Influences of Oxygen Partial Pressure on the Cu Metal Clustering

The influences of oxygen partial pressure on the local structure, magnetic, and transport properties of (In0.89Fe0.06Cu0.05)2O3 films deposited by dif...
1 downloads 0 Views 3MB Size
Article Cite This: J. Phys. Chem. C XXXX, XXX, XXX−XXX

pubs.acs.org/JPCC

Influences of Oxygen Partial Pressure on the Cu Metal Clustering and Room Ferromagnetism of the (In0.89Fe0.06Cu0.05)2O3 Films Xianke Sun,† Zhen Lin,‡ Zhonghua Wu,§ and Yukai An*,‡ †

College of Physics and Telecommunication Engineering, Zhoukou Normal University, Zhoukou 466001, China Key Laboratory of Display Materials and Photoelectric Devices, Ministry of Education; Tianjin Key Laboratory for Photoelectric Materials and Devices; School of Material Science and Engineering, Tianjin University of Technology, Tianjin 300384, China § Beijing Synchrotron Radiation Facility (BSRF), Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049, China Downloaded via DURHAM UNIV on August 5, 2018 at 06:52:11 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



ABSTRACT: The influences of oxygen partial pressure on the local structure, magnetic, and transport properties of (In0.89Fe0.06Cu0.05)2O3 films deposited by different Ar−O2 flow rate (10:0, 10:0.4, and 10:3.0) were studied systematically by X-ray diffraction (XRD), X-ray photoelectron spectroscopy (XPS), X-ray absorption spectroscopy (XAS), Hall effect, room temperature (RT) magnetic measurements, and first-principles calculations. The detailed structural analyses indicate that the doped Fe atoms substitute for In3+ sites of the In2O3 lattices with +2 and +3 mixture valences for all the films. However, the majority of codoped Cu atoms form the Cu metal clusters in the film deposited at oxygen flow rate of 0 sccm. With increasing oxygen partial pressure, the Cu metal clusters completely disappear and all the Cu atoms are incorporated substitutionally into the In2O3 lattices at In sites. All the films exhibit intrinsic RT ferromagnetism, and Mott variable range hopping (VRH) transport behavior. The saturation magnetization (Ms) of the films monotonically decreases with increasing oxygen partial pressure, implying that the Ms strongly correlates with the concentration of oxygen vacancies. The first-principles calculations suggest that the oxygen vacancy can act to mediate the superexchange interaction between Fe atoms, leading to a ferromagnetic ground state in Fe-doped In2O3 system. However, the Cu codoping cannot be responsible for the observed ferromagnetic ordering in Fe/Cu codoped In2O3 system. The stable ferromagnetic ground state only can be obtained by the coexistence of oxygen vacancy and Cu codoping in the system. Therefore, it can be concluded that the bound magnetic polarons (BMP) involving oxygen vacancy defects may be responsible for the observed intrinsic ferromagnetic coupling, which can be remarkably influenced by the localization effect of carriers in the (In0.89Fe0.06Cu0.05)2O3 films. These results provide a new sight for designing and manipulating the magnetic interaction of In2O3-based DMS systems.

1. INTRODUCTION In recent years, diluted magnetic semiconductors (DMSs) have been actively studied as promising materials for developing spintronic devices by effectively utilizing the spin and charge degrees of freedom simultaneously.1−3 Since the theoretical prediction of room temperature (RT) ferromagnetic ordering in wide-band-gap semiconductors by Dietl et al.,4 a great deal of oxide-based DMSs such as ZnO,5,6 TiO2,7,8 SnO2,9 and In2O310 have evolved and are expected to realize an intrinsic ferromagnetism above RT for practical applications. Indium oxide (In2O3) is a wide-band-gap (3.75 eV) oxide semiconductor with high optical transparency and high electric conductivity by Sn doping or introducing oxygen vacancies and has become a promising host material of DMSs for magnetooptical and -electronic devices. At present, transition-metal (TM)-doped In2O3-based DMSs have been investigated by many research groups,11,12 especially for Fe-doped In2O3 due to the good homogeneous and high solubility of Fe ions (∼20%) into the In2O3 lattice.13,14 Codoping has been reported to enhance the ferromagnetic coupling of the DMSs © XXXX American Chemical Society

by adjusting the position and occupancy of the Fermi level (EF).15−17 Cu has been considered as a good codopant in TMdoped In2O3-based DMSs due to the nonferromagnetic nature of Cu metal and Cu oxides (Cu2O and CuO), which can avoid the extrinsic ferromagnetism arising from the magnetic impurities brought by Cu itself. Recently, the codoping of Cu element into Fe doped In2O3 systems has received substantial interest due to the possibility of tailoring the band structures as well as tuning magnetic and transport properties by the interactions of the Cu/Fe codopants. Yoo et al. prepared the ferromagnetic Fe/Cu codoped In2O3 bulk ceramics with the Curie temperature of 750 K and confirmed that the observed magnetism derives from magnetic impurity phases.18 He et al. deposited RT ferromagnetic Fe doped In2O3 films with a small amount of Cu codoping (2 at. %). They found that the codoped Cu creates the mixed valence Received: May 27, 2018 Revised: July 20, 2018 Published: July 23, 2018 A

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

extend X-ray absorption fine structure (EXAFS), which were employed in the total fluorescence mode at 4B9A beamline of X-ray diffraction station in Beijing Synchrotron Radiation Facility (BSRF). The energy resolution ΔE/E was estimated to be about 10−4. This incident beam was collimated by slits and monochromatized by the Si (111) crystal detuned from 40% to 50% to suppress higher order harmonic content from the beam. A superconducting quantum interference device (SQUID) magnetometer was used for the magnetic measurements as a function of magnetic field (0 to ±2 T). The film resistivity (ρ) and Hall effect measurements were performed by conventional the van der Pauw four-probe configuration with a closed cycle cryo-cooler with the applied magnetic fields up to 1.1 T and temperature range of 10−300 K. The first-principles calculations on the total energy and electronic structure were performed in a supercell of In2O3 with cubic bixbyite structure using the MedeA-VASP code. The exchange correlation functional was approximated by the Perdew−Burke−Ernzerhof (PBE) general gradient approximation (GGA). A kinetic energy cutoff of 500 eV was considered for the plane-wave expansion. The lattice constants were optimized and obtained as a = b = c = 10.29 Å, which are in good agreement with experimental values.25 The Brillouin zone integrations were performed by 2 × 2 × 2 and 3 × 3 × 3 kpoint meshes within the Monkhorst−Pack scheme for genmetry optimization and electronic structure calculations, respectively, which make sure the total energy was relatively stable. The structures were fully relaxed until the maximum force acting on each atom and the total energy change between succeeding iterations were less than 0.02 eV/Å and 1.0 × 10−5 eV/atom, respectively.

