Photoemission and X-ray Magnetic Circular Dichroism Study of the Diluted Magnetic Semiconductor Zn1−xCoxO Master Thesis
Masaki Kobayashi Department of Physics, University of Tokyo January, 2005
Contents 1 Introduction 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1 1
1.2 Physical properties of Zn1−x Cox O . . . . . . . . . . . . . . . . . . 1.3 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5 7
2 Experiment and analysis 2.1 Principles of photoemission spectroscopy . . . . . . . . . . . . . .
9 9
2.1.1 2.1.2
Photoemission spectroscopy . . . . . . . . . . . . . . . . . Resonant photoemission spectroscopy . . . . . . . . . . . .
9 11
2.2 Configuration-interaction cluster model . . . . . . . . . . . . . . . 2.3 Principles of x-ray magnetic circular dichroism and sum rules . . . 2.3.1 X-ray absorption spectroscopy . . . . . . . . . . . . . . . .
12 14 14
2.3.2 2.3.3
X-ray magnetic circular dichroism . . . . . . . . . . . . . . XMCD sum rules . . . . . . . . . . . . . . . . . . . . . . .
3 Sample preparation
14 15 17
4 Photoemission spectroscopy study 19 4.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 4.2 Results of resonant photoemission and x-ray photoemission . . . . 4.3 CI cluster-model analysis for photoemission spectra . . . . . . . . 5 X-ray magnetic circular dichroism study
22 26 29
5.1 Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Magnetic field and temperature dependences of x-ray magnetic circular dichroism spectra . . . . . . . . . . . . . . . . . . . . . .
31
5.3 Comparison with calcualted spectra . . . . . . . . . . . . . . . . . 5.4 Results of XMCD sum rules and discussion . . . . . . . . . . . . .
32 34
6 Summary
29
37
i
Chapter 1 Introduction 1.1
Background
Diluted magnetic semiconductors (DMS’s), in which a portion of atoms in the non-magnetic semiconductor hosts are replaced by magnetic ions, are key materials in “spintronics” (spin electronics), which is intended to manipulate both the spin and charge degrees of freedom through coupling between the spins of the magnetic ions and the charge carriers. From the success of fabricating ferromagnetic III-V based DMS’s such as In1−x Mnx As [1] and Ga1−x Mnx As [2] by molecular beam epitaxiy (MBE), DMS has been extensively studied both experimentally and theoretically. These materials are considered to be intrinsic carrier-induced ferromagnetic materials. For example, Ohno et al. [3] reported that the Curie temperature (TC ) of Ga1−x Mnx As increases with increasing Mn concentration below x ∼ 0.05 (Fig. 1.2), and Koshihara et al. [4] demonstrated a ferromagnetic ordering induced by photogenerated carriers at the (In,Mn)As/GaSb interface (Fig. 1.3), both of which results suggest that ferromagnetism in DMS’s is related to carrier concentration. Indeed, using ferromagnetic DMS’s, it has been successful to achieve spin-related techniques that can be used for applications, such as spin injection [5] (Fig. 1.4), electrical manipulation of magnetization reversal [6], and current-induced domain-wall switching [7]. However, because the Curie temperature (TC ) of the prototypical ferromagnetic DMS Ga1−x Mnx As is below the room temperature (TC < 200 K), it is still difficult to utilize DMS’s in practical applications. Therefore, it has been strongly desired that ferromagnetic DMS’s having TC above the room temperature are realized. Since theoretical studies [8, 9] predicted that DMS’s based on wide gap semiconductors such as GaN and ZnO show room temperature ferromagnetism as shown in Fig. 1.5(a) and (b), it has started to search for room temperature fer1
Itinerant carriers Exchange interaction
Magneic imputiry
Figure 1.1: Schematic picture of carrier-induced ferromagnetism.
Figure 1.2: Manganese composition dependence of the Curie temperature TC of Figure 1.3: Temperature dependence of magGa1−x Mnx As determined by transport [3]. netization observed during cooldown in the dark (open circle) and warmup (solid circles) in (In,Mn)As/GaSb [4].
Figure 1.4: Electrical spin injection [5]: (left) Device structure. (right) Hysteretic electroluminescense polarization is a direct result of spin injection from the ferromagnetic (Ga,Mn)As layer.
2
romagnetic DMS’s by material design. Using the Zener model [10], in which the magnetic interaction between the local spins is assumed to be RudermanKittel-Kasuya-Yoshida (RKKY) interaction, Dietl et al. [8] have explained TC of Ga1−x Mnx As and that of II-VI counterpart Zn1−x Mnx Te, and predicted Mndoped p-type semiconductors to show TC exceeding room temperature, as shown in Fig. 1.5(a). The general tendency for greater TC value for lighter anions stems from the corresponding increase in the p-d hybridization strength and the reduction of the spin-orbit splitting between the Γ8 and Γ7 bands. Sato and Katayama-Yoshida [9] have investigated magnetism in III-V- and II-VI-based DMS’s based on ab initio calculations within the local-density approximation, and suggested that the origin of ferromagnetism in DMS’s is double exchange mechanism [11]. The ferromagnetic state is stabilized if there are carriers in the itinerant p-d hybridized antibonding band, which becomes in-gap states in wide gap semiconductors. Recently, oxide-based DMS’s [12] have been attracted much attention as candidates for room temperature ferromagnets and are expected to have potentiality to expand the range of applications due to the wide band gap. In addition, oxides are easy to operate in the atmospheric environment and are inexpensive. Ti1−x Cox O2 [13] is one of the most interesting material not only due to room temperature ferromagnetism but also to photocatalytic properties of the host TiO2 . One can control their ferromagnetic behavior by changing carrier concentration through the oxygen pressure during preparation [14]. Sn1−x Cox O2 also intrigues us due to giant magnetic moment 7.5 ± 0.5 µB /Co with TC about 650 K [15]. Sharma et al. have reported the first observation of ferromagnetic ordering with TC well above 425 K in uniform by Mn-doped ZnO bulk ceramics [16]. Zn1−x Vx O [17] and Zn1−x Cox O [18] are also reported to be ferromagnetic with TC above room temperature. However, the mechanisms of ferromagnetism in these materials including Ga1−x Mnx As is still under debate. It has even been controversial whether those ferromagnetic behavior are intrinsic properties of the DMS’s or due to impurity phases such as segregated metal clusters.
3
Figure 1.5: Theoretical predictions on ZnO-based DMS. (a) Curie temperature TC for various p-type semiconductors containing 5% of Mn and 3.4 × 1020 holes per cm3 [8]. (b) Stability of the ferromagnetic state against the spin-glass state in ZnO-based DMS. A positive energy difference indicates that the ferromagnetic state is more stable than the spin-glass state [19]. (c) Stability of the ferromagnetic state in Co-doped ZnO as a function of carrier concentration [19]. (d) Total energy differences between the ferromagnetic and antiferromagnetic states are plotted as a function of the amount of doped electrons for Co ions [20].
Figure 1.6: Magnetization curves of some room temperature ferromagnetic oxidebased DMS’s: (a) Ti1−x Cox O2 [13], (b) Sn1−x Cox O2 [15], (c) Zn1−x Mnx O [16], (d) Zn1−x Cox O [21].
