deposition

2 conditions.68 This provides the potential to tune the properties of TMN materials for a desired application.

1.1.1 Titanium Nitride

Titanium Nitride (TiN), a metallic TMN that inherits the refractory properties of the

TMN family of materials.15 As such, TiN finds uses in the manufacturing industry as a wear and corrosive resistant coating.59 However, interest in TiN for plasmonic applications as a replacement for gold has surged in recent years due to similar properties between the two.40, 68, 76 The use of TiN offers advantage over gold in the form of a harder and more chemically robust film. Additionally, the higher melting point of TiN allows for the use of higher powered lasers in plasmonic systems.40

TiN is a metallic TMN with a rock-salt lattice structure with a lattice constant of around 4.23Å.68 The rock-salt lattice structure is a face centered cubic (FCC) crystal structure with a two atom basis, as show in Figure 1.1 below.

Figure 1.1: Rock-salt crystal structure.

When examining Figure 1.1, it is possible to see the FCC crystal in the gray atoms.

The addition of the blue atoms can either be considered changing the FCC crystal to

3 having a two atom basis, or as intermixing an additional FCC crystal structure at a lateral offset. The lattice constant for TiN is the length of one of the sides of the FCC crystal.

Previous examination of TiN optimized the growth conditions in order to obtain high quality films. As TiN is a conducting TMN material, it is utilized as a conductive layer for the piezoelectric testing of AlN films.

1.1.2 Scandium Nitride

ScN is a TMN rocksalt semiconductor that was previously thought to be a semimetal, which is a material that has a slight band overlap.24 This state can also occur typically through excessive doping of a semiconductor such that the material acts more like a metal than a semiconductor. If in this state, the material is called a degenerate semiconductor.

ScN was thought to naturally occur at this state. This caused debate in the literature as to whether or not ScN truly had a direct band gap of 2.2-2.7 eV.16, 60 However, it has since been shown that this material is, in fact, a semiconductor that is simply doped excessively due to oxygen incorporation in the films.16

Recently, interest in ScN has increased for applications in optoelectronic devices16, 33,

47 as well as thermoelectrics.14, 35, 77 ScN itself is a material with an indirect band gap of approximately 0.9 eV and a direct band gap of 2.2-2.7 eV.14, 16, 24, 61 When grown, ScN films often have carrier concentrations in the range of 1018-1022 cm-3 with mobilites on the order of 30 to 180 cm2/V·s.14, 16, 35

In addition to interest in thermoelectrics, ScN is an interesting material for GaN based electronics. ScN has a lattice constant of ~4.503Å.16, 42 After rotating the ScN film to the

(111) orientation, this lattice constant becomes an in plane lattice constant of 3.184Å, which results in about 0.1% mismatch with the GaN in-plane lattice constant of around

3.189Å.16, 33, 48 As such, it is expected that GaN should grow significantly well on ScN films. This can be seen by a reduction in GaN dislocation densities for films of GaN grown on ScN.47, 49 Some even suggest using ScN as an interlayer material to incorporate

GaN films on Si wafers.42, 48, 62 This provides the potential for cheaper GaN growths and on chip incorporation, which decreases processing costs.

Due to the interest in this material stemming from either growth of GaN or as an optical material, there are a few important parameters. First, the crystallinity of the material is paramount, or else any regrown material would also have poor crystallinity.

Second, the understanding the optical properties of ScN (i.e. the dielectric constant) is necessary as many GaN based systems are for LED applications. Third, the electronic properties (i.e. mobility and carrier concentration) are also important as this material will be a part of electronic devices. Last, the impurity incorporation in ScN is discussed due to the dependence of the electrical properties on the incorporation of oxygen in the film.

1.1.3 Aluminum Nitride

Along with GaN, AlN is an extremely well known material. It is a wide bandgap semiconductor with a bandgap of 6.2eV.71 Having a wurtzite crystal structure, it easily alloys with GaN for bandgap engineering.41, 71 The wurtzite crystal structure differs significantly from the rocksalt structure of TiN and ScN as it is hexagonally symmetric rather than cubic. An image of the wurtzite structure can be seen in Figure 1.2 below.23

Figure 1.2: Wurtzite crystal structure

In Figure 1.2 above, one can clearly see the hexagonal nature of the wurtzite structure. In the case of AlN, the gray atoms represent Al and the yellow atoms represent the N.23 For pure AlN, the lattice constant is 3.1130Å in plane.41

In addition to having a wide bandgap, AlN is a piezoelectric material, or a material that strains when subjected to an electric field, with an extremely high maximum use temperature.7, 23, 31, 73 However, the piezoelectric coefficient, which dictates how much strain is observed when a bias is applied, is somewhat lackluster when compared to other crystalline materials such as lead zironate titanate (PZT) or lead zirconate niobate – lead titanate (PZN-PT).7 But, both PZT and PZN-PT have significantly lower maximum use temperatures, which can be seen below in Figure 1.3.7

Figure 1.3: Comparison of piezoelectric material maximum use temperatures Reprinted from Enhancement of Piezoelectric Response in Scandium Aluminum Nitride Alloy Thin Films Prepared by Dual Reactive Cosputtering, Akiyama, M. et al., Adv Mater 21, 593-596 (2009) with permission from John Wiley and Sons

In Figure 1.3 above, to save space and keep the graph readable AlN is labeled as

“AN” and PZN-PT is labeled “PZNT”. This figure, reproduced from Akiyama, shows the trend of materials with a higher maximum use temperature having a significantly lower piezoelectric coefficient.7 Note that the y-axis is in log scale.

For high temperature applications, incorporating Sc metal into AlN seems to pose a solution to the low piezoelectric coefficient of AlN.7, 78, 79 Some have shown an almost five times increase in the piezoelectric coefficient in Al1-xScxN films, where x reaches a maximum of 43%.7, 79 With the potential to not reduce the maximum use temperature of the films, this increase substantially benefits devices that make use of piezoelectric materials for high temperature applications.

7 a) b)

Figure 1.4: Al1-xScxN piezoelectric coefficients with varying Sc content a) Reprinted from Enhancement of Piezoelectric Response in Scandium Aluminum Nitride Alloy Thin Films Prepared by Dual Reactive Cosputtering, Akiyama, M. et al., Adv Mater 21, 593-596 (2009) with permission from John Wiley and Sons b) Reprinted from Effect of scandium content on structure and piezoelectric properties of AlScN films deposited by reactive pulse magnetrons sputtering, Zywitzki, O. et al., Surface and Coatings Technology 309, 417-422 (2017) with permission from Elsevier.

In Figure 1.4 above, one can clearly see that alloying AlN with Sc increases the piezoelectric coefficient of the AlN. The rapid drop in the piezoelectric coefficient after

43% Sc is attributed to the Al1-xScxN taking the ScN rocksalt crystal structure rather than the AlN wurtzite structure.

In this work, the first step of a series studying Al1-xScxN from x = 0 to x = 0.50 was conducted by examining AlN growth on Al2O3. The piezoelectric coefficient of AlN is applied along the c-axis.31 As such, growing (0001) oriented AlN is extremely important to obtaining the highest possible piezoelectric coefficient. Therefore, the crystallinity of the film represents one of the major optimization parameters, in addition to the actual piezoelectric coefficient obtained in the films.

1.2 Transition Metal Oxides

In addition to the TMN materials, this work also examines two transition metal oxide

(TMO) materials: vanadium oxide and titanium oxide. The work conducted on TiO2 is as a potential interlayer for multilayered growth of VO2. As such, the major applications for the TiO2 examined in this work are simply as a growth template. The VO2, however, finds significant use in optoelectronics as a switchable material, discussion of this phenomenon to follow.

1.2.1 Vanadium Oxide

In recent years vanadium dioxide (VO2) has garnered significant interest from the scientific community for its metal-insulator transition (MIT) that occurs at ~68°C.13, 25, 30,

75 After increasing the temperature of monoclinic VO2, it undergoes a structural phase transition to a tetragonal rutile structure.13, 25, 26, 39 Accompanied with this structural phase transition, VO2 undergoes a change in resistivity of multiple orders of magnitude,

19, 30, 37 with some of the best showing nearly four orders. This changes the VO2 from an insulating material, to a metallic material.

In addition to the change in resistivity, there is a change in the optical properties of

19, 32, 56 the VO2 film due to the increase in free carriers. As such, VO2 is an interesting material for smart windows,32, 56 optical switches,17 and even other switchable complex optical elements such a switchable wire grid polarizers.

As mentioned prior, VO2 undergoes a structural phase transition from the monoclinic phase to a tetragonal, rutile, phase at approximately 68°C. In order to effectively grow this material, one must understand both of these structures, as a crystalline substrate must

9 be chosen that matches well to both phases. These two phases of VO2 can be seen in

Figure 1.5 and Figure 1.6 below, which are both obtained from Goodenough.25

Figure 1.5: VO2 high temperature rutile phase Reprinted from The two components of the crystallographic transition in VO2, Goodenough, J. B., Journal of Solid State Chemistry 3, 490-500 (1971) with permission from Elsevier.

In the high temperature, rutile, phase of VO2, shown in Figure 1.5, the black spheres represent vanadium atoms, while the white spheres represent oxygen atoms. The spacing between vanadium atoms along the c-axis is around cr = 2.85Å with an in plane spacing

25, 45 of around ar = 4.55Å. However, after cooling down, the crystal distorts along the c- axis, which changes the interatomic spacing of the vanadium atoms along this direction, which can be seen in Figure 1.6 below.25

Figure 1.6: VO2 room temperature monoclinic phase Reprinted from The two components of the crystallographic transition in VO2, Goodenough, J. B., Journal of Solid State Chemistry 3, 490-500 (1971) with permission from Elsevier.

In this image, only the vanadium atoms are shown in order to maintain clarity. The rectangular prisms represent the original tetragonal crystal prior to the distortion along the c-axis as a reference.

The work on VO2 revolves mostly around electro-optic applications. As such, the change in resistivity as a function of temperature is extremely important. The magnitude of the resistivity change as well as the changes in the dielectric constant also represent a parameter to be examined. Last, the crystallinity and oxidation states of the film are examined to understand the different growth methods utilized to obtain VO2 films.

1.2.2 Titanium Oxide

The difficulty in growing VO2 films is that it requires growth on a substrate that matches well to the high temperature rutile phase. In addition, the thermal coefficient of expansion difference between the film and the substrate cannot be too great, or else the stresses could result in delamination of the VO2 film. As such, VO2 only grows

19 successfully on a few substrate materials. These include Al2O3, ZrO, and TiO2.

However, most people grow VO2 on single crystal Al2O3 as this produces high quality films, as discussed above.

When examining a material for an optical device, it is often necessary to ensure that the film can be made into a multilayered stack of materials, as this is how filters and DBR stacks are created. This presents a problem for incorporation of VO2 because crystalline

Al2O3 cannot be deposited as a thin film easily. As such, an additional material must be utilized to make multilayered structures of VO2.

36 TiO2 is a semiconducting material with low losses in the visible and near IR ranges.

Additionally, the high temperature (rutile) phase of VO2 is named after one of the phases of TiO2, which exists either as rutile, anatase, or brookite. The most common of the three are the anatase and rutile structures. These two phases of TiO2 can be seen in Figure 1.7 below.21

Figure 1.7: TiO2 rutile and anatase structures Reprinted from The surface science of titanium dioxide, Diebold, U., Surface Science Reports 48, 53-229 (2003) with permission from Elsevier.

Anatase and Rutile TiO2 have drastically different lattice constants and structures. In the rutile phase (which is tetragonal) the lattice constants are a = b = 4.584Å and c =

2.953Å. The anatase phase of TiO2 has lattice constants of a = b = 3.782Å and c =

9.502Å.21

When comparing the in-plane lattice mismatch between the TiO2 rutile phase and the

VO2 rutile phase (aVO2 = 4.55Å and aTiO2 = 4.584Å) it is possible to calculate the mismatch between them. This mismatch amounts to about 0.75%, with the rutile VO2

13 under tensile stress. This small lattice mismatch between the two of them suggests that the VO2 should grow well on rutile TiO2.

51 In fact, some have shown it is possible to grow VO2 on rutile TiO2, and some have

43 shown evidence that it may be possible to grow rutile TiO2 on silicon and SiO2. This makes TiO2 a potential material for incorporation of VO2 on silicon electronics. As such,

TiO2 is a promising material for creating multi-layered stacks of VO2, which would allow for significantly more versatile device fabrication.

The end goal for the TiO2 films in this study is as an interlayer for growth of VO2 films in optical devices. As such, three major properties of the TiO2 films must be obtained. First, the crystallinity of the TiO2 must be examined. In order to growth the

VO2 films, the TiO2 must first act as a template. As such, the TiO2 must be the right phase and orientation to grow the VO2 on top of. The optical properties of the TiO2 films must be examined, as they may vary significantly in the film as compared to the bulk parameters. Last, VO2 actually grown on the TiO2 should result in a MIT at around 68°C with a similar order of magnitude switch as if it was VO2 on Al2O3. If this is the case, then TiO2 thin films present the possibility to facilitate multilayered growth of VO2.

CHAPTER 2

GROWTH SYSTEMS

For fundamental materials research, i.e. examination of material properties for a particular application, one must maintain a high material purity in order to establish a baseline for a particular material. After establishing the fundamental properties of a particular material (such as but not limited to: transport properties, bandgap, thermal conductivity, and resistivity) it is possible to then examine the effect of material doping and alloys. As such, synthesis of materials for this thesis takes place in vacuum systems designed to maintain high material purity. Use of solid and gas source materials of

99.99% or greater allows for growth of crystalline materials with minimal impurity incorporation However, in some cases even the high purity target, gas sources, and ultra- high vacuum (UHV) chambers are not enough to obtain perfectly pure materials. For some of the TMNs, this fact is readily seen by the incorporation of oxygen in the grown films. However, the details of this problem follow. For now, it suffices to note that the materials are synthesized in vacuum chambers of different growth methods in order to obtain high purity materials.

2.1 DC Sputtering System

2.1.1 Sputter Deposition

Sputtering is a deposition tool used often for research and industrial growth of thin film coatings.20, 53, 54, 66, 70 This growth method utilizes atomic collisions to forcibly remove material from a source, often called a target. These removed atomic species then deposit on a material surface, known as the substrate.20, 53, 63, 70 In order to do this, the

15 sputter target is placed in the vacuum chamber and the system is then pumped to high vacuum. Depending on the system and the materials of interest, the base pressure can vary from around 10-6 Torr to as low as 10-12 Torr. Naturally, the lower the base pressure, the less impurities are incorporated in the film due to the background gas species. For the unbalanced magnetron sputtering system used in this work, the base pressure is between 1×10-8 Torr and 5×10-9 Torr whereas the Denton system has a base pressure of around 1µTorr.

After the target is placed in the vacuum chamber, and the substrate is loaded for material growth, the chamber is flooded with a non-reactive gas to pressures in the 20-

150 mTorr range.66 Typical gas used in sputtering systems is argon, as it is fairly massive but also prevalent and inexpensive. However, any noble gas could fill this role, but heavier atoms tend to provide higher sputter yields.

The Ar atoms that flood the vacuum chamber are then ionized by the sputter gun.

This creates a bias on the target which results in the Ar atoms accelerating towards the sputter target (cathode). The collision of the Ar atoms with the sputter target physically ejects atoms from the target surface. This collision between the gas atoms and the target atoms is the fundamental process by which sputtering takes place. A simple image of this process can be seen in Figure 2.1 below.70

Figure 2.1: Sputter deposition Reprinted from Materials Processing, Stadler, B. J. H. Chapter 7 - Vapor Processes. , 513-588, (2016) with permission from Elsevier.

2.1.2 Reactive Sputtering

Sometimes, one desires to deposit a material that is more than one element. In the case of some materials, it is possible to sputter using a target comprised of the desired material or alloy composition. For example, on can use a target comprised of Ti and O to obtain TiO2. However, purchasing a target of this alloy composition is not the only way to TiO2. It is also possible to flow a reactive gas into the system in addition to the Ar that is used for sputtering. This process is called reactive sputtering, which is the inclusion of a reactive species of gas in addition to the sputtering gas in order to deposit a material comprised of more than one element. The materials in this work all are deposited by some form of reactive deposition.

2.1.3 Magnetron Sputtering

One of the downsides of conventional sputtering is the low deposition rate. In order to remedy this low sputter rate, a magnetron system can be applied.66 A magnetron is a set of magnets used to create a magnetic field around the target. These magnets are arranged such that one set of poles, such as south, are placed in a ring around the edge of the target. The other pole, in this case the north pole, is placed in the center of the target.

This creates a magnetic field that wraps from middle to the edge of the target.20, 54, 70

Figure 2.2 shows the field lines created by these magnets in a cross section.

Figure 2.2: Magnetron sputtering field lines

The addition of these fields helps to increase the sputter rate of the target material, shown in blue in Figure 2.2, by trapping electrons in the magnetic field.20, 54, 70 The trajectory of a trapped electron in this field can be seen in Figure 2.3 below.20

Figure 2.3: Simulated electron trajectory in magnetron field Reprinted from Handbook of Deposition Technologies for Films and Coatings (Third Edition), Depla, D., Mahieu, S. & Greene, J. E., Chapter 5 – Sputter Deposition, 253-296 (William Andrew Publishing, Boston, 2010) with permission from Elsevier.

These captured electrons have the possibility to interact with neutral gas species to produce an ion.53, 70 This newly created ion can then be used in the sputtering process.

By introducing additional ions, one observes an increase in the deposition rate which allows for sputtering processes at lower pressures than conventional DC sputtering.54

This is because conventional sputtering requires higher pressures in order to obtain high enough sputter rates. The resulting plasma takes the form of a ring in the region of trapped electrons, as can be seen in Figure 2.4 below.20

Figure 2.4: Magnetron plasma at target Reprinted from Handbook of Deposition Technologies for Films and Coatings (Third Edition), Depla, D., Mahieu, S. & Greene, J. E., Chapter 5 – Sputter Deposition, 253-296 (William Andrew Publishing, Boston, 2010) with permission from Elsevier.

