betaFeedback

WorksPhysics and Astronomy

Emergence of Ferromagnetism Through the Metal-Insulator Transition in Undoped Indium Tin Oxide Films

Abstract

We present a detailed study of the emergence of bulk ferromagnetism in low carrier density samples of undoped indium tin oxide (ITO). We used annealing to increase the density of oxygen vacancies and change sample morphology without introducing impurities through the metal insulator transition (MIT). We utilized a novel and highly sensitive "Corbino-disk torque magnetometry" technique to simultaneously measure the thermodynamic and transport effects of magnetism on the same sample after successive annealing. With increased sample granularity, carrier density increased, the sample became more metallic, and ferromagnetism appeared as resistance approached the MIT. Ferromagnetism was observed through the detection of magnetization hysteresis, anomalous Hall effect (AHE), and hysteretic magnetoresistance. A sign change of the AHE as the MIT is approached may elucidate the interplay between the impurity band and the conduction band in the weakly insulating side of the MIT.

Introduction

The initial discovery of room-temperature ferromagnetism in Co-doped TiO2_2 [1] films was soon followed by the observation of similar behavior in other transition-metal (TM) doped oxides including Co-doped SnO2_2 [2], Mn-doped ZnO [3], Co-doped CeO2_2 [4], and Ni, Mo, Fe, and Mn-doped In2_2O3_3 [5, 6, 7, 8]. The origin of the observed thin film magnetism continues to be debated, particularly after the discovery of ferromagnetism in undoped oxides such as HfO2_2 [9], TiO2_2, and In2_2O3_3 [10]. At the heart of the puzzle is observation that the respective bulk undoped oxides exhibit only diamagnetism. Similarly, doping level in TM doped oxides was often too low to explain the observed strength, anisotropy, or thermal treatment sensitivity of the magnetic state. Correlation between the samples’ morphology and the occurrence of magnetism instead suggests an explanation based on a spontaneous formation of defects or oxygen vacancies [10]. The abundant surface vacancies of the nanoparticles which make up a granular structure provide unpaired electron spins which could interact via an exchange mechanism [11].

With the observation of ferromagnetism in In2_2O3_3 and SnO2_2, it is natural to explore whether the widely used solid-solution of tin-doped indium oxide (ITO) exhibits ferromagnetism. Ferromagnetism could be used to integrate this optoelectronic material into novel spintronics applications. Indeed, various TM doped [6, 12, 13, 14, 15] and undoped [5, 16, 17] ITO systems were shown to exhibit ferromagnetism that persisted to room temperature. Oxygen vacancy defects have been proposed as the source of magnetic moments rather than impurities [17, 18, 19], while itinerant carriers mediate the collective ferromagnetic state [15, 16, 20]. A persistent obstacle in proclaiming intrinsic ferromagnetism for ITO has been the difficulty in observing the anomalous Hall effect (AHE) and hysteretic magnetoresistance. For example, in Cr-doped ITO

AHE appears only at very high carrier concentrations exceeding 1021 cm−310^{21}\ \mathrm{cm}^{-3} [15], while in 12% Co-doped ITO (10% Sn) AHE appears above ∼109 cm−3\sim10^9\ \mathrm{cm}^{-3}. Samples of ITO with lower concentrations of magnetic elements generally do not show AHE, and the observed negative magnetoresistance is attributed to magnetic-field induced reduction in spin scattering [15, 19, 20]. Similar to other magnetic semiconductors (e.g. [21]), ITO is a heavily doped n-type semiconductor typically studied near the metal-insulator transition (MIT). In particular, undoped ITO close to the MIT may exhibit local magnetic moments that are typically self-generated in quenched disordered electronic systems due to interaction effects [22]. Thus carrier density may not be the only relevant parameter for magnetism [20, 23]. It is correspondingly important to elucidate the occurrence of ferromagnetism in undoped ITO within these contexts.

In this work we examine the interplay between morphology, proximity to the MIT, and the emergence of bulk magnetic properties in low carrier density samples of ITO. We used annealing to increase the density of oxygen vacancies and change sample granularity without introducing impurities, and annealed through the MIT [24, 25, 26, 27, 28, 29, 30, 31]. We utilized a novel and highly sensitive “Corbino-disk torque magnetometry” technique to simultaneously measure the thermodynamic and transport effects of magnetism on the same sample after successive annealing. Our results clearly show that: i) Starting with a highly insulating mostly amorphous ITO with no observed magnetism, successive annealing induces morphology change towards more granular structure. This change is accompanied by increase in carrier density and occurrence of ferromagnetism near the MIT. ii) The amplitude of magnetization hysteresis and saturating field both increase with decreased in sheet-resistance, indicating strengthening of the ferromagnetic state. iii) Undoped ITO thus is ferromagnetic, exhibiting anomalous Hall effect and hysteretic magnetoresistance for carrier density as low as ≈3×1018\approx3 \times10^{18} carriers/cm3^3. iv) Both AHE and hysteretic magnetoresistance are observed across the MIT. The sign of AHE changes from negative on the insulating side of the MIT to positive AHE on the metallic side of the MIT.

Experiment

Preparation of Samples

Hall bar and Corbino disk patterned ITO samples were prepared by RF magnetron sputtering using a 10% Sn target at 8 W/Inch2^2, 5 mTorr of Ar pressure, and 2.5% partial pressure of O2_2. Both Hall bar and Corbino disk samples were subjected to successive annealing schedules at 400 K and pressure of <10−5< 10^{-5} Torr with measurements performed after each anneal. A typical ITO sample was initially highly insulating and exhibited a starting sheet resistance of 70 kΩ\Omega/□ at room temperature and a thickness of 40 nm. Resistance increased to 460 kΩ\Omega/□ at the measurement base temperature of 4.2 K. Measurements were performed in a liquid helium cryostat with a 7 T superconducting magnet at temperature close to base temperature. Hall effect and magnetoresistance were measured on the Hall bars using standard 5 lead configuration. The Corbino disk samples were measured by torque magnetometry as described below.

