Citation
Epitaxial Growth and Characterization of Oxide Based Ferromagnetic Semiconductors for Spintronics Applications

Material Information

Title:
Epitaxial Growth and Characterization of Oxide Based Ferromagnetic Semiconductors for Spintronics Applications
Creator:
YANG, HYUCKSOO ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Anatase ( jstor )
Cobalt ( jstor )
Ferromagnetism ( jstor )
Magnetism ( jstor )
Magnets ( jstor )
Oxygen ( jstor )
Semiconductors ( jstor )
Silicon ( jstor )
Tin ( jstor )
X ray film ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Hyucksoo Yang. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
2/28/2005
Resource Identifier:
436098580

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

EPITAXIAL GROWTH AND CHARACTERIZATION OF OXIDE BASED FERROMAGNETIC SEMICONDUCTORS FOR SPINTRONICS APPLICATIONS By HYUCKSOO YANG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2004

PAGE 2

Copyright 2004 by HYUCKSOO YANG

PAGE 3

Dedicated to My Family

PAGE 4

ACKNOWLEDGMENTS I would like to express my sincere gratitude to my research advisor, Dr. Rajiv K. Singh, for his guidance, support and encouragement during the course of my research work. I would also like to thank Dr. S. J. Pearton, Dr. C. R. Abernathy, Dr. D. P. Norton, and Dr. A. F. Hebard for being on my supervisory committee and their interest in my research. My appreciation also goes to all my group members and colleagues for all the help and camaraderie: Chad, Jeremiah, Nabil, Joshua, Suho, Vishal, Wonseok, Seungmahn, Frank, Joodong, Kyose, Karthik, Seemant, Sanghyun, Michael, Prabhakar and Hanho. I especially thank to Jaeyoung and Wonseop for their collaboration and help. I extend my thanks to all the staff at the MAIC and especially to Kerry Siebein for her help. My special thanks go to Margaret Rathfon for her deep kindness and assistance. Finally, I would like to express my sincere gratitude to my family. With their endless love and affection, everything has been possible. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT.......................................................................................................................xi CHAPTER 1 INTRODUCTION........................................................................................................1 2 REVIEW OF LITERATURE.....................................................................................10 Diluted Magnetic Semiconductor (DMS)...................................................................10 Oxide Based Diluted Magnetic Semiconductors........................................................13 Properties of TiO 2 ................................................................................................14 TiO 2 -based DMS.................................................................................................16 Co-doped TiO 2 .............................................................................................16 Fe-doped TiO 2 ..............................................................................................20 V-doped TiO 2 ...............................................................................................20 ZnO-based DMS..................................................................................................21 Mn-doped ZnO.............................................................................................21 Co-doped ZnO..............................................................................................22 SnO 2 -based DMS.................................................................................................23 Mn-doped SnO 2 ............................................................................................23 Co-doped SnO 2 .............................................................................................24 Origin of Ferromagnetism..........................................................................................24 Carrier Mediated Exchange Interaction...............................................................26 Ferromagnetic Nanoclusters................................................................................28 3 EXPERIMENT...........................................................................................................35 Target Preparation......................................................................................................35 Substrate Material.......................................................................................................35 LaAlO 3 .................................................................................................................35 Silicon..................................................................................................................35 v

PAGE 6

Thin Film Fabrication.................................................................................................36 PLD (Pulsed Laser Deposition)...........................................................................36 Characterization Tools................................................................................................37 X-ray Diffraction (XRD).....................................................................................37 Atomic Force Microscopy (AFM) / Magnetic Force Microscopy (MFM).........38 XPS (X-ray Photoelectron Spectroscopy)...........................................................39 TEM (Transmission Electron Microscopy).........................................................40 SQUID (Superconducting Quantum Interference Device)..................................40 Magnetization measurements.......................................................................41 Zero field cooled (ZFC) – Field cooled (FC) measurement.........................42 Superparamagnetism....................................................................................42 4 EPITAXIAL GROWTH OF COBALT-DOPED ANATASE TiO 2 ON LaAlO 3 ......50 Overview.....................................................................................................................50 Experiment..................................................................................................................51 Results and Discussion...............................................................................................52 Summary.....................................................................................................................58 5 EPITAXIAL GROWTH OF COBALT-DOPED ANATASE TiO 2 ON SILICON...71 Overview.....................................................................................................................71 Epitaxial Growth of TiO 2 on Si...........................................................................71 Epitaxial Growth of TiN on Si............................................................................74 Epitaxial Growth of SrTiO 3 on TiN/Si................................................................75 Experiment..................................................................................................................76 Results and Discussion...............................................................................................77 Summary.....................................................................................................................79 6 EPITAXIAL GROWTH OF COBALT-DOPED RUTILE TiO 2 ON SILICON........95 Overview.....................................................................................................................95 Experiment..................................................................................................................96 Results and Discussion...............................................................................................97 Summary...................................................................................................................103 7 CONCLUSION.........................................................................................................120 LIST OF REFERENCES.................................................................................................124 BIOGRAPHICAL SKETCH...........................................................................................134 vi

PAGE 7

LIST OF TABLES Table page 1-1 Advantages of MRAM over the current memory devices.........................................7 3-1 Crystallographic information of LaAlO 3 , SrTiO 3 , TiN, TiO 2 and Si.......................46 4-1 FWHM values of a (004) anatase peak, coercive field and M r /M s at 300 K in the samples deposited at 400-700 o C.............................................................................64 6-1 Properties of the cobalt silicides.............................................................................109 vii

PAGE 8

LIST OF FIGURES Figure page 1-1 GMR Head in a computer hard disk drive.................................................................5 1-2 Two current model explaining how resistance changes.............................................6 1-3 Spin-FET proposed by Datta and Das........................................................................8 1-4 Spin-FET (Field Effect Transistor)............................................................................9 2-1 Schematic diagram of a diluted magnetic semiconductor where a sizable portion of atoms is substituted by a magnetic element.............................................................30 2-2 Predicted Curie temperature as a function of bandgap of semiconductors..............31 2-3 Calculated values of Curie temperature for various p-type semiconductors at Mn concentration of 5% and hole concentration of 3.510 20 /cm 3 ..................................32 2-4 Crystal structure of TiO 2 ..........................................................................................33 2-5 Combinatorial fabrication of TiO 2 films doped with transition metals....................34 3-1 Scanning electron microscopy image of the Co-doped TiO 2 ceramic target showing highly dense microstructure.....................................................................................45 3-2 Pulsed laser deposition system.................................................................................47 3-3 Change in coercive field (H ci ) as a function of particle diameter............................48 3-4 Estimated blocking temperature as a function of size of a magnetic particle..........49 4-1 The results of X-ray diffraction................................................................................60 4-2 X-ray phi scan diffraction patterns of the Ti 0.93 Co 0.07 O 2films on LaAlO 3 (001) grown at 400 o C by UVPLD and PLD, respectively................................................61 4-3 The results of SQUID measurement........................................................................62 4-4 Magnetization vs. magnetic field measurement at room temperature for the Ti 0.93 Co 0.07 O 2film grown at 400 o C by UVPLD....................................................63 viii

PAGE 9

4-5 Temperature dependence of zero-field-cooled (ZFC) and field-cooled (FC) magnetizations for the Ti 0.93 Co 0.07 O 2film grown (a) at 550 o C by UVPLD and (b) at 400 o C by UVPLD................................................................................................65 4-6 The results of TEM analysis.....................................................................................66 4-7 Reference photoelectron spectra of Ti 2p................................................................67 4-8 Reference photoelectron spectra of Co 2p...............................................................68 4-9 Photoelectron spectra of Ti 2p.................................................................................69 4-10 XPS spectra of the Co 2p and Ti 2p of the Ti 0.93 Co 0.07 O 2films grown by PLD and UVPLD, respectively...............................................................................................70 5-1 Crystallographic relationship between TiO 2 and Si (001).......................................81 5-2 X-ray diffraction pattern of the Ti 0.96 Co 0.04 O 2 films deposited at (a) HF-treated silicon and (b) just cleaned silicon having native oxide on the surface...................82 5-3 AFM data of Ti 0.96 Co 0.04 O 2 /Si deposited at (a) 400 o C and (b) 550 o C...................83 5-4 M-H and M-T curves of Ti 0.96 Co 0.04 O 2 /Si deposited at 400 o C [(a) and (b)] and at 550 o C [(c) and (d)]..................................................................................................84 5-5 Stability of binary oxides on silicon.........................................................................85 5-6 Isothermal section of the Ti-Si-O phase diagram (T=700-1000 o C)........................86 5-7 Stability of binary nitrides on silicon.......................................................................87 5-8 Crystallographic information of TiN and Si............................................................88 5-9 The results of X-ray diffraction................................................................................89 5-10 -2 X-ray diffraction pattern of the Co 0.04 Ti 0.96 O 2/SrTiO 3 /TiN/Si film structure90 5-11 The results of SQUID measurement........................................................................91 5-12 Atomic force and its corresponding magnetic force images....................................92 5-13 The results of TEM analysis.....................................................................................93 5-14 Enlarged TEM image of the Co:TiO 2 /SrTiO 3 region...............................................94 6-1 The results of X-ray diffraction..............................................................................104 6-2 Crystallographic relationship between rutile TiO 2 and TiN...................................105 ix

PAGE 10

6-3 Magnetization as a function of a magnetic field measured at 10 K, showing paramagnetic behavior...........................................................................................106 6-4 Cross-sectional TEM images of the Ti 0.96 Co 0.04 O 2 /TiN/Si structures....................107 6-5 Elemental mapping of the spike region..................................................................108 6-6 Additional results of TEM analysis........................................................................110 6-7 -2 X-ray diffraction pattern of (a) TiN/Si and (b) Ti 0.96 Co 0.04 O 2 film deposited on in situ oxidized TiN/Si...........................................................................................111 6-8 The results of SQUID measurements.....................................................................112 6-9 Elemental mapping of the film...............................................................................113 6-10 HAADF and line scan along the film depth...........................................................114 6-11 Imaging mechanism of Z-contrast transmission electron microscopy...................115 6-12 The results of SQUID measurement......................................................................116 6-13 Cross sectional TEM images of the CTO/thin TiN/Si...........................................117 6-14 Cross sectional TEM images of the Co 0.07 Ti 0.93 O 2 /thin TiN/Si..............................118 6-15 EDS spot analysis...................................................................................................119 x

PAGE 11

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EPITAXIAL GROWTH AND CHARACTERIZATION OF OXIDE BASED FERROMAGNETIC SEMICONDUCTORS FOR SPINTRONICS APPLICATIONS By Hyucksoo Yang August, 2004 Chair: Rajiv K. Singh Major Department: Materials Science and Engineering The utilization of the spin as an additional degree of freedom to electronic transport opens a new field for novel electronic devices combined with magnetic and optical properties. Semiconductors doped with some amount of the magnetic element have been developed as an effective injector of spin-polarized carriers owing to a good electrical conductivity match with semiconductors to be hybridized. Cobalt-doped TiO 2 is a promising candidate material due to properties like a high Curie temperature over 400 K and n-type electrical behavior which exhibits longer spin lifetime. Lattice matching oxide-based substrates such as LaAlO 3 , SrTiO 3 and Al 2 O 3 have been used for the epitaxial growth of Co-doped TiO 2 films, but the successful integration of a ferromagnetic semiconductor with silicon is expected to bring many technical advantages on the basis of existing silicon electronics technology. Direct growth of epitaxial Co-doped TiO 2 on silicon is, however, extremely difficult mainly because an xi

PAGE 12

amorphous silicon oxide layer inevitably forms at the interface during the overlying oxide layer deposition. Use of a non-oxide thermally stable buffer layer, TiN, was proposed to overcome the formation of such an interfacial layer, which has been reported to be grown epitaxially on silicon in terms of domain matching mechanism. For the epitaxial growth of a Co-doped anatase phase, one additional buffer layer, SrTiO 3 , was added, which also grows epitaxially on TiN. On the other hand, the epitaxial growth of a Co-doped rutile phase could be achieved by the direct deposition of Co:TiO 2 on TiN/Si. Depending on the film structures employed, the cobalt showed different behaviors in the Co:TiO 2 film; uniform distribution, formation of clusters/segregation and reaction with silicon forming cobalt silicide. The microstructure is very sensitive to the defect density, multilayer structure, oxygen atmosphere, and deposition temperatures. Different reactivity with oxygen between cobalt and titanium is one of the important factors to be considered. It was also found that cobalt atoms are highly mobile and diffusive even at the temperature as low as 450 o C. Circumventing the formation of cobalt silicide will be crucial for the development of the cobalt-based DMS spin injector into silicon. xii

PAGE 13

CHAPTER 1 INTRODUCTION The technology utilizing the degrees of freedom in both charge and spin, so-called spintronics (spin electronics), has raised much research interest by implementing novel functionalities in the community of electronics and magnetics. The interplay of charge and spin is expected to explore a new paradigm of devices. Most of the current semiconductor devices utilize an electron's charge to process information and it has been progressed by increasing the integration density. However, the rate of development predicted by Moore’s law the doubling of the number of transistors per integrated circuit every 18 months is approaching some physically fundamental obstacles. Moreover, power dissipation and heat generation are becoming critical issues. If we can harness the spin (the quantum particle of a well-defined spin state) of the carriers within a material, which has been an underappreciated characteristic of an electron, the devices exhibiting highly increased speed and storage density can be developed by the novel ways of signal processing. The spin of the electron has been already exploited in the area of data storage. Owing to the nature of the remanence of the magnetic material, it can be used as a nonvolatile memory; that is, it can keep information without the need of the electrical power. The inherent characteristics of the spin offer advantages of spintronics over conventional electronics. Specifically, these characteristics open the possibility of developing devices that can be nonvolatile, much smaller and faster with less consumption of electricity. 1

PAGE 14

2 The field of spintronics was initiated with the discovery of the giant magnetoresistive (GMR) effect in 1988 [Bai88]. Figure 1 shows a computer hard disk drive which is a typical example of GMR application. GMR is a quantum mechanical effect observed in magnetic thin film multi-layered structures that consist of alternating layers of ferromagnetic and nonmagnetic layers. As shown in Fig. 1-2, the change in resistance is caused by spin-dependent scattering at the interfaces and spin-dependent conductivity. In other words, when the magnetic moments of the ferromagnetic layers are parallel, the scattering of the electrons is minimized and the GMR film shows its lowest resistance. On the other hand, when the ferromagnetic layers are anti-aligned, it shows highest resistance due to increased scattering. The GMR effect has been applied in position sensors, non-volatile random access memory and read head sensors in computer hard disk drives. The discovery of the GMR effect revolutionized the information storage industry with the development speed outpacing Moore’s law and it is indeed the representative case of a rapid transition from discovery to commercialization [Pri98]. The examples of potential spintronic devices are non-volatile ultra-density magnetoresistive random access memory (MRAM) [Moo95], spin light emitting diodes (spin-LEDs) [Fie99], spin resonant tunneling diodes (spin-RTDs) [Ohn98a] and spin field effect transistors (spin-FETs) [Dat90]. Table 1-1 shows the advantages of MRAM over the current memory devices, which fully reflect the characteristics of the spin. MRAM combines the nonvolatility of flash with the high speed of SRAM and the high integration capacity of DRAM using magnetoresistance effect. If MRAM technology is applied to the computer, it will enable the instant-on computer without waiting for the booting time needed for retrieving information from the hard drive and extend the battery life due to

PAGE 15

3 the feature of less power consumption. In a magnetic tunnel junction, a very thin layer of an insulator (~1 nm) is sandwiched by magnetic metal layers. This is CPP (current perpendicular to plane) type of device compared to conventional CIP (current in plane) GMR. The magnetic orientation of the two magnetic layers can be switched by current flow, leading to different values of the tunneling current. This results in “bits” of computer memory. In 1990, Datta and Das proposed a design for a spin-polarized field-effect transistor (spin-FET) [Dat90], as shown in Fig. 1-3. In the spin-FET, both the source and the drain are ferromagnetic and are separated by a narrow semiconducting channel, the same as in a conventional FET. The source emits spin-polarized electrons into the channel, and this spin current flows easily if it reaches the drain unaltered. A voltage applied to the gate electrode produces an electric field in the channel, which forces the spins of fast-moving electrons to precess, or rotate due to the spin-orbit interaction. The drain impedes the spin current according to how far the spins have been rotated. The degree of precession by the gate field modulates the electron current. A spin-FET would have several advantages over a conventional FET. Flipping spins in this way takes much less energy and is much faster than the conventional FET process of pushing charges out of the channel with a larger electric field. However, the working device has not been demonstrated experimentally yet, because of difficulties in spin injection from a ferromagnetic metal into a semiconductor. For this device to work, the electron’s spin state should maintain its polarization when entering the channel and traveling through a semiconductor. Ohno et al. have demonstrated electrical field control of ferromagnetism [Ohn00]. Since the magnetic material has a tendency to retain its magnetization once magnetized,

PAGE 16

4 the ability to externally control the magnetic properties of the material is technologically important. A metal-insulator-semiconductor field effect transistor (FET) having an (In,Mn)As magnetic channel is shown in Fig. 1-4. The hole mediated magnetic interaction results in ferromagnetism in this material and by the application of electric fields the hole concentration can be controlled, thereby modifying the ferromagnetic properties. For example: a) applying a positive voltage to the gate creates electric fields that repel the positively charged holes, thus reducing hole concentration in the (In,Mn)As layer causing the Mn magnetic moments to orient randomly, and b) applying a negative voltage to the gate creates electric fields that attract the positively charged holes, thus increasing the hole concentration and making the system ferromagnetic and causing the Mn magnetic moments to align. Therefore, by simply applying a voltage to the gate electrode, the magnet can be reversibly turned on (ferromagnetic) and off (paramagnetic). This experiment proved that the magnetic properties of ferromagnetic semiconductors can be controlled using standard electronic techniques. Most magneto-electronic devices developed are based on metallic systems, where the magnetic layers are mainly composed of Fe, Co and Ni and their alloys. If an electron's spin can be harnessed within a semiconductor, the unprecedented ways of data processing and writing will increase speed and information storage densities. To utilize spins in semiconductor devices, it is essential to be able to generate, inject, transport, manipulate, and detect spins. The new material, called diluted magnetic semiconductor, is the key components for the purpose and the development of it with appreciable spontaneous magnetization at room temperature has become an important challenge of materials science.

PAGE 17

5 (d) (c) (b) (a) Conductor Ferromagnet Ferromagnet AntiFerromagnet Figure 1-1. GMR Head in a computer hard disk drive. (a) Computer hard disk drive. (b) Tip area of HGA (Head Gimbal Assembly). (c) The basic layer structure of a conventional GMR head. (d) Cross-sectional transmission electron microscopy image of the conventional spin-valve structure, Ta50/PtMn300/CoFe23/Cu30/CoFe20/NiFe40/Ta50/Al 2 O 3 600/Si substrate (thickness unit: ) [Yan01].

PAGE 18

6 Figure 1-2. Two current model explaining how resistance changes. The resistance changes depending on the relative magnetization directions of the magnetic layers.