cations of Fe2+/Fe3+, which are responsible for the observed RT ferromagnetism.19 Ho et al. found that the RT ferromagnetism of bulk Fe doped In2O3 decreases monotonically with increasing Cu concentration, which may originate from the mediated effects of Cu in Fe−Fe interactions.20 However, Chu et al. prepared the Fe/Cu codoped In2O3 nanocrystals and showed no evidence of RT ferromagnetic ordering by Cu codoping.21 Jiang et al. grew RT ferromagnetic Fe/Cu codoped In2O3 films by pulsed laser deposition and confirmed that the Cu codoping results in the forming of Fe metal clusters, which are responsible for the observed ferromagnetic ordering in the films.22 On the other hand, a great number of experimental investigations have confirmed that the existence form of TM dopants in In2O3 is also sensitive to the preparation methods, growth temperature, oxygen partial pressure, etc.23,24 The change of preparation conditions can result in different local structural characteristics around TM dopants, such as precipitation of TM-related metal or oxides, which can remarkably influence the magnetic properties of samples. Therefore, despite many reports on the advantages of codoping in improving the magnetic properties of the Fe/Cu codoped In2O3 systems, our understanding about the structural nature of codopants and the origin of ferromagnetic ordering is still far from satisfactory and needs to be further clarified. These are very critical to understand how to control the magnetic properties of In2O3-based DMSs by codoping of TM ions. In this article, we introduce Fe and Cu into the In2O3 lattice to fabricate Fe/Cu codoped In2O3based DMSs by the magnetron sputtering technique. Systematical experimental and theoretical studies on the local structures around Fe and Cu dopants as well as magnetic and transport properties of (In0.89Fe0.06Cu0.05)2O3 films deposited at different oxygen partial pressures were carried out. We aim to clarify the influences of oxygen partial pressure on the charge carriers, Cu metal clustering, oxygen vacancy defects, and ferromagnetic coupling in the films. These results give new insights into understanding the mechanism of magnetic interactions in the Fe/Cu codoped In2O3 films.

3. RESULTS AND DISCUSSION Figure 1 shows the XRD patterns of (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm.

2. EXPERIMENTAL DETAILS The (In0.89Fe0.06Cu0.05)2O3 films with a thickness of 1400 nm were fabricated on Si (001) substrates by the RF-magnetron sputtering technique. The sputter chamber was preliminarily evacuated to a background pressure of 8 × 10−5 Pa. The substrate temperature was set to 400 °C. To introduce different oxygen partial pressures, the (In0.89Fe0.06Cu0.05)2O3 films were sputtered in 0.9 Pa of pure Ar and O2 mixed gases (purity 99.999%). The argon flow rate of 10 sccm was kept constant for all the films, and the oxygen flow rate was changed to 0, 0.4, and 3 sccm. The microstructure analysis and phase identification were performed by the X-ray diffraction (XRD) technique with a θ−2θ scan, in which Cu Kα X-rays (λ = 0.15406 nm) were employed. The composition and valence state of the films were measured by X-ray photoelectron spectroscopy (XPS) which was performed on a PHI-1600 photoelectric spectrometer (Mg Kα, X-rays), and the corresponding nonstoichiometric formula for the films deposited at oxygen flow rate of 0, 0.4, and 3 sccm was determined to be (In0.89Fe0.06Cu0.05)2O2.76, (In0.89Fe0.06Cu0.05)2O2.85, and (In0.89Fe0.06Cu0.05)2O2.90, respectively. The local structures of Fe and Cu dopants were investigated by the Fe and Cu K-edge XAS measurements including X-ray absorption near-edge structure (XANES) and

Figure 1. XRD patterns of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm. The inset shows the enlarged view of (222) diffraction peaks.

Clearly, all the films show the same cubic bixbyite structure of In2O3 lattice with a preferred (222) orientation. Bragg peaks belonging to Fe and Cu cluster segregations, Fe and Cu oxides, or any other impurity phases were not be observed within the XRD detection limit, indicating that the Fe and Cu codoping does not change the cubic bixbyite structure belong to the space group Ia3̅. To investigate the influences of oxygen partial pressure, a detailed analysis of the position of diffraction peaks B

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 2. X-ray photoelectron spectra of (a) O 1s, (b) Fe 2p, (c) Cu 2p, and (d) Cu LMM for the (In0.89Fe0.06Cu0.05)2O3 film deposited at oxygen flow rate of 0 and 3 sccm.

the spectra have been calibrated by the C 1s peak at 284.6 eV. The O 1s XPS spectra are shown in Figure 2a, which can be best fitted by two peaks using Gaussian fitting, I and II. The strong peak I can be attributed to In2O3 lattice oxygen (In−O bond in In2O3), while the weak peak II on the higher binding energy side is assigned to the oxygen ions in the oxygen vacancy region of In2O3 matrix.27 The relative intensity of the peak II indicates that there exists a number of oxygen vacancies in the films. The intensity of peak II remarkably decreases with increasing oxygen partial pressure, which is consistent with the decrease in the concentration of oxygen vacancies. The binding energy of Fe 2p3/2 is located at 710.3 eV, which is between Fe2+ (709.9 eV) and Fe3+ (711.0 eV), as shown in Figure 2b. This implies that the Fe element exists in the ionic form with a mixed valence of +2 and +3 in the In2O3 matrix. Figure 2c shows the Cu 2p XPS spectra. One can see that the binding energies of Cu 2p3/2 for the films deposited at oxygen flow rate of 0 and 3 sccm are the same and located at 932.9 eV. As is well-known, the binding energy of Cu2+ atoms is located at 934.2 eV, and binding energy of Cu+ or Cu0 atoms is in the range 932.8−933.1 eV.28 The binding energies for Cu+ and Cu0 are almost the same, and their difference must depend on the AES spectrum. To evaluate the actual valence state of doped Cu ions for the two films, the corresponding Cu LMM spectra were measured and shown in Figure 2d. Clearly, the AES peak centered at 568 eV for the film with oxygen flow rate of 0 sccm is attributed to Cu0. However, with increasing oxygen flow rate to 3 sccm, the AES peak shifts to 570.6 eV, which corresponds to Cu+.29 These strongly suggest that the Cu0 atoms can be transformed into Cu+ ions by the increase of oxygen partial pressure. The XANES/EXAFS techniques have become a powerful method to understand the local structure of TM doped oxide

