Development of Magnetic Semiconductor Based on Spinel- Title type Oxide for Spintronics Application( Dissertation_全文 )
Author(s) Murase, Hideaki
Citation 京都大学
Issue Date 2010-03-23
URL https://doi.org/10.14989/doctor.k15388
Right 許諾条件により要旨・本文は2011-03-31に公開
Type Thesis or Dissertation
Textversion author
Kyoto University
Development of Magnetic Semiconductor Based on Spinel type Oxide for Spintronics Application
Hideaki Murase 2010
Contents
General Introduction...... 1
Chapter 1: Spintronics: spintronic applications and magnetic semiconductors ...... 8
Chapter 2: Magnetite ulvöspinel solid solution ...... 27
Chapter 3: Synthesis of magnetite ulvöspinel solid solution thin films .....45
3.1 Magnetite ulvöspinel solid solution thin films on α Al 2O3(0001) substrates ...... 45
3.2 Magnetite ulvöspinel solid solution thin films on MgO(100) substrates ...... 64
Chapter 4: Electronic and magnetic structure of magnetite ulvöspinel solid solution thin films...... 77
Summary...... 100
List of publications ...... 103
Acknowledgements ...... 104
General Introduction
Electronics has developed and supported modern information technology, which has caused drastic changes in the quality of our daily life. Nowadays, it becomes feasible that information with very high density is transmitted and processed in a very short time. The advancement in electronics has been accomplished by the development of various types of sophisticated devices which have been mainly made from semiconductors. Also, techniques to fabricate very tiny structures at nano to micro scales with precise dimensions have contributed to the invention of electronics devices. However, the technique for downsizing to obtain the devices like very large scale integrated circuits is now facing such a problem that the size of the devices will reach the limit of the present fabrication technique very soon. Alternative way to resolve such an issue is to use multifunctional materials instead of prevalent semiconductors like Si and to utilize the spin, which is another degree of freedom of carriers. It is known that metal oxides are one of the multifunctional materials. The metal oxides have been also ubiquitous in many fields of technology and industry since early times. They possess remarkable physical properties such as ferroelectric, ferromagnetic, metallic, semiconducting, superconducting, pyroelectric, piezoelectric, and multiferroic ones. The discovery of superconductivity with high critical temperature (Tc) in layered perovskite type cuprate oxides in 1986 [1] has opened up a new possibility of metal oxides from a point of view of both fundamental science and practical applications, and has also triggered findings of novel and exotic metal oxides involving high Tc superconductors. For instance, in 1994, Jin et al. [2] observed
‘colossal magnetoresistance’ (CMR) in ferromagnetic perovskite of La 1−xCa xMnO 3. In
2004, multiferroic TbMn 2O5, in which an interaction between ferroelectricity and magnetism is so strong that electrical polarization can be reversed by an external
1 magnetic field, was reported [3]. Very recently, a new type of high Tc superconductive properties were found in iron based layered oxides, LaO 1–xFxFeAs, and intense studies have been carried out from both aspects of solid state science and device applications [4,5]. Thus, metal oxides have been efficiently utilized in many fields of oxide electronics, and have potential applications in the future technologies involving new system of electronics. Among the metal oxides, perovskite type oxides and the related compounds are of interest from both fundamental and technical aspects. Perovskite type structure AB O3 can hold about 30 elements in the A sites and over half the periodic table in the B sites. Because of this malleable characteristic as host material, it is feasible to customize various physical properties. For instance, SrTi 1–xNb xO3 exhibits insulating, semiconducting, and metallic properties, depending on the concentration of Nb [6].
Incorporation of divalent cations such as Ca, Sr into LaMnO 3, i.e. La 0.7 A0.3 MnO 3, provides colossal magnetoresistance [1,7]. Also, Hg Ba Ca Cu O system with a layered perovskite type structure shows high superconducting transition temperature above 130 K [8]. On the other hand, recent technical advances in thin film growth have allowed intense studies on superlattices that are inaccessible by the conventional chemical methods. The precise control of thin film growth process has led to the discovery of unique electronic structure and properties of an interface between perovskite type oxides layers, for instance two dimensional electron gas (2D) with high mobility at the interface between LaAlO 3 and SrTiO 3 [9], giant 2D Seebeck coefficient of SrTiO 3 SrTi 0.8 Nb 0.2 O3 [10], and large magnetoresistance and magnetic hysteresis at the interface between LaAlO 3 and TiO 2 terminated SrTiO 3 [11] were prepared. Spinel type oxides containing transition metal (TM) elements are as unique and intriguing as perovskite type oxides from a viewpoint of structural and physical properties. In the spinel type structure, there are two kinds of cation sites (tetrahedral and octahedral sites) in the cubic closest packing of oxides ions, where various TM ions can occupy, so that spinel type oxides shows a wealth of the physical properties which
2 can be utilized in practical applications. For example, spinel type oxides containing iron ions as a main constituent are well known as ‘ferrite’, the properties of which have significantly contributed to the advancement in solid state physics, in particular, in magnetism [12–14]. Ferrites are mainly categorized into two types based on their magnetic coercivity: soft and hard ferrites. The former has been primarily used in transformers, inductors, recording heads, and microwave devices, and the latter in permanent magnet motors and storage media in magnetic recording devices [15].
LiTi 2O4, which also has a spinel type structure, is known as the first oxide superconductor with a relatively high critical temperature ( Tc ~ 12 K) and extended studies have been performed; for instant, studies about unusual characteristics in specific heat at the critical temperature [16,17]. Also, Cd 2SnO 4 is a transparent conductor, which can be used for high efficiency solar cells [18]. More recently, Yamasaki et al. have disclosed multiferroic behavior with a spontaneous magnetization in CoCr 2O4 crystal which has a conical spin structure [19]. The spinel type compounds have been mainly examined in the form of bulk material for many decades. In contrast, there are only a few reports about the synthesis, structural characterization, and physical properties of thin film form of spinel type compounds. Thin films are also important for the fabrication of sophisticated devices in electronics. On the other hand, semiconductors such as Si, Ge, GaAs, GaN, and SiC are essential for the fabrication of prevalent electronic devices (e.g., diodes and transistors), which mainly utilize only the charge of carriers. Recently, an attempt to use the spin of electrons and positive holes in addition to the charge has been made to gain new functionalities. This is a new emerging field of electronics called ‘spintronics’. Although a wide range of materials from metal to organic compounds have been studied to fabricate device in spintronics, magnetic oxide semiconductors have attracted considerable attention because of a variety of physical properties derived from the oxides as mentioned above. The main theme in the present thesis is the development of novel magnetic
3 semiconductor thin film which is applicable in spintronics. From a viewpoint of room temperature ferromagnetism, easiness to control the carrier type, and high spin polarization, spinel type TM oxides with intrinsic ferromagnetism are one of the promising materials for spintronic applications. Here, magnetite (Fe 3O4) and ulvöspinel (Fe 2TiO 4) solid solutions, (1–x)Fe 3O4 xFe 2TiO 4 (molar ratio: 0 < x < 1), are selected as a candidate material for semiconductor spintronics, and the possibilities of thin films of the solid solutions for spintronic applications are discussed. This thesis is composed of four chapters. The respective chapter outlines are described as follows. In chapter 1, the concept and history of spintronics are firstly described. Subsequently, main devices in spintronics technology are briefly overviewed, and then three types of magnetic semiconductors which can be candidate materials in semiconductor spintronics are summarized. In chapter 2, firstly, fundamental structure and properties as well as magnetic and transport functions applicable to spintronics for Fe 3O4 are summarized in order to understand the structure and properties of solid solutions based on Fe 3O4, and then the structural, electrical, and magnetic properties of (1–x)Fe 3O4 xFe 2TiO 4 solid solutions in the form of bulk material are presented. Finally, the future perspective of the solid solutions is discussed.
In chapter 3, the fabrication of (1–x)Fe 3O4⋅xFe 2TiO 4 ( x=0.6) solid solutions thin films using a pulsed laser deposition technique is described. Section 3.1 deals with epitaxial growth of the solid solution thin film on α Al 2O3 (0001). This is the first report on the epitaxial thin film of the solid solution, for which the structural, ferrimagnetic and semiconducting properties, and spin polarization are discussed. In section 3.2, epitaxial solid solution thin films with a flat and smooth surface at an atomistic level, deposited on MgO(100) substrates, are described. Also, physical properties of the solid solution thin films are discussed by comparing with those of films grown on α Al 2O3(0001).
4 In chapter 4, measurements of Ti L2,3 and Fe L2,3 edges x ray absorption near edge structure, x ray magnetic circular dichroism are performed for Fe 3O4 Fe 2TiO 4 solid solution thin films in order to clarify the electronic and magnetic structures of thin films. The experimental results are discussed on the basis of theoretical spectra derived by first principle calculations. Finally, the whole results and discussions are summarized in the last summary section.
5 References
[1] J. G. Bednorz and K. A. Müller, Z. Phys. B 64 , 189 (1986). [2] S. Jin, T. H. Tiefel, M. McCormack, R. A. Fastnacht, R. Ramesh, and L. H. Chen, Science 264 , 413 (1994). [3] N. Hur, S. Park, P. A. Sharma, J. S. Ahn, S. Guha and S. W. Cheong, Nature 429 , 392 (2004). [4] Y. Kamihara, T. Watanabe, M. Hirano, and H. Hosono, J. Am. Chem. Soc. 130 , 3296 (2008). [5] H. Takahashi, K. Igawa, K. Arii, Y. Kamihara, M. Hirano, and H. Hosono, Nature 453 , 376 (2008). [6] O. N. Tufte and P. W. Chapman, Phys. Rev. 155 , 796 (1967). [7] B. Nadgorny, I. I. Mazin, M. Osofsky, R. J. Soulen Jr., P. Broussard, R. M. Stroud, D. J. Singh, V. G. Harris, A. Arsenov, and Y. Mukovskii, Phys. Rev. B 63 , 184433 (2001). [8] A. Schilling, M. Cantoni, J. D. Guo, and H. R. Ott, Nature 363 , 56 (1993). [9] A. Ohtomo and H. Y. Hwang, Nature 427 , 423 (2004). [10] H. Ohta, S. Kim, Y. Mune, T. Mizoguchi, K. Nomura, S. Ohta, T. Nomura, Y. Nakanishi, Y. Ikuhara, M. Hirano, H. Hosono, and K. Koumoto, Nature Mater. 6, 129 (2007). [11] A. Brinkman, M. Huijben, M. Van Zalk, J. Huijben, U. Zeitler, J. C. Maan, W. G. Van Der Wiel, G. Rijnders, D. H. A. Blank, and H. Hilgenkamp, Nature Mater. 6, 493 (2007). [12] L. Néel, Ann. Phys. 3, 137 (1948). [13] F. Walz, J. Phys. : Condens. Matter 14 , R285 (2002). [14] J. García and G. Subías, J. Phys. : Condens. Matter 16 , R145 (2004). [15] J. Smit and H.P.J. Wijn, Ferrites: Physical Properties of Ferrimagnetic Oxides in Relation to Their Technical Applications, Cleaver Hume, London (1959).
6 [16] D. C. Johnston, J. Low. Temp. Phys. 25 , 145 (1976). [17] C. P. Sun, J. Y. Lin, S. Mollah, P. L. Ho, H. D. Yang, F. C. Hsu, Y. C. Liao, and M. K. Wu, Phys. Rev. B 70 , 054519 (2004). [18] S. B. Zhang and S. H. Wei, Appl. Phys. Lett. 80 , 1376 (2002). [19] Y. Tamasaki, S. Miyasaka, Y. Kaneko, J. P. He, T. Arima, and Y. Tokura, Phys. Rev. Lett. 96 , 207204 (2006).
7 CHAPTER 1 Spintronics: spintronic applications and magnetic semiconductors
1.1 Introduction to spintronics
The industry and technology of semiconductors are facing such a serious problem that the development of semiconductors based on the Moore’s law loses momentum. So far, the conventional semiconductor electronics mainly uses only the charge of carriers but has not controlled the spin, which is another degree of freedom of carriers. A new field of electronics simultaneously utilizing both the charge and the spin is called ‘spin electronics’ or ‘spintronics’. This technology will lead to the creation of next generation devices which are smaller and more versatile than the conventional devices and circuits based on silicon. In addition, spintronic devices will consume less energy compared with the conventional devices based on only the charge, because the energy needed to change the orientation of spin is much smaller than that typically required for charge transport. In the mid 1930’s, Mott proposed the famous two current model for various magnetoresistive phenomena [1], and this can be considered to be the beginning of the development of spintronics. In 1975, the tunneling magnetoresistance (TMR) effect for a magnetic tunnel junction was reported by Julliere [2]. Then, the breakthrough happened in 1988 when Fert et. al. [3] and Grünberg et. al. [4] independently discovered the giant magnetoresistance (GMR) effect in ferromagnetic metallic superlattices. For their findings, Grünberg and Fert won the 2007 Nobel prize in physics. Their
8 discovery brought about a rapid development of spintronics, resulting in the recent commercial success such as a hard disk drive accommodated in a computer. As described above, so far, spintronics has achieved great advances using ferromagnetic metals as various types of novel devices. On the other hand, spintronics based on semiconductor has recently attracted considerable attention, because the magnetic semiconductor has some advantages compared with magnetic metals. Firstly, the crystal growth and the integrating technologies in conventional electronics can be utilized for semiconductor spintronic devices. Secondly, since in many magnetic semiconductors the magnetism is related with the carriers, the magnetism can be altered by changing the bias voltage. In addition, carrier concentration and physical properties can be controlled by incorporation of aliovalent elements and light irradiation. In this chapter, spintronics is reviewed, and typical spintronic applications are introduced. Also, various magnetic semiconductors are overviewed as candidate materials for semiconductor spintronics.
1.2 Spintronic applications
Spintronics is one emerging technology that combines conventional electronics and magnetics. The interaction of spins with light can be also considered in the spintronics, although such a field is often called magneto optics. Novel phenomena relevant to the behavior of charge and spin of carries and atoms (ions) in solids and their interactions with light can lead to many types of functional materials. A conceptual diagram is shown in Fig. 1.1. Here, firstly, an explanation of ‘spin polarization’ is provided in order to understand spintronic applications. Then, TMR and GMR devices, which are essential for spintronic applications, are described, and spin FET is illustrated as another example of spintronic device.
9 Optics
Magnetics Semiconductor electronics Light
Spin Charge
Spintronics
Figure 1.1: A conceptual diagram of spintronics.
10 1.2.1 Spin polarization Spin polarization of electrons is one of the most important phenomena in the spintronics, because the degree of spin polarization has influence on the magnetoresistance. The spin polarization P is defined in terms of the density of states of the two spin bands at Fermi level ( EF):
N (E ) − N (E ) P = ↑ F ↓ F , (1) N↑ (EF ) + N↓ (EF )
where N is the density of states of up ( ↑) or down ( ↓) spin electrons at EF. A more general and useful definition is as follows:
N (E )w − N (E )w P'= ↑ F ↑ ↓ F ↓ , (2) N↑ (EF )w↑ + N↓ (EF )w↓
where w is the weighting factors in each of up and down electrons. Paramagnetic metals such as Cu and Au have the value P = 0, whereas ferromagnetic metals with P = +1 or –1 (i.e., the spin polarization is 100% or –100%, respectively) are called half metals, which have an unusual band structure where only half of the electrons are conducting. Schematic band diagrams of paramagnetic metals, ferromagnetic metals, and half metals are illustrated in Fig 1.2. For paramagnetic metals, because there are no magnetic interactions between electrons, up and down spins form the identical density of states to each other. As a result, the number of up and down spins occupying the states from the bottom of the energy band to EF is the same, leading to P = 0 [Fig.1.2a]. In the case of ferromagnetic metals, since the density of states of up and down spins is different, P varies in a range from 0 to 100% [Fig.1.2b]. Half metals are different from both paramagnetic and ferromagnetic metals; they are metals for electrons of one spin (either ↑ or ↓ spin) in their band structures, but are
11 (a) Paramagnetic (b) Ferromagnetic (c) Half metal metal metal
N (E ) N (E ) ↑ F ↓ F N↑ (EF ) N↓ (EF ) N↑ (EF ) EF
N
P = 0 P > 0 P = 1
Figure 1.2: Schematic band diagrams of paramagnetic metal, ferromagnetic metal, and half metal.
12 insulators for those of the opposite spin [Fig.1.2c]. For the determination of spin polarization, photoemission [5], magnetic tunnel junction [6], and point contact [7] are known as the common methods.
