ABSTRACT MCGREGOR, JR., DAVID ROSS. Growth Optimization and Characterization of Reactively Sputtered Zirconium Nitride Thin Films for III-V Buffer Layer Applications (Under the direction of Jerome J. Cuomo)

Zirconium nitride (ZrN) thin films were deposited by reactive dc magnetron sputtering to assess the effects of processing conditions upon film properties. Processing conditions and parameters were optimized to generate films of completely oriented (111)

ZrN on silicon to be used as buffer layers for the growth of gallium nitride A single and double Langmuir probe were used to determine trends in electron temperature, ion density, ionization fraction, and floating potential during reactive sputtering of zirconium in argon and nitrogen. Reactive gas concentration, deposition pressure, deposition temperature, cathode current, film thickness and substrate orientation were investigated as variable processing conditions. Four-point probe, scanning electron microscopy

(SEM), transmission electron microscopy (TEM), Raman spectroscopy, and x-ray diffraction (XRD) were used to characterize thin films produced.

The optimum growth conditions for the (111) oriented growth of ZrN, for this work,

were found to occur during reactive magnetron sputtering at a deposition temperature of

500°C, a constant cathode current of 0.5 ampere, a deposition pressure of 15 mTorr, a

reactive nitrogen gas concentration of 4% in argon, deposited on (111) oriented silicon,

with a thickness on the order of 600 nanometers. Gallium nitride was then deposited on

films of ZrN to assess the crystallinity of films produced. The lattice mismatch between

(111) oriented ZrN and c-axis oriented GaN was calculated at 1.6%. Microscopic

evaluation showed the films to be of columnar structure with dense grains and smooth

surfaces. A change in preferred orientation was noticed as a function of increasing film thickness and cathode current and was determined to be due to an increase in ion channeling and bombardment energy.

for Mer

“Chance favors the prepared mind.” -Louis Pasteur

ii BIOGRAPHY

DAVID ROSS MCGREGOR JR. was born July 27, 1976 in Greenville, North Carolina. He received his Bachelor of Science in Materials Science and Engineering from North Carolina State University in 1999. He continued his education, attending North Carolina State University in the Department of Materials Science and Engineering in August 1999. He was appointed a Research Assistantship in January of 2000 at the Center for Advanced Manufacturing Processes and Materials (CAMP-M), under the direction of Distinguished University Research Professor, Dr. Jerome J. Cuomo. He received a Masters of Science degree in Materials Science and Engineering from North Carolina State University in October 2002. He has accepted a position with CREE, Inc. in Raleigh, NC.

iii ACKNOWLEDGEMENTS I would first like to thank Dr. Jerome J. Cuomo for his continued support, guidance, and energy throughout the course of this research. The experience and knowledge I have gained at the Center for Advanced Manufacturing Processes and Materials (CAMP-M) has taught me to question all, hunt the truth, and thirst for more. I thank Dr. Mohamad Bourham for his countless hours of support and plasma science instruction. I thank Dr. D. Maher for his support and guidance through an undergraduate and graduate career. I would especially like to thank Mark Williams, Dr. Jung Won Cho, and Dr. Minseo Park for there invaluable time, teachings, and insights in plasma processing and characterization. I wish to express my sincere gratitude to my better half, Merri, for her overwhelming support, kindness and a strong willingness to put up with me. I would also like to thank my parents, Dave and Ruth, for there unwavering encouragement and support. Lastly, I would like to thank all past and present members of CAMP-M and NCSU, in particularly, Scott Camphausen, Robert Trussell, Peter Yancey, Brian Laughlin, Spalding Craft, Johnathon Melton, Dr. Dan Johnson, Scott Dillon, and Brent Thomas for their help and support around the laboratory.

iv Table of contents

Page

List of Tables ...... viii

List of Figures...... ix

1.0 INTRODUCTION...... 1

1.1 IV-B Transition Metal Nitrides ...... 2 1.2 The Sputtering Phenomena...... 3 1.2.1 Plasma Generation ...... 3 1.2.2 Ion Bombardment...... 8 1.2.3 Atomic Transport ...... 11 1.2.4 Nucleation and Growth...... 12 1.3 Practical Aspects of Sputtering...... 15 1.3.1 Epitaxial Growth...... 15 1.3.2 Reactive Sputtering ...... 16 1.3.3 Radio-frequency Sputtering ...... 17 1.3.4 Sputter Yield...... 18 1.4 References ...... 19

2.0 EXPERIMENTAL STRUCTURE ...... 23

2.1 Dual Magnetron...... 23 2.2 Self Sputter Magnetron...... 25 2.3 Magnetrons ...... 29 2.3.1 Balanced versus Unbalanced Magnetron Sputtering ...... 31 2.3.2 DC Magnetron Sputtering...... 32 2.3.3 Pulsed DC Sputtering ...... 33 2.4 Plasma Characterization...... 34 2.4.1 Pressure Effects on I-V Characteristics...... 34 2.4.2 Langmuir Probe Characterization...... 36

v 2.4.2.1 Single Probe...... 37 2.4.2.2 Double Langmuir Probe ...... 41 2.5 Thin Film Characterization...... 44 2.5.1 Raman Spectroscopy...... 45 2.6 Gallium Nitride...... 47 2.6.1 Lattice and Thermal Expansion Mismatch ...... 48 2.7 References ...... 50

3.0 ZIRCONIUM NITRIDE...... 54

3.1 Properties of IV-B Transition Metal Nitrides...... 54 3.2 Properties of Higher Zirconium Nitride ...... 55 3.3 Deposition of Zirconium Nitride ...... 56 3.4 GaN growth on ZrN...... 57 3.5 References ...... 59

4.0 OPTIMIZATION OF REACTIVE MAGNETRON SPUTTERING OF ZIRCONIUM NITRIDE ...... 61

4.1 DC Reactive Magnetron Sputtering of Zirconium Nitride and Gallium Nitride Thin Films...... 61 4.2 Processing Condition Effects upon the Resistivity of Zirconium Nitride Thin Films ...... 62 4.2.1 Effect of Reactive Gas Concentration upon the Resistivity of ZrN Thin Films ...... 62 4.2.2 Effect of Deposition Pressure upon the Resistivity of ZrN Thin Films...... 63 4.2.3 Effect of Cathode Current upon the Resistivity of ZrN Thin Films...... 64 4.2.4 Effect of Deposition Temperature upon the Resistivity of ZrN Thin Films... 65 4.3 Processing Condition Effects Upon the Crystallinity of Zirconium Nitride Thin Films ...... 66 4.3.1 Effect of Deposition Pressure upon the Crystallinity of ZrN Thin Films ..... 66 4.3.2 Effect of Cathode Current upon the Crystallinity of ZrN Thin Films...... 68 4.3.3 Effects of Deposition Temperature upon the Crystalinity of ZrN Thin Films70

vi 4.3.4 Effect of Substrate Orientation upon the Crystallinity of ZrN Thin Films ... 71 4.3.5 Effect of Film Thickness upon the Crystallinity of ZrN Thin Films...... 72 4.4 Microscopic and Spectroscopic Investigation of ZrN Thin Films ...... 74 4.5 Summary and Conclusions ...... 79 4.6 Future work...... 80 4.7 References ...... 81

vii LIST OF TABLES Page Table 2.1. Electron temperature and ion density at 0.5 ampere...... 41 Table 2.2. Electron temperature and ion density at 1.0 ampere...... 42 Table 2.3. Double Probe Characterization Results...... 45 Table 2.4. Gallium nitride material properties...... 49 Table 2.5. Lattice mismatch between substrate materials/buffer layers to gallium nitride...... 50 Table 2.6. Thermal Expansion Coefficients of Silicon, Sapphire (0001) and ZrN...... 51 Table 3.1. Properties of ZrN...... 56 Table 3.2. Deposition conditions of ZrN...... 58

viii LIST OF FIGURES Page Figure 1.1. Schematic of a typical sputter deposition system...... 5

Figure 1.2. Paschen plot...... 6

Figure 1.3. Schematic of regions of a diode plasma sputtering system...... 7

Figure 1.4. Gas discharge classification as a function of discharge current...... 8

Figure 1.5. Schematic of the three sputtering regimes...... 9

Figure 1.6. Collisional processes within the sputter process ...... 10

Figure 1.7. Ion - Target impact in the sputter process ...... 10

Figure 1.8. Schematic of the potential distribution in a typical diode plasma...... 11

Figure 1.9. Initial growth stages in the sputter deposition process...... 13

Figure 1.10. Schematic representation of the influence of substrate temperature and argon working pressure on the structure of sputtered thin films...... 14

Figure 1.11. Reactive sputtering hysteresis ...... 17

Figure 1.12. Sputter yield...... 18

Figure 2.1. Schematic of dual magnetron sputter deposition system ...... 23

Figure 2.2. CAMP-M dual magnetron sputter deposition system ...... 25

Figure 2.3. Schematic self sputter magnetron deposition system...... 26

Figure 2.4. The center for advanced manufacturing processes and materials self sputter magnetron...... 28

Figure 2.5. Cross-section of a planar magnetron ...... 29

Figure 2.6. Electron in the presence of magnetic and electric fields ...... 30

Figure 2.7. Cross-section of balance magnetron showing magnetic field lines and electron orbit...... 31

Figure 2.8. Unbalanced magnetron...... 32

Figure 2.9. Pulsed dc waveform ...... 34

ix Figure 2.10. Dual magnetron magnetic field arrangement ...... 34

Figure 2.11. Effect of applied voltage as a function of current for the Advanced Energy DC magnetron power supply operating in constant current mode sputtering...... 35

Figure 2.12. Circuitry schematics of CAMP-M Langmuir probe...... 36

Figure 2.13. Typical single Langmuir probe I-V curve...... 38

Figure 2.14. Single probe measurement of dc zirconium sputtering at 0.5 Amps, 5 mTorr at the substrate center ...... 40

Figure 2.15. Typical double Langmuir probe characteristics ...... 42

Figure 2.16. Nitrogen concentration effects upon the current-voltage characteristic of zirconium sputtering measured with double Langmuir probe ...... 42

Figure 2.17. Schematic of four point probe set up...... 44

Figure 2.18. Typical Raman spectroscopy plot ...... 46

Figure 3.1. Proposed structure for Zr3N4 ...... 56

Figure 3.2. NaCl (111) orientation and it’s relation to the hexagonal structure ...... 57

Figure 4.1. Resistivity of zirconium nitride as a function of reactive gas concentration at constant current ...... 62

Figure 4.2. Resistivity as a function of sputtering pressure at 500C, 4% nitrogen concentration, and 0.5 Amp. cathode current on Si(111)...... 63

Figure 4.3. Resistivity as a function of cathode current on silicon (111) and silicon (100) substrates ...... 64

Figure 4.4. Effect of ZrN deposition temperature on resistivity for films deposited at 0.5 ampere cathode current, 5mTorr, and 4% nitrogen concentration...... 65

Figure 4.5. Effect of pressure upon crystallinity of ZrN on (111) silicon substrates at 500° C, 0.5 ampere cathode current and 4% nitrogen...... 67

Figure 4.6. Effect of deposition pressure upon crystallinity of zirconium nitride deposited upon silicon (100) at 500°C, 4% nitrogen and 0.5 ampere cathode current ...... 68

Figure 4.7. Effect of cathode current upon crystallinity of ZrN on (100) Si, 500 °C, 5mTorr, 4% nitrogen concentration ...... 69

x Figure 4.8. Effect of cathode current on the crystallinity of ZrN reactively sputter deposited on silicon (111) ...... 70

Figure 4.9. Effects of temperature upon the crystallinity of ZrN thin films deposited on (111) silicon substrates...... 71

Figure 4.10. Substrate orientation upon the cyrstallinity of ZrN thin films deposited at 500°C, 4% N2, 5 mTorr...... 72

Figure 4.11. Effect of thickness upon the crystallinity of ZrN deposited on (111) silicon samples at 0.5 amp, 10 mTorr, and 4% nitrogen concentration...... 73

Figure 4.12.. Proposed critical thickness model for the titanium nitride system...... 74

Figure 4.13. Cross-sectional SEM micrograph of ZrN on silicon ...... 75

Figure 4.14. Transmission Electron Micrographs of GaN thin films grown on ZrN ...... 76

Figure 4.15. Raman spectroscopy analysis of ZrN as a function of deposition pressure 77

Figure 4.16. Raman Spectrum of Zr3N4 thin films deposited on (100) silicon at 500°C, and 0.5 amp, and 100% nitrogen, as a function of temperature3...... 78

Figure 4.17. Raman as a function of cathode current on (100) Si at 5mTorr, 0.5A, 15°C, higher nitride ...... 79

xi 1.0 INTRODUCTION

Over the course of the last several decades, the world has seen an exponential increase in technological advancement. Particularly, in the area of solid state physics and thin film processing to either improve or alter certain material properties, to enhance existing circuitry or to generate entirely new materials. As the need for improved film control with smaller feature sizes becomes necessary, manufacturing techniques must likewise be improved to accommodate. Techniques such as molecular beam epitaxy (MBE), ion implantation, physical vapor deposition (PVD), chemical vapor deposition (CVD), electron beam evaporation, plasma enhanced chemical vapor deposition (PECVD) and high vapor phase epitaxy (HVPE) have been developed to provide a means of manufacturing high quality materials and coatings. As these techniques have evolved in the global market place, academia has continued to use them to investigate a host of materials and processing capabilities. J. Georg Benorz and K. Alexander Müller, of IBM Research, were presented the Nobel Prize in 1987 for the study of a thin films problem1. To carefully control the properties of thin films, processing steps must also be carefully controlled. Investigations and examinations must be undertaken in order to understand the effects processing conditions have upon the outcome of any experiment or process. Designed experiments and statistical processes are increasingly utilized in industry for quickly identifying variables that affect a response of interest. As no ideal lattice or thermally matched substrates exist for the production of III-V material, buffer layers have been utilized to accommodate the defects and strain generated from the inherent problems during the vapor deposition of material. An investigation into processing conditional effects for growth of transition metal nitrides for the buffered growth of III-V compounds is the focus of the following investigation.