Corbino Disk Torque Magnetometry

Cantilever torque magnetometry utilizes a high-QQ resonator to detect the interaction between a magnetic dipole μ⃗\vec{\mu} and an external magnetic field B⃗\vec{B} [32, 33, 34, 35, 36], where the resulting torque is [34]

τ⃗=μ⃗×B⃗.(1)\vec{\tau}=\vec{\mu}\times\vec{B}. \tag*{(1)}

A schematic depiction of cantilever torque magnetometry is shown in Figure 1(a). The angular response θ\theta of a cantilever with moment of inertia AA, resonant frequency ω0\omega_0, and quality factor QQ subject to an external torque τ\tau may be approximated as a damped harmonic oscillator following [37]

Schematic of torque magnetometry and picture of the ITO Corbino disk cantilever

Figure 1. (a) Schematic of torque magnetometry where Hall currents in a Corbino disk induce the magnetic moment μ\mu. (b) Picture of the ITO Corbino disk cantilever. The center contact is connected to an underlying ground plane. The metal pad on the end of the cantilever is used for optical alignment.

Aθ¨+AQω0θ˙+Aω02θ=τ.(2)A\ddot{\theta}+\frac{A}{Q}\omega_0\dot{\theta}+A\omega_0^2\theta=\tau. \tag*{(2)}

The torque of a dipole parallel to the cantilever surface norm and a static zz-aligned magnetic field BB shifts resonant frequency as

Aω02→Aω02−μB,orΔω0ω0=μB2Aω02.(3)A\omega_0^2\rightarrow A\omega_0^2-\mu B,\quad\text{or}\quad\frac{\Delta\omega_0}{\omega_0}=\frac{\mu B}{2A\omega_0^2}. \tag*{(3)}

The circulating current in a Corbino disk [38] patterned on a cantilever forms such a magnetic dipole parallel to the cantilever surface norm [32]. Finally, cantilever resonant frequency is observed interferometrically and driven through laser radiation pressure [32].

Cantilevers with ITO Corbino disks were examined for magnetization and magnetic transport effects. Silicon cantilevers of dimension 200 × 600 × 3 μ\mum were fabricated with

Corbino disks as shown in Figure 1(b). The resulting devices exhibited A=3×10−18A=3\times10^{-18} kg-m2^2 and f0=ω0/2π=14.61f_0=\omega_0/2\pi=14.61 kHz. Magnetization is measured through the voltage-independent

Δf0(B)=f0(B)−f0(B=0)(4)\Delta f_0(B)=f_0(B)-f_0(B=0) \tag*{(4)}

and sample transport properties for Corbino disk voltage VV are measured through

δf0(±V,B)=f0(+V,B)−f0(−V,B).(5)\delta f_0(\pm V,B)=f_0(+V,B)-f_0(-V,B). \tag*{(5)}

The Hall effect is measured through the even in BB component of δf0(±V,B)\delta f_0(\pm V,B) while the odd component of δf0(±V,B)\delta f_0(\pm V,B) is proportional to the misalignment between the inner and outer Corbino disk contacts and longitudinal current. As shown by Mumford et al. [32], using Eqn. 3 the even component of δf0(±V,B)\delta f_0(\pm V,B) yields Hall conductivity

σxy=Cδf0BV,whereC=4Af0ln⁡(ro/ri)π(ro2−ri2)(6)\sigma_{xy}=C\frac{\delta f_0}{BV},\quad\text{where}\quad C=\frac{4Af_0\ln(r_o/r_i)}{\pi(r_o^2-r_i^2)} \tag*{(6)}

and ror_o and rir_i are the outer and inner radii of the disk. Thus, δf0(±V,B)\delta f_0(\pm V,B) is proportional to σxyB\sigma_{xy}B. The expected Hall contribution to δf0\delta f_0 is proportional to B2B^2 as σxy∝B\sigma_{xy}\propto B. A typical

B2B^2-like shift in δf0(±V,B)\delta f_0(\pm V,B) and background linear in BB shift in δf0(±V,B)\delta f_0(\pm V,B) due to misalignment are shown in Figure 2.

Even component of the frequency shift as a function of magnetic field

Figure 2. Example of extracting the even component of the frequency shift from which the Hall conductivity is calculated for non-magnetic metallic ITO. Inset shows the total frequency shift as a function of magnetic field.

Results and Discussion

Characterization of Samples

ITO grain size and granularity increased with annealing as shown in Figure 3. As deposited, the ITO was largely amorphous with a smooth surface profiled by atomic force microscopy (AFM). After annealing, the ITO subdivided into grains with height fluctuations nearly equal to the deposited ITO thickness as seen in Figure 3(a-Figure 3b). Complimentary scanning electron microscope (SEM) images also demonstrate a change in granularity with annealing. Two samples of ITO were analyzed by SEM. Sample 1 was patterned in a Hall bar and sample 2 was patterned in a Corbino disk. As deposited, the ITO exhibited either very small ITO grains or larger smooth ITO domains. The differences between the as-deposited ITO in the Corbino disk and Hall bar may be explained by a difference in deposited ITO morphology based on substrate [39]. After annealing, the ITO became granular with typical grain size on the order of 500 nm2^2. A clear increase in grain size with annealing is seen in Figure 4 which compares the grain distribution of unannealed and annealed ITO SEM images. The sheet resistance at base temperature also decreased with annealing, while the carrier density increased as is shown in Table I. Such a change in grain size [24, 25, 26, 27, 28, 29, 30], decreased resistivity, and increased carrier density [24, 26, 27, 28, 31] are consistent with previous observations on ITO with higher temperature thermal annealing.