PAGE 19

7 Table 1-1. Advantages of MRAM over the current memory devices. MRAM DRAM SRAM FLASH FRAM Cell structure 1 TR + 1 TMR 1 TR + 1 capacitor 6(4) TR 1 TR 1 TR Density high high low very high High Power for data none required required none none Refresh none required none none none Read speed Very fast ~3 ns Fast ~60 ns Very fast ~2 ns Fast ~10 ns Fast ~60 ns Write speed Very fast ~3 ns Fast ~60 ns Very fast ~2 ns Very slow 0.2s~200 ms Fast ~60 ns Power dissipation small small large very small small Non-volatility Y N N Y Y Application main memory main memory cache memory BIOS memory BIOS memory

PAGE 20

8 Figure 1-3. Spin-FET proposed by Datta and Das. (a) When the Gate is off. (b) When the Gate is on [Dat90, Sar01].

PAGE 21

9 Figure 1-4. Spin-FET (Field Effect Transistor). Ferromagnetism can be controlled by electrical method [Ohn00]. Figure 1-4. Spin-FET (Field Effect Transistor). Ferromagnetism can be controlled by electrical method [Ohn00]. 9

PAGE 22

CHAPTER 2 REVIEW OF LITERATURE Diluted Magnetic Semiconductor (DMS) Most magneto-electronic devices are based on metallic materials like Fe, Co, Ni and their alloy where magnetic moments originate from the unfilled 3d shells. Due to an imbalance in the number of spin-up and spin-down electrons at the Fermi level, the carriers are spin-polarized in these materials. In the mean time, it has been of great interest to combine the magnetic element with a conventional semiconductor in order to make use of their specific physical advantages as well as the existing semiconductor technology for spintronics. As an example, although metal-based magnetic devices are effective as a switch, they cannot amplify signal unlike semiconductor transistors. If spintronic devices could be fabricated based on semiconductors, they would principally offer amplifying ability and serve as multi-functional devices [Sar01]. More importantly, in order to utilize spins in semiconductor devices for the new functionality, it is required to inject, transport, manipulate and detect spin-polarized carriers. Spin injection can be done using ferromagnetic contact materials with high spin polarization. In the first place, ferromagnetic metal contacts on a nonmagnetic semiconductor have been investigated since electrons in ferromagnetic metals can produce a spin-polarized current at room temperature without an external magnetic field. In a ferromagnetic metal, there is a spin imbalance due to exchange splitting to lower the total energy of the system [Pri95]. However, little or no spin polarization of carriers has been found owing to a large mismatch in conductivities [Ham99]. This prevents the spin-polarized carrier injection 10

PAGE 23

11 when spins diffuse across the interface. Schmidt et al. reported that an efficient spin injection in the diffusive transport regime is difficult due to the dissimilar materials properties of a metal and a semiconductor, unless the magnetic material is nearly 100 % spin-polarized [Sch00]. Recently, enhanced spin injection efficiency was reported by inserting thin, insulating tunnel junctions at the interface [Zhu01]. This enables decoupling of the state of the metal and the semiconductor in terms of Schottky barrier and tunneling [Ras00]. Basically, if the spin injector is semiconducting, injecting spins into a nonmagnetic semiconductor should be easier because they can eliminate the conductivity mismatch that prevents diffusive spin injection from ferromagnetic metals into semiconductors. Additionally, semiconductor-based devices will be readily integrated with conventional semiconductors. Diluted magnetic semiconductors (DMS), the semiconductors doped with a small amount of a magnetic transition metal element, have been developed for effective spin transfer due to a good electrical conductivity match with semiconductors to be hybridized. The metal cation gives rise to localized magnetic moments in the semiconductor matrix as shown in Fig. 2-1 [Ohn98b]. In DMSs, it has been claimed that the magnetic coupling occurs by virtue of exchange interactions between the magnetic dopants that are mediated by free carriers in the semiconductor [Ohn98b]. The interaction can lead to antiferromagnetic or ferromagnetic coupling, depending on the concentration and the local structural environment of the magnetic impurity. The spontaneous magnetization in ferromagnetic semiconductors introduces imbalance of spin population

PAGE 24

12 in the carrier system as well. Thus, it can be used as a spin polarized carrier source for electrical spin injection. II-IV semiconductors like CdTe and ZnTe have been studied as a host material for DMS since they are able to incorporate a high concentration of magnetic elements due to the identical valence. Although it has an advantage in material synthesis, it has been difficult to dope to create nand p-type material. (Ga 1-X ,Mn X )As with Mn concentration in the range of 0.04
PAGE 25

13 Ohn96], and II-VI [Fer01] exhibited p-type electrical behavior, and low Curie temperature (T c <180 K) making them impractical for use in real devices. This means that the exchange interaction leading to ferromagnetic behavior is hole-mediated and a cooling system would be required for the device function. For successful spintronic applications, future efforts have to concentrate on fabricating ferromagnetic semiconductors in which ferromagnetism will persist at higher temperatures and have an n-type behavior for longer spin lifetime. Recently, it has been found that Curie temperature has close relationship with bandgap of semiconductor as shown in Fig. 2-2 and room temperature ferromagnetism has been observed in some wide-band-gap host semiconductors [Pea03a, Pea03b] such as GaN [Ree01], GaP [The02], ZnO [Ued01], and TiO 2 [Mat01a]. Among them, Co-doped TiO 2 discovered by Matsumoto et al. has been actively investigated due to properties like high T c over 400 K, excellent optical transmittance in the visible and near-infrared regions, and high n-type carrier mobility without intentional doping. While most III-V based DMSs are p-type, the source of spin-polarized electrons can provide more efficient electrical spin injection [Fie99, Par00] and much longer spin relaxation times [Kik99, Oes99]. These are desirable attributes for spintronic devices. Oxide Based Diluted Magnetic Semiconductors A ferromagnetic semiconductor combines both ferromagnetic and semiconducting properties and is a key component of emerging field of spintronics. A variety of diluted magnetic semiconductor materials has been developed based on group II-VI, III-V and IV semiconductors. Although active research has revealed the mechanism of ferromagnetism and has proposed new conceptual devices, most DMSs have a limit to be applied practically to the device owing to low Curie temperature. Dietl et al. [Die00]

PAGE 26

14 theoretically predicted Curie temperature of DMSs using Zener model [Zen51], which was originally proposed for the ferromagnetism in transition metals by the exchange interaction between carriers and localized spins. Accordingly, the model was developed based on exchange interactions mediated by delocalized holes in the ensemble of localized spins. Curie temperature is computed by minimizing the free-energy function with respect to magnetization at a given hole concentration. As a result, a mean field solution predicted the Mn-doped wide bandgap nitrides, p-(Ga,Mn)N, and oxides, p-(Zn,Mn)O, to be ferromagnetic at temperatures over 300 K, as shown in Fig. 2-3. This has stimulated considerable research interest in magnetically doped semiconducting nitrides and oxides. Since then, the room temperature ferromagnetism (RTFM) has been reported experimentally in some wide band-gap materials such as AlN [Fra03, Wu03], GaN [Ree01, Tha02] and GaP [The02]. In addition, ferromagnetic semiconductors based on nontraditional semiconductors such as semiconducting oxides have been found to have potential advantages including remarkably high T c , transparency for visible light and high n-type carrier concentration [Pre03]. Strong p-d exchange coupling is also expected due to the large electro-negativity of oxygen between band carriers and localized spins [Fuk04]. In the case of oxide-base DMS, some fractions of cation sites of the host semiconductor are substituted for the magnetic transition metal atoms such as V, Cr, Mn, Fe, Co and Ni, to possess the both ferromagnetic and semiconducting properties. Robust ferromagnetism has been reported in several oxide based DMSs like ZnO [Sha03], LaSrTiO 3 [Zha03], Cu 2 O [Kal03], SnO 2 [Oga03], and TiO 2 [Mat01a]. Properties of TiO 2 Titanium dioxide (TiO 2 ) has three polymorphs; rutile (tetragonal, a=4.585 and c=2.953 ), anatase (tetragonal, a=3.784 and c=9.515 ), and brookite (orthorhombic,

PAGE 27

15 a=9.184 , b=5.447 , c=5.145 ). Rutile (P4 2 /mnm) and anatase (I4 1 /amd) are the two most common allotropic forms of TiO 2 and their crystal structures are shown in Fig. 2-4. The anatase phase is less stable and is formed at lower temperature. The brookite phase is formed only under very constrained conditions. It occurs only when stabilized by some small amount of impurities. The most stable structure is rutile. If brookite and anatase are heated above 700 o C, they are generally converted to rutile. Although the anatase phase is difficult to make in bulk, it can be grown in thin film form on appropriate lattice-matched substrates. Both anatase and rutile structures can be viewed as a network of coordinated TiO 6 octahedra in which each Ti 4+ (3d 0 ) ion is surrounded by an octahedron of six O 2ions (coordination number (CN) of Ti is six). Each oxygen atom has three titanium neighbors (CN of Oxygen is three) and therefore belongs to three different octahedral. TiO 6 octahedra are interconnected differently for each phase, leading to different structures and symmetries. The two structures differ by the distortion inside each octahedron and by their assemblage. In an anatase structure, each octahedron is in contact with eight neighbors (four-edge sharing and four-corner sharing), while in a rutile structure, the coordination number is ten (two-edge sharing and eight-corner sharing) [San94]. The octahedral coordination is less dense in anatase than in rutile, leaving anatase in a more open and flexible structure. Titanium dioxide presents various attractive properties: a high refractive index (n = 2.35 at 550 nm), high dielectric constant ( r = 105 at 4.2 K), low tangential loss (tan = 10 7 at 4.2 K), an excellent optical transmittance (>85%) in the visible and in the near infrared, high chemical stability, high mechanical durability and photo-catalytic ability. In addition, TiO 2-X compounds have semiconductor properties (n-type) with a carrier

PAGE 28

16 density on the order of 10 19 /cm 3 via cation substitution, Ti interstitials [Yag96] or oxygen vacancies [For94]. Binary oxides of titanium are known to form a series of compounds that exhibit a wide range of Ti to O ratios depending on the oxygen environment under which they are synthesized. Depending on the conditions used, titanium oxide made from the evaporation of Ti metal, under 10 -3 -10 -6 Torr O 2 , is known to yield such oxides as Ti 2 O, TiO, Ti 2 O 3 , the magnelli phase series of Ti n O 2n-1 , and TiO 2 . Hall mobility of electron-doped anatase has been measured as high as 20 cm 2 /Vs with the effective mass of anatase, m*=m e [Pea03a, Tan94], while the mobility of rutile is in the 1 cm 2 /Vs range [Def83]. The band gap is slightly wider in anatase (~3.2 eV) than in rutile (~3.0 eV). The widespread applications of TiO 2 includes photo-catalysts for waste-water treatment, dye-sensitized photo-voltaics, solar-energy conversion, storage capacitors in DRAMs, insulators in MOS devices, electrochromic displays, gas and humidity sensors, waveguides in integrated optics, and antireflective protective coating on optical elements. For applications dependent on the semiconducting properties of TiO 2 , carrier transport will depend critically on crystallinity. Grain boundaries and/or secondary phase boundaries are unacceptable for optimizing mobility and conductivity. Furthermore, high crystalline quality is needed for spintronics application since defects cause spin-flip scattering reducing the spin polarization in the material [Cha02b]. Recent efforts have shown that phase-pure anatase thin films can be realized via epitaxial stabilization on lattice matching single crystal substrates. TiO 2 -based DMS Co-doped TiO 2 The interest in Co-doped TiO 2 was triggered early in 2001 by Matsumoto et al. [Mat01a] who synthesized Co X Ti 1-X O 2 anatase films (x=0.01-0.10) by combinatorial laser

PAGE 29

17 MBE [Koi00, Mat99, Ohs04]. It is one of the very few DMS materials demonstrated to exhibit ferromagnetic behavior above room temperature. Using high throughput nature of the combinatorial method demonstrated in Fig. 2-5, several transition metals were doped into TiO 2 and their magnetic property was screened by a scanning superconducting quantum interference device microscope. The ferromagnetism was serendipitously discovered only in cobalt-doped TiO 2 . A sizable amount of Co, up to ~8% was soluble in anatase TiO 2 and the magnetization was found to increase with Co content. The variation in lattice parameter with Co concentration explains Co substitution for Ti at cation sites. When the cobalt content reaches around 50%, other phases like CoTiO 3 and CoO formed [Mur04a]. The resistivity and carrier concentration of the Co X Ti 1-X O 2 films are about 0.1-1 ohmcm and 10 18 /cm 3 at room temperature, respectively, being rarely dependent on the Co doping level. The measured saturated magnetic moment was 0.32 B /Co, indicating low spin configuration of Co ions. Cobalt-doped TiO 2 is a promising candidate as a transparent ferromagnetic semiconductor due to properties like a high T c over 400 K, excellent optical transmission in the visible and near infrared regions, and high n-type carrier mobility. Unlike in many other ferromagnetic semiconductors, which are p-type doped, the source of spin-polarized electrons can provide more efficient spin injection [Fie99] with larger mobilities and much longer spin relaxation times [Oes99], which are desirable for spintronic devices [Zut02]. Room temperature ferromagnetism was also found in Co-doped rutile phase with a saturating magnetization of ~1.0 B /Co [Mat01b]. The cobalt is less soluble in rutile phase (~5%) [Mat02]. Chamber et al. [Cha01a] claimed that better magnetic properties can be obtained by using oxygen-plasma-assisted molecular beam epitaxy (OPA-MBE), showing a larger

PAGE 30

18 saturated magnetic moment of 1.26 B /Co atom and n-type carrier density of 10 19 /cm 3 . Using Co 2p photoemission and Co K-shell X-ray absorption near-edge structure (XANES) [Cha01b, Cha02a], the formal oxidation state of Co was found to be +2. He suggested that ferromagnetism occurs by virtue of electron mediated exchange interaction between Co +2 cations which replaced Ti +4 lattice sites. However, under certain conditions, highly Co-enriched clusters are formed on the epitaxial film surface. The relatively higher growth rate of MBE method (~0.4 /s) resulted in Co-enriched rutile clusters within the epitaxial anatase film [Cha02a] and even under the optimized growth rate condition (~0.1 /s), highly cobalt-segregated anatase clusters could sometimes nucleate on top of the film surface. The particles are characterized into ferromagnetic Co X Ti 1-X O 2-X (x=0.2-0.4), which is responsible for the ferromagnetism [Cha03]. The dependence of the magnetic properties of Co:TiO 2 films on oxygen partial pressure were examined by laser MBE [Kim02a]. As the oxygen partial pressures were decreased below 10 -5 Torr, the growth mode changed from a two dimensional layer-by-layer to a three dimensional island growth mode, and the conductivity and the magnetization of the film increased [Kim03b]. These systematic changes were attributed to the Co and/or Co intermetallic clusters, as observed by high resolution TEM images. The clusters were larger at the interface than the ones inside the film, with an overall average diameter of 11.7 6.0 nm. The larger saturation magnetization at lower P O2 was attributed to the increased number of Co clusters due to enhanced diffusion of the Co caused by oxygen vacancies in the anatase TiO 2 film. The origin of ferromagnetism was further investigated using X-ray absorption spectroscopy (XAS) and magnetic circular dichroism (MCD) at Co L 2,3 edges. The charge state of Co in TiO 2 matrix has high spin

PAGE 31

19 divalent state (Co +2 , 3d 7 , S = 3/2) like in CoO but the magnetic signal is almost identical to that of Co metal, showing the ferromagnetism is originated from clustered Co metal [Kim03a]. Ion implantation method was attempted as an alternative way to incorporate the cobalt dopant into TiO 2 [Kim03c]. It is a useful technique to screen the desirable doping element and host material due to the easy experimental setup [Heb04, Pea03c]. Magnetization versus temperature curve demonstrates the superparamagnetism in the film indicating the presence of the magnetic particles. The particles grow with annealing, thus increasing the blocking temperature [Kim04b]. Since titanium atoms are already bonded to oxygen atoms, it is unlikely to replace titanium site by cobalt due to smaller activity of cobalt with oxygen. In addition, the possible defects induced by ion implantation could be effective nucleation site for those clusters especially under the condition of high concentration dose [Kai02]. Sputtering technique, which is widely used in mass production, has been explored at a temperature as low as 350 o C [Par02a]. A single rutile phase was obtained, but the films were polycrystalline on unetched silicon and quartz. It was regarded as an intrinsic DMS having a low spin configuration of Co ions. But later, the origin of ferromagnetism was found to be caused by cobalt nanoparticles with a wide size distribution from the zero-field-cooled (ZFC)/field-cooled (FC) magnetic measurements [Pun03]. There have been few reports on the definite value of T c in Co-doped TiO 2 because T c is too high to be measured by conventional superconducting quantum interference devices. For the measurement of Curie temperature, vibrating sample magnetometry (VSM) with heating option was employed. Shinde et al. [Shi03] reported that the as

PAGE 32

20 grown low doped (~ 1% Co) sample as well as the annealed high-doped (7%) sample was indicated to be intrinsic DMS system with a high Curie temperature of 650-700 K. Fe-doped TiO 2 Ferromagnetism was also found in Fe-doped reduced rutile TiO 2 epitaxial film grown on -Al 2 O 3 substrate. Extraordinary Hall effect (EHE) was found in this film, indicating the interaction between itinerant carriers and localized electron spins of the Fe ions. It is claimed that EHE is one of the most conclusive factor for the intrinsic DMS [Wan03a]. But it is reported that the film having nano-sized magnetic particles can also exhibit EHE [Shi04]. The co-occurrence of superparamagnetism and EHE makes it difficult for the Hall effect to be a decisive tool to determine the intrinsic nature of ferromagnetism in DMS. The carrier density was calculated to about 10 22 /cm 3 but the carrier type is p-type contrary to the reported n-type in Co-doped anatase TiO 2 [Cha01a], suggesting RKKY type interaction [Wan03b]. Fe ions act as an acceptor impurity [Bal98]. The saturation magnetization was estimated to be 2.3-2.4 B /Fe. They ruled out the other possible magnetic sources such as Fe particles, iron oxides and Ti-Fe oxides, by magnetic moments and anticipated Curie temperature. (T c of Fe = 1045 K, magnetic moment of Fe 3 O 4 and -Fe 2 O 3 < 1.3 B /Fe, magnetic moment of FeTiO 3 and -Fe 2 O 3 << 1.0 B /Fe) In contrast, clusters of mixed rutile and Fe 3 O 4 were found on the surface of the Fe-doped TiO 2 films which was grown on rutile TiO 2 (110) substrate [Kim04a]. RTFM is attributed to the formation of inverse spinel Fe 3 O 4 on the surface, which is a ferrimagnet. V-doped TiO 2 Room temperature ferromagnetism was recently reported in V-doped anatase TiO 2 [Hon04]. The film was grown on LaAlO 3 by PLD. Although vanadium is a paramagnetic

PAGE 33

21 material in a bulk form, an isolated vanadium atom is known to have a permanent magnetic moment of 3 B . A very large magnetic moment of 4.23 B /V was obtained in this film and such a high value was explained by the unquenched orbital contributions, i.e., the orbital magnetic moment of vanadium maintains unquenched because the surrounding atoms get a moment through electronic effects. ZnO-based DMS ZnO is a direct band gap II–VI compound semiconductor with E g = 3.35 eV and a large exciton binding energy of 60 meV. The band gap can be engineered by appropriate doping of metal cations like Cd and Mg. The magnetic ions having +2 charge state is readily incorporated into the host materials by replacing Zn cations. However, it doesn’t contribute at all with a viewpoint of doping. Heavy electron doping (>10 21 cm -3 ) was readily achieved via group III substitutional doping in contrast to the other II-VI compound semiconductors. ZnO crystallizes in the wurtzite structure (hexagonal, with a = 3.25 and c = 5.12 ). Also, it has electron (n-type) conductivity naturally but p-type conductivity can also be induced by using a co-doping technique [Jos99, Yam99]. Owing to the high Curie temperature expectation from theoretical calculations [Die00], ZnO has been extensively studied as a host matrix for the spintronics applications [Pea04b]. Mn-doped ZnO Epitaxial wurtzite n-type Zn 1–X Mn X O films (x<0.36, far beyond the thermal equilibrium limit of x~0.13 at 600 C) were fabricated on sapphire substrate by PLD. Itinerant electrons could be increased over 10 19 cm by doping Al. The valence state of Mn ion was determined to be Mn 2 + having spins of S = 5/2 [Fuk99]. However, M(T) curve revealed a spin-glass behavior, showing a stronger antiferromagnetic coupling [Fuk01]. Recently, ferromagnetism above room temperature (above 425 K) has been

PAGE 34

22 reported in both bulk and thin films of Zn 1–X Mn X O with low Mn concentration (x<4 at. %) [Sha03]. High resolution transmission electron microscopy (HRTEM) and electron energy loss spectroscopy (EELS) found that the Mn atoms distributed homogeneously and the +2 valence state of Mn from the intensity ration of Mn-L 3 of L 2 . Ferromagnetism is suppressed or disappeared in the presence of Mn clusters and/or formation of secondary phases such as Zn 2 Mn 3 O 8 , ZnMn 3 O 7 , and Mn 2 O 3 , which is assumed to be favored either at high Mn concentration or at high growth temperature. Co-doped ZnO While Dietl et al. predicted high T c in Mn-doped p-type ZnO [Die00, Sat00], electron doping of Co-doped ZnO was anticipated to stabilize the ferromagnetic state ab initio calculations based on the local density approximation [Sat01]. Since n-type ZnO is easy to form, it has attracted much interest. Ueda et al. reported ferromagnetism in Zn 1–X Co X O film (x=0.05-0.25) with T c of 280 K. Aluminum (1 wt. %) is added to the target for electron doping. Cobalt atoms are regarded as substitutional dopants from the linear change in c-axis lattice constant with cobalt concentration. Above 50 % of cobalt, the phase separation occurs into ZnO and CoO (antiferromagnetic, T N =291 K). However, the reproducibility was less than 10 %. The ferromagnetism in transition metal-doped semiconductors is often induced from the secondary magnetic phase. The effect of oxygen partial pressure and substrate temperature was investigated in Zn 0.75 Co 0.25 O film grown by PLD [Kim02c]. The homogeneous film is paramagnetic whereas inhomogeneous film where rock salt CoO and hexagonal Co phases are mixed exhibits ferromagnetism. The inhomogeneous film is obtained at high substrate temperature (T>600 o C) and low oxygen partial pressure (P<10 -5 Torr). This condition generates more oxygen vacancy, leading to cobalt

PAGE 35

23 clustering through facile diffusion. In a film prepared by sol-gel method, cobalt clusters are formed when the x is over 0.12 in Zn 1–X Co X O [Par04]. It showed superparamagnetism with nanometer-sized clusters dispersed in ZnO matrix, as manifested by M(T) curve and HRTEM. Hexagonal and cubic structures of cobalt are present together. In an implanted sample, the cobalt is (110) oriented hexagonal phase in the ZnO matrix [Nor03, The03]. On the other hand, the room temperature ferromagnetism has been reported in intrinsic Co-doped ZnO films [Ram04, Rod03, Yan02]. The uniform distribution of cobalt in ZnO was proved by Auger mapping and cross-sectional HRTEM. This controversy between research teams may result from the growth method used and/or from the growth conditions (oxygen pressure, deposition temperature, etc). SnO 2 -based DMS Perceived by the high T c property in rutile TiO 2 -based DMS [Mat01b], other host oxides having the same rutile-type structure was examined. SnO 2 possesses interesting optical and electrical properties. It has a wide band gap of ~3.6 eV and can have large n-type carrier concentration up to ~10 21 /cm 3 which is potentially promising for spintronic devices. Mn-doped SnO 2 Epitaxial Sn 1-X Mn X O 2 :Sb (X < 0.34) films were fabricated on sapphire substrate by PLD. Mn ions are also soluble in SnO 2 films up to 30% moles, which is higher than the thermal equilibrium, mainly because of the non-equilibrium growth process of PLD. The high n-type carrier concentration (10 20 /cm 3 ) is obtained by Sb doping and it showed giant positive magnetoresistance as high as 60 % at 5K. However, it exhibits paramagnetic behavior down to 5 K [Kim01, Kim02b].

