Tuning Stoichiometry for Enhanced Spin‐Charge Interconversion in Transition Metal Oxides

Interconversion of spin and charge current provides a key route for low‐power spin memory and logic devices. Recent advances have revealed efficient spin‐charge interconversion in 4d and 5d transition metal oxides. However, the strategies to tune the conversion efficiency, essential for the generation and detection of spin‐current, are limited to engineering the crystalline structure of oxides. Here, a simple and broadly applicable approach by tuning the cation stoichiometry is reported. In the model system of 5d perovskite SrIrO3, it is shown that a significant Ir cation deficiency is induced by controlling the oxygen partial pressure during deposition. This off‐stoichiometry leads to an enhancement of the spin‐to‐charge conversion efficiency by around three times, accompanied by an increase of electrical resistivity at room temperature. Furthermore, a significant increase of inverse spin Hall voltage is observed by implementing the Ir‐deficient SrIr1‐xO3, highlighting the promising role of atomic defects in developing oxides for sensitive spin‐current detection. This work opens a new pathway to engineer the spin‐charge interconversion efficiency in oxides and offers new opportunities to integrate complex oxides in energy‐efficient spintronic devices.

In this study, we show that introducing cation vacancy provides a simple and effective approach to enhance the spin-tocharge conversion efficiency and electric resistivity in oxides.We use perovskite SrIrO 3 as the model system, given the versatile tunability and high compatibility with functional oxides such as ferroelectrics and multiferroics.By reducing the oxygen partial pressure during the deposition, we find that the Ir to Sr ratio (Ir/Sr) significantly deviates from one, suggesting the presence of Ir vacancies in the perovskite structure.This off-stoichiometry largely increases the resistivity of SrIr 1-x O 3 , likely due to the increase of carrier scattering.We further show that the vacancies also largely enhance the spin-to-charge conversion efficiency of SrIr 1-x O 3 from ≈0.16-0.22 to 0.55-0.63.Similar modulation effects have also been demonstrated in the 4d ruthenates, further revealing the general applicability of this approach in oxides.Owing to the simultaneous increase of conversion efficiency and resistivity, a clear increase of inverse spin Hall voltage is demonstrated in the heterostructures of ferromagnetic oxides and Irdeficient SrIr 1-x O 3 , suggesting the promising role for sensitive spin-current detection.Our results demonstrate a hitherto neglected route to develop transition metal oxides for spintronic applications.