AFM and SEM images characterizing the granularity of ITO with annealing

Figure 3. Characterization of the granularity of ITO with annealing. (a-b) AFM profile of ITO Hall bar before and after annealing. Sample height fluctuations and granularity increase after annealing. (c-d) SEM images of ITO Hall bar before and after annealing. (e-f) SEM images of Corbino disk ITO before and after annealing. The initially large patches of ITO or smooth ITO breaks down into granular areas after annealing.

Comparison of grain sizes between annealed and non-annealed ITO samples

Figure 4. Comparison of the grain sizes between the annealed and non-annealed ITO samples by SEM. Grains were identified by adaptive thresholding of the SEM images shown in Figure 3. Fluctuations in SEM brightness changed from local noise to distinct larger grains with annealing.

Resistivity (kΩ/□\Omega/\square)Configurationnn (101810^{18} cm−3^{-3})
3300Corbino Disk-0.08 ±\pm 0.2
460Corbino Disk0.17 ±\pm 0.04
185Corbino Disk1.7 ±\pm 1
48Corbino Disk3.4 ±\pm 0.5
32Corbino Disk-2.7 ±\pm 0.3
17Corbino Disk6.5 ±\pm 1.4
15Hall Bar9.33 ±\pm 0.02
8.5Corbino Disk22 ±\pm 10
4.6Hall Bar23.7 ±\pm 0.1

Table 1. Samples investigated in this study and their transport properties at base temperature.

Corbino Disk Torque Magnetometry of ITO

Hall Conductivity Annealing Across MIT

An initially highly resistive ITO cantilever (resistivity of 3300 kΩ/□\Omega/\square at base temperature) was annealed in 6 steps to investigate the variation of Hall conductivity through the MIT as well as the onset and evolution of magnetism. Figure 5(a) shows the extracted Hall conductivity of an ITO sample with

Figure 5. Behavior of σxyρxx\sigma_{xy}\rho_{xx} at 5 T measured on a series of annealed ITO samples through the MIT. The outlying sample very close to the MIT may be strongly influenced by the magnetic behavior as discussed in Secs. III B 2-III B 3. Additional data obtained through a Hall bar approach and discussed in Sec. III C is included as a reference.

annealing, while the respective carrier densities are given in Table I. Annealed Hall-bar configuration samples are included for comparison. Ferromagnetism emerged close to the MIT and thus complicated extracting the ordinary Hall effect from δf0(±V,B)\delta f_0(\pm V,B) as discussed in further detail below. This increased complexity increases the uncertainty in σxy\sigma_{xy} extracted near the MIT, but the observed carrier densities remain consistent with similarly prepared samples measured in a Hall bar configuration.

An important result to be drawn from the high resistance samples deep in the hopping regime is the tendency of σxy\sigma_{xy} to vanish with increasing disorder or increasing ρxx\rho_{xx}. The hopping regime Hall effect is governed by the self-interference effect of the electron wave function propagating along limited paths dictated by magnetic field and the impurity sites responsible for the impurity band. Early theoretical work by Friedman and Pollak [40] using the Holstein approach [41] concluded that the Hall resistivity diverges as the temperature tends to zero. This result was then reinforced by Entin-Wohlman, et al. [42], who attempted to explain the discrepancy with complementary calculations predicting an electronic state where the longitudinal resistance diverges at zero temperature, but the Hall resistivity remains constant [43, 44, 45].

Such a state is known as a Hall insulator. Indeed, while experimental results focused on the vicinity of the MIT confirmed divergence of ρxy\rho_{xy} on the insulating side [46, 47], in other experiments insulating states derived from quantum Hall states were consistent with a Hall insulator. The result for ρxy\rho_{xy} may be different depending on the order of temperature and frequency limits taken in the DC limit of the Hall resistivity. Similarly, it was emphasized in [42] that if both ρxy\rho_{xy} and ρxx\rho_{xx} are calculated and averaged over the disorder, they should both show divergence, while if the respective conductivities are averaged over the disorder the calculated ρxy\rho_{xy} may approach a constant. We directly measure σxy\sigma_{xy} and ρxx\rho_{xx}, which is a hybrid of the above two procedures.

We show σxyρxx\sigma_{xy}\rho_{xx} in Figure 2(b) to test the Hall behavior. The observed σxyρxx\sigma_{xy}\rho_{xx} is approximately a constant, excluding the very vicinity of the MIT. Far above the MIT weak localization corrections may alter the ratio [47] at a level smaller than the resolution of the experiment. However, on the insulating side, a Hall insulating behavior would require that σxyρxx∼ρxx−1\sigma_{xy}\rho_{xx} \sim\rho_{xx}^{-1} in contrast to the observed behavior. Such behavior supports that in the hopping regime ρxy∝ρxx\rho_{xy} \propto\rho_{xx}, or that both will show similar divergence as T→0T \to0.

Magnetism Near MIT

Successive annealing of highly insulating ITO resulted in increased granularity, lower resistivity, and, as the MIT is approached, the emergence of ferromagnetism. We will now profile the origin of the surprising magnetism, as well as its anisotropy and dependence on the film morphology, carrier density, and defect density.