4
1.2
Physical properties of Zn1−x CoxO
Zinc oxide (ZnO) is apt to became an n-type semiconductor because oxygen vacancies act as native donors (Zn vacancies acting as acceptors [22]). Its crystal structure is the wurtzite-type with the lattice parameters a = b = 3.2495 and ˚ and α = β = 90◦ and γ = 120◦ as shown in Fig. 1.7. ZnO is a very c = 5.2069 A important material in optical applications due to the wide band gap of ∼ 3.4 eV and the large exciton binding energy of ∼ 60 meV. Very recently, Tsukazaki et al. [23] have succeeded in fabricating high quality p-type ZnO and demonstrated violet electroluminescence from zinc oxide homostructual p-i-n junction at room temperature. Hence, the ZnO-based DMS’s have the possibility of expanding the range of applications due to the wide band gap of ZnO (∼ 3.4 eV) as, e.g., polarized light emitting diodes, optical isolators, and transparent magnets. Theoretical studies have predicted that intrinsic ferromagnetism of Co-doped ZnO can be stabilized by electron doping [9, 20] as shown in Fig. 1.5(c) and (d). However, the possibility of extrinsic origins of ferromagnetism such as precipitated Co metal clusters [24] and CoO nanoparticles [25] have not been excluded and the ferromagnetism of Zn1−x Cox O is still in strong dispute. It seems difficult to detect secondary phases by x-ray diffraction (XRD) due to an insufficient quantity and random distribution of these precipitates if they existed. Lee et al. [26] have pointed out that the confirmation of the existence of metal clusters using XRD is not conclusive and measurement by neutron scattering is necessary for the study of DMS materials. Schwartz et al [27] have observed that the aggregation of paramagnetic Co2+ :ZnO DMS quantum dots at room temperature give rise to robust ferromagnetic ordering with TC > 350 K. The S = 3/2 saturation magnetization behavior of the band gap magnetic circular dichroism (MCD) intensity indicates magnetization of the semiconductor due to interaction with the Co2+ dopant (the giant Zeeman effect) as shown in Fig. 1.8. Recently, Venkatesan et al. [28] have reported that ZnO-based DMS’s have a huge anisotropy of the magnetization and the magnetization of Zn1−x Cox O shows systematic dependence on oxygen pressure during preparation and Co concentration as shown in Fig. 1.9. Tuan et al. and Saeki et al. have reported that the magnetization of Zn1−x Cox O increases with increasing the temperature of rapid-thermal annealing in vacuum. If these phenomena do not originate from secondary phases or Co metal clustering, it seems that the ferromagnetism of Zn1−x Cox O is related to the carrier concentration and hence to the oxygen vecancies.
5
Figure 1.7: Wurtzite structure
Figure 1.8: MCD spectra of Co2+ :ZnO nanocrystals. (a) Absorption at 7 K and variablefield (0-6.5 T) MCD spectra at 5 K. (b) Variable-temperature, variable-field MCD intensities collected at 15600 cm−1 (open circle, Co2+ ligand-field region) and 27200 cm−1 (solid triangle, band gap region).
Figure 1.9: (a) Magnetic moment of Zn0.93 Co0.07 O measured at room temperature for films prepared at different oxygen pressures. (b) Magnetic moment of Zn1−x Cox O measured at room temperature for films prepared at 10−4 mbar for different x’s [28].
6
1.3
Motivation
From the fundamental physics point of view, the unique physical properties of DMS’s originate from the substituted transition-metal ions and, therefore, dd Coulomb interaction and hybridization between the ligand p orbitals and the localized d orbitals should play important roles. For such systems, photoemission spectroscopy (PES) and x-ray absorption spectroscopy (XAS) combined with configuration-interaction (CI) analysis using cluster model are powerful tools to investigate the electronic structure of the substituted transition-metal ions [29]. X-ray magnetic circular dichroism (XMCD), the difference in absorption spectra between right-handed and left-handed circularly polarized x-rays, is sensitive to magnetic polarization, and therefore enables us to extract the magnetic properties of the substituted transition-metal ions [30]. Since resonant PES, XAS, and XMCD experiments are element-specific, these make a strong combination to investigate both the electronic structure and the magnetic states of the doped magnetic ions in the DMS’s. Although the Co ions in Zn1−x Cox O are expected to be divalent and tetrahedrally coordinated by oxygen, the nature of the ferromagnetic component of Zn1−x Cox O has been unclear. To identify the electronic structure of the Co ions, espitially of the ferromagnetic component, we have performed PES and XMCD experiments on the ferromagnetic Zn1−x Cox O thin film.
7
Chapter 2 Experiment and analysis 2.1 2.1.1
Principles of photoemission spectroscopy Photoemission spectroscopy
Radiating monochromatic x-ray (or ultra-violet ray) on a material, electrons are emitted to outside of the material by absorption of the x-ray (photoelectric effect). To measure the kinetic energy of the electrons emitted out of the material (photoelectrons), one can obtain the information about the electronic structure of the materials. Analyzing the kinetic energy distribution of photoelectrons is called photoemission spectroscopy (PES). From the energy conservation law, the relationship between the kinetic energy V of the photoelectron Ekin relative to the vacuum level EV and the photon energy hν and the binding energy EB is given by V Ekin = hν − φ − EB ,
(2.1)
where φ is the work function of the sample. Using the kinetic energy Ekin relative to the Fermi level EF , the relationship between Ekin , EB , and hν is given by Ekin = hν − EB .
(2.2)
In one-electron approximation, the binding energy is equal to the negative Hartree-Fock orbital energy, EB = −k ,
(2.3)
where subscript k denotes the Bloch wave number of electron. This relationship is called Koopmans’ theorem [31]. This assumption is valid when the wave function of both the initial and final states can be expressed by single Slater determi9
nants of the N- and (N − 1)-electron systems, respectively, and the one-electron wave functions do not change by the removal of the electron. If we apply this approximation, the photoemission spectrum I(EB ) can be expressed as I(EB ) ∝
δ(EB + k ) ∝ N(−EB ).
(2.4)
k
Thus, when the one-electron approximation is valid, the photoemission spectrum is proportional to the density of states of the occupied one-electron states N(E). If many-body effects are taken into account, it need to more general description. Photoemission produces a final state that is lacking one electron with respect to the initial state. PES measures the energy difference between the total energies of a state with N electrons EiN and one with N − 1 electrons EfN −1 . In the most rigorous formulation, the binding energy of an electron measured by PES is therefore given by EB = EfN −1 − EiN ,
(2.5)
where subscripts i and f denote the initial and final state. Using Fermi’s golden rules, the photoemission spectrum, which now corresponds to the single-particle excitation spectrum of the system, is expressed I(ω) ∝
−1 N −1 |ΨN |ak |ΨN − EiN )), i |δ(ω − (Ef f
(2.6)
k −1 and ΨN where ΨN i denote the final and initial states, ak is the annihilation f
operator of the electron occupying the orbital k. E
PES spectra Ekin
Photo-electron
V
Ekin EV EF
Ekin
Intensity
hν Photon n
φ EB
DOS Figure 2.1: Schematic diagram of photoemission spectroscopy.
10
2.1.2
Resonant photoemission spectroscopy
With synchrotron radiation, one can use photons of continuously variable energy. A schematic diagram of resonant PES is depicted in Fig. 2.2. When the energy of the incident photon is equal to the energy difference between the p core level and the valence d states, beside the direct photoemission the process of a valence d electron, p6 dN + hν → p6 dN −1 + e− ,
(2.7)
the photo-absorption and subsequent Auger-type decay, called super CosterKr¨onig decay, p6 dN + hν → p5 dN + 1 → p6 dN −1 + e− ,
(2.8)
occur. The final states of these two processes have the same electron configuration, and therefore a quantum-mechanical interface occur. The photoemission intensity is resonantly enhanced and shows a so-called Fano profile [32]. Since this enhancement takes place only for the d orbitals, one can obtain the d partial density of states in the compound.
Jν
'
Normal Photoemission
'( final state Jν
Fano Resonance
initial state
Core absorption
Auger decay
final state
Figure 2.2: Schematic diagram of resonant photoemission spectroscopy.