1.2.2 TMR and GMR devices Magnetoresistance is a phenomenon that electrical resistance of a material is varied when an external magnetic field is applied, and is very important in the spintronics. Here TMR and GMR effects are explained in the following. TMR effect was firstly observed by Julliere in 1975 [2]. Figure 1.3 is the schematic illustration of a representative TMR device. When two types of ferromagnetic layers (a pinned layer and a free layer) are separated by a very thin layer of insulator (commonly, MgO, Al 2O3), the electrical resistance of the multilayer in direction perpendicular to the film surface changes depending on the direction of magnetizations of ferromagnetic thin layers. When the directions of the magnetizations of two ferromagnetic layers are parallel ( ↑↑), the number of electrons tunneling between the two ferromagnetic electrodes through the insulator layer becomes larger, resulting in a larger tunneling current. In contrast, when the magnetizations of two ferromagnetic layers are antiparallel (↑↓), the tunneling of electrons is suppressed. Thus, the tunneling current for the antiparallel magnetizations becomes smaller than that for the parallel magnetizations. This phenomenon is called tunneling magnetoresistance (TMR). Magnetoresistance ratio for TMR is defined as follows:
2 Rp − Rap 2P MR = = 2 , (3) Rp 1− P
where Rp and Rap are the tunneling resistance in the parallel ( p) and antiparallel ( ap ) configurations, respectively, and P is the spin polarization of the electrodes. When the
13 Low resistance High resistance (Parallel) (Antiparallel)
F1 F2 F1 F2
Insulator layer Insulator layer EE EE
N (E ) N (E ) N (E ) N (E ) ↑ F ↓ F N↑ (EF ) N↓ (EF ) ↑ F ↓ F N↑ (EF ) N↓ (EF )
Figure 1.3: Schematic illustrations of tunneling magnetoresistance (TMR) effect. F1 and F2 are ferromagnetic layers: (left) a pinned layer and (right) a free layer, respectively.
14 Ferromagnetic layer Nonmagnetic layer
(a) Parallel alignment (b) Antiparallel alignment
Figure 1.4: Schematic illustrations of giant magnetoresistance (GMR) effect.
15 ferromagnetic electrodes are half metal materials, the TMR ratio theoretically becomes infinity. This fact leads to vigorous research on TMR devices: in particular, studies to achieve a non volatile magnetic random access memory (MRAM). On the other hand, GMR effect was independently discovered by Fert et al. [3] and Grünberg et al. [4] in 1988. Schematic illustration of a typical GMR device is shown in Fig. 1.4. The GMR effect is observed in artificial lattice systems consisting of alternating ferromagnetic and nonmagnetic layers. When an external magnetic field is applied, the directions of magnetization of ferromagnetic layers are parallel (↑↑). In contrast, when an external magnetic field is zero, the directions of magnetization of ferromagnetic layers are antiparallel ( ↑↓) due to a weak antiferromagnetic coupling between ferromagnetic layers via a nonmagnetic layer. When the directions of magnetization of ferromagnetic layers are parallel [Fig. 1.4 (a)], a magnetic moment of which is parallel to the direction of magnetizations can move without the scattering by the electrons taking part in the formation of magnetizations in the ferromagnetic layers. Thus, the electrical resistivity for the parallel directions of magnetizations becomes smaller than that for the antiparallel magnetizations. However, GMR effect is not always observed in all artificial lattice systems, and following at least three requirements must be met: spins of ferromagnetic layers via a neighboring nonmagnetic layer are antiparallel with each other, the period of artificial lattice is smaller than mean free path of conduction electron, and the difference of spin dependent electrical resistance is large.
1.2.3 Spin FET Spin polarized field effect transistor, spin FET, was proposed by Datta and Das as the first idea of a semiconductor spintronic device in 1990 [8]. The spin FET is thought to be one of the most advanced devices in spintronics in the future. The schematic diagrams of spin FET without and with a gate voltage are illustrated
16 (a) Without Gate Voltage
Drain Source Gate Ferromagnet or Ferromagnet or Half metal Half metal
Spin polarized current 2D electrons gas (2DEG)
(b) With Gate Voltage Drain Source Gate Ferromagnet Ferromagnet or or Half metal Half metal
2D electrons gas (2DEG)
Figure 1.5: Schematic illustrations of spin FET (a) without and (b) with gate voltage.
17 in Fig. 1.5. The spin FET device consists of source and drain electrodes composed of ferromagnets or half metals, a gate electrode, and a two dimensional electron gas channel (2DEG). It is basically similar to a heterostructure of conventional FET. When the direction of spins in the 2DEG channel is parallel to that of the ferromagnetic drain electrode, the spin polarized electrons can flow into the ferromagnetic drain electrode [Fig. 1.5 (a)]. However, when the direction of spins in the 2DEG is flipped through the spin orbit interaction modified by external electric field as illustrated in Fig. 1.5 (b), the electrons cannot enter the drain electrode. In other words, the drain current is modulated by making the direction of spins of injected electrons parallel or antiparallel to the magnetization of drain electrodes. The spin FET is more advantageous than a conventional FET because of the following facts. If perfect spin polarized electrons are injected, the current can be easily controlled by application of much lower gate voltage which is used for only flipping of spins. In addition, magnetic electrodes can be utilized as a nonvolatile memory component. In spite of its intriguing properties and excellent functions, spin FET operating at room temperature has not been achieved yet.
1.3 Magnetic Semiconductor
The semiconductor spintronics instead of spintronics based on metals is expected to establish next generation technology. An important element for the achievement of semiconductor spintronics is the development of a new ferromagnetic material, in particular, magnetic semiconductor. Magnetic semiconductors can be divided into three types of materials: traditional magnetic semiconductors [Fig. 1.6 (b)], diluted magnetic semiconductors (DMS) [Fig. 1.6 (c)], and electrically conducting magnets [Fig. 1.6 (d)].
18 (a) (b)
(c) (d)
: nonmagnetic ions
: magnetic ions : metal (nonmagnetic and magnetic) ions
Figure 1.6: Four types of semiconductors: (a) a conventional nonmagnetic semiconductor, (b) a traditional magnetic semiconductor, (c) a diluted magnetic semiconductor (DMS), and (d) an electrically conducting magnet.
19 1.3.1 Traditional magnetic semiconductor Eu chalcogenides (EuO, EuS, and EuTe) have been intensively studied as a representative magnetic semiconductor. Schematic illustration of a traditional magnetic semiconductor is shown in Fig. 1.6 (b). These chalcogenides exhibit electrical conduction which is related to magnetic properties such as negative magnetoresistance, a maximum of resistivity near Curie temperature, T C, and magnetic polaron [9–11]. Cr chalcogenide spinels are also known as a magnetic semiconductor.
For example, FeCr 2S4 [12] is a ferrimagnet with a semiconducting like resistivity, and shows relatively large magnetoresistance at the transition temperature between paramagnetic and ferrimagnetic states. In spite of the numerous studies, there are no practical applications of the chalcogenides, because their TC lower than room temperature and extreme difficulty in growing these crystals have hampered the application and further studies on these materials. Hence, they are not so ideal for spintronic applications.
1.3.2 Diluted magnetic semiconductor (DMS) One of the methods to achieve a ferromagnetic semiconductor is to incorporate a small amount of magnetic atoms into non magnetic semiconductors [Fig. 1.6 (a)]. As a result, a part of the lattice is made up of substituted magnetic atoms [Fig. 1.6 (c)]. Such materials are called diluted magnetic semiconductors (DMS). Mn is the most common magnetic atoms for doping, but also Fe and Cr are widely used. The advent of II VI based DMS in the 1980’s caused the intense studies on magnetic semiconductors. Typical examples of DMS are (Cd,Mn)Te, (Zn,Mn)S, and (Zn,Mn)Se. II VI based DMS is usually antiferromagnetic, and no ferromagnetism has been reported until recently [13]. Moreover, difficulty to introduce magnetic atoms into II VI compounds has hampered the study of transport properties of II VI based DMS, so that magnetic and magneto optical properties have been mainly
20 investigated [14]. On the other hand, in 1989, the first ferromagnetic III V compound semiconductor, that is InAs heavily doped with Mn, was fabricated by nonequilibrium growth using low temperature molecular beam epitaxy (LT MBE) [15]. In 1996, a new type of ferromagnetic III V based DMS, p type (Ga,Mn)As, was reported by Ohno et al. [16,17]. The reason why these III V based DMS have attracted considerable attention is attributed to the ‘carrier induced ferromagnetism’, in which ferromagnetic interaction mediated by conduction carries plays an important role. This interaction was confirmed by some experimental results such as a decrease of magnetization in carrier compensated (In,Mn)As using Sn as a donor [18], ferromagnetism induced by photogenerated carriers in (In,Mn)As/GaSb [19], and disappearance of an anomalous Hall effect [19,20]. Afterward, the TC of Mn doped GaAs has been increased until 173 K by technological progress in thin film growth process [21]. However, the commercial applications require higher TC beyond room temperature, and therefore, further research is still needed. Because it became obvious that it would be difficult to obtain room temperature ferromagnetism in Mn doped GaAs, other materials were explored. Theoretical calculations by Dietl et al. predicted that Mn doped GaN and ZnO could be ferromagnetic even at room temperature [22]. Indeed, many research groups have reported room temperature ferromagnetism in Mn doped GaN [23–24] and ZnO [25–28]. However, the ferromagnetism experimentally observed could not be explained by Dietl’s model, and the origin of the ferromagnetism remains unclear. Another discovery of room temperature ferromagnetic semiconductor, i.e.,
Co doped TiO 2 (anatase phase) had a great impact on this field [29], and created a new trend [28–31]. This material is transparent in the visible region and has high TC ferromagnetic properties with appreciable conductivity. Nonetheless, some different models have been proposed to explain the ferromagnetism involving ‘carrier induced ferromagnetism’ like III V based DMS, and thus the precise mechanism is uncertain.
21 1.3.3 Electrically conducting magnet Another method to achieve a ferromagnetic semiconductor is to provide magnet with semiconducting properties by incorporation of aliovalent elements [Fig. 1.6 (d)].
Typical examples are LaMnO 3 based system, ilmenite hematite system, and magnetite based system. The origin of magnetism of these compounds can be explained by the popular theories of magnetism, leading to easier device designs for spintronics. Furthermore, because the existence of high spin polarized carriers in the ferromagnetic semiconductors is predicted by theoretical calculations, these materials may be useful as a source to inject spin polarized carriers.
Perovskite type manganites based on LaMnO 3, La 0.7 A0.3 MnO 3 ( A = Ba, Ca, Sr, Pb) with TC in a range of 280–380 K [32], have attracted considerable attention since the discovery of a colossal magnetoresistance in thin films of La 0.7 Ca 0.3 MnO 3 [33]. It was revealed that the tunnel junction, La 0.67 Sr 0.33 MnO 3/SrTiO 3/La 0.67 Sr 0.33 MnO 3, indicates very large magnetoresistance effect of 1800% at 4 K, from which a spin polarization of 95% is deduced using the Julliere formula (Eq. (3)) [34]. This experimental result suggests that La 0.67 Sr 0.33 MnO 3 is a half metal. However, the spin polarization and half metallic properties of La 0.7 Sr 0.3 MnO 3 have been controversial due to the facts that the measured magnetic moment is smaller than ideal one, and that the origin of spin polarization is entirely different from the case of a conventional half metal [35]. Thus, although the CMR in manganites based on LaMnO 3 is fascinating, it is difficult to use perovskite type manganites as a semiconductor device in spintronics, owing to temperature dependence of CMR in addition to low TC.
Both FeTiO 3 and α Fe 2O3 are known as antiferromagnetic insulators, whereas their solid solutions exhibit both semiconducting and ferrimagnetic properties [36,37]. Furthermore, in this system the conduction type can be controlled as either p or n type by simply changing the composition [37]. Also, the possibility of half metallic character in the solid solutions has been predicted by theoretical calculations [38]. Therefore, this system is expected to become one of candidate materials for spintronics
22 applications, and the fabrication of FeTiO 3 Fe 2O3 solid solution thin films has been recently attempted by some research groups [39–43]. For example, Hojo et al. reported that the solid solution in the form of thin film showed anomalous Hall effect in addition to both semiconducting and ferrimagnetic properties [41–43]. However, at this moment, there exist no spintronics devices based on the solid solution, so that further detailed investigation is required. Magnetite based system, in which both electronic and magnetic properties can be tailored, is intriguing from a point of view of semiconductor spintronics. This system is described in detail in chapter 2.
1.4 Conclusion
The concept, history, applications, and materials of spintronics are briefly overviewed in this chapter. The progress of spintronics has been achieved by making use of ferromagnetic metals so far. However, a key material to realize future spintronics is a magnetic semiconductor. Among them, ferromagnetic oxide semiconductors have attracted considerable attention because a wide variety of properties derived from oxides can be utilized for spintronic applications. Nonetheless, there still exist several issues to be solved concerning various magnetic oxide semiconductors. Therefore, intensive studies including the exploitation of novel magnetic oxide semiconductors must be performed to attain spintronic devices operating at room temperature.
23 References in CHAPTER 1
[1] N. H. Mott, Proc. Roy. Soc. A 15 3, 699 (1936). [2] M. Julliere, Phys. Lett. 54A , 225 (1975). [3] M. Baibich, J. M. Broto, A. Fert, F. Nguyen Van Dau, F. Petroff, P. Eitenne, G. Creuzet, A. Friederich, and J. Chazelas, Phys. Rev. Lett. 61 , 2472 (1988). [4] G. Binasch, P. Grünberg, F. Saurenbach, and W. Zinn, Phys. Rev. B 39 , 4828(R) (1989). [5] Y. Li, C. L. Chien, Y. Tomioka, and Y. Tokura, Phys. Rev. B 66 , 012410 (2002). [6] J. M. de Teresa, A. Barthélémy, A. Fert, J. P. Contour, F. Montaigne, and P. Seneor, Science 286 , 507 (1999). [7] M. J. M. de Jong and C.W. Beenakker, Phys. Rev. Lett. 74 , 1657 (1995). [8] S. Datta and B. Das, Appl. Phys. Lett. 56 , 665 (1990). [9] S. von Molnàr and S. Mathfessel, J. Appl. Phys. 38 , 959 (1967). [10] T. Kasuya and A. Yanase, Rev. Mod, Phys. 40 , 684 (1968). [11] A. mauger and C, Godart, Phys. Rep. 141 , 51 (1986). [12] A. P. Ramirez, R. J. Cava, and J. Krajewski, Nature 386 , 156 (1997). [13] A. Haury, A. Wasiela, A. Arnoult, J. Cibert, S. Tatarenko, T. Dietl, and Y. Merle d’Aubigné, Phy. Rev. Lett. 79 , 511 (1997). [14] J. K. Furdyna, J. Appl. Phys. 64 , R29 (1988). [15] H. Munekata, H. Ohno, S. von Molnar, A. Segmuller, L. Chang and L. Esaki, Phys. Rev. Lett. 63 , 1849 (1989). [16] H. Ohno, A. Shen , F. Matsukura, A. Oiwa, A. Endo, S. Katsumoto and Y. Iye, Appl. Phys. Lett . 69 , 363 (1996). [17] H. Ohno, Science 281 , 951 (1998). [18] H. Munekata, T. Penny and L. L. Chang, Surf. Sci. 267 , 342 (1992). [19] S. Koshihara, A. Oiwa, M. Hirasawa, S. Katsumoto, Y. Iye, C. Urano, H. Takagi and H. Munekata, Phys. Rev. Lett. 78 , 4617 (1997).