Tuning Cation Stoichiometry in SrIr 1-x O 3 Films
Epitaxial SrIrO 3 films were grown on DyScO 3 substrates with varying oxygen partial pressure ranging from 5 to 15 Pa.All films exhibit the perovskite structure as revealed by the X-ray diffraction (XRD) spectra in Figure 1a.The pseudo-cubic (pc) (002) Bragg peak of SrIrO 3 continuously shifts to the lower angle as decreasing the oxygen partial pressure, revealing the expansion of the out-of-plane pseudo-cubic lattice (c pc ) from 3.97 to 4.05 Å. Figure 1b shows the narrow full width at half max-imum (FWHM) of rocking curves, indicating the high crystallinity of all films.The high-quality growth of SrIrO 3 is further confirmed by the atomic force microscopy (AFM) measurements, showing the atomic-flat surfaces with step-terrace structures (Figure 1c; Figure S1, Supporting Information).Reciprocal space mapping (RSM) results reveal that the SrIrO 3 films show orthorhombic structure, with the orthorhombic axis c o lying inplane and parallel to the c axis of the DyScO 3 substrates (Figure S2, Supporting Information).All films remain fully strainclamped by the substrates, suggesting that the lattice expansion is not likely due to strain relaxation (Figure S2, Supporting Information).
To check the stoichiometry of the SrIrO 3 films, we employed Rutherford backscattering (RBS) measurements that reveal the Ir/Sr ratios by fitting the counts of backscattering 4 He (Figure S3, Supporting Information).Figure 1d shows that the Ir/Sr ratio largely decreases by decreasing the oxygen partial pressure, consistent with the trend observed in previous studies. [42,43]The presence of significant Ir deficiency is further supported by X-ray photoemission spectroscopy (XPS) data showing a large decrease of the Ir absorption peak intensity (Figure S4, Supporting Information).High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image (Figure 1e) confirms that the SrIrO 3 grown at 5 Pa still shows the perovskite structure without clear evidences of secondary phases, in accordance with the XRD data in Figure 1a.Therefore, these results strongly suggest that the low oxygen partial pressure induces the formation of Ir vacancies in the perovskite SrIrO 3 , likely attributed to the high volatility of iridium oxides and the propagation of laserablation plumes. [42]Note that the low pressure could also lead to the presence of oxygen vacancies in our films, although XPS data suggest a less important role and quantitative estimation remains challenging (Figure S4, Supporting Information).38][39][40][41] Next, we studied the electric transport properties of SrIrO 3 films grown under different oxygen pressures using the Hall bar device.Due to the spin-orbit splitting band and on-site Coulomb interaction, SrIrO 3 exhibits a correlated metallic state, [44] which can be tuned by strain, [45] dimensionality, [46] and back-gating. [47]he temperature dependence of resistivity shown in Figure 1f reveals the strong influence of vacancies on the transport properties of SrIrO 3 .The results show that SrIrO 3 films grown under high oxygen partial pressure exhibit lower resistivities as compared to those grown at reduced pressures.All films show weak temperature dependence along with a metal-to-insulator transition behavior.The transition temperature, characterized by the slope change in the temperature-dependent resistivity curve, is ≈260 K for films under high oxygen pressure and ≈40 K for those under low pressure.These results suggest a rise in carrier density with reduced oxygen partial pressures, which are further supported by the ordinary Hall resistivity measurements (Figure S5, Supporting Information).Therefore, the increased resistivity is mainly attributed to the decreased carrier mobility, stemming from the enhanced defect scattering by Ir vacancies and potential oxygen vacancies as oxygen pressure decreases below 10 Pa.