PAGE 36

24 Co-doped SnO 2 While Mn-doped SnO 2 films do not exhibit ferromagnetism, the Co-doped films (Sn 0.95 Co 0.05 O 2 film) exhibits high Curie temperature close to 650 K with giant magnetic moment of 7.5 0.5 B per cobalt atom [Oga03]. Cobalt atoms seem to be soluble up to 27 % and they distribute uniformly in the SnO 2 matrix, examined by HRTEM and EELS. The films are semiconducting and transparent. Rutherford backscattering spectroscopy (RBS) channeling data rarely exhibits cobalt channeling, indicating that significant amount of cobalt appears to reside in interstitial sites or in substitutional sites with incoherency due to the oxygen vacancy related distortions. The magnetic moment is as high as 7.5 0.5 B per cobalt atom compared to the value of cobalt metal (~1.67 B /Co), small cobalt clusters (~2.1 B /Co) [Buc91], and any of cobalt oxide compounds where the orbital moment is quenched. Such a giant magnetic moment can arise from unquenched orbital contributions. It has been reported that magnetic moment can be much larger than spin-only moments [Bec99]. In this case, the moment rapidly decreases with dopant-dopant association, leading to orbital moment quenching. After annealing of Sn 0.95 Co 0.05 O 2 film, the magnetic moment reduced to 2.5 B /Co, supporting this supposition. Origin of Ferromagnetism It has been reported that Co-doped TiO 2 is ferromagnetic well above room temperature but the origin of observed ferromagnetism remains uncertain. While carrier-induced interaction between the magnetic atoms (e.g. Mn, Co) is suggested as the cause of underlying ferromagnetism in the DMS [Die97], the precise mechanism in Co:TiO 2 is still controversial. It needs to be carefully identified whether all reported ferromagnetism

PAGE 37

25 are indeed intrinsic magnetic behavior or are arising from clustering and segregation. Generally, the magnetization measurement as a function of magnetic field has been used for the evidence of ferromagnetism. However, only magnetization measurement without thorough consideration of possible extrinsic magnetic effects from ferromagnetic inclusions or precipitates, often misleads into wrong results. Extremely careful characterizations and analysis are required to exclude the possibilities of the magnetic signal being provided from the precipitation of ferromagnetic compounds. One of the reasons to make it difficult is that one still cannot rule out the existence of ferromagnetic secondary phases, even when the impurity phases are not detected by X-ray diffraction due to the limited resolution limit. The efforts to find decisive and effective techniques have been made. One way is to examine ferromagnetic responses that are caused by charge carriers like magneto-optical effect and anomalous Hall effect. Fukumura et al. [Fuk04] employed magneto-optical spectroscopy, which gives the magneto-optical signal as a function of photon energy. The MCD signal is generally enhanced at the absorption edge of the host semiconductor of DMS [And01] as a result of carrier mediated exchange interaction between localized spins. Absorption and MCD spectra for Co-doped anatase TiO 2 at 300 K showed ferromagnetic hysteresis for the each photon energy. As for magnetotransport properties, ferromagnetic DMS such as (Ga,Mn)As [Ohn96], Mn:Ge [Paro2b] show anomalous Hall effect [Mat98, Ohn96]. The Hall resistance R Hall in magnetic materials is expressed as R Hall = (R 0 /d)B + (R s /d)M, where R 0 is the ordinary Hall coefficient, d the sample thickness, Rs(=c), the anomalous Hall coefficient (: resistivity and c: constant), B the magnetic field, and M the magnetization of the sample [Ohn96]. Ordinary component provides carrier density and mobility information

PAGE 38

26 of semiconductors and anomalous component is typically proportional to the magnetization [Pea04]. Since the anomalous Hall term is dominant, the shape of the Hall resistance (R Hall ) reflects the field dependence of M. Although the appearance of anomalous Hall effect has not been reported for Co-doped anatase TiO 2 yet [Shi04], it was very recently reported in the Co-doped rutile TiO 2 [Hig04, Shi04, Toy04a]. The anomalous Hall effect might be a convincing evidence of intrinsic DMS but anomalous Hall effect has also been reported in two immiscible granular systems like Co-Ag [Xio92] and even in the Co-doped rutile TiO 2 film containing superparamagnetic clusters [Shi04]. Carrier Mediated Exchange Interaction Carrier-induced exchange interaction is believed to be the origin of ferromagnetism in (Ga,Mn)As DMS system. The well-known Ruderman-Kittel-Kasuya-Yoshida (RKKY) [Rud54, Kas56, Yos57], has been applied to account for the magnetism observed in DMS. RKKY theory is the theory of indirect exchange, which is used for the explanation of magnetism in the material like rare earth metals, where unpaired 4f electrons are highly localized and thus their wave functions are do not overlap. This takes into account both conduction and localized electronic magnetic moment to understand the magnetic ordering in the material. The exchange coupling occurs via the polarized conduction electrons, leading to magnetism. It is demonstrated that the ferromagnetic transition temperature calculated from RKKY interaction using the exchange constant and the hole concentration has been in good agreement with the experimentally obtained T c in (Ga,Mn)As [Mat98]. Magnetic coupling occurs by virtue of exchange interactions between the magnetic spins and free carriers in the semiconductor. The interaction can occur via p-d or d-d exchange between

PAGE 39

27 carriers and transition metal d spins, and can lead to antiferromagnetic or ferromagnetic coupling, depending on the concentration and the local structural environment of the magnetic impurity. Due to the spin-dependent coupling between localized states and those in the valence and conduction bands, the carriers in diluted magnetic semiconductor mediate a ferromagnetic interaction between the localized spins [Die03]. For this reason, we are able to control magnetization by an electric method like gating, which affects the carrier density in semiconductors. Chamber et al. [Cha02b] claimed that Co-doped anatase TiO 2 is ferromagnetic by virtue of electron mediated exchange interaction between Co +2 cations that substitute for Ti +4 in the lattice. As expected by charge neutrality requirements, Co(II) substitution for Ti(IV) in the TiO 2 lattice results in oxygen vacancies. But it has been found that n-type semiconducting behavior and Co substitution are independent phenomena [Cha01b]. In other words, oxygen vacancies associated with substitutional Co do not contribute to the electrical conductivity of the material but the free electrons generated by oxygen vacancies are important, which are resulted from the growth conditions. Some highly resistive films are nonmagnetic despite having several percent of cobalt. Thus, substitutional Co(II) and oxygen vacancies in excess of those required to compensate Co(II) must be present for ferromagnetism to occur. The possible valence state of the Co ion is either Co +2 or Co +3 . If Co ions are in the high-spin state, the saturation magnetization of Co is ~3 B /Co while the low-spin state gives ~1 B /Co. The chemical state of Co is examined in the film by in situ XPS [Cha01b]. The Co x Ti 1-x O 2-x film spectrum matches the CoO reference spectrum very well, indicating that Co is in the +2 formal oxidation state. There is no photoemission at the energy measured for Co metal,

PAGE 40

28 indicating the absence of this species near the surface of Co x Ti 1-x O 2-x . Chamber et al. concluded that the ferromagnetism is induced by the interaction between the localized 3d spin of Co ion and the mobile 3d electron doped in the conduction band of the TiO 2 host, as expected in DMSs. Sullivan et al. [Sul03] performed calculations on various states of Co based on the density functional theory and found that Co ions in the interstitial sites had a spin state of S = 1 (Ms = 2 B /Co) and provided n-type carriers as measured by experiments while substitutional Co had a spin state of S = 1/2 (Ms = 1 B /Co). The spontaneous magnetization larger than 1 B /Co, therefore, can be accounted for by an additional contribution from interstitial Co ions. Experimentally, many seem to agree on that the ionic state of Co in Co:TiO 2 is divalent, but the spin states determined by XAS differ for different experiments. It was found to be in a low-spin state by Chambers et al. [Cha01a] and in a high-spin state by Kim et al. [Kim03a]. Ferromagnetic Nanoclusters The magnetic effects are roughly proportional to the concentration of the magnetic ions. However, formation of the secondary phase is favored when a high concentration of magnetic elements is introduced in excess of the solubility limit. Since magnetic ions typically exhibit low solubilities in their respective host semiconductors, the issues of secondary-phase formation must be carefully addressed. Kim et al. [Kim02] first reported the cobalt clusters as the source of the room temperature ferromagnetism. They are primarily formed at the interface between the film and SrTiO 3 substrate and become more pronounced in the films fabricated in less oxidizing environment [Kim03b]. They also investigated the origin of ferromagnetism using the X-ray magnetic circular dichroism (MCD) at the Co L 2,3 absorption edges. The

PAGE 41

29 substituted Co has a high spin Co +2 as in CoO, but the magnetic signal was found to be identical to that of cobalt metal, showing that the ferromagnetism is induced by a small amount of clustered cobalt [Kim03a]. Shinde et al. [Shi03] found evidence of several 20-50 nm cobalt clusters as well as a small concentration of Co incorporated into the remaining matrix in the Ti 0.93 Co 0.07 O 2sample, which showed a high T c of over 1180 K. The solubility of cobalt was limited to ~ 2% in the as-grown films and the formation of Co clusters was followed after that. The origin of ferromagnetism is attributed to the coexisting contributions of cobalt metal clusters and that of dispersed matrix-incorporated cobalt. High temperature treatment is shown to enhance the incorporation of Co into the matrix. Chambers et al.’s report calls attention in that Co:TiO 2 is very susceptible to inhomogeneities [Cha03]. Even though the film is deposited under the nominally identical condition, in some films highly Co-enriched TiO 2 anatase clusters nucleate on the epitaxial anatase TiO 2 grown on LaAlO 3. These nanoscale Co X Ti 1-X O 2-X anatase particles nucleate on top of the continuous epitaxial film and most ferromagnetism was focused on these particles, as evidenced by MFM.

PAGE 42

30 Figure 2-1. Schematic diagram of a diluted magnetic semiconductor where a sizable portion of atoms is substituted by a magnetic element [Ohn98b].

PAGE 43

31 Figure 2-2. Predicted Curie temperature as a function of bandgap of semiconductors [Pea04c].

PAGE 44

32 Figure 2-3. Calculated values of Curie temperature for various p-type semiconductors at Mn concentration of 5% and hole concentration of 3.510 20 /cm 3 [Die00].

PAGE 45

33 (a) (b) Figure 2-4. Crystal structure of TiO 2 . (a) Anatase and (b) Rutile.

PAGE 46

34 Figure 2-5. Combinatorial fabrication of TiO 2 films doped with transition metals.

PAGE 47

CHAPTER 3 EXPERIMENT Target Preparation Cobalt doped titanium oxide targets were made by conventional ceramic fabrication method. The high purity powders of Co 3 O 4 (99.9985%) and TiO 2 (99.995 %) were mixed, ball-milled and pressed uniaxially and isostatically into a one-inch diameter pellet. It was then sintered at 1300 o C for 24 h in air atmosphere. Highly dense and compact structure was verified by scanning electron microscope image as shown in Fig. 3-1. For the deposition of SrTiO 3 and TiN layer, commercial targets were used. The purity of both targets is 99.9 %. Substrate Material LaAlO 3 For the epitaxial growth of an anatase TiO 2 phase, two-side polished LaAlO 3 (LAO) was used as an oxide-based substrate since it shows extremely smaller lattice mismatch (-0.2%) with anatase TiO 2 . LAO has been widely used as a substrate for the superconductor film applied for high frequency and microwave. LAO has a rhombohedral (a=5.357 , c=13.22 ) structure at room temperature but can be regarded as pseudo-cubic with a=3.793 . It transforms into cubic structure above 435 o C. Silicon Two-inch diameter p-type Silicon (100) substrate was also used as a substrate. The resistivity of silicon is in the range of 1-10 cm and the thickness of Si is around 275 35

PAGE 48

36 m. The thin substrate is favored in order to reduce the signal from the substrate in magnetic property measurements. Just prior to transfer to the chamber, the silicon substrates were cleaned in acetone and methanol sequentially in an ultrasonic bath, followed by a treatment in 10 % HF solution in order to provide hydrogen terminated silicon surface. Crystallographic information of LaAlO 3 , SrTiO 3 , TiN, TiO 2 and Si such as crystal structure and lattice parameter is listed in Table 3-1. Thin Film Fabrication All layers were fabricated using a pulsed-laser deposition system equipped with a multi-target holder and an ultraviolet (UV) lamp. The schematic diagram of the system and the picture of inside of chamber were presented in Fig. 3-2. The rotating target was ablated by a KrF excimer laser (=248 nm) at a fluence of 0.5-1.5 J/cm 2 , a repetition rate of 5-10 Hz and an oxygen pressure of 3-5 10 -5 Torr. In the case of UV-assisted film growth, the film was irradiated by in situ UV radiation during laser ablation. A vacuum compatible, low pressure mercury lamp having a fused silica envelope which allows more than 85 % of the ~185 nm radiation to be transmitted was fitted in the chamber. Before the film deposition, UV lamp was warmed up for a couple of minutes. The temperature of the substrate was controlled and monitored by a thermocouple and an infrared pyrometer. PLD (Pulsed Laser Deposition) The Pulsed Laser Deposition (PLD) is simple and versatile technique especially for oxides and chemically complex material film deposition in which the photon energy of a laser beam is used to generate the atomic flux. This method was attracted after the

PAGE 49

37 successful growth of high temperature superconductor film without further processing [Dij87]. It opens the wide application of PLD mainly in the multicomponent metal oxide films due to simplicity and stoichiometric transfer. The wavelength of laser used in PLD is in the range of 190-400 nm (ArF: 193 nm, KrF: 248 nm, XeCl: 308 nm) since most materials show strong absorption in this spectral region. When the ultraviolet pulsed laser hits the rotating target material, its electromagnetic energy is absorbed by the solid surface of the target, being converted into electronic excitation and then into a combination of thermal, chemical, and mechanical energy. This energy results in the evaporation, excitation, ablation, and plasma plume generation (the atomic flux) from the target material. The evaporant thus obtained consists of a mixture of energetic species including neutral atoms, molecules and ions in both ground and excited states, electrons, clusters, micron-sized solid particulates, and molten globules having a very short collision mean free path. Due to this aspect, immediately after the laser beam impact with the target material, the plasma plume adiabatically expands perpendicular to the target surface forming a nozzle jet (velocity > 10 6 cm/s) [Low96], thus giving the PLD technique the advantages of flexibility, fast response, high energetic evaporants, and congruent evaporation. Depending on the substrate and its temperature, amorphous, polycrystalline, or epitaxial films can be grown by this method. Characterization Tools X-ray Diffraction (XRD) XRD has been used as a representative nondestructive analysis tool for the crystallographic information. In principle, when a monochromatic X-ray beam with a certain wavelength is incident on a crystalline material at a specific angle, diffraction

PAGE 50

38 occurs only when the distance traveled by the rays reflected from successive planes differs by a complete number of wavelengths. The crystalline structure and out-of-plane orientation of the film (/2 scan mode) were determined by Philips APD 3720 system. The standard diffraction geometries (-2 scan) provides information only for the parallel planes to the surface of the sample. Thus, high resolution XRD system (Philips X’ Pert MRD) was employed to examine the crystalline properties of the epitaxial film such as crystallinity (rocking curve), the in-plane orientation relationship between the deposited layers and the underlying substrate ( -scan), crystal symmetry (pole figure). All measurements were performed with Cu K radiation. Atomic Force Microscopy (AFM) / Magnetic Force Microscopy (MFM) AFM is widely used to measure the surface property like surface roughness and nano-sized patterned structure. AFM generally uses very sharply etched silicon (or silicon nitride) tip attached to a cantilever. The tip is placed close to the sample and raster-scans over the sample surface, experiencing a weak Van der Waals force. The interactions between the tip and sample surface is monitored by measuring changes in vertical deflection of the cantilever with high resolution by using a segmented photodiode detector. A feedback loop maintains a constant deflection between the cantilever and the sample by vertically moving the scanner at each data point to maintain a set point deflection. MFM is a secondary imaging mode derived from tapping mode that maps magnetic force gradient above the sample surface [Har99]. The silicon probes are sputter-coated with a ferromagnetic material like cobalt. The tip scanned the surface without contact through two-pass technique, lift mode. Lift mode first records topography by tapping

PAGE 51

39 mode and then magnetic property using the topographical information to track the probe tip at a constant height (lift height, typically 20-200 nm) above the sample surface. Magnetic field gradient yield different force on the tip and this response yields a magnetic force image. XPS (X-ray Photoelectron Spectroscopy) XPS is based upon a single photon in/electron out process. Photoelectron spectroscopy uses monochromatic sources of radiation (i.e. photons of fixed energy). The specimen considered is illuminated with monoenergetic soft X-ray (typically Mg K (h = 1253.6 eV) or Al K (h = 1486.6 eV)) and the photoelectric effect results in the ionization and the emission of core (inner shell) electrons. The kinetic energies of the emitted photoelectrons are in the range of 0-1500 eV . Owing to the very short inelastic mean free path, XPS is one of the most surface sensitive techniques. The number of emitted photoelectrons as a function of their kinetic energy is measured by electron energy analyzer and this is the photoelectron spectrum. In general, the binding energy of the orbital from which the electron was expelled is taken as a function from the difference between the incident photon energy (known value) and the electron kinetic energy (measured value). Since each core orbital has its own associated binding energy, each element produces a characteristic set of peaks in the photoelectron spectrum. In addition, it also provides the chemical state information of the elements present from small variations in the determined kinetic energies. Namely, photoelectron spectroscopy utilizes photo-ionization and energy-dispersive analysis of the emitted photoelectrons to study the composition and electronic states of the surface region of a sample.

PAGE 52

40 In situ Ar sputtering is typically used for cleaning the surface or depth profiling. The contribution of the different chemical species is determined by deconvolution of the spectra. In this study, the chemical state of constituent elements in a film was identified by X-ray photoelectron spectroscopy (Perkin Elmer 5100) with a monochromatic Mg K radiation (1253.6 eV). TEM (Transmission Electron Microscopy) The detailed microstructure and the cobalt distribution were revealed by transmission electron microscopy (JEOL 2010F) with an incident electron energy of 200 keV. The high resolution image is very helpful for the determination of possible cobalt nanoclusters, any interface reaction or interfacial layer, and the detailed atomic structure of films as well as the interface. The crystallographic relationship and single crystallinity of the films were also identified by electron diffraction. The film thickness was measured from low magnification bright field image. Samples were prepared by mechanical grinding, polishing, disc cutting and dimpling, followed by Ar + ion milling. TEM samples were cured with M-610 bond at 150 o C to minimize the heating effect. The ion milling was executed using the precision ion polishing system (PIPS). The guns were oriented at an angle of 3-8 o with respect to the sample, and a voltage of 2.5-4.0 keV was used during milling. SQUID (Superconducting Quantum Interference Device) Magnetic measurements were made with a Quantum Design superconducting quantum interference device (SQUID). The SQUID magnetometer is the most sensitive method used for measuring magnetic fields. The commercial rf-SQUID (Quantum Design MPMS-5S) has a range of 10 -8 emu to 2 emu. SQUID has a superconducting loop with one or two Josephson junctions in the loop’s current path. Because of the quantized

PAGE 53

41 states of the superconducting ring, SQUID are able to resolve changes in external fields ranging from small to large fields. The SQUID does not directly detect the magnetic field from the sample. As the sample moves through a series of superconducting detection coils, the magnetic moment of the sample induces a current in the detection coils, which is inductively coupled to the SQUID sensor. Any change in magnetic flux yields a change in the current flowing in the SQUID input coil, which is proportional to the output voltage [Van81]. The SQUID functions as an extremely sensitive linear current-to-voltage converter. Magnetization measurements Plastic tube container much like a drinking straw is used as a sample holder. As long as the sample holder is uniform along the scan distance, it induces no magnetic signal. The background signal (~10 -8 emu at low fields) is generally weaker by two orders of magnitude than the signal from the sample. Before the measurement, the sample is degaussed to remove any remanent magnetization. In most of the raw data in this study, hysteresis loops contained a negatively sloped component due to the diamagnetism of the LaAlO 3 and Si substrates. The slope ( dia < 0) of linear response was fitted and the only ferromagnetic component was extracted. Magnetic measurements were made with an applied field parallel to the film surface. Before the measurement, the silver paste on the backside of the substrate was removed and cleaned to avoid any signal from the contaminants. The Curie temperature was estimated from the magnetization versus temperature measurement. Most samples show the high Curie temperature over 350 K , which is the limit of our SQUID system.