24 [20] H. Ohno, Nature 408 , 944 (2000). [21] T. Jungwirth, K. Y. Wang, J. Mašek, K. W. Edmonds, Jürgen König, Jairo Sinova, M. Polini, N. A. Goncharuk, A. H. MacDonald, M. Sawicki, A. W. Rushforth, R. P. Campion, L. X. Zhao, C. T. Foxon, and B. L. Gallagher, Phys. Rev. B 72 , 165204 (2005). [22] T. Dietl, H. Ohno, F. Matsukura, J. Cibert and D.Ferrand, Science 287 ,1019 (2000). [23] M.Reed, N. El Masry, H. Stadelmaier,M. Ritums, M. Reed, C. Parker, J. Roberts and S. Bedair, Appl. Phys. Lett. 79, 3473 (2001). [24] S.Sonada, S. Shimizu, T. Sasaki, Y. Yamamoto and H. Hori, J. Crys. Growth 237-239, 1358 (2002). [25] K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79 , 988 (2001). [26] P. Sharma, A. Gupta, K. V. Rao, F. J. Owens, R. Sharma, R. Ahuja, J. M. O. Guillen, B. Johansson, and G. A. Gehring, Nature Mater. 2, 673 (2003). [27] S. W. Lim, M. C. Jeong, M. H. Ham and J. M. Myoung, Japan. J. Appl. Phys. 43 , L280 (2004). [28] R. Janisch, P. Gopal, and N. A. Spaldin, J. Phys.: Condens. Matter 17 , R657 (2005). [29] Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura, M. Kawasaki, P. Ahmet, T. Chikyow, S. Koshihara, and H. Koinuma, Science 291 , 854 (2001). [30] S. R. Shinde, S. B. Ogale, S. D. Sarma, J. R. Simpson, H. D. Drew, S. E. Lofland, C. Lanci, J. P. Buban, N. D. Browning, V. N. Kulkarni, J. Higgins, R. P. Sharma, R. L. Greene, and T. Venkatesan, Phys. Rev. B 67 , 115211 (2003). [31] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T. Hasegawa, H. Koinuma and M. Kawasaki, Nature Mater. 3, 221 (2004). [32] M. B. Salamon and M. Jaime, Rev. Mod, Phys. 73 , 583 (2001). [33] S. Jin, T. H. Tiefel, M. McCormack, R. A. Fastnacht, R. Ramesh, and L. H. Chen, Science 264 , 413 (1994).
25 [34] M. Bowen, M. Bibes, A. Barthélémy, J. P. Contour, A. Anane, Y. Lemaitre, and A. Fert, Appl. Phys. Lett. 82 , 233 (2003). [35] B. Nadgorny, I. I. Mazin, M. Osofsky, R. J. Soulen Jr., P. Broussard, R. M. Stroud, D. J. Singh, V. G. Harris, A. Arsenov, and Y. Mukovskii, Phys. Rev. B 63 , 184433 (2001). [36] Y. Ishikawa and S. Akimoto, J. Phys. Soc. Jpn. 12 , 1083 (1957). [37] Y. Ishikawa, J. Phys. Soc. Jpn. 13 , 37 (1958). [38] W. H. Butler, A. Bandyopadhyay, and R. Srinivasan, J. Appl. Phys. 93 , 7882 (2003). [39] F. Zhou, S. Kotru, and R.K. Pandey, Thin Solid Films 408 , 33 (2002). [40] T. Fujii, M. Kayano, Y. Takada, M. Nakanishi, and J. Takada, J. Magn. Magn. Mater. 272 , 2010 (2004). [41] H. Hojo, K. Fujita, K. Tanaka, and K. Hirao, Appl. Phys. Lett. 89 , 082509 (2006). [42] H. Hojo, K. Fujita, K. Tanaka, and K. Hirao, Appl. Phys. Lett. 89 , 142503 (2006). [43] H. Hojo, K. Fujita, K. Tanaka, and K. Hirao, J. Magn. Magn. Mater. , 310 , 2105 (2007).
26 CHAPTER 2 Magnetite-ulvöspinel solid solution
2.1. Introduction
Magnetite, Fe 3O4, is a promising material as a functional spintronics device which can operate efficiently at room temperature, because it has high Curie temperature
(TC ~ 860 K) and is theoretically predicted to exhibit high spin polarization [1–3].
However, it is difficult for Fe 3O4 to be utilized as a semiconductor, owing to the metallic band structure and the large amount of carrier concentration. Recently, some
2+ studies on the spinel type solid solutions based on Fe 3O4, Fe 3O4 Fe 2MO4 (M=Mn , Zn 2+ ), have indicated that their electrical and magnetic properties can be systematically tuned by adjusting the amount of doped metal ions [4–6]. In addition, the solid solutions exhibited anomalous Hall effect, confirming the spin polarization derived from half metallic properties of Fe 3O4. Thus, it is considered that the spinel type solid solutions systems are suitable materials for the realization of semiconductor spintronics.
The author has focused on the spinel type solid solutions between Fe 3O4 and Fe 2TiO 4 (ulvöspinel). In this chapter, the background necessary to understand the solid solutions based on
Fe 3O4 and the structural, electrical, and magnetic properties of (1–x)Fe 3O4 xFe 2TiO 4 solid solutions in bulk form are summarized. The future perspective of the solid solutions is presented as well.
27 2.2. Magnetite
Magnetite, Fe 3O4, is the first magnetic material discovered in the human history. It was known as a lodestone which could attract metallic iron more than 2500 years ago. Since then, the scientific field of ‘magnetism’ has evolved from a variety of studies carried out for Fe 3O4. For example, in a field of paleomagnetism, Fe 3O4 is important for understanding plate tectonics and useful as historic data for magnetohydrodynamics [7]. In spite of such a long period, Fe 3O4 has been still studied intensively, because it exhibits ferrimagnetic properties with a very high Curie temperature, high electrical conductivity, and high spin polarization.
2.2.1. Crystal structure
Spinel type compounds have the chemical formula AB 2O4, and can be observed in a number of naturally occurring minerals. The structure is cubic and belongs to the space group Fd3m . The unit cell contains 32 O 2– ions arranged in a face centered cubic packing. Within this fcc arrangement of O 2– ions, there are two types of cation sites. One is the tetrahedral (A) sites, which are made up of 8 sites per unit cell in tetrahedral coordination with four surrounding O 2– ions, and the other is octahedral (B) sites, which is composed of 16 sites per unit cell in octahedral coordination with six surrounding O 2– anions, as illustrated in Fig. 2.1. The spinel type structures are divided into two types of groups depending on the arrangement of cations in the A and B sites. In the normal spinel type structure, the 8 divalent cations occupy the A sites and the 16 trivalent cations occupy the B sites. In the inverse spinel type structure, half of the trivalent cations occupy the A sites, and the other half of trivalent cations and all divalent cations occupy the B sites. For Fe 3O4, the crystal structure is the inverse spinel type and there are 16 Fe 3+ ions and 8 Fe 2+ ions per unit cell, in which half of Fe 3+ ions occupy the A sites, and the other half of Fe 3+
28 octahedral site (B site)
tetrahedral site (A site)
Magnetite (Fe 3O4) 3+ 2+ 3+ [Fe ]A[Fe , Fe ]BO4 2 :Fe 3+ :Fe 2+ , Fe 3+ :O
Ulvöspinel (Fe 2TiO4) 2+ 2+ 4+ [Fe ]A[Fe , Ti ]BO4 2 :Fe 2+ :Fe 2+ , Ti 4+ :O
Magnetite ulvöspinel solid solution ((1 x)Fe 3O4⋅xFe 2TiO4) 2+ 3+ 2+ 3+ 4+ [Fe , Fe ]A[Fe , Fe , Ti ]BO4 2 :Fe 2+ , Fe 3+ :Fe 2+ , Fe 3+ , Ti 4+ :O
Figure 2.1: The crystal structure of magnetite, ulvöspinel, and magnetite ulvöspinel solid solution.
29 ions and all Fe 2+ ions occupy the B sites. Therefore, the general chemical formula can
3+ 2+ 3+ be written as [Fe ]A[Fe , Fe ]BO4.
2.2.2. Electrical properties
Fe 3O4 is known to exhibit relatively high electrical conductivity among the spinel type compounds. Schematic illustrations of the electronic configurations of iron ions (Fe 2+ and Fe 3+ ) and of the conduction mechanism are presented in Fig. 2.2. The Fe 2+ ion has an electronic configuration of [Ar]3d 6, while the Fe 3+ ion has [Ar]3d 5. It is the sixth paired electron in the d shell of Fe 2+ that hops onto the Fe 3+ ion. The conduction mechanism is electron hopping between Fe2+ and Fe 3+, namely, the thermally assisted transfer of the sixth down spin electron in the d shell of Fe 2+ at B site along the [110] direction.
2.2.3. Magnetic property
Fe 3O4 is well known as an archetypical ferrimagnetic compound with high Curie temperature, TC. The magnetic ordering was examined as an example to which Néel’s two sublattice model was applied [8]. In this model, it is assumed that the magnetic interaction (so called superexchange interaction) between the tetrahedral (A) and octahedral (B) iron ions is strongly negative and the interaction between the ions in the same sublattices (so called double exchange interaction) is weak and positive. These interactions induce such an arrangement that magnetic moments in sublattices A and B are antiparallel to each other, and consequently, the resultant magnetization is the difference between the A and B sublattice magnetizations. According to the ionic
3+ 3+ 2+ model, the magnetic moments of A (Fe ) and B (Fe /Fe ) sites are (A) = 5 B and
(B) = 9/2 B, respectively, so the net magnetic moment should be 4 B/f.u. The experimental value determined by Weiss and Forrer is 4.07 B/f.u. [9]. The good
30 Fe 2+ Fe 3+
eg
t2g
(I) Initial state
[Ar] [Ar]
Fe 2+ 3d 6 Fe 3+ 3d 5
(II) After one step of hopping
[Ar] [Ar]
Fe 3+ 3d 5 Fe 2+ 3d 6
O2– Fe 2+ Fe 3+ Ti 4+ Fe 2+ e [110]
Figure 2.2: Schematic illustrations of the electronic configuration of Fe 3+ and Fe 2+ ions and hopping conduction mechanism.
31 agreement between the theoretical magnetization predicted by the simple ionic model and the experimental value was considered as a proof of the validity of this model.
2.2.4. High spin polarization
The band structure of Fe 3O4 has been calculated by various methods such as the augmented plane wave (APW) method [1], the augmented spherical wave (ASW) method [10,11], and the linear muffin tin orbital method (LMTO) [3]. All of these calculations predicted that Fe 3O4 is a half metal with P = –100% [Fig. 2.3], indicating that the down spin electrons show conducting behavior while the up spin electrons exhibit insulating behavior. Namely, only down spin electrons exist at EF, indicating that the direction of the spin polarization is governed by the spin population of t2g orbital in the B site [Fig. 2.4]. On the other hand, experimental studies on the electronic structure of Fe 3O4 were mainly carried out by a photoelectron spectroscopy [12–15]. According to the experiments, the sign of the spin polarization is negative as suggested by the theoretical calculations, although the absolute value is less than –100%. Possible reasons for the difference between the calculated and the experimentally measured values of spin polarization were considered to be strain [3], nonstoichiometry [13], surface imperfections [15], and strong electron correlation effect [14]. In recent years, studies on the fabrication of a variety of heterostructures based on
Fe 3O4 and the magnetoresistance effect have been carried out. Ghosh et al. reported that a remarkable positive magnetoresistance was observed in Fe 3O4/SrTiO 3/
La 0.7 Sr 0.3 MnO 4 heterostructure below room temperature [16]. Also, Li et al. have fabricated the tunneling magnetic junctions only consisting of Fe 3O4 as ferromagnetic electrodes, i.e., Fe 3O4/MgO/Fe 3O4, although the TMR effect around the room temperature is much lower than would be expected for the bulk material [17].
32 EF
Figure 2.3: Density of states of up and down spins calculated for unstrained cubic Fe 3O4 at T = 0K [3]. A and B indicate Fe(A) and Fe(B) 3d dominant bands, respectively. The Fermi level is at zero energy.
33 Fe 2+
eg 4s t2g
E 3d F
2p
α β
Figure 2.4: Schematic illustration of density of states for magnetite with only β spin (down spin) electrons at EF.
34 2.3. Magnetite-ulvöspinel solid solution
The magnetite ulvöspinel solid solutions, (1–x)Fe 3O4 xFe 2TiO 4 (molar ratio), where x varies from 0 for magnetite to 1 for ulvöspinel, are one of the well known minerals along with the ilmenite hematite solid solutions, xFe 2TiO 3 (1–x)Fe 2O3. In a field of geophysics, the (1–x)Fe 3O4 xFe 2TiO 4 solid solutions have arrested considerable attention for a long time, because they are naturally occurring igneous rocks such as basalts and their characteristics of magnetization have provided a wealth of information on the magnetic history of the earth [7,18]. From a point of view of solid state physics,
4+ effect of doping with Ti ions on the physical properties of Fe 3O4 has been examined, and valuable information about its conduction mechanism has been deduced [19,20].
As for the electrical and magnetic properties of (1–x)Fe 3O4 xFe 2TiO 4 solid solutions, it is known that they exhibit both semiconducting and ferrimagnetic properties. Consequently, the solid solutions are one of the candidate materials for semiconductor spintronics.
2.3.1. Crystal structure
A series of (1–x)Fe 3O4 xFe 2TiO 4 solid solutions has inverse spinel structure with a face centered cubic close packed oxygen sublattice (space group: Fd3m ). The schematic illustrations of crystal structures for Fe 3O4, Fe 2TiO 4, and their solid solution are shown in Fig. 2.1. As described in section 2.2.1, Fe 3O4, one of the end members, consists of 16 Fe 3+ ions and 8 Fe 2+ ions per unit cell, in which half of Fe 3+ ions occupy the A sites, and the other half of Fe 3+ ions and all Fe 2+ ions are located at the B sites.
2+ 4+ In another end member, Fe 2TiO 4, there are 16 Fe ions and 8 Ti ions per unit cell, in which half of Fe 2+ ions occupy the A sites, and the other half of Fe 2+ ions and all Ti 4+ ions are located at the B sites. Therefore, for (1–x)Fe 3O4 xFe 2TiO 4 solid solutions, it is generally accepted that Ti 4+ ions substitute for Fe 3+ ions in the B sites according to the
35 reaction of 2Fe 3+ → Fe 2+ + Ti 4+ , in order to keep the charge neutrality. Most studies suggest that the fraction of B sites Ti 4+ ions can occupy is limited as indicated by magnetization measurements. Basically, four models are presented for cation ordering as follows. The first model proposed by Akimoto assumes an equal concentration of Fe 3+ ions in both A and B sites [21]:
A 2+ 3+ B 2+ 3+ 4+ (Fe 2+ xFe 1–x) (Fe Fe 1–xTi x)O 4 (0 ≤ x ≤ 1.0). The second model by Néel [22] and Chevallier et al. [23] assumes strong preference of Fe 3+ ions for the A sites for x ≤ 0.5:
A 3+ B 2+ 3+ 4+ (Fe ) (Fe 1+ xFe 1–2xTi x)O 4 (0 ≤ x ≤ 0.5),
A 2+ 3+ B 2+ 4+ (Fe 2x–1Fe 2–2x) (Fe 2–xTi x)O 4 (0.5 < x ≤ 1.0). The third model proposed by O’Reilly and Banerjee [24] is given by
A 3+ B 2+ 3+ 4+ (Fe ) (Fe 1+ xFe 1–2xTi x)O 4 (0 ≤ x ≤ 0.2),
A 2+ 3+ B 2+ 3+ 4+ (Fe x–0.2 Fe 1.2–x) (Fe 1.2 Fe 0.8–xTi x)O 4 (0.2 < x < 0.8),
A 2+ 3+ B 2+ 4+ (Fe 2x–1Fe 2–2x) (Fe 2–xTi x)O 4 (0.8 ≤ x ≤ 1.0), and the fourth model by Kakol et al. [25] is written as
A 3+ B 2+ 3+ 4+ (Fe ) (Fe 1+ xFe 1–2xTi x)O 4 (0 ≤ x ≤ 0.2),
A 2+ 3+ B 2+ 3+ 4+ (Fe 1.25 x–0.25 Fe 1.25–1.25 x) (Fe 1.25–0.25 xFe 0.75–0.75 xTi x)O 4 (0.2 < x ≤ 1). The latter two models were proposed on the basis of saturation magnetization measurements at 77 K. However, some studies argue that Ti 4+ ions may be located in the A sites [26,27], and thus the distribution of three types of cations (Fe 2+ , Fe 3+ , and Ti 4+ ) over A and B sites still remains as a controversial issue.
2.3.2. Electrical properties
The (1–x)Fe 3O4 xFe 2TiO 4 solid solutions except for Fe 3O4 ( x=0) exhibit semiconducting behavior and the conductivity decreases with increasing the amount of Ti 4+ ions as shown in Fig. 2.5 (a). Basically, the conduction mechanism is the thermally activated hopping at the B sites like Fe 3O4, and there are two types of
36 (a)
(b)
Figure 2.5: Temperature dependence of (a) electric conductivity (log σ) and (b) thermoelectric power for (1–x)Fe 3O4 xFe 2TiO 4 solid solution single crystals [28].