1 1.1 IV-B Transition Metal Nitrides Super-hard coatings, such as the IV-B transition metal nitrides, carbon nitrides and diamond-like carbons (DLC), have been used primarily to protect components in high friction environments from wear. Their high strength and low coefficient of friction permit the use of these materials in industries ranging from the food processing industry to decorative coatings and have recently been examined by the microelectronics industry. The IV-B transition metal nitrides, specifically, have been investigated for a number of years for a wide variety of applications. Produced as a thin film, they demonstrate high hardness2,3, chemical stability4, corrosion resistance5, high melting point6, low resistivity7 and low coefficient of friction8. Recently, these IV-B transition metal nitrides have been investigated as diffusion barriers9,10,11,12,13 in metallization schemes, superlattice constituents14,15,1, field emitters16, electrical contacts to III-V semi-conductors17, Josephson junctions18, and cryogenic thermometers19. Zirconium nitride (ZrN) and titanium nitride (TiN) coatings provide the look of gold with the strength of steel. The coating of watch bezels, drill bits, outdoor lighting and window glass are several of the more popular uses of these nitrides. With such a broad range of applications, it becomes increasingly important to clarify and quantify the effects processing parameters have on the growth and properties of ZrN and identify new uses of the resulting materials. Traditionally, vacuum processes that produce polycrystalline films have been used for the deposition of transition metal nitrides20. Little work has been done to determine the optimum growth conditions for single crystal thin films of transition metal nitrides21. Nor has much work been done to investigate the buffer or seed layers of transition metal nitrides for the growth of III-V material22. Magnetron sputtering is a useful method for the deposition of these materials due to high rate of deposition, reduced operating pressures and increased energetic bombardment of the substrate.

2 1.2 The Sputtering Phenomena Sputtering is a physical vapor deposition (PVD) process where material is physically removed from a target by energetic ion bombardment. The term sputtering comes from the Dutch “sputteren” meaning “to spit out in small particles and with a characteristic explosive sound”23. Sputtering was first observed by Groves in 1852 with Plücker first to suggest, in 1858, that this discovery be used as a tool to produce metallic films24. Groves initially called the phenomena “cathodic disintegration”, and it was renamed “spluttering” by Sir John Thompson in 192125. In 1923 Thompson dropped the “l” and the term “sputtering” has lasted to this day. Sputtering is referred to as a physical process as it is simply a momentum exchange between particles with no chemical reactions occurring. Early plasma processing found removal of material from the cathode to be an inconvenience, as the destruction of the cathode would result. Others, however, including Edison, Wright, Faraday, and Crookes, saw the potential of such a device and began investigating the technique26. Today, sputtering is a powerful tool used for the deposition of thin films, in chemical analysis, etching and cleaning. Sputtering is commonly utilized for thin film deposition, as the extremes for melting or chemically reacting high melting point materials are not required, as with evaporation or an electrochemical process. There are four main processes that occur during sputtering: plasma generation, ion bombardment, atomic transport, and nucleation and growth at the substrate. These will be individually discussed, as each of the four have effects upon the growing film.

1.2.1 Plasma Generation A plasma is defined as a “quasi neutral gas of charged and neutral particles that exhibits collective behavior”27. The phenomenon was first reported by Sir William Crookes and the “glowing ore” from the high voltage experiments was termed the “Fourth State of Matter”28. In 1928, Langmuir and Tonk coined the term “plasma” to describe the ionized gas28.

3 Pressure is related to number the density of particles per cubic meter by the perfect gas law29 P = nkT [1.1] where n is the density of particles per cm3, k is Boltzmans constant and T is temperature. As the pressure within a vessel is reduced, the number density of particles is proportionally reduced. Typically, an airtight chamber is evacuated using a system of vacuum pumps, to pressures on the order of 10-6 - 10-10 Torr, and backfilled with a high purity gas to relatively low pressures, usually in the milli-Torr (mTorr) range. A traditional planar diode sputtering device can be seen in figure 1.1.

4

Figure 1.1. Schematic of a typical sputter deposition system30.

In its basic form, a planar diode sputtering device consists of two opposing, flat electrodes in an evacuated chamber. A variety of power supplies are available for plasma generation, ranging from dc to ac with frequencies into the mega-hertz. In dc sputtering a potential difference of several hundred volts is generated between the two plates creating an intense electric field between them. The electric field is given by V E = [1.2] d where V is the applied voltage and d is the target to substrate distance31.

5 A free electron placed near the cathode, by a UV photon or cosmic ray, will be accelerated in the direction of the field32. As the electron is propelled toward the anode it will encounter gas atoms contained between the electrodes. This is generally a high purity noble gas at a reduced pressure. The impact of a high-energy electron with a gas atom will ionize the atom by the ejection a weekly bound outer shell electron. The ejected electron is subjected to the same field and will also be accelerated toward the anode. This avalanche effect of electrons is responsible for the breakdown and sustained ionization of the gas.

The voltage at which this occurs is known as the breakdown voltage (Vb). The minimum breakdown voltage can found from plotting breakdown voltage versus pressure times electrode spacing, as seen in figure 1.2, described by a Paschen law24 pd Vb = a [1.3] log pd + b

10 where Vb is known as the breakdown voltage (typically on the order of 1-6kV ), p is the gas pressure, d is the separation distance of the electrodes, and a and b are constants11.

Figure 1.2. Paschen plot33.

There are four important processes that occur within the plasma in order to sustain a “quasi-neutral gas of charged particles”23: Ionization, Excitation, Relaxation and Recombination. The most prevalent of which is electron impact ionization as seen here: e-+ Aro → Ar+ + 2e- [1.4]

6 As can be seen in equation 1.4, with each electron collision two electrons and an ion will be produced. As the process continues, the generation of two electrons from a single electron impact leads to an avalanche effect (Townson discharge) and will only sustain a plasma if every collision emits more than one electron on average. Excitation is a process of electron promotion to a higher energy state, which is short lived due to recombination and photon generation. e- + Ar → Ar* + e- [1.5] Relaxation occurs as the electron relaxes from an excited state to its ground state. As the electron relaxes a photon is given off with an energy corresponding to an energy difference that is elementally specific. It is this process that is a function of all plasmas and responsible for the term “glow discharges”33. Recombination occurs as ions and electrons within the plasma recombine to form neutral molecules.

This avalanche effect provides a self sustaining nature, and charge equilibrium will be found with the procurement of characteristic dc sheaths as seen in figure 1.3.

Figure 1.3. Schematic of regions of a diode plasma sputtering system23.

7 The cathode sheath is a region of large voltage drop with low electron density. It is across the cathode dark space where a majority of the applied target voltage is dropped23. The cathode sheath contains few electrons, as they are repelled as soon as they enter the sheath. Ions, however, are drawn out of the plasma edge and are accelerated across the sheath as seen in Fig 1.6. There are a variety of ion-cathode interactions, which will be covered in the next section. There are four distinct glow regimes defined by the discharge current as seen in Figure 1.4. Below a discharge current of 10-9 A, the secondary electron emission is so low that there is not enough ionization to produce a self-sustained discharge. As the discharge current increases, the glow will eventually engulf the cathode in the abnormal glow regime. As discharge current increases further the discharge will move into the arc regime.

Figure 1.4. Gas discharge classification as a function of discharge current33.

1.2.2 Ion Bombardment In 1908, Stark was the first to propose the momentum transfer theory, which states that target atoms are ejected when momentum is transferred from bombarding ions to target atoms24. In the 1930’s and 40’s, however, it was believed that the thermal vaporization theory, developed by Hippel in 1926, was the most important sputtering mechanism24. The momentum transfer processes occurring at the surface are today believed to be caused by a collision cascade, which was proposed by Sigmund34, within the surface layers24.

8 Powell and Rossnagel state there are four different results obtained depending upon the kinetic energy (E) of the bombarding ion25. The modern theory of sputtering, developed by Peter Sigmund34, states that at low kinetic energies, or the “subthreshold region”, where E = 0-50 electron volts (eV), the ion does not have enough energy to dislodge target atoms. At moderate energies or the “knock-on sputtering regime”, where E = 50-1000 eV, ions impact and dislodge “knock-on” atoms into the target, these knock on atoms impact and dislodges other target atoms. This results in a collision cascade that ejects atoms, ions, electrons and neutrals from the first 10 to 50Å of surface of a target35. Several researchers have shown that the ions must exceed four times the binding energy of the target atoms to induce sputtering25, 35. It is within this regime that most practical PVD systems are operated. At high energy, where E = 1-50 keV, impacting ions penetrate into the lattice and are implanted. At very high energies, E >50 keV, ions penetrate deep into the lattice and are implanted within the bulk. While these latter two processes are used for material doping and implantation, it is virtually irrelevant to sputtering. The three regimes of sputtering by elastic collisions are seen here. For metallic targets, the elastic collision processes are the most important34.

a. single-knock-on b. linear cascade c. spike Figure 1.5. Schematic of the three sputtering regimes34.

Ions located near the edge of the plasma are accelerated across the electrical cathode sheath toward the negative cathode. Similar to the break event in billiards, target atoms are ejected as seen in figures 1.6 and 1.7.

9

Figure 1.6. Collisional processes within the sputter process30.

Figure 1.7. Ion - Target impact in the sputter process30.

Momentum transferred from ion impact to target atoms will drive the surface target atoms into the target. Sputtering, therefore, requires a sequence of collisions. A collision cascade arises from multiple interactions of bombarding gas atoms with target atoms, as the ions impart their energy over a distance within the surface of the target. Industrial processing plasmas are considered weekly ionized with ionization fractions typically between 10-5-10-1 33. Electrons within the plasma are orders of magnitude lighter (9.11*10-31 kg) than their ionic counterparts (~10-27 kg) and will respond more quickly to electric fields. Typical energies range from 1-10 eV for electrons and from 0.02-0.1 eV for the colder ions25.

10 In a steel vacuum chamber, fast moving electrons within the plasma are drawn out toward the chamber walls and other grounded surfaces. The more mobile electrons will be removed from the plasma at a rate greater than that of the heavier ions, leaving the plasma more positive than its surroundings. If an electrically isolated medium is inserted into a plasma, the same situation arises; however, a negative charge will build and upon reaching equilibrium leaves the medium at a negative potential known as the floating potential. This is given as the negative of the plasma potential

 kTe   2meπ 1+ Ti  Φf = −0.5 ln   [1.6]  e   mi  Te  where k is the Boltzmans constant, Te and Ti are the electron temperature and ion temperature respectively, me and mi are electron and ion mass respectively and can be seen graphically in figure 1.8.

Figure 1.8. Schematic of the potential distribution in a typical diode plasma30.

These electrically isolated surfaces are at a negative potential relative to the plasma. Electrons will initially be repelled from these surfaces until equilibrium is reached such that the plasma potential (Vp) is several volts more positive than it’s surroundings

1.2.3 Atomic Transport In sputtering, target atoms originate from the solid phase. These particles are then subject to aspects of the plasma before reaching the substrate. The transport mechanism is largely dependent upon the sputtering pressure utilized. Sputtered atoms traversing the

11 distance between cathode and anode (d) have a statistical probability of collision with gas atoms, known as the collision cross-section. The mean distance a gas molecule can travel before collision with another gas molecule is known as the “mean free path” (l) and is given by: 1 l = [1.7] 2πd 2 n where n is the molecular concentration (molecules/m3) and d is the diameter of the molecule36. At a pressure of 10-2 Torr, the mean free path is roughly 5 mm37. At low pressures, or the ballistic transport regime25, sputtered particles will have few gas phase collision and arrive at the substrate with their initial kinetic energy. These energetic bombarding species can have beneficial or deleterious effects upon the substrate and the growing film. At higher pressures, or shorter mean free paths, sputtered atoms collide with gas atoms and transfer energy through elastic collisions, thereby thermalizing with the gas, the outcome of which is a decrease in sputtered atom energy as a function of increasing pressure or increasing distance between cathode and anode33. This type of transport is referred to as diffusive transport.

1.2.4 Nucleation and Growth Thin films form through a nucleation and growth process. The process begins as sputtered particles arrive at the substrate surface and become physically absorbed. Initially these particles are not in thermal equilibrium with the substrate and will move and interact with other species. The mobility of atoms arriving at the substrate depend upon the binding energy of the atom to the substrate and the temperature of the substrate. At higher substrate temperatures and lower binding energies, the greater the motion of the adatom upon the surface. Clusters of particles that form that are not thermodynamically stable will be desorbed from the surface. Thermodynamically stable clusters are said to have overcome the nucleation barrier24 and will begin to grow. These nuclei grow across the surface of the substrate by atom surface diffusion until islands grow large enough to coalescence and begin to grow vertically by the arrival of sputtered species.

12 The coalescence growth stage occurs when these islands grow large enough to touch and continues until the film reaches continuity38. The islands start to coalesce to reduce the overall surface area of the growing film. This tendency to form larger islands is known as agglomeration24. When the islands grow together, the film is said to go from a discontinuous islands type to a porous network type24. This type of growth will continue until all islands have grown together and will a continuous film. Depending upon the conditions for deposition, there are three initial nucleation and growth stages. Films grown by coalescence of islands are known as Volmer-Weber type24. Volmer-Weber occurs when sputtered atoms are bound more tightly to each other than the substrate, such as in the growth of metal films on insulators39. Films that grow layer by layer are known as Frank-van der Merwe type. Frank-van der Merwe growth occurs when lattice mismatch is small and the binding energy of sputtered atoms is equal to or less than the binding energy of sputtered atoms to the substrate39. Those that grow by a mixture of the two are called Stranski-Krastanov type24. A graphical summary of the initial stages of growth can be seen in figure 1.9.