Spin-To-Charge Conversion in SrIr 1-x O 3
Spin-to-charge conversion efficiencies were assessed by using two independent experimental techniques: spin-torque ferromagnetic resonance (ST-FMR) and in-plane second harmonic Hall (SHH) measurements at room temperature.Figure 2a depicts the setup for ST-FMR measurement, where a radio frequency (rf) current is applied to induce resonance of the magnetic moment, resulting in an oscillation in resistance due to anisotropic magnetoresistance (AMR). [48]By sweeping the external in-plane magnetic field, the DC rectifying voltage (V mix ) is measured, which can be fit by: [48] V whereas H ext and H res represent the external magnetic field and resonance field, respectively, and ΔH denotes the linewidth.As shown in Figure 2b, the rectifying voltage signal primarily consists of two components: the antisymmetric component (V a ) that is related to Oersted filed and filed-like torque ( FL ), and the symmetric component (V s ) that is mainly determined by the damping-like torque ( DL ).Therefore, the ratio between V s and V a serves as an indicator of the spin-to-charge efficiency.Figure 2c shows the angular dependence of V s and V a , both conform well to the cos(Φ B )sin(2Φ B ), where Φ B is the angle between rf current and external magnetic field (Figure 2a).Note that the ratio between V s and V a remains constant regardless of the magnetic field direction.As shown in Figure 2d, the V s become more dominant as reducing the oxygen partial pressure, indicating an increase of  DL .Furthermore, the ST-FMR signals for each sample were measured within a frequency range of 5-10 GHz (Figure 2e), all of which exhibit a good fit to the Lorentzian function.By neglecting the much weaker  FL , the spin-to-charge conversion efficiency () measured at different frequencies can be calculated by using the following equation: [48] whereas e, h, and M s represent the electron charge, reduced Planck constant, and the saturation moment of Py, respectively.Moreover, the thicknesses of Py (t Py ) and SrIrO 3 (t SIO ) are controlled as 5 nm and ≈18 nm, respectively, to balance the spinorbit torque and Oersted field.We further used X-ray reflectivity measurements to precisely determine the layer thickness in each sample.As summarized in Figure 2f, the spin-to-charge conversion efficiencies, independent with the frequency, increase from 0.16 to 0.63 as decreasing oxygen partial pressure.
To further qualify the spin-to-charge conversion efficiency, the second harmonic Hall voltage (V xy 2 ) was measured while rotating the in-plane magnetic field respect to the alternating current (Figure 3a).The external magnetic field, Oersted field and spinorbit torque collectively determine the position of magnetic moment, thereby influencing the Hall resistance through the planar and anomalous Hall effect. [49]The angular dependence of the first harmonic Hall signal in Figure 3b exhibits the sin(2Φ B ) angular dependence due to the planar Hall effect, consistent with whereas B Oe denote the external magnetic field and the Oersted field induced by current, R AHE , R PHE , and R ANE are anomalous Hall, planar Hall, and anomalous Nernst resistance, respectively, and  stands for the gyromagnetic ratio.In Equation 3, the cos(Φ B ) component is related to  DL and anomalous Nernst effect.The cos(Φ B )cos(2Φ B ) component is induced by  FL and Oersted field. [49]The sin(2Φ B ) component mainly arises from a thermal gradient along with the current. [25]Due to the shunting effect in the bilayer, ≈25% to 30% of the total current flows through the SrIrO 3 layer grown at high pressures, and reduces to ≈12% by increasing Ir vacancies.Nevertheless, all samples show reasonably good second harmonic signals (Figure S6, Supporting Information).As shown in Figure 3c, the measured second harmonic signals as a function of the angle (Φ B ) are well fitted by the three contributions in Equation 3, demonstrating a good agreement with theoretical analysis.To extract  DL , the second harmonic resis-tances were measured with varying magnetic fields to eliminate the contribution from anomalous Nernst effect (R ANE ). Figure 3d shows the dependence of the coefficient of the cosΦ B term on 1/(B+μ 0 M s ), all exhibiting good linear correlations.Here, B+μ 0 M s is the effective field applied to Py, whereas μ 0 M s corresponds to the demagnetizing field and the magneto-crystalline anisotropy field of Py is negligible.The spin-to-charge conversion efficiency can be calculated by using the linear slope ( DL /2) and following equation: whereas the J SIO is the current density in the perovskite iridates.Figure 3e summarizes the spin-to-charge conversion efficiencies obtained from these two measurement methods (ST-FMR and SHH) and the resistivity of the SrIr 1-x O 3 in the bilayer.Despite of variations in exact magnitudes (Section SVII, Supporting Information), both datasets consistently demonstrate a similar trend in their dependence on the oxygen partial pressure.Remarkably, the conversion efficiencies largely increase from 0.16-0.22 to 0.55-0.63,accompanied by the increase of resistivity by approximately four times at room temperature.This enhancement is likely correlated to the presence of vacancies, which act as barriers and enhance the spin-to-charge conversion efficiency via additional disorder scatterings. [2,50]To further demonstrate the applicability of this strategy, we also studied the 4d oxides SrRuO 3 grown under varying oxygen pressures, showing simi- lar changes in structure and spin-to-charge conversion efficiency (Figure S7, Supporting Information).Figure 3f shows the dependence of the measured spin Hall conductivity  sh on electric conductivity  xx , whereas  sh is calculated by using  sh =  xx .The magnitude of  sh gradually reduces as  xx decreases, suggesting that the SrIr 1-x O 3 is located near the boundary of the clean and dirty metal region, [51][52][53][54] in which the  sh is mainly attributed to the intrinsic interband transition and  can be enhanced by increasing scatterings. [55,56]