37 conduction mechanisms depending on temperature. In a high temperature range the relation between resistivity ( ρ) and temperature ( T) exhibits Arrhenius type behavior, i.e., thermally activated hopping [28]:
ρ = ρ0 exp(E g / k BT ) , (3.1)
where ρ0 , Eg, and kB are the preexponential term, the activation energy, and the
Boltzmann constant, respectively. The Eg estimated from the slope of the linear
–1 relationship between log ρ and T varies in a range from 0.05 ( x=0; Fe 3O4) to 0.24 eV
4+ (x=1; Fe 2TiO 4) depending on the content of Ti ions. In a low temperature range, the log ρ T–1 plot clearly shows deviation from the linear relationship. Instead, the following Mott formula ( T–1/4 law): [29]
4/1 ρ = ρ0 exp(T0 /T ) , (3.2)
where T0 is the Mott temperature, can describe the experimental results. In other words, variable range hopping (VRH) is observed in a low temperature range, suggesting that the charge carriers in the solid solutions are localized at low temperatures. Interestingly, the carrier type in bulk solid solutions can be controlled by simply changing the composition x [28]. The temperature dependence of the thermoelectric power for (1–x)Fe 3O4 xFe 2TiO 4 solid solution single crystals is in Fig. 2.5 (b). The carrier type for compositions 0 ≤ x ≤ 0.6 is n type, while for compositions 0.7 ≤ x ≤ 1.0, it is p type.
2.3.3. Magnetic properties
The (1–x)Fe 3O4 xFe 2TiO 4 solid solutions are ferrimagnetic, while the end members,
38 (a)
(b)
Figure 2.6: (a) Variation of saturation magnetization with composition x for various models of cation distributions [25], open and solid circles: ref. 24 and 25, dotted lines (1)–(3): ref. 21–23, solid line: data in ref. 25. (b) Temperature dependence of calculated Ms for (1–x)Fe 3O4 xFe 2TiO 4 solid solutions [30].
39 Fe 3O4 and Fe 2TiO 4, are ferrimagnetic and antiferromagnetic, respectively. The magnetic ordering of the solid solutions basically obeys the Néel’s two sublattice model similarly to Fe 3O4. However, the saturation magnetization is somewhat complicated when compared with Fe 3O4 as shown in Fig. 2.6 (a), because the cation distribution at the A and B sites in the spinel structure varies with x in the solid solutions as described in section 2.3.1. Lyberatos [30] calculated the temperature dependence of spontaneous magnetization, Ms, using the Néel’s molecular field theory, and revealed that for composions x > 0.14, the spontaneous magnetization increases initially with an increase in temperature and exhibits a maximum at a certain temperature below TC [Fig. 2.6 (b)]. This behavior is in good agreement with experimental results [25, 31].
Also, this system has relatively high TC because of the solid solutions based on
Fe 3O4. It was reported by Akimoto et al. [31] that the TC of the solid solutions is given by
TC = 858 – 656 x (K).
This indicates that the solid solutions are ferrimagnetic at room temperature up to the composition of approximately x=0.8. Also, Shmidbauer [32] experimentally indicated that the solid solution in the form of bulk polycrystal with compositions x=0.6 and 0.8 possess TC of 434 and 292 K, respectively.
40 2.4. Future perspective for application of magnetic semiconductors in spintronics
As mentioned above, the Fe 3O4 Fe 2TiO 4 solid solutions have the unique properties. For instance, the conduction type can be easily controlled by changing the chemical composition x; the compositions of x ≤ 0.6 show n type conduction, while p type conduction is observed for the compositions of x ≥ 0.7. Also, the solid solutions have superiority that TC of some compounds ( x < 0.8) is beyond room temperature regardless of the conduction type. In addition, it is anticipated that the solid solutions have high spin polarized carriers because the compounds are based on Fe 3O4. In spite of such intriguing properties, there are no reports on fabrication of the Fe 3O4 Fe 2TiO 4 solid solutions in the form of thin film. The fabrication of the solid solution thin films is of great importance for applications, and the physical properties presumably lead to novel semiconductor spintronics devices which efficiently operate above room temperature. Recently, a few studies have been reported on the fabrication of semiconductor spintronics devices. Ziese et al. have successfully fabricated Schottky barrier at the
Fe 3O4/Nb:SrTiO 3, in which Fe 3O4 acts as a semiconductor and Nb:SrTiO 3 substrate as a metal, and have demonstrated that the magnetoresistance depends on the bias current through the junction [33]. Moreover, they have proposed the measurement method to determine the transport spin polarization in Fe 3O4 using Schottky barrier formation.
On the other hand, Satoh et al. [34] have fabricated (Fe,Mn) 3O4/Nb:SrTiO 3 heterostructure and investigated its I–V characteristics. They have demonstrated that the Schottky diode behavior can be controlled by adjusting the composition of ferromagnetic Fe 3− xMn xO4 and the carrier concentration in semiconductor Nb:SrTiO 3 substrate. The effective barrier height is also tuned by controlling the Mott gap originating from the strong electron–electron correlation in Fe 3− xMn xO4 system.
Furthermore, the spin polarization of carriers in Fe 2.5Mn 0.5O4 estimated using
Fe 2.5Mn 0.5O4/Nb:SrTiO 3 Schottky barrier is P=0.7 at 200 K, suggesting that the solid
41 solutions also have high spin polarized carriers. All of the applications are based on the compounds in the form of thin film with n type carriers. So far, there have been no reports on Fe 3O4 based compounds in the form of a film with p type carriers at room temperature, and therefore it is expected that the fabrication of p type ferrimagnetic semiconductor based on Fe 3O4 Fe 2TiO 4 solid solution thin films will expand the possibility of semiconductor spintronics.
2.5. Conclusion
In this chapter, the structural, electrical, and magnetic properties of Fe 3O4 and
Fe 3O4 Fe 2TiO 4 solid solutions are described. These are of interest from a point of view of solid state chemistry and physics and are important information from a standpoint of practical applications in spintronics as well. The fact that the solid solutions ( x < 0.8) possess TC above room temperature regardless of the conduction type makes this system interesting for spintronic applications. The solid solutions are promising candidate materials as semiconductor spintronics devices.
42 References in CHAPTER 2
[1] A. Yanase and K. Siratori, J. Phys. Soc. Jpn. 53 , 312 (1984). [2] Z. Zhang and S. Satpathy, Phys. Rev. B 44 , 13319 (1991). [3] H. T. Jeng and G. Y. Guo, Phys. Rev. B 65 , 094429 (2002). [4] M. Ishikawa, H. Tanaka, and T. Kawai, Appl. Phys. Lett. 86 , 222504 (2005). [5] J. Takaobushi, H. Tanaka, and T. Kawai, S. Ueda, J. J. Kim, M. Kobata, E. Ikenaga, M. Yabashi, K. Kobayashi, Y. Nishino, D. Miwa, K. Tamasaku, and T. Ishikawa, Appl. Phys. Lett. 89 , 242507 (2006). [6] D. Venkateshvaran, M. Althammer, A. Nielsen, S. Geprägs, M. S. R. Rao, S. T. B. Goennenwein, M. Opel, and R, Gross, Phys. Rev. B 79 , 134405 (2009). [7] R. F. Butler, Paleomagnetism : magnetic domains to geologic terranes , Blackwell Scientific Publications, Boston (1992). [8] L. Néel Ann. Phys. 3, 137 (1948). [9] P. Weiss and R. Forrer, Ann. Phys. 12 , 279 (1929). [10] R. A. Groot and K. H. J. Buschow, J. Magn. Magn. Mater. 54-57 , 1377 (1986). [11] M. Pénicaud, B. Siberchicot, C. B. Sommers, and J. Kübler, J. Magn. Magn. Mater. 103 , 212 (1992). [12] S. F. Alvarado, W. E. F. Meier, D. T. Pierce, K. Sattler, H. C. Slegmann, and J. P. Remaika, Phys. Rev. Lett. 34 , 319 (1975). [13] Y. S. Dedkov, U. Rüdiger, and G. Güntherodt, Phys. Rev. B 65 , 064417 (2002). [14] D. J. Huang, C. F. Chang, J. Chen, L. H. Tjeng, A. D. Rata, W. P. Wu, S. C. Chung, H. J. Lin, T. Hibma, and C. T. Chen, J. Magn. Magn. Mater. 239 , 261 (2002). [15] E. Vescovo, H. J. Kim, J. M. Ablett, and S.A. Chambers, J. Appl. Phys. 98 , 084507 (2005). [16] K. Ghosh, S. B. Ogale, S. P. Pai, M. Robson, E. Li, I. Jin, Z. Dong, R. L. Greene, R. Ramesh, and T. Venkatesan, Appl. Phys. Lett. 73 , 689 (1998).
43 [17] X. W. Li, A. Gupta, G. Xiao, W. Qian, and V. P. Dravid, Appl. Phys. Lett. 73 , 3282 (1998). [18] A. F. Buddington and D. H. Lindsley, J. Petrology 5, 310 (1964). [19] F. Walz, J. Phys.: Condens. Matter 14 , R285 (2002). [20] F. Walz, V. A. M. Brabers, J. H. V. J. Brabers, and H. Kronmüller, J. Phys.: Condens. Matter 17 , 6763 (2005). [21] S. Akimoto, J. Geomagn. Geoel. 6, 1 (1954). [22] L. Néel, Adv. Phys. 4, 191 (1955). [23] K. Chévallier, J. Bolfa, and S. Mathieu, Bull. Soc. Fr. Mineral. Cristall. 78 , 307 (1955). [24] W. O’Reilly and S.K. Banerjee, Phys. Lett . 17 , 237 (1965). [25] Z. Kakol, J. Sabol, and J. M. Honig, Phys. Rev. B 43 , 649 (1991). J.Appl. Phys. 69 , 4822 (1991). [26] R. H. Forster and E. O. Hall, Acta Cryst. 18 , 857 (1965). [27] M. Z. Stout and P. Bayliss, Can, Miner. 18 , 339 (1980). [28] V. A. M Brabers, Physica B 205 , 143 (1994). [29] N. F. Mott, J. Non Crystal. Solids 1, 1 (1968). [30] A. Lyberatos, J. Magn. Magn. Mater. 311 , 560 (2007). [31] S. Akimoto, T. Katsura, and M. Yoshida, Adv. Phys. 9, 165 (1957). [32] E. SchmidBauer, and P.W. Readman, J. Magn. Magn. Mater. 27 , 114 (1982). [33] M. Ziese, U. Köhler, A. Bollero, R. Höhne, and P. Esquinazi, Phys. Rev. B 71 , 180406(R) (2005). [34] I. Saitoh, J. Takaobushi, H. Tanaka, and T. Kawai, Solid State Comm. 147 , 397 (2008).
44 CHAPTER 3 Synthesis of magnetite-ulvöspinel solid solution thin films
Section 3.1. Magnetite-ulvöspinel solid solution thin films on
ααα-Al 2O3(0001) substrates
3.1.1. Introduction
Recently, materials having high spin polarization of conducting charge carriers have attracted considerable attention due to their potential application in spintronics, a technology simultaneously utilizing both the charge and spin degrees of freedom of carriers [1]. In particular, magnetite (Fe 3O4) is a promising candidate as functional spintronics devices that operate efficiently at room temperature, because it has high
Curie temperature ( TC ~ 860 K) and is theoretically predicted to exhibit high spin polarization [2,3]. Fe 3O4 has inverse spinel type structure ( Fd3m ) with a face centered cubic (fcc) oxide ion sublattice, in which half of Fe3+ ions occupy the tetrahedral (A) sites, and the other half of Fe 3+ ions and all Fe 2+ ions are located at the octahedral (B) sites. The superexchange interaction between iron cations in the A and B sites plays the most important role in magnetic structure and transition of Fe 3O4, while the electrical conduction stems from the electron hopping between Fe 3+ and Fe 2+ ions in the
B sites. The device implementation such as magnetic tunnel junctions using Fe 3O4 as ferromagnetic electrodes has been demonstrated [4], although the tunnel magnetoresistance (TMR) effect near room temperature is much lower than would be expected for the bulk. Very recently, spinel type Fe 3O4 Fe 2MO4 solid solutions
45 (M=Mn 2+ , Zn 2+ ), where M 2+ ions preferentially occupy the A sites in the spinel structure, have been exploited for application in semiconductor spintronics [5–7].
Here the author has focused on the spinel type solid solutions of Fe 3O4 and Fe 2TiO 4
(ulvöspinel). According to previous studies, bulk specimens of (1–x)Fe 3O4⋅xFe 2TiO 4 (molar ratio) forms a complete series of solid solutions in a range of 0 ≤ x ≤ 1 [8–10]. The uniqueness of this solid solution system lies in the fact that the conduction type can be easily controlled by changing the chemical composition x; n type conduction is obtained for the compositions of x ≤ 0.6, while the compositions of x ≥ 0.7 show p type conduction [10]. In addition, the solid solution system has superiority that TC of some compounds ( x < 0.8) is above room temperature regardless of the conduction type [8]. In spite of such intriguing properties, there exist no reports on preparation of the
Fe 3O4 Fe 2TiO 4 solid solution in the form of thin films, to the best of our knowledge. In this section, 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin films have been prepared on sapphire substrates using a pulsed laser deposition (PLD) technique, and their structural, electrical, and magnetic properties have been also investigated.
3.1.2. Experimental procedures
3.1.2.1. Sample preparation
Thin films with 0.4Fe 3O4 0.6Fe 2TiO 4 composition (molar ratio) were grown on the
(0001) surface ( c plane) of sapphire ( α Al 2O3 single crystal) substrates by a PLD technique. A schematic diagram of PLD experiments in this study is shown in Fig. 3.1. The c sapphire substrates were annealed at 1000 ºC in air for 3 hours to obtain atomically flat surface. The target for PLD was prepared from reagent grade α Fe 2O3 and TiO 2 by the solid state reaction; first, the mixture of α Fe 2O3 and TiO 2 powders was sintered at 1200 ºC for 12 hours in air, and then heat treated in CO(40 %) CO 2(valance)
46 Substrate KrF excimer Heater laser
O2 Lens
Plume
Target
Figure 3.1: Schematic diagram of PLD experiment.
47 atmosphere at 800 ºC for 24 hours in order to convert some of Fe 3+ to Fe 2+ ions. The deposition chamber had a base pressure of ~10 –6 Pa, and the target to substrate distance was 3.5 cm. A focused KrF excimer laser with a wavelength of 248 nm and a pulse duration of 20 ns was used as a light source for PLD. The repetition frequency was 2 Hz, the laser fluence was about 1.3 J/cm 2, and the deposition rate was 0.67 nm/min.
To obtain optimal deposition conditions, oxygen partial pressure, PO2, and substrate –5 –3 temperature, TS, were varied from 1.0×10 to 1.0×10 Pa and from 500 to 700 ºC, respectively.
3.1.2.2. Characterization The thickness and chemical composition of thin film were determined by Rutherford backscattering spectroscopy (RBS) using 2.0 MeV He 2+ [Fig. 3.2]. The analysis of RBS spectra with SIMNRA simulation program revealed that the film thickness was 40 nm and that the ratio of Fe to Ti was typically 0.799 : 0.201, consistent with the value expected from the chemical composition of 0.4Fe 3O4 0.6Fe 2TiO 4. X ray diffraction (XRD) analysis with Cu Kα radiation (Rigaku, SLX2500K and ATX G) was performed in 2 θ/ω scans (out of plane), 2 θχ/φ scans (in plane), ω scans (rocking curve), and φ scans. Magnetization as a function of external magnetic field and temperature was measured by a superconducting quantum interference device (SQUID) magnetometer (Quantum design, MPMS). Temperature dependence of electric resistivity was explored by the van der Pauw method (Resitest8300, TOYO). Hall effect measurements were performed at room temperature for photolithographically patterned Hall bars using a physical property measurement system (Quantum design, PPMS). To cancel the effect of magnetic field on the longitudinal resistivity, the magnetic field was swept to a positive region and then to a negative region so that only the transverse resistivity, i.e., Hall resistivity, could be obtained by half the difference between the signals measured at the positive and negative fields. In order to determine the major
48 2.0 MeV He 2+ 1000 Experimental Fe Simulation (SIMNRA) 800 O 600
Al sub Counts 400
200 Ti
0 500 1000 1500 Energy (keV)
Figure 3.2: The RBS experimental spectrum of the solid solution of
the Fe 3O4 Fe 2TiO 4 solid solution thin film with that from simulation program SIMNRA.
49 conduction type in thin films, Seebeck coefficient was determined by using two R type thermocouples attached to both ends of the film, while utilizing a heater to induce a temperature difference of 2 K at room temperature.