Anisotropy in bulk magnetization may be observed through cantilever torque magnetometry [33, 34, 35, 36]. Hysteresis in magnetization due to ferromagnetic ordering results in hysteresis in Δf0(B)\Delta f_0(B) observed through Eqs. (3) and (4). Hysteresis in magnetization arose from ferromagnetism after annealing near the MIT transition point as shown in Figure 6. All tested cantilevers with patterned metallic layers displayed low-field shifts in f0f_0 as discussed in Appendix A, thus only the hysteresis in magnetization for ∣B∣>1.5|B| > 1.5 T can be attributed to the ITO. The

Figure 6. Hysteresis in magnetization Δm\Delta m attributed to magnetization for three annealed ITO resistivities. We calculate the hysteretic magnetic moment per “formula unit” (f.u.), approximated as the unit cell volume of crystalline In2O3\mathrm{In}_2\mathrm{O}_3. The amplitude of hysteresis and saturating field increase with with longer annealing as shown in Table II. Ramping in magnetic field to ±7\pm7 T was performed in 0.7 T steps.

Resistivity (kΩ/□)BsB_s (T)α\alpha (mHz/T2^2)
323±0.13.6±0.7
173.6±0.15.4±0.5
8.56±0.23.1±0.3

caption: Hysteretic magnetization fit coefficients obtained through torque magnetometry. Magnetization was more pronounced as BsB_s increased with increased annealing and lower resistance

:::

amplitude of the magnetic hysteresis loop increased with annealing and confirms that the observed Δf0\Delta f_0 hysteresis arose from changes in the ITO. To rule out systematic factors, the order of data-taking was switched between the two 32 kΩ/□ datasets, and a 6 hour delay at +7 T was inserted between the sweep up and sweep down runs for the 17 kΩ/□ dataset. The lack of hysteresis at high magnetic field confirms that hysteresis in Δf0(B)\Delta f_0(B) arose from changes in the ITO instead of systematic factors. No hysteresis in magnetization was found for ITO with sheet resistance > 32 kΩ/□.

Magnetization hysteresis or hysteresis in Δf0\Delta f_0 above 1.5 T was fit to Δf0,hyst=α(B−BS)B\Delta f_{0,\mathrm{hyst}}=\alpha(B-B_S)B for saturating field BSB_S and proportionality constant α\alpha. Fit coefficients are provided in Table II. The amplitude of magnetization and BSB_S increased with annealing, as previously observed in ITO [16]. Note that stronger zz-axis ferromagnetism raises f0f_0, while in-plane magnetism can decrease f0f_0. The lower f0f_0 during the ramp down in magnetic field is thus consistent with in-plane ferromagnetism. In-plane ferromagnetism has been observed previously in ITO, and the higher training fields are consistent with a largely out-of-plane applied field instead of a purely in-plane field [16, 19].

The most annealed 8.5 kΩ\Omega/□ sample displayed a hysteretic magnetization saturation of Δm≈0.002μB\Delta m \approx0.002\mu_B/f.u. where f.u. is the In2O3\mathrm{In}_2\mathrm{O}_3 unit cell volume. If we assume that the associated moment with each defect is of order ∼1μB\sim1\mu_B, we obtain an approximate magnetic defect density of 1.3×10191.3 \times10^{19} defects/cm3^3, within a factor of 2 of the carrier density shown in Table I. However, the strong in-plane anisotropy observed in AHE measurements as well as the large anisotropy field suggest that the relevant oxygen defects and vacancies concentrate at surfaces and interfaces, presumably due to the granular structure as shown in Fig. 3. If we distribute the moments only at the surface of the film, and assume that because of surface roughness the effective surface area is increased by a factor of two, we obtain a surface magnetization of ∼0.1μB\sim0.1\mu_B/f.u.. One out of 10 unit cells at the surface introduces a magnetic moment if each surface defect produces ∼1μB\sim1\mu_B. Finally, the anisotropy energy KuK_u can be approximated from the saturation magnetization and field. The approximate Ku=BSΔmS/2≈2×104K_u=B_S\Delta m_S/2\approx2\times10^4 J/m3^3 is about 10 times smaller than a typical anisotropy energy of a thin film of polycrystalline ferromagnetic Fe3O4\mathrm{Fe}_3\mathrm{O}_4 of similar thickness [48].

Anomalous Hall Conductivity Near MIT

The most direct demonstration of intrinsic ferromagnetism is the emergence of the anomalous Hall effect in the transverse channel. Using Corbino disk torque magnetometry, we can detect both the Hall and anomalous Hall effects. However, unlike a standard Hall bar configuration, the sample is voltage-biased and transverse conductivity σxy(B)\sigma_{xy}(B) is directly measured by the magnetic moment created by circulating Hall currents. For a material with zz-component of magnetization MzM_z and coefficients aa and bb,

σxy=aB+bMz⇒δf0(±V,B)∝B2+βBMz.(7)\sigma_{xy}=aB+bM_z \Rightarrow\delta f_0(\pm V,B)\propto B^2+\beta BM_z. \tag*{(7)}

The extracted and fit σxy\sigma_{xy} from δf0(±V,B)\delta f_0(\pm V,B) is shown in Fig. 7. The even δf0(±V,B)\delta f_0(\pm V,B) is fit to Eqn. 7 assuming a diamagnetic sigmoidal MzM_z appearing at the saturating field listed in Table II. Such behavior would be expected from a magnetic transition from an in-plane ferromagnet to an out of plane paramagnet. The region of varying anomalous Hall signal corresponds with the hysteresis loop bounds shown in Fig. 6 as expected. The amplitude of the anomalous Hall signal also increased in the more conductive device as is expected based on the increased strength of magnetism with annealing.