Figure 1.9. Initial growth stages in the sputter deposition process1.

In most cases, thin films are formed by the coalescence of islands. When these islands grow together, various defects and grain boundaries are formed. Grains that coalesce with randomly oriented directions will produce a polycrystalline film. Movchan and Demchishin, followed later by Thornton and Messier32, investigated the structures of thick films and determined that film structure could be expressed as a

13 30 function of deposition temperature, melting temperature (T/Tm) and deposition pressure . The model consists of four zones as seen in figure 1.10.

Figure 1.10. Schematic representation of the influence of substrate temperature and argon working pressure on the structure of sputtered thin films30.

Zone 1 films occur when adatom mobility is not enough to overcome the effects of shadowing and results in a fibrous film separated by voids. As T/Tm increases, so too does the crystallite size. Zone T is known as the transition region where the adatom mobility is sufficient to overcome the shadowing effects and results in a smoother surface than in Zone 1 films. Zone 2 is a region dominated by surface diffusion growth with increasing grain size separated by intercrystalline boundaries30. Zone 3 is a region characterized by elevated growth temperatures. At these temperatures adatom mobility results in bulk diffusion and equiaxed grains. At sufficiently high substrate temperatures and sufficiently lattice matched substrates, epitaxial growth can be achieved30. Sputter deposition results in columnar growth40, which Thornton describes as a “fundamental consequence of forming a solid from a vapor flux [arriving] from one direction”30.

14 1.3 Practical Aspects of Sputtering

1.3.1 Epitaxial Growth The term epitaxy comes from the Greek meaning “to arrange” (taxia) “upon” (epi)39. Epitaxial growth is essentially the reproduction of the structure of a substrate in a thin film. Epitaxy can be achieved in two ways: either through the growth of a material upon a substrate of that identical material (homoepitaxy), or the growth of a material upon a substrate of different material (heteroepitaxy). The first successful attempt of which was performed by Frankenheim, in 1836, with parallel oriented growth of sodium nitrate on calcite, with a systematic study later performed by Barker41. The discovery of the diffraction of X-rays in 1912, led to the development of x-ray diffraction devices and transmission electron microscopy (TEM) which allowed for the quantitative examination of epitaxial produced material.

One of the most important conditions for growth of epitaxial films is the substrate temperature. There exists a critical temperature, known as the epitaxial temperature (Te), at which epitaxy will occur41. Epitaxial deposition temperatures must be relatively high to allow for large scale atomic movement and arrangement upon the substrate. This requires that the thermal expansion coefficients of the material system be well matched. At low temperatures, atoms generally condense where they arrive as thermal energy is quickly withdrawn. However, with elevated substrate temperature, the addition of thermal energy to the substrate allows for increased adatom movement and the time necessary for atomic ordering. Epitaxial deposition of metallic films on semiconductors is of interest for metal based transistors and three-dimensional integrated circuitry42.

The growth of material from the gas phase is a complex process and many reactions and formations can occur. For a structure that is grown upon a substrate that is not of closely matching atom spacing, the interfacial energy (γi) is reduced by increasing the density of bonds across the interface31. The growing film will achieve the lowest free

15 energy state and will strain the first few mono-layers in order to accommodate the films lattice constant. This effect is known as pseudomorphism, or pseudomorphic growth31. The use of single crystal silicon substrates is the basis of today’s chip technology. As the push for high power, high frequency circuitry has stretched the limits of silicon processing technology; other materials capable of meeting these needs are being developed. Single crystal wide bandgap semiconductors, such as gallium nitride (GaN), are currently being developed to meet the needs of tomorrows chip processing technology. GaN, however, suffers from a lack of suitably thermal and lattice matched substrates. Sapphire and silicon carbide wafers are today the most widely used substrates for these applications. The expense of these wafers has fueled research of buffer layers grown on Si that can accommodate the strain produced during thin film growth and result in closely matched surfaces. Nitronex43 is one such company using buffer layers on silicon to produce GaN based high power, high frequency circuitry.

1.3.2 Reactive Sputtering DC reactive sputtering has been used to generate a host of high quality nitrides: (44) (45,46) (47,48) (41) (38) NbN , TiN , AlN ; and oxides: PtO , TiO2 . Reactive sputtering is an advantageous method for formation of complex products from the reaction of elemental metal targets and reactive gases. This eliminates the need for construction of expensive compound targets or the use of highly volatile gases. Sputtering in an inert environment yields no chemical reaction between sputtered species and ionic sputtering species. If a reactive gas is introduced into the sputtering environment, reactions between sputtered species and reactive gas molecules, or atoms, will occur. At relatively low partial pressures, the whole of the reactive gas will be consumed through reactions with sputtered species. This process is synonymous with the getter pumping action of an ion pump and is referred to as the metallic mode of sputtering. As the partial pressure of reactive gas increases and the supply of reactive gas exceeds the reaction rate with sputtered species, reactions will then occur at the surface of the target, resulting in compound formation on the surface of the target. If this material is electrically insulating, or the sputter yield of the compound is less than that of the pure

16 metal, a decrease in deposition rate will occur. This is referred to as the non-metallic or poisoned mode of sputtering as the target is said to be “poisoned”49. The effect is generally described in terms of a hysteresis examining the deposition rate versus reactive gas concentration as seen in figure 1.11. Once poisoning of the target occurs, and the partial pressure of reactive gas is reduced, the deposition rate will not increase again until the reacted material on the surface of the cathode has been removed.

120

100 Increasing Reactive Gas 80 Concentration 60 Decreasing 40 Reactive Gas Concentration 20 Deposition Rate (nm/min.) 0 02468101214 Reactive Gas Concentration (%)

Figure 1.11. Reactive sputtering hysteresis.

1.3.3 Radio-frequency Sputtering The sputtering of insulating compounds does not permit the use of dc power as charge will accumulate on the surface of the target resulting in electrical discharging, or an arcing event. Arcing is detrimental to the growth of high quality films as it results in macro-molecule ejection of target species which will be incorporated into the growing film. The use of high frequency power supplies allows for the sputtering of insulating, conducting, or semi-conducting material by defeating the build up of charge on the surface of the target. The Federal Communications Commission (FCC) has assigned a radio frequency of 13.56 MHz for industrial, educational and medical uses33, and is one of the frequencies

17 typically utilized for RF sputtering. As electrons are more mobile than ions, they will respond to the field more quickly and oscillate with the applied frequency resulting in the ionization of the gas. The use of RF sputtering results in a decrease in deposition rate as the negative bias on the target is only applied on the half cycle. The positive cycle alleviates any charge from the surface. RF can be coupled through any kind of impedance permitting the sputtering of insulators33.

1.3.4 Sputter Yield Sputter yield is given as the mean number of atoms remitted from a target per incident ion. The mass and energy of the incident atoms, angle of incidence, mass of target atoms, and target crystallinity will all affect the rate by which material is sputtered30. Thornton investigated the sputter yield of most commonly sputtered elements as a function of ion energy, which is provided as a quick reference guide30.

Figure 1.12. Sputter yield30.

18 1.4 References

1K. Reichelt, Vacuum, v. 38, p. 1083, 1988.

2C. P. Constable, J. Yarwood, and W. D. Münz, Surface and Coating Technology, v. 116, p. 155, 1999.

3W. D. Sproul, Thin Solid Films, v. 107, p. 141, 1983.

4G. V. Samsonov, Soviet Physics-Technical Physics, v. 1, p. 695, 1967.

5P. Panjan, B. Navinšek, A. Cvelbar, A. Zalar, and I. Milošev, Thin Solid Films, v. 298, p.281, 1996

6K. Schwarz, A. R. Williams, J. J. Cuomo, J. H. E. Harper, and H. T. G. Hentzell, Phys. Rev. B, v. 32(12), p. 8312, 1985.

7L. E. Toth, Transition Metal Carbides and Nitrides, Academic, New York, 1971.

8I. Milošev, H. H. Strehblow, and B. Navinšek, Thin Solid Films, v. 303, p. 246, 1997.

9L. Krusin-Elbaum, M. Wittmer, C. –Y. Ting, and J. J. Cuomo, Thin Solid Films, v. 104, p. 81, 1983.

10M. Östling, S. Nygren, C. S. Petersson, H. Norström, P. Wiklund, R. Buchta, H. –O. Blom, and S. Berg, J. Vac. Sci. Technol. A, v. 2, p. 281, 1984.

11N. Kumar, K. Pourrezaei, B. Lee, and E. C. Douglas, Thin Solid Films, v. 164, p. 417, 1988.

12M. Y. Kwak, D. H. Shin, T. W. Kang, and S. Kim, Thin Solid Films, v. 339, p. 290, 1999.

13J. P. Noël, D. C. Houghton, G. Este, F. R. Shepherd, and H. Plattner, J. Vac. Sci. Technol. A, v. 2(2), p. 284, 1984.

14M. –S. Wong, G. –Y. Hsiao, and S. –Y. Yang, Surface and Coatings Technology, v. 133-134, p. 160, 2000.

15M. Setoyama, M. Irie, H. Ohara, M. Tsujioka, Y. Takeda, T. Nomura, and N. Kitagawa, Thin Solid Films, v. 341, p. 126, 1999.

16M. Endo, H. Nakane, and H. Adachi, J. Vac. Sci. Technol. B, v. 14(3), p. 2114, 1996.

19 17S. Ruvimov, Z. Lilental-Weber, J. Washburn, K. J. Duxstad, E. E. Haller, Z. R. Fan, S. N. Mohammad, W. Kim, A. E. Botchkarev, and H. Morkoç. Appl. Phys. Lett., v. 69, p. 1556, 1996.

18K. Schwarz, A. R. Williams, J. J. Cuomo, J. H. E. Harper, and H. T. G. Hentzell, Phys. Rev. B, v. 32(12), p. 8312, 1985.

19T. Yotsuya, M. Yoshitake, and T. Kodama, Cryogenics, v. 37, p. 817, 1997.

20R.G. Gordon, Mat. Res. Soc. Symp. Proc., v. 335, p. 9, 1994.

21B. O. Johansson, J. –E. Sundgren, J. E. Greene, A. Rockett, and S. A. Barnett, J. Vac. Sci. Technol. A, v. 3, pg. 303, 1985.

22J. Ito et al., United States Patent #6,426,512, issued 2002.

23B. Chapman, Glow Discharge Processes, John Wiley & Sons, Inc., New York, p. 177, 1980.

24K. Wasa and S. Hayakawa, Handbook of Sputter Deposition Technology: Principles, Technology and Applications, Noyes, New Jersey, 1992.

25R. A. Powell and S. M. Rossnagel, Thin Films: PVD for Microelectronics: Sputter Deposition Applied to Semiconductor Manufacturing, Academic Press, San Diego, 1999.

26D. M. Mattox, The History of Vacuum Coating Technology, 2002, from http://www.svc.org.

27F. F. Chen, Introduction to Plasma Physics and Controlled Fusion, Plenum Press, New York, 1984.

28http://my.engr.ucdavis.edu/~das/ctix/plasma.html

29J. R. Roth, Industrial Plasma Engineering: Principles, Institute of Physics Publishing, London, 1995.

30J. A. Thornton, Deposition Technologies for Films and Coatings: Development and Applications, ed. R. F. Bunshah, Noyes Publications, NJ, 1982.

31J. L. Cecchi, Handbook of Plasma Processing Technology: Fundamentals, Etching, Deposition, and Surface Interactions, ed. S. M. Rossnagel, J. J. Cuomo, W. D. Westwood, Noyes Publications, Park Ridge, NJ, 1990.

20 32Thesis, NC State University, N. M. Williams, Enhanced Reactive Magnetron Sputtering of Aluminum Nitride, 1998.

33S. M. Rossnagel, Thin Films Processes II, ed. J. L. Vossen, W. Kern, Academic Press, New York, 1991.

34P. Sigmund, Topics in Applied Physics: Sputtering by Particle Bombardment I, ed. R. Behrisch, v. 47, Springer-Verlag, Berlin, 1981.

35G. K. Wehner, G. S. Anderson, Vacuum Technology, Thin Films, and Sputtering: an Introduction, ed. R. V. Stuart, Academic Press, New York, 1983.

36D. L. Smith, Thin-Film Deposition: Principles and Practice, McGraw-Hill, New York, 1995.

37J. Floro, C. Jahnes, and D. J. Mikalsen, Vacuum Technology, Laboratory Procedures and Deposition Processes, T. J. Watson Research Center, NY, 1986.

38W. D. Westwood, Progress in Surface Science, v. 7, p. 71, 1976.

39L. Hultman, Growth of Single- and Polycrystalline TiN and ZrN Thin Films by Reactive Sputtering; Influence of Low Energy Ion Irradiation, Thesis, Linköping University, Sweden, 1988.

40A. G. Dirks, H. J. Leamy, Thin Solid Films, v. 47, p. 219, 1977.

41D. W. Pashley, Epitaxial Growth: Part A, ed. J. W. Matthews, Academic Press, New York, 1975.

42S. A. Barnett, L. Hultman, J. –E. Sundgren, Appl. Phys. Lett., v. 53, p. 400, 1988.

43http://www.nitronex.com

44M. S. Wong, W. D. Sproul, X. Chu, and S. A. Barnett, J. Vac. Sci. Technol. A, v. 11, pg. 1528, 1993.

45S. Adachi and M. Takahashi, J. Appl. Phys., v. 87, pg. 1264, 2000.

46C. Carney and D. Durham, J. Vac. Sci. Technol. A, v. 17, pg. 2850, 1999.

47F. Englemark, G. Fucntes, I. V. Katardjiev, A. Harsta, U. Smith, and S. Berg, J. Vac. Sci. Technol. A, v. 18, pg. 1609, 2000.