3.1.3. Results and discussion
3.1.3.1. Structural Analysis Figure 3.3 (a) shows the XRD 2 θ/ω scan profiles of the thin films grown at –5 TS = 600 ºC and under varied PO2. The thin film grown under PO2 = 1.0×10 Pa can be identified as a single phase of (111) oriented spinel type Fe 3O4 Fe 2TiO 4 solid solution –4 without any impurity phases. When PO2 is increased to 1×10 Pa, another crystalline phase ascribed to the solid solution of ilmenite (FeTiO 3) and hematite (Fe 2O3) appears –3 in addition to the Fe 3O4 Fe 2TiO 4 solid solution. At PO2 = 1.0×10 Pa, the diffraction peaks due to the Fe 3O4 Fe 2TiO 4 solid solution disappear, and only diffraction peaks due to the FeTiO 3 Fe 2O3 solid solution are detected. Figure 3.3 (b) summarizes the change in crystalline phase with the deposition conditions ( TS and PO2). Both lower PO2 and 2+ higher TS, which promote the formation of Fe ions, are effective in stabilizing the spinel phase of Fe 3O4 Fe 2TiO 4 solid solution without the precipitation of FeTiO 3 Fe 2O3 solid solution. Figure 3.4 (a) is displayed the rocking curve for the 222 reflection of –5 the Fe 3O4 Fe 2TiO 4 solid solution thin film grown under PO2 = 1.0×10 Pa and at
TS = 600 ºC. A single peak is observed with a full width at half maximum ( ω) of 0.072 º. This value is typical for most of the films with single phase Fe 3O4 Fe 2TiO 4 –5 solid solution, except for the sample prepared under PO2 = 1.0×10 Pa and at TS = –5 500 ºC. For the thin film grown under PO2 = 1.0×10 Pa at TS = 600 ºC, the out of plane lattice parameter was calculated to be 0.8483 nm from the peak position of the 222 reflection shown in Fig. 3.3 (a). In the 2 θχ/φ scan measurements for the same film (Fig. 3.4 (b)), the in plane lattice parameter was estimated to be 0.8496 nm by
50 TS = 600 ºC (a) 0006 222 333 111 444 0009
PO2 (Pa) –5 1.0 ×10
–4 1.0 ×10 Log Intensity (arb.units) 0006 00012 –3 1.0 ×10 20 40 60 80 2θ (degree)
(b) 10 3
10 4
10 5 Oxygen (Pa) pressure 10 6 400 500 600 700 800 Substrate temperature ( oC)
Figure 3.3: (a) XRD 2 θ/ω scan of thin films grown under varied oxygen pressure
at 600 ºC. : Fe 3O4 Fe 2TiO 4 solid solution, ■: FeTiO 3 Fe 2O3 solid solution,
●: α Al 2O3 (substrate). (b) Variation in the crystalline phase with the deposition
conditions ( TS and PO2), : Fe 3O4 Fe 2TiO 4 solid solution with high crystallinity,
○: Fe 3O4 Fe 2TiO 4 solid solution with low crystallinity, : Coexistence of
Fe 3O4 Fe 2TiO 4 and FeTiO 3 Fe 2O3 solid solutions, ×: FeTiO 3 Fe 2O3 solid solution.
51 (a) ω=0.072 º Intensity (arb. units) Intensity
18.5 19 ω (degree)
(b) (111)
(220) 3300 440 Intensity(arb. units) 220
20 40 60 80 2θ(degree)
Figure 3.4: (a) The rocking curve of 222 reflection in the Fe 3O4 Fe 2TiO 4 solid solution thin film grown under 1.0×10 –5 Pa and at 600 ºC. (b) XRD
2θχ/φ scan of the thin film.
52 0.4Fe 3O4⋅0.6Fe 2TiO4 004
_ α Al 2O3 10 14
Intensity(arb. units) ×1/100
0 50 100 150 200 250 300 350 φ (degree)
Figure 3.5: φ scan of 004 reflection in the solid solution thin film grown under 1.0×10 –5 Pa and at 600 ºC . For comparison, φ scan
_ of 1014 reflection in sapphire substrate is also shown.
53 _ using the 220 reflection. The lattice parameters of the film are very close to that of bulk specimen, i.e., the PLD target (0.8489 nm). The small deviation between out of plane and in plane lattice parameters is presumably due to a substrate induced strain. To examine the in plane alignment for the thin films, XRD φ scan was carried out using the 004 reflection (2 θ = 42.7 º, ψ = 54.7 º) of the solid solution. A typical result is depicted in Fig. 3.5 for the thin film composed of the single phase grown under PO2 =
_ –5 1.0×10 Pa and at TS = 600 ºC. Also shown in the figure is the φ scan in the 1014 reflection (2 θ = 35.141 º, ψ = 38.236 º) of the c sapphire substrate. The (0001) surface of substrate has threefold symmetry, so that only three peaks are observed in the φ scan from 0 ° to 360 °. If the resultant film is a perfect single crystal, only three peaks should be detected at every 120 º in the φ scan due to the threefold symmetry relative to the [111] direction. For the present film, however, six peaks corresponding to the sixfold symmetry are observed, indicating that the thin film has two crystallographic domains turned by 60 º or 180 º with each other on the surface of c sapphire substrate. The formation of twinned in plane alignment is ascribable to the large lattice mismatch (10.7 %) between the solid solution thin film and the c sapphire substrate [11]. Based on the results shown in Figs. 3.3–3.5, the epitaxial relation can be evaluated to be;
Fe 3O4 Fe 2TiO 4 solid solution (111) || α Al 2O3 (0001) and Fe 3O4 Fe 2TiO 4 solid solution
_ _ _ [110] || α Al 2O3 [1100] and [1100].
3.1.3.2. Physical Properties Figure 3.6 shows the temperature dependence of magnetization, M, for the solid –5 solution thin film grown under PO2 = 1.0×10 Pa and at TS = 600 ºC. The measurements were performed under field cooling condition with an external magnetic field, H, of 8500 Oe applied parallel to the thin film surface. One can see that M exhibits a plateau at 5 to 300 K and begins to decrease above 300 K, but does not reach
54 ) 3 H = 8500 Oe 100 ) 3 150 100 300 K 50 50 0 H 50 100 150 Magnetization(emu/cm 40000 20000 0 20000 40000 Magnetic field (Oe) 0 Magnetization (emu/cm 100 200 300 400 Temperature (K)
Figure 3.6: Temperature dependence of magnetization, M, for the solid solution thin film measured under field cooling condition at H = 8500 Oe. Inset shows r oom temperature M H curve.
55 zero even at 400 K. This observation indicates that the solid solution thin film shows ferrimagnetism with TC above 400 K. The inset of Fig. 3.6 displays the in plane M H curve at room temperature for the solid solution thin film. The M H curve exhibits an obvious hysteresis loop with saturation magnetization of 0.81 B/mol at room temperature (1.0 B/mol at 5 K, not shown here). The saturation magnetization of the present thin film is smaller compared with the value of single crystal (1.5 B/mol at 77 K) [9]. This is generally explained by the presence of antiphase boundaries (APBs) where antiferromagnetic interactions are dominant in the films [12,13]. APBs are structural defects formed during the growth process, and are observed when growing
Fe 3O4 epitaxial thin films [14–16].
In Fig. 3.7 (a) is depicted the temperature dependence of electric resistivity, ρxx , for –5 the solid solution thin film grown under PO2 = 1.0×10 Pa and at TS = 600 ºC. A typical semiconducting behavior is observed between 77 and 325 K. The value of ρxx at 300 K is 0.87 cm, which is almost consistent with those of single crystals [10]. For the purpose of evaluating the conduction mechanism, two types of analyses were performed as shown Figs. 3.7 (b) and (c). In a high temperature range between 100
–1 and 325 K, the relation between logarithmic ρxx and reciprocal temperature ( T ) exhibits Arrhenius type behavior as shown in Fig. 3.7 (b):
ρ xx = ρ0 exp(E g / k BT ) , (3.1)
where ρ0 , Eg, and kB are the preexponential term, the activation energy, and the
Boltzmann constant, respectively. The Eg is estimated to be 0.055 eV from the slope
–1 of the line. On the other hand, the log ρxx T plot in a low temperature range obviously shows deviation from the linear relation. In a temperature range from 77 to 160 K, a linear relationship is observed between log ρ and T–1/4 as shown in Fig. 3.7 (c), following the Mott formula ( T–1/4 law): [17]
56 (a) 10 6 10 5 10 4 cm) 3
Ω 10
( 2
xx 10 ρ 10 1 10 0 10 1 100 200 300 Temperature (K)
(b) 10 6 10 5 10 4 cm) 3 Ω 10 ( 2 xx 10 ρ 10 1 10 0 10 1 2 4 6 8 10 12 14 1000/ T (K 1 )
(c) 10 6 10 5 10 4 cm) 10 3 Ω
( 2
xx 10 1 ρ 10 10 0 10 1 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 T 1/4 (K 1/4 )
Figure 3.7: (a) Temperature dependence of electric resistivity, ρxx , for
–1 –1/4 solid solution thin film. Dependence of log ρxx on (b) T and (c) T .
57 )
200 3 5 2 1 cm) Ω 5 0 100 (10 H cm)
ρ 1 Ω 2 0.4 0.2 0 0.2 0.4 5 μ 0 0H (T) 300 K 0
(10 H H
ρ 100 5
200 (emu/cm Magnetization 5 0 5 μ0H (T)
Figure 3.8: Magnetic field dependence of –ρH and magnetization (open circles) for solid solution thin film. The inset shows a magnified view of
magnetic field dependence of –ρH.
58 4/1 ρ xx = ρ0 exp(T0 /T ) , (3.2)
where T0 is the Mott temperature. This conducting behavior is derived from the variable range hopping (VRH). From these plots, the conduction behavior is found to change from the VRH to the simple thermal activated hopping at a critical temperature (141 K), suggesting that the charge carriers in the solid solution thin film localize at low temperatures. The Seebeck coefficient of the thin film was evaluated to be –6.0 V/K, which is coincident with the value of bulk sample as reported previously [10]. The negative Seebeck coefficient indicates that the charge carrier is an electron ( n type).
Figure 3.8 shows the magnetic field dependence of the Hall resistivity, ρH, measured –5 at room temperature for the thin film grown under PO2 = 1.0×10 Pa and at TS = 600 ºC. The magnetic field was applied perpendicular to the thin film surface. We found that
ρH is negative at room temperature (this is why –ρH is plotted in the ordinate of Fig. 3.8). The plot of out of plane M H curve is also illustrated in Fig. 3.8. The hysteresis behavior of out of plane M H curve is smaller than that of in plane shown in Fig. 3.6, suggesting the presence of easy axis of magnetization in the in plane direction. In ferromagnetic materials, the Hall resistivity is generally expressed as [18]
ρ H = RO 0 H + RA 0 M , (3.3)
where RO is the ordinary Hall coefficient, RA is the anomalous Hall coefficient, and 0 is the vacuum permeability. The first term of Eq. (3.3), proportional to H, describe the ordinary Hall effect (OHE), and the second term, in general much larger than the first one, denotes the anomalous Hall effect (AHE), which is proportional to M of the material. The OHE arises from the Lorentz forces acting on charge carriers. On the other hand, although the origin of AHE has been a controversial issue for decades [19–22], it has been considered that AHE in a ferromagnetic material provides a strong evidence of intrinsic ferromagnetism caused by spin polarized charge carriers that
59 mediate ferromagnetic exchange interaction between localized spins of magnetic ions distant from each other [22]. As shown in Fig. 3.8, ρ H exhibits a behavior similar to that of out of plane M H curve, especially in the low magnetic fields. The result suggests that 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film has spin polarized charge carriers at room temperature. From the slope of the line at high magnetic fields caused
–2 3 by OHE, we can obtain RO = –1.17×10 cm /C at room temperature, indicating n type conduction, consistent with the Seebeck effect measurement. The carrier density, n,
20 –3 determined by using the relation RO = 1 / en is 5.35×10 cm . The Hall mobility
2 –1 –1 derived from RO and conductivity (1/ ρxx ) is 0.01 cm V s . Substituting the saturation
3 3 magnetization (102 emu/cm ) into the second term of Eq. (3.3) yields RA = –2.48 cm /C at room temperature. The negative value of RA has been commonly observed in Fe 3O4
3 (RA = –0.25 cm /C) and related compounds [23]. In order to clarify the mechanism of
AHE in more detail, a relation between ρH and ρxx should be investigated in a systematic fashion as reported in literatures [24–26].
3.4. Conclusion
We have fabricated 40 nm thick 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin films with spinel type structure on c sapphire substrates by using a PLD technique, and examined their structural, electrical, and magnetic properties. The solid solution thin films are epitaxially grown on the sapphire substrate, although they have a preferential in plane orientation due to the large lattice mismatch between the thin film and the substrate.
The crystallographic relationship is Fe 3O4 Fe 2TiO 4 solid solution (111) || α Al 2O3
_ _ _ (0001) and Fe 3O4 Fe 2TiO 4 solid solution [110] || α Al 2O3 [1100] and [1100]. The solid solution thin film is ferrimagnetic with TC higher than 400 ºC, and shows n type conduction. The conduction behavior changes from the simple thermal activated
60 hopping to the VRH at a critical temperature (141 K) due to the localization of electrons at low temperatures. The presence of spin polarized charge carriers is suggested by the Hall effect measurement at room temperature.
61 References in Section 3.1.
[1] J. M. D. Coey and C. L. Chien, MRS Bull. 28 , 720 (2003). [2] A. Yanase and K. Siratori, J. Phys. Soc. Jpn. 53 , 312 (1984). [3] Z. Zhang and S. Satpathy, Phys. Rev. B 44 , 13319 (1991). [4] X. W. Li, A. Gupta, G. Xiao, W. Qian, and V. P. Dravid, Appl. Phys. Lett. 73 , 3282 (1998). [5] M. Ishikawa, H. Tanaka, and T. Kawai, Appl. Phys. Lett. 86 , 222504 (2005). [6] J. Takaobushi, H. Tanaka, and T. Kawai, S. Ueda, J. J. Kim, M. Kobata, E. Ikenaga, M. Yabashi, K. Kobayashi, Y. Nishino, D. Miwa, K. Tamasaku, and T. Ishikawa, Appl. Phys. Lett. 89 , 242507 (2006). [7] I. Saitoh, J. Takaobushi, H. Tanaka, T. Kawai, Solid State Comm. 147 , 397 (2008). [8] E. SchmidBauer, and P.W. Readman, J. Magn. Magn. Mater. 27 , 114 (1982). [9] Z. Kakol, J. Sabol, and J. M. Honig, Phys. Rev. B 43 , 649 (1991). [10] V.A.M Brabers, Physica B 205 , 143 (1994). [11] I. Yamaguchi, T. Terayama, T. Manabe, T. Tsuchiya, M. Sohma, T. Kumagai, and S. Mizuta, J.Solid State Chem. 163 239 (2002). [12] D. T. Margulies, F. T. Parker, F. E. Spada, R. S. Goldman, J. Li, R. Sinclair, and A. E. Berkowitz, Phys. Rev. B 53 , 9175 (1996). [13] D. T. Margulies, F. T. Parker, M. L. Rudee, F. E. Spada, J. N. Chapman, P. R. Aitchison, and A. E. Berkowitz, Phys. Rev. Lett. 79 , 5162 (1997). [14] F. C. Voogt, T. T. M. Palstra, L. Niesen, O. C. Rogojanu, M. A. James, and T. Hibma, Phys. Rev. B 57 , R8107 (1998). [15] W. Eerenstein, T. T. M. Palstra, T. hibma, and S. Celotto, Phys. Rev. B 66 , 201101(R) (2002). [16] A. V. Ramos, J. –B. Moussy, M. –J. Guittet, M. Gautier Soyer, M. Viret, C. Gatel, P. Bayle Guillemaud, and E. Snoeck, J. Appl. Phys. 100 , 103902 (2006). [17] N. F. Mott, J. Non Crystal. Solids 1, 1 (1968).
62 [18] C. L. Chien and C. R. Westgate, The Hall Effect and its Applications (Plenum, New York, 1980). [19] R. Karplus, and J. M. Luttinger Phys. Rev. 95 , 1154 (1954). [20] J. Smit, Physica 24 , 39 (1958). [21] L. Berger, Phys. Rev. B 2, 4559 (1970). [22] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chikyow, T. Hasegawa, H. Koinuma, and M. Kawasaki, Nature Mater. 3, 221 (2004). [23] D. Reisinger, P. Majewski, M. Opel, L. Alff, and R. Gross, Appl. Phys. Lett. 85 , 4980 (2004). [24] T. Miyasato, N. Abe, T. Fujii, A. Asamitsu, S. Onoda, Y. Onose, N. Nagaosa, and Y. Tokura, Phys. Rev. lett. 99 , 086602 (2007). [25] D. Venkateshvaran, W. Kaiser, A. Boger, M. Althammer, M. S. R. Rao, S. T. B. Goennenwein, M. Opel, and R. Gross, Phys. Rev. B 78 , 092405 (2008). [26] A. Fernandez Pacheco, J. M. De Teresa, J. Orna, L. Morellon, P. A. Algarabel, J. A. Pardo, and M. R. Ibarra, Phys. Rev. B 77 , 100403(R) (2008).