Finally we comment on the sign change of the 32 kΩ\Omega/□ sample Hall conductivity shown in Fig. 7(a). Here we note that this sample was close to but on the insulating side of the MIT and thus at the onset the variable range hopping regime. A sign change of the anomalous Hall effect is then possible as a function of the impurity band filling [23]. As our films experienced strong variations in oxygen vacancies and defects as the morphology changed through annealing, such an effect may be anticipated. Indeed, such sign reversal was previously observed in (In0.27Co0.73)2O3−v(\mathrm{In}_{0.27}\mathrm{Co}_{0.73})_2\mathrm{O}_{3-v} (vv denotes the oxygen vacancies) ferromagnetic semiconductors [49] in the VRH regime, particularly as the temperature was reduced and localization effects dominate. As materials like ITO are based on the In2O3\mathrm{In}_2\mathrm{O}_3 oxide system where oxygen vacancies and morphology contribute to the creation of the impurity band, the observation of sign change in our ITO samples may have the same origin.

Fits to the even component of $\delta f_0(\pm V,B)$

Figure 7. Fits to the even component of δf0(±V,B)\delta f_0(\pm V,B) assuming a sigmoidal kink in anomalous Hall signal at BsB_s. For comparison, BsB_s as fit from the closing of the Δf0\Delta f_0 hysteresis in Figure 6 is shown with a vertical line.

Hysteretic Transport

The detection of AHE is also accompanied by hysteresis in δf0(±V,B)\delta f_0(\pm V,B). As shown in Figure 8, there is a hysteretic component of δf0(±V,B)\delta f_0(\pm V,B) with peaks coinciding with the closing of the f0f_0 hysteresis loops. The smoothly varying non-hysteretic component of δf0(±V,B)\delta f_0(\pm V,B) is the average between the up and down magnetic field sweep values of δf0(±V,B)\delta f_0(\pm V,B) and is used to extract Hall conductivity and magnetoresistance. The hysteretic component of δf0(±V,B)\delta f_0(\pm V,B) is the difference between the smoothly varying background and δf0(±V,B)\delta f_0(\pm V,B) in each sweep direction. Hysteresis in transport closes above BSB_S and qualitatively peaks near BSB_S for annealed magnetic ITO cantilever datasets. The hysteretic δf0(±V,B)\delta f_0(\pm V,B) is approximately 0 at low field, although such a closing may be due to the lack of dipole sensitivity at B=0B = 0. While hysteretic δf0(±V,B)\delta f_0(\pm V,B) signal cannot be simply ascribed to a single source as both magnetoresistance and Hall conductivity have contributions from sample magnetization, the fact that the hysteresis profile is not entirely even or odd in BB suggests that both components are hysteretic. Finally, the δf0(±V,B)\delta f_0(\pm V,B) hysteresis coincided with the hysteresis in magnetization and its evolution with annealing, confirming that both the transport and magnetization hysteresis originate from the ITO.

Hysteresis in transport or $\delta f_0(\pm V,B)$

Figure 8. Hysteresis in transport or δf0(±V,B)\delta f_0(\pm V,B) for 32 kΩ\Omega/□ and 17 kΩ\Omega/□ annealed devices. Peaks near the closing of the hysteresis loops are highlighted in boxes. The Corbino disk voltage was ±0.3\pm0.3 V. Hysteresis could appear due to both magnetoresistance for B≲BsB \lesssim B_s and the anomalous Hall effect as B≈BsB \approx B_s.

C ITO Hall Bar Transport

To confirm our findings using Corbino disk torque magnetometry and highlight the power of the technique, we performed complementary measurements using a standard Hall bar configuration. For a given temperature, the sample was current-biased, and the longitudinal and transverse voltages were recorded as a function of magnetic field. However, detection of the anomalous Hall effect in the transverse channel in such a configuration may be difficult due to sample issues such as strong in-plane anisotropy or high sample resistance [15, 19, 20]. Negative magnetoresistance due to the reduction in spin-related scattering thus is typically used to relate transport to direct magnetization measurements in ITO. Hysteretic magnetoresistance would provide convincing evidence of ferromagnetism, pointing to the presence of magnetic domains that need to be flipped in direction when the magnetic field is reversed.

Initially resistive ITO was sputtered in a Hall bar pattern and successively annealed to detect the onset of magnetism. Note that Hall bar measurements were performed on more conductive ITO films of the same thickness as the Corbino disk samples due to the difficulties in measuring high resistance samples using a Hall bar technique. Relative magnetoresistance or δR/R(0)≡[R(B)−R(0)]/R(0)\delta R/R(0) \equiv[R(B)-R(0)]/R(0) is shown in Figure 9(a). The B=0B=0 resistance was 15 kΩ/□\Omega/\square for the first anneal and 4.6 kΩ/□\Omega/\square for the second anneal. More strongly annealed or conductive samples displayed more dramatic decreases in δR/R0\delta R/R_0 at ∣B∣<2|B|<2 T and asymptotic δR/R0\delta R/R_0 for ∣B∣>5|B|>5 T. Such changes with annealing are consistent with previous studies of ITO with changing carrier density and temperature [50]. Magnetoresistance of the ITO Hall bar mounted at a 20° angle is also shown in Figure 9(a) as a test for anisotropic magnetoresistance. However, any changes in δR/R0\delta R/R_0 with mounting angle were minimal and could be explained by slight differences in annealing between the vertically and angle-mounted Hall bars. Resistivity and carrier density of the Hall bar samples are included in Table I.

Figure 9. (a) Magnetoresistance for a lightly annealed and more strongly annealed ITO sample. (b) Hysteresis in magnetoresistance after two stages of annealing for a vertically and 20° mounted Hall bar. Hysteresis is slightly more pronounced for the vertically mounted sample. Magnetoresistance was observed by a continuous sweep in BB for the 20° angle mounted dataset and by stopping at set BB for the vertically mounted dataset.