PAGE 54

42 The vibrating sample magnetometry having high temperature measurement option will be suitable for the precise measurement of the Curie temperature. Zero field cooled (ZFC) – Field cooled (FC) measurement Since the hysteresis can be also observed from the superparamagnetism, the ferromagnetic hysteresis behavior was manifested by zero-field-cooled/field-cooled (ZFC/FC) set of measurements [The02]. In addition to the hysteresis in M(H) curves at fixed temperature, magnetization behavior as a function of temperature can also show a dependence on the magnetic field history if magnetization is not saturated. One way of determining whether irreversibility exists or not is to do a ZFC/FC set of measurement. After demagnetizing the sample at 300 K, it is cooled to the lowest measurement temperature (T=10 K) in zero field (H=0). After stabilization, a certain magnetic field is applied and the moment is measured as a function of temperature increasing the temperature (ZFC measurement) Once the intended highest temperature is reached, the sample is cooled down again with the same field and the magnetization is recorded again with increasing temperature (FC measurement). During the whole ZFC/FC experiment the magnetization (M) along with the field is measured. This ZFC/FC method gives information on whether the system is irreversible in the temperature range considered. Superparamagnetism When the size of the magnetic particle decreases, the thermal energy exceeds the magnetic anisotropy energy below a certain size and the superparamagnetism occurs, where there are non-correlated orientations of magnetic moments. This phenomenon is getting important especially in ultrahigh density recording as the magnetic bit size is decreasing to increase the areal density [Sku03]. As shown in Fig. 3-3, with decrease of magnetic volume size, the magnetic energy per particle sustaining the magnetic moment

PAGE 55

43 along certain directions becomes comparable to the thermal energy. The thermal fluctuations randomize the magnetic moment of the particles. They lose their magnetic order making them superparamagnetic. In some diluted magnetic ferromagnetic semiconductor system, there has been reported that the ferromagnetism is originated from the magnetic precipitates or clusters because of the limited solid solubility of the magnetic element in a semiconductor host. Since the doping amount is typically small, the diameter of clusters may be in a nanometer size regime. Below some critical size, a magnetic particle can be viewed as a single large spin and this magnetic moment aligns along one of the easy axes of the particle at low temperature. This situation is referred that the dispersion is blocked. Above blocking temperature, thermal fluctuations can overcome the anisotropy barrier, randomizing the magnetic moment [Hel00]. In the conventional superparamagnetic system, magnetization (M) increases aligning with the applied field with increasing temperature in a ZFC measurement until the blocking temperature (T b ) is reached. At T b , the thermal energy forces the magnetic moments to fluctuate reducing the net magnetization. Therefore, M decreases with increasing temperature above T b (in the superparamagnetic regime). In the FC experiment, M coincides with the ZFC curve after the blocking temperature is reached. With the known value of anisotropy constant, the T b of cobalt-based superparamagnetic system is assumed, as shown in Fig. 3-4. Depending on the stability criteria (typically K u V=25k B T, where K u : uniaxial anisotropy constant, V: volume of a particle, k B : Boltzmann’s constant, T: absolute temperature) and the anisotropy constant, it shows some variations. If there is any magnetic cluster in cobalt-based DMS host, it

PAGE 56

44 might be able to be detected by using ZFC/FC curve measurements. If the size of cobalt particle is, however, larger than approximately 10 nm, the blocking temperature readily exceeds room temperature. Furthermore, blocking temperature is also affected by strength of dipolar interaction [Han95, Nie03, Sun99]. In dense particle systems, the mutual interactions between the magnetic particles make an influence on dispersion, thus increasing T b [Gar00]. Though the only ZFC/FC measurement may be insufficient to determine if the DMS is intrinsic, it can provide some insight whether magnetic particles are present.

PAGE 57

45 Figure 3-1. Scanning electron microscopy image of the Co-doped TiO 2 ceramic target showing highly dense microstructure.

PAGE 58

46 Table 3-1. Crystallographic information of LaAlO 3 , SrTiO 3 , TiN, TiO 2 and Si. Material Crystal system Lattice parameter Thermal expansion coefficient % Lattice mismatch LaAlO 3 Rhombo-hedral a=5.357 , c=13.22 (pseudo-cubic with a=3.793 ) ~ 10.010 -6 / o C (a LAO -a anatase ) / a LAO = 0.2 % SrTiO 3 Cubic a=3.905 ~ 11.110 -6 / o C (a STO -a anatase ) / a STO = 3.1 % Si Cubic a=5.431 ~ 4.010 -6 / o C TiN Cubic (NaCl) a=4.242 ~ 9.010 -6 / o C (a Si -a TiN ) / a Si = 22 % Rutile-TiO 2 Tetragonal a=4.593 , c=2.959 ~ 8.210 -6 / o C Anatase-TiO 2 Tetragonal a=3.785, c=9.514

PAGE 59

47 (b) (a) Figure 3-2. Pulsed laser deposition system. (a) Schematic diagram of pulsed laser deposition system. (b) Side view of ultraviolet assisted pulsed laser deposition chamber.

PAGE 60

48 igure 3-3. Change in coercive field (Hci) as a function of particle diameter. Ds and Dp denote the diameter at which domain becomes single and superparamagnetism begins, respectively [Cul72]. Stable Multi-Domain Sin g le-Domain Superparamagnetic Unstable Hci D p Ds 0 Particle F

PAGE 61

49 Figure 3-4. Estimated blocking temperature as a function of size of a magnetic particle. 0102030405060708090100110010020030040050600700 0 SuperparamagneticK K=4.5*106 erg/cm3 K=2.7*106 erg/cm3StableBlocking Temperature ()Critical Diameter of a Magnetic Particle (A)

PAGE 62

CHAPTER 4 E PITAXIAL GROWTH OF COBALT-DOPED ANATASE TiO2 ON LaAlO3 Overview temterials 0thod momray absorption analysis [Cha02a]. He claimed that the ferromagnetism originated from the electron-mediated exchange interaction between Co+2 cations that substitute for Ti+4 in the lattice, arguing against the formation of cobalt metallic nanoclusters [Kim02a]. Most of the growth of the cobalt-doped TiO2 film has been carried out at around 700 oC, a relatively high temperature [Cha01a, Mat01a], where undesired diffusion of cobalt atoms and substantial interdiffusion were reported [Cha01a, Cha02b]. A low temperature process will have many advantages for device fabrications such as the retardation of the undesired diffusion of constituent atoms and the minimization of reaction of the film with the underlying substrate and the suppression of phase separation. Ferromagnetic semiconductors have been actively investigated for spin injectors into semiconductor heterostructures and materials having Curie temperatures above roomperature are promising for practical applications. Some wide band gap DMS mbased on ZnO and GaN were predicted theoretically from the Zener model [Die0cobalt doped TiO2 was experimentally discovered by a combinatorial deposition m[Mat01a] for the high Tc materials. Chamber et al. [Cha01a] reported improved properties like higher magnetic ents of 1.26 B/Co atom and larger remanence when activated oxygen, such as oxygen plasma, was used during the film growth process. The oxidation state of cobalt in the film was found to be +2 from both Co 2p core level photoemission and Co L-edge Xa ] while e 50

PAGE 63

51 We have reported many beneficial effects of ultraviolet assisted pulsed laser deposition (UVPLD) upon the crystalline quality and properties of the oxide films [Cra00a, Kum02, Sin02, Sri99] and this effect was more pronounced at relatively low temperatures [Cra99, Cra00e reaction of metal oxidation (M + O MO) and the oxygen dissociation reaction (O2 2O) on the substrate surface should occur faster than the film growth rate [Cha00]. However, even if the growth rate of metal oxide film is slow, the molecular oxygen source can not be active enough to fully oxidize metal atoms during the growth [Cha00]. The active oxygen source will enhance the reactivity toward metal atoms. Irradiating in situ ultraviolet light in an oxygen atmosphere is one of the methods which can generate the reactive oxygen species. The UV light which has the wavelength less than 243 nm, corresponding to the binding energy of oxygen (5.1 eV), helps to break down the -3h 2e en at g the 2 b, Yan03]. In general, in order to control the stoichiometry of the oxide film, t h molecular oxygen into radicals such as atomic oxygen and ozone that exhibit high reactivity during the deposition process. At the same time, it increases surface adatom mobility and reactivity toward metal cations. In my study, the main interest was focused on the properties in the films deposited at relatively low temperature in the range of 400550 oC. Experiment LaAlO was chosen as a substrate due to the smaller lattice mismatch (-0.2%) witanatase TiO [Mur01a]. Since the metastable anatase phase can be stabilized on a latticmatching substrate through a template effect [Gor02], the anatase phase maintains ev800 oC. This indicates that substrate structure can play a significant role in controllinheteroepitaxial growth of TiO films. Crystallographic matching is important because

PAGE 64

52 spin decoherence due to spin-flip scattering is correlated with defect density at the interface of diluted magnetic semiconductor and conventional semiconductor [Cha02b]. Generally speaking, the structure of the obtained films is strongly dependent on the substrate used, substrate temperature, and the gas pressure. The substrates were cleanedin acetone and methanol sequentially, followed by a rinse in deionized water. The two-side polished LaAlO3 (001) substrates were mounted to a substrate holder using silvpaste for the efficient heat transfer. Ti0.93Co0.07O2films with a thickness of several tof nanometers were prepared at temperatures ranging from 400 up to 700 oC. The ablation was performed with a KrF excimer laser (=248 nm) at a fluence of 1 J/cm2, a repetition rate of 10 Hz and an oxygen pressure of 3 10 -5 Torr. The effect of Ultravioleon the film property was investigated by comparing the properties of the film which wagrown by PLD with the ones by UVPLD. When the film was grown by UVPLD mit was irradiated by in situ UV radiation during laser ablation. A vacuum compatiblpressure Hg lamp having a fused silica envelope which allows more than 85% of the emitted 184.9 nm radiation (~6% of the output) to be transmitted, was added to the PLD system. Before the film deposition, UV lamp was warmed up for a couple of minutes. er ens t s ethod, e, low Results and Discussion The structural properties of the flyzed by -2 scan, rocking curve and p ilms were ana ole figu r e measurements with a high resolution X-ray diffractometer (Philips X-pert MRD). Figure 4-1(a) shows the X-ray diffraction pattern and the rocking curve of a UVPLD grown film at a substrate temperature of 550 oC. All samples exhibited similar diffraction profiles, which were deposited in the substrate temperature of 400-700 oC. XRD patterns revealed that all deposited Ti0.93Co0.07O2films on LaAlO3 (001)

PAGE 65

53 substrates had an orientation of c-axis normal to the surface of the substrate. No secondary phase like cobalt metal and anatase phase were detected within the resolution limit. In order to determine the crystalline quality of the films, rocking curves was performed. The full width at half maximum value (FWHM) of (004) rocking curve of thesample deposited at 550 oC was found to be 0.18o, whereas the corresponding one fromPLD grown films was in the regime of 0.20-0.25o. The FWHM value was 0.46o in the UV-assisted film deposited at 400 oC. .9o) of (101) plane that are inclined at ~ that are parallel to the substrd al e VPLD It is n ote d th at this is larger than the value (0.26o) of the film deposited at 700 oC without UV assistance but smaller than tho s e (0.8~0films grown on (001) SrTiO3 substrate [Cha01a]. Since the -2 scan provides information only for the parallel planes to the surface of the sample, the degree of in-plane alignment in the films was analyzed by pole figure measurement. As shown in Fig. 4-1(b), the profiles showed four poles corresponding to the {101} plane with narrow intensity distribution and exhibited four-fold rotational symmetry, where each peak is separated by 90o. This result indicates that the film has an epitaxial relationship to thesubstrate. The in-plane orientation of the grown films was also assessed by phi scans of the 68 from the (001) planes ate. As shown in Fig. 4-2, UVPLD film grown at 400 oC exhibits much sharper anhigher intensity peaks, which are 90 apart, the signature of four-fold symmetry. The observed four-fold symmetry indicates that this film grows in a cube-on-cube epitaxiorientation with the [001] normal to the sample surface. The PLD grown film exhibited some epitaxy, with peaks separated by 90, but several other peaks from regions th a t arnot well aligned with the substrate were als o v isib le. T he higher crystalli n ity in U

PAGE 66

54 grown film is attributed to the fact that short wavelength UV radiation dissociatesmolecular oxygen into atomic oxygen and ozone which are more reactive gaseous speciesin addition to enhanced adatom surface mobility. Individual Ti, Co and O atoms appto reside their respective sites of minimum energy, which are regular cation and anion sites in the crystal lattice ears ata as ut 350 Oe and the 45% remanence (Mr) of the magnetization saturation. In compd m . Magnetic measurements were made with a Quantum Design superconducting quantum interference device (SQUID) with an applied field parallel to the film surface. Before the measurement, the silver paste on the backside of the substrate was remove d and cleaned to avoid any signal from the contaminants. The measured original dinclude the magnetic moment not only from the film but also from the substrate. The diamagnetic contribution from the LaAlO3 substrate was subtracted from the measurement for the correction. The magnetization versus magnetic field graph, measured at room temperature, of a Ti0.93Co0.07O2film grown by UVPLD at 550 oC wshown in Fig. 4-3(a). The curve showed a well-defined hysteresis loop with coercivity (Hc) of abo arison, Hc and Mr in the PLD grown film under the same condition were 190 Oe an33%, respectively. Figure 4-3(b) shows the variation of coercivity as a function of temperature which was measured from the M-H loop at different temperatures. Higher coercivity was measured with decreasing temperature. Over the temperature range fro10 K to 300 K, the UVPLD grown film showed stronger hysteresis behavior than its counterpart grown by PLD. Figure 4-4 shows the magnetization as a function of magnetic field at room temperature for the Ti0.93Co0.07O2film grown at 400 oC by UVPLD [Yan03]. The curve

PAGE 67

55 exhibited hysteresis behavior and the measured coercive field and the remanence wapproximately 180 Oe and 17% of the saturation magnetization, respectively. When the film was deposited by conventional PLD, a very noisy hysteresis curve was obtained. TheFWHM values of a (004) anatase peak, coercive field and Mr/Ms at 300 K in the sampledeposited at 400-700 oC wa ere s s listed in Table 4-1. -as with the (FC C data surface, examined by atomic force and magnetic force micro Since the hysteresis can be also observed from the spin glass or superparamagnetism, the ferromagnetic hysteresis behavior was manifested by zero-fieldcooled/field-cooled (ZFC/FC) set of measurements [The02]. After demagnetizing the sample, the sample was cooled down to 10 K without a field and the magnetization wmeasured with increasing temperature under the field of 500 Oe (ZFC measurement). Once the highest temperature was reached, the sample was cooled down againsame field and the magnetization was recorded again with increasing temperaturemeasurement). The subtraction between the fie ld co oled (FC) and zero field cooled (ZFC) magnetization eliminates paraand diamagnetic contributions and indicates the presence of hysteresis if the difference is non-zer o [The02]. Figure 4-5 shows ZFC/FC measurements and the difference between them (inset) as a function of temperature in the UVPLD grown film. They clearly show nonzero difference between the FC and ZFindicating the hysteresis. Similar graphs were obtained in the PLD grown film. Distinct blocking behavior, which is characteristic of the nanosized magnetic particle system was not found and the hysteresis persisted over 300 K. In addition, we could not find segregated clusters on the film scope (AFM/MFM) images. These results suggest that the film possesses an intrinsic diluted magn etic se mi conductor characteristic.

PAGE 68

56 The Curie temperature was estimated to be well over 350 K in all samples from thmagnetization versus temperature measurement. The vibrating sample magnetometry having high temperature measurement option will be suitable for the precise measurement of the Curie temperature. Compositional depth profiling was perform e ed using Auger electron spectroscopy (AES instant oxygen ly ed for cially in transition metal oxide ). No cobalt segregation was observed at the surface owing to thegas stop after termination of the deposition [Cha01a]. The distribution of cobalt as well as other elements was uniform all the way through the depth of the film at the substrate temperature of 400 oC. At higher deposition temperature such as 700 oC, the significant interdiffusion between the Ti0.93Co0.07O2film and the LaAlO3 substrate was observedbecause of the increased diffusivity. The distribution of cobalt in the film was investigated by cross-sectional TEM analysis. As shown in Fig. 4-6, no cobalt clusters and segregation were observed. Epitaxial growth was confirmed by selected area electron diffraction patterns. The nearuniform distribution of cobalt and titanium was verified by energy dispersive spectroscopy (EDS) dot mapping of the selected film area. The chemical state of cobalt in a film was identified by X-ray photoelectron spectroscopy (XPS, Perkin Elmer 5100) with a monochromatic Mg K radiation (1253.6eV). The photoelectron take-off angle was set at 45o. XPS provides information on the chemical state of elements comprising the film and ion sputtering is frequently ussurface cleaning and depth profiling. However, the physical removal of surface atoms with energetic ions can induce substantial damage, surface morphology change andpreferential sputtering in multi-component systems. Espe

PAGE 69

57 systemter in situ fter the measurements, all binding energies were e amples, the oxidation states of Co and Ti were clearle .50.1 ing energy resence of some reduced statesiO2, nal s, the preferential sputtering of oxygen can change the chemical state of elements in the near surface region [Sor93]. In order to see if Ar ion bombardment has any influence, the XPS analysis was performed in an as-received form and then afAr sputtering with the energy of 4 keV. A calibrated by the carbon 1s peak position (284.6 eV). Figures 4-7 and 4-8 show the reference spectra of Ti(0), Ti(IV), Co(0), and Co(II). At first, the spectra of TiO2 were obtained without Ar sputtering and after every 2 min sputtering. In Fig. 4-9, it is clearly seen that the Ar sputtering makes a drastic impact on the valence state of Ti cation. Thfraction of reduced state increases with sputtering time. Figure 4-10 shows the XPS spectra obtained from both PLD and UVPLD grown Ti0.93Co0.07O2films. In as-received s y identified to be +2 and +4, respectively in both PLD and UVPLD samples. Thbinding energies of Co (II) 2p3/2 and Ti (IV) 2p3/2 were measured to be around 780eV, 458.1.1 eV, respectively, which are very close to the reference values [Wag79]. After Ar ion bombardment, however, the shoulders appeared at the lower bindsides near both the Co 2p3/2 and Ti 2p3/2 peaks, indicating a p . Those new peaks are assigned to metallic cobalt and Ti +3, respectively. In Tthe appearance of lower oxidation states after Ar ion bombardment is a well-known phenomenon due to the preferential removal of oxygen during the sputtering [Sul91]. Meanwhile, the appearance of metallic cobalt signal can be either from the metallic cobalt clusters, one of the most discussed issues in Co-doped TiO2 system, or from the reduction of cobalt ions because of the sputter-induced oxygen loss. The cross-sectioTEM data and ZFC/FC measurement implied effective incorporation of cobalt into TiO2

PAGE 70

58 matrix. Additionally, it has been reported that the metallization due to ion-bombardment-induced reduction can occur in a cobalt oxide system like CoO where the cobalt is+2 formal oxidation state [Cho89]. The preferential removal of oxygen makes the film surface oxygen-depleted and under this reducing environment, the cobalt atoms seem have less probability to bind to oxygen than the titanium atoms do, in terms of oxide heat of formation (H of (TiO2) = -944.1 kJ/mol, H of (CoO) = -239.1 kJ/mol) [Sm in the to is Ar ion sputtering. the d the i 76]. Thhas some analogy to the fact that a metallic cobalt phase formation was promoted in a lower oxygen partial pressure growth condition [Kim02a]. For these reasons, the metalliccobalt signal was attributed to the effect of the energetic Even though the preferential sputtering of oxygen was found in both PLD and UVPLD grown samples, there was a difference in the degree of reduction. Comparingspectra between PLD and UVPLD grown samples, the UVPLD grown samples exhibitea smaller fraction of reduced state indicating stronger binding surroundings between oxygen and metal cations. The ratios of Co (0):Co (II) and Ti (III):Ti (IV) were calculated to be 24% and 26% in PLD, and 5% and 16% in UVPLD grown films, respectively. This suggests that the re ac tive oxygen atmosphere during the UVPLD process results in the growth of a more stable oxide phase, forming stronger and homogeneous binding environment [Yan04a]. Summary The effect of in situ ultraviolet irradiation during the pulsed laser deposition onTi0.93Co0.07O2film grown on the LaAlO3 (001) was investigated at a substrate temperature of 400 oC and 550 oC. High quality epitaxial Ti0.93Co0.07O2films having enhanced magnetic properties could be achieved. Ti0.93Co0.07O2films were grown

PAGE 71

59 epitaxially at this low temperature with photonic assistance and hysteresis behavior was observed at room temperature. Enhanced crystalline and magnetic properties were thought to be due to the energetic photons and reactive gaseous species present during tdeposition. When the films are bombarded by Ar ions, oxygen atoms are preferentially removed from the surface and the subsequent atomic rearrangement is made in such a manner that the oxygen binds more favorably to the titanium making the cobalt atoms metallic. Much stronger binding coordination between oxygen and cations (Co and Ti) inferred from the smaller fraction of reduced states in the UVPLD grown films. No surface segregation was found at the surface of the film and the interdiffusio he is n between the Ti oC. 0.93Co0.07O2film and the LaAlO3 substrate was significantly suppressed at 400The successful epitaxial film growth at a temperature as low as 400 oC is expected to bring many technical advantages to spintronic device fabrication. These include kinetic restriction of reactions and interdiffusion between the layers comprising the device.