The increased deposition in the area enclosed by the magnetic field lines results in target erosion as the increased deposition rate is no longer uniform across the target surface. This area of erosion is known as the racetrack, which can be seen in Figure 2.5.

Figure 2.5: Magnetron sputter target

As mentioned prior, sometimes one desires to create films which consist of multiple atoms, such as TiN or ScN. It is possible to sputter multiple materials at the same time through the use of an alloyed target. However, it is often difficult to maintain stoichiometry while sputtering from an alloyed target. Additionally, some materials

(such as TiO2) are difficult to sputter. As such, it is often easier to reactively sputter these types of materials.20, 63, 70 This allows one to sputter a material that consists of both the reactive gas and the target material, which can improve stoichiometry of the sputtered films and allow one to sputter materials that would otherwise be extremely difficult.

The materials examined in this work utilized this concept in order to make binary materials. The titanium, scandium, aluminum, or vanadium target is placed in the chamber and then reactive gas is included in the argon process gas. This allows the materials to react with the process gas and form the binary compounds during deposition.

2.1.4 Sputtering System

The conventional magnetron sputter system utilized in this work is outfitted with three confocal sputter guns, allowing for deposition of up to three different materials simultaneously. In order to maintain uniformity of the film composition and thickness, the substrate holder is outfitted to rotate at a controllable rate. This chamber is a commercially available vacuum system: the Denton Vacuum Explorer 14. A picture of this system can be seen in Figure 2.6 below.

Figure 2.6: Denton Vacuum Explorer 14 sputter system

The system is pumped to a base pressure of around 1µTorr. Without a load lock system, the chamber itself must be evacuated with each loading of a sample. Due to the use of this system being mostly oxide films, or films where oxygen contamination is not a significant drawback, this base pressure is certainly adequate.

2.2 Controllably-Unbalanced DC Reactive Magnetron Sputtering

The second system utilized in this work is the controllably-unbalanced reactive magnetron sputtering system, which supplied most of the samples examined. This system allows for the growth of the TMN samples. As such, it is important to understand each of the deposition parameters used for optimization of the film growth, the growth method itself, and the limitations of this growth system.

2.2.1 Unbalanced Magnetron Sputtering

When performing magnetron sputtering, it is sometimes desirable to decrease the plasma confinement at the target surface. Doing so enables the once completely confined electrons to travel along the magnetic field towards the substrate.20, 54, 58 This is referred to as unbalanced magnetron sputtering. The difference in magnetic field lines between balanced and unbalanced magnetron sputtering can be seen in Figure 2.7 below.54

Figure 2.7: Balanced and unbalanced field lines Reprinted from Materials Science of Thin Films (Second Edition), Ohring M., Chapter 5 – Plasma and Ion Beam Processing of Thin Films, 203-275 (Academic Press, San Diego, 2002) with permission from Elsevier.

The balanced magnetron field lines are labeled as the intermediate in Figure 2.7. The unbalanced systems used in this work take the type II configuration of Figure 2.7 above.

This allows the magnetic field to extend towards the substrate. Figure 2.8 below shows the substrate included in the magnetic field lines.70 23

Figure 2.8: Unbalanced magnetron field lines Reprinted from Materials Processing, Stadler, B. J. H. Chapter 7 - Vapor Processes. , 513-588, (2016) with permission from Elsevier.

A simulation of the electron trajectories in a type II unbalanced configuration can be seen in Figure 2.9 below.20

Figure 2.9: Simulated electron trajectories in an unbalanced system Reprinted from Handbook of Deposition Technologies for Films and Coatings (Third Edition), Depla, D., Mahieu, S. & Greene, J. E., Chapter 5 – Sputter Deposition, 253-296 (William Andrew Publishing, Boston, 2010) with permission from Elsevier.

The resulting plasma from an unbalanced system can be seen in Figure 2.10 below.20

Figure 2.10: Plasma in an unbalanced system Reprinted from Handbook of Deposition Technologies for Films and Coatings (Third Edition), Depla, D., Mahieu, S. & Greene, J. E., Chapter 5 – Sputter Deposition, 253-296 (William Andrew Publishing, Boston, 2010) with permission from Elsevier.

One direct impact to changing the system from a balanced magnetron system to an unbalanced system is noticed by examining both the incident ion energy Ei and the ion to

57, 58 neutral flux ratio, Ji/JMe. The product of these two parameters, Ei and Ji/JMe, is the average energy deposited per atom (Ed), which is a parameter that is extremely useful in understanding film growth mechanics.57

Sputter deposition often takes place in the kinetic growth regime, resulting in high incident energy atoms which can damage the substrate and film during growth. Driving the growth into a thermally limited growth regime allows for high-flux, low-energy deposition, which mitigates these drawbacks. One achieves this type of growth by adjusting the ion energy or the ion to neutral flux. Unbalancing the system allows for adjustment of the ion to neutral flux.

Typically, unbalanced systems are created by changing the relative strength of the center magnet as compared to the outer magnets.70 As such, they are fixed in a particular

25 configuration and one cannot change from balanced to unbalanced without adjusting the magnetron, which requires venting the system. Therefore, it is beneficial to have a system that allows for controllably unbalancing the magnetron through a process other than the adjusting of the internal magnets in the magnetron. The use of external

Helmholtz coils, concentric with the magnetron systems, has been shown to unbalance a system, controllably, through the induced magnetic field by an applied current.58 This allows for controllably adjusting the ion to neutral flux ratio in order to obtain high quality growth.

2.2.2 Controllably-Unbalanced Sputtering System

The controllably-unbalanced DC reactive magnetron sputtering system utilized in this work takes advantage of an external, axially applied, magnetic field, to unbalance the magnetrons. In order to better understand the physical system, a cross-sectional view can be seen in Figure 2.11, below.

Figure 2.11: Controllably-unbalanced reactive magnetron sputtering system

In Figure 2.11 it is possible to see that the magnetron and the external magnetic coil are both concentric. A duplicate set, not pictured due to the cross-section, is placed on the other side of the system to create a uniform field. As such, there are only two magnetrons in the system, both applied on axis to the sample. This prevents cosputtering, but is absolutely necessary for the use of the external magnetic field. The chamber is equipped with a load lock system that allows the main chamber to maintain

UHV pressures while samples are loaded. This also helps to mitigate oxygen contamination in the grown chamber.

The sample is resistively heated through current applied to a Ta foil. This value is converted to a temperature at the surface of the substrate through a calibration obtained through a thermocouple measurement that was performed without plasma induced heating. This calibration curve can be seen in Figure 2.12.

T Avg 1000 y = 18.364x + 19.667 900 R² = 0.9926 800

C) 700 ° 600 500 T Avg 400

300 Linear (T Avg) Temperature( 200 100 0 0 10 20 30 40 50 60 Applied Current (A)

Figure 2.12: Substrate heating calibration

The calibration curve in Figure 2.12 was obtained using a thermocouple clamped onto an Al2O3 substrate. This allowed recording of the temperature at the substrate surface, where film growth happens, due to only the applied current through the Ta foil element.

However, it is important to note that the thermocouple is not present during actual growth in the system and all temperatures reported from this system are in reference to this calibration.

Deviations from this calibration can arise due to differences in 1) manufacture of the

Ta foil, 2) contact of the Inconel holder to the heating forks, 3) differences in contact with the substrate to the Ta foil, and 4) changes in resistance due to deposition on the Ta foil through continued use. These differences, however, are not expected to exceed 50°C.

Additionally, any temperature reported for films grown in this system do not include the addition of plasma heating. Again, this is change is also not expected to exceed 50°C and should be relatively consistent between growths of the same material, as it is not expected that the plasma changes significantly between growths similar materials at similar deposition powers.

The external Helmholtz coils apply an additional magnetic field induced by an applied current. This current is typically applied in the range of 4-8A. The coils are water cooled to prevent damage due to resistive heating of the copper wires inside the coils. The magnetic field inside the chamber was simulated in COMSOL for both the balanced and unbalanced configurations. These simulations can be seen in Figure 2.13 below.

28 a) No applied field b) Externally applied field

Figure 2.13: COMSOL 퐵⃗ Simulation

As mentioned prior, areas with high magnetic field confine the plasma. In Figure

2.13 above, which is a simulation of a cross-sectional view of the system, it is possible to see that without the additional magnetic field due to the external coils, the areas of high magnetic field (and thereby plasma confinement) are mostly around the targets in the typical ring shape for magnetron sputtering. However, application of the external magnetic field dramatically increases the field intensity at the substrate, which is placed in the center between the two coaxial magnetrons. This corresponds to an increase in

58 plasma confinement at the substrate, which corresponds to an increase in Ji/JMe.

However, this does not change the ion energy, Ei, but instead changes the average energy,

Ed.

The imaged plasma for both the balanced and unbalanced state of the system can be seen in Figure 2.14 below.

a) Balanced magnetron system b) Unbalanced magnetron system

(no additional field) (additional field applied)

Figure 2.14: Plasma comparison of balanced and unbalanced configuration

The TMN materials grown in this work are prepared in this controllably-unbalanced reactive magnetron sputtering system. As such, it is important to understand the parameters that can be controlled in the chamber. The growth pressure is maintained at

20 mTorr for all depositions in this system. However, the pressure ratio of N2 to Ar (fN2) is not fixed and is a parameter used to obtain the desired phase of the TMN material. The substrate temperature is controlled through the use of applied current through the Ta foil resistive heating element, IH. The strength of the axially applied magnetic field is also controlled by adjustments in the applied current, IC. Last, the magnetron power, Wt, is also controlled. These parameters are used to obtain the correct phase of the TMN material as well as to induce highly crystalline and smooth film growth.

2.3 Ion Assisted Evaporation System

In some cases, it is often difficult to sputter some types of materials. In this work, both of the oxides (VO2 and TiO2) are difficult to sputter without reactive methods. As such, an addition growth mechanism are necessary. 30

2.3.1 Thermal Evaporation

It is fairly well known that when materials increase in temperature, the vapor pressure also increases. This is the pressure of atoms expelled from a material at a given temperature. In thermal evaporation, this principle is exploited to evaporate the source material, which is the material you wish to grow, in order to create a thin film. The substrate is placed in the path of the evaporated atoms which then coalesce on the substrate, thereby growing the film.54, 63, 66, 70

Traditional evaporation processes typically utilize a solid source of material inside a holder, known as a crucible. This crucible is then heated, thereby heating the material it is holding. This heating can be through resistive methods or otherwise. An example of this can be seen in Figure 2.15.70

Figure 2.15: Thermal evaporation Reprinted from Materials Processing, Stadler, B. J. H. Chapter 7 - Vapor Processes. , 513-588, (2016) with permission from Elsevier.

The downside to heating the material through the crucible in this way is that the hottest part of this system is the outer edge of the crucible and the edges of the material of interest. As such, a lot of the energy is wasted through heating the crucible and the 31 underside of the evaporation material. As such, it is desirable to instead heat the surface of the material of interest, while maintaining a cool crucible. This can be done through the use of an electron beam.63, 66 The electron beam is used to locally heat the surface of the evaporation material, thereby only heating a small portion of the source material.

This process, known as electron beam evaporation, produces an area of evaporated atoms in a cosine squared distribution as can be seen in Figure 2.16 below.66

Figure 2.16: Electrom beam evaporation Reprinted from Handbook of Deposition Technologies for Films and Coatings (Third Edition), Shah, S. I., Jaffari, G. H., Yassitepe, E. & Ali, B. Chapter 4 – Evaporation Processes, 135-252 (William Andrew Publishing, Boston, 2010), with permission from Elsevier.

Unlike sputtering, this process does not require the use of additional gasses. It is possible to perform evaporation in pure vacuum with no process gasses, as the only thing required is a source material under heat. However, it is sometimes desired to preform evaporation processes in a gaseous environment, despite the lack of necessity. One reason for this is that thickness non-uniformities due to the cosine squared distribution of evaporated atoms can be mitigated by having the evaporated species undergo a large

32 number of collisions prior to reaching the substrate. This usually requires pressures in the range of 5-200 mTorr.66 Additionally, one may desire to grow films of more than one atomic element. Just like sputtering, it is possible to deposit materials in a reactive gaseous environment as well, which results in reactive evaporation. Materials such as

TiO2, which are typically difficult to sputter, can be deposited with ease using a Ti metal source and reactive oxygen gas in the evaporation chamber.

2.3.2 Ion Assisted Evaporation

One potential drawback to evaporation systems is that the films typically grow in a less dense manner than other methods.63 This can result in properties that differ significantly from the bulk material properties. Additionally, incorporation of the reactive species is often difficult. As such, methods exist in order to make evaporated materials significantly denser and improve stoichiometry. This is done through the use of an ion assist system resulting in ion assisted deposition (IAD).44

In an ion assisted evaporation system, a stream of high-energy ions are focused on the substrate surface. In most cases, a reactive gas (such as oxygen) are used as the ion source. These ions impact the film, which effectively forces the atoms together, thereby densifying the film. In addition to densifying the film, these reactive ions also help to improve stoichiometry in the films, as any unbound film atoms will react with the ion beam. In Figure 2.17 below, one may see the additional ion beam introduced into the evaporation system from the ion gun on the left side of the chamber.44

Figure 2.17: Ion assisted evaporation Reprinted from Handbook of Physical Vapor Deposition (PVD) Processing (Second Edition), Mattox, D. M., Chapter 9 – Ion Plating and Ion Beam-Assisted Deposition, 301-331 (William Andrew Publishing, Boston, 2010), with permission from Elsevier.

2.3.3 IAD System

The ion assisted electron beam evaporation system was used in this work in order to grow the TMO materials. The particular system used in this work is evacuated to less than 1µTorr prior to deposition. Due to the chamber’s relatively small size, it is simple to load samples directly into the chamber without a load lock mechanism. In addition, the materials grown with this system often include oxygen, so using a highly reactive material as a “getter” is not necessary. The physical chamber can be seen in Figure 2.18 below.

Figure 2.18: Ion assisted evaporation system

This system allows for growth of samples on substrates of 2in. diameter. These substrates are heated using a quartz lamp during deposition, with the temperature measured by a thermocouple. In addition, a quartz crystal microbalance (QCM) is used to record film thickness during deposition. A QCM is, at its most fundamental nature, is a harmonic oscillator made from quartz. The frequency of this oscillator is well understood and deposition of an additional material on the crystal essentially adds a weight to the oscillator. By recording the frequency of the oscillator, it is possible to back out the thickness of the additional material. This results in an in situ measurement of film thickness.

In this system it is also desired to get materials of the correct phase, crystallinity, and stoichiometry. The parameters that can be adjusted in this system to achieve the desired growth are 1) deposition rate, controlled through the electron beam, 2) IAD ion energy 35 and, 3) the IAD ion flux. The background pressure during the deposition is not actively controlled during the deposition. These are the parameters that can be explored to obtain crystalline growth of TMO materials grown with the IAD system.

CHAPTER 3

CHARACTERIZATION TECHNIQUES

After creating samples of different materials, it is important to characterize the film properties. This either gives the basic properties of the material, if one is conducting fundamental research, or could show how film properties may be optimized if the material is of interest for a particular application. The characterization techniques performed on the films of interest for this discussion are outlined in this section to not only examine the technique advantages, the information obtained from each technique, but also to outline how each technique’s fundamental properties are utilized to obtain relevant information about the sample of interest. As such, each technique will be discussed in some depth.

3.1 X-Ray Diffractometry

X-Ray Diffractometry (XRD) enables one to probe the inner atomic structure of a material system and is, by far, the most used technique during this work due to its extreme versatility, relative ease of operation, non-destructive nature, and the speed at which data can be taken and processed. As such, nearly every film created in this work undergoes examination using XRD in at least one type of scan.

Probing the atomic structure of a crystalline material allows one to obtain crystalline structure, atomic spacing,38 layer strain,22 dislocation densities,28, 29 and overall crystalline quality of the material.18 Each material examined in this work requires some sort of crystalline structure (polycrystalline or single crystal) to obtain the desired properties.

The nitride material growth focuses on epitaxial growth, or the growth of crystalline films

37 that have the same orientation as the substrate material. Therefore, XRD provides a way to ensure that the grown materials not only mimic the substrate in orientation, but also do not include any different orientations or phases. This allows for optimization of deposition parameters with respect to crystal structure. As mentioned previously, the electronic an optical properties of the deposited materials often rely heavily on the crystal structure. As such, this examination of the crystal structure also allows for some optimization in the electronic and photonic properties of the materials of interest.

For the duration of this work, a PANalytical X’Pert pro materials research diffractometer (MRD) system provides the XRD scans. The specific optics used for each type of scan are discussed later.

3.1.1 Introduction to XRD

Prior to explaining each type of XRD scan, it is important to have a fundamental understanding in the basics of XRD. This allows one to understand physically how each scan differs in addition to the different pieces of information that can be obtained from each scan variant.

At a fundamental level, XRD is simply diffraction of an x-ray beam off of a three dimensional diffraction grating (the crystal). The process by which this occurs is most simply understood when examining a set of atomic planes under incident plane wave illumination. An example of such can be seen in Figure 3.1 below.

Figure 3.1: Bragg diffraction

In Figure 3.1, the atomic spacing is given by a distance dhkl. The hkl subscript corresponds to the Miller indices for the particular set of atomic planes. The atomic spacing and the incident angle, θ, dictate if and where diffraction will occur. In most cases, however, the incident angle may not be known (in the case of powder samples).

Instead, it is significantly easier to measure the diffracted angle, 2θ.

As diffraction in this case is best understood by interference, the additional distance traveled by the waves hitting the second plane of atoms must equal an integer multiple of wavelengths for constructive interference to occur in the diffracted beam. If this is not an integer multiple of wavelengths, then destructive interference will occur and no diffracted beam will be observed.

It is possible to show that the additional length traveled by the ray diffracting off the second plan of atoms is equivalent to:

푙 = 2푑ℎ푘푙푠푖푛(휃)

If one includes the fact that for constructive interference this distance must be equal to an integer number of wavelengths, then one immediately obtains Bragg’s Law:

푛휆 = 2푑ℎ푘푙푠푖푛(휃)

Where λ is the wavelength of the x-ray source, dhkl is the interatomic spacing, and θ is half of the diffracted angle. In most cases, Cu Kα radiation is used as the x-ray source.