48H. Jensen, J. Sobota, and G. Sorensen, J. Vac. Sci. Technol. A, v. 15, pg. 941, 1997.

21 49W. D. Westwood, Handbook of Plasma Processing Technology: Fundamentals, Etching, Deposition, and Surface Interactions, ed. S. M. Rossnagel, J. J. Cuomo, W. D. Westwood, Noyes, New Jersey, 1992.

22 2.0 EXPERIMENTAL STRUCTURE

The physical vapor deposition sputtering systems utilized for deposition of ZrN and GaN include a dual and a single source magnetron sputter system. Both systems were assembled at the Center for Advanced Manufacturing Processes and Materials and are unique to this research environment. The mechanisms for thin film deposition are different for each system and will be individually discussed.

2.1 Dual Magnetron The dual magnetron system seen in figure 2.1 consists of four main components: the deposition chamber, pumping system, pressure monitor and control, and power supplies.

Power

H20 in H20 out Mass Flow Controller Array (Ar, N2)

Nude Convectron gauge Ion Gauge N S N N S N

Cryogenic Pump

Plasma

Substrate

Nude Ion Gauge

Convectron Gauge Bias Substrate Heater Baratron gauge Gate Valve H20 in H20 out Figure 2.1. Schematic of dual magnetron sputter deposition system.

23 The deposition chamber is a 50 liter stainless steel vacuum chamber, which can either be heated or cooled by external heating and cooling jackets. Two six-inch diameter magnetrons are mounted face down and are isolated from the top plate. The targets can be biased by DC, RF, or AC biasing voltage. A six inch linear-motion mobile substrate holder is mounted at a distance of three inches below the targets and can be heated or cooled. The substrate holder is equipped with external bias capability. The pumping system consists of an Edwards QDP40 Drystar vacuum pump used to evacuate the chamber to a pressure of 10-2 Torr (1.3 Pa.). A CTI Helix Technology Cryotorr10 cryogenic pump, with a pumping speed of 5,000 liters per second for argon, was used with a water-cooled CTI-Helix Technology 9600 He compressor to achieve a base pressure of 10-7 Torr (1.3 * 10-5 Pa.) for all depositions. The pressure monitoring system consists of an MKS type 390HA-0001 capacitance manometer Baratron gauge with a range of 1 Torr and was used to monitor pressure during deposition. Both the chamber and the pumping system were monitored with a Granville-Phillips thermocouple Convectron gauge and a Troy-Onic, Inc. nude ionization gauge from Duniway Stockroom (Part # I-100-P), in conjunction with a Granville- Phillips 303 vacuum process controller. Two MKS Type 260 mass flow controllers, with a maximum flow of 100 sccm of nitrogen, were used to control the flow of argon and nitrogen to the system. An Advanced Energy MDX Series Magnetron Drive direct current (DC) power supply was used for all DC experimentation. The MDX power supply is capable of supplying 10 kW in a constant current, voltage or power mode. An RFPP RF10S radio frequency power supply was used for capacitively-coupled experimentation. The RF10s operated at the standard 13.56 MHz with a maximum power output of 1000 W into a 50 ohm load. The RF10S can be pulsed with a specified frequency or operated in a constant operation mode. This system was primarily used for the deposition of zirconium nitride (ZrN), higher zirconium nitride (Zr3N4). A photo of the device is shown in Figure 2.2.

24

Figure 2.2. CAMP-M dual magnetron sputter deposition system

2.2 Self Sputter Magnetron The self-sputter magnetron system seen in figure 2.3 consists of four main components: the deposition chamber, pumping system, pressure monitor and control, and power supplies.

25 Heater Power

Substrate Heater Substrate Holder

Plasma

Target Pool

Power

Chilled water lines

Gate Valve

Turbo Pump

To roughing pump Figure 2.3. Schematic self sputter magnetron deposition system.

The deposition chamber is a 25 liter stainless steel vacuum chamber with a four inch quartz window sealed with a Viton® O-ring. A four inch water cooled target is mounted directly below the substrate. The substrate holder can be heated or cooled with target to

26 substrate distance of 8 inches. The system is equipped with a hollow cathode emission source for increased ionization of gaseous species.

The pumping system is comprised of a Welch Duo-seal rotary vein pump used to evacuate the chamber to a pressure of 10-2 Torr (1.33 Pa.). A Leybold-Heraeus turbo pump was used to achieve a base pressure of 10-7 Torr (1.33 * 10-5 Pa.) for all depositions.

The pressure monitoring system consists of an MKS 122AA-00010AB capacitance manometer Baratron gauge with a range of 10 Torr and was used to monitor pressure during deposition. Both the chamber and the pumping system were monitored with Granville-Phillips thermocouple convectron gauges and an Electron Technology Inc. 4336K/1 nude ionization gauge with a Varian 843 vacuum ionization gauge. Two MKS Type 260 mass flow controllers with a maximum flow of 100 sccm nitrogen were used to control the flow of argon and nitrogen to the system. An ENI RPG-50 Pulsed DC power supply was primarily used for deposition.

This system was utilized to deposit gallium nitride, a photo of the system is shown in Figure 2.4. A four inch water cooled molybdenum target holder was used to hold a high purity (99.999%) liquid gallium target. A four inch substrate was mounted directly above the liquid gallium pool at a distance of eight inches. The substrate was heated with a carbon serpentine heater capable of raising temperatures to > 1000°C with the use of a Variac high voltage source. High purity (99.999%) argon and nitrogen gases were used for all experimentation. A photo of the device is shown in figure 2.4.

27

Figure 2.4. The Center for Advanced Manufacturing Processes and Materials self sputter

magnetron.

28 2.3 Magnetrons F. M. Penning was the first to propose and then patent1 magnetron sputtering in 19352. A prototype of the planar magnetron was invented by Wasa in 19672,3. Magnetrons utilize magnetic confinement of charged particles to increase ionization within regions close to the target. Such confinement allows for electron trapping within the magnetic field flux lines, but also allows ions to travel across the sheath and gain acceleration towards the target. Magnetrons are generally operated in a diode mode with rare earth magnets, such as barium ferrites, alnico alloys, cobalt-rare earth alloys4, or electro-magnets located behind the cathode in a circular or closed loop pattern as seen in Fig. 2.5.

Figure 2.5. Cross-section of a planar magnetron2.

The magnetic field is orientated parallel to the cathode surface such that the ExB drift path forms a closed loop5 in the shape of a toriod, or a “race track”. The generated ExB field causes electrons to orbit along field lines confining primary and secondary electrons within this magnetic field loop. This increases the path length of the electrons thereby increasing the probability of an ionization event and increases the density of ionized species close to the cathode. The force upon an electron in a magnetic field is

→ → Fm = q v× B [2.1] For particles moving perpendicular to this field, the force causes the particles to orbit with a radius, known as the Larmour radius, that is given by mυ r = ⊥ [2.2] qB

29 2 where the perpendicular velocity is the component in the kinetic energy ½mυ ⊥, and is equal to the potential energy by conservation; thus

r = 2m(kE) [2.3] qB where m is the electron mass, k is the Boltzmans constant, E in the strength of the electric field, q is the elemental charge of an electron, and B is the strength of the magnetic field.

Figure 2.6 shows the effects of magnetic and electric fields on electron orbit. Fig 2.6a depicts an electron rotating around a magnetic field coming perpendicularly out of the page. 2.6b shows the electron orbital path along a magnetic field oriented parallel to the page. 2.6c shows a representation of electron impact and the impacts effect on an electron traveling along a magnetic field oriented parallel to the page. 2.6d and 2.6e show electron motion within an electric and magnetic field oriented perpendicular to one another. The electron will “drift” and orbit around the ExB direction under the influence of the magnetic and electric field.

Figure 2.6. Electron in the presence of magnetic and electric fields6.

30 Secondary electrons emitted from the target become trapped within the magnetic toriod and are confined to this closed loop. An increase in the number of the active species close to the cathode results in an increased rate of deposition. The only limitation in the deposition rate of magnetron systems is the ability to cool the cathode4. Cathode cooling becomes an issue due to the high power dissipated at the cathode where 55%- 70% of the applied power is lost to heating4. The high electron density in regions near the cathode results in a high current, low voltage discharge4. Due to increased plasma density, magnetron discharges can be sustained at lower operating pressures, which results in less gas scattering.

A wide variety of magnetrons have been developed in two basic geometrical groups: planar4 and cylindrical7. All magnetron sputtering geometries utilize a closed loop ExB field for electron confinement.

2.3.1 Balanced versus Unbalanced Magnetron Sputtering The magnetic field line arrangement in a magnetron discharge will have an effect upon the operation and properties of a plasma. A magnetron is said to be balanced when all of the field lines emanating from the north poles of the magnet arrangement are re- connected to the south poles of the exterior magnet as shown in figure 2.7.

Figure 2.7. Cross-section of balance magnetron showing magnetic field lines and electron orbit.8

31 Unbalanced magnetrons utilize an unmatched set of magnets such that the magnetic field lines bleed out to the substrate as seen in figure 2.8.

Figure 2.8. Unbalanced magnetron8.

Unbalanced magnetrons are advantageous for increasing the energy available at the substrate by channeling ions to the surface of the substrate, which increases ion bombardment of the substrate and will affect the properties of the growing film.

2.3.2 DC Magnetron Sputtering Direct current (dc) sputtering has become a preferred method for metal deposition. Providing high deposition rates and uniform coverage, dc magnetron sputtering provides the ability to quickly deposit large amounts of material. A typical magnetron sputtering system can be seen in Fig.2.1. DC magnetron sputtering systems are typically operated in argon at cathode potentials from 300-700 V and pressures between 1-10 mTorr4. For ideal magnetic confinement, several researchers4,5 state that the voltage current characteristics follow I = kV n [2.4] where I is the cathode current, V is the cathode potential and k and n are constants with n indicating the efficiency of electron trapping, where n ranges anywhere from below 1 to just above 10. Some researchers9 have found that this expression overestimates the current at low voltages and have proposed the relationship

32 3 / 2 I = C()V −Vo [2.5] for magnetron discharges, where Vo is the minimum voltage at which the discharge can be maintained which governed by the Paschen law, and 2e3 / 2 1 C ≈ 2()1+ γ ε K [2.6] o M L2 where M is the atomic weight of the sputtering ion, γ is the secondary electron coefficient, L is the distance of the dark space and K is a pressure dependent9. The cathode in dc sputtering is referred to as cold as it does not rely on the thermionic emission of secondary electrons to sustain the discharge10.

2.3.3 Pulsed DC Sputtering DC sputtering is used primarily for the sputtering of conducting electrodes, which limits the process to the sputtering of metals and low resistivity metallic alloys. Pulsed DC sputtering is a relatively new method of sputter deposition. Pulsing the magnetron discharge in the mid frequency range (10-200 kHz) stabilizes the plasma, prevents arcing, and yields high quality insulating compounds11. Initial work began on pulsed magnetron sputtering in the early 1990’s. Pulsed magnetron sputtering utilizes a small positive pulse, on the order of microseconds to “sweep” the charge from the surface of the target12. A schematic of the waveform utilized can be seen here12.

33

Figure 2.9. Pulsed dc waveform12.

The actual waveform for pulsed magnetron sputtering varies due to inherent characteristics of the plasma.

2.4 Plasma Characterization

2.4.1 Pressure Effects on I-V Characteristics A Bell 640 Incremental guassmeter was used to measure the magnetic field strength across the surface of the dual magnetron target and at the surface of the substrate. The strength of the magnetic field arrangement is depicted in Figure 2.10.

Target

N S N 350G 900G 350G

3 inches

7G 10G 7G

Substrate

Figure 2.10. Dual magnetron magnetic field arrangement.

34 The dc magnetron plasma was characterized by examining the applied current-voltage characteristics to investigate the efficiency of electron trapping and the effect of deposition pressure on the electron trapping for the dual magnetron system. The Advanced Energy power supply was operated in constant current mode and current was incrementally adjusted from 0.25 to 2 amps with applied voltage read at each incremental step. The applied voltage was displayed by the Advanced Energy MDX magnetron power supply. The I-V characteristics measured at 5 and 10 mTorr can be seen in figure 2.11. As the operating pressure is increased the applied voltage is noticeably decreased for the same current. These measurements show that magnetron voltage will increase slowly even upon large changes in cathode current. Rossnagel6 shows this same effect for Cu, Al sputtering in argon that as gas pressure increases, gas density increase and the voltage required is reduced for the same current.

4

3.5

3

y = 4E-24x9.5 2.5 R2 = 0.998 y = 7E-26x10.5 2 R2 = 0.998

1.5 Current (amps) 10mTorr 5mTorr 1

0.5

0 200 250 300 350 400 Voltage (Volts)

Figure 2.11. Effect of applied voltage as a function of current for the Advanced Energy DC magnetron power supply operating in constant current mode sputtering

35 Figure 2.11 is designed to document the performance of the dual magnetron system and can be used a reference for power consumption determination during operation. For experimentation performed at 5 mTorr and 10 mTorr, the current/voltage characteristics can be calculated. Power law trendlines were fit to the experimental data points for both experiments conducted and R2 values achieved were at least 0.998. The results show an extremely small value of k, and a fluctuation in n of 0.9 by doubling the pressure.

2.4.2 Langmuir Probe Characterization An electric (Langmuir) probe was utilized to examine several plasma parameters.

Electron temperature (Te), electron density (ne), ion density (ni), plasma and floating potential (Vp and Vf, respectively) and ion flux (Γi) were found from measurements taken with a Langmuir probe. The circuit of the Langmuir probe is shown in figure 2.12. The probe was constructed of a thin tungsten wire with a diameter of 0.24 mm, an exposed length of 6.64 mm, and a cross-sectional area of 5.052 * 10-6 m2. Measurements were taken during sputtering with mixtures of argon and nitrogen gases.