PAGE 72

60 (a) CoO2 (LaA3 (002La (0Co-T2 (0La3 (0 -Ti004)lO)AlO303)iO08)AlO01) Intensity (a.u.)2 (deg.) 17181920 =0.18oFWHM Figure 4-1. The results of X-ray diffraction. (a) X-ray diffraction pattern of the epitaxial Co-doped anatase TiO2 film grown on LaAlO3 (001) at 550 oC by UVPLD. The inset shows the rocking curve of the (004) diffraction of the Ti0.93Co0.07O2film. (b) A pole figure showing fourfold symmetry of {101} poles of the film. 2030405060708090

PAGE 73

61 Figure 4-2. X-ray phi scan difilms on LaAlO3 03600500100015002000 ffraction patterns of the Ti0.93Co0.07O2(001) grown at 400 oC by UVPLD and PLD, respectively. 60120180240300 PLDPhi (degree) 90 o90 o90 o90 oTsub=400 oCUVPLDntensity (a.u.) I

PAGE 74

62 Figure 4-3. The results of SQUID measurement. (a) Magnetization vs. magnetic field measurement measured at room temperature for the Ti0.93Co0.07O2film grown at 550 oC by UVPLD. (b) Temperature dependence of the coercivity for the Ti0.93Co0.07O2film grown at 550 oC by PLD and UVPLD, respectively for the temperature range from 10 K to 300 K. 6.0x10-5 -5000-2500025005000-6.0x10-5-3.0x10-50.03.0x10-5 at 300 KM (emu)H (Oe) 5007501000 (b) UVPLD PLD Hc (Oe) (a) 0100200300250 0 Temperature (K)

PAGE 75

63 erature for the 5000 Figure 4-4. Magnetization vs. magnetic field measurement at room tempTi0.93Co0.07O2film grown at 400 oC by UVPLD. -5000-4000-3000-2000-100001000200030004000-9.0x10-5-6.0x10-5-3.0x10-50.03.0x10-56.0x10-59.0x10-5 at 300 KM (emu)H ( Oe )

PAGE 76

64 at 300 K in PLD oC Table 4-1. FWHM values of a (004) anatase peak, coercive field and Mr/Msthe samples deposited at 400-700 oC. Tsub UVPLD @ 400 oC PLD @ 550 oC UVPLD @ 550 oC @ 700 FWHM of a (004) anatase peak 0.46 o 0.22 o 0.18 o Hc (Oe) at 300 K 180 190 350 0.26 o 290 M/M (%) at 300 K 17 33 45 35 rs

PAGE 77

65 Figure 4-5. Temperature dependence of zero-field-cooled (ZFC) and field-cooled (FC) magnetizations for the Ti0.93Co0.07O2film grown (a) at 550 oC by UVPLD and (b) at 400 oC by UVPLD. The inset shows the difference between FC and ZFC magnetizations. 0501001502002503004.0x10-68.0x10-61.2x10 -5 1.6x10-5 FC ZFCH = 500 OeTemperature (K) M (emu) Tem atur e T (K)M 0501001502002503000.04.0x10-68.0x10-61.2x10-5 0501001502002503001.0x10-51.5x10-5 T (K)MFCZFCH = 500 OeM (emu)per (K) 01002003000.05.0x10-6 (a) (b)

PAGE 78

66 Figure 4-6. The results of TEM analysis. (a) High resolution TEM image of film on ly LaAlO3 Co:TiO2 5 nm A(002)(b) (c) Co Ti (d) L(001) L(010) A(011) A(013) (a) LaAlO3 deposited at 400 oC. (b) SADP of anatase TiO2/LAO. (d) SADP of LAO. (d) EDS dot mapping of cobalt and titanium which are nearly uniformdistributed in the film.

PAGE 79

67 (a) (b) Figure 4-7. Reference photoelectron spectra of Ti 2p. (a) Ti(0) and (b) Ti(IV).

PAGE 80

68 Figure 4-8. Reference photoelectron spectra of Co 2p. (a) Co(0) and (b) Co(II). (b) (a)

PAGE 81

69 F igure 42040608010001200 Intensity 9. Photoelectron spectra of Ti 2p. The reduced states of Ti appeared with Ar sputtering. Ar sputtering effect was investigated by collecting the spectrum after every 2 min etching. (The red one: after 2min etching and the purple one: after 16 min etching). 470.0467.5465.0462.5460.0457.5455.000000000000 Ti(0) Ti(+4) Ar sputteredas-receivedTi 2pBinding Energy (eV)

PAGE 82

70 igure 4s grown by asAr ion 3/2 -10. XPS spectra of the Co 2p and Ti 2p of the Ti0.93Co0.07O2filmPLD and UVPLD, respectively. The lower spectra were obtained fromreceived samples while the upper ones from the samples after in situsputtering. (A: Co (II) 2p3/2, B: Co (0) 2p3/2, C: Ti (IV) 2p3/2, D: Ti (III) 2pbinding energy position). 805800795790785780775 PLDUVPLD 805800795790785780775 BACDDABCBinding Energy (eV) 468464460456 Co 2p Ti 2p468464460456 Intensity (a.u.) F

PAGE 83

CHAPTER 5 EPITAXIAL GROWTH OF COBALT-DOPED ANATASE TiO2 ON SILICON semcomthe basis of existing microelectronics process technologies and to make multifunctional devices on a sihigh quality Coubstrates. Generally, heteroepitaxy depends on lattice match, thermal expansion coefficients, and chemical reaction or interdiffusion with the underlying layer at the process temperature. The epitaxial growth of TiO2 was anticipated on Si (001) when anatase (001) and rutile (110) lattice planes are rotated by 45 about the silicon surface normal [001] axis, as shown in Fig. 5-1. For the direct heteroepitaxial growth of oxides on silicon, control of the silicon/oxide interface is Overview Epitaxial Growth of TiO2 on Si Room temperature ferromagnetism has been reported in both anatase [Mat01a] and rutile [Mat01b] phases of TiO2 thin films doped with cobalt, which are promising ferromagnetic semiconductors for practical spin-based electronic devices. To date, appropriate oxide-based substrates have been employed for the epitaxial growth of TiO2 phases using a template effect. For instance, SrTiO3 and LaAlO3 substrates [Cha02a, Mur01a] have been used for anatase, and Al2O3 for rutile phase stabilization. [Che93, Mat01b] The integration of a ferromagnetic semiconductor into a conventional iconductor like silicon is, however, of great interest from the viewpoint of the bination of novel spin functionality with current well-established semiconductor devices. In addition, it is expected to advance commercialization of spintronic devices on ngle chip. Many researchers are, therefore, interested in the preparation of :TiO2 films on silicon s 71

PAGE 84

72 critical. Thus, it is necessary to remove any native silicon oxide from the silicon wafer and to preclude its regeneration during the heating of substrate in oxygen environment of the TiO2 growth. When TiO2 film is formed on HF treated Si, the resultant film structure was, however, polycrystalline due to an amorphous silicon oxide layer that is typically formesubstrate. Ttreatmand mwas forsametempformcobalt-segregaclearly identifi 550 C. Root mean square (RMS) roughness value was small (0.261 nm) in the film deposited at 400 oC and it is not easy to tell if there is any foreign phase particle. The formation of magnetic particles was confirmed by the magnetization measurement as a function of temperature (M-T curve), as shown in Fig. 5-4. The transition temperature (Tb, blocking temperature) of zero field cooled curve ed at the interface, hindering the crystallographic matching between the film and thhe reason for the formation of a polycrystalline film was verified by depositing the TiO2 film on two differently treated substrates: one was prepared with a ent of HF acid (no native oxide on Si) and the other was just cleaned in acetone ethanol (native oxide on Si). Both substrates were loaded at the same time and filmmed under the identical condition. The X-ray diffraction profiles were almost the for both cases except some intensity difference, as shown in Fig. 5-2. The smaller intensity was obtained on the film grown on the silicon substrate having a native oxide. It was also found that the film exhibited different film texture depending on the substrate erature employed. In order to limit the growth of the interfacial disordered layer, the initial film growth was made without flowing oxygen gas but the film still grew in a polycrystalline . Instead, the initial oxygen-poor condition was found to promote the nucleation of ted particles [Cha03, Kim02a]. As shown in Fig. 5-3, some particles were ed on the film deposited ato

PAGE 85

73 (ZFC) of the sample deposited at 400 oC is observed below 300 K exhibiting some plateau, while the ZFC curve of the sample deposited at 550 oC does not show distinct Tb below 300 K. The blocking temperature is closely related with the size of the magnetic partiparticles (V) and anisotropy constant (Kbreases with increased diameter of the ements, it was found that the diameter of the magnetic particles was dependent upon the substrate temperature employed. The broad range of Tb in Fig. 5-4(b) is presumably due to the distribution of the particle size [Nie03]. In order to limit the regrowth of this interfacial disordered silicon oxide layer during the critical period of initial film growth, the following method was employed. The film was first deposited under vacuum conditions until about 2-5 nm of film growth whereupon the oxygen partial pressure will be increased to the value for the remainder of the Co:TiO2 deposition, but the film still grew in a polycrystalline form. Instead, the initial oxygen-poor condition was found to promote the nucleation of cobalt-segregated particles [Cha03, Kim02a]. Another potential issue is that there is a significant driving force for most dielectrics to react with silicon. Figure 5-5 shows stability of binary oxide on silicon and TiO2 has reactivity with silicon [Hub96, Sch02]. A thermodynamic approach was used to assess the thermodynamic stability of TiO2 in contact with silicon [Bey84a]. The calculated phase diagram is shown in Fig. 5-6. Ti-Si-O ternary phase diagram is one of the systems that are free of ternary phases. For a binary oxide to be stable in contact with silicon, a tie line must exist between MOx (M=metal) and silicon [Bey84b]. The existence of such a stable tie line constrains the rest of the phase diagram, since the crossing of tie cles [Mur01b]. Since the magnetic energy is the product of the volume of the ), the T inc particle. From the ZFC-FC measur

PAGE 86

74 lines is forbidden in terms of the Gibbs phase rule. The stable two-phase tie lines are established by calculating the Gibbs free energy of reaction at the point where two possible tie lines would cross. From the phase diagram, TiO2 was found to have a significant driving force to react with silicon at all temperatures between room temperature and 1000 oC. Thermodynamic prediction is a necessary, but insufficient criterion to conclude that a reaction will occur at the interface. It predicts the presa driving force for reaction, but it may be possible for a kinetic barrier to limit the extenof reaction. Epitaxial Growth of TiN on Si Due to the aforementioned difficulties, the use of a template layer is proposed as analternative way. High-quality epitaxial films can be grown on silicon through the use of structural templates and chemically stable buffer layers between the silicon substrate and the film. Ideally, a buffer layer for the growth of an oxide on silicon should provide a nonreactive layer and a s ence of t table nucleation template for the subsequent epitaxial growth of the ove of e N [Nar03]. In the plane, four unit cells of TiN match with three unit cells of silicon with erlying desired oxide. From this viewpoint, the proposed TiN layer will be onthe best candidate materials. TiN film has been investigated due to properties like wear resistance, corrosion resistance and low resistivity. It has especially been used as a diffusion barrier for the aluminum metallization in semiconductor devices. TiN is one of the few nitrides that are thermodynamically stable in contact with silicon, as shown in Fig. 5-7 [Hub96, Sch02]. Figure 5-8 shows the crystal system of TiN and Si which havthe same cubic structure with lattice parameters 4.242 and 5.431 , respectively. Even though there is a large lattice mismatch (~22%) between TiN and silicon, Tiis known to be able to grow epitaxially on silicon in terms of domain matching epitaxy

PAGE 87

75 less than 4.0% misfit. This domain matching epitaxy provides a new mechanism of epitaxial growth in systems with large lattice misfits. Cross s e ctional transmission electron microscopy revealed that the interface between TiN and silicon was very smootand free from any perceptible interdiffusion. The lattice misfit strain between the Tlayer and the silicon substrate was found to be r h iN Epitats of nce of mutually stable compounds, Si:SrSi2:SrO before the SrTiOt iO3 has e lieved by the introduction of dislocations at the interface [Cho96]. The epitaxial nature of the films was confirmed by the X-ray phi scan and the orientation relationship was revealed to be TiN(100)//Si(100) and TiN[001]//Si[001]. The TiN film aligned in a cube-on-cube manner. xial Growth of SrTiO3 on TiN/Si SrTiO3 has a simple cubic perovskite structure with a= 3.905 and about 1.7% lattice mismatch with Si (a=5.431 ), if the unit cell is rotated 45o. The lattice consisalternate layers of SrO and TiO2 stacked along [001].The direct epitaxial growth of a SrTiO3 layer on silicon was also reported by preceding it with a thin layer of SrO [Mck98]. Since SrTiO3 is not thermodynamically stable in contact with silicon, it is achieved by growing a seque 3. In this case, the SrTiO3 (001) is rotated by 45o on Si (001) surface, so tha[110]Si//[100]STO. The epitaxial growth of SrTiO3 on TiN/Si was reported in terms of dom a in matching epitaxy as well [Vis96, Wu00]. In contrast, the SrTiO3 film grown on TiN/Si has a cube-on-cube orientation rela t ionship with TiN and Si. Since the SrTa relatively small lattice mismatch (~3.1%) with anatase TiO2, it will act as an effective buffer for the epitaxial growth of anatase TiO2.

PAGE 88

76 In this chapter, we report on the growth of the Co0.04Ti0.96O2(CTO)/SrTiO3 (STO)/TiN/Si heteroepitaxial structure and the origin of ferromagnetism in the Co-dTiO2 layer. Experiment The silicon substrates were cleaned in acetone and methanol sequentially in an ultrasonic bath, followed by a treatment in 10% HF solution in order to provide a hydrogen-terminated silicon surface. HF treatment provides hydrogen termination on silicon surfac oped e atoms, which is known to have the effect of protecting the silicon surface soon as the wafer was cleaned and dried, it was loaded to a d g dation ater crysta from oxidation for some time. As eposition chamber and evacuated immediately to prevent regrowth of the native oxide. The ablation was carried out using a KrF excimer laser (=248 nm) at a fluence of1 J/cm2, and a repetition rate of 5 Hz. TiN film was first deposited at a substrate temperature of 650 C under an evacuated atmosphere (P <810-6 Torr) without flowinnitrogen gas since the film prepared with nitrogen gas flow reduces the kinetic energy of the evaporated target material by increasing the gas phase collision, resulting in relativelypoor crystalline film. The oxygen environment during the subsequent oxide layer growthcan cause transformation of TiN into TiO2 on the surface due to the fact that the oxiof TiN is thermodynamically favorable. Thus, in order to suppress the oxidation of TiN, aSTO film was deposited without flowing oxygen gas for 2 min and then the remaining film was grown in an oxygen pressure of 410-4 Torr at 650 C. This two-step grown STO film showed a much more intense (002) X-ray diffraction peak, indicating gre llinity than the film that was fully grown without oxygen gas flow. Figure 5-9 (a) shows the -2 X-ray diffraction profile of the STO/TiN/Si heterostructures. The in-plane

PAGE 89

77 orientation relationship was examined using -scan measurement. In Fig. 5-9(b), TheSTO buffer layers were found to grow with a cube-on-cube orientation with respectsilicon substrate The substrate temperature was then lowered to 500 C and the Co-doped TiO2 filwas deposited in an oxygen pressure of 310-5 Torr. All depositions were carried out sequentially without breaking the vacuum. X-ray diffraction (Philips X’ Pert MRD) was used to determine the crystallographic relationship between the deposited layersunderlying substrate. Magnetic properties were measured with a Quan to the m and the tum Design ice (SQUID) and magnetic force microscopy (VeecSi are lue d c measurement temperature of 300 K. The Hc value is remarkably high when compared to typical ferromagnetic semiconductors. No distinct blocking behavior, which is a typical signature of a magnetic nanoparticle superconducting quantum int e rference dev o Dim e nsion 3100). The detailed microstructure and the cobalt distribution were investigated by transmission electron microscopy (JEOL 2010F). Results and Discussion Figure 5-10 shows the X-ray diffraction pattern of the CTO/STO/TiN/multilayer structure. All layers show only (00l) diffractions, indicating that the films strongly oriented along the film normal. The crystalline quality of the C T O film was assessed by the rocking curve of a CTO (004) peak. Full width at half maximum vawas approximately 3. Figure 5-11 shows magnetization (M) measurements as a function of magnetic fiel(H) and temperature (T). All measurements were made with an applied field parallel to the film surface. The M-H curve exhibited well-defined hysteresis behavior with a coercive field (H) of approximately 607 Oe a t a

PAGE 90

78 systemnce the blocking temperature easily excesses room temperature in the cobalt nanoparticle system when the diamete goes over around 10 nm due to its high athe res 5-es . as e , was observed in the zero field cooled/field cooled se t of m e asurements. However, one still cannot rule out the possibility of the presence of the magnetic particles in the film si r of the particle nisotropy. To see if there exist cobalt-segregated magnetic particles in the film, surface was investigated by atomic/magnetic force microscopy (AFM/MFM). Figu12(a) and 5-12(b) show the AFM and corresponding MFM images of the CTO film deposited on the STO/TiN buffered silicon. It is clearly seen that the spherical particles are present on the surface with diameters ranging from 30 to 120 nm. In the MFM image, it is hard to see a domain structure in a continuous film. The most magnetic signal comfrom the specific spherical regions. Figures 5-13(a) and 5-13(b) show cross-sectional transmission electron microscopy images of the CTO/STO/TiN/Si(100) heterostructureAs expected from AFM/MFM data, the particles are clearly seen on the film surface. In a magnified image, all interfaces are quite flat and sharp without observable interfacial reactions. It should be noted that the formation of an interfacial layer is significantly suppressed at the interface between STO and TiN and that atomically sharp heteroepitaxy interface is obtained, owing to the two-step growth of the STO layer, as shown in Fig. 5-13(c). When the STO layer was grown in an oxygen atmosphere, it was reported that the interfacial oxinitride layer was formed [Vis96]. The epitaxial orientation relationship winferred from the selected area diffraction pattern taken in the CTO/STO region, as shown in Fig. 5-13(d). The incident electron beam is parallel to the [110] zone axis of thsilicon substrate. The epitaxial relationship between the CTO and the STO thin film is determined to be (001)CTO //(001)STO and [-110]CTO //[-110]STO. The films are identified

PAGE 91

79 as single crystalline from the clearly defined spot patterns. As expected from the AFMdata, the spherelike shape clusters are evidently observed in the CTO layer. The magnified image of the CTO/STO region is shown in Fig. 5-14. It is seen that the clusters nucle, induce the formation of these clusters. Once this occu effectively relieved by the formae e clusters that are assumed to form to relieve the strain [Yan04b]. d n to the ated after a certain growth of film and protruded from the surface. Though the anatase phase is favored on STO, there exists a misfit strain due to the lattice mismatchgiving rise to some defects like dislocations which act as preferential nucleation sites. Even though both SrTiO3 and LaAlO3 provide the template for the anatase phase, due to the difference in lattice mismatch, it is reported that the anatase film on the SrTiO3 has many defects and rough interface between the film and substrate induced by stress, whereas defects are much reduced and the sharp interface is observed in the film on LaAlO3 [Mat02]. Above the certain thickness, the strain is likely to rs, the strain appears to be tion of the clusters [Cha02a] after a critical film thickness has been reached. Thcomposition of clusters was examined by energy dispersive spectroscopy in a TEM. The cobalt mole fractions in CTO measured with an electron beam focus on the cluster and the continuous film were 0.4800.05 and 0.0070.002, respectively. Most of the cobalt atoms diffused into thos Summary Epitaxial growth of a Co-doped anatase TiO2 thin film on silicon has been achieveby using SrTiO3/TiN bilayers as a buffer. X-ray diffraction and transmission electromicroscopy studies clearly demonstrated high-quality epitaxy of films on the silicon. Ferromagnetism was observed at room temperature with a coercivity of around 607 Oe. However, the ferromagnetism arising from the Co-doped TiO2 layer was attributed

PAGE 92

80 presence of the cobalt enriched clusters. The clusters nucleate after a critical thickfilm is reached to relieve the strain. Compositional analysis data revealed that the clusters behaved as a sink for the cobalt dopants. Thus, cobalt atoms mostly diffused into the clusters during the film growth. ness of

PAGE 93

81 c = 9.515 c = 2.959 Anatase TiO2 Rutile TiO2 Silicon Crystal system Tetragonal Tetragonal Cubic (Diamond) Lattice parameter a = 3.785 a = 4.593 a = 5.431 Thermal coefficient of expansion 19 10-6 / oC 8.19 10-6 / oC 8. Figure 5-1. Crystallographic relationship between TiO2 and Si (001).