was done. A shift in the (222) diffraction peaks to lower 2θ angles can be distinctly observed with increasing oxygen partial pressure, as shown in the inset of Figure 1, suggesting a dspacing expansion, namely a gradual increase in the average lattice parameter. As is well-known, the increase of oxygen partial pressure during deposition can effectively decrease the concentration of oxygen vacancies in the In2O3 lattices and also change the oxidation valence states from Fe2+ to Fe3+. The decrease in the concentration of oxygen vacancies can cause the d-spacing expansion, but the increase in the number of Fe3+ ions has also the opposite effect because Fe3+ ion (RFe3+ = 0.79 Å) has a smaller atomic radius than Fe2+ ion (RFe2+ = 0.92 Å) and In3+ (RIn3+ = 0.94 Å).26 Therefore, these results imply that the change in the concentration of oxygen vacancies plays an important role on tuning the average lattice parameter of the films. Moreover, it is also observed that the intensity of (222) diffraction peaks remarkably increases and the line width becomes narrower with increasing oxygen partial pressure. The mean crystallite size, D, of the (In0.89Fe0.06Cu0.05)2O3 films is 0.9λ estimated using the Debye−Scherrer equation: D = β cos θ , where λ is the wavelength of X-ray radiation (λ = 0.15406 nm), θ is the Bragg angle, and β is the full width at half-maximum. The corresponding mean crystallite size was calculated to be 17.8, 19.4, and 45.7 nm for the films deposited at oxygen flow rate of 0, 0.4, and 3 sccm, respectively. These suggest that the increase of oxygen partial pressure can decrease the concentration of oxygen vacancies and improve crystal quality of the films, resulting in an obvious increase in mean crystallite size of the films. Figure 2a−d shows the typical core-level XPS spectra of O 1s, Fe 2p, Cu 2p, and Cu LMM for the (In0.89Fe0.06Cu0.05)2O3 film deposited at oxygen flow rate of 0 and 3 sccm, after etching with Ar+ ions to remove the surface oxidized layer. All

C

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C systems. The XANES spectrum is a fingerprint of local TM environment and dopant charge state. Figure 3a shows the Fe

7124 eV, respectively, which are located between those for FeO (7117.9 eV) and Fe2O3 (7125.8 eV). These results again reveal that the Fe element exists in the ionic form with the valences of +2 and +3 in In2O3 matrix in comparison with the threshold energies of various kinds of Fe oxides. Figure 4a shows the Cu K-edge XANES spectra of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0,

Figure 3. (a) Fe K-edge XANES spectra of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm as well as Fe, FeO, Fe2O3, and Fe3O4 foils. (b) Enlarged views of the pre-edge absorption features of Fe K-edge XANES spectra for the films and reference compounds.

Figure 4. (a) Cu K-edge XANES spectra of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm as well as Cu, Cu2O, and CuO foils. (b) Enlarged views of the pre-edge absorption features of Cu K-edge XANES spectra for the films and reference compounds.

K-edge XANES spectra of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm as well as some reference compounds (Fe metal foil and Fe oxides). Distinctly, the shape and position of the two characteristic peaks, A and B, are rather different from Fe metal foil, FeO, Fe2O3, and Fe3O4, implying that the local environments of Fe atoms in the (In0.89Fe0.06Cu0.05)2O3 films are different from the Fe metal and Fe-related oxides. The enlarged views of pre-edge absorption features of XANES spectra for the films and reference compounds are shown in Figure 3b. As is wellknown, the energy of the edge jump is used to quantify the oxidation state of Fe, which can be obtained by secondderivative analysis of the XANES spectral edge. It is clear that Fe oxidation state differs from the metallic Fe for the (In0.89Fe0.06Cu0.05)2O3 films. The position of absorption edge of the films lies between that of FeO and of Fe2O3, suggesting that there exists a mixed-valent state (Fe2+/Fe3+), which is consistent with the XPS results. The inset of Figure 3b shows the dependence of threshold energy (E0) on the Fe-valence state for the reference samples, Fe, FeO, Fe3O4, and Fe2O3. These reference samples provide a monotonically increasing relationship between E0 and Fe valence state. The threshold energies for the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm are 7122.8, 7123.7, and

0.4, and 3 sccm as well as standard Cu metal and Cu oxides. One can see that the experimental Cu K-edge XANES spectra for the films deposited at oxygen flow rate of 0.4 and 3 sccm are similar to each other but are rather different from those of Cu metal and Cu oxides, indicating a different local structure. As for the film deposited at oxygen flow rate of 0 sccm, the XANES spectrum has a weak oscillation feature and is also different from those of standard samples. However, this spectrum shows more similar features (labeled as A, B, and C) with the reference Cu metal; this needs to be further verified by EXAFS analysis. Figure 4b shows the enlarged views of the pre-edge absorption features of Cu K-edge XANES spectra for the films and reference compounds. It is obvious that the intensity of pre-edge peak remarkably increases and a strong shoulder peak is observed in comparison with the pre-edge features of Cu metal, implying the existence of metallic Cu clusters in the film deposited at oxygen flow rate of 0 sccm. The threshold energies for the films and reference compounds are also obtained according to the energy of the edge jump. It is found the E0 values for the reference Cu, Cu2O, and CuO are 8978.98, 8981.56, and 8983.62 eV, respectively, as shown in the inset of Figure 4b, and for the films deposited at oxygen D

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 5. (a) Fe and (b) Cu K-edge Fourier transforms of EXAFS spectra of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm as well as standard samples. The fitted curves are the dashed lines.

flow rate of 0, 0.4, and 3 sccm E0 values are 8979.3, 8981.24, and 8982.88 eV, respectively. The threshold energies of the films show a monotonically increasing with increasing oxygen flow rate, suggesting that introducing oxygen partial pressure can increase the oxidation states of doped Cu atoms from metallic Cu to Cu1+ or Cu2+. Figure 5a shows the results of Fe K-edge Fourier transforms of EXAFS spectra for the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm. As references, Fe or In K-edge functions of standard Fe, FeO, Fe2O3, and Fe3O4 foils as well as pure In2O3 are also plotted. The Fourier transforms of the Fe K-edge EXAFS spectra can provide the radial distribution function around Fe atoms. It is clear that there exist the similar characteristics in the Fe K-edge Fourier transforms of EXAFS spectra for the films deposited at oxygen flow rate of 0, 0.4, and 3 sccm; namely, three strong peaks at 1.52, 2.62, and 3.25 Å can be observed, which are due to the first (Fe−O), second (Fe−In), and third (Fe−In) nearest-neighboring coordination shells of Fe. Distinctly, the positions of these peaks are not consistent with metallic Fe (2.24 and 3.63 Å), Fe2O3 (1.40 and 2.58 Å), Fe3O4 (1.40, 2.57, and 3.08 Å), and FeO (1.59 and 2.62 Å) but are very close to those of In atoms in pure In2O3, implying that the existence of Fe metal and Fe-related oxide secondary phases in the films can be safety excluded and the doped Fe ions may well substitute for the In3+ sites of the In2O3 lattices. The Cu Kedge Fourier transforms of EXAFS spectra are shown in Figure 5b. A strong peak at about 2.24 Å and two weak peaks at about 1.48 and 3.22 Å can be observed for the film deposited at oxygen flow rate of 0 sccm. The strong peak obviously comes from the Cu−Cu coordination shell, proving the presence of a large amount of metallic Cu in the film. The positions of two weak peaks are very close to the first (In−O) and third (In− In) coordination shells of pure In2O3, implying that only a small number of Cu atoms are incorporated into In2O3 lattice. With increasing oxygen partial pressure, the strong peak at about 2.24 Å completely disappears. At the same time, three coordination shell peaks clearly appear, which are consistent with those of In atoms of pure In2O3. These suggest that all the Cu atoms are incorporated into the In3+ sites of In2O3 lattice after introducing oxygen partial pressure.