11
2.2
Configuration-interaction cluster model
The transition-metal impurities substituted for the cation in the II-VI and III-V semiconductors are tetrahedrally coordinated by the anions and can be described by the tetrahedral T Y4 cluster model (T is the transition-metal ion and Y is an anion). In cluster model, the 3d-3d Coulomb interaction and the 3d-anion hybridization are taken into account. One-electron transfer integrals between the 3d orbitals and the anion ligand orbitals are given by the Slater-Koster parameters (pdσ) and (pdπ) as Tt2 ≡ t2 |H|Lt2 =
8 4 (pdσ)2 + (pdπ)2, 3 9
(2.9)
√ 2 6 Te ≡ e|H|Le = (pdπ), (2.10) 3 where Lt2 and Le are ligand orbitals with T2 and E symmetry of the Td point group, respectively [33, 34].
and
One has to take into account the hybridization between the 3d and the ligand orbitals. Then, the ground state ψg expanded as a linear combination of the ligand-to-3d charge-transfered electronic configurations: ψg = a0 ψ(dn ) + a1 ψ(dn+1 L) + a2 ψ(dn+2 L2 ) + ... ,
(2.11)
where L denotes a hole in a anion ligand orbital. In this picture, covalency is regarded as the interaction between different electronic configurations which arise from the ligand-to-3d charge transfer. This picture is called configuration interaction (CI). In the CI cluster-model analysis of PES spectra, the final states wave function of core-level photoemission ψc and valence-band photoemission ψv are also expanded by linear combinations of the charge-transfered states as ψc = b0 ψ(cdn ) + b1 ψ(cdn+1 L) + b2 ψ(cdn+2 L2 ) + ... ,
(2.12)
ψv = c0 ψ(dn−1 ) + c1 ψ(dn L) + c2 ψ(dn+1 L2 ) + ... ,
(2.13)
where c denotes a core hole. CI interaction influences the binding energy and the photoemission spectrum as the satellite structure through Eqn.(2.5) and (2.6). The ligand-to-3d charge-transfer energy is defined by ∆ ≡ E(dn+1 L) − E(dn ), 12
(2.14)
and the on-site d-d Coulomb interaction energy by U ≡ E(dn−1 ) + E(dn+1 ) − 2E(dn ),
(2.15)
where E(dn L ) is the center of gravity of the dn L multiplet. It is also possible to define the charge-transfer energy ∆eff and the Coulomb interaction energy Ueff with the lowest term of each multiplet. The multiplet splitting is expressed using Racah parameters A, B, and C or Kanamori parameters u, u , j, and j for the multiplet splitting of the dn configuration due to intra-atomic Coulomb and exchange interactions. There is relationship between these four parameters; u = u − 2j, j = j. The charge-transfer energy ∆ and Coulomb energy U as well as the Slater-Koster parameters are adjustable parameters in the cluster-model calculation.
Figure 2.3: CoO4 cluster
13
2.3
Principles of x-ray magnetic circular dichroism and sum rules
2.3.1
X-ray absorption spectroscopy
The measurements of photo-absorption by excitation of a core-level electron into unoccupied states as a function of photon energy is called x-ray absorption spectroscopy. The photo-absorption intensity is given by I(hν) =
|f |T |i|2δ(Ei − Ef − hν),
(2.16)
f
where T is the dipole transition operator. In the 3d transition-metal compounds, transition-metal 2p XAS spectra reflect the 3d states such as the valence, the spin state and the crystal-field splitting. There are two measurement modes for XAS, the transmission mode and the total-yield mode. In the transmission mode, the intensity of the x-ray is measure before and after the sample and the ratio of the transmitted x-rays is counted. Transmission-mode experiments are standard for hard x-rays, while for soft xrays, they are difficult to perform because of the strong interaction of soft x-rays with the sample. In the present work, the total electron-yield mode was employed.
2.3.2
X-ray magnetic circular dichroism
When the relativistic electrons in the storage ring are deflected by the bending magnets that keep them in a closed circular orbit, they emit highly intense beams of linearly polarized x-rays in the plane of the electron orbit (bremsstrahlung) but they emit circularly elliptically polarized light out of the plane. Currently, an number of alternative sources for circularly polarized synchrotron radiation are under development. The most notable are so-called insertion devices like helical wigglers [35] and crossed [36] undulator, which are complex arrays of magnets with which the electrons in a storage ring are made to oscillate in two directions perpendicular to their propagation direction, with the result that they emit circularly polarized light. Using circularly polarized x-rays in XAS, x-ray magnetic circular dichroism (XMCD) is defined as the difference in absorption spectra between right-handed and left-handed circularly polarized x-rays when the helicity of the x-rays are parallel and antiparallel to the magnetization direction of the magnetic materials such as ferromagnet or ferrimagnet. XMCD is sensitive to magnetic polarization, 14
and therefore enable us to study the magnetic properties of particular orbitals on each element.
2.3.3
XMCD sum rules
XMCD reflects the spin and orbital polarization of local electronic states. Using integrated intensity of the L2,3 -edge XAS and XMCD spectra of a transitionmetal atom, one can separately estimate the orbital [37] and spin [38] magnetic moments by applying XMCD sum rules given by 4
L3 +L2
(µ+ − µ− )dω
(10 − Nd ), Morb = − 3 L3 +L2 (µ+ + µ− )dω Mspin + 7MT = −
6
(2.17)
+ − (µ − µ )dω − 4 (µ+ − µ− )dω L3 L3 +L2 (10 − Nd ), (2.18) (µ+ + µ− )dω L3 +L2
where Morb and Mspin are the spin and orbital magnetic moments in units of µB /atom, respectively, µ+ (µ− ) is the absorption intensity for the positive (negative) helicity, Nd is the d electron occupation number of the specific transitionmetal atom. The L3 and L2 denote the integration range. MT is the expectation value of the magnetic dipole operator, which is small when the local symmetry of the transition-metal atomic site is high and is neglected here with respect to Mspin .
15
Chapter 3 Sample preparation A Zn1−x Cox O (x = 0.05) thin film was epitaxially prepared on a α-Al2 O3 (0001) substrate by the pulsed laser deposition technique using an ArF excimer laser with energy density 1.0 J/cm2 . During the deposition, the substrate was kept at temperature of ∼300 ◦ C in an oxygen ambient pressure of 1.0×10−5 mbar. The A on a 500 ˚ A ZnO buffer total thickness of the Zn1−x Cox O layer was ∼ 2000 ˚ layer. X-ray diffraction confirmed that the thin film had the wurtzite structure and no secondary phase was observed. All samples have been fabricated by Saeki et al. Detail of the sample fabrication are given in Ref. [39]. Ferromagnetism with TC above the room temperature was confirmed by magnetization measurements using a SQUID magnetometer (Quantum Design, Co. Ltd.). Figure 3.1 shows the magnetization (M) curves, where the magnetic field (H) was applied perpendicular to the sample plane (parallel to the c-axis). Figure 3.1(a) shows the raw data of the magnetic field dependence of magnetization. It seems that there are ferromagnetic, paramagnetic, and diamagnetic components. The diamagnetic components comes from the ZnO buffer layer and the Al2 O3 substrate. Subtracting the liner component at high magnetic field, one can obtain the ferromagnetic component as shown in Fig. 3.1(b). The saturation magnetization slightly decreases with increasing temperature (T ). Figure 3.1(c) and (d) show enlarged plots of Fig. 3.1(b) near the origin at T = 5.0 and 300 K. The residual magnetization and coercivity decreased with increasing temperature. Figure 3.1(e) shows an M-T curve at H = 1.0 T without the diamagnetic component of the ZnO buffer layer and the sapphire substrate. The temperature dependence of the magnetization was week above ∼ 100 K. Because the magnetization is rapidly increased at low temperatures, it seemes that the paramagnetic component obeys the Curie law. The magnetization of the film was as small as ∼ 0.3 µB . This is probably due to insufficient electron carrier concentra17
tion (n 1020 cm−3 ) because no Al doping [18] nor high-temperature vacuum annealing [21, 39] was performed.
Zn0.95Co0.05O
10
(a)
5 0 @5K @20K @45K @100K @220K @300K
-5 -10 -15 -15
-10
Magnetization (µB/Co)
-5
Magnetization x10 (emu)
15
(b)
0.4 0.2 0
@5K @20K @45K @100K @220K @300K
-0.2 -0.4
-5
0
5
10
15
-15
-10
Magnetic Field (kOe)
0
5
10
15
0.4 at T = 5 K 0.2
(c)
Hdec
0
Magnetization (µB/Co)
Magnetization (µB/Co)Magnetization (µB/Co)
-5
Magnetic Field (kOe)
Hinc
-0.2 -0.4 0.2
at T = 300 K
(d)
Hdec
0 Hinc
0.6
at H = 1.0 T
0.5
(e)
0.4 0.3 0.2 0.1
-0.2
0 -4
-2 0 2 Magnetic Field (kOe)
4
0
100
200
300
Temperature (K)
Figure 3.1: Magnetization curves of the Zn0.95 Co0.05 O thin film. (a) Magnetic field dependence including the diamagnetism of the substrate. Units are in emu. (b) Magnetization with the linear component subtracted in the high magnetic field region. (c) Enlarged plot of the M -H curve at T = 5.0 K. (d) Enlarged plot of the M -H curve at T = 300 K. (e) M -T curve at H = 1.0 T with the diamagnetic component of the substrate subtracted.