Inverse Spin Hall Voltage of SrIr 1-x O 3
The inverse spin Hall effect, which transforms spin-current to electric voltage, is commonly used for spin-current detection.In a simplified model, the inverse spin Hall voltage (V ISHE ) is expected to be proportional to the spin-to-charge conversion efficiency () and the electric resistivity ( xx ).Therefore, the spin Hall resistivity (defined as  sh =  xx ) could provide an indicator to evaluate the materials for spin-current detection. [6]As compared to heavy metals such as Pt [48,52] and W, [57] the transition metal oxides generally show higher  sh due to the large  and high  xx .Figure 4a summarizes of the  sh and  xx of typical transition metal oxides, revealing that the 5d oxide SrIrO 3 provides a promising material candidate.More importantly, our strategy to introduce the Ir cation vacancy further increases the  sh of SrIr 1-x O 3 owing to the increases of both  and  xx , which could further improve the sensitive detection of spin-current.
To assess the role of cation vacancy, we conducted spin pumping measurement [24,58,59] using all-oxide heterostructures comprised of La 0.66 Sr 0.33 MnO 3 (LSMO, 15 nm) and SrIr 1-x O 3 (SIO, 5 nm).The ferromagnetic La 0.66 Sr 0.33 MnO 3 was opted as the spin-current source layer due to the minimal spin rectification effects as compared to the highly conductive Py. [59] Considering the lattice mismatch, SrTiO 3 substrate was used to grow La 0.66 Sr 0.33 MnO 3, showing excellent ferromagnetic properties at room temperature (Figure S8, Supporting Information).We further demonstrate that SrIrO 3 on SrTiO 3 substrates also shows the similar structure changes by reducing the growth oxygen partial pressure, further suggesting that the control of Ir vacancies by growth is universal on different substrates (Figure S9, Supporting Information).As displayed in Figure 4b, the heterostructures with SrIrO 3 close to stoichiometry exhibit a relatively low V ISHE ≈0.49 μV.As we introduce cation vacancy in SrIr 1-x O 3 , the V ISHE largely increases to 1.21 μV.Meanwhile, the resonance magnetic fields exhibit no clear shift, suggesting that the magnetic properties of La 0.66 Sr 0.33 MnO 3 remains almost the same.Note that the charge current induced by inverse spin Hall effect is shunted by the conductive La 0.66 Sr 0.33 MnO 3 layer.][61] We find that the relative increase of R is smaller than the enhancement of V ISHE from 0.49 to 1.21 μV, revealing the increased  that is consistent with ST-FMR and SHH results.Furthermore, we show that the V ISHE is further enhanced to 1.49 μV by inserting a thin layer (2 nm) of antiferromagnetic insulator LaFeO 3 (LFO) between La 0.66 Sr 0.33 MnO 3 and SrIr 1-x O 3 , which is likely attributed to the suppression of the shunting effect and may also be contributed by the enhancement of interfacial transparency.The increase of V about three suggests that engineering cation stoichiometry provides an efficient route for optimizing oxides for sensitive spin-current detection.

Conclusion
In conclusion, we report a hitherto neglected approach to tune the spin-to-charge conversion and electrical transport property of transition metal oxides, by controlling the cation stoichiometry.We show that the low oxygen partial pressure during growth leads to a significant Ir deficiency in the perovskite SrIr 1-x O 3 .The offstoichiometry largely increases both the longitudinal resistivity and the spin-to-charge conversion efficiency (from 0.16-0.22 to 0.55-0.63) of SrIr 1-x O 3 at room temperature.Moreover, we show that the use of Ir-deficient SrIr 1-x O 3 also increases the inverse spin Hall voltage induced by spin pumping, highlighting the pivotal role of engineering cation vacancy for sensitive spin-current detection.Our results provide an efficient and applicable strategy to develop oxide materials for further advancements in oxide spin-orbitronics.

Experimental Section
Film Growth: High-quality SrIrO 3 films were grown by pulsed laser deposition using a 248 nm KrF excimer laser at varying oxygen pressures ranging from 5 to 15 Pa.For ST-FMR and SHH measurements, DyScO 3 substrates with small lattice mismatch were used, which are less likely to introduce artifacts to the ST-FMR measurements at radio frequency, as compared to high dielectric substrates. [13,25]The films were grown at temperature 700 °C with laser energy density of ≈1.2 J cm −2 , then cooled down to room temperature at 10 °C min −1 in a 3000 Pa oxygen atmosphere.Permalloy (Py) was deposited on SrIrO 3 by DC magnetron sputtering at 3 mtorr Ar pressure.An extra 1.2 nm TiO x capping layer was added to prevent oxidation of Py films.For single layer SrIrO 3 film, an extra 2 nm SrTiO 3 capping layer was grown to prevent degradation during the fabrication process of device. [62]For inverse spin hall measurements, the heterostructures of La 0.66 Sr 0.33 MnO 3 and SrIrO 3 were grown on SrTiO 3 substrates.La 0.66 Sr 0.33 MnO 3 was grown at the same temperature with 25 Pa oxygen pressure.
Structure and Chemical Characterization: The crystal structure of SrIrO 3 film were characterized by X-ray diffraction (XRD, Malvern PANalytical Empyrean) with Cu-K radiation ( = 0.1542 nm).The atomic scale high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) images were performed by Cs-corrected JEM ARM200CF microscope operated at 200 kV using a high-angle annular detector for Zcontrast imaging with a collection angle of 90 -370 mrad.The chemical composition was measured by using Rutherford backscattering (RBS).
Electric Transport Measurement: The samples were patterned into Hall bars (16 μm × 80 μm) for electric transport measurements using photolithography and ion beam etching.Ti (5 nm)/Pt (60 nm) bilayers were deposited as electrodes of both devices.The resistivity and Hall effect were measured in Physical Properties Measurement System (PPMS, Quantum Design).
Spin-To-Charge Conversion Measurement: The spin-to-charge conversion efficiencies were measured by using spin-orbit torque ferromagnetic resonance (ST-FMR) and second harmonic Hall (SHH) measurements by using 16 μm × 80 μm stripes and Hall bars.In ST-FMR measurement, the radio frequency currents were supplied though a microwave source with nominal power of 15 dBm, modulated by an AC voltage (1 Vrms, 1713 Hz).