63 Section 3.2. Magnetite-ulvöspinel solid solution thin films on
MgO(100) substrates
3.2.1. Introduction
Development of magnetic oxide semiconductors is a subject of strongly growing interest mainly due to their potential applications in spintronics [1]. Among various oxides, intensive studies have been performed on magnetite (Fe 3O4) [2,3] and its related materials such as Fe 3O4 Fe 2MO4 ( M = Co, Mn, Zn, Ni) [4–7] because of their high spin polarization and high Curie temperature (typically TC > 300 K). In this study, our attention is focused on the (1–x)Fe 3O4 xFe 2TiO 4 (molar ratio: 0 < x < 1) solid solutions. The uniqueness of this solid solution system lies in the fact that the conduction type can be easily controlled by changing the chemical composition x; n type conduction is obtained for the compositions of x ≤ 0.6, while the compositions of x ≥ 0.7 show p type conduction [8]. The solid solution system has also superiority that for x < 0.8, TC is above room temperature regardless of the conduction type [9]. As mentioned in section 3.1, ferrimagnetic and semiconducting 0.4Fe3O4 0.6Fe 2TiO 4 solid solution thin films can be grown on the α Al 2O3(0001) substrates by a pulsed laser deposition (PLD) technique. Although the solid solution thin films were epitaxially grown on
α Al 2O3(0001) substrates, twin domain boundaries were formed, presumably due to the large lattice mismatch (10.7%) between the thin film and the substrate. In addition, the film surface was rough as a result of a three dimensional growth. For device implementation, it is highly desirable to produce thin films with a flat surface so that the heterostructures with a smooth interface can be obtained. A single crystalline MgO substrate has been frequently utilized for the cube on cube epitaxial growth of Fe 3O4 and the related solid solutions, which is due to the small lattice mismatch (typically < 1%) between the thin film and the substrate [10, 11]. In this section, the epitaxial growth of the (1–x)Fe 3O4 xFe 2TiO 4 ( x=0.6, 0.7) solid solution thin
64 films on MgO(100) substrates have been demonstrated by the PLD technique, and their magnetic and electrical properties of the resultant thin films are also presented.
3.2.1. Experimental Procedure
3.2.1.1. Sample preparation
Thin films based on magnetite ulvöspinel solid solutions, (1–x)Fe 3O4 xFe 2TiO 4 (x=0.6, 0.7), were grown on the MgO(100) substrates by a PLD technique [Fig. 3.9]. The target for PLD was prepared by the conventional solid state reaction; first, the mixture of reagent grade α Fe 2O3 and TiO 2 powders was sintered at 1200 ºC for 12 h in air, and then reduced in CO (40, 50% for x=0.6, 0.7, respectively) CO 2(valance) atmosphere at 800 ºC for 24 h in order to convert some of Fe 3+ to Fe 2+ ions. Laser pulses launched from a KrF excimer laser ( λ = 248 nm, 2 Hz) were focused on the resultant targets at a fluence of 3 J/cm 2. The film growth was performed under an –5 –3 oxygen partial pressure ( PO2) from 1.0×10 to 1.0×10 Pa and at a substrate temperature ( TS) of from 400 to 600 ºC.
3.2.1.2. Characterization An energy dispersive x ray spectrometry (EDX) (HORIBA, EMAX) revealed that the cation ratio of thin films was almost the same as that of target (Fe:Ti=2.4:0.6 and 2.3:0.7). The crystal structure was characterized by high resolution X ray diffraction (XRD) with Cu Kα radiation (Rigaku, SLX2500K and ATX G) was performed in 2 θ/ω scans (out of plane), 2 θχ/φ scans (in plane), and pole figure measurements. The surface morphology of films was observed by an atomic force microscope (AFM) (SII Nano Technology, SPI3800N). Magnetization was measured by using a superconducting quantum interference device (SQUID) magnetometer (Quantum design,
65 Plume Substrate
O2
Heater
Lens Target
KrF excimer laser
Figure 3.9: Schematic diagram of PLD experiment.
66 MPMS). The measurement of electric resistivity was carried out by the van der Pauw method. The major conduction type was determined through the measurement of Seebeck coefficient at room temperature (Resitest8300, TOYO). In order to determine the major conduction type in thin films, Seebeck coefficient was determined by using two R type thermocouples attached to both ends of the film, while utilizing a heater to induce a temperature difference of 2 K at room temperature.
3.2.2. Results and Discussion
3.2.2.1. Structural Analysis
Figure 3.10 (a) shows the 2 θ/ω (out of plane) XRD patterns of (1–x)Fe 3O4 xFe 2TiO 4 (x=0.6, 0.7) solid solution thin films. The diffraction peaks observed at around 2θ=42.4º and 92.5º are ascribed to the solid solution 400 and 800 reflections. The appearance of Pendellösung fringe patterns around the 400 and 800 reflections indicates the growth of a high quality thin film. By analyzing the fringe pattern, the both films thickness is evaluated to be about 60 nm. In the 2θχ/φ (in plane) XRD pattern [Fig. 3.10 (b)], two diffraction peaks assigned to 220 and 440 reflections due to the solid solution are observed without any impurity phases. A tetragonal distortion is epitaxially induced on MgO substrates because the lattice constant of solid solution single crystal (0.8472, 0.8490 nm for x=0.6, 0.7, respectively) [12] is just a little larger than twice the lattice constant of MgO (0.4213 nm ×2 = 0.8426 nm). The in plane lattice constant for both solid solution films is shortened (0.8428 nm) to match twice the lattice constant of MgO, while the out of plane lattice constants for x=0.6 and
0.7 are elongated to 0.8530 and 0.8541 nm, respectively, due to the compressive strain. To evaluate the in plane orientation relationship between the solid solution thin film for x=0.6 and the MgO(100) substrate, the pole figure measurements were further performed using the 222 reflection [Fig. 3.11]. Both pole figures of film and MgO
67 (a) 200S 400S 400F 800F
x=0.6
x=0.7 Log Intensity Log(arb. units) Intensity 40 41 42 43 44 45 9146 92 93 94 95 96 97 2θ (degree)
(b) 440F 220F
x=0.6
x=0.7 Log Intensity (arb. units) (arb. Intensity Log 20 40 60 80
2θχ (degree)
Figure 3.10: (a) XRD 2θ/ω scan (out of plane) of the solid solution thin film
grown on MgO(100) substrate. (b) XRD 2θχ/φ scan (in plane) of the thin film.
: Fe 3O4 Fe 2TiO 4 solid solution (F), ●: substrate (S), □: W L1.
68 0 (a) 2θ = 36.67º
α 90 270 20 40 60 80
β
180
0.4Fe 3O4 0.6Fe 2TiO 4
0 (b) 2θ = 78.58º
α 90 270 20 40 60 80
β
180 MgO substrate
Figure 3.11: Pole figures of the 222 reflection for (a) 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution film (upper) and (b) MgO substrate (lower).
69 (a) 5.53 m) Length Length ( 0 0 0.27 Height (nm) (C)
(b) 6.66 m) Length Length ( 0 0 0.38 Height (nm)
Figure 3.12: AFM images (5 ×5 m2) for (a) MgO(100) substrate and (b) the
0.4Fe 3O4 0.6Fe 2TiO 4 solid solution film grown on MgO(100) substrate (left) and the cross sectional profile along the solid line in the AFM image (right).
70 substrate exhibit only four spots at every 90º corresponding to the fourfold symmetry. Namely, the atomic arrangement in the [111] direction exactly matches between the film and the substrate. This is due to the small lattice mismatch (0.68%) between the film and the substrate. In contrast to the case of using the α Al 2O3(0001) substrate, the cube on cube epitaxy can be achieved by the use of MgO substrate; the solid solution thin film is epitaxially grown on MgO(100) without formation of twin domain boundaries. The orientation relationship between the film and the substrate is as follows; 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution(100)[111] || MgO(100)[111]. Figure 3.12 (a) and (b) depict a typical AFM image (5×5 m2) for the MgO(100) substrate and the 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film, respectively. The flat and smooth surface morphology can be observed without pits and islands (the root mean squared (RMS) roughness is ~0.07 nm), reflecting the information of MgO substrates (RMS is ~0.05 nm). The generation of smooth surface strongly suggests the epitaxial growth in a layer by layer mode.
3.2.3.2. Physical Properties Figure 3.13 shows the temperature dependence of magnetization, M(T), for the
(1–x)Fe 3O4 xFe 2TiO 4 ( x=0.6, 0.7) solid solution thin films. The measurements were performed under a field cooling condition at an external magnetic field (H) of 15000 Oe applied along the in plane direction of the film. Strong M can be obtained in the temperature region below 400 K, indicating that the solid solution film is ferrimagnetic with a TC above 300 K. A close look at Fig. 3.13 reveals a maximum in the M T curve around 150 K ( x=0.6) and 220 K ( x=0.7), respectively. According to the Néel’s two sublattice model of ferrimagnetism [13], the observed behavior may be caused by a different T dependence of M between two sublattices (in the present case, the tetrahedral and octahedral sites in the spinel structure). The M H curve as depicted in the inset of Fig. 3.13 clearly displays the ferrimagnetic behavior at room temperature;
71 400 ) 3 )
3 H = 15000 Oe 100 300 K
0 300 H x=0.7
100 x=0.6 Magnetization (emu/cm Magnetization 40000 20000 0 20000 40000 200 Magnetic field (Oe)
100 x=0.6
Magnetization (emu/cm Magnetization x=0.7 0 100 200 300 400 Temperature (K)
Figure 3.13: Temperature dependence of M for the
(1–x)Fe 3O4 xFe 2TiO 4 ( x=0.6, 0.7) solid solution thin films measured under a field cooling condition at H = 15000 Oe. Inset shows the M H curve at 300 K.
72 4 10 4 10 10 3 cm) 2
3 Ω 10 ( 10 1 ρ 10 cm) 10 2 10 0 1 Ω 10 2 4 6 8 10 12 14 ( 10 1 1000/ T (K 1 ) ρ 10 0 10 1 100 200 300 Temperature (K)
Figure 3.14: Temperature dependence of log ρ for the
0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film. Inset shows log ρ as a function of T –1.
73 for (1–x)Fe 3O4 xFe 2TiO 4 ( x=0.6, 0.7) solid solution thin films, the coercive forces are 30
3 Oe, and the saturation magnetizations are 160 (1.3 B/f.u.) and 54 (0.5 B/f.u.) emu/cm , respectively. It is worth noting that the saturation magnetization for the
0.4Fe 3O4 0.6Fe 2TiO 4 solid solution films grown on MgO(100) is significantly increased compared to that for the films grown on α Al 2O3(0001) (0.8 B/f.u.), presumably because the use of substrate with a smaller lattice mismatch leads to the improvement in crystallinity.
The temperature dependence of electric resistivity, ρ(T), for the 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film is shown in Fig. 3.14. The ρ T curve exhibits a typical semiconducting behavior between 77 and 325 K. The value of ρ at 295 K is 0.52 cm, comparable to that of bulk single crystal [8]. It should be noted that the ρ value for the films grown on MgO(100) is smaller than that for the film grown on α Al 2O3(0001) (0.87 cm at 295 K), which again confirms an improvement in the quality of films. The conduction behavior in a high temperature range obeys a thermal activated hopping of the Arrhenius type behavior, ρ=ρ0exp( Eg/kBT), where ρ0, Eg, and kB are the preexponential term, the activation energy, and the Boltzmann constant, respectively.
The Eg estimated from the slope of the linear part is 0.045 eV [see the inset in Fig. 3.14]. The log ρ vs. T –1 plot at low temperatures deviates from the linear relationship.
1/4 Instead, the ρ T curve follows the Mott formula, ρ =ρ0exp( T0/T) , meaning a variable range hopping (VRH). The change in the conduction behavior from the thermal activated hopping to the VRH is due to the localization of electrons, which is also observed for the film grown on α Al 2O3(0001) as well as bulk single crystal. The Seebeck coefficient is evaluated to be –32 V/K, indicating that the charge carrier is an electron ( n type).
74 3.4. Conclusion
We have fabricated spinel type (1–x)Fe 3O4 xFe 2TiO 4 ( x=0.6, 0.7) solid solution thin films on MgO(100) substrates by a PLD technique. The solid solution thin films are epitaxially grown with cube on cube relationship on the MgO(100) substrate and possess the flat and smooth surface, due to the small lattice mismatch between the film and the substrate. The crystallographic relationship is Fe 3O4 Fe 2TiO 4 solid solution (100)[111] || MgO(100)[111]. We also have demonstrated that the
0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film grown on MgO(100) substrates exhibits the ferrimagnetic behavior with TC above 400 K and is an n type semiconductor and that their physical properties are improved compared to that of films grown on
α Al 2O3(0001). As temperature decreases, the conduction behavior was observed to change from the thermal activated hopping to the VRH due to the localization of conduction carriers.
75 References in section 3.2.
[1] J. M. D. Coey and C. L. Chien, MRS Bull. 28 , 720 (2003). [2] A. Yanase and K. Siratori, J. Phys. Soc. Jpn. 53 , 312 (1984). [3] Z. Zhang and S. Satpathy, Phys. Rev. B 44 , 13319 (1991). [4] D. Tripathy, A. O. Adeyeye, C. B. Boothroyd, and S. N. Piramanayagam, J. Appl. Phys. 101 , 013904 (2007). [5] M. Ishikawa, H. Tanaka, and T. Kawai, Appl. Phys. Lett. 86 , 222504 (2005). [6] J. Takaobushi, H. Tanaka, and T. Kawai, S. Ueda, J. J. Kim, M. Kobata, E. Ikenaga, M. Yabashi, K. Kobayashi, Y. Nishino, D. Miwa, K. Tamasaku, and T. Ishikawa, Appl. Phys. Lett. 89 , 242507 (2006). [7] U. Lüders, M. Bibes, J. F. Bobo, M. Cantoni, R. Bertacco, and J. Fontcubetra, Phys. Rev. B 71 , 134419 (2005). [8] V. A. M. Brabers, Physica B 205 , 143 (1994). [9] E. SchmidBauer, and P. W. Readman, J. Magn. Magn. Mater. 27 , 114 (1982). [10] D. T. Margulies, F. T. Parker, F. E. Spada, R. S. Goldman, J. Li, R. Sinclair, and A. E. Berkowitz, Phys. Rev. B 53 , 9175 (1996). [11] A. Bollero, M. Ziese, R. Höhne, H. C. Semmelhack, U. Köhler, A. Setzer, and P. Esquinazi, J. Magn. Magn. Mater. 27 , 114 (1982). [12] F. Bosi, U. Hålenius, and H. Skogby, Am. Miner. 94 , 181 (2009). [13] A. H. Morrish, The Physical Principles of Magnetism ,Wiley (1965).
76 CHAPTER 4 Electronic and magnetic structure of magnetite-ulvöspinel solid solution thin films
4.1. Introduction
Magnetic oxide semiconductors composed of transition metals have attracted much attention as key materials to realize semiconductor spintronics [1,2]. Spinel type magnetite ulvöspinel, (1–x)Fe 3O4 xFe 2TiO 4 (molar ratio: 0 < x < 1), solid solutions are as a promising material for functional semiconductor spintronics devices efficiently working at room temperature, because of a room temperature ferrimagnetic character [3] and a spin polarization. Also, the uniqueness of this solid solution system lies in the fact that the conduction type (either n or p type) can be controlled depending on the chemical composition x [4]. Such intriguing properties are mainly caused by the cation distribution of Fe and Ti ions at the tetrahedral (A) and octahedral (B) sites in the spinel structure. So far, the cation distribution in the solid solution has been proposed by most studies based on magnetic investigation for Fe ions under restriction that Ti 4+ ions exist at the B sites [5,6]. However, some studies have suggested that Ti 4+ ions might be located at the A sites [7,8], and the distribution of three kinds of cations (Fe 2+ , Fe 3+ , and Ti 4+ ) over A and B sites has still remained a controversial issue. Thus, this clarification is critical to arrange material designs and to develop semiconductor spintronics devices. X ray absorption spectroscopy (XAS) and x ray magnetic circular dichroism
77 (XMCD) are a powerful element selective method to clarify electronic and magnetic structure directly, and are also very useful to determine the oxidation state and the occupation site in magnetic materials [9,10].