The wavelength of this radiation is around 1.542Å.38

The interatomic spacing is dependent on the crystal structure of the material undergoing examination. Due to the diffracted angle being intricately tied to the interatomic spacing, it is possible to examine the observed diffraction peaks in a sample and compare it to a known set of diffraction peaks. This allows one to determine an unknown sample’s crystalline structure and phase.69 This process is often done with powder samples. If some information is known about the sample, such as what chemicals may be present, it is then possible to obtain chemical makeup. However, it is best if

XRD is use in tandem with other systems that can obtain chemical makeup of the material. For example, it is possible to obtain chemical makeup of TiO2 using X-ray

Photoelectron Spectroscopy (XPS). However, TiO2 exists in multiple crystalline phases

(Rutile and Anatase). The difference between these phases can be determined through

XRD techniques, as each phase will have characteristic diffraction peaks.

The process of obtaining a powder diffraction pattern entails taking a crystalline sample and crushing it into a fine powder or obtaining a powder through some chemical process. Then it is possible to mount the sample in the XRD system. A general diagram of an XRD system can be seen in Figure 3.2 below:

Figure 3.2: Powder XRD system diagram

The basic diagram of Figure 3.2 was modified from Speakman’s powder diffraction presentation to shows the four major angles in an XRD system.69 The two most important are the angle between the sample and the X-Ray source (ω) and the diffraction angle (2θ). The other two angles (χ and φ) are used mostly for single crystal work.

In the case of powder diffraction, χ and ω will be used to flatten the sample, ω will be fixed, and then the detector will sweep a range of degrees in 2θ. A powder diffraction scan (2θ scan) assumes that a number of crystallites will be oriented with respect to the source beam to produce diffraction. Of these oriented crystallites, it is assumed that equal numbers of every possible orientation of the crystal will be aligned. For visualization of this, a diagram from Speakman’s presentation is included below as

Figure 3.3.69

Figure 3.3: Powder diffraction scan

In order to better visualize each plane, Figure 3.3 keeps the normal of the powder sample constant. In reality, the incident beam is constant and the crystal grain orientations are what changes. As such, only grains that are oriented with respect to the incident beam such that Bragg’s law is satisfied will diffract. This is why discrete peaks are observed in the 2θ scan, rather than having a constant intensity. Figure 3.4 below shows an example of two different orientations in a polycrystalline diffraction scan.

Figure 3.4: Powder diffraction – differing planes

The blue arrow above in Figure 3.4 denotes the direction normal to the entire powder sample whereas the black dashed line denotes the atomic plane normal. The two crystallites depicted refer to two different crystals orientations that are present in the powder sample. In both crystallites, the incident beam is set at 30° from the horizontal.

However, in the powder sample, both the (100) and (110) crystals would be present in orientations as depicted. As such, one would expect to see diffraction peaks at different angles from each other as depicted by the diffracted beams in the figure. Although this is just two examples, it can be seen from these that because there are a limited number of possible interatomic separations (dhkl), and the incident beam is fixed, that there will be a finite number of discrete peaks that form in the 2θ scan of a powder sample. These peaks correspond to each allowed diffraction peak.

3.1.2 Grazing Incident X-Ray Diffraction

Grazing Incident X-Ray Diffraction (GIXRD), sometimes called a glancing angle scan, is a particular type of scan that can be performed on an XRD system. This scan is designed for thin film materials whose crystallinity is unknown. A GIXRD scan is, essentially, a polycrystalline scan designed to interact with a thin film material and not the substrate.

For this scan, the user flattens the sample using ω and χ scans. After this, the user fixes the incident angle (ω) at approximately 0.9°. This incident angle increases the interaction volume of the x-rays within the film material. Thus allowing the diffracted beams to stem from the film rather than the substrate. After these steps are performed, a

2θ scan is carried out. If any peaks appear in this scan, then the film is sufficiently polycrystalline. In some applications, such as single crystal work, peaks that appear in this type of scan are undesirable. If no peaks appear in this scan then the film is: 1) highly oriented or 2) amorphous. If the film is highly oriented, then the in plane rotation may affect the ability to see polycrystalline peaks or the film is oriented such that no grains will diffract at the 0.9° incident beam. The latter is the case for single crystal work.

An example of a GIXRD scan of polycrystalline Ge deposited on Si can be seen in

Figure 3.5.

Figure 3.5: Example of a GIXRD scan

Note that multiple peaks appear in this GIXRD scan. This implies that the material is polycrystalline. For this scan the XRD is not set into a position such that a single crystal material should diffract. The peaks all correspond to different orientations of Ge. At

27.3° the (111) plane of Ge diffracts. One also sees the (022) peak at 45.3°. The additional peaks in the image correspond to other diffraction planes for the Ge crystal. If the sample was single crystal or amorphous, there would be no discernable diffraction peaks.

The GIXRD scans in this work are obtained using an x-ray mirror as the incident optics and a 0.09° parallel plate collimator for the diffracted optics. The divergence slit is adjusted in size from 1/16° to 1/2° based on the size of the sample.

3.1.3 Coupled Scans

The coupled 2θ-ω (sometimes called θ-2θ or ω-2θ) scan is the first example of a specifically single crystal, or highly oriented, type of scan. In this scan the incident and diffracted beams are both swept in a coupled manner such that the incident beam angle is adjusted at half of the rate of the diffracted beam angle. In other words:

훥휔 = 훥휃

For a concrete example, if ω is adjusted by 1° then the diffracted angle (2θ) is adjusted by 2°. Although this relationship is true of all coupled 2θ-ω scans there are two different kinds of 2θ-ω scans: the symmetric and asymmetric coupled scans. These scans refer to the position in reciprocal space one is examining and the orientation of the film grown.

For the coupled scans performed in this work, the 0.09° parallel plate collimator was utilized with the hybrid monocrometer and x-ray mirror incident optics. Most scans use a

1/2° divergence slit, for consistency between scans.

3.1.3.1 Symmetric Coupled Scans

As mentioned prior, there are two different kinds of coupled (2θ-ω) scans. Of these scans, the symmetric is the most straightforward. This scan probes the reciprocal lattice points along the sample normal direction, which is also known as the symmetric axis in reciprocal space. This is where the name “symmetric coupled scan” comes from. For a concrete example of the symmetric axis, it is possible to view reciprocal lattice points in a two dimensional image using software. In the PANalytical Data Collector program, one can plot a two dimensional graph of reciprocal space for a lattice defined in the program.

Below, in Figure 3.6, an example of this can be seen for a crystal of Al2O3 oriented with the (0001) set of planes grown along the sample surface.

Figure 3.6: Reciprocal space map of Al2O3

In Figure 3.6 each diffraction spot for the given crystal is represented by a red point.

The origin of reciprocal space, or the point 000, is at the center of the bottom axis between the two large gray semi-circles. These semi-circles show the limitations of the instrument, meaning that any points in these regions are inaccessible based on the diffractometer’s capabilities. The symmetric axis of reciprocal space is the axis that extends directly from the origin. This axis is clearly marked in Figure 3.7 below.

Figure 3.7: Reciprocal space map of Al2O3 – symmetric axis

The region enclosed by the red rectangle show the available symmetric diffraction peaks for Al2O3 oriented along the c-axis which is the (0001) family of planes. For every set of planes along this direction, the incident and reflected beams are related such that:

휔 = 휃

This is ignoring any slight offset that would be present due to sample mounting tilt

(which should be less than 0.5°). This relationship holds only with the symmetric coupled scans, because the planes examined are along the sample normal direction.

The interest in a symmetric scan is two-fold. First, they provide insight into the orientation of the crystal film grown on the substrate. For example, rocksalt TMNs grow

(111) planes on c-axis Al2O3 substrates. As such, one would expect (111), or another set in that family of planes, such as the (222), to grow if attempting to grow rocksalt TMN materials on c-axis Al2O3. Second, symmetric coupled scans provide information on out of plane lattice constants. The out of plane lattice constant can be paired with in plane lattice constants to calculate strain in the film.

If the film is strained, it could be either pseudomorphic (strained to the substrate lattice constant) or in some state of partial relaxation. If the film is relaxed or partially relaxed, then there is the possibility of the inclusion of Lomer dislocations, which occur along the interface, or other less desirable dislocations (such as threading dislocations).

These are often means of film relaxation. The 2θ value of the diffracted peak gives some insight into the strain in the material as this value will change based on the interplanar spacing. A material compressively strained in-plane will exhibit an increasing dhkl spacing out of plane, thereby changing the 2θ value at which diffraction occurs.

An example of a coupled 2θ-ω scan can be seen in Figure 3.8 below.

Figure 3.8: Thin TiN coupled scan

When examining XRD scans, it is important to understand conventions for displaying data. In the case of coupled scans, there are a few ways to present the data. In the case of

Figure 3.8, the value labeled on the x-axis is 2θ-ω. This convention notes that the scan is 49 a coupled scan, but the 2θ value is plotted. It is also possible to plot an ω-2θ, where the ω value is plotted. Otherwise, it is also possible to only label the x-axis as 2θ, but mention it in the caption as a coupled scan.

Many pieces of information can be obtained from this coupled scan in Figure 3.8.

First, it is possible to note that the diffraction peaks for both the TiN (111) and the Al2O3

(0001) orientations are observed in this image. In addition, they are at the diffracted angles that are expected for these materials. This implies that the TiN film layer is not significantly strained. Additional XRD scans will show that the TiN lattice is, in fact, completely relaxed. Second, the Full Width Half Maximum (FWHM) of these peaks can also be used to understand the quality of the film growth. Note that the Substrate peak is relatively narrow in angular resolution and has a FWHM on the order of 0.002° (or around 7-10 arcseconds). For thicker grown films, the FWHM should narrow with increasing crystal quality. This relationship does not hold if the film is thinner than about a few hundred nanometers where the film peak broadens simply due to film thickness.

Third, one notices satellite fringes appearing around the TiN peak. These fringes are known as Pendellösung fringes, or thickness fringes, as the period of the fringes relate to the thickness of the film layer. These fringes indicate high quality growth of the material as they are an interference effect between both the top surface of the film and the interface. As such, if the roughness of the film surface or the interface increases significantly, these fringes disappear. Additionally, increased scattering in the film from defects results in a loss of these fringes. So, one often interprets Pendellösung fringes as a sign of high quality growth.

As mentioned prior, Pendellösung fringes allow for obtaining the thickness of the film layer. The film thickness is inversely related to the spacing between the fringes, as shown in the relationship below.72

휆 sin(휃퐵 + 휑) ∆휃푝 = 푡 sin(2휃퐵)

Where Δθp is the Pendellösung fringe spacing, θB is the Bragg angle, t is the thickness of the layer, and φ is the angle between the Bragg planes and the crystal surface. In the case of the (111) oriented TMN materials grown on Al2O3, this additional φ term is zero.

It is important to note that the fringe spacing is inversely related to both the film thickness and sin(2θB). This means that the fringe period will decrease both with increasing diffracted angles and increased thickness. As such, it is possible to resolve these fringes for thicker films if their interatomic spacing (dhkl) is larger (i.e. the diffracted angle is smaller).

The previous TiN film, seen in Figure 3.8, is nearly 50nm thick, according to the fringe spacing. A coupled scan of a thicker TiN film (about 100nm) can be seen in

Figure 3.9.

Figure 3.9: Thick TiN coupled scan

When one compares Figure 3.8 and Figure 3.9, it is possible to note that the fringes in

Figure 3.9 are closer to one another, as expected of a thicker film. Note that if the film thickness continues to increase the ability to resolve the fringes will be lost. As such, one should not expect to see these fringes much past a few hundred nanometers.

3.1.3.2 Asymmetric Coupled Scans

Any coupled scan that is not along the symmetric axis of reciprocal space is called an asymmetric coupled scan. These scans involve planes that are no longer along the sample normal direction. As such:

휔 ≠ 휃

In this case, ω and θ are still scanning in a coupled manner but, because the planes are no longer along the sample normal direction, the sample must be offset in ω. An example can be seen in Figure 3.10 of the asymmetric region in reciprocal space.

Figure 3.10: Reciprocal space map of Al2O3 – asymmetric axis

Once again, the diffraction peaks enclosed by the red lines in Figure 3.10 show all the diffraction peaks for this orientation of Al2O3 that would be classified as asymmetric peaks.

As mentioned prior, symmetric scans allow one to probe the out of plane lattice spacing. This is because the symmetric axis of planes does not include any in-plane component. Asymmetric planes, however, include in-plane components in the sets of planes probed. As such, when paired with a symmetric scan, it is possible to obtain all the lattice constants for the crystalline film. These cannot be assumed to be the bulk values as, if the film is strained in any way, they could differ from the bulk.

Additionally, the crystal structure itself could differ. As an example, it was observed that

53 the TiN crystal grown Al2O3 was no longer a perfect cube, but had distorted. This could be observed by taking a symmetric coupled scan from the (111) family of planes and pairing it with an asymmetric scan such as the (113) planes. In which case, the two interatomic spacings allow one to obtain the lattice constant of the material and the change in bond angle. For the case of a rocksalt material, two scans are enough information. However, more complicated structures require using additional asymmetric scans.

3.1.4 Rocking Curves

Whereas a coupled 2θ- ω scan couples ω and 2θ together, a rocking curve (or ω scan) is performed at a particular diffraction condition. This involved finding the maximum intensity in the coupled direction (2θ-ω) and then “rocking” the sample in ω. This scan provides information about the quality of a film.

In particular, this type of scan looks at how well oriented a crystal is in a particular direction. If the crystal has any slightly misoriented grains, they would appear in this scan as broadening in the FWHM. Additionally, different types of dislocations are attributed to broadening of the FWHM of film rocking curves in particular direction.28, 29

High quality crystals, such as good substrate materials, often have extremely low FWHM values of their rocking curve measurements (on the order of 10s of arcseconds).

An example of two rocking curves, FWHM of 7 and 30 arcseconds, can be seen in

Figure 3.11 below.

Figure 3.11: Rocking curve FWHM comparison

The rocking curves in Figure 3.11 are examples of ScN grown on Al2O3 at differing deposition powers. These rocking curves show that the ScN grown at 50W possesses better crystalline film quality through two methods. First, the FWHM is 7 arseconds for

50W sample as compared to 30 arcseconds for 100W sample. Additionally, the background in the scan of 50W sample is higher despite nominally similar scan conditions.

Due to the rocking curve varying ω, the exact peak location is unimportant, as the sample itself is not necessarily placed onto the XRD exactly flat. In the case of any of the materials grown on Al2O3 in this work, XRD scans are often performed on these samples with them affixed to the system through the use of double sided tape. This guarantees that the ω-offset will differ between each sample. Thus, so will the exact

55 position of the rocking curve peak. However, the FWHM is not dependent on the sample tilt, so that value can be compared between samples. Additionally, if the same optics and diffractometer are used for each sample, the backgrounds can also be compared.

Rocking curve scans in this work are performed using the hybrid monocrometer and x-ray mirror incident optics with a three bounce Ge crystal detector, known as a triple axis detector. Divergence slits are typically held at 1/2° for consistency between scans.

3.1.5 Pole Figures

The types of scans mentioned prior to this all scanned for either the dhkl spacing of the sample or the distribution of dhkl spacings in the sample. With the pole figure (sometimes called a texture map), the XRD is set up at a specific dhkl spacing and then that spacing is examined. As such, 2θ and ω are fixed at the desired diffraction angle for the peak undergoing examination. Afterwards, φ and χ are varied from 0-360° and ~0-90°, respectively. This allows one to view the texture of the crystal.

Crystalline texture refers growth of a crystalline material in a preferential direction. It is possible to grow materials that are either amorphous, polycrystalline, highly oriented, or single crystal. There refer to the symmetries in the film. Amorphous films show no crystalline symmetries and, as such, are often isotropic in their properties. Many optical thin film applications, such as optical filters, take advantage of the isotropic nature of these materials. Additionally, films can be deposited in a purely polycrystalline nature, such that there are distinct XRD peaks in the GIXRD scans, but they are oriented in a completely random manner. These materials have no dependence on the φ, χ, or ω values and are inherently isotropic as well. As they have no φ or χ dependence, one would not

56 expect to see any particular peaks in an XRD scan that examines those features. This can be seen in a pole figure of polycrystalline Ge grown on Si in Figure 3.12 below.

Figure 3.12: Purely polycrystalline pole figure

In Figure 3.12, it is important to note that there is a significant amount of intensity over the entire range of the pole figure. Additionally, there is no visible peaks, which indicates no preferential growth direction. The circular pattern noted from about χ = 5°-

40° results from an increasing of the interaction volume in the film. As the χ angle increases, more of the x-rays interact with the film and are diffracted from the polycrystalline grains. After about χ = 40°, part of the x-ray beam no longer interacts with the polycrystalline film, thereby reducing the intensity.

Some materials, however, grow in a preferential manner such that they may not be single crystal, but still have a particular dependence on the φ or χ angles. These materials 57 are denoted as textured polycrystalline materials. ZnO is an example of a material that grows in a preferential direction on glass slides.64 Most materials examined in this work are either highly textured polycrystalline, highly oriented, or single crystals. These materials often display a high degree of dependence on φ and χ as only particular orientations will provide any sort of diffraction peaks. An example of a single crystal pole figure can be seen in Figure 3.13 below.

Figure 3.13: Single crystal pole figure

This is a pole figure of epitaxial TiN grown on Al2O3. As can be seen in this figure, a single well defined diffraction peak appears in the center of the figure. This corresponds to TiN (111). Additionally, six well defined peaks are highlighted in different colored circles. These correspond to the (-111), (1-11), and (11-1) peaks of TiN. The six fold symmetry implies twinning in the TiN. A pole figure of a highly textured film would look similar to Figure 3.13 but the peaks may be significantly broader due to the slight orientation differences in the grains.