Figure 1 .
Figure 1.Structure and electric property of Ir-deficient SrIr 1-x O 3 .a) X-ray diffraction (XRD) spectra of SrIrO 3 films grown on DyScO 3 (110) substrates under different oxygen partial pressures.b) The rocking curve of SrIrO 3 films.c) The atomic force microscopy (AFM) image of SrIrO 3 /Py bilayer grown under 10 Pa of oxygen pressure.d) The Ir/Sr ratio measured by using Rutherford backscattering (RBS) and the pseudo-cubic lattice constants along the c-axis (c pc , out of plane) for different samples.e) The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image of SrIrO 3 films grown under 5 Pa of oxygen pressure, showing the single perovskite phase and good crystallinity.f) The resistivity as a function of temperature for different samples.

Figure 2 .
Figure 2. Spin-torque ferromagnetic resonance (ST-FMR) measurements.a) The schematic of ST-FMR measurement set-up.b) The typical ST-FMR spectra of a SrIrO 3 /Py bilayer measured at 7 GHz, which consists of both symmetric and antisymmetric contributions.c) The angular dependence of symmetric and antisymmetric components, exhibiting a typical sin(2Φ B )cos(Φ B ) dependence with respect to the magnetic field.The oxygen partial pressure for SrIrO 3 in (b) and (c) is 12.5 Pa.d) The rectifying voltage for different samples at 9 GHz with Φ B = 45°.The data is normalized for better visualization.e) The ST-FMR spectra for the SrIrO 3 /Py bilayer (12.5 Pa) measured within a frequency range of 5-10 GHz.f) The spin-to-charge conversion efficiency for different samples at frequencies of 5-10 GHz.

Figure 3 .
Figure 3. Second harmonic Hall (SHH) measurements.a) The schematic of SHH measurement.b) Angular dependence of the first harmonic Hall signal for the SrIrO 3 /Py bilayer (12.5 Pa).c) Angular dependence of the second harmonic Hall signal under an external magnetic field of 500 Oe for the SrIrO 3 /Py bilayer (12.5 Pa).d) The linear relationship between the cos(Φ B ) component of the coefficient and 1/(B+μ 0 M s ).e) The spin-to-charge conversion efficiency of different samples measured by both ST-FMR and SHH, together with the of SrIrO 3 .f) The calculated spin Hall conductivity plotted against the electrical conductivity.Solid lines are guide to eye.

Figure 4 .
Figure 4. Inverse spin Hall measurements.a) The spin Hall resistivity and electric resistivity of typical transition metal oxides.b) The inverse spin Hall voltage (V ISHE ) of all-oxide heterostructures with SrIr 1-x O 3 grown at different oxygen partial pressure.Inset shows the schematic of measurements set-up, where H is the external field and V ISHE is the measured inverse spin Hall voltage across the sample.The spin-current is generated by the spin pumping from La 0.66 Sr 0.33 MnO 3 to SrIr 1-x O 3 layer.All the data were measured at room temperature. substrate