18
Chapter 4 Photoemission spectroscopy study 4.1
Experimental
Ultraviolet photoemission (UPS) measurements were performed at BL-18A of Photon Factory. Spectra were taken in a vacuum below 7.5×10−10 Torr. The total resolution of the spectrometer (VG CLAM hemispherical analyzer) including temperature broadening was ∼ 200 meV. X-ray photoemission (XPS) measurements were performed using a Gammadata Scienta SES-100 hemispherical analyzer and an AlKα source (hν = 1486.6 eV) in a vacuum below 1.0×10−9 Torr. In both UPS and XPS measurements, photoelectrons were collected in the angle integrated mode at room temperature. Sample surface was cleaned by cycles of Ar+ -ion sputtering at 1.5 kV for 10 min and annealing at 250 ◦ C for 10 min. For some oxides, e.g. TiO2 , they are reduction by Ar ion sputtering [40]. However, ZnO is known as a stable material against Ar ion bombardment as confirmed by XPS [41] (see Fig. 4.2). It was performed to anneal the damaged surface for repairing. Cleanliness of the
Figure 4.1: Schematic view of the BL-18A.
19
sample surface was checked by low energy electron diffraction (LEED) and by the absence of a high binding-energy shoulder in the O 1s spectrum and C 1s contamination by XPS. Figure 4.3 shows the LEED images of the surface of ZnO and Zn0.95 Co0.05 O treated by Ar ion sputtering and annealing. The LEED image of ZnO shows clear spots with six fold symmetry. On the other hand, spots are poorly resolved in the LEED image of Zn0.95 Co0.05 O. This indicates that although it was possible to obtain a clean surface for ZnO by Ar ion sputtering and annealing, the similarly treated surface of Zn0.95 Co0.05 O was not completely ordered. It seemes that the degradation of the sample is induced by doping the Co impurities. The same trend has also been observed in Ga1−x Mnx N [42]. Figure 4.4 shows core-level XPS spectra of the contaminated and the treated surface of Zn0.95 Co0.05 O. Before the treatment, the signal from C1s was very strong. Green curves in Fig. 4.4 show the XPS spectra after 3 hours from the surface treatment. This indicates that the surface of Zn0.95 Co0.05 O obtained by the treatment remained uncontaminated for 3 hours in the vacuum. Figure 4.5 shows the ratio of the XPS intensity of Zn2p3/2 to that of O1s estimated after each surface treatments. This indicates that the treatment did not change the ratio of Zn to O.
Figure 4.2: Change of various oxide surfaces by Ar ion sputtering, according to theclassification by the enthalpy of atomization and the ionicity. (I): Additional peaks appearing in core-level XPS spectrum/reduced region, (II): full width half maximum of the core-level XPS spectrum broadened/amorphous region, (III): stable region. [41]
20
1.5x10
Zn0,95Co0.05O Zn2p3/2 AlKα
6
no treatment treated surface after 3 hours
1.0
(a)
0.5 1028
1026
1024
1022
1020
1018
1016
Binding Energy (eV) Counts/Sec
400x10
3
Zn0,95Co0.05O O1s AlKα
no treatment treated surface after 3 hours
300 200
(b)
100 536
534
532
530
528
526
Binding Energy (eV) Counts/Sec
50x10
3
40 30
Zn0,95Co0.05O C1s AlKα
no treatment treated surface after 3 hours
(c)
20 290
288
286
284
282
Binding Energy (eV)
Figure 4.4: Zn2p3/2 (a), O1s (b), and C1s (c) corelevel XPS spectra under different surface conditions. Red and blue curves show the spectra from contaminated and treated surfaces. Green curves show the spectra 3 hours after the surface treatment.
21
Zn 2p3/2 intensity (arb. units)
Counts/Sec
Figure 4.3: LEED images of ZnO (left) and Zn0.95 Co0.05 O (right). For ZnO, the exposure time was 15 min and the applied voltage was -56 V. For Zn0.95 Co0.05 O, the exposure time was 15 min, and the applied voltage was -107 V.
Zn0.95Co0.05O 1500
AlKα 1000
500
Ratio of XPS intensity 0 0
50 100 150 200 250
O 1s intensity (arb. units) Figure 4.5: Ratio of the XPS spectral intensity of O1s to that of Zn2p3/2 . The dashed line is a guide to the eye.
4.2
Results of resonant photoemission and xray photoemission
Figure 4.6(a) shows the Co 2p3/2 XPS spectrum of Zn0.95 Co0.05 O compared with those of several Co oxides. All the spectra consist of two structures, i.e., a strong peak on the lower binding energy side and a satellite on the higher binding energy side. The spectrum of Zn0.95 Co0.05 O is similar to that of CoO. In the case of LiCoO2 (Co3+ ), a main peak around EB ∼ 780 eV is sharper than that in Zn0.95 Co0.05 O. Thus, the valency of the Co ion in ZnO is considered to be divalent, the same as that of CoO. Figure 4.7(a) shows the valence-band (VB) UPS spectra of Zn0.95 Co0.05 O taken at various photon energies in the Co 3p→3d core-excitation region. The absorption spectrum in the same energy region is shown the inset. Binding energies (EB ’s) were referenced to the Fermi energy (EF ) of a metallic sample holder which was in electrical contact with the sample. The absorption spectrum shows that Co 3p→3d absorption occurs at hν∼61 eV. Constant-initial-state (CIS) spectra at various EB ’s as shown in Fig. 4.7(b) indicate that the Co 3d partial density of
Co 2p3/2 (a) 2+
Zn0.95Co0.05O (AlKα) CoO (MgKα)
3+
Co3O4 (Co -Co mixed)
Co 2p
Intensity (arb. units)
Intensity (arb. units)
(b) 3+
LiCoO2 (Co )
2+
CoO (Co )
2p3/2 2p1/2
Zn0.95Co0.05O
Co (LMM)
810
800
790
780
770
Binding Energy (eV) 790
785
780
Binding Energy (eV) Figure 4.6: XPS spectra of the Co 2p core level. (a) Comparison of the Co 2p3/2 XPS spectrum of Zn0.95 Co0.05 O with those of several Co compounds: CoO [43], LiCoO2 [44], and Co3 O4 [44]. (b) Co 2p XPS spectra of Zn0.95 Co0.05 O and CoO [43]
22
Zn0.95Co0.05O
O 2p hν = 65 eV 63 62 61.5 61 60.5 60 59 58 57 55
12
8
4
0
Binding Energy (eV)
On-Resonance Off-Resonance
(c)
Intensity (arb. units)
40 50 60 70 80 Photon Energy (eV)
CIS Intensity (arb. units)
Intensity (arb. units)
Zn 3d
EB=7.0 EB=6.5 EB=5.0 EB=3.1 EB=1.5
(b)
Absorption (a.u.)
(a)
satellite Co 3d
Difference x10 50
55
60
65
70
8
6
4
2
0
Binding Energy (eV)
Photon Energy (eV)
Intensity (arb. units)
Intensity (arb. units)
Figure 4.7: (a) A series of photoemission spectra of Zn0.95 Co0.05 O for photon energies in the Co 3p→3d core-excitation region. Inset: Absorption spectrum recorded in the total electron yield mode. (b) Constant-initial states spectra at various EB ’s. Allows show enhancement of the photoemission intensity. (c) On-resonance (hν = 61.5 eV) and off-resonance (hν = 60.0 eV) spectra (see text for normalization). Difference between these spectra represents the Co 3d PDOS.
O 2p
hν =50eV hν =49eV
ZnO Zn0.95Co0.05O 10
8 6 4 2 0 Binding Energy (eV) Zn0.95Co0.05O hν = 61.5 eV
Co 3d
Co 3d PDOS x10
O 2p band
5
4
Band Gap of ZnO ~3.4 eV
3
2
1
0
Binding Energy (eV) Figure 4.8: On-resonance spectra and Co 3d PDOS (open circle) near EF . The dashed lines are the guide for eyes. Inset: VB spectrum of Zn0.95 Co0.05 O for hν = 50 eV (red color) and that of ZnO for hν = 49 eV (blue color).