3.2 X-Ray Photoelectron Spectroscopy

Although crystal structure is important in any grown film, so is the stoichiometry of the film itself. X-Ray Photoelectron Spectroscopy (XPS) is one of the techniques that provides information about film stoichiometry.

XPS, as the name suggests, employs the use of X-rays to probe information about film constituents. This is done by illuminating the material surface with x-ray radiation which is usually generated using an aluminum, magnesium, or zirconium sources.6, 8

This x-ray energy is then absorbed by electrons in the material. These now excited electrons eject from their atoms with a kinetic energy that is related to the binding energy of that particular material and the incident photon energy:

퐾퐸 = ℎ푓 − 퐵퐸

Where h is Planck’s constant and f is the frequency of the x-ray light. The binding energy (BE) of the state can be determined, then, by measuring the kinetic energy (KE) of the ejected electron. As the binding energy of each electron varies by the material, it is possible to use this information to identify the present atomic species to a minimal atomic concentration of about 1%.

The obtained XPS spectra can even distinguish between different binding states of a material.9, 10, 46, 74 As such, it is possible to not only obtain stoichiometry in general, but it is also possible to obtain information about different phases of material present. For example, it is possible to distinguish between peaks that are attributed to a material in the

4+ state vs the 5+ state, as the binding energies of the electrons in these two configurations are often different.46, 74 Peak fitting allows one to use the obtained spectra in order to quantify the relative amount of each concentration incorporated in the film.

One of the most common programs in order to perform XPS spectral analysis is

Computer Aided Surface Analysis for X-ray Photoelectron Spectroscopy, or CasaXPS.

This program allows for fitting of the various spectral features obtained in an XPS scan to various line shapes. An example of peak fitting using CasaXPS can be seen in Figure

3.14 below.

Figure 3.14: CasaXPS peak fitting

The peaks in Figure 3.14 are obtained examining the vanadium peak in a VOx film.

Each peak observed in the spectrum result from the V contribution in the V-O bonds.

The peak set near 523eV and 517eV correspond to the 2p1/2 and 2p3/2 binding energies.

However, it is noted that the fit of the 2p3/2 binding energy results in three peaks, attributed to the V5+, V4+, and V3+ binding states.46, 74 Using the areas of these fit peaks, it is possible to pull out the relative concentration of V atoms in each binding state.

5+ 4+ These can be correlated to the relative amounts of V2O5, VO2, and V2O3 for the V , V , and V3+ peaks, respectively.

One downside of XPS as compared to some other chemical analysis techniques is that the electrons ejected often cannot escape the film or material or interest if they are removed from an atom more than 10nm from the surface.5, 67 As such, XPS is extremely surface sensitive but is unable to do bulk characterization of a material without using a depth profile. A depth profile involves taking an XPS spectra and then sputtering away the material and re-taking the XPS spectra. This allows one to obtain some information about the material stoichiometry, even away from the original film surface.

Due to sputtering the material of interest, an XPS depth profile is an inherently destructive technique. Additionally, there is no guarantee that the material constituents will sputter at the same rate. As such, it is possible to have a misrepresentation of the stoichiometry in the film due to differences in sputter yields of the two constituent atoms.

For example, V and O in a VO2 may not sputter at the same rates. In the case of O sputtering faster than V, the VO2 film may progress to a VO1.5 film by the end of a depth profile, as an extreme example. Although, typically the sputter yield difference is not nearly this pronounced.

In addition to differences in sputter yield presenting a problem in depth profiles of

XPS, the process of sputtering away the material of interest can damage the crystal structure of the material. This results with a surface of a crystalline material becoming amorphous. As such, the potential to lose any differences in bonding states (such as a 5+ and 4+ state of the same material) arises.

Despite these drawbacks, however, this technique is extremely quick to perform and is often powerful enough to give a general understanding of the film stoichiometry and constituent materials. Additionally, it is possible to obtain information about contaminations included in the film, provided the inclusion is around 1%.

3.3 Secondary Ion Mass Spectrometry

Secondary Ion Mass Spectrometry (SIMS) is a destructive characterization technique that allows for the understanding the materials that make up a sample.3, 4, 50 Like XPS, it is possible to use this technique to quantify the stoichiometry of a material, but often

SIMS is used as a way to see impurity incorporation in a sample, as it is typically more sensitive than XPS.

SIMS characterizes a material by utilizing a high energy ion beam to eject atomic species from a sample. Often these species are ionized in the process of interacting with the primary ion beam (usually made up of Cs or O atoms).50 These newly ionized atoms, referred to as secondary ions, are accelerated away from the sample and analyzed by a mass spectrometer.3, 4, 50 This allows one to examine the composition of constituent atoms in a material to extreme accuracy, down to even parts per million incorporation.3

However, the majority of the atoms that the primary ion beam ejects from the surface are neutral atoms. These atoms are not accelerated towards the mass spectrometer, due to no charge, and are unused in the analysis.50 As such, it is often required to have a standard to perform extremely accurate SIMS measurements. A standard is a material with known quantities of impurities or dopants that one can compare to. This allows for more accurate quantification of impurities in a material. Often, these standards must be of a similar material. For example, if one wants to examine incorporation of oxygen in

ScN, it is helpful to have a ScN material standard of known oxygen content to act as a standard.

Although SIMS is more sensitive than XPS, it also has its own sets of drawbacks.

First, SIMS is a destructive technique and cannot be anything but. The primary ion beam will sputter away the surface of the material. Additionally, SIMS often naturally includes a depth profile. This means that the film will be sputtered away in the process of obtaining SIMS measurements. Second, SIMS requires detailed analysis of the obtained data in order to accurately determine the concentrations of atomic species.4 Third, because SIMS requires rigorous analysis, it is often provided as a service and can be relatively expensive as compared to other characterization techniques. All of the SIMS done in this work was obtained by sending out the samples and having SIMS provided, rather than actually performing the measurement and analysis in-house.

3.4 Spectroscopic Ellipsometry

In order to use a material in an optic application, one must first understand the refractive index of the material. One way to obtain the refractive index of a given material is through the use of Spectroscopic Ellipsometry (SE). This non-destructive, optical characterization technique allows one to examine how a sample interacts with light of different polarizations.1, 32, 36

Fundamentally, reflection SE measures the ratio of two parameters known as Fresnel reflection coefficients.1 When examining the Fresnel reflection coefficients a difference arises when examining Transverse Electric (TE) and Transverse Magnetic (TM) polarizations of light. These polarizations are often referred to as s- and p-polarizations

63 when examining work with SE. As such, s- for TE-polarized light and p- for TM- polarized light will be used for the duration of this work.

As mentioned prior, the measured data for an SE scan is the ratio of the two Fresnel reflection coefficients. This ratio, ρ, is a complex valued parameter that is displayed as two angles psi (Ψ) and delta (Δ).32 The complex valued ellipsometric parameter is defined as:

푅̃푝 휌 = = tan(Ψ)푒푗Δ 푅̃푠

In this definition above, the pseudo-Fresnel reflection coefficients for the s- and p-

1 polarized light are 푅̃푝 and 푅̃푠, respectively. The tilde above the values represent only that each value is complex. The angles Ψ and Δ are the values reported by the SE.

When examining a film grown on a substrate material, one can obtain an effective index for that combination of materials. This effective index gives rise to effective reflection coefficients. These reflection coefficients are referred to as pseudo-Fresnel reflection coefficients as they are not a reflection coefficient for a single material, but the combination of substrate and film.1 In the case of a bulk material, these would simply be the Fresnel reflection coefficients.

As mentioned prior, the Fresnel reflection coefficients are related to the index of refraction of the material undergoing examination. For a single material, the complex

Fresnel reflection coefficients are:1

푛̃1 cos(휃̃0) − 푛̃0 cos(휃̃1) 푟푝̃ = 푛̃1 cos(휃̃0) + 푛̃0 cos(휃̃1)

푛̃0 cos(휃̃0) − 푛̃1 cos(휃̃1) 푟푠̃ = 푛̃0 cos(휃̃0) + 푛̃1 cos(휃̃1)

Where the subscripts are defined as the light propagates from material 0 to material 1.

Note that according to Snell’s Law, it is possible to have complex angles of propagation in the case of materials with complex indices of refraction.

The pseudo-Fresnel reflection coefficients are a combination of multiple film and substrate reflection coefficients. In the simplest case, one can define the pseudo-Fresnel reflection coefficient for a single film layer on a substrate material. An example of optical propagation in this scenario can be seen in Figure 3.15 below.

Figure 3.15: Optical propagation in a film

It is possible to see that at every interface both a reflection and transmission of the optical wave occurs. When examining the reflected optical wave, one must take into account not only the initial reflection, but also all the reflections from the film-substrate interface. Reflections off the backside of the substrate are not included due to both low intensity and the lack of coherence with the incident wave after traveling through the substrate. After adding the contributions of each reflected wave and simplifying, one can obtain the pseudo-Fresnel reflection coefficient for this system:1

푟̃ + 푟̃ 푒−푖2훽 ̃ 01 12 푅 = −푖2훽 1 + 푟01̃ 푟12̃ 푒

Where β is the optical thickness of the film material. This expression provides the pseudo-Fresnel coefficients for both the s- and p-polarized light, provided one remembers to use the correct reflection coefficient definitions. Although relatively simple to define for a single film layer on a substrate, this problem becomes increasingly complex with additional film layers. However, it is possible from this expression to see how the effective properties of the film and substrate stacks are related back to the individual properties of the materials. As such, it is possible to see how both Ψ and Δ are related to the indices of the film(s) and substrate.

Although SE is a powerful technique, it comes with one drawback: the Ψ and Δ values obtained from SE do not, in of themselves, have much physical meaning. As such, in order to obtain the refractive index, one must utilize the Ψ and Δ values to create a fit for the index of the material. This is where the difficulty in using the SE system is found, as fitting the values is not necessarily straightforward. In addition, the fits are not unique. It is very much possible to fit an SE scan with extremely low mean standard error (MSE) values, but have the fit itself make no sense. For example, many electron transitions that are observed take the form of a Lorentzian peak. As such, in some materials it is expected to find electron transitions as absorption peaks in the SE fits. But, a negative amplitude on a Lorentzian peak makes no sense as this would suggest gain in the system. Without pumping, this should not exist. So, if one wants to draw conclusions about the material, such as the location of the bandgap or defect bands, one must make sure that the parameters used to make the fit also make physical sense.

Without this, it is possible to obtain a fit with an extremely low MSE, but not have the

66 correct reasons for the features seen in the scan and, therefore, an incorrect refractive index.

Like mentioned before, it is possible to use particular functions in the SE fit that correspond to particular material properties. One example property is the bandgap of the material of interest. In the case of the J.A.Woollam software, it is possible to use a parametric semiconductor (PSEMI) function to model the band edge of a material. The

PSEMI-0 function is often used for this application.

In addition, it is possible to obtain information about the free carriers in a material through the use of a Drude model. This model acts as similarly to an exponentially increasing function at low energies. This model takes the form:1

−ħ2 휀퐷푟푢푑푒 = 2 휀0휌(휏퐸 + 푖ħ퐸)

Where

푚∗ 1 휌 = = 푁푞2휏 푞휇푁

In the expressions above, E is the photon energy (ħω), ρ is the resistivity of the material, and τ is the scattering time. The physical constants included in this model are the reduced Plank’s constant (ħ), permittivity of free space (ε0), and electron charge (q).

Both ρ and τ are the parameters used to fit the Drude model. If one either assumes or measures the effective mass of the material (m*) it is possible to obtain both the carrier concentration (N) in cm-3 and the mobility (µ) in cm2V-1s-1. This function allows for fitting of metals and doped semiconductor materials.

As seen prior, the film thickness plays a role in the pseudo-Fresnel reflection coefficients through the inclusion of the optical path length. As such, it is possible to

67 obtain film thickness values from SE. One method in order to do this is to examine a film material in the transparency window, also known as a wavelength region of low loss.

By doing so, one can fit a Cauchy dispersion model to the film layer and back out the film thickness. This dispersion model acts as a slowly decaying function in wavelength.

The permittivity associated with a Cauchy dispersion model is described as:1

2 휀퐶푎푢푐ℎ푦 = (푁 + 푗퐾)

Where

퐵 퐶 푁 = 퐴 + + 휆2 휆4

And

퐾 = 훼푒훽(퐸−훾)

Six fitting parameters are used in this model: A, B, C, α, β and γ. Once again, E is the photon energy and λ is the photon wavelength. Using this model, one can usually obtain a thickness value for the film layer. However, this model often fails to adequately fit materials with loss. As such, it is often best to use this model to obtain a thickness in the transparency window and then later fit the whole dielectric constant using other methods with the obtained thickness.

These are just a few of the additional properties that can be obtained from the SE system. However, they are all utilized in this work for varying materials. As such, SE is instrumental in understanding multiple material properties of the films grown in this work.

3.5 Hall Effect and Transport

In order to utilize a material in an electronic application, one must understand the resistivity, carrier concentration, and mobility of the carriers. Knowing these properties,

68 one can find the best application for a given material. As mentioned prior, it is possible to obtain at least the carrier concentration and mobility of a material from the Drude term in an SE fit. However, it is often easier to preform Hall Effect measurements on the samples of interest to obtain these values and then fit the effective mass in the SE. As such, Hall Effect measurements were used in order to characterize the majority carriers and mobility of the films grown in this work.

The underlying principle of the Hall Effect stems from the Lorentz force, where a moving charge in a magnetic field feels a force.2, 27, 65 This force is:

푭⃗⃗ = 푞풗⃗⃗ × 푩⃗⃗

Where q is the charge, v is the velocity vector of that charge, and B is the magnetic field. In the cases examined in this work the majority charge carrier is the electron. As such, the charge is the electron charge. In order to obtain a moving charge, in most applications an electric field will be applied to obtain a current through a sample. The full Lorentz force is then:

푭⃗⃗ = 푞(푬⃗⃗ + 풗⃗⃗ × 푩⃗⃗ )

Where E is the additional field applied.

When a magnetic field is applied to a sample with current running through it in the configuration of Figure 3.16, the electrons in the sample feel a force and move toward one side of the sample.

Figure 3.16: Hall Effect configuration

This forced movement of electrons creates a voltage difference between the sides of the samples. This voltage difference is known as the Hall voltage (VH). VH is related to both the sheet concentration and the mobility of the carriers. As such, knowing the applied magnetic field, current, sheet resistance, and the measured Hall voltage, one can obtain the carrier concentration and mobility as follows:2, 65

퐼퐵 푛푠 = 푞|푉퐻|

And

|푉 | 휇 = 퐻 푅푠퐼퐵

Where ns and µ are the sheet concentration and mobility, respectively, I is the current,

B is the magnitude of the applied magnetic field, q is again the electron charge, VH is the measured Hall voltage and Rs is the sheet resistance. Due to the velocity directions differing between holes and electrons, the charges congregate on the same side of the sample. As such, the sign of the measured Hall voltage is negative for n-type materials and positive for p-type materials. In this way, it is possible to determine the majority carriers using this measurement.

One downside to Hall Effect measurements is that it is often required to pair it with a resistivity measurement in order to obtain the sheet resistance. As such, contacts must be deposited onto the sample. In the van der Pauw method, which is a resistivity measurement that is often paired with Hall Effect, these contacts could simply be on the corners of the sample, but this still requires some sample prep that may be irreversible.2

However, using these methods allows for accurate measurement of carrier concentration and mobility, even to low temperatures.

CHAPTER 4

SCANDIUM NITRIDE GROWTH AND CHARACTERIZATION

As mentioned prior in 1.1.2, ScN is a rocksalt TMN material that is interesting for use in GaN based electronics due to its extremely close lattice match to GaN. However, it is important to note that GaN is a material that has a hexagonal crystal structure, so the

(111) orientation of ScN is preferred for GaN regrowth.

In order to obtain (111) ScN, it is possible to use c-axis oriented Al2O3, which is the

(0001) orientation of Al2O3. The benefits of using Al2O3 in this case are that it is a fairly inexpensive substrate and is often used for growth of GaN already. However, it is well known that a film material’s lattice structure must match well to the substrate lattice structure in order to grow well on that substrate. ScN has a lattice constant of about

16 4.503Å and online suppliers mark Al2O3 with an in-plane lattice constant around

4.780Å. It is important to note that these lattice constants cannot be directly compared, as the ScN lattice constant is along the edge of the cubic cell. This means one must first convert the ScN lattice constant into a number that is equivalent for the hexagonal system. This can be done by rotating the rocksalt crystal into the (111) orientation as shown in Figure 4.1 below.

Figure 4.1: ScN hexagonal structure

The marked distance “d” in Figure 4.1 is the distance along the cube diagonal. As such, one can obtain the hexagonal in-plane constant fairly simply:

푑 = 푎√2

Where a is the actual lattice constant (4.503Å) and d is the hexagonal lattice constant.

If one performs this calculation, the hexagonal lattice constant for ScN can be obtained.

Using the 4.503Å, the (111) in plane lattice constant is 6.368Å. This is a value that can be directly compared to the 4.780Å in-plane lattice constant of Al2O3. It is possible to calculate the lattice mismatch of the system by comparing the substrate and film lattice constants:

푑퐴푙 푂 − 푑푆푐푁 훿 = 2 3 푑퐴푙2푂3

If this calculation is carried out, one obtains a lattice mismatch of -33%, where the negative sign in this case simply shows that the film is under compressive strain. A lattice mismatch of this value is typically considered extremely high. However, there is an additional orientation in which ScN can grow in this system. In Figure 4.2 below, one

73 can see both the lattice that was obtained simply by using the (111) orientation of ScN and the lattice that can be obtained with an in-plane 30° rotation, with respect to the

Al2O3 substrate.

Figure 4.2: In plane rotation

The crystal on the left in Figure 4.2 is the crystal that is obtained simply using the in- plane lattice constant for (111) ScN. However, an additional hexagonal structure can be seen on the right of Figure 4.2 which includes a 30° rotation from the structure on the left. This 30° can create a new constant:

푑30° = 푑111 ∗ cos (30°)

Where d30° is the rotated lattice constant and d111 is the standard (111) lattice constant.