The In2O3 cell is composed of 80 atoms, including 8 In1 atoms and 24 In2 atoms, which correspond to trigonally distorted octahedral b sites and tetragonally distorted octahedral d sites, respectively. To determine the accurate location of Fe and Cu atoms, the EXAFS data were fitted by the Athena and Artemis interfaces to the IFEFFIT program package.30 The Artemis routine was exploited to fit EXAFS data in R space within the window of 0−4 Å. During the fitting process, the multiple-scattering paths were included, while the E0, bond length (R), and Debye−Waller (DW) factors (σ2, including thermal vibration and static disorder) varied. It was found that the Fe K-edge Fourier transform curves for all the films can be well fitted by the simple Fe substitutional model assuming the Fe atoms substitution for In1 sites of the In2O3 lattices, as shown by the dashed lines in Figure 5a. The Cu Kedge Fourier transform curves for the films deposited at oxygen flow rate of 0.4 and 3 sccm can be also well fitted by the simple Cu substitutional model. Of the attempts to fit the Fourier transform curve for the film deposited at an oxygen flow rate of 0 sccm, only the Cu substitutional model failed (not shown here). The best fit to Fourier transform curves can be obtained by assuming the Cu atoms substitution for In1 sites of the In2O3 lattice and coexistence with metallic Cu, namely, a twophase fitting. These results further provide strong evidence that the doped Fe atoms are incorporated into the In2O3 lattice for all the films. On the contrary, the doped Cu atoms form a large number of Cu metal clusters for the film deposited at oxygen flow rate of 0 sccm. Only when oxygen partial pressure is introduced, all the Cu atoms are incorporated into the In2O3 lattice. Larde et al. and Fukuma et al. reported the aggregation of a nanoscaled Co metal clusters in the Co doped ZnO thin films;31,32 they concluded that the formation of TM metal clusters has strong correlation with the presence of oxygen vacancy defects in the films. Similar results were also found in Co doped TiO2 films under vacuum or low partial oxygen pressure.33,34 Therefore, to avoid the formation of TM metal clusters, the increase of oxygen partial pressure is useful during deposition. Figure 6 presents the dependence of resistivity ρ and carrier concentration nc on oxygen flow rate. One can see that the semiconductivity of the films is strongly dependent on the oxygen partial pressure. For the film deposited at oxygen flow E

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

exp(Ed/KBT) plays an important role in the transport mechanism. In the low-temperature region, thermal activation energy is insufficient to promote conduction through delocalized carriers. Therefore, the variable range hopping (VRH) conduction mechanism is activated and takes place. The temperature dependence of the VRH conduction can be expressed by Mott’s equation: ρ ∼ exp[(T0/T)p]. The exponent p varies from p ∼1/4 when the density of states across the Fermi level is constant to p ∼ 1/2 in the case of significant electron−electron interactions that lead to the formation of a coulomb gap.39,40 It is clear that the ρ−T curves can be well expressed as the sum of three-dimensional Mott VRH conduction and thermal activation conduction, i.e., ρ = ρ0 exp[(T0/T)1/4] + ρ1 exp(Ed/KBT), where ρ0 and ρ1 are constants and T0 and Ed are the characteristic hopping temperature as well as the activation energy for thermal activation conductivity, respectively.40 The corresponding fitting curves of ln ρ versus T−1/4 are shown in Figure 7b. The Mott parameter, average hopping distance (RMott), of carriers was calculated by RMott = (2γ/Nd)1/3(Tv/T0)1/4, where γ is a constant, Nd is the donor concentration, and Tv is the onset temperature of Mott VRH.41 The corresponding fitted results and calculated parameters with the Mott VRH and thermal activation conduction models are given in Table 1. The average hopping distance RMott of carriers decreases, and the characteristic hopping temperature T0 increases with increasing oxygen partial pressure, suggesting that the change of oxygen partial pressure can effectively adjust the localization effect of carriers. To indentify the influence of oxygen partial pressure on the magnetic properties of the (In0.89Fe0.06Cu0.05)2O3 films, the magnetization versus magnetic field (M−H) curves for the films deposited at oxygen flow rate of 0, 0.4, and 3 sccm, measured at 300 K with a maximum field of ±2 T, are shown in Figure 8. The diamagnetic background from the substrates has been subtracted from the M−H curves. All the films show symmetric hysteresis loop at 300 K, suggesting the existence of RT ferromagnetism. The dependence of saturation magnetic moments (Ms) on the oxygen flow rate is plotted in the inset of Figure 8. Clearly, the Ms monotonously decreases from 2.58 to 0.93 emu/cm3 with increasing the oxygen flow rate. The decrease in Ms has strong correlation with the concentration of oxygen vacancies, indicating that the variation of oxygen partial pressure can effectively tune the magnetic properties of the (In0.89Fe0.06Cu0.05)2O3 films. However, this magnetic mechanism need be further clarified, as discussed later. To understand the magnetic mechanism and spin polarization of Fe/Cu codoped In2O3, we further employed firstprinciples calculations based density functional theory to investigate the electronic structures and magnetic interactions between Fe ions. The Fe/Cu codoped In2O3 system was modeled with a cubic bixbyite In2O3 supercell consisting of 32

Figure 6. Resistivity ρ and carrier concentration n of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm.

rate of 0 sccm, the type of carriers is confirmed to be n-type, as expected for the film growth in an oxygen-deficient environment. With increasing oxygen flow rate from 0.4 to 3 sccm, the carrier concentration nc decreases by more than 2 orders of magnitude, and the resistivity ρ also increases. It maybe due to that with increasing oxygen partial pressure the oxygen vacancies are largely compensated, resulting in remarkably decreasing the concentration of oxygen vacancies. The product of mean free path l and Fermi wave vector kF is estimated using the formula kFl = ℏ(3π2)2/3/(e2ρnc1/3), where e is the electron charge, ℏ is the Planck constant, ρ is the resistivity and nc is the carrier concentration.35 It is clear from Table 1 that the calculated values of kFl gradually decrease from 2.66 to 1.88 × 10−4 with increasing oxygen partial pressure, implying that the localization effect of carriers in the (In0.89Fe0.06Cu0.05)2O3 films is strongly increased.36 The temperature dependence of resistivity ρ(T) for the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0 and 0.4 sccm are shown in Figure 7a. The ρ−T curve for the film deposited at oxygen flow rate of 3 sccm is not measured for its high resistivity beyond the limit of the setup. It is obvious that the two films show different ρ(T) behavior. There exists a crossover from metallic to semiconducting transport behavior, namely, a minimum in the ρ−T curve near ∼125 K for the film deposited at oxygen flow rate of 0 sccm. The crossover behavior is also often observed in Sn doped In2O3 films and Cr doped In2O3 thin films.37,38 As for the ρ−T curves for the film deposited at oxygen flow rate of 0.4 sccm, the semiconducting transport behaviors can be observed in the whole temperature range. Generally, the charge transport of semiconductor is due to that charge carriers hop from occupied states to unoccupied states above the Fermi level EF. In the high-temperature region, thermal activation band conduction from donor levels near the conduction band, ρ ∼