A V A V

Single Probe Configuration Double Probe Configuration

Figure 2.12. Circuitry schematics of CAMP-M Langmuir probe.

The electric probe technique was developed by Langmuir as early as 192413. An electric probe is a thin electrode, typically a wire, connected to an ammeter, voltmeter

36 and power supply capable of biasing the probe positively or negatively. When inserted into a plasma, the current the probe collects as a function of applied voltage is then used to gather information about the condition of the plasma. The advantage of this technique is that localized measurements can be made, where other techniques give average information over a large volume. When the probe is biased positively, negative charge is collected until saturation occurs. As the negative charge carriers are typically electrons, this portion of the I-V curve is referred to as electron saturation. When the probe is biased negatively, positive charge carriers are collected until saturation is reached. As the positive charge carriers are ions, this portion of the I-V curve is referred to as ion saturation. The Langmuir probe, while relatively simple in design, is an effective means for determining plasma parameters. Electron probes provide a means for determining plasma parameters with little disturbance of the plasma. Any disturbance within a plasma is localized by the plasma and will not affect the overall behavior. While there is some error associated with the use of Langmuir probes, they are useful for a relatively inexpensive method of quickly characterizing plasmas.

2.4.2.1 Single Probe With no bias, the single probe is at a negative potential relative to the plasma floating potential and will collect ions, or negative current. With increased negative biasing, the current that flows is that of ions withdrawn from the plasma until ion current saturates. It is this ion saturation current that is used to calculate ion density (ni). The bias voltage is increased positively until the probe draws no current indicating that the probe is at the the same potential as the plasma, or the floating potential (Vf) of the plasma. The bias voltage is increased positively until electron saturation is reached and electron density

(ne) can be calculated. A typical single probe current versus voltage plot can be seen in figure 2.13.

37

Figure 2.13. Typical single Langmuir probe I-V curve14.

From evaluation of the curve the electron temperature (Te) can be calculated from d ln | I | e = [2.7] dV kTe where k is the Boltzmann constant, and e is the electron charge. Thus, for a log plot, the slope dV/d(lnI) is a measure of the plasma electron temperature: e dV T = [2.8] e k d(ln I) From electron and ion saturation currents, electron and ion density can be found from

1 2  8kT  I = 1 en A e  [2.9] e 4 e    πme 

1  kT  2  e  I i = eni A  [2.10]  mi  This technique is valid when the probe radius is large compared to the sheath distance given by the Debye length

1/ 2  kTe  λD =   [2.11]  4πne 2 

38 Using a technique suggested by Dr. Bourham, the floating potential can be used to calculate electron temperature through

1/ 2 kT  m  e  i  V f = − ln  [2.12] e  4πme  The ion density can then be calculated through

1.7I i ni ≅ [2.13] kT eA e p m + The ionization fraction can then be found by calculating the ratio of ion density to total number of particles. Single Langmuir probe measurements were taken during reactive magnetron sputtering of zirconium with argon and nitrogen. The probe was located at the center of the substrate during all measurements. The probe was shuttered during target pre- cleaning; the shutter was then removed and measurements taken. Current versus applied voltage was plotted and can be seen in figure 2.14. Ion current and floating potential was found and electron temperature and ion density were calculated at 3, 5, 7 mTorr, and 0.5 ampere and 1.0 ampere cathode current15. The reduced mass (µ) of the gas was calculated to be 6.52 * 10-26 kg.

39 2.5

2

1.5

1 Current (micro A) 0.5

0 -30 -20 -10 0 10 20

-0.5 Voltage (V)

Figure 2.14. Single probe measurement of dc zirconium sputtering at 0.5 amps, 5 mTorr at the substrate center.

Table 2.1. Electron temperature and ion density at 0.5 ampere. -3 0.5 Ampere Te (eV) Ii (m ) 3 mTorr 2.18 8.29*1017 5 mTorr 1.94 8.29*1017 7 mTorr 1.62 8.96*1017

Table 2.2. Electron temperature and ion density at 1.0 ampere.

1.0 Ampere Te(eV) Ii(m-3) 3 mTorr 1.94 1.4*1018 5 mTorr 1.78 1.37*1018 7 mTorr 1.68 1.30*1018

40 Ion density, ion current and ionization fraction were found to increase with cathode current. Ion density, ion current and ionization fraction were found to decrease with pressure. The Debye length was calculated and found to be much smaller than the probe diameter, validating the use of the equations. As cathode current increases, the density of ionized species increase, and the number of those transferred to the substrate are also increased. As pressure is increased, the mean free path, the average distance a particle can travel, is decreased. This decrease in mean free path reduces the energy of ions arriving at the substrate. The mean free path is calculated from RT MFP= [2.14] 2 2πd N A P where R is the gas constant, T is temperature, d is particle diameter, NA is Avogadro’s number and P is pressure. The MFP decreases from 6 cm. to 1.5 cm. when pressure is increased from 5 mTorr to 20 mTorr.

2.4.2.2 Double Langmuir Probe The double probe technique uses two identical probes, one to draw electron current and one to draw ion current. The two together, however, draw no net current. Double probes are used to quickly determine electron temperature (Te) from

dI ii = [2.15] dV V =0 2Te Electron temperature can also be estimated by dividing the potential difference between “knees” by four. A schematic of a typical double Langmuir can be seen in Figure 2.15.

41

Figure 2.15. Typical double Langmuir probe characteristics13.

Double probe measurements were taken during reactive magnetron sputtering of zirconium in argon and nitrogen. The double probe was located at the center of the substrate for all measurements. The sputtering pressure was 5 mTorr and the cathode current was 0.5 amperes. A plot of the current versus applied voltage can be seen in figure 2.16.

8 6 4 2 0 -2

Current (micro-A) Current -4 4% N -6 100% N -8 -60 -40 -20 0 20 40 60 Voltage (V)

Figure 2.16. Nitrogen concentration effects upon the current-voltage characteristic of zirconium sputtering measured with double Langmuir probe.

42 Electron temperature can be determined from double probe evaluation by examining the slope of the curve through zero as 1 Slope = [2.16] kTe or estimated from the voltage difference between the “knees” ∆ T = [2.17] e 4 Utilizing electron temperature, electron density can be calculated

− 1 i  2kT  2 p  e  ne ≈   [2.18] 0.4eAp  mi  where Ap is the area of the probe, mi is ion mass, and ip is probe current and assuming a quasi-neutral plasma ne ≈ ni, the ion flux can be calculated from

1  kT  2  e  Γi = niυi = ni   [2.19]  mi  1/2 and υi is the Bohm velocity (kTe/mi) .

The data obtained from double probe experimentation for the deposition of ZrN (4% N2) and Zr3N4 (100% N2) and equations 2.16-2.18 is summarized in table 2.3.

Table 2.3. Double probe characterization results.

4% N2 100% N2

Te 3.18 eV 3.11 eV 15 3 15 3 ne 3.38 * 10 /m 3.44*10 /m 18 2 19 2 Γi 9.67*10 /m s 1.12*10 /m s

It was found that the sputtering of higher ZrN resulted in a slightly lower electron temperature, with little to no change in electron density. It was found that there was negligible change in Te upon change in nitrogen concentration from 4 to 100%.

43 The difference noted between single and double probe measurements is due to the method of collection. The single probe is biased with respect to ground and collects the true current which is directly related to charge carrier density and electron temperature. The double probe is floating and collects only circuit current. The error in density measurements of the double probe may as high as 40-50%, but is a useful technique for quickly determining electron temperature.

2.5 Thin Film Characterization

Sheet resistance (Rs) was measured utilizing a four point probe apparatus comprised of Keithley 224 programmable current source, HP 3456A digital voltmeter and a Signatone four point probe. The probe injects current from two electrodes across the film with two other electrodes measuring the voltage collected as seen in the figure 2.17.

V

Figure 2.17. Schematic of four point probe set up.

If the spacing of the probes is constant and the edges of the sample are more than four times this distance away from the measurement sight, sheet resistance (Rs) can be found from

Vs Rs = 4.532 ∗ [2.20] Is which is the two dimensional sheet resistance (Ω/ ) and is the resistance between two 16 electrodes on opposite sides of a theoretical square . Is is the current injected and Vs is

44 the voltage measured. Film thickness (d) and resistivity (ρ) are related to sheet resistance

(Rs) through. ρ R = [2.21] s d Therefore, if the thickness of the film is known, and the sheet resistance has been measured, the resistivity can be calculated. All of the film thicknesses presented were measured with a Dektac 3030 profilometer from silicon samples that had been partially masked during deposition to create a step. The reported accumulation rates were calculated from these measurements. Profilometer scans were run three times and an average thickness was obtained.

Thin film crystal orientation was measured with a Rigaku diffractometer using Cu Kα radiation (λ = 1.543 Å). θ-2θ scans were performed in order to determine the orientation of the films. A Hitachi S-3200N environmental scanning electron microscope was used to examine the structure of several of the films. Accelerating voltages were between 10 to 25 keV. A Topcon transmission electron microscope (TEM) was utilized to obtain high resolution TEM images of atomic structure

2.5.1 Raman Spectroscopy17 Crystals can be thought of as a group of balls connected by springs, which can vibrate with a frequency dependent upon the mass of the balls and the strength of the springs. Diatomic molecules (two balls/one spring) have only one fundamental vibrational frequency. These vibrational frequencies are high, 1012-1014 Hz, or the infrared region. The Raman spectrum comes from the indirect coupling of high-frequency radiation with the electrons that make up the chemical bond. Therefore, Raman spectroscopy is sensitive to atomic arrangement and chemical bonding, but is used primarily for structural characterization, because it is more sensitive to bonding than chemical composition. Raman spectroscopy is performed by exposing a material, or thin film, to an intense monochromatic light beam. The electric field of the incident radiation distorts the electron cloud and when the wave passes, the electron cloud relaxes and the stored

45 energy is reradiated. Rayleigh scattering occurs when energy is given off at the same frequency as the incident light. A small portion of this energy is transferred to the sample and is said to excite the vibrational modes. The separation of the Raman lines from the Rayleigh line is a direct measure of the vibrational frequencies of the sample. The Raman process that excites molecular and crystal vibrations is called Stokes scattering, and the process that annihilates existing vibrations is known as anti-Stokes scattering. The anti-Stokes scattering depends upon existing thermal vibrations and is highly temperature dependent. Thus, anti-Stokes scattering is rarely measured, and it is the Stokes intensities that are typically measured as they are only weakly temperature dependent. Figure 2.18 shows a typical spectrum obtained from Raman analysis.

Figure 2.18. Typical Raman spectroscopy plot17.

Scattering from materials with metallic properties is restricted by prohibitive selection rules and low penetration depth due to high reflectivity18. In a perfect fcc crystal, every ion is at a site of inversion symmetry and first-order Raman spectra are forbidden18. However, ZrN is inherently defective, and has been found to show an induced Raman effect of the first order19.

46 2.6 Gallium Nitride Gallium nitride (GaN) was first reported by Johnson et al. in 1928 and was referred to as “Gallic nitride”20. Gallium nitride is member of the III-V semi-conductors family that consists of three commercially valuable materials and their alloys, for device applications: AlN, GaN, and InN. GaN is a direct wide band gap semiconductor that is today finding uses in blue and UV light emitters and detectors, secure space communications, and is currently being investigated for short wavelength laser applications20. It has been found that Schottky barriers on GaN have barrier heights depending upon the difference in work function between the contact and GaN21. A metal with a work function less than that of GaN will yield an ohmic contact and a higher work function will yield a rectifying contact21. GaN is slated to be the substrate of choice for the future high temperature, high power, high frequency devices21. Methods of film deposition include molecular beam epitaxy (MBE)22,23,24, chemical vapor deposition (CVD)25,26, reactive sputtering27,28, pulsed laser deposition (PLD)29,30,31, metal-organic chemical vapor deposition (MOCVD), high vapor phase epitaxy (HVPE)32,33. The III-V semiconductors form a continuous alloy system capable of generating light from the red to UV.20,29 GaN has a thermal conductivity (κ) of 1.3 W/cm K, a dielectric 20 constant of 9.5, a band gap (Eg) of 3.4 eV . GaN lattice constants and thermal expansion information can be found in table 2.4.

Table 2.4. Gallium nitride material properties.20 Lattice spacing Coefficient of thermal expansion (α) a: 3.189Å 5.59 * 10-6/K c: 5.185Å 3.17 * 10-6/K

GaN is the basis for blue and blue-green light emitting diodes (LED’s). The advent of the blue LED, the last of the primary color LED’s to be developed, will be incorporated into green and red LED’s to produce white light LED’s that will eventually replace the light bulb. The increased lifetime, decreased power consumption, and

47 reduced heat generation of LED’s allows them to replace traditional tungsten bulbs in application. The most noticeable of which is in stoplights across the nation.

2.6.1 Lattice and Thermal Expansion Mismatch Gallium nitride (GaN) grown as thin films suffers from a lack of suitably lattice and thermally matched substrates. Buffer layers of aluminum nitride (AlN), zinc oxide (ZnO) have been investigated34,35,36 to reduce the mismatch. Substrates of silicon carbide (SiC), gallium arsenide (GaAs), and sapphire are also widely used37,38,39,40. Sapphire is currently the substrate of choice since it is widely available in larger wafer sizes, is thermally stable, has a hexagonal crystal structure, is easily cleaned and relatively inexpensive. Large thermal and lattice mismatches between substrate and film often result in strained GaN films with large defect densities41. The high cost of these substrates, with respect to silicon, has pushed many research groups to attempt to develop methods to grow high quality GaN films using silicon substrates29,42,43,44,45,46,47,48. Table 2.5 summarizes lattice mismatch between several substrates/buffer layers and GaN thin films. Mismatch calculations were performed using equation 2.21 given by Pashley49 who states that epitaxial growth occurs when the lattice mismatch between substrate and film is less than 15%. The relation is obtained through

 as − af  Lattice Mismatch =   *100 [2.22]  as  49 where af and as are the film and substrate lattice spacing, respectively .