The newly obtained lattice constant is 5.515Å which results in a new lattice mismatch of

-15%. This value is still substantial, but is significantly less than the previously calculated value, which makes this growth configuration energetically more favorable than simply growing in the non-rotated (111) orientation. This is, in fact, how the ScN grows on the Al2O3 surface. One sees this in XRD pole figures performed on optimized

ScN films. Figure 4.3 below shows the obtained pole figure on (111) oriented ScN.

Figure 4.3: ScN (111) pole figure

In Figure 4.3, one notices a large central peak and six additional outer peaks. These peaks correspond to the (-111), (1-11), and (11-1) peaks of ScN. However, the six peaks show that there is twinning of the ScN domains on the Al2O3 surface. A pole figure of the (1-12) peak of Al2O3 substrate of this film, run at the same time as the ScN pole figure, shows how the two crystals are oriented relative to one another. The (1-12) peak of Al2O3 should align with the (-111) family of peaks in the ScN, if there is no misorientation between the two crystals. This pole figure can be seen in Figure 4.4 below.

Figure 4.4: Al2O3 (1-12) pole figure

As can be seen when comparing Figure 4.3 and Figure 4.4, there is approximately a

30° difference in φ orientation of the peaks. This implies a 30° in-plane rotation between the two crystals.

4.1 Nitrogen Fraction

When growing a material for the first time, one of the most important parameters in a reactive system is the fraction of reactive gas in the plasma. This parameter dictates the phase and stoichiometry of the deposited film. As such, this is often one of the parameters that is varied first when optimizing a new material. For all of the TMN samples, this is reported as the nitrogen fraction (fN2).

For this set of samples, some of the optimized TiN growth parameters were used as a starting point for the ScN growth. These parameters included: substrate temperature

IH=30A (550°C) and coil current (Ic=4A). Additionally, 200W was chosen for the deposition power. Some TMN materials, like ZrN, grow well at low fN2, so an entire range of nitrogen gas fractions were chosen from 1%-75%. Once again, the chamber was held at around 20mTorr background pressure and these percentages are in reference to that value.

During the initial optimization of film growth, both AFM surface measurements and

XRD scans allowed optimization of the film roughness and crystallinity. Due to their speed and relevant information, these techniques were ideal for the optimization sets. As such, both the AFM RMS roughness and XRD coupled 2θ-ω scans were used to optimize the fN2 value of the ScN growth. However, crystallinity dictated the major optimization parameter.

When examining a material for the first time, a coupled scan that covers a significant range of angles (e.g. 10° - 110°) provides ample information. One can use this scan to see the different crystal phases grown on a sample due to the large range used.

Additionally, this ensures that one will obtain any available coupled peaks in the sample.

The survey scans for the different samples grown with varying nitrogen fractions can be seen below in Figure 4.5.

Figure 4.5: ScN nitrogen fraction surveys

The different peaks that appear in Figure 4.5 correspond to different crystals present in the grown film and substrate. The peaks around 42° and 91° correspond to the Al2O3 substrate. The additional peak that appears around 35° corresponds to the desired ScN

(111) crystal structure. As such, both the 50% N2 and the 75% N2 growth conditions provide single phase ScN (111) films. However, the 75% N2 peak is both sharper and more intense. This is more easily seen if the graph is zoomed in around the first order peaks, as seen in Figure 4.6 below.

Figure 4.6: ScN nitrogen fractions – zoomed in

Note that the steps in the nitrogen fraction were not extremely fine, but this gave single phase materials to then adjust the other deposition parameters to further optimize the films. The asymmetry in the single phase ScN peaks could result from strain in the crystal, which was expected as the correct growth parameters were not yet fully optimized.

4.2 Magnetron Power

Previously, the magnetron power for the ScN samples were fixed at 200W. However, this value may not have been the optimized value for this material growth. After finding the nitrogen fraction that provided single phase ScN, it was necessary to optimize the deposition power in order to create uniform films. Once again, XRD coupled 2θ-ω scans were used to optimize the film growth parameters. For this optimization step, the

79 magnetron power was varied from 50W to 200W. The resulting XRD coupled scans can be seen below in Figure 4.7.

Figure 4.7: ScN original power series surveys

Throughout this coupled scans, the Al2O3 peaks are observed at about 42° and 91°.

The film shows single phase throughout all of the different deposition powers, however it is easier to see the intensity and FWHM differences of the peaks if the area around the

(111) and (0006) peaks are examined more closely. Figure 4.8 below shows this region in more detail.

Figure 4.8: ScN power series – zoomed in

After examining Figure 4.8, one notices that the ScN peak becomes more intense as the deposition power decreases. Additionally, the 50W deposition shows the beginning of a fringe pattern on the left side of the ScN peak. This indicates the start of

Pendellösung fringes in this film, implying the 50W growth provides the best film of these deposition conditions.

4.3 Substrate Temperature

In addition to the deposition power and the nitrogen fraction, the substrate temperature can be adjusted in order to provide more (or less) energy to the deposited material. This allows the material either more energy to find the most energetically stable position in the lattice, or less to force the deposited atoms to maintain position. As such, substrate temperature constitutes an extremely important parameter for film growth. 81

Additionally, the optimum growth value can vary significantly from material to material.

In the case of ScN, it is necessary to vary the substrate temperature from 30A to 50A

(550°C to 885°C) in order to obtain the highest quality films. The survey scans for these films can be seen in Figure 4.9 below.

Figure 4.9: ScN temperature series surveys

In all the samples other than the film grown at the low 550°C temperature, the ScN grew in a single phase. Note that the previous ScN samples were deposited at the temperature of 550°C. However, a significant portion of time passed between the growth of this sample and the previous ScN samples. Due to the high oxygen affinity of Sc, most likely the target oxidized between growths and this peak is an inclusion of Sc2O3, rather than an addition ScN phase.

Despite no clear trend showing in the survey scans, a more detailed examination of the ScN (111) peaks leads to some readily available conclusions. These detailed scans can be seen in Figure 4.10 below.

Figure 4.10: ScN temperature series detailed scans

Pendellösung fringes can be seen to appear in the samples with deposition temperature higher than 550°C. However, the peaks below 860°C are highly asymmetric, with fringes only on one side of the ScN peak. The 860°C are also asymmetric, but this film at least has fringes on both sides of the peak. This means that ScN is extremely sensitive to deposition temperature where the range of acceptable ScN growth is a fairly narrow range of about 50°C around 860°C. However, the growth at 860°C (IH = 47A) provides the best quality film in this set.

4.4 Final Optimization

Once each individual parameter is examined and optimized, it is necessary to once again go through the various optimization parameters in order to obtain the highest quality films. For this process the AFM RMS roughness now become a useful optimization parameter because the crystal itself should be of relatively good quality, but the surface could be improved upon for subsequent film growth.

After realizing that the ScN films grew better at 860°C, the nitrogen fraction was once again examined in a range of 25% to 75%. The resulting symmetric coupled scans of these films can be seen in Figure 4.11 below.

Figure 4.11: ScN nitrogen fraction at 860°C substrate temperature

It is possible to see in Figure 4.11 that, once again, the ScN film prefers growth at higher nitrogen fractions, around 75%. However, one may note what looks like the start

84 of Pendellösung fringes in the 50% deposition, but they are not nearly as defined as the

75% deposition.

In addition to re-examining the nitrogen fraction, the deposition power was also further optimized. Samples were grown from 25W to 200W in order to find the best optimized deposition power. The original set of samples grown to vary the power can be seen in Figure 4.12 below

Figure 4.12: Power series optimization at high temp – original set

The lack of Pendellösung fringes in the 50W and 100W films is suspect due to the previous films including fringes. Upon examination of the observed current and voltage values for the substrate heater, the temperature seems to be at a higher value for these films, as the product of the current and voltage is larger. As such, regrowth of these films

85 allows for a more accurate examination of this property. The XRD scans of these samples can be seen in Figure 4.13 below.

Figure 4.13: ScN power series optimization at higher temperature

After examining the samples grown in Figure 4.13, it is possible to realize that growth of ScN also favors higher deposition powers. The 75W and 200W films show well defined Pendellösung fringes, indicating high quality growth at these values of deposition power. This suggest that ScN crystallinity improves at higher deposition powers. In addition, it was possible to obtain film thicknesses from these Pendellösung fringes showing that all of the films were nominally 35-40nm thick.

However, in order to use the ScN for regrowth of GaN, the surface of the ScN must also be extremely smooth. As such, AFM surface analysis on these films can help to

86 further define the best growth parameters. The AFM surface analysis of each of these samples can be seen in Figure 4.14 below.

25W 40W

RMS Roughness: 2.602nm RMS Roughness: 1.926nm

50W (1) 50W (2)

RMS Roughness: 2.079nm RMS Roughness: 0.529nm

75W 100W (1)

RMS Roughness: 0.290nm RMS Roughness: 1.372nm

100W (2) 200W

RMS Roughness: 0.088nm RMS Roughness: 0.121nm

Figure 4.14: Surface AFM measurements of ScN power series

In Figure 4.14 it is important to differentiate between the samples of the original series as compared to the regrown samples. These samples are differentiated as (1) for the original 50W and 100W and as (2) for the regrown samples. Note that a clear difference between the first 50W and 100W and the regrown samples in the form of

88 dramatically lower RMS roughness values. These values show that the original samples were significantly rougher, implying lower quality growth.

In addition to examining the RMS roughness of the films, one must also examine the actual shape of the growth. The most clear image is the 100W (2) surface where clear atomic steps are observed. This implies high quality step flow growth. If one also examines the 200W, 100W (1), 75W, and 50W (2), then steps can also be seen, but they are not as clear as in the 100W (2) film.

In addition to examining the coupled XRD scans and the AFM, it is also important to examine the rocking curves of the symmetric XRD peaks, which can shed some light into the best quality growth. An example of these rocking curves can be seen in Figure 4.15 below.

Figure 4.15: Rocking curve of 75W ScN sample

Note that this peak in Figure 4.15 is extremely narrow. The FWHM of this peak is on the order of 7 arcseconds, which is comparable to the substrate FWHM. This suggests extremely high quality growth of these films. Each additional film grown in this power series underwent examination of rocking curves, and the plotted FWHM for these scans can be seen in Figure 4.16 below.

Figure 4.16: ScN power series rocking curve FWHMs

In all cases, the rocking curve FWHM is extremely low. The largest seen value in this case is the 100W (1) sample, with a FWHM of around 28 arcseconds. This is still a fairly low value, but it is the worst in the study. The other FWHM values are on the order of the substrate FWHM, meaning they are extremely well oriented.

Using both the information obtain from the XRD and the AFM, it is possible to note that high quality growth of ScN occurs above 50W. The films between 75W and 200W all show relatively high quality growth. This means that it is possible to use a range of

90 deposition powers and still obtain high quality growth, allowing for some tuning of other properties, such as optical, by changing the deposition power. However, the film that seems to perform the best taking into account both the XRD and the AFM measurements is the 100W (2) film. This implies that the best growth conditions for the crystallinity of

ScN is 100W deposition power, 75% fN2, 47A IH (860°C), and 4A IC.

4.5 ScN Optical and Electronic Properties

The use of ScN in GaN based electronics requires knowledge of both the optical and electronic properties of ScN. However, many of the properties of TMN materials can change due to differences in growth technique, which allows for some tuning of the properties. In the case of ScN, the examination of the optical and electronic properties of the final power series allows one to understand the role that the deposition power plays in the film properties.

The optical properties of the ScN films were examined using SE measurements on two J.A.Woollam VASE systems. The two systems utilized were both the UV-Vis system, outfitted with an extended IR detector, and the IR-VASE. These systems allowed for examination of 193nm-3.2µm and about 1µm-30µm, respectively. However, it achievement of much shorter wavelengths than 300nm in the UV-Vis system without changing out the light source frequently proved difficult. As such, most scans ended at about 275nm, in order to not introduce significant noise into the data. Additionally, the bandgap of ScN at around 2.1-2.4 eV provided an ideal range of interest from around

20µm to 275nm, or about 0.06 eV to 4.5 eV. The obtained n and k values for ScN over this range can be seen in Figure 4.17.

Figure 4.17: ScN full range optical constants

As mentioned prior, the first set of 50W and 100W seemed to not produce the same quality of crystal as expected. However, these values are included in the image as 50W

(1) and 100W (1), denoting the first set. Likewise, the second set of 50W and 100W are labeled as 50W (2) and 100W (2).

In the full range it is important to note two different features that appear in the response of the ScN. First, at low energies, a large change in n and k can be observed.

This feature is a direct response Drude term, or the free carrier absorption in the ScN.

Using this term, it is possible to fit for both the mobility and the carriers of the ScN film.

Additionally, one notices a large increase in k at around the 2.1-2.4 mark, which is due to the absorption of the film starting after the bandgap of the ScN film. It is also important to note that the response of the film at this bandgap seems to change with deposition power. Figure 4.18 below shows a clearer image of the bandgap change that can be observed at differing deposition powers.

Figure 4.18: ScN bandgap examination of power series

As can be seen in Figure 4.18, the feature in k that is attributed to the band edge shifts to higher energies as the deposition power increases, aside from the 25W and 50W (1) samples. Additionally, the 50W (2) and 100W (2) samples also have a noticeably higher energy band edge feature.

The shift in the bandgap of ScN can be due to multiple factors. One factor to examine is the strain of the crystal, which will be examined using XRD imaging of the asymmetric peak of the film. Additionally, the shift in the bandgap could also be due to the Moss-Burstein effect, where the apparent absorption edge of a semiconductor is pushed to higher energies due to a majority of the states near the conduction band being occupied. These two effect could possibly explain the observed change in the bandgap of the ScN film.

Hall Effect measurements were performed on the ScN films in order to examine the carrier density and mobility of films. These results were compared to the obtained SE results, which provided good agreement to the trends seen. The Hall Effect results can be seen in Figure 4.19 below.

Figure 4.19: ScN room temperature Hall Effect results

As can be seen by the carrier concentrations from the Hall Effect plots, the carriers for all of the original set are around the same value of about 5-6x1020 cm-3. However, the two regrown samples show carrier concentrations of nearly double. This suggests that the trend of increasing bandgaps seen in the original set is not due to a Moss-Burstein shift. However, the increased value of carriers in the new set should show some effect due to a Moss-Burstein shift.

As for why the carrier concentration of the regrown samples are so much larger, the removal and replacement of the Sc target between the growths plays a factor. This allows for additional oxygen incorporation into the ScN target, which has been shown to be an n type donor in ScN.16 Secondary Ion Mass Spectrometry (SIMS) results of similarly grown ScN films show oxygen incorporations on the order of 1020-1021 cm-3, which is similar to what was observed for the carrier concentration in the Hall Effect results. This

SIMS analysis, performed by EAG, can be seen in Figure 4.20.

1E+23 AFRL: Sample ScN RX-39 1E+03

1E+22 Sc-> Al (a.u.)-> 1E+02

1E+21 N-> 1E+01 H O 1E+20 1E+00 C 1E+19 F Ta 1E-01 1E+18

1E-02

1E+17 (atom%) CONCENTRATION Sc,N

O,F,C,H,Ta CONCENTRATION (atoms/cc) CONCENTRATION O,F,C,H,Ta Profile tails may be compromised by morphology / roughness 1E+16 1E-03 0 50 100 150 200 250 300 350 DEPTH (nm) Y0JZM839_YA_08 Sample ScN RX-39_overlay

Figure 4.20: ScN SIMS analysis

When performing analysis on a sample using SIMS, it is important to understand how the observed impurities incorporate into the film. In the case of oxygen and fluorine in

ScN films, this is understandable due to the fact that Sc reacts readily with oxygen.

Additionally, fluorine is found in the solution used to process the Sc targets. The small amount of tantalum observed in the film arises from the tantalum heating element in the magnetron sputtering system. The carbon and hydrogen, however, are more difficult to explain. It is possible that there are incorporations into the target prior to loading into the chamber. Or, perhaps, the 99.999% pure Ar and N2 gases used in the magnetron sputtering system still include some amount of H2O, leading to the inclusion of these elements into the film. Regardless, it has been shown that the oxygen incorporation in the films is the most likely source of the additional carriers.16 The fact that this agrees

95 with the order of the carriers observed by the Hall Effect measurements, helps to corroborate the fact that removal of the target between growth sets contributes the additional carriers observed.

However, if the observed shift in the bandgap does not arise from a change in the carriers in the original series, there must exist some other phenomenon that contributes to this change in the bandgap. One such possibility is lattice strain or relaxation. In order to examine this, one must examine both the in plane and out of plane lattice constants in the film. This can be done by both a symmetric coupled scan and an asymmetric coupled scan. The particular sets of planes examined are the symmetric ScN (111) and the asymmetric (113) planes. A summary of the obtained lattice spacings can be seen in

Figure 4.21.

Figure 4.21: ScN power series planar spacings

The examination of the ScN film layers via symmetric and asymmetric coupled scans shows a trend of increasing lattice spacings as the deposition power increases. The bulk

ScN crystal values for the d111 and d113 spacings are 2.5981Å and 1.3568Å, respectively.

These values are marked by the horizontal blue lines in Figure 4.21. The films deposited

96 at higher deposition powers have lattice spacing values closer to the bulk properties than those deposited at lower deposition powers. However, if this resulted from varying states of strain in the films, one would expect an opposite relationship between the d111 and d113 spacings due to compressive or tensile strain in one direction and the opposite in the other. For example, if the ScN was under compressive strain in plane, one would expect a decrease in the d113 spacing, but an increase in the d111 spacing. This implies that the films themselves have differing lattice constants from the bulk 4.503Å value.

Normally, a change in lattice constants results from increased impurity incorporation in the films or changes in stoichiometry in the films. Although these films did not undergo any compositional analysis, examination of other ScN films grown in the same chamber can be used to understand this phenomenon. These films, grown on MgO, were examined by Cetnar et. al.16 and they showed, via SIMS, that the films deposited at higher powers incorporated more impurities from the target. This could result in the changes observed in the lattice constant of the film. However, the lattice constant nearing the bulk ScN lattice constant at higher powers seems to possibly disagree with this theory. Regardless, future would should examine the film stoichiometry and impurity incorporation to fully understand exactly why these lattice constants change due to the deposition power.