23
states (PDOS) are primarily located at EB ∼3.0 eV and ∼ 7.0 eV. Figure 4.7(c) shows the Co 3d PDOS of Zn0.95 Co0.05 O, which has been obtained by subtracting the off-resonance spectrum (hν = 60 eV) from the on-resonance (hν = 61.5 eV) one. Here, the off-resonance spectrum was multiplied by the integrated intensity ratio (integration range: 0
(4.1)
where Q is the charge on the ion relative to that of the neutral atom, K is an empirical constant, and V is the Madelung constant. Therefore, in comparison between the elements having the same valencies in a chemically and structurally the same environment, the difference between core-level binding energies is dominantly caused by the EB (atom) term in Eqn.(4.1). EB (atom) increases with the atomic number of the element. Following this trend, it is seem more reasonable to consider that the EB (atom) of Co 3d is not near the main peak of Co 3d PDOS around the VB maximum but is near the satellite because the Zn 3d core level show a peak as deep as EB ∼10.0 eV. The main structure of the Co 3d PDOS is then considered to be charge-transfered states (Sec. 2.2). It is therefore expected
Figure 4.9: Electronic structure in VB expected from resonant PES.
24
the VB electronic structure as shown in Fig. 4.9. The satellite structure of Co 2p XPS also reflects the original core-level position and the main peak is due to charge-transfered states. Although the valency of the Co ion in ZnO is the same as CoO, the electronic structures would be different judged from the energy difference between the main peak and the satellite. Figure 4.10 shows the VB and CIS spectra of CoO for photon energies in the Co 3p→3d core-excitation region [43]. The energy difference between the main peak and the satellite of the Co 3d PDOS is as large ∼9 eV in CoO [43] while it is only ∼4 eV in Zn1−x Cox O, probably because of the different coordination of oxygen atoms (or crystal field symmetry) between CoO and Zn1−x Cox O. Indeed, the crystal field at the Co site in CoO has octahedral symmetry (Oh ) and that Zn site in ZnO has tetrahedral symmetry (Td ). Since the energy difference between the O 2p band maximum (lowest EB ) and EF is nearly equal to the band gap of ZnO as shown in Fig. 4.8, EF is supposed to be located near the conduction band minimum (composed of Zn 4s and possibly of Co d8 states), meaning that the sample is n-type. Although localdensity approximation (LDA) calculations have predicted that ferromagnetism is mediated by carriers and therefore needs a high density of states (DOS) at EF [9], we could not clearly observed a finite DOS at EF , probably due to the
Intensity (arb. units)
CoO valence band
O 2p
Co Main
(a) Co satellite hν = 80 eV 64 61 60 59 58 56 50 15
10
5
CIS intensity (arb. unit)
low carrier density.
Co3d Main Co3d satellite
(b)
40
0
Binding Energy (eV)
50
60
70
80
Photon Energy (eV)
Figure 4.10: A series of photoemission spectra of CoO for photon energies in the Co 3p→3d core-excitation region. (a) Energy distribution curves. (b) CIS spectra of the main peak and the satellite of Co 3d. [43]
25
4.3
CI cluster-model analysis for photoemission spectra
In order to explain the similarities and differences between the spectra of CoO and Zn1−x Cox O, we have applied CI cluster-model analysis to the Co 2p XPS and the Co 3d PDOS spectra (Fig. 4.12) and estimated the electronic structure parameters : the ligand-to-3d charge-transfer energy ∆, the d-d Coulomb interaction energy U, and the Slater-Koster parameter (pdσ) (a transfer integral between O 2p and Co 3d). In the CI cluster-model analysis, Racah parameters were fixed at the values of the free ion: B = 0.138 eV and C = 0.54 eV. The Co 3d-2p core hole Coulomb attraction energy Q is assumed to be related to U through U = βQ, where at β = 0.7. The ratio between (pdσ) and (pdπ) was fixed (pdσ)/(pdπ) = −2.16 [47]. It was assumed that the Co ions are in the highspin state and divalent in the octahedral crystal field for CoO ([Co2+ (O2− )6 ]−10 ) cluster). For Zn0.95 Co0.05 O, the valency of Co tetrahedrally coordinated by oxygen was assumed to be divalent (i.e. [Co2+ (O2− )4 ]6− cluster) and the electronic configuration was assumed to be the high-spin state. Figure 4.11 shows the results of the CI cluster-model analysis for CoO. Here, the Co 3d PDOS was simply the difference between the on-resonance (hν = 61 eV) and off-resonance (hν = 60 eV) spectra in Fig. 4.10 without normalization. Electronic structure parameters were estimated as ∆ = 5.2 eV, U = 6.9 eV, and (pdσ) = −1.1 eV. There are several reports about electronic structure parameters of transition-metal mono-oxides [48–50] as listed in Table. 4.1. Obtained electronic structure parameters for CoO are consistent with the previous report [50]. Figure 4.12 shows that the main structure of the spectra dominantly consist of charge-transferred states, i.e. cd8 L for the Co 2p XPS spectrum and d7 L for the Co 3d PDOS, where c and L denote a hole in Co 2p orbital and ligand 2p orbitals, respectively. Electronic structure parameters for Zn0.95 Co0.05 O were estimated ∆ = 5.0 eV, U = 6.0 eV, and (pdσ) = −1.6 eV. These parameters are in agreement with those obtained from the previous XAS study [51] and are consistent with the chemical trend in transition-metal-doped II-VI DMS [34] as listed in Table. 4.2. ∆ decreases from Ti to Ni due to the difference in the energy position of the transition-metal 3d orbitals. U increases with increasing transition-metal atomic number.
26
2p3/2
Experiment Cluster calc. Back Ground
(a)
2p1/2
Co 3d PDOS Cluster calc. Back Ground
Co 2p XPS
Valence Band Co 3d PDOS CoO
(b)
sat.
Intensity (arb. units)
sat.
cd
7
d
8
7
cd L
dL
9 2
8 2
cd L
810
800
790
780
6
dL
77015
10
5
0
-5
Binding Energy relative to VBM (eV)
Binding Energy (eV)
Figure 4.11: CI cluster-model analysis for CoO. (a) Co 2p XPS spectrum [43]. (b) Co 3d PDOS [43]. [Co2+ (O2− )6 ]−10 cluster was assumed. c and L denote a hole in a Co 2p orbital and a ligand 2p orbital. VBM stands for the valence-band maximum.
Table 4.1: Electronic structure parameters ∆ and U for transition-metal mono-oxides (in eV).
∆ CuO 2.8 NiO 6.2 CoO 5.5 FeO 7.0 MnO 8.8
U 3.5 6.9 5.4 5.7 4.0
Ref. [48] [48] [48] [48] [48]
∆ 3.0 4.7 6.5 7.0 8.0
U Ref. 7.8 [49] 7.3 [49] 6.5 [49] 6.0 [49] 5.5 [49]
27
∆ 2.2 5.0 5.2 5.5 7.0
U Ref. 8.8 [50] 7.5 [50] 6.9 this work 6.5 [50] 6.0 [50]
Experiment Calculation Back Ground
(a)
2p1/2
Valence Band Co 3d PDOS Zn0.95Co0.05O
(b)
sat. Co(LMM)
sat.
Intensity (arb. units)
Co PDOS Calculation Back Ground
Co 2p XPS
2p3/2
cd
7
d
7
8
dL
8 2
dL
cd L
8 2
cd L
810
800
790
780
6
770
8
Binding Energy (eV)
6
4
2
0
Binding Energy (eV)
Figure 4.12: CI cluster-model analysis for Co ions in Zn0.95 Co0.05 O. (a) Co 2p XPS spectrum. (b) Co 3d PDOS. [Co2+ (O2− )4 ]−6 cluster was assumed. c and L denote a hole in a Co 2p orbital and a ligand 2p orbital.
Table 4.2: Electronic structure parameters ∆, U , (pdσ) for substituted transition-metal atoms in II-VI semiconductors ZnSe, ZnS, and ZnO (in eV).