48 Table 2.5. Lattice mismatch between substrate materials/buffer layers to gallium nitride. Substrate/ a – axis (111) GaN Buffer layer Structure (Å) spacing (Å) % mismatch a AlN Wurtzite 3.11(20) N/A 2.5 GaAs Cubic 5.65(20) 4.00 20.3 6H-SiC Hexagonal 3.08(20) N/A 3.5 Sapphire (0001) Corundum 4.75(20) N/A 13.54 ZnO Wurtzite 3.25(20) N/A 1.88 Silicon Diamond Cubic 5.43(20) 3.84 17.0 ZrN NaCl 4.58(54) 3.24 1.57

The lattice mismatch between sapphire and gallium nitride is 13.54%, between silicon and GaN is 17%. Silicon with a (111) orientation has been investigated as an epitaxial substrate50,51. The thermal expansion coefficient of silicon, sapphire and ZrN are shown in Table 2.6

Table 2.6. Thermal expansion coefficients of silicon, sapphire (0001) and ZrN. Substrate Thermal Expansion Coefficient Silicon 2.33 * 10-6/°C(52) Sapphire 7.50 * 10-6/°C(53) ZrN 7.24 * 10-6/°C(54)

The thermal expansion coefficient of ZrN is closer to GaN than either sapphire or silicon.

49 2.7 References

1United States Patent # 2,146,025

2K. Wasa and S. Hayakawa, Handbook of Sputter Deposition Technology: Principles, Technology and Applications, Noyes, New Jersey, 1992.

3K. Wasa and S. Hayakawa, Review of Scientific Instruments, v. 40, p. 693, 1969.

4R. K. Waits, J. Vac. Sci. Technol., v. 15, p.179, 1978.

5S. M. Rossnagel, Handbook of Plasma Processing Technology: Fundamentals, Etching, Deposition, and Surface Interactions, ed. S. M. Rossnagel, J. J. Cuomo, W. D. Westwood, Noyes, New Jersey, 1992.

6J. A. Thornton, Deposition Technologies for Films and Coatings: Development and Applications, ed. R. F. Bunshah, Noyes Publications, NJ, 1982.

7J. A. Thornton, J. Vac. Sci. Technol., v. 15, pg. 171, 1978.

8D. L. Smith, Thin-Film Deposition: Principles and Practice, McGraw-Hill Inc., New York, 1995.

9W. D. Westwood, S. Maniv, and P. J. Scanlon, J. Appl. Phys., v. 54, p. 6841, 1983.

10J. R. Roth, Industrial Plasma Engineering: Principles, Institute of Physics Publishing, London, 1995.

11P. J. Kelly and R. D. Arnell, Vacuum, v. 56, p. 159, 2000.

12P. J. Kelly, P. S. Henderson, and R. D. Arnell, G. A. Roche, D. Carter, J. Vac. Sci. Technol. A, v. 18, p. 2890, 2000.

13F. Chen, Plasma Diagnostic Techniques, ed. R. H. Huddlestone, S. L. Leonard, Academic Press, New York, 1965.

14L. Schott, American Vacuum Society Classics: Plasma Diagnostics, ed. W. Lochte- Holtgreven, AIP Press, New York, 1995.

15J. W. Cho, unpublished results.

16http://four-point-probes.com/short.html

17W. B. White, Encyclopedia of Materials Characterization, ed. C. R. Brundle, C. A. Evans Jr., managing editor L. E. Fitzpatrick, Butterworth-Heinemann, Boston, 1992.

50 18C. P. Constable, J. Yarwood, W. –D. Münz, Surface and Coatings Technology, v. 116, pg. 155, 1999.

19W. Spengler, and R. Kaiser, Solid State Communications, v. 18, pg. 881, 1976.

20S. Strite and H. Morkoç, J. Vac Sci. Technol. B, v. 10, pg. 1237, 1992.

21J. S. Foresi, and T. D. Moustakas, Appl. Phys. Lett. v. 62, pg. 2859, 1993.

22N. Grandjean and J. Massies, Appl. Phys. Lett., v. 71, pg. 1816, 1997.

23N. Grandjean and J. Massies, Appl. Phys. Lett, v. 71, pg. 1078, 1998.

24R. C. Powell, N. –E. Lee, Y. –W. Kim, and J. E. Greene, J. Appl. Phys., v. 73, pg. 189, 1993.

25M. Ilegems, J. Cryst. Growth, v. 13/14, pg. 360, 1972.

26K. Naniwae, S. Itoh, H. Amano, K. Itoh, K. Hiramatsu, and I. Akasaki, J. Cryst. Growth, v. 99, pg. 381, 1990.

27J. Ross, M. Rubin, Materials Letters, v. 12, pg. 215, 1991.

28P. Singh, J. M. Corbett, J. B. Webb, S. Charbonneau, F. Yang, and M. D. Robertson, J. Vac. Sci. Technol. A, v. 16, pg. 786, 1998.

29S. C. Jain, M. Wilander, J. Narayan, and R. Van Overstraeten, J. Appl. Phys., v. 87, pg. 965, 2000.

30R. D. Visute, H. Wu, and J. Narayan, Appl. Phys. Lett., v 67, pg. 1549, 1995.

31R. F. Xiao, X. W. Sun, and H. S. Kwok, Applied Surface Science, v. 127-129, pg. 425, 1998.

32L. Akasaki, and H. Amano, Tech. Dig. Int. Electron Devices, Meet, v. 69, pg. 231, 1996.

33S. D. Hersee, J. C. Ramer, and K. J. Malloy, MRS Bull. July, v. 22, pg. 45, 1997.

34Yasuo Ohba and Susumu Iida, Jpn. J. Appl. Phys. Part 2, v. 41, pg. L615, 2002.

35Shulin Gu, Rong Zhang, Jingxi Sun, Ling Zhang, and T. F. Kuech, Appl. Phys. Lett., v. 76, pg. 3454, 2000.

36X. W. Sun, R. F. Xiao, and H. S. Kwok, J. Appl. Phys., v. 84, pg. 5776, 1998.

51

37D. D. Koleske, R. L. Henry, M. E. Twigg, J. C. Culbertson, S. C. Binari, A. E. Wickenden, and M. Fatemi, Appl. Phys. Lett., v. 80, pg. 4372, 2002.

38T. M. Katona, M. D. Craven, P. T. Fini, J. S. Speck, and S. P. DenBaars, Appl. Phys. Lett. V. 79, pg. 2907, 2001.

39Mitsuru Funato, Shuichiro Yamamoto, Kiyohiro Kaisei, Koichiro Shimogami, Shizuo Fujita, and Shigeo Fujita, Appl. Phys. Lett., v. 79, pg. 4133, 2001.

40M. J. Manfra, N. G. Weimann, J. W. P. Hsu, L. N. Pfeiffer, K. W. West, and S. N. G. Chu, Appl. Phys. Lett., v. 81, pg. 1456, 2002.

41J. Keckes, J. W. Gerlach, R. Averbeck, H. Riechert, S. Bader, B. Hahn, H. –J. Lugauer, A. Lell, V. Härle, A. Wenzel, and B. Rauschenbach, Appl. Phys. Lett., v. 79, pg. 4307, 2001.

42http://www.nitronex.com

43M. A. Khan, Q. Chen, J. Yang, M. Z. Anwar, M. Blasingame, and M. S. Shur, Tech. Dig. Int. Electron Devices Meet., v. 96, pg. 27, 1996.

44R. D. Vispute, J. Narayan, H. Wu, and K. Jagannadham, J. Appl. Phys. Lett., v. 77, pg. 4724, 1995.

45A. V. Blant, T. S. Cheng, C. T. Foxon, J. C. Bussey, S. V. Novikov, and V. V. Tret’yakov, III-V Nitrides: Symposium held December 2-6, 1996, ed. F. A. Ponce, T. D. Moustakas, I. Akasaki, MRS, Pittsburgh, PA, 1997.

46S. C. Binari, J. M. Redwing, and W. Kruppa, Electron Lett., v. 33, pg. 242, 1997.

47K. A. Brown, S. A. Ustin, L. Lauhon, and W. Ho, J. Appl. Phys., v. 79, pg. 7667, 1996.

48A. Barski, U. Rössner, J. L. Rouviere, and M. Arlery, MIJ-NSR, v. 1, article 21, 1996.

49D. W. Pashley, Epitaxial Growth: Part A, ed. J. W. Matthews, Academic Press, New York, 1975.

50S. Barnett, L. Hultman, J. –E. Sundgren, F. Ronin, and S. Rohde, Appl. Phys. Lett., v. 53, pg. 400, 1988.

51W. J. Meng, J. Heremans, and Y. T. Cheng, Appl. Phys. Lett., v. 59, pg. 2097, 1991.

52A. G. Milnes, and D. L. Feucht, Heterojunctions and Metal-semiconductor Junctions, Academic, New York, 1972.

52

53T. Kozawa, T. Kachi, H. Kano, H. Nagase, N. Koide, and K. Manabe, J. Appl. Phys., v. 77, pg. 4389, 1995.

54L. E. Toth, Transition Metal Carbides and Nitrides Academic, New York, 1971.

53 3.0 ZIRCONIUM NITRIDE

3.1 Properties of IV-B Transition Metal Nitrides Best known for extremely high hardness and aesthetically pleasing gold coloration: zirconium nitride (ZrN) is a IVB-V compound that exhibits properties desirable for a variety of industrial and consumer applications. The brilliant golden color and resistance to wear make ZrN extremely desirable as decorative coatings, and has achieved Federal Drug Administration (FDA) approval for use in surgical applications and food preparation. High-end watches and the jewelry industry use ZrN for coatings with the look of gold and increased wear resistance. The tool and die industry uses these coatings for tool wear protection. Table 3.1summarizes the major technical aspects of ZrN.

Table 3.1 Properties of ZrN ZrN Hardness (Knopp) 4510 kg/mm2(1) Melting Temp. 2950 °C(2) Density 7.09 g/cc(3) Resistivity 13.6µΩ-cm(4) Lattice Parameter 4.58 Å(5) (6) Superconducting Temp. 10 K

The Hägg rule states that the structure of transition metal hydrides, carbides, nitrides, and borides is determined by calculating the ratio of atomic radii of the interstitial element to the transition metal, and then comparing that to 0.597. If the ratio is less than 0.59, simple structures will be formed. If larger, then complex structures form. ZrN satisfies the Hägg rule and crystallize in a B1-NaCl crystal structure.

Zirconium nitride has recently gained attention from the microelectronics industry for exploitation of its electrical and chemical properties. As the feature size of multi-layer

54 films decreases, the possibility for interactions between materials increases. It has been shown that barrier layers can be used to prevent these materials interactions2. Today, aluminum or copper is used as the metal electrode in metal oxide semiconductor (MOS) integrated circuits and to prevent the low temperature inter-diffusion of aluminum or copper and silicon during contact sintering, passivation or device packaging, barrier layers are used2,8. Zirconium nitride has been produced by a variety of deposition techniques: ion assisted deposition9, reactive magnetron sputtering10, dual ion beam sputtering11,12, ion plating12, and plasma enhanced chemical vapor deposition (PECVD)13.

3.2 Properties of Higher Zirconium Nitride Juza et al. are credited with the discovery of a metastable phases of ZrN in 1964 by ammonolysis of zirconium tetraiodide14, in which the stoichemistry favored that of the (15) nitrogen such that the ratio of N/Zr>1.2 or the chemical formula Zr3N4 . This new metastable phase of ZrN, referred to as “higher nitride”6, was found to the same structure (iso-structural) and the same chemical make up (iso-chemical) as stoichiometric ZrN. The electrical and optical properties, however, are opposite that of stoichiometric ZrN(16). The higher nitride phase is a semiconductor with a transparent straw coloration. Juza4 et.al. discovered that Zr3N4 had no metallic conductivity, was diamagnetic with a density 3 (17) of 5.9 g/cm . The band gap of Zr3N4 has been calculated to be 2.2 eV , with others calculating an optical band gap of 3.1 eV(13). Schwarz et al. have proposed a structural model consisting of a highly defective NaCl structure Zr vacancies18, shown in figure 3.1.

55

(18) Figure 3.1. Proposed structure for Zr3N4

R. Manaila et al. have found that in stoichiometric ZrN and TiN, the nitrogen occupies octahedral sites, in Zr3N4 films, excess nitrogen seems to locate randomly on small tetrahedral interstices of the cubic network, inducing lattice dilation14.

3.3 Deposition of Zirconium Nitride Zirconium nitride was deposited by dc reactive magnetron sputtering utilizing the magnetron system described in Chapter 2. DC deposition conditions can be seen in table 3.2. Table 3.2. Deposition conditions of ZrN. Argon 0-98% Nitrogen 2-100% Pressure 3-25 mTorr Cathode 0.5-1.0 Current Amperes Temperature 15-500°C

56 3.4 GaN growth on ZrN Some authors19-23 have reported on the (111) preferred orientation of ZrN and TiN thin films. The (200) plane is the lowest energy plane in TiN19, and (111) oriented films are thought to arise from kinetics of moderate ion bombardment19. Thin films with a (111) orientation were investigated as seed layers for the epitaxial growth of GaN(24). The relationship between the (111) orientation of cubic structure to the hexagonal structure can be seen in figure 3.2.

Figure 3.2. NaCl (111) orientation and it’s relation to the hexagonal structure.

57 In examining the lattice spacing of (111) oriented ZrN thin films, we find that lattice mismatch between (111) ZrN and c-axis GaN is 1.6%. We hypothesized that ZrN used a buffer layer will result in improved c-axis oriented bulk GaN as compared silicon. Ito et al. have recently published a patent on the growth of bulk GaN on (111) oriented transition metals25.