Examining of the optical properties of ScN it reveals that varying the deposition power results in a change of the bandgap of the ScN. Additionally, this change in the bandgap seems to arise independently from any Moss-Burstein shift observed in the films. This is corroborated by the later grown samples where a very large shift is observed by doubling the carriers. However, the resulting shift in the bandgap of the

97 original series of ScN films is accompanied by a shifting of increased d111 and d113 spacings toward the bulk crystal values. This implies a relaxation of the crystal toward the bulk parameters at the higher values of deposition power and, as such, an increase in the lattice constant of the ScN films towards the bulk.

CHAPTER 5

ALUMINUM NITRIDE GROWTH AND CHARACTERIZATION

In the case of AlN, the films are desired mostly for piezoelectric applications. As such, the films must be highly crystalline to preserve the piezoelectric effect. As mentioned prior, AlN has a wurtzite crystal structure, which means that it should grow well on other hexagonal crystal structures such as Al2O3. This is the typical substrate used for GaN, AlN, and AlGaN materials when using the c-plane of Al2O3.

Once again, in order to grow a crystalline material on a substrate, the crystals much match well. This can again be examined using the lattice mismatch between the AlN and

Al2O3 in-plane lattice constants. These are 3.1130Å and 4.780Å, respectively. This leads to a lattice mismatch of about 35%. Like the ScN, this value can also be reduced to about

13% by using a 30° in plane rotation and using two times the lattice constant of AlN. As such, one can use Al2O3 to grow AlN in the (0001) orientation. Having AlN oriented in this way allows also for convenient testing of the piezoelectric coefficient as the (0001) direction of AlN has the largest piezoelectric coefficient.

5.1 Nitrogen Fraction

Once again, it was first necessary to optimize the nitrogen gas ratio in order to obtain the correct phase of the desired material. The deposition of AlN in this work was done in the controllably unbalanced reactive magnetron sputtering system designed for nitride growths. As such, the total background pressure was once again held constant at

20mTorr and the ratio of nitrogen to argon was varied.

However, the other deposition parameters were chosen from previous work done by the group with AlN. These deposition parameters were chosen as: IH = 40A (760°C), IC

= 8A, and Wt = 100W. This allowed starting with nearly single phase AlN.

One important factor to consider when growing AlN in a sputtering system is that

AlN is an insulating material. This means that DC sputtering can struggle to deposit this material due to charge build up on the surface of the target during growth. Observation of this phenomena occurs during the original depositions of AlN when performed at high nitrogen fractions, such as 50% N2. In order to sputter at these nitrogen fractions, AC sputtering provides a method of reducing charge buildup. However, it is possible to sputter AlN in a DC system without apparent breakdown on the surface of the Al target simply by reducing the fN2 of the plasma. As such, the depositions of AlN take place at significantly lower values of fN2 than the other TMN materials examined thus far. After about 35%, breakdown on the surface of the AlN target is observed, meaning that all growths must be performed below this value. Therefore, fN2 was varied from 5% to 35%.

The symmetric coupled scans from the samples grown at these values can be seen in

Figure 5.1 below.

100

Figure 5.1: AlN nitrogen fraction coupled scans

From Figure 5.1, it is possible to note that above 5% fN2 the samples are single phase with the AlN growing in the (001) orientation. However, it is fairly difficult to notice the differences between the AlN (002) peak after the 5%, as such, a plot of the (002) diffraction peak is overlaid in Figure 5.2 below.

101

Figure 5.2: AlN overlaid nitrogen fraction coupled scans

From Figure 5.2, all the values of fN2 tested gave relatively good intensities as well as

FWHM values. However, no peak looks the most optimized as the 25%, 15%, and 10% all look similar. As such, one can also examine the AFM measurements in order to gain a better understanding of the film quality. The obtained surface morphologies can be seen in Figure 5.3 below.

102

5% 10%

15% 20%

25% 30%

35%

Figure 5.3: AlN fN2 AFM analysis

103

From these AFM images, it is possible to see that increasing the fN2 decreases the features seen on the surface of the film and, therefore, decreases the RMS roughness of these samples. This implies that the film quality increases directly with the fN2.

However, greater values of fN2 also present problems with DC deposition, as arcing can be observed at higher fN2.

The combination of both the XRD and AFM measurements allows one to draw conclusions on the best growth parameters. The AFM measurements suggest that films grown above 20% fN2 result in smoother surfaces and less large features. XRD coupled scans show that above 25% and above films result in fairly good crystal quality. The slight asymmetry while still maintaining a high peak intensity in the 25% fN2 growth could have resulted from the beginning of fringes or from strain in the film.

5.2 Substrate Temperature

As with any new material, the substrate temperature must also be optimized in order to obtain high quality film growth. In the case of AlN, the previous fN2 growths were performed using a substrate heater current of 40A, which corresponds to 760°C. It was anticipated that the growth of AlN might require higher substrate temperatures, so the heater current was varied from 40A - 50A (760°C - 885°C). The power and fN2 remained unchanged for these growths, as they were performed prior to the analysis of the previous films. The coupled scans obtained from these films can be seen in Figure 5.4 below.

104

Figure 5.4: AlN substrate temperature stacked coupled scans

Unlike with the ScN samples, Pendellösung fringes were not forthcoming in the second step of optimization. As such, again the peak intensity and FWHM were used as the indication of highest quality growth for these films. In order to compare, the AlN peak is again overlaid in Figure 5.5 below.

105

Figure 5.5: AlN overlaid substrate temperature optimization

Once again, no peaks particularly outperform the others. As such, AFM measurements can be used in tandem with the XRD coupled scans in order to quantify the quality of the film. The surface morphology results from the AFM can be seen in

Figure 5.6 below.

106

760°C 810°C

840°C 860°C

885°C

Figure 5.6: AlN substrate temperature series AFM

107

As can be seen in Figure 5.6, the increase in substrate temperature reduces the surface roughness of the films. This is desirable for incorporation of these films into devices.

However, the XRD FWHM does not decrease with increasing temperature, but the asymmetry in the coupled scans is still present for the higher temperature films. In addition, the 885°C film has a higher intensity. This implies that more planes are diffracting in the higher temperature films, provided that all of the films are nominally the same thickness. As such, it is likely that growing at higher temperatures provides better quality growth.

5.3 Coil Current

In addition to optimizing substrate temperature and fN2 for the growth of AlN, one potential way to grow high quality material is to adjust the Ji/JMe ratio. This is done by changing the current in the external magnetic coils. This can help to provide energies to the incoming atomic species that are favorable for growth. Adjusting the coil current while holding the other parameters constant at 100W, 25% fN2, and IH = 40A allows understanding of the effect of the coil current. As characterization of the previous films had not finished, the optimized parameters from the previous studies are not utilized.

Once again, these parameters were varied and the resulting films were examined using XRD symmetric coupled scans and AFM surface analysis. The obtained coupled scans can be seen in Figure 5.7 below.

108

Figure 5.7: AlN stacked external coil current coupled scans

Again, no fringes are observed in this optimization. As such, it is best to optimize to the film peak intensity and FWHM. To this end, the AlN peak overlay can be seen in

Figure 5.8 below.

109

Figure 5.8: AlN overlaid coil current coupled scans

From this, one can see that the changes of the external magnetic coil current only seems to change the overall intensity of the coupled scan peak. The resulting film peaks are still slightly asymmetric but no major difference in film peak intensity or FWHM is noted, except for the coil current at 8A. However, film surface morphology also play a role in selecting an optimized condition. These can be seen in the obtained AFM scans in

Figure 5.9 below.

110

0A 4A

6A 8A

9A 10A

Figure 5.9: AlN coil current series AFM measurements

111

These AFM images show that all coil currents examined in this series still grow in the island growth mode (i.e. three dimensionally). Additionally, all of the AFM images have large features of this growth mode. The RMS roughness of each film is between 4.2nm and 8.1nm. However, examination of the XRD coupled scans and the AFM surface morphology suggests that growth between 8A and 10A provides the most crystalline samples. Tuning the additional growth parameters to the more optimized values may provide better quality films.

5.4 Magnetron Power

The last potential growth parameter that can be adjusted in the unbalanced magnetron sputtering system is the actual magnetron sputter power. This changes the energies of the incoming atomic species. A set of films were deposited at 25% fN2, IH = 40A, and IC =

8A. The deposition powers were adjusted from 50W to 200W in order to examine the role of the magnetron sputter power. These films were also studied by XRD coupled scans and tapping mode AFM. The resulting coupled scans for this set of data can be seen below in Figure 5.10.

112

Figure 5.10: AlN power series stacked coupled scans

One can clearly see that these peaks both change shape and intensity as the deposition power varies. However, a clearer image of this change is presented in Figure 5.11 below.

Figure 5.11: AlN power series overlaid coupled scans

113

One challenge when comparing each of these coupled scans arises from the lack of

Pendellösung fringes. This challenge is the fact that the film thickness is unknown.

Thus, the deposition rate is unknown. As such, one cannot simply compare the peak intensity values in this set of scans. This is because the deposition time between the films is the same. However, varying the deposition power ensures that the film thickness is not the same value. As such, the shape of the coupled scan peak and the AFM surface provides the most information for optimizing these films. Note that, once again, the

100W sample is slightly asymmetric, and has a high intensity. However, both the 150W and the 200W also have high intensities, with a lower FWHM value.

Once again, the AFM surface analysis must be used in order to better understand the most optimized conditions for film growth. The obtained AFM images for this set of films is available in Figure 5.12 below.

114

50W 100W

200W 150W

Figure 5.12: AlN power series AFM

The film surfaces in this optimization set varies dramatically from film to film. The

50W and 100W have relatively large features across the surface. In addition, the 200W has features of many different sizes. Note that for this film, the scale bar is an order of magnitude larger than the other films. However, the 150W film has features that are almost reminiscent of step flow growth. This is a puzzling result, as often films with these features will at least have a symmetric coupled scan peak that includes at least some fringes. Often, these fringes are on one side of the peak that is highly asymmetric, but they should still be there.

115

The AlN growth rates had previously been assumed to be comparable to the TiN and

ScN growth rates such that 20 minute depositions would result in films on the order of

50nm thick. Film of this thickness should have shown indications of Pendellösung fringes if the growth quality allows. The fact that none of the optimization sets observed these fringes was concerning. In order to understand the reasons behind this, the UV-Vis

SE was used to examine the optical properties of the AlN layer and try to fit the thickness of the film layer. A Cauchy dispersion model was used, as the data was taken in the non- absorbing region for the AlN layer. A simple fit of this gave a thickness on the order of

500nm. This explains the lace of fringes in every optimization set.

5.5 Final Optimization

Once the film thickness was understood, it became necessary to use the information gained in the optimization sets and attempt to grow significantly thinner films. The growths resulting in films on the order of 500nm were 20 minutes long. As such, growths resulting in thicknesses on the order of 50nm were performed for approximately

2 minutes long.

As the parameter that seem to affect the surface morphology the most, the sputter power was first varied for these films. The power was varied from 50W-200W, again, with deposition times varying from 8 minutes to 1.25 minutes, in an attempt to have similar thicknesses in these films. The remaining deposition parameters were: Ic = 6A, fN2 = 20%, and IH = 40A (840°C). The symmetric coupled XRD scans for these films can be seen in Figure 5.13 below.

116

Figure 5.13: AlN thin film power series coupled scans

Each of these films result in some features that resemble Pendellösung fringes.

However, the 200W sample is the only film that has fringes on both sides of the symmetric peak. This implies high quality crystal growth at this deposition power. The surface morphology of each of these films as obtained by AFM can be seen in Figure

5.14 below.

117

50W 100W

150W 200W

Figure 5.14: Thin AlN power series surface morphology

After examining the different surface morphologies, one notices that the large features seen in the 200W and 150W growths seem to decrease in frequency as moving toward the lower power growths. This could indicate that the higher power growths facilitate coalescence by increasing the energy of the incoming atomic species. However, these morphologies do not show any step flow growth. This indicates that additional optimization of these thinner films is necessary in order to obtain the highest quality film growth possible. However, RMS roughness values on the order of 500-800pm is still

118 relatively smooth, indicating at least high quality growth despite the films not being single crystal.

However, in order to fully optimize the films, one must also look at the desired properties, being the piezoelectric response. Maximizing this parameter is the desired outcome and, therefore, it should be included in the optimization of the films.

5.6 Piezoelectric Properties

After finding growth conditions that result in acceptable films, the piezoelectric coefficient of the AlN film must be tested in order to ensure high quality film growth.

Additionally, the AlN examined in this work is a baseline for the Al1-xScxN films that are planned. As such, understanding the AlN piezoelectric coefficient for films grown in this system allows for understanding the improvement that is obtained by alloying the AlN with Sc.

5.6.1 TiN on Al2O3

One potential way to obtain the piezoelectric coefficient of a film is through the use of a conductive tipped AFM measurement known as piezoelectric force microscopy. By applying a voltage across the film using a conductive AFM tip, it is possible to measure the deflection in the AFM tip from the expansion of the film. This allows one to measure the piezoelectric coefficient directly. One difficulty in this measurement arises due to the necessity of a conductive layer beneath the AlN.

Previously, the AlN layers were grown on non-conductive Al2O3. As such, additional films were grown on TiN (111) on Al2O3. The growth of TiN was optimized prior to this work for deposition on c-plane Al2O3. As such, only optimized TiN was used in this set.

119

However, the AlN had been optimized for growth on Al2O3. In order to potentially obtain the best film properties, multiple films of AlN/TiN/Al2O3 were grown.

The TiN typically grown in this system benefits from extensive work and a relative ease of deposition. These films typically have multiple fringes and extremely smooth surfaces. As mentioned prior, TiN is a rocksalt TMN material with a lattice constant of around 4.23Å. The in-plane lattice constant for the TiN is 5.982 Å which results in a mismatch of about -25% with Al2O3. Once again, a 30° in-plane rotation can be used to mitigate some of this mismatch. The resulting mismatch is about -8%. As such, the TiN films grow fairly well on Al2O3. An example of a typical symmetric coupled scan for standard TiN (IH = 30A, IC = 4A, Wt = 100W, and fN2 = 50%) can be seen in Figure 5.15 below.

Figure 5.15: TiN on Al2O3 example

120

In addition to having a suitable crystal structure for AlN nucleation, the optimized

TiN has extremely low RMS roughness values. One example of a TiN surface can be seen in Figure 5.16 below.

Figure 5.16: TiN surface morphology

The RMS roughness of this surface is on the order of 860pm, which is extremely smooth. This should allow for high quality AlN film growth on the TiN surface. The additional triangular shapes seen in this image are attributed to un-coalesced nucleation points for the TiN. As such, they do not present a problem for regrowth.

5.6.2 AlN on TiN on Al2O3

As for the growth of AlN on TiN, it is important to understand the lattice mismatch between the two materials. AlN, as mentioned prior, has an in-plane lattice constant of about 3.113Å. This corresponds to a lattice mismatch to the TiN of about 4% if one takes into account the potential of domain matching two AlN unit cells to the single TiN unit

121 cell. As such, it should be possible to grow AlN layers on the TiN layer. Additionally,

TiN layers are extremely smooth, so the AlN should nucleate well on TiN templates.

However, the AlN layers were optimized for growth on Al2O3 and the surface free energies should differ between the Al2O3 and the TiN. As such, it is necessary to reexamine some deposition parameters in order to obtain good quality growth of AlN on

TiN. As such, the samples grown on TiN examined a set of deposition powers (50W to

200W) with other parameters of: IH = 45A, fN2 = 20%, and IC = 6A. Additionally, two growths examined the effect of fN2 by growing two films: one at 15% and the other at

20%. The additional deposition parameters in these films were: IH = 45A, IC = 6A, and

Wt = 200W. In all of these films, the TiN layer was grown to achieve a 30nm film whereas the AlN films was grown to achieve about 200nm.

In order to ensure high quality materials for piezoelectric testing, the films are examined with AFM and XRD measurements. The films examining the deposition power can be seen in Figure 5.17 below.

Figure 5.17: AlN/TiN/Al2O3 power series stacked coupled scans

122

The resulting film stacks show single film peaks for both the TiN and the AlN. The

Pendellösung fringes observed correspond to the TiN film layer. Due to the thickness of the AlN layers, fringes are not expected for the AlN layers. However, once again, AFM analysis can provide insight into the quality of each layer. As such, the films were also examined using AFM measurements, which can be seen in Figure 5.18 below.

50W 100W

150W 200W

Figure 5.18: AlN/TiN stacks power series surface morphology

In these 5µm x 5µm scans, one notices that not only does the film surface roughness decrease with increasing deposition power, but the actual topography changes. The films grown at 50W show significant island growth and large ball-like features on the surface.

123

However, the films at 200W seem to have a significantly smoother surface with a reduction in large balled features.

An additional parameter that may affect the piezoelectric coefficient is the fN2 value.

The resulting XRD scans from the films deposited to examine this effect can be seen in

Figure 5.19 below.

Figure 5.19: AlN/TiN stacks fN2 coupled scans

124

15% 20%

Figure 5.20: AlN/TiN stacks fN2 surface morphology

Once again, the surface morphologies change when examining these two films. First, the RMS roughness reduces from about 12nm to about 10nm when increasing the nitrogen partial pressure. Additionally, the large island and craters seen in the 15% AFM images disappear when increasing the fN2 to 20%. However, the cliff like features seen in the 15% fN2 growth are reminiscent of steps. But this could also indicate columnar growth, which is sometimes common for AlN grown on Al2O3.

After obtaining films grown on the conductive TiN, it is possible to examine the piezoelectric coefficient of the AlN layers. These values can be seen in Figure 5.21 below for the power series samples.

125

Figure 5.21: AlN power series piezoelectric coefficients

Note that the piezoelectric coefficient for all of these films is in the neighborhood of

3.5 pm/V, which is about what is expected for AlN.7 However, the piezoelectric properties seem to indicate that that growing around 100W provides the best film for this application.