In this chapter, Fe L2,3 edge and Ti L2,3 edge XAS and XMCD measurements have been preformed, and the electronic and magnetic structures of the solid solutions are discussed by comparison of the experimental results with theoretical ones obtained by first principles multi electron calculation based on density functional theory (DFT) and configuration interaction (CI) method.
4.2. X-ray absorption spectroscopy and x-ray magnetic circular dichroism
4.2.1. X-ray absorption spectroscopy (XAS) X ray absorption spectroscopy, XAS, is an experimental technique to investigate the local geometry and/or the electronic structure of materials. XAS spectra are obtained by measuring an energy variation of the absorption intensity arisen from excitation of a core level electron to conduction and unfilled valence bands. This electron transition follows the dipole selection rule, meaning electric dipole allowed transitions of azimuth state number to unfilled orbitals (i.e. l = ±1). Also, every element has the different core level energy. Therefore, only a specific electron transition of a targeted element in materials can be investigated by tuning x ray energy needed to the excitation of a core level electron.
In the L2,3 edge absorption spectroscopy, the absorption spectrum is dominated by dipole transition from the core 2 p level to the empty 3 d states, and because of the large Coulomb interaction between these levels, it depends on the local electronic structure.
Thus, analysis of the L 2,3 edge absorption structure may provide information about the oxidation state and the symmetry of the 3 d metal cations.
78 4.2.2. X-ray magnetic circular dichroism (XMCD) X ray magnetic circular dichroism (XMCD) provides valuable information on the electronic and magnetic structures. XMCD is defined as the difference of absorption intensities between XAS spectra taken with right and left circularly polarized x rays. It is simply explained in the following section by a two step model based on the transition of electrons from a 2 p core state to a 3 d valence state [Fig. 4.1] [11]. The first step is the excitation of electrons from the 2 p core state by circularly polarized x rays. When the right or left circularly polarized x rays enter into the sample, a 2 p core hole is formed by circularly polarized absorption, and in consequence, the spin orbit coupling is created. The spin orbit coupling results in the 2 p core state splits into two levels of 2p3/2 ( L3 edge) and 2p1/2 ( L2 edge), and the coupling between r spin ( s) and angular ( l) momenta on these levels are opposite: l+s and l–s. Since l r and s are parallel for the 2p3/2 initial state, transfer of the angular momentum +h by right circularly polarized x ray excites more up spin than down spin electrons. The r r reverse holds for left circularly polarized x ray. On the other hand, l and s are antiparallel for the 2 p1/2 initial state, and therefore right and left circularly polarized x ray preferentially excite down and up spin electrons, respectively. The second step is the detection of excited and polarized electrons by the imbalanced unoccupied up and down spin states of 3d character. Here, the spin split valence shell acts as a “detector” for the spin and/or orbital momentum of the excited photoelectron. The quantization axis of the detector is given by the magnetization direction which needs to be aligned with the photon spin direction. If the material is ferromagnetic, an imbalance in empty up and down spin states exists, and hence the valence shell acts as a “spin detector”. Similarly, the valence shell acts as an “orbital momentum detector”, if there is an imbalance of states with different magnetic quantum numbers ml. Note that L3 and L2 edge have opposite sign, reflecting the opposite spin orbit coupling of the 2 p3/2 and 2 p1/2 levels.
79 E
2nd step
EF
Right circularly Left circularly polarized x ray polarized x ray
1st step 2p3/2
2p1/2
Figure 4.1: Schematic illustration of the two step model for XMCD.
80 4.3. A first-principle multiplet approach
The charge transfer multiplet method is the most prevalent theoretical approach for the analysis of TM L2,3 edge XAS. Although this has been successfully reproduced many experimental spectra, it cannot predict multiplet structure a priori , because of the use of the adjustable parameters. So, the Tanaka’s group has developed the hybrid method of a first principle calculation based on density functional theory (DFT) and configuration interaction (CI) method. In this approach, experimental spectra from a variety of compounds having different d electron numbers and coordination numbers can be successfully reproduced without any empirical parameters [12–14]. As the first step of an a first principle CI calculation, a relativistic molecular orbital (MO) calculation is carried out using a model cluster composed of a TM ion and neighboring ligand ions within the local density approximation (LDA). The cluster is embedded into a mesh of point charges with formal valences, in order to take into account the effective Madelung potential. Thus, the spatial distributions and the energy levels of MOs depend on the crystalline structure and the covalency between the TM ion and ligands. Electronic correlations among 3 d electrons and a 2 p hole are rigorously calculated by taking the Slater determinants made by the DFT MOs. Furthermore, electronic correlations among particularly important orbitals are also taken into account by the CI scheme, while correlations among other orbitals are approximately treated within the framework of DFT. Therefore, this method can be applied to TM compounds with arbitrary atomic arrangement and symmetry. The charge transfer (CT) from ligands to the TM ions plays an important role in the spectral shapes of the TM L2,3 edge XAS of some 3 d TM compounds. In principle, two or more ligand metal CT channels or metal ligand CT can be included by taking the corresponding electronic configurations in the DFT CI method. Also, the metal–metal CT can be included if a cluster model including multiple TM ions is used and take the additional electronic configuration corresponding to the metal–metal CT in the CI.
81 However, at this moment, this approach is only available for a small number of systems since, in most cases, the number of Slater determinants, i.e. the size of the Hamiltonian matrix, becomes too large to solve with the present computational power.
4.4. Experimental procedure
4.4.1. XAS and XMCD measurements
The Fe 3O4 and the solid solution thin film with 0.4Fe 3O4 0.6Fe 2TiO 4 composition thin film were epitaxially grown on the MgO(100) substrates using a PLD method. The oxygen partial pressure was set to 1.0×10 –4 and 1.0×10 –5 Pa, and the substrate temperature was 400 and 500 ºC during the deposition. In order to prevent a surface deterioration, in situ Au capping (~1.5 nm) was performed after the thin films fabrication. X ray absorption intensities (XAS) were recorded through a total electron yield method at the twin helical undulaters soft x ray beam line BL25SU of Spring 8. The experimental configuration for XAS and XMCD measurements is shown in Fig. 4.2. XMCD spectra were obtained using a helicity switching technique (1 Hz) with right and left circularly polarized x rays ( p = ±0.96) by taking the difference of x ray absorption spectra from each polarization at 150 K. In addition, the XMCD intensity was obtained by averaging the recorded values in an external magnetic field of ±1.9 T in order to minimize artifacts. The directions of the applied magnetic field was aligned with the x rays propagation direction, i.e., along the helicity vector. The angle between x rays and the surface normal of the samples was set to 70 degrees.
82
bias voltage liquid He
A – e– e and 1.9 T
X ray
magnet 1.9 T
Au 70 deg.
MgO(100)
(1–x)Fe 3O4 x Fe 2TiO 4
Figure 4.2: Schematic illustrations of experimental configuration for XAS and XMCD.
83 4.4.2. Computational procedure
In the present study, first principles multi electron calculations for Fe 3O4 Fe 2TiO 4 solid solution were made for two kinds of clusters, i.e., MO 4 and MO 6, meaning A site and B site respectively. The total number of electrons in the cluster was obtained from formal charges. Possible valence states in this system structure are +3 and +2 for Fe, and +4 and +3 for Ti, respectively. Therefore, the clusters can be expressed as
5– 6– 4– 9– FeO 4 for Fe(III), FeO 4 for Fe(II), and TiO 4 for Ti(IV) in A sites, while FeO 6 for
10– 8– 9– Fe(III), FeO 6 for Fe(II), TiO 6 for Ti(IV) and TiO 6 for Ti(III) in B sites. Because of the difficulty in constructing proper model for the crystal structure of Fe 3O4 Fe 2TiO 4 solid solution, atomic positions including the lattice constants were obtained from the experimentally reported values of the end member Fe2TiO 4 [15] without lattice relaxation due to substitution of Ti 4+ ions. The cluster is embedded into a mesh of point charges with formal valences, placed at the atomic sites of 150,000 around clusters, in order to take account of effective Madelung potential. The valence number in A sites is set at +2, while that in B sites is set +3 determined by averaging +4 for Ti and +2 for Fe in view of random distribution of cations in B sites. The incident angle of x ray was set to the surface normal of a thin film, i.e. tilted at 90 degrees to the surface, probably due to no dependence on azimuth direction. For the XMCD calculation, magnetic field of 50, 100, and 200 T was introduced to induce a Zeeman splitting, and the temperature was set at 150 K. The direction of the magnetic field was aligned with that of the incident x ray. First, fully relativistic molecular orbital calculations were carried out by solving Dirac equations with the local density approximation (LDA) using the code that was originally described in Ref. 16. In this code, four component relativistic molecular orbitals (MO) are expressed as linear combination of atomic orbitals (LCAO). The numerically generated four component relativistic atomic orbitals (1 s 4p for TM and 1s 2p for O) were used as basis functions for MO. After the one electron calculations of relativistic MO, multi electron calculations
84 are made. Slater determinants were constructed using the relativistic MO as the components. Multi electron eigenstates can be computed by diagonalizing the multi electron Hamiltonian. Multi electron wave functions were formed by linear combination of Slater determinants as given by
M Ψ i = ∑CipΦp , (1) p=1
where Ψi is the ith multi electron wave function, Φp is the pth Slater determinant, and
Cip is the coefficient. The maximum number of Slater determinants, M, is LCn where L is the number of MO for the cluster and n is the number of electrons. In the present study, calculations have been done using the set of MO for not only TM 2p and TM 3d orbitals but also O 2p orbitals forming valence bands. The exchange and correlations not only among TM 2p and TM 3d, but also those among O 2p can be explicitly included in this approach. The oscillator strength of the electric dipole transition averaged over all directions is given by
2 2 n Iif = E( f − Ei ) Ψi ∑rk Ψ f , (2) 3 k=1
where Ψi and Ψf are multi electron wave functions for the initial state and the final state, while Ei and Ef are their energies. The electric dipole transition from initial state to all excited states has been considered. Selection rule for the dipole transition was not considered explicitly. However, the oscillator strength for some final states became zero when they are forbidden. The quadrupole or higher order transitions were not included. Theoretical spectrum was made by broadening the oscillator strengths using Lorentzian functions with FWHM = 0.6 eV.
85 This procedure was pointed out to overestimate the absolute transition energy typically by less than 1% [17]. In Ref. [17], the energy was corrected by taking orbital energy difference between single electron orbitals for the Slater’s transition state as a reference. The same procedure is adopted in the present study. In the present calculation, the energy scales between the experimental spectra and each theoretical spectrum slightly have an error (less than 1 eV), because the structural relaxation induced by the substitution of Ti ions is ignored.
4.5. Results and Discussions
Figure 4.3 shows Fe L2,3 edge XAS spectra of the end member Fe 3O4 and
2+ 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film, together with those of FeO (Fe ) and
3+ α Fe 2O3 (Fe ) as the references to guide valence states of Fe ions [18]. The energy scales of each reference spectrum are slightly translated (less than 1 eV) with respect to the peak positions of experimental spectra. The line shapes of Fe L3 edge XANES spectra of Fe 3O4 and the solid solution indicates characteristic two peaks at around the transition energy of 708.5 eV and 710 eV, and those of both Fe 3O4 and the solid solution
3+ 2+ are similar to that of α Fe 2O3 (Fe ), but different from that of FeO (Fe ), reflecting that Fe ions are in the mixed valence state of Fe 2+ and Fe 3+ . However, it is noteworthy that the intensity of A/B in the solid solution and Fe 3O4 is higher compared to α Fe 2O3, suggesting existence of more Fe 2+ ions.
Figure 4.4 shows the Fe L2,3 edge XAS experimental spectra of Fe 3O4 and the solid solution thin films, together with the theoretical spectra obtained from a first principle calculation based on the DFT CI method. The four calculated spectra represent
3+ 3+ 2+ 2+ Fe (Td), Fe (Oh), Fe (Td), and Fe (Oh). The peak denoted as A at L3 egde in the
3+ 2+ 2+ experimental spectra is produced by adding up Fe (Oh), Fe (Td), and Fe (Oh), while
3+ 3+ the peak denoted as B is composed of Fe (Td) and Fe (Oh). One can see that the peak
86 B A
0.4Fe 3O4⋅0.6Fe 2TiO 4
Fe 3O4
α Fe 2O3 Intensity (arb. units) Intensity FeO
705 710 715 720 725 730 Photon energy (eV)
Figure 4.3: Comparison of Fe L2,3 edge XAS spectra of Fe 3O4 and
2+ 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film with those of FeO (Fe )
3+ and α Fe 2O3 (Fe ) as the references.
87 B A 0.4Fe 3O4⋅0.6Fe 2TiO 4
Fe 3O4
3+ Fe (Td)
3+ Fe (Oh)
Intensity (arb. units) (arb. units) Intensity 2+ Fe (Td)
2+ Fe (Oh) 705 710 715 720 725 730 Photon energy (eV)
Figure 4.4: Comparison of the experimental Fe L2,3 edge XAS spectra of Fe 3O4 and 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film with the
3+ 3+ 2+ 2+ theoretical spectra of Fe (Td), Fe (Oh), Fe (Td), and Fe (Oh).
88 0.4Fe 3O4⋅0.6Fe 2TiO 4
Fe 3O4
3+ Fe (Td)
3+ Fe (Oh)
2+ Fe (Td) Intensity (arb. units) (arb. units) Intensity
2+ Fe (Oh)
705 710 715 720 725 730 Photon energy (eV)
Figure 4.5: Comparison of the experimental Fe L2,3 edge XMCD spectra of Fe 3O4 and 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution thin film with
3+ 3+ 2+ 2+ the theoretical spectra of Fe (Td), Fe (Oh), Fe (Td), and Fe (Oh).
89 2+ A of the solid solution is sharper than that of Fe 3O4, indicating the increase of Fe , and
3+ that the peak B is broader compared to that of Fe 3O4, reflecting the decrease of Fe . These results are quite natural, because it is generally accepted that Ti 4+ ions in the solid solution substitute for Fe 3+ ions with a variation of 2Fe 3+ → Fe 2+ + Ti 4+ , in order to keep charge valance neutrally.
Figure 4.5 shows comparison of the experimental Fe L2,3 edge XMCD spectra of the
3+ 3+ Fe 3O4 and the solid solution thin films with the theoretical spectra of Fe (Td), Fe (Oh),
2+ 2+ Fe (Td), and Fe (Oh). The characteristic three peaks observed at around 709eV,
2+ 3+ 3+ 710eV, and 710.8eV are assigned to Fe (Oh), Fe (Td), and Fe (Oh), respectively.
For the experimental spectra of Fe 3O4 and the solid solution thin films, the minor differences between Fe 3O4 and the solid solutions are seen in the part of arrow (↓) in
2+ 3+ Fig. 4.5, reflecting increase of Fe (Td and Oh) and decrease of Fe ( Oh). In the present study, however, the determination of precise amount at occupation sites based on comparison of XMCD intensities with each other is difficult, because the accurate estimation of internal magnetic field at each site has not been considered.
The Ti L2,3 edge XAS experimental spectrum of the solid solution thin film and the
4+ 4+ 3+ theoretical spectra of Ti (Oh), Ti (Td), and Ti (Oh) are shown in Fig. 4.6. The characteristic four peaks of Ti L2,3 edge XAS spectrum result from the spin orbit splitting of the Ti 2 p and the crystal field splitting of the Ti 3 d, and 457.8eV, 460.3eV,
463.0eV, and 465.5eV are ascribed to L3 edge t2g , L3 edge eg, L2 edge t2g , L2 edge eg, respectively. Also, a small and broad satellite peak around 470 eV is ascribed to the ligand metal charge transfer [19]. The line shape and peak positions of experimental
4+ spectrum are similar to the theoretical of Ti (Oh), but are quite different from those of
3+ Ti (Oh). Also, one can see a shoulder peak at around 460 eV, implying the presence
4+ of Ti (Td). Thus, these results indicate that Ti ions in the solid solution essentially have the valence number of +4 and mainly exist at B sites and slightly at A sites in the spinel structure.
Figure 4.7 shows comparison of the Ti L2,3 edge XMCD experimental spectrum of
90 L3 L2 eg eg t2g t2g
0.4Fe 3O4 0.6Fe 2TiO 4
4+ Ti (Td)
Ti 4+ (O ) Intensity (arb. units) (arb. units) Intensity h
3+ Ti (Oh) 455 460 465 470 475 Photon energy (eV)
Figure 4.6: Comparison of Ti L2,3 edge XAS spectrum of the solid
4+ solution and theoretical spectra of Ti with Oh and T d symmetry and
3+ Ti with Oh symmetry.