58 3.5 References 1W. D. Sproul, J. Vac. Sci. Technol. A, v. 4, pg. 2874, 1986.

2M. Wittmer, J. Vac. Sci. Technol. A, v. 3, pg. 1797, 1985.

3V. R. Juza, A. Rabenau, and I. Nitschke, Z. anorg. Allg. Chemie, v. 332. p. 1, 1964.

4L. Krusin-Elbaum, and M. Wittmer, Thin Solid Films, v. 107, p. 111, 1983.

5M. Konuma, and O. Matsumoto, Journal of the Less-Common Metals, v. 56, pg. 129, 1977.

6K. Schwarz, A. R. Williams, J. J. Cuomo, J. H. E. Harper, and H. T. G. Hentzell, Phys. Rev. B, v. 32, p. 8312, 1985.

7J. –E. Sundgren, and H. T. G. Hentzell, J. Vac. Sci. Technol. A, v. 4, p. 2259, 1986.

8M. B. Takeyama, A. Noya, and K. Sakanishi, J. Vac. Sci. Technol. B, v. 18, pg. 1333, 2000.

9S. Horita, T. Tujikawa, H. Akahori, M. Kobayashi, T. Hata, J. Vac. Sci. Technol. A, v. 11(5), pg. 2452, 1993.

10S. Jin, X. Y. Wen, Z. X. Gong, and Y. C. Zhu, J. Appl. Phys., v. 74, p. 2886, 1993.

11T. Pichon, A. Girardeau, F. Straboni, P. Lignou, and J. P. Guérin, Applied Surface Science, v. 150, p. 115, 1999.

12S. Inoue, K. Tominaga, R. P. Howson, and K. Kusaka, J. Vac, Sci. Technol. A, v. 13, p. 2808, 1995.

13L. M. Atagi, J. A. Samuels, D. C. Smith, and D. M. Hoffman, Mat. Res. Coc. Symp. Proc., v. 410, pg. 289, 1996.

14R. Manaila, D. Biro, A. Devenyi, D. Fratiloiu, R. Popescu and J. E. Totolici, Applied Surface Science, v. 134, pg. 1, 1998.

15V. R. Juza, A. Gabel, H. Rabenau, and W. Klose, Z. anorg. Allg. Chem., v. 329, p. 136, 1964.

16B. O. Johansson, H. T. G. Hentzell, J. M. E. Harper, and J. J. Cuomo, J. Mater. Res., v. 1, pg. 442, 1986.

59 17Prieto, F. Yeubero, E. Elizalde, and J. M. Sanz, J. Vac. Sci. Technol. A, v. 14, p. 3181, 1996.

18Schwarz, A. R. Williams, J. J. Cuomo, and J. M. E. Harper, Phys. Rev. B. v. 32, pg. 8312, 1985.

19C. –H. Ma, J. –H. Huang, and H. Chen, Surface and Coatings Technology, v. 133-134, pg. 289, 2000.

20J. H. Je, D. Y. Noh, H. K. Kim, and K. S. Liang, J. Appl. Phys., v. 81, pg. 6126, 1997.

21J. E. Greene, J. –E. Sundgren, L. Hultman, I. Petrov, and D. B. Bergstrom, Appl. Phys. Lett., v. 67, pg. 2928, 1995.

22L. Hultman, J. –E. Sundgren, J. E. Greene, D. B. Bergstrom, and I. Petrov, J. Appl. Phys., v. 78, pg. 5395, 1995.

23J. W. Gerlach, U. Preckwinkel, H. Wengenmair,T. Kraus, and B. Rauschenbach, Appl. Phys. Lett., v. 68, pg. 2360, 1996.

24H. Bairamov, O. Gürdal, A. Botchkarev, H. Morkoç, G. Irmer, and J. Monecke, Phys. Rev. B, v. 60, pg. 16741, 1999.

25J. Ito et al., United States Patent #6,426,512, 2002.

60 4.0 OPTIMIZATION OF REACTIVE MAGNETRON SPUTTERING OF ZIRCONIUM NITRIDE

4.1 DC Reactive Magnetron Sputtering of Zirconium Nitride and Gallium Nitride Thin Films DC magnetron sputtering of zirconium nitride and higher zirconium nitride was performed in a 50 liter, stainless steel vacuum chamber evacuated with a CTI CryoTorr 10. A schematic of the deposition system can be seen in Figure 2.1 and is discussed in detail in chapter two of this work. Samples were cleaned with acetone, ethyl alcohol, and blown dry with nitrogen. Etched silicon samples were acid etched in a 10/1 de-ionized water/hydrofluoric acid solution for 1 minute to remove any native oxide from the surface. Samples were transferred to vacuum within 5 minutes to insure the cleanliness of the surfaces. All experiments were performed with high purity (99.999%) argon and nitrogen gas. Deposition pressures were measured with a Baratron capacitance manometer. The chamber was pumped down to at least 5*10-7 Torr before deposition. One and three inch silicon (100) and (111) oriented wafers were used as substrates, along with other samples (glass, Al2O3) depending upon the particular experiment. A radiant heater with a maximum temperature of 500°C was used to degas the substrates during pump down and as a substrate heater during deposition. Substrate temperatures were measured by a thermocouple inserted into the molybdenum substrate. The zirconium target (purity 99+ %, including hafnium), manufactured by Pure Tech Inc., was pre- sputtered for 30 minutes with the substrates shuttered to remove any contamination upon the surface of the cathode and getter pump the chamber. Sputtering power, substrate temperature, chamber pressure and reactive gas concentration were varied in order to optimize the growth of zirconium nitride and higher zirconium nitride. Pulsed DC sputtering of gallium nitride was performed in a 25 liter magnetron system described in chapter 2. Chamber pressures of 6*10-7 were reached prior to deposition. The substrates were heated to deposition temperature while the gallium target was pre- sputtered for 30 minutes with the substrates shuttered. All substrates used were pre- deposited with buffer layers of ZrN.

61 4.2 Processing Condition Effects upon the Resistivity of Zirconium Nitride Thin Films

4.2.1 Effect of Reactive Gas Concentration upon the Resistivity of ZrN Thin Films Thin films of zirconium nitride were deposited by direct current sputtering at a constant chamber pressure of 5 mTorr, 0.5 amps, and 15°C on glass substrates and characterized as a function of increasing reactive gas concentration. The reactive gas concentration was determined from the ratio of partial pressures of gases. The sheet resistance of all films was measured by the four point probe technique described earlier. The thickness of the samples was measured with a Dectak 3030 profilometer. As reactive gas concentration is increased the resistivity also increases. Figure 4.1 show the resistivity increase as films of ZrN move from stoichiometric to insulating

Zr3N4. The lowest resistivity occurred for films deposited at a nitrogen concentration of 4%.

2500

2000 y = 41.371x R2 = 0.9905 1500

1000

500

0 Resistivity (micro-ohm-cm) 0 204060 Nitrogen concentration (%)

Figure 4.1. Resistivity of zirconium nitride as a function of reactive gas concentration at constant current.

62 4.2.2 Effect of Deposition Pressure upon the Resistivity of ZrN Thin Films The resistivity of ZrN thin films was measured as a function of sputtering pressure. As the deposition pressure is increased, the resistivity of the films is increased as seen in figure 4.2.

34000 4 3 2 29000 y = -1.9347x + 77.476x - 881.32x + 3782.3x - 5322.1 R2 = 1 24000

19000

14000

9000

4000 Resistivity (micro-ohm-cm) -1000 0 2 4 6 8 10121416182022 Deposition Pressure (mTorr)

Figure 4.2. Resistivity as a function of sputtering pressure at 500C, 4% nitrogen concentration, and 0.5 amp. cathode current on Si(111).

The resistivity of the films was found to increase as a function of deposition pressure. As deposition pressure is increased, the flux of higher energy bombarding species is reduced, due to the reduction in mean free path (MFP). The reduction in MFP, will increase the loss of kinetic energy through gas phase collisions. Low energy ion bombardment is known to change the properties of sputtered films and the addition of energy to growing films reduces impurity concentrations and results in a more crystalline higher conductivity film.

63 The drastic resistivity increase observed above a deposition pressure of 10 mTorr could also be attributed to a change in the conductance of the pumping system required to achieve pressures greater than 10 mTorr. The gate valve of the system was lowered to an almost closed position to reach these pressures. The partial pressure ratio was kept the same, however, the flow rates had to be significantly reduced.

4.2.3 Effect of Cathode Current upon the Resistivity of ZrN Thin Films The resistivity of ZrN thin films as a function of cathode current was investigated. ZrN thin films were grown at 0.5 and 1.0 amp cathode currents, on (100) and (111) oriented silicon wafers, at 5 mTorr, with a nitrogen concentration of 4%. As can be seen in figure 4.3, the resistivity for films grown on both (100) and (111) oriented silicon, shows a decrease in resistivity when cathode current was increased.

40

35 (111) y = -20.835x + 42.247 30

25

20 (100) 15 y = -8.8601x + 23.181 10

5

Resistivity (micro-ohm-cm) 0 0 0.2 0.4 0.6 0.8 1 1.2 Cathode current (Amps)

Figure 4.3. Resistivity as a function of cathode current on silicon (111) and silicon (100) substrates.

The difference in the results shown in Fig. 4.4 are attributed to the resistivity difference between the (111) and (100) oriented substrates, which was 0.015 ohm-cm and 0.01 ohm-cm, respectively. A decrease in resistivity as a function of cathode current is

64 noticed for both substrates and is of the same order indicating that there is a decrease in resistivity for films deposited from 0.5 to 1.0 amp. The increase in cathode current results in increased energetic bombardment of the substrate. The addition of energy to the substrate during growth allows for increased atomic ordering and results in films of reduced resistivity.

4.2.4 Effect of Deposition Temperature upon the Resistivity of ZrN Thin Films The resistivity of ZrN thin films was measured as a function of deposition temperature and can be seen in figure 4.4. The films were deposited at a 0.5amps, 5 mTorr, and at a 4% nitrogen concentration.

450 400 350

300 y = -0.6308x + 412.66 R2 = 0.9807 250 200 150 100

Resitivity (micro-ohm-cm) 50 0 0 100 200 300 400 500 600 Deposition Temperature (degrees C)

Figure 4.4. Effect of ZrN deposition temperature on resistivity for films deposited at 0.5 ampere cathode current, 5mTorr, and 4% nitrogen concentration.

The reduction in resistivity as a function of deposition temperature is attributed to the additional energy available at the substrate, permitting increased atomic ordering upon the substrate surface and results in a more crystalline film with low resistivity.

65 4.3 Processing Condition Effects Upon the Crystallinity of Zirconium Nitride Thin Films

4.3.1 Effect of Deposition Pressure upon the Crystallinity of ZrN Thin Films The effect of deposition pressure upon the crystallinity of zirconium nitride was investigated over a range of pressures, 3-20 mTorr, keeping all other variables constant. At lower pressures, sputtered particles have few gas phase collisions between the target and substrate. Sputtering at low pressures results in more energetic deposition as particle energy is not lost to collisions. As deposition pressure increases, the mean free path decreases, and the sputtered particles will have gas phase collisions, reducing the energy with which they arrive at the substrate. This reduction in arrival energy will limit adatom mobility on the substrate and result in a less crystalline films. The effects of pressure upon crystallinity of zirconium nitride films can be seen in figure 4.5. Experiments were performed at 500°C, with a nitrogen concentration of 4%, a constant target current of 0.5 amperes, and a constant target to substrate distance of 3 inches. Deposition time for all samples was 30 minutes, which produced films on the order of 400 nm in thickness, or an accumulation rate of 130 Å/min. The silicon orientations of (100) and (111) were etched in a 10:1 solution of de-ionized water and hydrofluoric acid (HF) for 1 minute prior to pump down to remove the native oxide from the surface.

66 ZrN (111) (200) 25 mTorr

20 mTorr

15 mTorr

Intensity (arbitrary units) (arbitrary Intensity 10 mTorr 5 mTorr 3 mTorr

30 35 40 45 50 55 60 65 70 75 80 Two-theta (degrees)

Figure 4.5. Effect of pressure upon crystallinity of ZrN on (111) silicon substrates at 500° C, 0.5 ampere cathode current and 4% nitrogen.

The results are in good agreement with Wu et al. who have shown that ZrN grows preferentially (111) even on (100) oriented silicon substrates1. The (111) and (100) zirconium nitride peaks are the only listed as the other peaks are second order reflections of ZrN ((311) at 67.85 and (222) at 71.30) and Si ((222) forbidden reflection at 59.15). As can be seen from figure 4.5, the pressure at which produces the greatest (111) crystallinity occurs is at 15 mTorr for (111) substrates.

Figure 4.6 shows the effects of deposition pressure upon the crystallinity of zirconium nitride deposited upon 10:1 (DI-H2O:Hydroflouric acid) etched (100) silicon substrates. The sample deposited at 10 and 15 mTorr show only a zirconium nitride (111) orientation indicating that the film consists of grains of (111) oriented ZrN. As mentioned previously, the (111) direction is the preferred growth direction for zirconium nitride at low cathode currents.

67 (111) ZrN (200) 25 mTorr

20 mTorr

15 mTorr

10 mTorr

5 mTorr Intensity (arbitrary units) (arbitrary Intensity

3 mTorr

30 35 40 45 50 55 60 65 Two-theta (degrees)

Figure 4.6. Effect of deposition pressure upon crystallinity of zirconium nitride deposited upon silicon (100) at 500°C, 4% nitrogen and 0.5 ampere cathode current.