2+
Cr Mn2+ Fe2+ Co2+ Ni2+
ZnSe ∆ U (pdσ) Ref. 3.5 3.5 -1.1 [34] 2.5 4.0 -1.2 [34] 1.5 4.5 -1.05 [34] 1.0 5.0 -1.0 [34] 0.5 5.5 -1.1 [34]
ZnS ∆ 4.0 3.0 2.0 1.5 1.0
U (pdσ) Ref. 3.5 -1.2 [34] 4.0 -1.3 [34] 4.5 -1.2 [34] 5.0 -1.1 [34] 5.5 -1.2 [34]
28
ZnO ∆ 6.5 5.5 5.0 -
U 5.0 5.5 6.0 -
(pdσ) Ref. -1.6 [51] -1.6 [51] -1.6 this work -
Chapter 5 X-ray magnetic circular dichroism study 5.1
Experimental
Co 2p→3d (Co L2,3 edge) XMCD measurement was performed in the totalelectron yield mode at beam line BL23SU [52, 53] of SPring-8. A double array variable undulator, so-called APPLE-2 (advanced planar polarized light emitter) or Sasaki-type which produces both linearly and circularly polarized soft x-rays is used as a light source (Fig. 5.1) [54]. The superconducting magnet system as shown in Fig. 5.2 are equipped to apply the magnetic field to the sample at BL23SU. The monochromator resolution was E/∆E > 10000. Right-handed (µ+ ) and left-handed (µ− ) circularly polarized x-ray absorption spectra were obtained by reversing photon helicity at each photon energy. External magnetic field was applied perpendicular to the sample surface. Surface clearing was made in the same way as in the photoemission experiments. Circularly polarized x-ray
Figure 5.1: Magnetic structure of the APPLE II consists of Figure 5.2: Schematic view of the superconducting magnet two pairs of arrays of permanent magnets. [52] system. [53]
29
absorption spectra under each experimental conditions have been normalized to the Co L2,3 edge XAS [= (µ+ + µ− )/2] area. The background of the XAS spectra was assumed to be a hyperbolic tangent function as shown in Fig. 5.3.
Absorption (arb. units)
Figure 5.3: Schematic optical layout of the BL23SU at SPring-8. [55] Zn1-xCoxO (x=0.05) Co L2,3 edge
2.0 +
1.5
µ
1.0
µ Temperature : 20 K Back Ground
−
Magnetic Field : 4.5 T
0.5 0 770
775
780
785
790
795
800
Photon Energy (eV)
Figure 5.4: µ+ and µ− spectra of Zn0.95 Co0.05 O. Background is assumed to be a hyperboric tangent function. (a)
Zn0.95Co0.05O Co L2,3
XAS (arb. units)
3.0
(b)
Co L3
H=2.0 T, T=20 K H=4.5 T, T=20 K H=7.0 T, T=20 K H=7.0 T, T=100 K H=7.0 T, T=220 K
2.5 2.0 1.5 1.0 0.5 0 770
775
780
785
790
795
Photon Energy (eV)
800 776
778
780
782
Photon Energy (eV)
Figure 5.5: L2,3 XAS spectra of Zn0.95 Co0.05 O. (a) Normalized XAS spectra under various experimental conditions. (b) Enlarged set of data at the Co L3 edge.
30
5.2
Magnetic field and temperature dependences of x-ray magnetic circular dichroism spectra
Figure 5.6(a) shows the Co L2,3 edge XMCD spectra (µ+ − µ− ) of Zn0.95 Co0.05 O taken at 20 K under various applied magnetic fields (H’s). One can see that the intensity of XMCD increases with applied magnetic field keeping the same line shape. Figure 5.6(b) shows the difference spectra of these XMCD spectra. The difference of the intensity of XMCD between H = 4.5 and 2.0 T is almost the same as that between H = 7.0 and 4.5 T, and is half of that between H = 7.0 and 2.0 T. Thus, the intensity of XMCD increases linearly with the magnetic field. However, extrapolating the XMCD intensity to H = 0 T, there is a finite intensity, corresponding to the ferromagnetic behavior of our sample (see Fig. 5.11). XMCD spectra under various magnetic fields and the difference of these spectra have identical structure as shown in Fig. 5.6(c), suggesting that all the magnetic signals come from Co ions in the same chemical and structural en-
(a) (c) H = 2.0 T H = 4.5 T H = 7.0 T
-0.05
-0.10 Co L3 777
-0.15
XMCD (arb. units)
20
778
779
780
Diff. (4.5T-2.0T) Diff. (7.0T-4.5T) Diff. (7.0T-2.0T)
(b)
H = 2.0 T
Zn0.95Co0.05O Co L2,3 at 20 K
781
Co L2,3
0 -20
H = 4.5 T
Normalized XMCD (arb. units)
XMCD (arb. units)
0
H = 7.0 T
Diff. (4.5T-2.0T)
Diff. (7.0T-4.5T)
-40 Co L3
-60 777
-80x10
778
779
780
Co L3
781
Diff. (7.0T-2.0T)
-3
775
780
785
790
795
Photon Energy (eV)
800
776
778
780
782
Photon Energy (eV)
Figure 5.6: (a) XMCD spectra under various applied magnetic fields at T = 20 K. (b) Difference XMCD spectra between different magnetic fields. Inset: Enlarged set at the Co L3 edge. (c) XMCD spectra normalized to the Co L3 peak area.
31
Co L2,3
Zn0.95Co0.05O Co metal
T = 20 K T = 100 K T = 220 K
Zn0.95Co0.05O Co L2,3 at 7.0 T
XMCD (arb. units)
XMCD (arb. units)
0
-0.05
-0.10
Co L3
Co L3 777
-0.15 775
780
778
785
779
780
790
795
776
781
800
775
Photon Energy (eV)
780
778
785
780
790
795
782
800
Photon Energy (eV)
Figure 5.7: Temperature dependence of XMCD Figure 5.8: XMCD spectra compared with spectra at H = 7.0 T. that of Co metal [56].
vironment. Figure 5.7 shows XMCD spectra under various temperature (T s) at H = 7.0 T, indicating that the temperature dependence of XMCD is small. This behavior shall be discussed in Sec. 5.4. All the XMCD spectra show multiplet structure both at the L3 and the L2 edges, unlike those of Co metal as compared in Fig. 5.8. This indicates that the ferromagnetism in Zn1−x Cox O is not caused by Co metal cluster.
5.3
Comparison with calcualted spectra
To study the valence and the crystal field for the Co ion, the XAS and XMCD spectra are compared with spectra calculated using atomic multiplet theory in Fig. 5.9. All calculations for XAS and XMCD spectra have been practiced by A. Tanaka. The calculation was carried out for Co2+ and Co3+ with the positive and negative 10Dq representing the octahedral symmetry (coordinated by six oxygen) and tetrahedral symmetry (coordinated by four oxygen) at the Co site, respectively. The calculated multiplet splitting for Co3+ is more spread than the experimental one. The calculated ones for Co2+ with 10Dq = −0.7 eV best reproduces the experimental spectra of both XAS and XMCD. To obtain more details in the XAS and XMCD line shape, cluster-model calculation have been performed as shown in Fig. 5.10. We have used electronic structure parameters of ∆ = 5.0 eV, U = 5.0 eV, (pdσ) = −1.6 eV, almost the same as those estimated from the PES experiments. This result gives further evidence that the substituted Co ions are divalent under tetrahedral crystal field and responsible 32
for the ferromagnetism in Zn1−x Cox O. Zn0.95Co0.05O Co L2,3 edge at H=4.5 T, T=20 K
(a)
3+
(b)
7
Co (d ), 10Dq=+0.5 eV
2+
3+
8
2+
7
2+
7
Co (d ), 10Dq=-0.5 eV
XMCD (arb. units)
XAS (arb. units)
8
Co (d ), 10Dq=-0.5 eV
2+
8
8
Co (d ), 10Dq=+0.5 eV
3+
3+
Co (d ), 10Dq=+0.5 eV
Co (d ), 10Dq=+0.5 eV
Co (d ), 10Dq=-0.7 eV
7
Co (d ), 10Dq=-0.7 eV
Experiment
Experiment 775
780
785
790
795
800
775
Photon Energy (eV)
780
785
790
795
800
Photon Energy (eV)
Figure 5.9: Experimental XAS (a) and XMCD (b) spectra (open circle) compared with atomic multiplet calculation (solid curves), in which the valency of Co and the sign and magnitude of the crystal-field splitting are varied.