91 L2 L3 0.4Fe 3O4 0.6Fe 2TiO 4
Ti 4+ (T + O ) X 2.5 d h
4+ X 2.5 Ti (Td)
4+ X 2.5 Ti (Oh) Intensity (arb. units) (arb. units) Intensity 3+ Ti (Oh)
455 460 465 470 475 Photon energy (eV)
Figure 4.7: Comparison of Ti L2,3 edge XMCD spectra of the solid
4+ 3+ solution and theoretical spectra of Ti with Oh and Td symmetry, Ti
4+ with Oh symmetry, and sum spectrum of Ti with O h and Td symmetry.
92 4+ 4+ 3+ the solid solution thin film with the theoretical spectra of Ti (Oh), Ti (Td), and Ti
(Oh). In the present calculation based on the DFT CI method, the Ti L2,3 edge XMCD theoretical spectra of Ti 4+ are obtained by the transition from the initial electronic configuration (2 p)6(3 d)0 to the final electronic configuration (2 p)5(3 d)1, while the initial and final states of Ti 3+ are (2 p)6(3 d)1 and (2 p)5(3 d)2, respectively. However, it is worth noting that the MCD signal is clearly observed for the solid solution thin film. If the valence number of Ti ions in the solid solutions is +3, it is expected to have magnetic moment due to the electronic configuration (2 p)6(3 d)1. But, one can see the experimental spectrum with different line shape and peak positions from those of
3+ 3+ theoretical of Ti (Oh). This result indicates there are no Ti ions in the solid solution as well as the above XAS results. On the other hand, since the Ti 4+ nominally
4+ 4+ has no 3 d electrons ( d0), i.e. the ground state of Ti is a singlet state, Ti should not basically cause the exchange splitting, and therefore the magnetic polarization should not occur. It is, nevertheless, noteworthy that not only experimental MCD signal but also theoretical ones are observed for the two kinds of theoretical spectra of Ti 4+
(Oh and Td). Comparing the experimental spectrum of the solid solution thin film with
4+ theoretical of Ti (Oh), their line shape and peak positions are slightly different. The experimental spectrum exhibits positive signs at around 461eV and 466 eV, while signs
4+ in the theoretical of Ti (Oh) are negative. In contrast, another theoretical spectrum of
4+ Ti ( Td) exhibits positive signs at the same photon energies. To understand easily, a
4+ spectrum obtained by summing theoretical spectra of Ti with Oh and Td symmetry is shown in Fig. 4.7. Obviously, its spectral shape and peak positions are in good agreement with those of the experimental spectrum. Therefore, it is believed that Ti 4+ ions exist not only B sites but also A sites in the spinel structure, and it is consistent with the XAS results. This suggests the random distribution of Ti 4+ ions, presumably due to a nonequilibrium gas phase process of PLD method.
Next, the origin of Ti L2,3 edge XMCD is discussed in two viewpoints of the charge transfer and the final state effect. According to the molecular orbital calculation [20],
93 (a)
(b) 4s, 4p conduction E α β e Ti ( g)
e t Fe ( g) Ti ( 2g ) 3d overlap t Fe ( 2g ) e Fe ( g)
t Fe ( 2g )
2p(O) valence
Figure 4.8: (a) Wave function contours for the Fe( t2g ) and the Ti( t2g ) orbitals.
14– Molecular orbital diagram for (FeTiO 10 ) cluster [20]. (b) Molecular
14– orbital diagram for (FeTiO 10 ) cluster. Energy levels indicated by dashed lines are unoccupied.
94 4+ 4+ Ti (Oh) Ti (Td)
Intensity (arb. units) Intensity 200 T 100 T 50 T
455 460 465 470 475 455 460 465 470 475 Photon energy (eV)
Figure 4.9: The magnetic field strength dependence of theoretical XMCD
4+ spectra of Ti with Oh and Td symmetry.
95 14– in the (FeTiO 10 ) clusters, there is a weak Fe–Ti bonding formed by overlap of the Fe 3d and Ti 3 d orbitals across the shared polyhedral edge [Fig. 4.8 (a)]. The Fe–Ti
2+ bonding results in a slight delocalization of the Fe (t2g ) β spin electron onto the Ti site [Fig. 4.8 (b)], i.e. Fe 2+ →Ti 4+ charge transfer. However, even though there exist such a charge transfer of delocalized electrons, it is considered that it does not have influence on the MCD spectral shape so much. This is due to the fact that the theoretical spectrum reproduced by the calculation without taking account of metal metal charge transfer is almost corresponding to the experimental spectrum, as shown in Fig. 4.7.
The magnetic field strength dependence of theoretical Ti L2,3 edge XMCD is shown in Fig. 4.9. According to Fermi’s golden rule, relative MCD intensity is simply proportional to the product of electron transition probability and unoccupied states number of 3 d level. However, one can clearly see that theoretical MCD intensities of
4+ 4+ both Ti (Oh) and Ti (Td) depend on the magnetic field strength, indicating that the magnetic polarization of Ti is not intrinsic. Hence, it is believed that the origin of Ti
L2,3 edge XMCD is mainly attributed to the exchange splitting caused by the final state effect based on the final electronic configuration (2 p)5(3 d)1.
96 4.6. Conclusion
In this chapter, the electronic and magnetic structure of 0.4Fe 3O4 0.6Fe 2TiO 4 solid solution has been investigated using Ti L2,3 edge and Fe L2,3 edge XAS and XMCD measurement. Comparison of Fe L2,3 edge XAS and XMCD with first principle calculations based on the DFT CI method has clearly indicated the increase of Fe 2+ at A
4+ and B sites by the substitution of Ti . Although the results of Ti L2,3 edge XAS have elucidated that the valence state of Ti ions in the solid solution thin film is +4, and interestingly, Ti 4+ has exhibited the MCD behavior. Based on comparison of the Ti
L2,3 edge XMCD spectra with theoretical ones, it has been demonstrated that the origin of MCD signal is due to the final state effect from the final electronic configuration (2 p)5(3 d)1, and that Ti 4+ ions exist not only B sites but also A sites in the spinel structure, meaning the random distribution.
97 References in CHAPTER 4
[1] J. M. D. Coey and C. L. Chien, MRS Bull. 28 , 720 (2003). [2] U. Lüders, A. Barthélémy, M. Bibes, K. Bouzehouane, S. Fusil, E. Jacquet, J. P. Contour, J. F. Bobo, J. Fontcuberta, and A. Fert, Adv. Mater. 18 , 1733 (2006). [3] E. SchmidBauer, and P. W. Readman, J. Magn. Magn. Mater. 27 , 114 (1982). [4] V. A. M. Brabers, Physica B 205 , 143 (1994). [5] W. O’Reilly and S.K. Banerjee, Phys. Lett. 17 , 237 (1965). [6] Z. Kakol, J. Sabol, and J. M. Honig, Phys. Rev. B 43 , 649 (1991). [7] R. H. Forster and E. O. Hall, Acta Cryst. 18 , 857 (1965). [8] M. Z. Stout and P. Bayliss, Can, Miner. 18 , 339 (1980). [9] H. J. Lee, G. kim, D. H. Kim, J. S. Kang, C. L. Zhang, S. W. Cheong, J. H. Shim, S. Lee, H. Lee, J. Y. Kim, B. H. Kim, and B. I. Min, J. Phys.: Condens. Matter 20 , 295203 (2008). [10] J. Takaobushi, M. Ishikawa, S. Ueda, E. Ikenaga, J. J. Kim, M. Kobata, Y. Takeda, Y. Saitoh, M. Yabashi, Y. Nishino, D. Miwa, K. Tamasaku, T. Ishikawa, I. Satoh, H. Tanaka, K. Kobayashi, and T. Kawai, Phys. Rev. B 76 , 205108 (2007). [11] J. Stöhr, J. Electron. Spectrosc. Relat. Phenom. 75 , 253 (1995). J. Magn. Magn. Mater. 200 , 470 (1999). [12] H. Ikeno, I. Tanaka, Y. Koyama, T. Mizoguchi, and K. Ogasawara, Phys. Rev. B 72 , 075123 (2005). [13] H. Ikeno and I. Tanaka, Phys. Rev. B 77, 075127 (2008). [14] H. Ikeno, F. M. F. de Groot and I. Tanaka, J. Phys.: Condens. Matter 21 , 104208 (2009). [15] B. A. Wechsler, D. H. Lindsley, and C. T. Prewitt, Am. Miner. 69 , 754 (1984). [16] A. Rosén, D.E. Ellis, H. Adachi, and F.W. Averill, J. Chem. Phys. 65 , 3629 (1976). [17] K. Ogasawara, T. Iwata, Y. Koyama, T. Ishii, I. Tanaka, and H. Adachi, Phys. Rev. B 64 , 115413 (2001).
98 [18] T. J. Regan, H. Ohldag, C. Stamm, F. Nolting, J. Luning, J. Stohr, and R. L. White, Phys. Rev. B 64 , 214422 (2001). [19] K. Okada, T. Uozumi, and A. Kotani, Japan J. Appl. Phys. 32 , 113 (1993). [20] D. M. Sherman, Phys. Chem. Miner. 14 , 364 (1987).
99 Summary
In the present thesis, with the aim of opening up a new field of electronics, the author explored the possibilities of applying magnetite ulvöspinel solid solutions, which are one of the spinel type oxides, to the field of semiconductor spintronics. Optimization of growth conditions to attain high quality epitaxial thin film was performed, and the structural, electrical, and magnetic properties of the resultant solid solution thin films were examined in detail. Moreover, in order to obtain a guideline for the development of semiconductor spintronics devices, electronic and magnetic structures of the solid solution thin films were investigated using XAS and XMCD measurements along with theoretical calculations. The results obtained by the present study are summarized as follows In chapter1, the concept and history of spintronics were briefly overviewed. So far, spintronics has achieved great advances in fabrication of various types of novel devices using ferromagnetic metals, but a key to realize next generation electronics is to establish semiconductor spintronics. Ideal for the semiconductor spintronic applications is to possess physical properties such as high spin polarization, room temperature ferromagnetism, and chemical stability in air. As one candidate material, ferromagnetic oxide semiconductors composed of transition metal elements have attracted considerable attention because a wide variety of properties derived from the oxides can be utilized in the spintronic application. In chapter 2, first, basic structural and physical properties as well as possible spintronic applications of Fe 3O4 are summarized in order to understand the solid solutions based on Fe 3O4. Secondly, intriguing properties of (1–x)Fe 3O4 xFe 2TiO 4 solid solutions were presented. They are easiness to control the carrier type (n or p type), room temperature ferromagnetism regardless of the conduction type ( x < 0.8),
100 and the probability of high spin polarization. Finally, the future perspective of the solid solutions in application to spintronics was discussed. In chapter 3, Synthesis of the magnetite ulvöspinel solid solution thin films was
described. The (1–x)Fe 3O4 xFe 2TiO 4 ( x = 0.6) solid solutions thin films were grown
on α Al 2O3 (0001) and MgO(100) substrates using a pulsed laser deposition technique, and the structural, electrical, and magnetic properties of the resultant thin films were examined.
In section 3.1, the fabrication of solid solution thin films on α Al 2O3 (0001) substrates was performed. This is the first report on the epitaxial growth of solid solution thin films in the magnetite ulvöspinel system. They show a preferential in plane orientation due to the large lattice mismatch between the thin film and the substrate. The crystallographic relationship is Fe3O4 Fe 2TiO 4 solid solution (111) ||
_ _ _ α Al 2O3 (0001) and Fe 3O4 Fe 2TiO 4 solid solution [110] || α Al 2O3 [1100] and [1100]. The solid solution thin film has a Curie temperature higher than room temperature. In other words, the room temperature ferrimagnetism is attained. Also, semiconducting behavior with n type conduction carriers is observed. Furthermore, the presence of spin polarized charge carriers at room temperature was confirmed by the Hall effect measurement. These results indicate that the solid solution thin film can be a promising candidate as a semiconductor spintronics device. In section 3.2, epitaxial solid solution thin films with flat and smooth surface were grown on MgO(100) substrates by optimizing the oxygen partial pressure and substrate temperature. Also, the solid solution thin films were epitaxially grown with cube on cube relationship on the MgO(100) substrate, due to the small lattice mismatch between the film and the substrate (0.5%). The crystallographic relationship is
Fe 3O4 Fe 2TiO 4 solid solution (100)[111] || MgO(100)[111]. Their magnetization and electric conductivity of the solid solution thin films were increased compared to those of films grown on α Al 2O3(0001) substrates, owing to the small lattice mismatch. These results indicate that it is feasible to construct heterostructures with a smooth interface,
101 suggesting that the epitaxial solid solution thin films on MgO(100) substrates are promising for device implementation.
In chapter 4, for the Fe 3O4 Fe 2TiO 4 solid solution thin films, Ti and Fe L 2,3 edges XAS and XMCD measurements were performed in order to clarify the electronic and magnetic structures of the thin films. The results of Ti L2,3 edge XAS revealed that the valence state of Ti ions in the solid solution thin films is essentially +4. From the
4+ results of Fe and Ti L2,3 edge XAS and XMCD, it was found that the Ti ions are randomly distributed at octahedral and tetrahedral sites in the spinel structure. Despite the fact that the valence number of Ti is +4, for Ti L2,3 edge XMCD, the MCD behavior was observed. A comparison of the Ti L2,3 edge XMCD spectra with theoretical ones obtained by first principle calculations based on the DFT CI method shows that the origin is due to the final state effect derived from the final electronic configuration (2 p)5(3 d)1.
102 List of Publications
CHAPTER 3 “Epitaxial Growth of Room Temperature Ferrimagnetic Semiconductor Thin Films
Based on Fe 3O4 Fe 2TiO 4 Solid Solution” Hideaki Murase, Koji Fujita, Shunsuke Murai, and Katsuhisa Tanaka Materials Transactions 50 , 1076 1080 (2009).
“Epitaxial Growth of Ferrimagnetic Semiconductor 0.4Fe 3O4 0.6Fe 2TiO 4 Solid Solution Thin Films on MgO(100) Substrates” Hideaki Murase, Koji Fujita, Shunsuke Murai, and Katsuhisa Tanaka Journal of Physics : Conference Series 200 , 062013 (2010).
“Fabrication of Room Temperature Ferrimagnetic Semiconductor 0.3Fe 3O4 0.7Fe 2TiO 4 Solid Solution Thin Films” Hideaki Murase, Koji Fujita, Shunsuke Murai, and Katsuhisa Tanaka to be submitted to Journal of the Ceramic Society of Japan.
CHAPTER 4 “X ray Magnetic Circular Dichroism Study of Ferrimagnetic Semiconductor Thin Films
Based on Fe 3O4 Fe 2TiO 4 Solid Solution” Hideaki Murase, Koji Fujita, Hidekazu Ikeno, Shunsuke Murai, Tetsuya Nakamura, Isao Tanaka, and Katsuhisa Tanaka to be submitted to Applied Physics Express .
103 Acknowledgement
The present thesis has been carried out under the direction of Professor Katsuhisa Tanaka at Graduate School of Engineering, Kyoto University.
The author wishes to express his sincere gratitude to Professor Katsuhisa Tanaka for his continuous encouragement and valuable advice all through the duration of the present work.
The author is also grateful to Professor Toshinobu Yoko and Professor Kazuyuki Hirao for their guidance and discussion in preparing the present thesis. The author would like to acknowledge
Professor Atsushi Nakahira of Osaka Prefecture University for his understanding. The author is greatly indebted to Associate Professor Koji Fujita for his informative discussion, helpful advice, and sincere supports in the present work. The author thanks Research Associate
Shunsuke Murai for his valuable discussion and helpful assistance. The author also thanks Dr.
Takanori Okada for his fruitful discussions.
Experimental supports for Hall effect measurements and informative discussions by
Professor Minoru Suzuki is greatly acknowledged. The author would like to thank
Dr. Hidekazu Ikeno for the theoretical calculation of XAS and XMCD. The author also thanks
Dr. Hajime Hojo, Dr. Hirofumi Akamatsu, Dr. Takashi Kubo, and Mr. Kazuma Kugimiya, for their experimental supports, fruitful discussions, and hearty encouragements.
Hearty thanks are made to all the students of Tanaka’s laboratory, Hirao’s laboratory, and
Nakahira’s laboratory for their collaboration and everyday activities. Especially his colleagues,
Dr. Xiangeng Meng, Mrs. Sakiko Ukon, and Mrs. Yanfua Zong contributed to his happiest time.
Finally, the author would like to express his sincere gratitude to his parents, Mr. Katsuhide
Murase, Mrs. Junko Murase, and his family for their understanding, supports, and hearty encouragements.
Kyoto, 2010 Hideaki Murase
104