Hall effect and magnetoresistance data were further examined for evidence magnetism. Hysteretic magnetoresistance below 2 T for both the vertically mounted and angled ITO Hall bars after annealing is shown in Figure 9(b). The sign of hysteresis is consistent with previously observed in-plane magnetic ordering [5, 16, 51] and appeared only with annealing or increased granularity. No hysteretic magnetoresistance was observed in the more lightly annealed dataset likely due to either increased magnetism with annealing or a lower slope in δR/R(0)\delta R/R(0) near B=0B=0. No discernible anomalous Hall contribution could be detected, and the absence of AHE is consistent with strong in-plane magnetic anisotropy.

The Corbino disk cantilevers exhibit significantly higher BsB_s than the hysteresis-closing BB in the Hall bar samples shown in Figure 9(b). The saturating field observed in the more conductive Hall bar ITO sample is lower than any of the saturating fields of the magnetic Corbino disk samples but is closest to that of the more lightly annealed 32 kΩ/□\Omega/\square samples. Magnetization in ITO thus is not dictated purely by longitudinal resistivity and results from an interplay between granularity and carrier density.

Conclusions

We have shown unambiguous evidence of in-plane ferromagnetism and its effect on the transport properties of low carrier density ITO annealed through its MIT. Using a novel Corbino disk torque magnetometry technique [32], hysteresis in the voltage-independent shift in f0f_0 with BB provides a direct measurement of in-plane magnetic ordering, while a finite voltage δf0(±V,B)\delta f_0(\pm V,B) exhibits AHE and hysteretic transport. In particular, the observation of AHE provides direct evidence of inherent magnetism in ITO. Examination of both the Hall and anomalous Hall contributions to δf0(±V,B)\delta f_0(\pm V,B) confirm the breaking of an in-plane magnetic ordering. Magnetism arose with annealing in ITO and thus with changes to the magnetic oxide morphology. The observation of intrinsic magnetism in ITO correlated with changes in surface morphology supports the hypothesis that magnetism arises from surface oxygen vacancies. Such Hall and magnetization measurements may be performed for deeply insulating materials due to the direct measurement of σxy\sigma_{xy}. Initial results of σxy\sigma_{xy} in resistive ITO contradict expectations for both a Hall insulator and standard models of Mott variable range hopping [52, 53, 54]. Additionally, Corbino disk torque magnetometry provides the ability to simultaneously measure the bulk magnetic and transport properties of a material across the MIT, opening avenues for new physics.

Acknowledgments

This work was funded by the Army Research Office grant W911NF1710588 and by the Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division, under Contract DEAC02-76SF00515.

A Low Field f0f_0 Dependence

The full profile of Δf0\Delta f_0 with magnetic field for 4 annealing strengths is shown in Figure 10. Hysteresis appears as fluctuations in Δf0\Delta f_0 between neighboring BB points when both sweep directions are plotted together. Without subtracting the background change in f0f_0 such hysteresis is most clear in the 8.5 kΩ\Omega sample.

Figure 10. Voltage-independent Δf0\Delta f_0 for four cooldowns. There is rapid increase in Δf0\Delta f_0 in the yellow highlighted region of BB near 0 T seen in all cooldowns regardless of sample magnetization. The low field Δf0\Delta f_0 is therefore not attributed to the ITO and is not fit in Figure 6.