The additional set of films, examining the nitrogen partial pressure, also can be seen in Figure 5.22 below.

Figure 5.22: AlN fN2 piezoelectric coefficients

126

As can be seen in Figure 5.22, the piezoelectric coefficient seems to increase at decreasing fN2. Combining the two fN2 piezoelectric coefficients with the power series coefficients shows that the ideal film for the piezoelectric properties of AlN is still not optimized. This may be due to the fact that the previous optimization of AlN grown on

Al2O3 does not compare to growth of AlN on a TiN backside contact. However, despite the fact that additional optimization is needed for the AlN on TiN films, the actual AlN films show extreme promise. The obtained piezoelectric coefficient for these films ranged from 1.86 to 3.85 pm/V, which is acceptable for AlN growth that is still not optimized. It is expected that another optimization set for AlN on TiN would result in larger piezoelectric coefficients and, therefore, a better baseline for understanding the improvement in the films due to alloying with Sc.

127

CHAPTER 6

TITANIUM OXIDE GROWTH AND CHARACTERIZATION

For many optical applications, such as optical filters and distributed Bragg reflectors, one requires multilayered stacks of the materials of interest at specific thicknesses. This could simply be amorphous or polycrystalline material. However, in the case of VO2, the unique properties of the material require highly crystalline material. As such, it is necessary to find materials that allow for the growth of crystalline VO2 in a multilayered stack. This work included TiO2 as a potential candidate for regrowth of VO2.

The deposition of TiO2 takes place in the IAD deposition system. Growth parameters consist of a discharge current of 1A, discharge voltage of 300V, O2 gas flow at 5.7sccm, and a dose of 37.5 (µA/cm2)/(A/s). The beam current was 110mA with a power of 825W and pressure of 150µTorr. These deposition conditions allowed for growth of TiO2 on

Al2O3 (0001) substrates.

As mentioned prior, two stable phases of TiO2 exist: anatase and rutile. Regrowth of

VO2 requires the rutile phase of TiO2, as the crystal structures match well in the high temperature VO2 phase. As such, it was necessary to verify the phase of the TiO2 deposited by reactive sputtering. In order to do this, XRD coupled scans were performed on the deposited TiO2, which can be seen in Figure 6.1 below.

128

Figure 6.1: TIO2 on Al2O3 coupled scans

When examining Figure 6.1 it is necessary to understand that the two peaks around

42° and 91° correspond to the Al2O3 substrate. The additional points of high intensity around 21° and 94° degrees are not allowed diffraction planes for Al2O3. Impurities in the crystal substrate allow for low intensity diffraction at these positions. The two TiO2 peaks noted are at 38.5° and 82.4°. When comparing to known diffraction planes of

TiO2, these could correspond to the (020) and (040) planes of rutile TiO2 or the (112) and

(224) planes of anatase TiO2. However, it has been shown that rutile TiO2 grows well on

34 Al2O3 (0001) oriented substrates due to the low lattice mismatch. Additionally, anatase

11 TiO2 is a metastable state that often grows in highly polycrystalline orientations. As such, it is highly unlikely that the films grown are anatase TiO2.

129

In addition to obtaining the necessary crystal structure, one must also understand the optical properties of the TiO2 in order to incorporate it in a multilayered optical stack. As such, the TiO2 was examined through SE.

Figure 6.2: TiO2 optical properties

The TiO2 optical properties show typical behavior including low loss in the visible spectrum and out into the near IR. This shows that TiO2 is a promising candidate for use with the VO2 films because only the VO2 films will contribute to loss in the multilayered stacks. As such, it is necessary to examine growth of VO2 on TiO2 in order to determine if TiO2 could allow for incorporation of multilayered stacks of VO2.

130

CHAPTER 7

VANADIUM OXIDE GROWTH AND CHARACTERIZATION

All of the work on TiO2 aims to better incorporate VO2 as a material for optical devices. This material is typically considered due to its structural phase transition that is accompanied by dramatic changes in the resistivity, as outlined in 1.2.1. But, conventionally this material is not incorporated due to its difficulty to deposit due to deposition conditions and lattice matched substrates. As such, this work examines both the potential for incorporation of VO2 films on TiO2 as well as a novel approach to depositing VO2 in order to mitigate some of the challenges in the deposition of this material.

In most of the samples grown in this work, crystallinity of the film is paramount for the application. Although the crystallinity of the VO2 films is still important, it is not the most important property. In fact, the VO2 is not expected to be single crystal, but rather a highly oriented polycrystalline film. This is due to the stress that builds up in the crystal due to the structural phase transition. As such, the most important property for the VO2 films is the switching characteristics, i.e. how much the resistivity switches between the high and low temperature VO2 films. In addition, as this material is examined for optical applications, it is also necessary to have the optical properties of these films once the resistivity switching properties are optimized.

7.1 Deposition of VO2 on Al2O3

The deposition of VO2 on Al2O3 (0001) orientation had already been optimized for

IAD in order to obtain the highest change in resistivity due to the metal-insulator

131 transition of VO2. In order to obtain high quality VO2, the sample was deposited at

600°C in an O2 environment. The background pressure in the system was between 120 and 140µTorr for the films examined. The IAD parameters were: discharge current at

2 1A, discharge voltage at 300V, O2 flow at 6sccm, and a dose at 50(µA/cm )/(A/s). These films were deposited in order to obtain 50nm of material according to the QCM. A standard resistivity plot for these materials can be seen in Figure 7.1 below.

Figure 7.1: Resistivity switiching of VO2 on Al2O3

The major metric for understanding the quality VO2 film growth is the orders of magnitude change in the resistivity values. In this example, the resistivity changes by about 2.5 orders of magnitude after transitioning from the insulating phase to the metallic phase. This is not a terrible value for transition, but it is certainly not the best seen values. However, it could be enough of a change in order to use this film in optical devices.

132

Examination of the film optical properties allows for one to understand the applicability of these films in optical devices. SE provides a useful way of obtaining film optical properties over a wide range of wavelengths. In addition, the SE is equipped with a heater stage that allows for heating a sample up to about 300°C. As such, it is possible to obtain both the room temperature (25°C) and high temperature (80°C) optical properties of the film. These can be seen in Figure 7.2 and Figure 7.3 below.

Figure 7.2: Room temperature VO2 optical properties

133

Figure 7.3: High temperature (80°C) VO2 optical properties

Although some changes can be seen to n, the major difference in heating the sample appears in the increasing value of the absorption coefficient, where it nearly doubles at longer wavelength values. This shows that the switching value of the VO2 dramatically changes the absorption of the light in the film layer. This is the major principle for using

VO2 as an optical limiter.

However, the use of VO2 in optical devices is still dependent on obtaining multilayered stacks of this material as many different optical applications call for this capability. Deposition of crystalline Al2O3 is not only extremely expensive, but also very difficult. As such, Al2O3 is not an acceptable material for growing multilayered stacks of

VO2. Therefore, TiO2 is examined as a potential growth layer as it shares the crystal structure of VO2 in its high temperature state.

7.2 Deposition of VO2on TiO2

After examining the depositions of VO2 on an Al2O3 (0001) sample, a baseline is established. If the VO2 deposited on TiO2 has a similar resistivity change, the different

134 material used as the growth template does not drastically effect the relevant properties of the VO2. Additionally, it is possible to get rutile TiO2 on Al2O3 (0001). As both TiO2 an

VO2 (high temperature) share a similar crystal structure, the growth of the VO2 on TiO2 should succeed.

The growth of the TiO2 on Al2O3 (0001) follows the parameters outlined in Chapter 6

Titanium Oxide Growth and Characterization. This way the growth of the TiO2 results in rutile (020) oriented TiO2 films. A subsequent growth of VO2 using the same parameters as outlined in 7.1 allows obtains a VO2 film on the TiO2/Al2O3 heterostructure. 4-point probe measurements allows one to examine the quality of the VO2 film on TiO2 through examination of the resistivity curve. The resistivity curve for the VO2/TiO2/Al2O3 sample can be seen in Figure 7.4 below.

Figure 7.4: VO2/TiO2/Al2O3 resistivity curve

As can be seen by this resistivity curve, the VO2 still successfully switches after deposition on TiO2. Additionally, the switching nearly achieves around 2.5 orders of magnitude, which makes it comparable to the VO2 deposited on bare Al2O3. For a more

135 apparent comparison, one can see both the VO2/Al2O3 (sample 1490) and

VO2/TiO2/Al2O3 (sample 1493) resistivity curves in Figure 7.5 below.

Figure 7.5: Comparison of VO2 films grown on Al2O3 and TiO2

As can be seen in Figure 7.5, the resistivity curves of each of the VO2 samples are very similar. The growth on TiO2, does have a slightly higher resistivity in both the low temperature and high temperature state, but this should not significantly affect device functionality.

The change in the transition temperature, however, could have a more noticeable effect if one expects to operate these films at a value close to the transition temperature.

However, if one picks a value that is far enough away from the transition temperature

(80°C for example) the films should act very similarly. This suggests that multilayered stacks of VO2 on TiO2 could succeed. As such, further growth is required to see if increasing the number of layers leads to film delamination or strain build up.

136

7.3 VO2 Films by Thermal Oxidation

The previous mentioned depositions utilize the Ion Assisted Deposition system in order to grow VO2 films. However, traditional VO2 films are generally difficult to pattern and etch and the process by which to obtain VO2 films through reactive deposition is rather difficult. As such, it is desirable to obtain a process where it is possible to easily make VO2 patterned structures. This allows for creating switchable optical systems that are more complicated than simply switchable films. One example is switchable wire grid polarizers.

Whereas VO2 is difficult to etch, vanadium itself is relatively easy to etch. As such, if one simply deposits V and then later anneal the system in an oxygen environment to crystalize the V into VO2, an etching step between deposition and crystallization allows for patterned VO2. In addition to the ease of etching, sputter deposition of V is significantly easier to control and perform as compared to IAD of VO2. A step of oxidation a tube furnace allows for wafer scale thermal annealing of the V samples. Thus providing an easier deposition method as well as an easier processing method. The only downside to this process is the additional step introduced by including the thermal oxidation. However, the ease of sputtering the films rather than using IAD more than makes up for the additional time introduced.

For this process, a series of V films were deposited on Al2O3 substrates using the

Denton magnetron sputtering system. All the films were deposited to obtain about 50nm of vanadium and then annealed. The annealing step took place in an MTI OTF-1200X tube furnace that was evacuated using a turbo pump. The oxygen gas pressure, temperature, and duration at maximum temperature were varied in order to obtain

137 material with large resistance switching. The oxygen gas pressure was varied from 5 to

20mTorr. The maximum temperature was varied from 450°C to 550°C. The time at maximum temperature varied from 1hr to 4hrs. Each of these films underwent 4-point probe resistivity measurements in order to understand the switching characteristics of the films.

Of the films examined, three samples were chosen for additional characterization through the use of XPS and XRD analysis. These films were selected due to drastic differences in the resistivity measurements. These films were thought to be over- oxidized (sample 122), under-oxidized (sample 113), and near optimum (sample 116).

The resistance switching of these three films can be seen in Figure 7.6 below.

Figure 7.6: Thermally annealed VO2 sheet resistance curves

138

When examining the change in resistivity of the samples, it is possible to note that samples 113 and 122 which are under-oxidized and over-oxidized, respectively, behave as expected. An under-oxidized film would act as a metallic layer and, therefore, have an extremely low resistivity. However, over oxidizing the film results in insulating V2O5 and V2O3 as compared to VO2. The film grown with near optimum conditions, 116, shows resistivity switching of about 3.5 orders of magnitude (from about 10-1 to 5*10-5), which is already comparable to those grown using the IAD method.

As VO2 undergoes a structural phase transition, it is necessary to understand the crystallinity of the VO2 films. In order to do this, GIXRD and symmetric coupled scans were performed on each of the films undergoing examination. The combination of the two provides insight into the polycrystalline nature of each film. In addition, higher quality VO2 is expected to match well with Al2O3 substrate and, as such, should diffract in a coupled scan.

The obtained GIXRD scans for the selected samples can be seen in Figure 7.7 below.

139

Figure 7.7: VO2 GIXRD scans

The GIXRD scans show that each of the samples are at least partially polycrystalline.

Sample 122, which is over oxidized, shows the most peaks. Some of these peaks correspond to V2O5, VO2, V2O3 and many other possible combinations of vanadium and oxygen. Samples 113 and 116 show significantly less peaks in the GIXRD. This could be due to higher degrees of crystallinity as well as the potential of partially amorphous films.

In addition to understanding the polycrystalline nature, one can understand the high degree of orientation through the use of symmetric coupled scans. The coupled scans for these films can be seen in Figure 7.8 below.

140

Figure 7.8: VO2 symmetric coupled scans

In the coupled scans, one can see three very different scans. Sample 113 has two diffraction peaks where one would expect VO2 to diffract, but both are extremely wide and asymmetric. This is common for samples that have some crystallinity as well as some amorphous material and a high degree of misorientation. This would be consistent with the idea that the sample is likely under-oxidized, leaving some metallic vanadium at the interface between the film and substrate. Additionally, 122 only has diffraction peaks in the positions expected for VO2, but when paired with the GIXRD it is well understood that some VO2 should be in the films in addition to many other phases of VxOy. Last, 116 has two additional peaks that appear due to other phases of VxOy, despite it being considered near optimal conditions. However, this is not unexpected due to the GIXRD scans, as some additional phases are present in those as well.

141

Although it is possible to identify some of the phases possible in the films using

XRD, this requires comparison to known crystals. Additionally, this is difficult due to the many different phases of VxOy. As such, in order to understand the chemical composition of these films, x-ray photoelectron spectroscopy was performed on the samples. Fitting the vanadium peak allows for understanding the oxidation states of the vanadium in the sample. The obtained high resolution XPS spectra of the vanadium peak can be seen in Figure 7.9 below.

113 116

122

Figure 7.9: XPS of V peak in VO2 films

142

The two major peaks that appear in the measured data correspond to vanadium 2p 3/2 and 2p 1/2. The fitting of the vanadium 2p 3/2 peak for all the samples was performed in casaXPS using asymmetrically modified Gaussian-Lorentzian peaks. This is to compensate for electronic shake-up-like events that tend to broaden metal peaks towards the high energy. Fitting was only performed on the V 2p 3/2 peak due to the exclusion of the oxygen peak in the spectra. When including the oxygen peak, the background function changes and allows for fitting between both the V 2p 3/2 and the V 2p 1/2 peaks.

Without this, the constant area ratio rules between the V 2p 3/2 and 2p 1/2 peaks grossly overfits the 2p 1/2 peak. However, it is possible to examine just the V 2p 3/2 peak.

The resulting fits show differing concentrations of V3+, V4+, and V5+ in these films.

These correspond to the oxidation states of V2O3, VO2 and V2O5, respectively. Using the area of the fits one can approximate the relative composition amounts of each present in the films. However, these results are only applicable at the surface of the sample. In order to understand the bulk properties, one could depth profile the sample. As mentioned prior, this results in a loss of the resolution between the different oxidation states. Regardless, it is at least possible to make some qualitative conclusions about the samples. First, the metal peak in sample 122 has a significantly lower intensity than the other two. This could result from oxidation around clumps of metal, thereby shielding the metal on the inside. Samples 113 and 122 both have inclusions of V2O3 on the surface, whereas 116 does not. This could arise due to decomposition of V2O5 into V2O3 during the annealing process. Sample 116 has nearly equal parts of VO2 and V2O5. The large resistivity switching that occurs in this film suggests that the V2O5 is mostly

143 concentrated to the surface of the film. Additionally, the lack of V2O3 in this film suggests more uniform oxidation, as no V2O5 decomposed.

Using the information gained from each of these samples, as well as the resistivity measurements for all the samples, it is possible to find an optimized VO2 film. This film was grown by thermally annealing an amorphous vanadium layer by thermal annealing.

The optimized annealing process involved heating the sample up to 500°C for 2 hours in

10mTorr of O2. The O2 gas remained flowing during the cool down process for this film.

This allowed the vanadium to fully to VO2 and not loose oxygen while cooling down from 500°C.

Once again, this film is examined using the 4-point probe measurement in order to obtain the temperature dependent resistivity characteristics. The resulting hysteresis curve can be seen in Figure 7.10 below.

Figure 7.10: Optimized thermally annealed VO2 resistance switching

144

This VO2 film shows a resistivity switching of nearly 4 orders of magnitude, as compared to the 3.5 noted by the previous near optimum film. A comparison of these two film resistivity curves can be seen in Figure 7.11 below

Figure 7.11: VO2 optimum and near optimum resistivity switching

Three major differences between these resistivity switching curves can be seen when examining the near-optimum (116) and optimum (135) films. First, the optimized film shows a larger degree of switching (4 orders of magnitude as compared to 3.5).

Additionally, the resistivity of sample 135 in the low temperature state is lower than that of sample 116, suggesting the potential for lower incorporations of V2O5 (or other insulating phases) in the film. Additionally, the hysteresis of the phase transition is significantly smaller in the optimized film, indicating higher quality material.

145

After examining this film through the 4-point probe measurements, XRD and XPS measurements were also performed in order to compare to the previously grown films and to explain the difference in resistivity curves as compared to the near-optimum film.

The coupled scan for this film can be seen in Figure 7.12 below.

Figure 7.12: Optimized thermally annealed VO2 coupled scan

As can be seen in Figure 7.12, the VO2 (020) peak appears clearly without any major asymmetry. Additionally, no other peaks appear in this image other than the VO2 (020),

VO2 (040), and the Al2O3 substrate peaks. This implies that, compared to the near- optimum conditions, this VO2 sample is most likely more pure. This can explain the differences seen in the resistivity switching between the two films.

Despite seeing clear differences between the films in XRD, XPS measurements also help to understand the chemical differences in the film makeup between the two samples.

The obtained XPS high resolution spectra can be seen in Figure 7.13 below.