Table 1. Parameters of (In0.89Fe0.06Cu0.05)2O3 Films Obtained from Hall Effect Measurement [Resistivity (ρ), Electron Concentration (n), and Mean Free Path (l) Calculated Using Drude Formula (l = (h/2π)kF/ne2ρ)] and Standard for FourProbe Measurement [Characteristic Hopping Temperature (T0), Localization Radius (ξ), Onset Temperature of Hopping (Tv), and Hard-Band Energy (Ed)] samples Ar:O2 = 10:0 Ar:O2 = 10:0.4 Ar:O2 = 10:3

ρ (Ω cm) 2.27 × 10 14.65 584.01

−3

n (cm−3)

kFl

Tv (K)

T0 (K)

ξ (nm)

Ed (meV)

2.79 × 1020 2.72 × 1018 4.65 × 1016

2.66 1.93 × 10−3 1.88 × 10−4

15 17

151 38416

2.596 0.36

0.02 0.035

F

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 7. (a) Resistivity−temperature (ρ−T) curves and (b) plots of ln ρ versus T−1/4 for the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0 and 0.4 sccm.

ΔEFM−AFM for the (1)−(3) configurations is −22.6, 72.5, and −56.2 meV, respectively, indicating that the ferromagnetic state is the ground magnetic ordering for the (1) and (3) configurations. As is well-known, Fe−O−Fe superexchange interaction will be antiferromagnetic coupling in nature despite the hybridization between the Fe 3d and O 2p states. It is obvious that after introducing oxygen vacancy into the Fe doped In2O3 system, the antiferromagnetic coupling between Fe atoms transforms into the ferromagnetic coupling, where the ΔEFM−AFM is a negative value (−22.6 meV), implying an energetic weak ferromagnetic favorable ground state. These strongly suggest that the oxygen vacancy defects can well mediate the magnetic interaction for the ferromagnetic coupling between Fe atoms. One can see that for the (2) configuration, namely Fe−O−Fe−O−Cu configuration, in the absence of oxygen vacancy, the antiferromagnetic coupling still predominates over the magnetic interaction between Fe atoms although there exists the additional codoping of Cu into the In2O3 system. These results are consistent with some theoretical predictions that Cu codoping cannot be responsible for the observed ferromagnetic ordering of Fe/Cu codoped In2O3 systems.19,42 However, for the (3) configuration involving both oxygen vacancy and Cu codoping, a more stable ferromagnetic ground state with the ΔEFM−AFM of a magnitude of −56.2 meV was observed in comparison with the Fe−Vo−Fe configuration. Figure 9a,b shows the density of states (DOSs) of (1) and (3) configurations in ferromagnetic ordering. As shown in Figure 9a, the DOS spectra of spin-polarized Fe and O atoms for the Fe−VO−Fe configuration show that the Fe 3d electrons are mainly distributed in the valence band, which leads to the unoccupied minority states in the conduction band. Clearly, the 3d states of Fe significantly overlap with the 2p bands of O around the Fermi level (EF), which can mediate the magnetic interaction between Fe dopants and contribute to an indirect ferromagnetic coupling. The hybridization introduces a total magnetic moment of 7.41 μB, which mainly comes from the Fe 3d orbital, as shown in the spin density distribution of Figure 9c. Figure 9b shows the DOS spectra for the Fe−Vo−Fe−O− Cu configuration. The Fe and Cu 3d partial DOSs are strongly hybridized with the O 2p state. The hybridization leads to the strong splitting of the energy levels and introduces a higher magnetic moment of 7.65 μB, as shown in Figure 9d. Although the introduction of oxygen vacancy results in more electrons compensating the holes induced by the Fe/Cu codoping, the

Figure 8. M−H curves at 300 K of the (In0.89Fe0.06Cu0.05)2O3 films deposited at oxygen flow rate of 0, 0.4, and 3 sccm.

In atoms and 48 O atoms. The supercell size is large enough to allow us to investigate different various configurations of Fe doping and Fe/Cu codoping configurations. In the In2O3 supercell, the two In atoms were substituted by two Fe atoms. The Fe−Fe position was selected to the “closed” (3.91 Å) configuration, in which two Fe atoms substitute for the nearest-neighbor In atoms and are separated by a single O atom. Different doping configurations have been considered: (1) Two Fe atoms substitute for the nearest-neighboring In atoms, and an oxygen vacancy is introduced by removing one oxygen atom adjacent to neighboring Fe atom in In2O3 lattice, denoted the Fe−VO−Fe configuration. (2) Two Fe atoms substitute for the nearest-neighboring In atoms in the In2O3 lattice, and a Cu atom is substituted for the In atom at the site near Fe−O−Fe pair, denoted the Fe−O−Fe−O−Cu configuration. (3) Two Fe atoms substitute for the nearestneighboring In atoms, and a Cu atom is substituted for the In atom at the site near Fe−Vo−Fe pair, denoted the Fe−Vo− Fe−O−Cu configuration. The ferromagnetic stability can be well obtained by the magnetization energy difference (ΔEFM−AFM) between the antiferromagnetic and ferromagnetic configurations. A negative value of ΔEFM−AFM means that the doped Fe atoms prefer ferromagnetic spin order. We have calculated the energies of ferromagnetic and antiferromagnetic couplings between two Fe atoms for the (1)−(3) configurations. It was found that the magnetization energy difference G

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Figure 9. DOS curves for (a) Fe−VO−Fe and (b) Fe−Vo−Fe−O−Cu configurations and (c, d) corresponding spin density distribution in the Fe/ Cu codoped In2O3 systems. The vertical dotted line denotes the position of the Fermi level. Red and blue isosurfaces correspond to spin-up and spin-down regions, and the large red and blue balls represent the Fe and Cu atoms, respectively.