References

References

  1. [1]Y. Matsumoto, Room-temperature ferromagnetism in transparent transition metal-doped titanium dioxide, Science 291, 854 (2001).pubmed.ncbi.nlm.nih.gov/11228146
  2. [2]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. Das Sarma, H. D. Drew, R. L. Greene, and T. Venkatesan, High temperature ferromagnetism with a giant magnetic moment in transparent co-doped sno₂−δ, Phys. Rev. Lett. 91, 077205 (2003).DOI
  3. [3]P. Sharma, A. Gupta, K. V. Rao, F. J. Owens, R. Sharma, R. Ahuja, J. M. Guillen, B. Johansson, and G. A. Gehring, Ferromagnetism above room temperature in bulk and transparent thin films of mn-doped zno, Nature Materials 2, 673 (2003).pubmed.ncbi.nlm.nih.gov/14502276
  4. [4]A. Tiwari, V. M. Bhosle, S. Ramachandran, N. Sudhakar, J. Narayan, S. Budak, and A. Gupta, Ferromagnetism in co doped ceo2: Observation of a giant magnetic moment with a high curie temperature, Applied Physics Letters 88, 142511 (2006).
  5. [5]N. H. Hong, J. Sakai, N. T. Huong, and V. Brizé, Room temperature ferromagnetism in laser ablated ni-doped in2o3 thin films, Applied Physics Letters 87, 102505 (2005).DOI
  6. [6]G. Peleckis, X. Wang, and S. X. Dou, High temperature ferromagnetism in ni-doped in2o3 and indium-tin oxide, Applied Physics Letters 89, 022501 (2006).DOI
  7. [7]O. D. Jayakumar, I. K. Gopalakrishnan, S. K. Kulshreshtha, A. Gupta, K. V. Rao, D. V. Louzguine-Luzgin, A. Inoue, P.-A. Glans, J.-H. Guo, K. Samanta, M. K. Singh, and R. S. Katiyar, Structural and magnetic properties of in2o3 system: Prepared by gel combustion method, Applied Physics Letters 91, 052504 (2007), https://doi.org/10.1063/1.2757589.
  8. [8]C.-Y. Park, S.-G. Yoon, Y.-H. Jo, and S.-C. Shin, Room-temperature ferromagnetism observed in mo-doped indium oxide films, Applied Physics Letters 95, 122502 (2009), https://doi.org/10.1063/1.3232243.
  9. [9]M. Venkatesan, C. B. Fitzgerald, and J. M. Coey, Unexpected magnetism in a dielectric oxide, Nature 430, 630 (2004).pubmed.ncbi.nlm.nih.gov/15295588
  10. [10]N. H. Hong, J. Sakai, N. Poirot, and V. Brizé, Room-temperature ferromagnetism observed in undoped semiconducting and insulating oxide thin films, Phys. Rev. B 73, 132404 (2006).
  11. [11]A. Sundaresan, R. Bhargavi, N. Rangarajan, U. Siddesh, and C. N. Rao, Ferromagnetism as a universal feature of nanoparticles of the otherwise nonmagnetic oxides, Physical Review B 74, 10.1103/physrevb.74.161306 (2006).
  12. [12]J. Philip, N. Theodoropoulou, G. Berera, J. S. Moodera, and B. Saptati, High-temperature ferromagnetism in manganese-doped indium-tin oxide films, Applied Physics Letters 85, 777 (2004).DOI
  13. [13]T. Ohno, T. Kawahara, H. Tanaka, T. Kawai, M. Oku, K. Okada, and S. Kohiki, Ferromagnetism in transparent thin films of fe-doped indium tin oxide, Japanese Journal of Applied Physics 45, 10.1143/jjap.45.1957 (2006).DOI
  14. [14]M. Venkatesan, R. D. Gunning, P. Stamenov, and J. M. Coey, Room temperature ferromagnetism in mn- and fe-doped indium tin oxide thin films, Journal of Applied Physics 103, 10.1063/1.2835476 (2008).
  15. [15]H. S. Kim, H. Ji, H. Kim, S.-K. Hong, D. Kim, Y. E. Ihm, and W. K. Choo, Observation of ferromagnetism and anomalous hall effect in laser-deposited chromium-doped indium tin oxide films, Solid State Communications 137, 41 (2006).DOI
  16. [16]A. M. Hakimi, F. Schoofs, M. G. Blamire, S. Langridge, and S. S. Dhesi, Intrinsic and extrinsic ferromagnetism in co-doped indium tin oxide revealed using x-ray magnetic circular dichroism, Advances in Condensed Matter Physics 2017, 1 (2017).
  17. [17]H. S. Majumdar, S. Majumdar, D. Tobjörk, and R. Österbacka, Ferromagnetism in indium tin-oxide (ito) electrodes at room temperature, Synthetic Metals 160, 303 (2010).DOI
  18. [18]J. M. Coey, P. Stamenov, R. D. Gunning, M. Venkatesan, and K. Paul, Ferromagnetism in defect-ridden oxides and related materials, New Journal of Physics 12, 053025 (2010).DOI
  19. [19]A. M. Hakimi, The quest for a dilute magnetic oxide, in Magnetism and spin transport studies on indium tin oxide (University of Cambridge, 2011) pp. 63–96.
  20. [20]J. Stankiewicz, F. Villuendas, and J. Bartolomé, Magnetic behavior of sputtered co-doped indium-tin oxide films, Phys. Rev. B 75, 235308 (2007).DOI
  21. [21]L.-F. m. c. Arsenault, B. Movaghar, P. Desjardins, and A. Yelon, Magnetotransport in the insulating regime of mn-doped gaas, Phys. Rev. B 78, 075202 (2008).DOI
  22. [22]M. A. Paalanen, S. Sachdev, R. N. Bhatt, and A. E. Ruckenstein, Spin dynamics of nearly localized electrons, Phys. Rev. Lett. 57, 2061 (1986).pubmed.ncbi.nlm.nih.gov/10033622
  23. [23]A. A. Burkov and L. Balents, Anomalous hall effect in ferromagnetic semiconductors in the hopping transport regime, Phys. Rev. Lett. 91, 057202 (2003).DOI
  24. [24]N. M. Ahmed, F. A. Sabah, H. Abdulga four, A. Alsadig, A. Sulieman, and M. Alkhoaryef, The effect of post annealing temperature on grain size of indium-tin-oxide for optical and electrical properties improvement, Results in Physics 13, 102159 (2019).DOI
  25. [25]A. M. Hakimi, The indium oxide system, in Magnetism and spin transport studies on indium tin oxide (University of Cambridge, 2011) pp. 41–60.
  26. [26]S. Song, T. Yang, J. Liu, Y. Xin, Y. Li, and S. Han, Rapid thermal annealing of ito films, Applied Surface Science 257, 7061 (2011).DOI