XAS
Zn0.95Co0.05O Co L2,3 edge
XAS (arb. units)
Experiment Cluster Calc. Co L3
XMCD (arb. units)
776 778 780 782
Co L3
776
775
780
778
785
780
782
790
XMCD 795
800
Photon Energy (eV) Figure 5.10: CI cluster-model analysis of XAS and XMCD spectra at the Co L2,3 edge.
33
5.4
Results of XMCD sum rules and discussion
By applying XMCD sum rules to the XAS and XMCD spectra, the orbital magnetic moment (Morb ) and the spin magnetic moment (Mspin ) have been estimated [56] as shown in Fig. 5.11. Magnetic field and temperature dependences of the total magnetic moment (Mtot = Mspin + Morb ) is the same as that of the XMCD intensity. On the other hand, Morb remains small and dose not change under various conditions, due to the quenching of orbital moment in a crystal field. This is consistent with the high spin Co2+ ion in the crystal field of tetrahedral oxygen coordination and not of octahedral one since Co2+ should have a large orbital moment in the octahedral crystal field. Mtot at H = 0.0 T obtained by extrapolating the XMCD intensity from high magnetic field is small compared with that estimated from SQUID (see Fig. 3.1(b)). The origin of this discrepancy is not known at present. Since Mtot has a component which increases linearly with the magnetic field and is independent of temperature, we conclude that Co ions with temperatureindependent paramagnetism and ferromagnetic Co ions coexist, although the two kinds of Co ions have the same valence and indistinguishable crystal-field splitting. The temperature-independent paramgnetism indicates that there are magnetic interaction between Co ions because, if Co ions do not interact with each other, they may behave as isolated magnetic ions, which show paramganetism
Average Moment (µB/Co)
0.16
(a) Morb Mspin Mtot
0.14 0.12
(b)
0.10
Morb Mspin Mtot
0.08 at T=20 K
0.06
at H=7.0 T
0.04 0.02 0 0
2
4
6
0
Magnetic Field (T)
50
100 150 200
Temperature (K)
Figure 5.11: Magnetic field (a) and temperature (b) dependences of average magnetic moment estimated from XMCD sum rules. The dashed line is the guide for eyes.
34
obeying the Curie law. While the ferromagnetic Co ions are minority, the remaining majority Co ions may be couple antiferromagnetically with each other. The temperature-independent paramagnetic component may arise from the susceptibility of antiferromagnetic Co ions having a N´eel temperature above the room temperature. Such antiferromagnetic moments would give rise to a magnetic susceptibility of χexp ∼ 1.43 × 10−2 (µB /T per Co), because the magnetic susceptibility is estimated to about g 2 S(S +1)/TN ∼ 8.4×10−3 (µB /T per Co) for g = 2, S = 3/2, and TN = 400 K. The paramagnetism originating from the Pauli paramagnetism of the conduction electrons is seem to be negligible because its value ∼ 2.5 × 10−6 (µB /T per Co) estimated from one-electron approximation is several orders of magnitude smaller than the experimental value. It is hard to consider that this temperature-independent paramagnetism originate from van Vleck paramagnetism in insulator or orbital paramagnetism in metal, both which are caused by orbital polarization effect, because Morb estimated from XMCD sum rules is small. In order to clarify the origin of the “paramagnetic” signal, PES and XMCD experiments on samples with various carrier concentrations showing larger magnetization are necessary.
35
Chapter 6 Summary In summary, we have performed Co 3p→3d resonant PES, XPS, XAS, and XMCD experiments on the diluted ferromagnetic semiconductor n-type Zn1−x Cox O (x = 0.05). The Co 3d PDOS obtained by Co 3p→3d resonant PES shows a peak at EB ∼3.0 eV and a satellite at EB ∼7.0 eV. By applying CI cluster-model analysis to the photoemission spectra, we have estimated electronic structure parameters for Zn1−x Cox O which are consistent with the chemical trend in II-VI DMS and confirmed the difference of electronic structure between Zn1−x Cox O and CoO. The XMCD spectra show a multiplet structure, characteristic of the Co2+ ion tetrahedrally coordinated by oxygen. The magnetic field and temperature dependences of the XMCD intensity suggest that there are magnetic interaction between Co ions.
37
Acknowledgements First of all, I would like to express my sincere gratitude to Prof. Atsushi Fujimori, who has given me this work, for his lot of valuable advice and enlightening discussion. I also extend my hearty thanks to Prof. Takashi Mizokawa for indispensable discussion. In particular, I owe a great deal in the cluster-model calculation to him. I am much obliged to Dr. J. Okamoto and Dr. K. Mamiya for helpful advice about this work. The experiments at Photon Factory and SPring-8 were supported by a number of people. I am particularly indebted to the members of Kinoshita group, Dr. T. Okuda, Ms. A. Harasawa, Dr. T. Wakita, and Prof. T. Kinoshita, for their valuable technical support during the beamtimes at Photon Factory. I would like to make my acknowledgement to Dr. Y. Takeda, Dr. T. Okane, Dr. Y. Saitoh, and Prof. Y. Muramatsu for their helpful technical support during the beamtimes at SPring-8. I am deeply thankful to Dr. H. Saeki, Prof. H. Tabata, and Prof. T. Kawai for providing me with such interacting and excellent samples of Zn1−x Cox O with valuable advice, and suggestions. I express to Dr. A. Tanaka my cordial thanks for his theoretical support about CI cluster-model analysis of the x-ray absorption spectra and enlightening discussions. I was encouraged by collaboration with Prof. D. D. Sarma on a related subject. Thanks are also due to Dr. S. Ray. I am very grateful to the members of Fujimori-Mizokawa Group. Mr. Yukiaki Ishida provided me with useful advice about this work and helped me at Photon Factory and SPring-8. I was always encouraged by his attitude toward the research. Mr. J. I. Hwang has taught me how to use the photoemission instruments at the beam lines and always answered my questions about DMS. Mr. H. Wadati taught me how to use and condition the photoemission spectrometer. Mr. M. Takizawa always cooperated with me in the maintenance of the photoemission spectrometer. Mr. Y. Osafune assisted me in the experiments at the beam lines. 39