ferromagnetic ordering is still under debate. Accounting for the strong localization effect of carriers in our (In0.89Fe0.06Cu0.05)2O3 films, the free carrier-mediated RKKY exchange interaction can be safety ruled out. The magnetic characteristics indicate that the maximum Ms is observed at oxygen flow rate of 0 sccm. As the oxygen partial pressure increases, the Ms remarkably decreases, indicating a strong correlation between the concentration of oxygen vacancies and Ms. Recently, some experimental and theoretical studies have proved that TM doped oxide based DMSs exhibit intrinsic structural defect-induced ferromagnetic coupling involving exchange magnetic interaction of TM dopants. For examples, Weng et al. predicted that Co doped TiO2 materials with oxygen vacancies result in the enhancement of ferromagnetic coupling and the change of the electronic structure of TiO2.43 Peleckis et al. concluded that the non-negligible concentration of oxygen vacancies in Ni doped In2O3 polycrystalline samples prepared under an argon atmosphere should be responsible for the observed RT ferromagnetism.23 Liu et al. by first-principles calculations suggest that oxygen vacancy defects can induce long-ranged RT ferromagnetism between nearest-neighbor Co ions.44 Ciatto et al. provided direct evidence of the existence of cobalt-vacancy complexes in Zn1−xCoxO epilayers and concluded that the Co−VO complexes make a significant contribution to the observed ferromagnetism.45 For our (In0.89Fe0.06Cu0.05)2O3 films, experimentally, the nonequilibrium sputtering process makes it possible that there exist a great deal of intrinsic oxygen vacancy defects in the In2O3 lattices. At the same time, the oxygen vacancies can also be

Cu still form an acceptor-type impurity state about 0.12 eV above EF. Therefore, it can be confirmed that the enhanced ferromagnetic coupling for Fe−Vo−Fe−O−Cu configuration has strong correction with the electrons induced by oxygen vacancy as well as the holes induced by Cu codoping, which mediate the indirect ferromagnetic coupling between Fe atoms. The variation of oxygen partial pressure during deposition remarkably changes the concentration of oxygen-related defects in the (In0.89Fe0.06Cu0.05)2O3 films. The detailed investigations of the films were presented by different experimental techniques and theoretical calculation. The XRD, XPS, and XAS measurements demonstrate that the Fe atoms substitute for the In sites of In2O3 lattices for all the films, while the majority of Cu atoms are separated to form the Cu metal clusters for the film films deposited at oxygen flow rate of 0 sccm. With increasing oxygen partial pressure, all the Cu atoms have substitutional occupancy at the In3+ sites of In2O3 lattices in the 1+ oxidation state. The observed longrange ferromagnetic ordering in the (In0.89Fe0.06Cu0.05)2O3 films can arise from some possibilities, such as Fe metal, Fe2O3, and Fe3O4, as well as intrinsic magnetic interaction between Fe or Cu atoms. Although the formation of nonferromagnetic Cu metal clusters in the film deposited at oxygen flow rate of 0 sccm, the absence of any Fe metal and Fe oxide magnetic phases can exclude the possibility of ferromagnetic order due to the extrinsic origin. Therefore, the ferromagnetic ordering is expected to arise from the intrinsic magnetic interaction between Fe or Cu atoms in the (In0.89Fe0.06Cu0.05)2O3 films. The exact mechanism of intrinsic H

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

films is intrinsic and originates from electrons bound in defect states associated with oxygen vacancies.

generated by substitutional Fe and Cu ions due to the chargecompensation effects. According to the results of Hall measurements, the electron concentration nc remarkably decreases from 2.79 × 1020 to 4.65 × 1016 cm−3 with increasing oxygen partial pressure. The large decrease by more than 4 orders of magnitude for the electron concentration implies an obvious reduction in the concentration of oxygen vacancies, although more Cu atoms are dissolved in the In2O3 lattice at high oxygen flow rates of 0.4 and 3 sccm, resulting in an increase in the concentration of oxygen vacancies. Therefore, the formation of bound magnetic polarons (BMP) involving oxygen vacancies, which form F-centers with trapped electrons, can account for the origin of ferromagnetic ordering. Based on first-principles calculation and experimental results, the ferromagnetic coupling can be only obtained by the coexistence of oxygen vacancies and Cu codoping in the Fe/Cu codoped In2O3 systems. Although more Cu atoms are incorporated into the In2O3 lattice with increasing oxygen partial pressure, the codoped Cu atoms cannot well activate the ferromagnetism interaction due to the lack of oxygen vacancies. On the other hand, with increasing oxygen partial pressure, the concentration of oxygen vacancies and the localization radius RMott of carriers remarkably decrease. These effects can to some extent change the disturbance of BMPs from the hopping of carrier and shrink the radius of magnetic polaron rBMP. Therefore, more doped Fe ions are outside the ferromagnetic domains (effective rBMP), and the oxygen vacancy mediated ferromagnetism coupling is not achieved effectively due to the small effective rBMP. Therefore, the long-range ferromagnetic ordering could not be activated and results in the monotonous decrease in Ms with increasing oxygen partial pressure. These results point toward a strong relationship among the oxygen vacancies, Cu codoping, and magnetic interactions. However, further experimental and theoretical investigations are needed to address this issue.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (Y.A.). ORCID

Yukai An: 0000-0003-1678-7330 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported by Tianjin Natural Science Foundation of China (Grant 17JCYBJC17300), Natural Science Foundation of Henan Province of China (Grant 182300410271), Program for Science & Research Innovation Foundation of Zhoukou Normal University (Grant ZKNUA201803), and the Beijing Synchrotron Radiation Laboratory.