PAGE 94

82 Figure 5-2. X-ray diffraction pattern of the Ti0.96Co0.04O2 films deposited at (a) HF-treated silicon and (b) just cleaned silicon having native oxide on the surface. Both substrates were loaded at the same time and films were formed under the identical condition. 20406080100 (a)(10)0) R10)R(22400 oCR(200)Si(004)R(211)R(21R(111)R(110)R(400)550 oC Intensity (arb. unit)2 (degree) Si(001) Co:TiO2 20406080100 R(210)(b)R(200)R(110)550 oCR(400)R(220)R(211)R(111)Si(004)R(110)400 oCIntensity (arb. unit)2 (degree) Co:TiO2 a-SiO2 Si(001)

PAGE 95

83 Figure 5-3. AFM data of Ti0.96Co0.04O2 /Si deposited at (a) 400 oC and (b) 550 oC. Note that the scan-size and z-range are different. RMS=0.261 nm RMS=1.745 nm (a) (b)

PAGE 96

84 0.96Co0.04O2 /Si deposited at 400 oC [(a) and (b)] 000010000000001 (a) T=300 K0.08.0x10-6 (b)FCZFCM (emu) -5000-4000-3000-2000-100001000200030004000500000010.00000.0001 (c) T=300 KH (Oe)052.0x10-54.0x10-5 (d)FCZFCM (emu) Figure 5-4. M-H and M-T curves of Ti-0. and at 550 oC [(c) and (d)]. M (emu)0100150200250300 H=200 OeT (K)M Temperature (K) 0501001502002503000.01.0x10-52.0x10-5 3.0x10-5 -5000-4000-3000-2000-1000010002000300040005000-0.0. 0. M (emu)H (kOe)050100150200250300 T (K)MH = 200 OeTemperature (K) 0501001502002500.0 3003.0x10-6 6.0x10-6

PAGE 97

85 Figure 5-5. Stability of binary oxides on silicon. The yellow elements are the experimentally demonstrated stable oxides [Sch02]. Oxides

PAGE 98

86 Figure 5-6. Isothermal section of the Ti-Si-O phase diagram (T=700-1000 oC) [Bey84a].

PAGE 99

87 Figure 5-7.ents are the experimentally demonstrated stable nitrides [Sch02]. Nitrides Stability of binary nitrides on silicon. The yellow elem

PAGE 100

88 TiN Si Crystal system Cubic (NaCl) Cubic (Diamond) Lattice parameter a = 4.242 a = 5.431 Thermal coefficient of expansion 9.0 10-6 / oC 4.0 10-6 / oC Figure 5-8. Crystallographic information of TiN and Si.

PAGE 101

89 Figure 5-9. The results of X-ray diffraction. (a) -2 X-ray diffraction pattern of STO/TiN/Si heteroepitaxial structures. (b) X-ray -scan showing cube-on-cube orientation relationship between STO and Si substrate. 20406080100 TiN(400)TiN(200)STO(200)STO(100)Intensity (arb. unit)2 (degree) STO (111) Intensity (arb. unit) (b) (a)Si(004)060120180240300360 Si (111) (degree)

PAGE 102

90 Figure 5-10. -2 X-ray diffraction pattern of the Co0.04Ti0.96O2-/SrT3ilm structure. All films are grown with a c-axis normal to the surface of substrate. 20406080100 STO(002)Si(004) STO(003)TiN(004)CTO(008)STO(001)TiN(002)CTO(004) Intensity (arb. unit)2 (degree) iO/TiN/Si f

PAGE 103

91 igure 5-11. The resnd (b) M-T curve of the Co0.04Ti0.96O2-/SrTiO3/TiN/Si. ults of SQUID measurement. (a) M-H a H (Oe)050100150200250-6-5 (b)T (K)MFCZFCH = 200 OeTemperature (K) 0501001502002500.05.0x10-61.0x10-5 -500005000-20-1001020 (a) 10 K 300 KM (emu*10) -63000.05.0x101.0x10 300 M (emu) F

PAGE 104

92 e (a). Figure 5-12. Atomic force and its corresponding magnetic force images. (a) AFM imagof Co0.04Ti0.96O2-/SrTiO3/TiN/Si film structure. It is clearly seen that particles are randomly distributed over the film. (b) Corresponding MFM image of (a) (b) The lateral dimensions are 5 m5 m.

PAGE 105

93 Figure 5-13. The results of TEM analysis. (a) Cross-sectional transmission elecmicroscopy image of Co0.04Ti0.96O2-/SrTiO3/TiN/Si heteroepitaxial stru(b) Magnified image of (a). (c) High resolution image of SrTiO3region. Negligible interfacial layers are formed at the SrTiO3/TiN interf(d) Selected area electron diffraction pattern taken from a CTO/STO region with the electron beam parallel to the [110] Si. The anatase phase is epitaxially grown on SrTiO3/TiN buffered Si substrate. ‘A’ and ‘S’ denote SrTiO3 TiN A(004) S(001) S(110) TiN SrTiO3 Co:TiO2 100 nm (b) 5 nm (d) A(112) (a) Si (c) Si tron cture. /TiN interface ace. cobalt-doped anatase and SrTiO3, respectively.

PAGE 106

94 Figure 5-14. Enlarged TEM image of the Co:TiO2/SrTiO3 region. It is apparent that the clusters nucleate during the film growth and most of the cobalt diffused into them. On the Co:TiO2 film On the cluster

PAGE 107

CHAPTER 6 EPITAXIAL GROWTH OF COBALT-DOPED RUTILE TiO2 ON SILICON rtronic on above roomsemanatasexide-atching temical form of TiO easier to oborts epitax, Al2hould be used for the stabilizatioration of a robusviewpocommeon technolwill be more resistant to the severe thermal asperity during the device fabrication in comparison with the anatase phase. In addition, the anomalous Hall effect has been recently reported only in rutile Overview the development of the realistic spinductors retaining feearch on Co:TiO2 ferromag phase of TiO2 that is grown o3 [Cha02a, Cha03, Kim02a]. Lattice m One of the essential prerequisites fodevices is to develop ferromagnetic semic temperature. To date, most of the resiconductor has been focused on the based substrates such as SrTiO3 and LaAlO rromagnetism netic n o substrates have been employed to stabilize thplate effect [Che93]. On the other hand, 2, is much more stable and ison the epitaxial rutile phase of TiO2ial growth of rutile phase of TiO2fact that specific substrate s e less stable anatase phase by utilizing a rutile phase, which is another chemtain than anatase. There are a few rep doped with cobalt [Mat01b, Shi04, Toy04a]. For the O3 substrate has been used [Mur04b]. The n of anatase or rutile phase substantially limits the fabrication of TiO2-based devices. In this regard, the integt ferromagnetic semiconductor with silicon is of great importance from the int of multi-functionality on a chip as well as an expectation of prompt rcialization of spintronic devices on the basis of well-established silicogy. Moreover, the thermally stable rutile phase 95

PAGE 108

96 phase [Shi04, Toy04a, Toy04b] which is a useful property for the spintronic device applications. The main obstacle to the epitaxial growth of TiO2 on silicon is amorphous silicon oxide layer at the interface which hinders the crysand silicon. Even if the silicon substrate isoxygen environment during the oxide film growthotes the formation of such an interfacial layer. The presencsults in the growth of polycrystalline or mixed phases of TiO2is the reaction of TiO2 with silicon to form titanium silicide and silicon d to be kinetically delayed in typical experimental conditions in terms of reliability. Thus, a use of a buffeof Co-doped rutile TiO2 on silicon. The selection of TiN otivated by the following reasons: 1) TiN can be grown easily oxidized into TiO2, which is the same host m91, Wit81]. TiN has been used for a diffusion barrindeed one of the few nitrides that are stableEven though there exists a large lattice mismatch (~ 22%) between TiN and silicon (001), epitaxi] with a cube-on-cube orientation relationship between them. In this chapter, we report on the structural property of the Co-doped rutile TiO2 film and the behavior of the cobalt atoms during the deposition, which plays a critical role in the magnetic property. Experiment The silicon (001) substrates were cleaned in acetone and methanol sequentially, followed by a removal of native oxide in 10 % HF solution just prior to transfer to a tallographic matching between the film prepared without a native oxide layer, the at elevated temperatures prome of this layer typically re. Another possibility ioxide [Bey84a]. This appearsbut can be a potential problem r layer was proposed for the epitaxial growth as a buffer layer was mepitaxially on silicon [Nar92], 2) TiN is aterial for Co:TiO2 layer [Tomier in microelectronics industries and is in direct contact with silicon [Hub96, Sch02]. al growth can be obtained via the domain matching epitaxy mechanism [Nar03

PAGE 109

97 chamber. The ablation was carried out using a KrF excimer laser ( = 248 nm) at a fluence of 1 J/cm2, and a repetition rate of 5-10 Hz. The TiN film was deposited using a hot-pressed TiN target at 650 oC under an evacuated atmosphere (P < 8 10-6 Torr) witho0.960.042 an oxygen pressure of 4 10-5 Torr at substrate temperatures ranging from 500 to 725 oC. The temperature was monitored by a thermocouple and an infrared pyrometer. All depositions (TiN and Co:TiO2) were carried out in a chamber sequentially without breaking vacuum. The crystalline structure was examined by X-ray diffraction (Philips APD 3720 and X’ Pert MRD) with Cu K radiation. X-ray scan measurements were performed to determine the in-plane orientation relationship between the film and substrate. Magnetic properties were measured with a Quantum Design superconducting quantum interference device. The detailed microstructure and the cobalt distribution were investigated by transmission electron microscopy (JEOL 2010F) with an incident electron energy of 200 keV. Results and Discussion Figure 6-1(a) shows the -2 X-ray diffraction (XRD) pattern of the Ti0.96Co0.04O2/TiN/Si heterostructures, where the Co:TiO2 film was grown at 725 oC. Similar diffraction profiles were obtained in the samples deposited at 600 oC and 550 oC. The XRD result indicates that the Co:TiO2 film is a single rutile phase with a well-defined {110} family of planes parallel to the silicon substrate. No anatase and secondary phases like a cobalt oxide were detected. The in-plane orientation of the Co:TiO2 film was assessed by scan of the (211) plane that is inclined at ~ 39 from the (110) plane. The scan exhibits eight peaks, which are separated by 45, instead of four peaks, as ut flowing nitrogen gas. The TiCoOfilmswere then dire c tly deposited in

PAGE 110

98 shown in Fig. 6-1(b). This suggests that Co:TiO2 film is composed of two equivalentstructural dom ains which are perpendicular to each other [Lee95]. Schematic illustration is shoe iO2 film is directly deposited on TiN/Si atratures, a majority of cobalt atoms diffus wn in Fig. 6-2. The two domains appear to be favored to reduce the lattice misfit. The in-plane orientations are determined to be TiO2[001]//TiN[110] and TiO2[001]//TiN[-110]. Magnetization was measured as a function of a magnetic field applied parallel to the film. Since room temperature ferromagnetism was also found in the rutile TiO2 phas[Mat01b], the ferromagnetic property was also expected from the Co-doped TiO2 rutile phase. However, as shown in Fig. 6-3, no ferromagnetic hysteresis behavior was observed down to a measurement temperature of 10 K in all samples. The detailed microstructure and the reason for the paramagnetism were revealed by cross-sectional TEM an alysis. In Fig. 6-4, it is clearly seen that some foreign phase was formed in the silicon substrate having pyramid, hut and spike morphology. Energy dispersive spectroscopy (EDS) analysis demonstrated that this secondary phase was composed of cobalt and silicon, as shown in Fig. 6-5. Quantitative calculation suggeststhat it is close to CoSi2. Some physical properties of cobalt silicides are listed in Table 6-1. It is also noted that a negligible amount of cobalt was detected in the Co:TiO2 film. Figures 6-6(a) and (b) show the high resolution image of Co:TiO2 film and its selected area electron diffraction pattern with an e-beam focus on the Co:TiO2 layer. The image was taken with the e-beam aligned along the [001] direction with respect to the rutile. The clear spots indicate the epitaxial nature of the film. As a result, while the Co:T elevated tempe e into silicon through the TiN layer and react with silicon forming a silicide

PAGE 111

99 [Yan04c]. The formation of silicide is preferred along the {111} atomic planes of Sishown in Figs. 6-6 (c) and (d), because of its lowest surface energy. When the foreign phase is formed, the shape is generally determined by the surface free energy , as ution cewest t ffer rather than the diffusion barrier property. The better crystallinity of TiN was obtained without flowing nitrogen gas [Cho96, Xu98] due to higher kinetic energhe film is film o. thermally stable rutile phase providing a same host material for the overlying Co:TiO contribto the total energy. One way to reduce surfa energy is to minimize the surface area. This frequently results in a sphere shape since it has the lowest surface to volume ratio among any three dimensional object. The other way is the crystallite exposes the losurface energy, as in this case. Consequently, the cobalt silicides has pyramid shapewhich is composed of equivalent {111} planes and spikes along {111} atomic plane. It should also be noted that voids were formed on top of the TiN layer. To make the TiO2thin film semiconducting (n-type), it is typically synthesized under oxygen-deficienenvironment. The vacant sites caused by rapid cobalt diffusion are likely to assemble as void defects and remain along the top interface of TiN. Though TiN acts generally as a diffusion barrier, the growth condition of TiN in our experiment is tailored for the effective epitaxial bu y of impinging atoms caused by less collision in the gas phase. Thus, tpossibly sub-stoichiometric [Oba99] and its ability to function as a diffusion barrier might be poor [Jos92]. The behavior of cobalt atoms was further elucidated in a somewhat modified structure, i.e., Ti0.96Co0.04O2/TiO2(oxidized TiN)/Si. TiN film deposited on Si under identical conditions was in situ oxidized in an oxygen pressure of 100 mTorr at 700 CThe oxidation of TiN is thermodynamically favorable [Wit81] and it transforms into 2

PAGE 112

100 film. In the entire temperature range considered (25-1000 oC), the TiO2 oxide phase isalways more stable than the TiN nitride phase. Gibb’s free energy change for the oxidation of TiN is represented by the following equation [Lee94]. TiN + O2 TiO2 + N 2 G = 604.3 + 81.510-3 T (kJ/mol) After oxidation, the (100) oriented gold colored TiN on Si transforms into (110) oriented transparent rutile TiO2, accompanying volume expansion. The thickness increased almost twice and root mean square roughness value increased from 0.22 nm (TiN/Si) to 2.9 nm (TiO2/Si). In Fig. 6-7, X-ray diffraction profiles are shown before and after the oxidation of the TiN film which is epitaxially grown on Si. After oxidation, TiN transformed into rutile TiO2 phase with (110) preferred orientation, as expected. The oxidation temperature was relatively high (650-750 C), thus leading to rutile phase which is the most thermally stable phase among the three polymorphs of TiO2. The resultant TiO2 films might serve as a template layer for the growth of the Co-doped TiO2 layer. As a final step, the Co:TiO2 layer was deposited in an oxygen pressure of 410 Torr at 450 oC. This temperature was chosen to be low enough to limit the undesirable diffusion of cobalt but high enough to still provide appropriate thermal energy for the crystallization. The XRD result was quite similar to Fig. 6-1(a) except for the extinction of TiN peaks due to oxidation. However, room temperature ferromagnetism was observed in this film structure with a coercive field (Hc) of about 285 Oe, as shown in Fig. 6-8. No distinct blocking temperature was found in M-T plot and Curie temperature was estimated over 350 K. The origin of the ferromagnetism was unveiled by the investigation of a o-5

PAGE 113

101 distribution of cobalt. In Fig. 6-9, the EDS mapping clearly shows some segregation of cobalt mainly found in oxidized TiN region. Further evidence was provided by the highangle annular dark field (HAADF) method. Figure 6-10(a) shows HAADF image of the Ti0.96Co0.04O2/TiO2(oxidized TiN)/Si heterostructures. HAADF technique is similar to backscattered electron imaging on the scanning electron microscopy. However, in HAADF, the incident electrons interacted with atomic nuclei (therefore sensitive to atomic number) are scattered in forward direction, in contrast to backscattered electrowhich mainly occurs at tens of eV electron beam energy range [Ott91]. HAADF acceptance angles are generally 2.5 to 6 degrees. Schematic imaging mechanism is depicted in Fig. 6-11. The brightness in HAADF is roughly proportional to Z2 (Z: atonumber) leading to compositional contrast. Hence, the brighter contrast area in the imagerepresents the higher Z element, in this case, cobalt atom position (Z(Ti)=22, Z(Co)=27,Z(O)=8). It is apparent that significant amount of cobalt atoms diffused into the TiO(oxidized TiN) and some cobalt-enriched TiO2 clusters are present in the Co:TiO2 region as well. The top Co:TiO2 layer acts a cobalt source into the TiO2 layer, and cobalt prefersto segregate rather than substitute Ti lattice sites when it diffuses into TiO2. This is presumably due to the larger affinity of titanium toward oxygen. Breaking the strongO bond and replacing th ns mic 2 Ti-e Ti site may not be probable. Likewise, when a Co:TiO2 film is formetals all d, Ti will preferentially react with O compared to Co. The oxygen-deficient condition also seems to help the formation of cobalt clusters. It is also known that medeposited on metal oxides typically grow as three-dimensional clusters because adatom-adatom interactions are stronger than adatom-substrate interactions [Cha02c]. Almost

PAGE 114

102 metals deposited on oxide films tend to form three-dimensional (3D) nonlaminar island[Bu99]. This probably can be another cause. Figures 6-10(b) and (c) show an EDS line scan of Co K along s the film depth. This is in a good agreement with the HAADF result. The humps in the Co:TiO2 film region correspond to the cobalt-enriched clusters. As the third experiment, very thin TiN layer is introduced not only to decrease the TiO (oxidized TiN) thickness but also to get highly oriented rutile structure. Two or three monolayer of TiN was deposited and then Co:TiO layer was directly deposited on it. By inserting thin TiN layer, the crystallographic orientation drastically changes from polycrystalline to highly (110) oriented film. Figures 6-12(a) shows M-H curves measured at 10 K and 300 K. In this case, ferromagnetism was examined at low measurement temperature but the coercive field becomes almost zero at 300 K, suggesting superparamagnetic state of the film. In Fig. 6-12(b), the solid symbols refer to the ZFC procedure in which the sample is cooled in zero applied field and the indicated field is then applied before making the measurement, while the open symbols refer to the FC procedure in which the sample is cooled in the presence of the indicated field. All measurements were taken on heating. The hump in zero field cooled curve shifts slightly to the low temperature with increasing applied field, indicating the presence of magnetic particles. This was verified by cross-sectional TEM analysis. Figure 6-13 shows formation of both cobalt silicide and cobalt enriched clusters. These clusters are mainly located near the interface between the film and substrate. The size of particles is in the range of 7-11 nm. Assuming the sphere shape of clusters, this is in good agreement with blocking temperature calculated on the stability criteria of magnetic particles [Mur01a]. 22

PAGE 115

103 This phenomenon was prominent in the Co:TiO2 film deposited at 650 oC with 7%Co-doped TiO2 target. One can clearly see that the spikes extended deep into the silicon substrate in Fig. 6-14. The cluster size was relatively large (~15 nm) due to high cobalt concentration. HAADF image apparently shows the cobalt enriched clusters which are located at the interface. As shown in Fig. 6-15, negligible cobalt was found in the continuous film. Instead, they formed clusters at the interface and silicides into the sias evidenced by EDS. Summary In summary, the epitaxial Co-doped rutile TiO2 thin film has been successfullygrown on silicon using a TiN buffer layer. X-ray diffraction and transmission electron microscopy clearly demonstrate high-quality epitaxy of the film on the silicon substrate. We found that cobalt atoms are highly mobile and diffusive at a temperature of as low as 450 oC, resulting in a nonuniform distribution. They exhibit strong tendency toward licon, omic silicide formation and readily segregate into clusters rather than substitute Ti lattice sites. Room temperature ferromagnetism arising from the Co-doped rutile TiO2 film is attributed to the presence of the cobalt-enriched clusters. Careful control o f cobalt atmovement, defect control and oxygen atmosphere will be crucial in order to effectively incorporate cobalt into the TiO2 matrix.