146

Figure 7.13: Optimized thermally annealed VO2 XPS spectrum

In Figure 7.13 the XPS spectra of both the vanadium peaks and the oxygen peaks are included in the graph simply to allow for fitting of both the V 2p3/2 and the V 2p1/2 peaks, simultaneously. This is because the Shirley background does not function correctly if only the vanadium peak is examined. However, the differences in fitting with both the V 2p3/2 and V 2p1/2 as compared to just the V 2p3/2 peak are minimal. Once again, the V 4+ and V 5+ peaks are nearly equivalent in this film. As such, it is difficult to draw conclusions simply from the XPS spectra. However, the combination of XPS,

XRD, ad resistivity measurements suggest that the bulk material is mostly VO2 with some V2O5 on the surface of the film, which is not unexpected as V2O5 is a more stable compound than VO2.

In addition to understanding the crystal structure and chemical makeup of the VO2 film, it is paramount to understand the optical properties. This allows one to incorporate

147 the film into a multi-layered stack as well as understand how the properties of a patterned system changes with thermal switching of the material. Using the SE allows one to obtain these optical properties for the VO2 film.

The optical properties of the VO2 sample were examined both at low temperature

(25°C) and high temperature (80°C) in order to examine the effect of the phase transition.

The wavelength range of interest was around 400nm to 2.4µm, as most applications for

VO2 fall in the near IR range. The obtained optical properties of the VO2 can be seen in

Figure 7.14 and Figure 7.15 below.

Figure 7.14: VO2 25°C optical properties

148

Figure 7.15: VO2 80°C optical properties

When examining these two sets of optical properties, one immediately realizes that an increase in the absorption coefficient appears at the longer wavelength range. Using a

Drude term allows one to fit this particular phenomena, which is consistent with the increasing in free carriers due to the structural phase transition of the material. However, it is also important to note that the absorption coefficient for the 25°C sample is also fairly low in the long wavelength range. As such, it is possible to switch the VO2 from a material that would mostly transmit light in this range to a material that would mostly absorb light in this range, as is typically expected with the VO2 films. Additionally, this change is not a small value, but rather large as it switches from around 0.05 to about 2.9 at 2.4µm.

Through all of this characterization, one notes that not only is it possible to thermally oxidize vanadium films and achieve extremely high resistivity switching but also that these films correspond well to the VO2 films deposited by other methods. As such, the

149 fact that the thermal annealing method provides high quality, switchable VO2 in addition to the relative ease of operation and scalability of the technique suggests that this method could be utilized for wafer scale growth of VO2 films. This would mitigate many of the issues with the tight growth parameters of VO2 in typical vacuum deposition systems.

150

CONCLUSIONS AND FUTURE WORK

This work focuses on the growth and characterization of multiple TMN and TMO materials namely: ScN, AlN, TiO2, and VO2. Understanding the growth parameters, useful properties, and varying methods of deposition for these materials allows one to incorporate them into next generation electrical and optical devices. As such, it is paramount to understand how to grow these materials and the degree to which their useful properties are tunable via growth parameters and methods.

Through this work, multiple materials were successfully grown and characterized.

The desired films were obtained with the correct crystal structures and properties.

However, much work is still yet to be done. Incorporation of VO2 on TiO2 to create multilayered stacks of VO2 is still underway. Although possible to grow VO2 on rutile

TiO2, the question remains: how many stacks of these materials can be deposited prior to either loss of desired properties, degradation in crystal structure, or strain buildup in the film layers. Additionally, work of GaN growth on ScN is still underway. Optimization of growth parameters for this regrowth requires a significant amount of effort and time in order to improve growth quality and reduce the dislocation densities in the GaN film.

Last, examination of Al1-xScxN films requires growth and optimization of each differing alloy concentration in order to examine the increase in the piezoelectric constant of the

Al1-xScxN films as compared to AlN. Additionally, one must understand if the maximum use temperature is effected by the alloy concentration. As such, much work is still yet to be had on these materials.

151

REFERENCES

1. Guide to Using WVASE Spectroscopic Ellipsometry Data Acquisition and Analysis Software (J.A. Woollam Co., Inc., Lincoln, NE, USA, 2012).

2. https://www.nist.gov/pml/engineering-physics-division/popular-links/hall-effect/hall- effect.

3. https://www.cameca.com/products/sims/technique.

4. https://www.eag.com/resources/tutorials/sims-tutorial-theory/.

5. http://www.xpsfitting.com/search/label/Detection%20Limits.

6. http://www.xpsfitting.com/search/label/X-ray%20Sources.

7. Akiyama, M. et al. Enhancement of Piezoelectric Response in Scandium Aluminum Nitride Alloy Thin Films Prepared by Dual Reactive Cosputtering. Adv Mater 21, 593- 596 (2009).

8. Bancroft, G. M. et al. Toward a comprehensive understanding of solid-state core-level XPS linewidths: Experimental and theoretical studies on the Si 2p and O 1s linewidths in silicates. Phys. Rev. B 80, 075405 (2009).

9. Biesinger, M. C. et al. Resolving surface chemical states in XPS analysis of first row transition metals, oxides and hydroxides: Cr, Mn, Fe, Co and Ni. Applied Surface Science 257, 2717-2730 (2011).

10. Biesinger, M. C., Payne, B. P., Lau, L. W. M., Gerson, A. & Smart, R. S. C. X-ray photoelectron spectroscopic chemical state quantification of mixed nickel metal, oxide and hydroxide systems. Surf. Interface Anal. 41, 324-332 (2009).

11. Brandt, I. S. et al. Influence of substrate on the structure of predominantly anatase TiO2 films grown by reactive sputtering. RSC Adv. 8, 7062-7071 (2018).

12. Briggs, J. A. et al. Fully CMOS-compatible titanium nitride nanoantennas. Appl. Phys. Lett. 108, 051110 (2016).

13. Budai, J. D. et al. Metallization of vanadium dioxide driven by large phonon entropy. Nature 515, 535 (2014).

14. Burmistrova, P. V. et al. Thermoelectric properties of epitaxial ScN films deposited by reactive magnetron sputtering onto MgO(001) substrates. J. Appl. Phys. 113, 153704 (2013).

152

15. Catellani, A. & Calzolari, A. Plasmonic properties of refractory titanium nitride. Phys. Rev. B 95, 115145 (2017).

16. Cetnar, J. S. et al. Electronic transport in degenerate (100) scandium nitride thin films on magnesium oxide substrates. Appl. Phys. Lett. 113, 192104 (2018).

17. Chen, S. et al. A novel structural VO2 micro-optical switch. Opt. Quant. Electron. 35, 1351-1355 (2003).

18. Chierchia, R. et al. Microstructure of heteroepitaxial GaN revealed by x-ray diffraction. J. Appl. Phys. 93, 8918-8925 (2003).

19. Chiu, T., Tonooka, K. & Kikuchi, N. Influence of oxygen pressure on the structural, electrical and optical properties of VO2 thin films deposited on ZnO/glass substrates by pulsed laser deposition. Thin Solid Films 518, 7441-7444 (2010).

20. Depla, D., Mahieu, S. & Greene, J. E. in Handbook of Deposition Technologies for Films and Coatings (Third Edition) (ed Martin, P. M.) 253-296 (William Andrew Publishing, Boston, 2010).

21. Diebold, U. The surface science of titanium dioxide. Surface Science Reports 48, 53- 229 (2003).

22. Fewster, P. F. & Andrew, N. L. Strain analysis by X-ray diffraction. Thin Solid Films 319, 1-8 (1998).

23. Fu, Y. Q. et al. Advances in piezoelectric thin films for acoustic biosensors, acoustofluidics and lab-on-chip applications. Progress in Materials Science 89, 31-91 (2017).

24. Gall, D. et al. Electronic structure of ScN determined using optical spectroscopy, photoemission, and ab initio calculations. Phys. Rev. B 63, 125119 (2001).

25. Goodenough, J. B. The two components of the crystallographic transition in VO2. Journal of Solid State Chemistry 3, 490-500 (1971).

26. Goodenough, J. B. Direct Cation- -Cation Interactions in Several Oxides. Phys. Rev. 117, 1442-1451 (1960).

27. Hall, E. H. On a New Action of the Magnet on Electric Currents. American Journal of Mathematics 2, 287-292 (1879).

28. Heinke, H., Kirchner, V., Einfeldt, S. & Hommel, D. X-ray diffraction analysis of the defect structure in epitaxial GaN. Appl. Phys. Lett. 77, 2145-2147 (2000).

153

29. Heying, B. et al. Role of threading dislocation structure on the x‐ray diffraction peak widths in epitaxial GaN films. Appl. Phys. Lett. 68, 643-645 (1996).

30. Ji, Y., Niklasson, G. A., Granqvist, C. G. & Boman, M. Thermochromic VO2 films by thermal oxidation of vanadium in SO2. Solar Energy Materials and Solar Cells 144, 713-716 (2016).

31. Kamohara, T., Akiyama, M., Ueno, N., Nonaka, K. & Kuwano, N. Influence of aluminum nitride interlayers on crystal orientation and piezoelectric property of aluminum nitride thin films prepared on titanium electrodes. Thin Solid Films 515, 4565- 4569 (2007).

32. Kana Kana, J. B., Ndjaka, J. M., Vignaud, G., Gibaud, A. & Maaza, M. Thermally tunable optical constants of vanadium dioxide thin films measured by spectroscopic ellipsometry. Optics Communications 284, 807-812 (2011).

33. Kappers, M. J. et al. Interlayer methods for reducing the dislocation density in gallium nitride. Physica B: Condensed Matter 401-402, 296-301 (2007).

34. Kazunori Fukushima and Gikan H Takaoka and,Isao Yamada. Epitaxial Growth of TiO 2 Rutile Thin Films on Sapphire Substrates by a Reactive Ionized Cluster Beam Method. Japanese Journal of Applied Physics 32, 3561 (1993).

35. Kerdsongpanya, S. et al. Anomalously high thermoelectric power factor in epitaxial ScN thin films. Appl. Phys. Lett. 99, 232113 (2011).

36. Kischkat, J. et al. Mid-infrared optical properties of thin films of aluminum oxide, titanium dioxide, silicon dioxide, aluminum nitride, and silicon nitride. Appl. Opt. 51, 6789-6798 (2012).

37. Kusano, E., Theil, J. A. & Thornton, J. A. Deposition of vanadium oxide films by direct current magnetron reactive sputtering. Journal of Vacuum Science & Technology A 6, 1663-1667 (1988).

38. Lee, M. in X-Ray diffraction for Materials Research: From Fundamentals to Applications (Apple Academic Press, Waretown, NJ, 2016).

39. Lee, S., Ivanov, I. N., Keum, J. K. & Lee, H. N. Epitaxial stabilization and phase instability of VO2 polymorphs. Scientific reports 6, 19621; 19621-19621 (2016).

40. Li, W. et al. Refractory Plasmonics with Titanium Nitride: Broadband Metamaterial Absorber. Adv Mater 26, 7959-7965 (2014).

41. Liu, N. X. et al. AlGaN layers grown on AlGaN buffer layer and GaN buffer layer using strain-relief interlayers Ser. 6841, 6715).

154

42. Lupina, L. et al. Zero lattice mismatch and twin-free single crystalline ScN buffer layers for GaN growth on silicon. Appl. Phys. Lett. 107, 201907 (2015).

43. Magnus, F. et al. Growth of TiO2 thin films on Si(001) and SiO2 by reactive high power impulse magnetron sputtering. MRS Proceedings 1352, mrss11-1352-gg10-26 (2011).

44. Mattox, D. M. in Handbook of Physical Vapor Deposition (PVD) Processing (Second Edition) (ed Mattox, D. M.) 301-331 (William Andrew Publishing, Boston, 2010).

45. McWhan, D. B., Marezio, M., Remeika, J. P. & Dernier, P. D. X-ray diffraction study of metallic VO2. Phys. Rev. B 10, 490-495 (1974).

46. Mendialdua, J., Casanova, R. & Barbaux, Y. XPS studies of V2O5, V6O13, VO2 and V2O3. Journal of Electron Spectroscopy and Related Phenomena 71, 249-261 (1995).

47. Moram, M. A., Zhang, Y., Kappers, M. J., Barber, Z. H. & Humphreys, C. J. Dislocation reduction in gallium nitride films using scandium nitride interlayers. Appl. Phys. Lett. 91, 152101 (2007).

48. Moram, M. A., Joyce, T. B., Chalker, P. R., Barber, Z. H. & Humphreys, C. J. Microstructure of epitaxial scandium nitride films grown on silicon. Applied Surface Science 252, 8385-8387 (2006).

49. Moram, M. A. et al. Growth of dislocation-free GaN islands on Si(111) using a scandium nitride buffer layer. Journal of Crystal Growth 308, 302-308 (2007).

50. https://serc.carleton.edu/research_education/geochemsheets/techniques/SIMS.html.

51. Muraoka, Y. & Hiroi, Z. Metal insulator transition of VO2 thin films grown on TiO2 (001) and (110) substrates. Appl. Phys. Lett. 80, 583-585 (2002).

52. Narayan, J. & Larson, B. C. Domain epitaxy: A unified paradigm for thin film growth. J. Appl. Phys. 93, 278-285 (2003).

53. Ohring, M. in Materials Science of Thin Films (Second Edition) (ed Ohring, M.) 145- 202 (Academic Press, San Diego, 2002).

54. Ohring, M. in Materials Science of Thin Films (Second Edition) (ed Ohring, M.) 203- 275 (Academic Press, San Diego, 2002).

55. Olaya, J. J., Huerta, L., Rodil, S. E. & Escamilla, R. Superconducting niobium nitride films deposited by unbalanced magnetron sputtering. Thin Solid Films 516, 8768-8773 (2008).

155

56. Paone, A., Sanjines, R., Jeanneret, P. & Schüler, A. Temperature-dependent multiangle FTIR NIR–MIR ellipsometry of thermochromic VO2 and V1−xWxO2 films. Solar Energy 118, 107-116 (2015).

57. Petrov, I., Adibi, F., Greene, J. E., Hultman, L. & Sundgren, J. ‐. Average energy deposited per atom: A universal parameter for describing ion‐assisted film growth? Appl. Phys. Lett. 63, 36-38 (1993).

58. Petrov, I., Adibi, F., Greene, J. E., Sproul, W. D. & Münz, W. ‐. 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 10, 3283-3287 (1992).

59. Saha, B. et al. Enhanced hardness in epitaxial TiAlScN alloy thin films and rocksalt TiN/(Al,Sc)N superlattices. Appl. Phys. Lett. 105, 151904 (2014).

60. Saha, B. et al. Electronic and optical properties of ScN and (Sc,Mn)N thin films deposited by reactive DC-magnetron sputtering. J. Appl. Phys. 114, 063519 (2013).

61. Saha, B. et al. Temperature-dependent thermal and thermoelectric properties of n - type and p -type Sc1−xMgxN. Phys. Rev. B 97, 085301 (2018).

62. Sana, P. et al. Growth of ScN(111) on Sc2O3(111) for GaN integration on Si(111): Experiment and ab-initio calculations. J. Appl. Phys. 120, 135103 (2016).

63. Sarangan, A. in Fundamentals and Applications of Nanophotonics (ed Haus, J. W.) 149-184 (Woodhead Publishing, 2016).

64. Sato, K., Hoshino, H., Mian, M. S. & Okimura, K. Low-temperature growth of VO2 films on transparent ZnO/glass and Al-doped ZnO/glass and their optical transition properties. Thin Solid Films 651, 91-96 (2018).

65. Schroder, D. K. in Semiconductor Material and Device Characterization, Third Edition (John Wiley & Sons, Inc., Hoboken, New Jersey, 2006).

66. Shah, S. I., Jaffari, G. H., Yassitepe, E. & Ali, B. in Handbook of Deposition Technologies for Films and Coatings (Third Edition) (ed Martin, P. M.) 135-252 (William Andrew Publishing, Boston, 2010).

67. Shard, A. G. Detection limits in XPS for more than 6000 binary systems using Al and Mg Kα X-rays. Surf. Interface Anal. 46, 175-185 (2014).

68. Smith, H. A. et al. Epitaxial titanium nitride on sapphire: Effects of substrate temperature on microstructure and optical properties. Journal of Vacuum Science & Technology A 36, 03E107 (2018).

156

69. Speakman, S. A. Basics of X-Ray Powder Diffraction.

70. Stadler, B. J. H. Chapter 7 - Vapor Processes. , 513-588 (2016).

71. Strite, S. & Morkoç, H. GaN, AlN, and InN: A review. Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures Processing, Measurement, and Phenomena 10, 1237-1266 (1992).

72. Tanner, B. K. & Halliwell, M. A. G. Interference structures in double-crystal X-ray rocking curves from very thin multiple epitaxial layers. Semiconductor Science and Technology 3, 967-972 (1988).

73. Turner, R. C., Fuierer, P. A., Newnham, R. E. & Shrout, T. R. Materials for high temperature acoustic and vibration sensors: A review. Applied Acoustics 41, 299-324 (1994).

74. Ureña-Begara, F., Crunteanu, A. & Raskin, J. Raman and XPS characterization of vanadium oxide thin films with temperature. Applied Surface Science 403, 717-727 (2017).

75. Yang, M. et al. Resistance switching of epitaxial VO2/Al2O3 heterostructure at room temperature induced by organic liquids. AIP Advances 5, 037114 (2015).

76. Zakomirnyi, V. I. et al. Refractory titanium nitride two-dimensional structures with extremely narrow surface lattice resonances at telecommunication wavelengths. Appl. Phys. Lett. 111, 123107 (2017).

77. Zebarjadi, M. et al. Thermoelectric Transport in a ZrN/ScN Superlattice. J Electron Mater 38, 960-963 (2009).

78. Zukauskaite, A. et al. Microstructure and dielectric properties of piezoelectric magnetron sputtered w-ScxAl1âˆ’xN thin films. J. Appl. Phys. 111, 093527 (2012).

79. Zywitzki, O., Modes, T., Barth, S., Bartzsch, H. & Frach, P. Effect of scandium content on structure and piezoelectric properties of AlScN films deposited by reactive pulse magnetron sputtering. Surface and Coatings Technology 309, 417-422 (2017).

157