I would also like to thank Dr. T. Yoshida, Dr. J.-Y. Son, Dr. J. Quilty, Mr. K. Tanaka, Mr. H. Yagi, Mr. D. Asakura, Mr. T.-T. Tran, Mr. K. Ebata, Mr. M. Hashimoto, Mr. A. Shibata, Mr. Y. Fujii, Mr. K. Takubo, Mr. M. Ikeda, Mr. S. Hirata, Ms. R. Toriumi, Ms. Y. Shimazaki, and Ms. A. Fukuya for a lot of encouragement and supports. Finally, I wish to extend special thanks to my family and friends for their various supports. January 2005 Masaki Kobayashi
40
References [1] H. Ohno, H. Munekata, T. Penny, S. von Moln´ar, and L. L. Chang, Phys. Rev. Lett. 68, 2664 (1992). [2] H. Ohno, A. Shen, F. Matsukura, A. Oiwa, A. Endo, S. Katsumoto, and Y. Iye, Appl. Phys. Lett. 69, 363 (1996). [3] H. Ohno and F. Matsukura, Solid State Commun. 117, 179 (2001). [4] S. Koshihara, A. Oiwa, M. Hirosawa, S. Katsumoto, Y. Iye, C. Urano, H. Takagi, and H. Munekata, Phys. Rev. Lett. 78, 4617 (1997). [5] Y. Ohno, D. K. Young, B. Beschoten, F. Matsukura, H. Ohno, and D. D. Awschalom, Nature 402, 790 (1999). [6] D. Chiba, M. Yamanouchi, F. Matsukura, and H. Ohno, Science 301, 943 (2003). [7] M. Yamanouchi, D. Chiba, F. Matsukura, and H. Ohno, Nature 428, 539 (2004). [8] T. Dietl, H. Ohno, F. Matsukura, J. Cibert, and D. Ferrand, Science 287, 1019 (2000). [9] K. Sato and H. Katayama-Yoshida, Jpn. J. Appl. Phys. 40, L334 (2001). [10] C. Zener, Phys. Rev. 81, 440 (1950). [11] C. Zener, Phys. Rev. 82, 403 (1951). [12] S. J. Pearton, W. H. Heo, M. Ivill, D. P. Norton, and T. Steiner, Semicond. Sci. Technol. 19, R59 (2004). [13] Y. Matsumoto, M. Murakami, T. Shono, T. Hasegawa, T. Fukumura, M. Kawasaki, P. Ahmet, T. Chikyow, S. Y. Koshihara, and H. Koimura, Science 291, 854 (2001). 41
[14] H. Toyosaki, T. Fukumura, Y. Yamada, K. Nakajima, T. Chiyow, T. Hasegawa, H. Koimuna, and M. Kawasaki, Nat. Mater. 3, 221 (2004). [15] S. B. Ogale, R. J. Choudhary, J. P. Buban, S. E. Lofland, S. R. Shinde, S. N. Kale, V. N. Kulkarni, J. Higgins, C. Lanci, J. R. Simpson, N. D. Browning, S. D. Sarma, H. D. Drew, R. L. Greene, and T. Venkatesan, Phys. Rev. Lett. 91, 077205 (2003). [16] P. Sharma, A. Gupta, K. V. Rao, F. J. Owens, R. Sharma, R. Ahuja, J. M. O. Guillen, B. Johanssion, and G. A. Gehring, Nat. Mater. 2, 673 (2003). [17] H. Saeki, H. Tabata, and T. Kawai, Solid State Commun. 120, 439 (2001). [18] K. Ueda, H. Tabata, and T. Kawai, Appl. Phys. Lett. 79, 988 (2001). [19] K. Sato and H. Katayama-Yoshida, Semicond. Sci. Technol. 17, 367 (2002). [20] E. C. Lee and K. J. Chang, Phys. Rev. B 69, 085205 (2004). [21] A. C. Tuan, J. D. Bryan, A. B. Pakhomov, V. Shutthanandan, S. Thevuthasan, D. E. McCready, D. Gaspar, M. H. Engelhard, J. W. Rogers, Jr., K. Krishnan, D. R. Gamelin, and S. A. Chambers, Phys. Rev. B 70, 054424 (2004). [22] F. Tuomisto, V. Ranki, K. Saarinen, and D. C. Look, Phys. Rev. Lett. 91, 205502 (2003). [23] A. Tsukazaki, A. Ohtomo, T. Onuma, M. Ohtani, T. Makino, M. sumiya, K. Ohtani, S. F. Chichibu, S. Fuke, Y. Segawa, H. Ohno, H. Koimura, and M. Kawasaki, Nat. Mater. 4, 42 (2005). [24] J. H. Kim, H. Kim, D. Kim, Y. E. Ihm, and W. K. Choo, J. Euro. Ceram. Soc. 24, 1847 (2004). [25] C. F. J. Flipse, C. B. Rouwelaar, and F. M. F. de Groot, Eur. Phys. J. D 9, 479 (1999). [26] H. J. Lee, G. H. Ryu, S. K. Kim, S. A. Kim, C. H. Lee, S. Y. Jeong, and C. R. Cho, Phys. Stat. Sol. 241, 2858 (2004). [27] D. A. Schwartz, N. S. Norberg, Q. P. Nguyen, J. M. Parker, and D. R. Gamelin, J. Am. Chem. Soc. 125, 13205 (2003). [28] M. Venkatesan, C. B. Fitzgerald, J. G. Lunney, and J. M. D. Coey, Phys. Rev. Lett. 93, 177206 (2004). 42
[29] T. Mizokawa, T. Nambu, A. Fujimori, T. Fukumura, and M. Kawasaki, Phys. Rev. B 65, 085209 (2002). [30] D. J. Keavney, D. Wu, J. W. Freeland, E. Johnston-Halperin, D. D. Awschalom, and J. Shi, Phys. Rev. Lett. 91, 187203 (2003). [31] T. A. Koopmans, Physica 1, 104 (1933). [32] U. Fano, Phys. Rev. 124, 1866 (1961). [33] T. Mizokawa and A. Fujimori, Phys. Rev. B 48, 14150 (1993). [34] T. Mizokawa and A. Fujimori, Phys. Rev. B 56, 6669 (1997). [35] S. Yamamoto, H. Kawata, H. Kitamura, M. Ando, N. Saki, and N. Shiotani, Phys. Rev. Lett. 62, 2672 (1989). [36] H. Onuki, N. Saito, and T. Saito, Appl. Phys. Lett. 52, 173 (1988). [37] B. T. Thole, P. Carra, F. Sette, and G. van der Laan, Phys. Rev. Lett. 68, 1943 (1992). [38] P. Carra, B. T. Thole, M. Altatelli, and X. Wang, Phys. Rev. Lett. 70, 694 (1993). [39] H. Saeki, H. Matsui, T. Kawai, and H. Tabata, J. Phys.: Condens. Mat. 16, S5533 (2004). [40] S. Hashimoto and A. Tanaka, Surf. Interface Anal. 34, 262 (2002). [41] S. Hashimoto, K. Hirokawa, Y. Fukuda, K. Suzuki, T. Suzuki, N. Usuki, N. Gennai, S. Yoshida, M. Koda, H. Sezaki, H. Horie, A. Tanaka, and T. Ohtsubo, Surf. Interface Anal. 18, 799 (1992). [42] J. I. Hwang, Master Thesis, University of Tokyo (2004). [43] Z. X. Shen, J. W. Allen, P. A. P. Lindberg, D. S. Dessau, B. O. Wells, A. Borg, W. Ellis, J. S. Kang, S. J. Oh, I. Lindau, and W. E. Spicer, Phys. Rev. B 42, 1817 (1990). [44] J. van Elp, J. L. Wieland, H. Eskes, P. Kuiper, G. A. Sawatzky, F. M. F. de Groot, and T. S. Turner, Phys. Rev. B 44, 6090 (1991). [45] S. C. Wi, J. S. Kang, J. H. Kim, S. B. Cho, B. J. Kim, S. Yoon, B. J. Suh, S. W. Han, K. H. Kim, K. J. Kim, B. S. Kim, H. J. Song, H. J. Shin, J. H. Shim, and B. I. Min, Appl. Phys. Lett. 84, 4233 (2004). 43
[46] S. H¨ ufner, Photoelectron Spectroscopy (Springer-Verlag, Berlin, 2003). [47] W. A. Harrison, Electronic Structure and the Properties of Solids (Dover, New York, 1989). [48] J. van Elp, PhD Thesis, University of Groniingen . [49] A. Tanaka and T. Jo, J. Phys. Soc. Jpn. 63, 2788 (1994). [50] A. Fujimori, A. E. Bocquet, T. Saitoh, and T. Mizokawa, J. Electron Spectrosc. Relat. Phenom. 62, 141 (1993). [51] J. Okabayashi, K. Ono, M. Mizuguchi, M. Oshima, S. S. Gupta, D. D. Sarma, T. Mizokawa, A. Fujimori, M. Yuri, C. T. Chen, T. Fukumura, M. Kawasaki, and H. Koinuma, J. Appl. Phys. 95, 3573 (2004). [52] SPring-8 Beamline Handbook 112 (2003). [53] J. Okamoto, K. Mamiya, S. I. Fujimori, T. Okane, Y. Saitoh, Y. Muramatsu, A. Fujimori, S. Ishikawa, and M. Takano, AIP Conference Proceedings 705, 1110 (2004). [54] S. Sasaki, Nucl. Inst. Meth. A 347, 83 (1994). [55] Y. Saitoh, T. Nakatani, T. Matsushita, A. Agui, A. Yoshigoe, Y. Teraoka, and A. Yokoya, Nucl. Inst. Meth. A 474, 253 (2001). [56] C. T. Chen, Y. U. Idzerda, H. J. Lin, N. V. Smith, G. Meigs, E. Chaban, G. H. Ho, E. Pellegrin, and F. Sette, Phys. Rev. Lett. 75, 152 (1995). [57] Y. Teramura, A. Tanaka, and T. Jo, J. Phys. Soc. Jpn. 65, 1053 (1996).
44