REFERENCES

(1) Saadaoui, H.; Luo, X.; Salman, Z.; Cui, X. Y.; Bao, N. N.; Bao, P.; Zheng, R. K.; Tseng, L. T.; Du, Y. H.; Prokscha, T.; et al. Intrinsic Ferromagnetism in the Diluted Magnetic Semiconductor Co:TiO2. Phys. Rev. Lett. 2016, 117, 227202. (2) Pellicer, E.; Menendez, E.; Fornell, J.; Nogues, J.; Vantomme, A.; Temst, K.; Sort, J. Mesoporous Oxide-Diluted Magnetic Semiconductors Prepared by Co Implantation in Nanocast 3D-Ordered In2O3-y Materials. J. Phys. Chem. C 2013, 117, 17084−17091. (3) Babu, S. H.; Kaleemulla, S.; Rao, N. M.; Krishnamoorthi, C. Indium Oxide: A Transparent, Cconducting Ferromagnetic Semiconductor for Spintronic Applications. J. Magn. Magn. Mater. 2016, 416, 66−74. (4) Dietl, T.; Ohno, H.; Matsukura, F.; Cibert, J.; Ferrand, D. Zener Model Description of Ferromagnetism in Zinc-Blende Magnetic Semiconductors. Science 2000, 287, 1019−1022. (5) Cao, Q.; Fu, M. X.; Zhu, D. P.; Cai, L.; Zhang, K.; Liu, G. L.; Chen, Y. X.; Kang, S. S.; Yan, S. S.; Mei, L. M.; et al. Enhancing s, p-d Exchange Interactions at Room Temperature by Carrier Doping in Single Crystalline Co0.4Zn0.6O Epitaxial Films. Appl. Phys. Lett. 2017, 110, 092402. (6) Phan, T.-L.; Yu, S. C. Optical and Magnetic Properties of Zn1−xMnxO Nanorods Grown by Chemical Vapor Deposition. J. Phys. Chem. C 2013, 117, 6443−6453. (7) Zhang, L.; Zhu, L. P.; Hu, L.; Li, Y. G.; Song, H.; Ye, Z. Z. Interfacial Effect on Mn-Doped TiO2 Nanoparticles: from Paramagnetism to Ferromagnetism. RSC Adv. 2016, 6, 57403−57408. (8) Prajapati, B.; Kumar, S.; Kumar, M.; Chatterjee, S.; Ghosh, A. K. Investigation of the Physical Properties of Fe:TiO2-Diluted Magnetic Semiconductor Nanoparticles. J. Mater. Chem. C 2017, 5, 4257−4267. (9) Mazloom, J.; Ghodsi, F. E.; Golmojdeh, H. Synthesis and Characterization of Vanadium Doped SnO2 Diluted Magnetic Semiconductor Nanoparticles with Enhanced Photocatalytic Activities. J. Alloys Compd. 2015, 639, 393−399. (10) Tandon, B.; Yadav, A.; Nag, A. Delocalized Electrons Mediated Magnetic Coupling in Mn-Sn Codoped In2O3 Nanocrystals: Plasmonics Shows the Way. Chem. Mater. 2016, 28, 3620−3624. (11) Farvid, S. S.; Sabergharesou, T.; Hutfluss, L. N.; Hegde, M.; Prouzet, E.; Radovanovic, P. V. Evidence of Charge-Transfer Ferromagnetism in Transparent Diluted Magnetic Oxide Nanocrystals: Switching the Mechanism of Magnetic Interactions. J. Am. Chem. Soc. 2014, 136, 7669−7679. (12) Ramesh, S.; Venugopal, P.; Mosquera, E. Experimental and Theoretical Investigation of Bixbyite (Mn0.8Ni0.2)2O3 Nanoparticles for Magnetic and Electrochemical Applications. J. Magn. Magn. Mater. 2017, 443, 45−50.

4. CONCLUSION In summary, the (In0.89Fe0.06Cu0.05)2O3 films were deposited by RF-magnetron sputtering technique with different argon− oxygen flow rates (10:0, 10:0.4, and 10:3). The influences of oxygen partial pressure on the local Fe and Cu structures as well as magnetic and transport properties of the films were systematically investigated by experimental measurements and the first-principles calculations. The XRD, XPS, and XAS analyses reveal that Fe dopants substitute In3+ sites of In2O3 lattice with a mixture valence of +2 and +3 for all the films, while the majority of Cu dopants exist as Cu metal clusters in the (In0.89Fe0.06Cu0.05)2O3 films without oxygen partial pressure. After introducing oxygen partial pressure, all the Cu atoms are incorporated substitutionally into the In3+ sites of In2O3 lattices with the 1+ oxidation state. The typical RT ferromagnetism is observed for all the (In0.89Fe0.06Cu0.05)2O3 films. The Ms and carrier concentration nc decrease monotonically with increasing oxygen partial pressure. The Mott VRH and hard band gap hopping transport behavior dominates the conduction mechanism of the films, confirming that the carriers are strongly localized. The first-principles calculations suggest that the oxygen vacancy defects can well mediate the magnetic interaction for the ferromagnetic coupling between Fe atoms. The coexistence of oxygen vacancies and Cu codoping can lead to a more stable ferromagnetism ground state in the Fe/Cu codoped In2O3 systems. Therefore, the observed RT ferromagnetism in the (In0.89Fe0.06Cu0.05)2O3 I

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX

Article

The Journal of Physical Chemistry C

Zn0.9Co0.1O Thin Films Using Atom Probe Tomography. J. Am. Chem. Soc. 2011, 133, 1451−1458. (32) Fukuma, Y.; Asada, H.; Yamamoto, J.; Odawara, F.; Koyanagi, T. Large Magnetic Circular Dichroism of Co Clusters in Co-Doped ZnO. Appl. Phys. Lett. 2008, 93, 142510. (33) Kim, D. H.; Yang, J. S.; Lee, K. W.; Bu, S. D.; Noh, T. W.; Oh, S. J.; Kim, Y. W.; Chung, J. S.; Tanaka, H.; Lee, H. Y.; Kawai, T. Formation of Co Nanoclusters in Epitaxial Ti0.96Co0.04O2 Thin Films and their Ferromagnetism. Appl. Phys. Lett. 2002, 81, 2421−2423. (34) Shinde, S. R.; et al. Co-occurrence of Superparamagnetism and Anomalous Hall Effect in Highly Reduced Cobalt-Doped Rutile TiO2‑δ Films. Phys. Rev. Lett. 2004, 92, 166601. (35) Xu, Q. Y.; Hartmann, L.; Schmidt, H.; Hochmuth, H.; Lorenz, M.; Spemann, D.; Grundmann, M. s-d Exchange Interaction Induced Magnetoresistance in Magnetic ZnO. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 134417. (36) Song, C.; Geng, K. W.; Zeng, F.; Wang, X. B.; Shen, Y. X.; Pan, F.; Xie, Y. N.; Liu, T.; Zhou, H. T.; Fan, Z. Giant Magnetic Moment in An Anomalous Ferromagnetic Insulator: Co-Doped ZnO. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 73, 024405. (37) Ederth, J.; Johnsson, P.; Niklasson, G. A.; Hoel, A.; Hultåker, A.; Heszler, P.; Granqvist, C. G.; van Doorn, A. R.; Jongerius, M. J.; Burgard, D. Electrical and Optical Properties of Thin Films consisting of Tin-Doped Indium Oxide Nanoparticles. Phys. Rev. B: Condens. Matter Mater. Phys. 2003, 68, 155410. (38) Kharel, P.; Sudakar, C.; Sahana, M. B.; Lawes, G.; Suryanarayanan, R.; Naik, R.; Naik, V. M. Room temperature ferromagnetism in Cr-doped In2O3 on high vacuum annealing of thin films and bulk samples. J. Appl. Phys. 2007, 101, 09H117. (39) Efros, A. L.; van Nguyen, L.; Shklovskii, B. I. Impurity Band Structure in Lightly Doped Semiconductors. J. Phys. C: Solid State Phys. 1979, 12, 1869. (40) Mott, N. F.; Davies, E. A. Electronic Processes in Noncrystalline Materials, 2nd ed.; Oxford University: New York, 1979. (41) Lisunov, K. G.; Arushanov, E.; Thomas, G. A.; Bucher, E.; Schon, J. H. Variable-Range Hopping Conductivity and Magnetoresistance in n-CuGaSe2. J. Appl. Phys. 2000, 88, 4128. (42) Yu, Z. G.; He, J.; Xu, S.; Xue, Q.; van’t Erve, O. M. J.; Jonker, B. T.; Marcus, M. A.; Yoo, Y. K.; Cheng, S.; Xiang, X. D. Origin of Ferromagnetism in Semiconducting (In1‑x‑yFexCuy)2O3‑σ. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 165321. (43) Weng, H.; Yang, X.; Dong, J.; Mizuseki, H.; Kawasaki, M.; Kawazoe, Y. Electronic Structure and Optical Properties of the CoDoped Anatase TiO2 Studied from First Principles. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 69, 125219. (44) Liu, E. Z.; Jiang, J. Z. O-Vacancy-Mediated Apin-Apin Interaction in Co-Doped ZnO: First-Principles Total-Energy Calculations. J. Appl. Phys. 2010, 107, 023909. (45) Ciatto, G.; Di Trolio, A.; Fonda, E.; Alippi, P.; Testa, A. M.; Bonapasta, A. A. Evidence of Cobalt-Vacancy Complexes in Zn1‑xCoxO Dilute Magnetic Semiconductors. Phys. Rev. Lett. 2011, 107, 127206.