PAGE 116

104 he Ti0.96Co0.04O2/TiN/Si heterostructures. “R” denotes a rutile phase of TiO2 film. (a)R(110) Figure 6-1. The results of X-ray diffraction. (a) -2 X-ray diffraction pattern of tdoped with cobalt. (b) X-ray scan for the {211} reflection of the Co:TiO2 20406080100 TiN(004)R(330)R(220)060120180240300360 2Intensity (arb. unit) (degree) TiN(002)Si(004)Intensity (arb. unit)2 (degree) (b)TiO:Co (211)

PAGE 117

105 Figure 6-2. Crystallographic relationship between rutile TiO2 and TiN.

PAGE 118

106 F igure 6-3. M-5000-1 agnetization as a function of a magnetic field measured at 10 K, showing paramagnetic behavior. 5000-2500025000-5 T=10 KH (Oe) 0510 M (emu*10-6)

PAGE 119

107 igure 6-4. Crctures. Cobalt pyramids. F oss-sectional TEM images of the Ti0.96Co0.04O2/TiN/Si strusilicide is formed in the silicon region with various types of spikes, huts, and (c) 100 nm Si TiN Co:TiO2 50 nm (a) (d) (b) Si Si Si 100 nm 100 nm

PAGE 120

108 Figure 6-5. Elemental mapping of the spike region. (a) STEM image of the Ti0.96Co0.04O2/TiN/Si. (b) EDS spot analysis on the Ti0.96Co0.04O2. Cobalt was not detected within the resolution limit, indicating downward diffusion tendency of cobalt into silicon. (c) EDS mapping of the region including spikes. The spikes were identified as cobalt silicide. (b) Co K position (a) Co Si Ti (c)

PAGE 121

109 Table. 6-1. Properties of the cobalt silicides. Silicide Phase Crystal Structure Lattice Constant Tf (oC) (cm) DDSa) ic a=4.918 , b=3.738 .109 110 Co CoSi Cubic a=4.444 375 147 Si CoSi2 Cubic a=5.365 500 15-20 Co a) DDS-dominant diffusing species during formation Co2Si Orthorhomb , 350 c=7

PAGE 122

110 (a) 5 nm (b)(010) (110)(100) Si TiN 5 nm {111} 54.7o {001} (c) (d) TiN Si 54.7o 5 nm {001} {111} age of (d) ne. Figure 6-6. Additional results of TEM analysis. (a) High resolution TEM imTi0.96Co0.04O2. (b) Selected area electron diffraction pattern taken on the Ti0.96Co0.04O2 film. The electron beam is parallel to the [110] Si. (c) and The cobalt silicide pyramid and spike which form along {111} atomic pla

PAGE 123

111 igure 6-7. -2 X-ray diffraction pattern of (a) TiN/Si and (b) Ti0.96Co0.04O2 film deposited on in situ oxidized TiN/Si. 0 204 60 80 100 TiN 004) 2 (deg b) (a)Si (004) (TiN (002)Intensity (arb. unit)ree)20406080100 ((330)(220)Si (004)Intensity (arb. unit)2 (degree) TiO2:Co (110) F

PAGE 124

112 -5-4-3-2-1012345-80-40040 (a) at 10 K at 300 KM (emu*10-6)H (kOe)0100200300010 H=200 Oe M (e*10-6)T (K)05010002002503003500.00.20.40.81.01.2 (b)Normed Remanence T Figure 6-8.M-80 he res ults of SQUID measurements. (a) Magnetization-magnetic field (H) hysteresis loop measured at 10 K and 300 K. (b) Curie temperature is estimated over 350 K, which is the limit of SQUID used. 5 mu 0.6 15 aliz Temperature (K)

PAGE 125

113 (a) F igure 6-9. Eure). lemental mapping of the film. (a) STEM image of the (b) Ti0.96Co0.04O2/TiO2(TiN)/Si. (b) EDS mapping of the sqare gion of (aCobalt segregations were clearly seen.

PAGE 126

114 Silicon 2 (b) (a) TiO2 Co:TiO Ti K Co K Si TiO2 Co:TiO2(c) t un Co O K Si K Figure 6-10. HAADF and line scan along the film depth. (a) High angle annular dark field image (HAADF) of the Ti0.96Co0.04O2/TiO2/Si heterostructures. (b) and (c) EDS line scan of Co K along the film depth. The small humps in the Ti0.96Co0.04O2 film region correspond to the cobalt-enriched clusters.

PAGE 127

115 Figure 6-11. Imaging mechanism of Z-contrast transmission electron microscopy.

PAGE 128

116 Figure 6-12. The results of SQUID measurement. (a) M-H curves of CTO/thin TiN/Si -4-1012345-20 10 K 300 KH (Oe)0501001502002503000.0000000.0000050.0000100.0000150.000020 (b)H=1000 OeH=500 OeH=50 OeH=100 OeH=200 Oe FC ZFCM (emu)Temp (K) -5-30 -100102030 (a) M (emu*10-6) -3-2 measured at 10 K and 300 K. (b) M-T curves of CTO/thin TiN/Si measured under the different magnetic fields.

PAGE 129

117 (a) 10 nm Si (d) (c) (b) 50 nm 5 nm 5 nm Si SiSi Co:TiO2 F igure 6-13. Cross sectional TEM images of the CTO/thin TiN/Si. (a) Low magnification. (b) High magnification. (c) Cobalt silicide is also seen. (d) cobalt enriched clusters (circled) are found mainly near the interface.

PAGE 130

118 Figure 6-140 C. (a) Low magnification. (b) HAADF image. It is apparent that cobalt enriched clusters are located at the interface. (c) Magnified image at the interface between the film and substrate. Both the cobalt enriched cluster and cobalt silicide spikes are shown. (d) High resolution image of a cluster. (b) (c) (d) 10 nm 2 nm (a) . Cross sectional TEM images of the Co0.07Ti0.93O2/thin TiN/Si. The deposition temperature of CTO is 65o

PAGE 131

119 silicon. Figure 6-15. EDS spot analysis. Negligible cob are detected in the continufilm region. Instead, they fd cluthe interface and silicides into the alt atomssters at ous orme

PAGE 132

CHAPTER 7 CONCLUSION In this study, the experiment and characterization were designed with a view to understand and unveil the critical issues in the area of oxide based ferromagnetic semiconductors, and to gain insight into the process conditions for intrinsic ferromagnetic semiconductor which is the key material in spintronics. Whereas most diluted magnetic semiconductors show very low magnetic transition temperatures, room temperature ferromagnetism has been reported in both anatase and rutile phases of TiO2 thin films doped with cobalt. As high crystalline quality film is required for the effective spin transfer, lattice-matching oxide based substrates have been employed for the epitaxial growth of TiO2 phase through template effect. The optimal growth conditions of Co:TiO2 were obtained using LaAlO3 and the effect of in situ ultraviolet irradiation during the Co:TiO2 film growth was investigated at the same timeEnhanced crystalline and magnetic properties were observed at relatively low process temperature in UVPLD grown samples, possibly due to the energetic photons and reactive oxygen species present during the deposition. Much stronger binding . coordination between oxygen and metal cations (Co and Ti) was inferred in the UVPLD grown filmTo dLaAlO3 ananatase or rutile phase, substantially restricting the fabrication of TiO2-based devices. The integration of a ferromagnetic semiconductor with silicon is of great importance from the viewpoint of multi-functionality on a chip as s from the result of X-ray photoelectron spectroscopy analysis. ate, most of research has been doing in limited substrates such as SrTiO3, d Al2O3 for the stabilization of 120

PAGE 133

121 well as an expectation of prompt commercialization of spintronic devices on the basis of w ell-established silicon microelectronics technology. Although the epitaxial growth of iO2 was anticipated on silicon, polycrystalline rutile film was obtained due to the evitablich hinders the rystallolly stable i-baseduffer (Cudies cwever, room mpresenceicknesis data vealedn TiN/Shase wi an measurement reveals that Co:TiO2 film is composed of two equivalent structural ferromagnetic hysteresis behavior even at a measurement temperature of 10 K. The ason for the paramagnetism was unveiled by cross-sectional TEM analysis. While the o:TiO2 film is directly deposited on TiN/Si at elevated temperatures, a majority of obalt atoms diffuse into silicon through the TiN layer and react with silicon forming a silicide. The formation of silicide is preferred along the {111} atomic planes of Si, and T in e formation of amorphous silicon oxide layer at the interface, whgraphic matching between the film and the substrate. Nonoxide therma material, TiN, was introduced for the epitaxial growth of Co:TiO2. For the epitaxial growth of anatase Co:TiO2, SrTiO3/TiN bilayer was used as a TO/STO/TiN/Si). X-ray diffraction and transmission electron microscopy learly demonstrated epitaxial nature of the films on the silicon. Hoerature ferromagnetism arising from the Co-doped TiO2 layer was attributed to the of the highly cobalt enriched TiO2 clusters. The clusters nucleate after a critical ss of film has been reached so as to relieve the strain. Compositional analy that the clusters behaved as a sink for the cobalt dopants. For the epitaxial growth of rutile Co:TiO2, the Co:TiO2 film was directly deposited i (CTO/TiN/Si). XRD results indicate that the Co:TiO2 film is a single rutile th a well-defined {110} family of planes parallel to the silicon substrate. X-ray c T b st te p th re o p sc domains which are perpendicular to each other. Interestingly, this film showed no re C c

PAGE 134

122 negligible cobalt is detected in the intended Co:TiO2 film region within the resolution limit. The behavior of cobalt atoms was further elucidated in somewhat modified film structures. TiN film, which was epitailicon substrate, was in situ oxidiz22 2o:TiO2 the TiO2 layer is very thin, cobalhe 2 2sample O2 material; eitherence 3, xially grown on s ed into rutile TiO and then the deposition of Co:TiO film was followed. In thiscase, no silicide was formed but cobalt atoms diffused into TiO layer during the Cfilm deposition, forming cobalt enriched magnetic clusters. These are attributed to the source of room temperature ferromagnetism. In case that t silicide as well as cobalt enriched clusters were found since the diffusion length of cobalt was long enough to react with silicon. Because a partial amount of cobalt introduced was consumed by silicide formation, the diameter of magnetic clusters in tCo:TiOfilm region decreased down to 7-11 nm showing superparamagnetic behavior. This explains the almost zero coercive fie l d at 300 K. Cobalt atoms were found to be highly mobile and diffusive at a temperature of as low as 450 oC, resulting in a nonuniform distribution; They exhibited strong tendency toward silicide formation and readily segregate into clusters rather than substitute Ti lattice sites when diffuse into TiO. The mechanisms for the observed magnetic behavior in several different structures are quite complex and appear to depend on a number of factors. As debated, there are mainly two possible causes for the ferromagnetism in Co-doped Ti intrinsic DMS phenomenon (carrier mediated exchange interaction) or the p r esof the cobalt-enriched clusters (Co-Ti-O compound or metallic Co). Among the film structures studied here, the intrinsic film was only achieved in the film grown on LaAlO

PAGE 135

123 which has very small lattice mismatch (-0.2 %) with anatase TiO2. The defects induced during the film growth are likely to promote the cluster formation by providing the facilnucleation site. Furthermore, it is known that the interaction of cobalt with oxide is weak that it would normally ball up on oxides. Since adatom-adatom (Co-Co) interactions are relatively stronger than adatom-substrate interactions, it has a high possibility to grow as three-dimensional clusters. e so d r is applicable. n, Even though the Co:TiO2 film is grown on the same substrate, the origin for the magnetism can be different depending on the growth conditions (oxygen pressure antemperature) and the growth technique (sputtering, pulsed laser deposition, and moleculabeam epitaxy etc.) employed. As a result, the microstructure of the film can have single phase alloy or multiphase film containing magnetic inclusions. Therefore, it is necessary to carefully analyze and characterize the film to relate observed magnetic properties with the film structure. Not until then can we conclude which mechanism Finally, careful control of cobalt atomic movement, defects and process conditions will be needed in order to effectively incorporate cobalt into the TiO2 matrix. In additiocircumventing the nonuniform formation of the cobalt silicide will be crucial for the development of a cobalt-based DMS spin injector into silicon.

PAGE 136

d Bal98 A. R. Bally, E. N. Korobeinikova, P. E. Schmid, F. Levy, and F. Bussy, J. . (1984). Buc91 J. P. Bucher, D. C. Douglass, and L. A. Bloomfield, Phys. Rev. Lett. 66, er and U. Diebold, Appl. Phys. Lett. 79, 3467 (2001). Cha01b S. A. Chambers, S. M. Heald, R. F. C. Farrow, J.–U. Thiele, R. F. Marks, McCready, A. S. Lea, V. Shutthanandan, and C. F. Windisch, Thin Solid nce 297, 827 (2002). Cha03 S. A. Chambers, T. Droubay, C. M. Wang, A. S. Lea, R. F. C. Farrow, L. Folks, V. Deline, and S. Anders, Appl. Phys. Lett. 82, 1257 (2003). LIST OF REFERENCES And01 K. Ando, H. Saito, Z. Jin, T. Fukumura, M. Kawasaki, Y. Matsumoto, anH. Koinuma, Appl. Phys. Lett. 78, 2700 (2001). Bai88 M. Baibich, J. Broto, A. Fert, F. Nguyen Van Dau, and F. Petroff, Phys. Rev. Lett. 61, 2472 (1988). Phys. D 31, 1149 (1998). Bu99 M. Bumer and H.-J. Freund, Prog. Surf. Sci. 61, 127 (1999). Bec99 H. Beckmann and G. Bergmann, Phys. Rev. Lett. 83, 2417 (1999)Bey84a R. Beyers, J. Appl. Phys. 56, 147 (1984). Bey84b R. Beyers, R. Sinclair, and M. E. Thomas, J. Vac. Sci. Technol. B 2, 781 3052 (1991). Cha00 S. A. Chambers, Surf. Sci. Rep. 39, 105 (2000). Cha01a S. A. Chambers, S. Thevuthasan, R. F. C. Farrow, R. F. Marks, J. U. Thiele, L. Folks, M. G.. Samant, A. J. kellock, N. Ruzycki, D. L. Eder M. F. Toney and A. Chattopadhyay, cond-mat/0208315. Cha02a S. A. Chambers, C. M. Wang, S. Thevuthasan, T. Droubay, D. E. Films 418, 197 (2002). Cha02b S. A. Chambers, Materials Today, April issue, 34 (2002). Cha02c S. A. Chambers, T. Droubay, D. R. Jennison, and T. R. Mattsson, Scie 124

PAGE 137

125 Che93 S. Chen, M. G. Mason, H. J. Gysling, G. R. Paz-P u jalt, T. N. Blanton, T. Cohen, and J. F. Evans, J. Vac. Sci. Technol. A 11, 2419 (1993). Chi03 D. Chiba, K. Takamura, F. Matsukura, and H. Ohno, Appl. Phys. Lett. 821185 (1989). Cho96 R. Chowdhury, R. D. Vispute, K. Ja Castro, K. M. Chen, C. P. Fictorie, W. L. Gladfelter, A. Franciosi, P. I. , 3020 (2003). Cho89 T. Choudhury, S. O. Saied, J. L. Sullivan, and A. M. Abbot, J. Phys. D. 2, gannadham, and J. Narayan, J. Mater. Res. 11, 1458 (1996). Cra99). Electrochem. Soc. 147, 1077 (2000). Cul72 B. D. Cullity, “Introduction to magnetic materials”, Addison-Wesley Pub. Die03, Dij87 D. Dijkkamp, T. Venkatesan, X. D. Wu, S. A. Shaheen, N. Jisrawi, Y. H. Min-Lee, W. L. Mclean, and M. Croft, Appl. Phys. Lett. 51, 619 (1986). Fer01D. Ferrand, J. Cibert, A. Wasiela, C. Bourgognon, S. Tatarenko, G. Fishman, T. Andrearczyk, J. Jaroszynski, S. Kolesnik, and T. Dietl, Phys. Rev. B 63, 085201 (2001). Fie99 R. Fiederling, M. Keim, G. Reuscher, W. Ossau, G. Schmidt, A. Waag and L. W. Molenkamp, Nature 402, 787 (1999). For94 L. Forro, O. Chauvet, D. Emin, L. Zuppiroli, H. Berger, and F. Levy, J. Appl. Phys. 75, 633 (1994). V. Craciun and R. K. Singh, Electrochem. Solid-State Lett. 2, 446 (1999Cra00a V. Craciun and R. K. Singh, Appl. Phys. Lett. 76, 1932 (2000). Cra00b V. Craciun, R. K. Singh, J. Perriere, J. Spear, and D. Craciun, J. Co. (1972). Dat90 S. Datta and B. Das, Appl. Phys. Lett. 56, 665 (1990). Def83 J. W. DeFord and O. W. Johnson, J. Apply. Phys. 54, 889 (1983). T. Dietl, Nature Materials 2, 646 (2003). Die97 T. Dietl, A. Haury, and Y. Merle d’Aubigne, Phys. Rev. B. 55, R3347 (1997). Die00 T. Dietl, H. Ohno, F. Matsukura, J. Cibert and D. Ferrand, Science 2871019 (2000).

PAGE 138

126 Fra03 R. M. Frazier, J. Stapleton, G. T. Thaler, C. R. Abernathy, S. J. Pearton, R. Rairigh, J. Kelly, A. F. Hebard, M. L. Nakarmi, K. B. Nam, J. Y. Lin, H. X. Jiang, J. M. Zavada, and R. G. Wilson, J. Appl. Phys. 94, 1592 (2003). Fuk01 T. Fukumura, Zhengwu Jin, M. Kawasaki, T. Shono, T. Hasegawa, S. Koshihara, and H. Koinuma, Appl. Phys. Lett. 78, 958 (2001). Fuk04 T. Fukumura, Y. Yamada, H. Toyosaki, T. Hasegawa, H. Koinuma, and M. Fuk99 T. Fukumura, Zhengwu Jin, A. Ohtomo, H. Koinuma, and M. Kawasaki, Gar00 J. Garcia-Otero, M. Porto, J. Rivas, and A. Bunde, Phy. Rev. Lett. 84, 167 Gor02 O. Y. Gorbenko, S. V. Samoilenkov, I. E. Graboy, and A. R. Kaul, Chem. Lett. 83, 203 (1999). Han95 M. Hanson, C. Johansson, M. S. Pedersen, and S. Morup, J. Phys. 7, 9269 Har99 U. Hartmann, Annu. Rev. Mater. Sci. 29, 53 (1999). Heb04 R. P. Rairigh, J. G. Kelly, S. J. Pearton, C. R. Abernathy, S. N. G. Chu, and R. G. Wilson, J. Phys. D 37, 511 (2004). Hel00 G. A. Held, G. Grinstein, H. Doyle, S. Sun, and C. B. Murray, Phys. Rev. Hig04 Jsan, and R. L. Greene, Phys. Rev. B 69, 073201 (2004). Hon04 , 2602 (2004). Jos92 s 221, 228 (1992). , L1205 (1999). Kai02 uller, J. L. Grazul, A. Chuvilin, and M. Kawasaki, Nature materials, 1, 102 (2002). Kawasaki, Appl. Surf. Sci. 223, 62 (2004). Appl. Phys. Lett. 75, 3366 (1999). (2000). Mater. 14, 4026 (2002). Ham99 P. R. Hammar, B. R. Bennet, M. J. Yang and M. Johson, Phys. Rev. (1995). A. F. Hebard, B 64, 012408 (2000). . S. Higghins, S. R. Shinde, S. B. Ogale, T. Venkate N. H. Hong, J. Sakai, and A. Hassini, Appl. Phys. Lett. 84 Hub96 K. J. Hubbard and D. G. Schlom, J. Mater. Res. 11, 2757 (1996). H. Joswig and W. Pamler, Thin Solid Flim Jos99 M. Joseph, H. Tabata, and T. Kawai, Jpn. J. Apply. Phys. 38 U. Kaiser, D. A. M

PAGE 139

127 Kal03 S. N. Kale, S. B. Ogale, S. R. Shinde, M. Sahasrabuddhe, V. N. Kulkarni, R. L. Greene, and T. Venkatesan, Appl. Phys. Lett. 82, 2100 (2003). Kas56 T. Kasuya, Prog. Theor. Phys. 16, 45 (1956). Kik99 and D. D. Awschalom, Nature (London) 397, 139 (1999). anaka, H. Y. Lee and T. Kawai, Appl. Phys. Lett. 81, 2421 (2002). Kim02b H. Kimura, T. Fukumura, M. Kawasaki, K. Inaba, T. Hasegawa, and H. Phys. 92, u, T. W. Noh, H. J. Lin, H. -H. Hsieh, C. T. Chen, Phys. Rev. Lett. 90, 017401 (2003). Kim03b D. H. Kim, J. S. Yang, K. W. Lee, S. D. Bu, D. -W. Kim, T. W. Noh, S.-J. J. Y. Won, S. H. Park, and J. C. Lee, J. Appl. Phys. 93, 6125 (2003). Kim03c D. H. Kim, J. S. Yang, K. Y. S. Kim, D. -W. Kim, T. W. Noh, S. D. Bu, s. , S. A. Chambers, R. P. Sears, B. Taylor, and B. Sinkovic, 84Kim04b Y. -W. Kim, Y. D. Park, S. J. Pearton, and J.-H. Park, Ann. Phys. 13, 70 Koi00 T. Itoh, A. Ohtomo, M. Murakami, Z. Jin, and Y. Matsumoto, Physica C 335, 245 (2000). Ku03 , P. Schiffer, N. Sarmath, M. J. Seong, A. Mascarenhas, E. Johnston-Halperin, R. C. Myers, A. C. J. M. Kikkawa Kim01 H. Kimura, T. Fukumura, H. Koinuma, and M. Kawasaki, Physica E 10, 265 (2001). Kim02aD. H. Kim, J. S. Yang, K. W. Lee, S. D. Bu, T. W. Noh, S.-J. Oh, Y. –W. Kim, J.-S. Chung, H. T Koinuma, Appl. Phys. Lett. 80, 94 (2002). Kim02c J. H. Kim, H. Kim, D. Kim, Y. E. Ihm, and W. K. Choo, J. Appl.6066 (2002). Kim03aJ. Y. Kim, J.-H. Park, B.-G. Park, H. -J. Noh, S. -J. Oh, J. S. Yang, D. -H. Kim, S. D. B Oh, Y. -W. Kim, J.-S. Chung, H. Tanaka, H. Y. Lee and T. Kawai, Y. -W. Kim, Y. D. Park, S. J. Pearton, Y. Jo, and J.-G. Park, Appl. PhyLett. 83, 4574 (2003). Kim04aY. J. Kim, S. Thevuthasan, T. Droubay, A. S. Lea, C. M. Wang, V. Shutthanandan Appl. Phys. Lett. , 3531 (2004). D. H. Kim, J. S. Yang, K. Y. S. Kim, Y. J. Chang, T. W. Noh, S. D. Bu, (2004). H. Koinuma, M. Kawasaki, K. C. Ku, S. J. Potashnik, R. F. Wang, S. H. Chun Gossard, and D. D. Awschalom, Appl. Phys. Lett. 82, 2302 (2003).