The samples deposited at 10 mTorr showed the highest (111) crystallinty for films deposited on etched (100) silicon. As can be seen, at low cathode currents, films of completely oriented (111) ZrN are produced. Figure 4.6 and 4.7 show that at low pressures, the energy of bombarding species is too high and damages the crystal structure of the growing film. As the pressure is increased, mean free path decreases, reducing the energy of bombarding ions to conditions that are ideal for the growth of oriented films. As the pressure is further increased, ion energy is lost through collisions and not enough energy is available at the substrate for film orientation.

4.3.2 Effect of Cathode Current upon the Crystallinity of ZrN Thin Films ZrN thin films were reactively sputtered on silicon substrates at 500°C, 5 mTorr, 4% nitrogen concentration at 2 different cathode currents. The crystallinity of the films show a (111) preferred growth orientation at 0.5 ampere constant current and a (100) preferred growth orientation at 1.0 ampere cathode current. At higher cathode currents, an increase in plasma density will be noticed, resulting in an increased deposition rate and the increase in energy of atoms arriving at the substrate as shown by the Langmuir probe

68 data. The results shown in figure 4.8 are attributed to an ion channeling effect2. Geometrically, the (200) plane is more open to ions than the (111) plane and as the energy of the bombarding ions increased the (111) oriented grains are damaged by these collisions2. The (200) oriented grains are subjected to less radiation damage and those that survive will act as seed layers for the growth of (200) oriented grains. The results displayed in figure 4.7 could also be attributed to an increased deposition rate as the rate at which atomic species arrive at the substrate is greater than the time required for adatoms to arrange in the preferred (111) orientation.

(200)

(111)

1.0 Ampere Intensity (arbitrary units) 0.5 Ampere

30 35 40 45 50 55 Two theta (degrees)

Figure 4.7. Effect of cathode current upon crystallinity of ZrN on (100) Si, 500 °C, 5mTorr, 4% nitrogen concentration.

An identical experiment using (111) oriented silicon substrates was conducted to determine the effect of a more closely matched substrate and to test the idea that energy plays a dominant role in the crystallization of ZrN. The results of which can be seen in figure 4.8.

69 (200)

(111) 1.0 Ampere Intensity (arbitrary units) 0.5 Ampere

30 35 40 45 50 55 Two-theta (degrees)

Figure 4.8. Effect of cathode current on the crystallinity of ZrN reactively sputter deposited on silicon (111).

The sample sputtered at 1.0 amperes crystallized with a (100) preferred orientation for ZrN films deposited on (111) and (100) oriented silicon substrates. This is a result of the energy addition to the substrate which pushes the ZrN thin films to orient with the lowest surface energy (200) oriented grains.

4.3.3 Effects of Deposition Temperature upon the Crystallinity of ZrN Thin Films The effect of deposition temperature upon the crystallinity of ZrN thin films was investigated. Films were reactively magnetron sputtered onto silicon (100) oriented substrates. Temperature dependent crystallinity investigations were performed on (111) oriented silicon and can be found in figure 4.9. As the deposition temperature is increased, the mobility of adatoms on the surface also increased. This increased mobility promotes atomic ordering and growth along a direction corresponding to substrate orientation.

70 (111) ZrN

(200)

500 C

Intensity (arbitrary units) 400 C

300 C

20 25 30 35 40 45 50 55 60 65 Two-theta (degrees)

Figure 4.9. Effects of temperature upon the crystallinity of ZrN thin films deposited on (111) silicon substrates.

4.3.4 Effect of Substrate Orientation upon the Crystallinity of ZrN Thin Films The effect of substrate orientation upon the crystallinity of ZrN thin films was investigated. Samples of (100) and (111) oriented silicon wafers were used to examine the effect upon the crystallinity of films deposited at 500°C, 0.5 amp, 5mTorr, and 4% nitrogen. The results are shown in figure 4.10.

71 (111) ZrN

(111) Silicon Intensity (arbitrary units) (100) Silicon

30 35 40 45 50 Two-theta (degrees)

Figure 4.10. Substrate orientation upon the crystallinity of ZrN thin films deposited at

500°C, 4% N2, 5 mTorr.

X-ray diffraction is mainly used for phase identification, but by comparing the diffraction intensities from substrates of (100) and (111) orientation we see that, as expected, the growth (111) material upon a substrate of (111) orientation results in higher quality (111) oriented grains. There is an effect of substrate orientation upon thin film preferred orientation, but this effect was not found to be a driving force for the preferred orientation.

4.3.5 Effect of Film Thickness upon the Crystallinity of ZrN Thin Films ZrN thin films were deposited on (111) oriented etched silicon wafers at 10 mTorr, 0.5 amps, 500°C and a nitrogen concentration of 4% at thicknesses of 600 nm and 1200 nm. X-ray diffraction was performed to investigate the effect of film thickness upon the (111) orientation of ZrN thin films.

72 (200) ZrN

(111)

1200 nm Intensity (arbitrary units) 600 nm

30 35 40 45 50 55 60 65 70 75 80 Two-theta (degrees)

Figure 4.11. Effect of thickness upon the crystallinity of ZrN deposited on (111) silicon samples at 0.5 amp, 10 mTorr, and 4% nitrogen concentration.

The results shown in figure 4.11 show the addition of a strong (100) orientation at greater thicknesses. The films are thought to grow to a thickness where strain layer epitaxy permits the films to achieve the lowest free energy state and move from a (111) oriented film, to a (100) oriented film. This effect has been observed for the titanium nitride system. Pelleg et al.3 and Oh and Je4 claim that the orientation of titanium nitride is determined by a competition between strain energy and surface energy and assume that the strain energy increases linearly with thickness5. They4 state that at small thicknesses the surface energy is dominant resulting in orientations of the lowest free energy, which for the TiN system has been found to be the (200). As thickness increases, a change from (200) to (111) is said to occur at a critical thickness as seen in figure 4.12.

73

Figure 4.12.. Proposed critical thickness model for the titanium nitride system4.

Greene et al5. claim that this theory does not account for several other critical growth factors such as energy distribution, nitrogen overpressure and impurities and have shown growth TiN films with controlled orientations up to any thickness by controlling the 6 incident N2 ion to Ti ratio. Patsalas et al . have shown that when TiN internal film stress is increased, the preferred orientation changes from (111) to (200) indicating that strain energy is not the cause for the preferred orientation. The results of this work disagree with Pelleg’s TiN thickness theory3, but agrees well with the work done by Greene5 and Patsalas6; showing that thickness is not the dominant factor for controlling the preferred orientation of transition metal nitride films.

4.4 Microscopic and Spectroscopic Investigation of ZrN Thin Films Films deposited at 500°C, 5 mTorr, 4% nitrogen concentration and a constant cathode current of 0.5 amperes, were examined with a Joel field emission scanning electron microscope to investigate the structure of the deposited films. The working distance utilized was 12 mm with a 5 kV operating voltage. A cross-sectional scanning electron micrograph of ZrN on silicon can be seen in figure 4.13. The film is dense with small void concentrations, has a columnar structure with a smooth surface. This type of structure is formed when a vapor flux arrives perpendicular to the substrate with little adatom movement7.

74

Figure 4.13. Cross-sectional SEM micrograph of ZrN on silicon

High resolution transmission electron micrographs were taken of samples of bulk gallium nitride grown on a zirconium nitride thin film buffer layers. The use of ZrN as a buffer layer is thought to be useful as the backside metallization in future device development.

75

Figure 4.14a Figure 4.14b. Figure 4.14. Transmission Electron Micrographs of GaN thin films grown on ZrN.

Figure 4.14b. shows a “first attempt” of GaN growth on a ZrN buffer layer. The interface of ZrN (1) with GaN (2) can be seen and the films are polycrystalline. The growth of GaN on ZrN shows a continuous film with a columnar structure.

Raman analysis was performed upon thin films of ZrN and Zr3N4 as a structural investigation tool. The longitudinal-optical (LO) mode of Si at 522.5cm-1 is allowed for (100) Si substrates8.

The structure of ZrN it is not expected to produce any first order Raman lines and only broad spectral features due to second order (two-phonon) processes have been observered9. The disorder induced by N vacancies causes these crystals loose their “translational invariance”, which gives rise to a Raman spectrum close to that of a single phonon density of states9. The peak 150-200 cm-1 is associated with disorder induced single acoustic phonons, 330 and 400 are due to second order (two acoustic phonons) processes, and 500 is due to disorder induced scattering9. Cassinese et al. state that disorder induced scattering is likely to be proportional to the number of vacancies9.

76 ZrN was deposited on silicon (111) at 500°C, 4% N2 and 0.5 amps. The effect noticed upon investigating the Raman spectrum of ZrN thin films as a function of deposition pressure is shown in figure 4.15. ZrN Raman activity begins in films deposited at 15 mTorr or greater. This is attributed to the atomic disorder of these films as seen in the x-ray diffraction and resistivity study of films deposited at these pressures. The generation of vacancies is thought to be the cause of the Raman activity in films deposited at or above 15 mTorr and the effects seen in the resistivity and x-ray diffraction study.

5 mTorr

10 mTorr

TO/LO TA/LA 2A A+O Intensity (arbitrary units) (arbitrary Intensity

15 mTorr 2O

20 mTorr

160 260 360 460 560 660 760 860 960 1060 1160 Raman Shift (cm-1)

Figure 4.15. Raman Spectroscopy analysis of ZrN as a function of deposition pressure10.

77 Figure 4.16 shows Raman spectrum as a function of deposition pressure for Zr3N4 thin films deposited on (100) silicon at 500°C, 100% nitrogen and 0.5 ampere. Films deposited at temperatures from 15-500°C showed no ZrN Raman activity. This could be attributed to the transparency of the Zr3N4 thin films.

LO 500 Degrees C

400 Degrees C Intensity (arbitrary units) (arbitrary Intensity

15 Degrees C

H2O Cooled

100 200 300 400 500 600 700 800 900 1000 1100 1200 Raman Shift (cm-1)

Figure 4.16. Raman Spectrum of Zr3N4 thin films deposited on (100) silicon at 500°C, and 0.5 amp, and 100% nitrogen, as a function of temperature10.

Upon examination of the Raman spectrum of films of Zr3N4 deposited on (100) silicon as a function of cathode current showed only silicon LO activity.

78 1.0 Amp Intensity (arbitrary units) (arbitrary Intensity

0.5 Amp

100 200 300 400 500 600 700 800 900 1000 1100 1200 Raman Shift (cm-1)

Figure 4.17. Raman as a function of cathode current on (100) Si at 5mTorr, 0.5A, 15°C, higher nitride10.

4.5 Summary and Conclusions Zirconium nitride thin films were deposited by dc magnetron sputtering on two different orientations of silicon: (111) and (100). Processing conditions and parameters were optimized to generate films of completely oriented (111) ZrN thin films. The lattice mismatch between (111) oriented ZrN and c-axis oriented GaN is 1.6%. The optimum condition for the growth of (111) ZrN with the CAMP-M dual magnetron sputter deposition system were found to be 500°C, a constant cathode current of 0.5 amperes, a 10 mTorr argon and nitrogen mixture of 4% nitrogen on silicon (111) wafers. Four-point probe, Raman spectroscopy, scanning electron microscopy, transmission electron microscopy and x-ray diffraction were used to characterize thin films produced. Analysis of ZrN films with the lowest resistivity were found to occur at the same conditions for the most highly oriented (111) growth. This is due to the fact that the most

79 highly oriented, stoichiometric thin films produce the lowest resistivity material. Langmuir probe and I-V measurements were used to characterize the plasma utilized for experimentation. It is known that the ion flux, ion density, and ion energy play a role in crystallization changes11. Ion energy and ion density were calculated and their effects have been correlated to the crystallization changes in ZrN thin films.

The growth of GaN upon thin film buffer layers of ZrN has been investigated with transmission electron microscopy and it is believed that ZrN will become a valuable buffer layer for the growth of bulk GaN with the added benefit of back side metallization. To the author’s knowledge, this is the first time ZrN has been used for the buffered layer growth of c-axis oriented bulk GaN growth.

4.6 Future work Future work needs to be performed in several key areas. The growth of GaN and higher ZrN on oriented buffer layers or substrates of ZrN will need to be investigated further. The effect of improved thermal contact by the use of thermal compound or other technique must be investigated. Growth of GaN on higher nitride should be performed. X-ray diffraction of GaN upon buffer layers of ZrN must be performed and an optimization of the growth of GaN upon the films produced from this endeavor should be investigated. A more detailed understanding and investigation into plasma parameters should be performed; including optical absorption spectroscopy. An investigation into the thickness effects for the ZrN system should also be undertaken.

80 4.7 References 1D. Wu, Z. Zhang, W. Fu, X. Fan, and H. Guo, Appl. Phys. A, v. 64, p. 593, 1997

2J. –H. Huang, C. –H. Lin, Haydn Chen, Materials Chemistry and Physics, v. 59, pg. 49, 1999.

3J. Pelleg, L. Z. Zevin, and S. Lungo, Thin Solid Films, v. 197, pg. 117, 1991.

4U. C. Oh, J. H. Je, J. Appl. Phys. v. 74, pg. 1692, 1993.

5J. E. Greene, J. –E. Sundgren, L. Hultman, I. Petrov, D. B. Bergstrom, Appl. Phys. Lett. v. 67, pg. 2928, 1995.

6P. Patsalas, C. Charitidis, S. Logothetidis, Surf. Coat. Technol., v. 125, pg. 335, 2000.

7J. A. Thornton, Annual Review Material Science, v. 7, p. 239. 1977.

8D. H. Lee, Y. S. Cho, W. I. Yi, T. S. Kim, J. K. Lee, and H. J. Jung, Appl. Phys. Lett., v. 66, pg. 815, 1995.

9A. Cassinese, M. Iavarone, R. Vaglio, M. Grimsditch, and S. Uran, Phys. Rev. B, v. 62, pg. 13915, 2000.

10M. Park, unpublished results.

11Guruvenket and G. M. Rao, J. Vac. Sci. Technol. A, v. 20, pg. 678, 2002.

81