(13) Garnet, N. S.; Ghodsi, V.; Hutfluss, L. N.; Yin, P. H.; Hegde, M.; Radovanovic, P. V. Probing the Role of Dopant Oxidation State in the Magnetism of Diluted Magnetic Oxides Using Fe-Doped In2O3 and SnO2 Nanocrystals. J. Phys. Chem. C 2017, 121, 1918−1927. (14) Green, R. J.; Regier, T. Z.; Leedahl, B.; McLeod, J. A.; Xu, X. H.; Chang, G. S.; Kurmaev, E. Z.; Moewes, A. Adjacent Fe-Viacancy Interactions as the Origin of Room Temperature Ferromagnetism in (In1‑xFex)2O3. Phys. Rev. Lett. 2015, 115, 167401. (15) Kuroda, S.; Nishizawa, N.; Takita, K.; Mitome, M.; Bando, Y.; Osuch, K.; Dietl, T. Origin and Control of High-Temperature Ferromagnetism in Semiconductors. Nat. Mater. 2007, 6, 440. (16) Lathiotakis, N. N.; Andriotis, A. N.; Menon, M. Codoping: A Possible Pathway for Inducing Ferromagnetism in ZnO. Phys. Rev. B: Condens. Matter Mater. Phys. 2008, 78, 193311. (17) Lisenkov, S.; Andriotis, A. N.; Sheetz, R. M.; Menon, M. Effects of Codoping on the Ferromagnetic Enhancement in ZnO. Phys. Rev. B: Condens. Matter Mater. Phys. 2011, 83, 235203. (18) Yoo, Y. K.; Xue, Q. H.; Lee, C.; Cheng, S.; Xiang, X. D.; Dionne, G. F.; Xu, S.; He, J.; Chu, Y. S.; Preite, S. D.; et al. Bulk Synthesis and High-Temperature Ferromagnetism of (In1‑xFex)2O3‑σ with Cu Co-Doping. Appl. Phys. Lett. 2005, 86, 042506. (19) He, J.; Xu, S.; Yoo, Y. K.; Xue, Q.; Lee, H. C.; Cheng, S.; Xiang, X. D.; Dionne, G. F.; Takeuchi, I. Room Temperature Ferromagnetic n-Type Semiconductor in (In1‑xFex)2O3‑σ. Appl. Phys. Lett. 2005, 86, 052503. (20) Ho, H. W.; Zhao, B. C.; Xia, B.; Huang, S. L.; Tao, J. G.; Huan, A. C. H.; Wang, L. Magnetic and Magnetotransport Properties in Cu and Fe co-Doped Bulk In2O3 and ITO. J. Phys.: Condens. Matter 2008, 20, 475204. (21) Chu, D.; Zeng, Y. P.; Jiang, D. Abnormal Phase Transition and Magnetic Properties in Cu, Fe co-Doped In2O3 Nanocrystals. Appl. Phys. Lett. 2008, 92, 182507. (22) Jiang, F. X.; Feng, Q.; Quan, Z. Y.; Ma, R. R.; Heald, S. M.; Gehring, G. A.; Xu, X. H. The Role of Cu Codoping on the Fe metal Clustering and Ferromagnetism in Fe-Doped In2O3 Films. Mater. Res. Bull. 2013, 48, 3178−3182. (23) Peleckis, G.; Wang, X. L.; Dou, S. X. High Temperature Ferromagnetism in Ni-Doped In2O3 and Indium-Tin Oxide. Appl. Phys. Lett. 2006, 89, 022501. (24) Samariya, A.; Singhal, R. K.; Kumar, S.; Xing, Y. T.; Sharma, S. C.; Kumari, P.; Jain, D. C.; Dolia, S. N.; Deshpande, U. P.; Shripathi, T.; et al. Effect of Hydrogenation vs. Re-Heating on Intrinsic Magnetization of Co doped In2O3. Appl. Surf. Sci. 2010, 257, 585− 590. (25) Gonzalez, G. B.; Mason, T. O.; Quintana, J. P.; Warschkow, O.; Ellis, D. E.; Hwang, J.-H.; Hodges, J. P.; Jorgensen, J. D. Defect Structure Studies of Bulk and Nano-Indium-Tin Oxide. J. Appl. Phys. 2004, 96, 3912. (26) Xu, X. H.; Jiang, F. X.; Zhang, J. X.; Fan, C.; Wu, H. S.; Gehring, G. A. Magnetic and Transport Properties of n-Type FeDoped In2O3 Ferromagnetic Thin Films. Appl. Phys. Lett. 2009, 94, 212510. (27) Yan, S. M.; Ge, S. H.; Qiao, W.; Zuo, Y. L. Synthesis of Ferromagnetic Semiconductor 0.67FeTiO3-0.33Fe2O3 Powder by Chemical Co-Precipitation. J. Magn. Magn. Mater. 2010, 322, 824− 826. (28) Singhal, S.; Kaur, J.; Namgyal, T.; Sharma, R. Cu-Doped ZnO Nanoparticles: Synthesis, Structural and Electrical Properties. Phys. B 2012, 407, 1223−1226. (29) Raimondi, F.; Geissler, K.; Wambach, J.; Wokaun, A. Hydrogen Production by Methanol Reforming: Post-Reaction Characterisation of a Cu/ZnO/Al2O3 Catalyst by XPS and TPD. Appl. Surf. Sci. 2002, 189, 59−71. (30) Ravel, B.; Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: Data Analysis for X-ray Absorption Spectroscopy sing IFEFFIT. J. Synchrotron Radiat. 2005, 12, 537−541. (31) Larde, R.; Talbot, E.; Pareige, P.; Bieber, H.; Schmerber, G.; Colis, S.; Pierron-Bohnes, V.; Dinia, A. Evidence of Superparamagnetic Co Clusters in Pulsed Laser Deposition-Grown J

DOI: 10.1021/acs.jpcc.8b05058 J. Phys. Chem. C XXXX, XXX, XXX−XXX