PAGE 140

128 Kum02 D. Kumar, S. V. Pietambaram, V. Craciun, R. K. Singh, J. Perriere, and J. Sankar, J. Vac. Sci. Technol. A 20, 198 (2002). Lee94 M. B. Lee, M. Kawasaki, M. Yoshimoto, M. Kumagai and H. Koinuma, Lee95 M. B. Lee, M. Kawasaki, M. Yoshimoto, B. K. Moon, H. Ishiwara and H. Low96 D. H. Lowndes, D. B. Geohegan, A. A> Puretzky, D. P. Norton, and C. M. Mat98 F. Matsukura, H. Ohno, A. Shen, and Y. Sugawara, Phys. Rev. B 57, Mat99 Y. Matsumoto, M. Murakami, Z. Jin, A. Ohtomo, M. Lippmaa, M. Mat01a Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura, M. (2001). a, T. Fukumura, M. Kawasaki, S.-Y. Koshihara, and H. Koinuma, Jpn. J. Appl. Phys. 40, L1205 (2001). Mat02 Y. Matsumoto, M. Murakami, T. Hasegawa, T. Fukumura, M. Kawasaki, Lett. 74, 3273 (1995). Mur01a M. Murakami, Y. Matsumoto, K. Nakajima, T. Makino, Y. Segawa, T. ys. Lett. 78, 2664 (2001). Mur01b rray, S. Sun, H. Doyle, and T. Betley, MRS Bulletin, 12, 985 (2001). Mur04a . Nagano, T. Hasegawa, M. Kawasaki, and H. Koinuma, Appl. Surf. Sci. 223, 245 (2004). Jpn. J. Appl. Phys. 33, 6308 (1994). Koinuma, Jpn. J. Appl. Phys. 34, 808 (1995). Rouleau, Science, 273, 898 (1996). R2037 (1998). Kawasaki, and H. Koinuma, Jpn. J. Appl. Phys. 38, L603 (1999). Kawasaki, P. Ahmet, T. Chikyow, S.-Y. Koshihara and H. Koinuma, Science 291, 854 Mat01bY. Matsumoto, R. Takahashi, M. Murakami, T. Koida, X.–J. Fan, T. Hasegaw P. Ahmet, K. Nakajima, T. Chikyow and H. Koinuma, Appl. Surf. Sci. 189, 344 (2002). Mck98R. A. McKee, F. J. Walker, and M. F. Chisholm, Phys. Rev. Lett. 81, 3014(1998). Mol91 S. von Molnar, H. Munekata, H. Ohno, and L. L. Chang, J. Magn. Magn. Mater. 93, 356 (1991). Moo95 J. S. Moodera, L. R. Kinder, T. M. Wong, and R. Meservey, Phys. Rev. Chikyow, P. Ahmet, M. Kawasaki, and H. Koinuma, Appl. Ph C. B. Mu M. Murakami, Y. Matsumoto, M

PAGE 141

129 Mur04bM. Murakami, Y. Matsumoto, T. Hasegawa, P. Ahmet, k. Nakajima, T. Chikyow, H. Ofuchi, I. Nakai, and H. Koinuma, J. Appl. Phys. 95, 53 30 (2004). Nar03 J. Narayan and B. C. Larson, J. Appl. Phys. 93, 278 (2003). Nar92 J. Narayan, P. Tiwari, X. Chen, J. Singh, R. Chowdhury and T. Zheleva, Naz03 A. M. Nazmul, S. Sugahara, and M. Tananka, Phys. Rev. B 67, 241308 Nie03 , E. I. Meletis, L. D. Tung, and L. Spinu, J. Appl. Phys. 93, 4750 (2003). Nor03 , K. Pruessner, J. D. Budai, L. A. Boatner, M. F. Chisholm, J. S. Lee, Z. G. Khim, Y. D. Park, and R. G. Oba99 K. Obata, K. Sugioka, K. Toyoda, H. Takai, and K. Midorikawa, Appl. Oes99 don) 402, 735 (1999). tesan, Phys. Rev. Lett. 91, 077205 (2003). Ohn00 London) 408, 944 (2000). Ohn96 . Oiwa, A. Endo, S. Katsumoto, and Y. Iye, Appl. Phys. Lett. 69, 363 (1996). Ohn98a l. tt. 73, 363 (1998). Ohn99 H. Ohno, J. Magn. Magn. Mater. 200, 110 (1999). Ohs04 Ott91 M. T. Otten, J. Elec. Micro. Tech. 17, 221 (1991). Appl. Phys. Lett. 61, 1291 (1992). (2003). X. Nie, J. C. Jiang D. P. Norton, M. E. Overberg, S. J. Pearton Wilson, Appl. Phys. Lett. 83, 5488 (2003). Phys. A 69, S715 (1999). M. Oestreich, Nature (Lon Oga03 S. B. Ogale, R. J. Choudhary, J. P. Buban, S. E. Lofland, S. R. Shinde, S. N. Kale, V. N. Kulkarni, J. Higgins, C. Lanci, J. R. Simpson, N. D. Browning, S. D. Sarma, H. D. Drew, R. L. Greene, and T. Venka H. Ohno, D. Chiba, F. Matsukura, T. omlya, E. Abe, T. Dietl, Y. Ohno and K. Ohtanl, Nature ( Ohn02 H. Ohno, J. Magn. Magn. Mater. 242, 105 (2002). H. Ohno, A. Shen, F. Matsukura, A H. Ohno, N. Akiba, F. Matsukura, A. Shen, K. Ohtani, and Y. Ohno, AppPhys. Le Ohn98bH. Ohno, Science 281, 951 (1998). T. Ohsawa, Y. Matsumoto, and H. Koinuma, Appl. Surf. Sci. 223, 84 (2004).

PAGE 142

130 Par00 Y. D. Park, B. T. Jonker, and B. R. Bennett, G. Itskos, M. Furis, G. Kioseoglou, and A. Petrou, Appl. Phys. Lett. 77, 3989 (2000). Par02a W. K. Park, R. J. Ortega-Hertogs, J. S. Moodera, A. Punnoose and M. S. Seehra, J. Appl. Phys. 91, 8093 (2002 ). nos, B. T. Jonker, Science 295, 651 (2002). Par04 . Ryu, and Y. M. Kim, Appl. Phys. Lett. 84, 1388 (2004). Pea03a R. Abernathy, M. E. Overberg, G. T. Thaler, D. P. Norton, N. Theodoropoulou, A. F. Hebard, Y. D. Park, F. Ren, J. Kim, and Pea03b S. J. Pearton, C. R. Abernathy, D. P. Norton, A. F. Hebard, Y. D. Park, L. Pea03c thy, G. T. Thaler, R. Frazier, F. Ren, A. F. Hebard, Y. D. Park, D. P. Norton, W. Tang, M. Stavola, J. M. Zavada, and Pea04a Pea04c S. J. Pearton, C. R. Abernathy, G. T. Thaler, R. M. Frazer, D. P. Norton, F. J. Phys. 16, R209 (2004). Pri95 G. Prinz, Physics Today, April, 58 (1995). Pri98 2, 1660 (1998). s. Ram04 S. Ramachandran, A. Tiwari, and J. Narayan, Apply. Phys. Lett. 84, 5255 Ras00 E. I. Rashba, Phys. Rev. B 62 R16267 (2000). Par02b Y. D. Park, A. T. Hanbicki, S. C. Erwin, C. S. Hellberg, J. M. Sullivan, J. E. Mattson, T. F. Ambrose, A. Wilson, G. Spa J. H. Park, M. G. Kim, H. M. Jang, S S. J. Pearton, C. L. A. Boatner, J. Appl. Phys. 93, 1 (2003). A. Boatner and J. D. Budai, Mat. Sci. Eng. R 40, 137 (2003). S. J. Pearton, C. R. Aberna R. G. Wilson, Physica B 340, 39 (2003). S. Pearton, Nature Materials, 3, 203 (2004). Pea04b S. J. Pearton, D. P. Norton, K. Ip, Y. W. Heo, and T. Steiner, J. Vac. Sci. Technol. B 22, 932 (2004). Ren, Y. D. Park, J. M. Zavada, I. A. Buyanova, W. M. Chen, and A. F. Hebard, Pre03 W. Prellier, A. Fouchet, and B. Mercey, J. Phys. 15, R1583 (2003). G. A. Prinz, Science 28 Pun03 A. Punnoose, M. S. Seehra, W. K. Park, and J. S. Moodera, J. Appl. Phy93, 7867 (2003). (2004).

PAGE 143

131 Ree01 M. K. Reed, N. A. El-Masry, H. H. Stadelmaier, M. K. Ritums, M. J. Reed, C. A. Parker, J. C. Roberts, and S. M. Bedair, Appl. Phys. Lett. 79, 3473 (20 01). Sat00 K. Sato and H. Katayama-Yoshida, Jpn. J. Appl. Phys. 39, L555 (2000). Sat01 tayama-Yoshida, Jpn. J. Appl. Phys. 40, L334 (2001). Sch02 D. G. Schlom and J. H. Haeni, MRS Bulletin 27, 198 (2002). Sha03 . Rao, F. J. Owens, R. Sharma, R. Ahuja, J. M. O. Guillen, B. Johansson, and G. A. Gehring, Nature Materials, 2, 673 . L. Shi04 S. R. Shinde, S. B. Ogale, J. Higgins, H. Zheng, A. J. Millis, V. N. . yis, D. Givord, and J. Nogues, Nature (London), 423, 850 (2003). Sim02 e, Y. Zhao, S. B. Ogale, and T. Venkatesan, cond-mat/0205626. Sin02 un, H. Hofmann, and R. K. Singh, Electrochem. Solid-State Lett. 5, A198 (2002). Smi76 Metals Reference Book, 5 edition, edited by C. J. Smithells (1976). Sor93 L. Soriano, M. Abbate, J. Vogel, J. C. Fuggle, A. Fernandez, A. R. . Sri99 A. Srivastava, V. Craciun, J. M. Howard and R. K. Singh, Appl. Phys. Rud54 M. A. Ruderman and C. Kittel, Phys. Rev. 96, 99 (1954). San94 R. Sanjines, H. Tang, H. Berger, F. Gozzo, G. Margaritondo and F. Levy,J. Appl. Phys. 75, 2945 (1994). Sar01 S. D. Sarma, Amer. Sci. 89, 516 (2001). K. Sato and H. Ka Sch00 G. Schmidt, D. Ferrand, L. W. Molenkamp, A. T. Filip, B. J. van Wees, Phys. Rev. B 62, R4793 (2000). P. Sharma, A. Gupta, K. V (2003). Shi03 S. R. Shinde, S. B. Ogale, S. D. Sarma, S. E. Lofland, C. Lanci, J. P. Buban, N. D. Browning, V. N. Kulkarni, J. Higgins, R. P. Sharma, RGreene, and T. Venkatesan, Phys. Rev. B 67, 115211 (2003). Kulkarni, R. Ramesh, R. L. Greene, and T. Venkatesan, Phys. Rev. Lett92, 166601 (2004). Sku03 V. Skumryev, S. Stoyanov, Y. Zhang, G. Hadjipana J. R. Simpson, H. D. Drew, S. R. Shind D. Singh, W. Kim, V. Craci th Gonzalez-Elipe, M. Sacchi, and J. M. Sanz, Surf. Sci. 290, 427 (1993) Lett. 75, 3002 (1999).

PAGE 144

132 Sul03 J. M. Sullivan and S. C. Erwin, Phys. Rev. B 67, 144415 (2003). Sul91 J. L. Sullivan, S. O. Saied, and I. Bertoti, Vacuum 42, 1203 (1991). Sun99 . 85, 4325 (1999). Tha02 G. T. Thaler, M. E. Overberg, B. Gila, R. Frazier, C. R> Abernathy, S. J. , 64 (2002). , S. J. J. Pearton, and R. G. Wilson, Solid-State Electron. 47, 2231 (2003). Tom91 Toy04a H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T. ature Materials 3, 221 (2004). Toy04b . Yamada, K. Nakajima, T. Chikyow, T. Hasegawa, H. Koinuma, and M. Kawasaki, cond-mat/0307760. Ued01 Van81 T. Van Duzer and C. W. Turner, Principles of Super-conducting Devices . Jagannadham, N. Parikh, A. Suvkhanov, and J. D. Budai, J. Appl. Phys. 80, 6720 (1996). Wag79 C. D. Wagner, W. M. Riggs, L. E. Davis, and J. F. Moulder, Handbook of ie, MN (1979). tt. 83, 518 (2003). Spinu, J. Appl. Phys. 93, 7870 (2003). Wit81 M. Wittmer, J. Noser, and H. Melchior, J. Appl. Phys. 52, 6664 (1981). S. Sun and C. B. Murray, J. Apply. Phys Tan94 H. Tang, K. Prasad, R. Sanjines, P. E. Schmid, and F. Levy, J. Appl. Phys.75, 2042 (1994). Pearton, J. S. Lee, S. Y. Lee, Y. D. Park, Z. G. Khim, J. Kim, and F. RenAppl. Phys. Lett. 80, 39 The02 N. Theodoropoulou, A. F. Hebard, M. E. Overberg, C. R. AbernathyPearton, S. N. G. Chu, and R. G. Wilson, Phys. Rev. Lett. 89, 107203 (2002). The03 N. A. Theodoropoulou, A. F. Hebard, D. P. Norton, J. D. Budai, L. A. Boatner, J. S. Lee, Z. G. Khim, Y. D. Park, M. E. Overberg, S. H. G. Tompkins, J. Appl. Phys. 70, 3876 (1991). Hasegawa, H. Koinuma, and M. Kawasaki, N H. Toyosaki, T. Fukumura, Y K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79, 988 (2001). and Circuits, Elsevier, New York (1981). Vis96 R. D. Vispute, J. Narayan, K. Dovidenko, K X-ray Photoelectron Spectroscopy (edited by G. E. Muiledberg), Perkin Elmer, Eden Prair Wan03aZ. Wang, W. Wang, J. Tang, L. D. Tung, L. Spinu and W. Zhou, Appl. Phys. Le Wan03bZ. Wang, J. Tang, L. D. Tung, W. Zhou, and L.

PAGE 145

133 Wu00 W. Wu, K. H. Wong, and C. L. Choy, Thin Solid Films 360, 103 (200S. Y. Wu, H. X 0). Wu03 . Liu, L. Gu, R. K. Singh, L. Budd, M. V. Schilfgaarde, M. R. McCartney, D. J. Smith, and N. Newman, Appl. Phys. Lett. 82, 3047 . M. Jyumonji, J. Mater. Sci. 33, 1777 (1998). Yag96 E. Yagi, R. R. Hasiguti, and M. Aono, Phys. Rev. B 54, 7945 (1996). Yam99 T. Yamamoto and H. Katayama-Yoshida, Jpn. J. Appl. Phys. 38, L166 Yan01 H. S. Yang, K.-I. Min, Y. K. Kim, S. R. Lee, G.-S. Park, and S. A. Song, Yan03 H. S. Yang, J. Choi, V. Craciun, and R. K. Singh, J. Appl. Phys. 93, 7873 Yan04a m. Solid-State Lett. 7. C4-C6 (2004). Yan04b Yan04c H. S. Yang and R. K. Singh, J. Appl. Phys (accepted). Yos57 . ary, V. N. fland, C. Lanci, J. P. Buban, N. D. Browning, S. D. Sarma, and A. J. Millis, Appl. Phys. Lett. herr, (2003). Xio92 P. Xiong, G. Xiao, J. Q. Wang, J. Q. Xiao, J. S. Jiang, and C. L. Chien, Phys. Rev. Lett. 69, 3220 (1992) Xu98 S. Xu, L. Du, K. Sugioka, K. Toyoda, and (1999). J. Magn. Magn. Mater. 226, 2070 (2001). Yan02 S. G. Yang, A. B. Pakhomov, S. T. Hung, and C. Y. Wong, IEEE trans. Magn. 38, 2877 (2002). (2003). H. S. Yang, J. Y. Choi, S. J. Song, and R. K. Singh, Electroche H. S. Yang and R. K. Singh, J. Appl. Phys. 95. (2004). K. Yosida, Phys. Rev. 106, 893 (1957). Zen51 C. Zener, Phy. Rev. 81, 440 (1951); Phy. Rev. 82, 403 (1951); Phy. Rev83, 299 (1951). Zha03 Y. G. Zhao, S. R. Shinde, S. B. Ogale, J. Higgins, R. J. ChoudhKulkarni, R. L. Greene, T. Venkatesan, S. E. Lo 83, 2199 (2003). Zhu01 H. J. Zhu, M. Ramsteiner, H. Kostial, M. Wassermeier, H. –P. SchnK. H. Ploog, Phys. Rev. Lett. 87, 016601 (2001). Zut02 I. Zutic, J. Fabian and S. D. Sarma, Phys. Rev. Lett. 88, 06603 (2002).

PAGE 146

Hyucksoo Yang was born in Seoul, Korea, on January 16, 1971. After graduating from Posung HMetallurgical Engineering. He received a B. S. degree in 1994 and then earned a M. S. degree in metagraduation, he worked for five and a half years as an assistant manager at Samsung, in Korea. While aried out through on) and ion implantation, combined with analysis of the structural, electrical and magnetic ce, he decided to pursue the Ph.D. degree in the United States. In 2001, he enrolled at the University of Florida in the Department of Materials Science and Engineering, specializing in are in the areas of growth and understanding the relationships between atomic-level structure and properties. BIOGRAPHICAL SKETCH igh School, he enrolled at Yonsei University in the Department of llurgical engineering from Seoul National University in 1996. Upon t Samsung, he was involved in the development of fuel cells, giant magneto-resistance heads, and semiconductors. Research was car materials processing including thin film depositions (sputtering and ion beam depositi properties. After completion of his duty at Samsung as a substitution for military servi electronic materials. His research interests characterization of novel ferromagnetic semiconductor films with emphasis on the 134