  27. [27]J. H. Lee, Y. H. Kim, S. J. Ahn, T. H. Ha, and H. S. Kim, Grain-size effect on the electrical properties of nanocrystalline indium tin oxide thin films, Materials Science and Engineering: B 199, 37 (2015).DOI
  28. [28]L. Kerkache, A. Layadi, E. Dogheche, and D. Rémiens, Annealing effect in dc and rf sputtered ito thin films, The European Physical Journal Applied Physics 39, 1 (2007).DOI
  29. [29]D. Raouf, A. Kiasatpour, H. R. Fallah, and A. S. H. Rozatian, Surface characterization and microstructure of ito thin films at different annealing temperatures, Applied Surface Science 253, 9085 (2007).
  30. [30]M. Gulen, G. Yildirim, S. Bal, A. Varilci, I. Belenli, and M. Oz, Role of annealing temperature on microstructural and electro-optical properties of ito films produced by sputtering, Journal of Materials Science: Materials in Electronics 24, 467 (2012).
  31. [31]W.-F. Wu and B.-S. Chiou, Effect of annealing on electrical and optical properties of rf magnetron sputtered indium tin oxide films, Applied Surface Science 68, 497 (1993).DOI
  32. [32]S. Mumford, T. Paul, S. H. Lee, A. Yacoby, and A. Kapitulnik, A cantilever torque magnetometry method for the measurement of hall conductivity of highly resistive samples, Review of Scientific Instruments 91, 045001 (2020).DOI
  33. [33]K. A. Modic, M. D. Bachmann, B. J. Ramshaw, F. Arnold, K. R. Shirer, A. Estry, J. B. Betts, N. J. Ghimire, E. D. Bauer, M. Schmidt, and et al., Resonant torsion magnetometry in anisotropic quantum materials, Nature Communications 9, 10.1038/s41467-018-06412-w (2018).DOI
  34. [34]M. Perfetti, Cantilever torque magnetometry on coordination compounds: from theory to experiments, Coordination Chem-
  35. [35]J. Chiaverini, K. Yasumura, and A. Kapitulnik, Microcantilever studies of angular field dependence of vortex dynamics in bi₂Sr₂CaCu₂O₈−x, Phys. Rev. B 64, 014516 (2001).DOI
  36. [36]A. C. Bleszynski-Jayich, W. E. Shanks, B. Peaudecerf, E. Gignossar, F. von Oppen, L. Glazman, and J. G. E. Harris, Persistent currents in normal metal rings, Science 326, 272 (2009), http://science.sciencemag.org/content/326/5950/272.full.pdf.DOI
  37. [37]E. Finot, A. Passian, and T. Thundat, Measurement of mechanical properties of cantilever shaped materials, Sensors 8, 3497 (2008).pubmed.ncbi.nlm.nih.gov/27879891
  38. [38]O. M. Von Corbino, Phys. Z. 12, 561 (1911).
  39. [39]N. Manavizadeh, F. A. Boroumand, E. Asl-Soleimani, F. Raissi, S. Bagherzadeh, A. Khodavari, and M. A. Rasouli, Influence of substrates on the structural and morphological properties of rf sputtered ito thin films for photovoltaic application, Thin Solid Films 517, 2324 (2009), thin Film Chalogeni de Photovoltaic Materials (EMRS, Symposium L).DOI
  40. [40]L. Friedman and M. Pollak, The hall effect in the variable-range-hopping regime, Philosophical Magazine B 44, 487 (1981), https://doi.org/10.1080/01411868108222858.
  41. [41]T. Holstein, Ann. Phys. (N.Y.) 8, 343 (1959).
  42. [42]O. Entin-Wohlman, A. G. Aronov, Y. Levinson, and Y. Imry, Hall resistance in the hopping regime: A "hall insulator"?, Physical Review Letters 75, 4094 (1995).
  43. [43]O. Viehweger and K. Efetov, Phys. Rev. B 44, 1168 (1991).
  44. [44]S. Kivelson, D.-H. Lee, and S.-C. Zhang, Global phase diagram in the quantum hall effect, Phys. Rev. B 46, 2223 (1992).pubmed.ncbi.nlm.nih.gov/10003898
  45. [45]Y. Imry, Zero-temperature frequency-dependent hall conductivity of the anderson insulator, Phys. Rev. Lett. 71, 1868 (1993).pubmed.ncbi.nlm.nih.gov/10054520
  46. [46]P. F. Hopkins, M. J. Burns, A. J. Rimberg, and R. M. Westervelt, Magnetic-field-induced localization in degenerately doped n-type ge, Phys. Rev. B 39, 12708 (1989).pubmed.ncbi.nlm.nih.gov/9948141
  47. [47]D. W. Koon and T. G. Castner, Hall effect near the metal-insulator transition, Phys. Rev. B 41, 12054 (1990).pubmed.ncbi.nlm.nih.gov/9993657
  48. [48]E. Liu, Z. Huang, J.-G. Zheng, J. Yue, L. Chen, X. Wu, Y. Sui, Y. Zhai, S. Tang, J. Du, and H. Zhai, Texture induced magnetic anisotropy in fe3o4 films, Applied Physics Letters 107, 172403 (2015), https://doi.org/10.1063/1.4934744.
  49. [49]R. M. Qiao, S. S. Yan, T. X. Wu, M. W. Zhao, Y. X. Chen, G. L. Liu, W. L. Yang, R. K. Zheng, and L. M. Mei, Anomalous hall effect in variable range hopping regime: Unusual scaling law and sign reversal with temperature (2014), arXiv:1406.5672 [cond-mat.mes-hall].arxiv.org/abs/1406.5672
  50. [50]A. Fujimoto, K. Yoshida, T. Higaki, Y. Kimura, M. Nakamoto, Y. Kashiwagi, M. Yamamoto, M. Saitoh, T. Ohno, S. Furuta, and et al., Negative magnetoresistance of indium tin oxide nanoparticle thin films grown by chemical thermolysis, Journal of the Physical Society of Japan 82, 024710 (2013).DOI
  51. [51]M. Majumder, M. Schmidt, H. Rosner, A. A. Tsirlin, H. Yasuoka, and M. Baenitz, Anisotropic ru3 4d5 magnetism in the alpha-rucl3 honeycomb system: Susceptibility, specific heat, and zero-field nmr, Physical Review B 91, 10.1103/physrevb.91.180401 (2015).
  52. [52]R. Németh and B. Mü llschlegel, Hopping Hall conductivity in disordered and granular systems, Solid State Communications 66, 999–1001 (1988).DOI
  53. [53]D. Koon and T. Castner, Variable-range-hopping and the hall coefficient in si:as, Solid State Communications 64, 11–14 (1987).DOI
  54. [54]M. Gruenewald, H. Mueller, P. Thomas, and D. Wuertz, The hopping hall mobility — a percolation approach, Solid State Communications 38, 1011–1014 (1981).DOI

Paper details

vlnc:W_tdma7nc8
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents