Influence of single-ionized oxygen vacancies on the generation of ferromagnetism in SnO2 and SnO2:Cr nanowires

In this work, we report the influence of single-ionized oxygen vacancies (VO′\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V^{\prime}_{{\text{O}}}$$\end{document}) as a spin ½ system in the ferromagnetic response of undoped and Cr-doped SnO2 nanowires. For this study, undoped and Cr-doped SnO2 nanowires were synthesized by a thermal evaporation method. Raman, Auger, and X-ray photoelectron spectroscopies confirmed the incorporation of Cr3+ ions in the SnO2 lattice. Electron paramagnetic resonance measurements demonstrated the presence of single-ionized oxygen vacancies (VO′\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V^{\prime}_{{\text{O}}}$$\end{document}) in undoped and Cr-doped nanowires. Complementarily, cathodoluminescence measurements confirmed the presence of VO defects in the samples. Magnetic measurements revealed FM behavior from the undoped SnO2 and Cr-doped SnO2 nanowires, showing magnetization saturation values (MS) of ± 1 × 10–3 and ± 1.6 × 10–3 emu/g, respectively, and magnetic coercivity values (HC) of 180 and 200 Oe. We assign the FM response of nanowires to the presence of single ionized VO′\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V^{\prime}_{{\text{O}}}$$\end{document} acting as a spin ½ system and to the alignment of magnetic moments of Cr3+ ions, finding that VO′\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$V^{\prime}_{{\text{O}}}$$\end{document} defects dominate in the FM generation.


Introduction
Tin oxide (SnO 2 ) is a semiconductor that has received attention due to its use to fabricate gas sensors, solar cells, optoelectronic devices, and catalysts [1][2][3]. In addition, this semiconductor has also received importance in spintronics because it exhibits ferromagnetism (FM) at room temperature, induced by doping with non-magnetic and magnetic impurities [4][5][6]. SnO 2 doped with some transition metals, such as chromium [7][8][9], exhibit this property, however, with an origin not straightforward. Several authors have proposed that FM is due to exchange interactions between magnetic impurities and native point defects type oxygen vacancies (V O ) [7,10]. However, there are reports of first principle calculations predicting that tin-vacancy (V Sn ) defects generate FM in undoped SnO 2 and that oxygen vacancy (V O ) does not contribute to it [11,12]. In addition, other authors have suggested that ferromagnetism is a universal feature of metal oxide nanoparticles, assuming its origin to exchange interactions between unpaired electron spins and surface V O [13]. For SnO 2 doped with magnetic impurities, Garcia-Tecedor et al. demonstrated that chromium doping of SnO 2 nanostructures promotes rising in the luminescence associated with V O [14]. Furthermore, Duhan et al. recently compared the FM response of SnO 2 , SnO 2 :Cr, and SnO 2 :Fe, Cr, finding that the magnetic saturation increases in co-doped samples assigning this effect to the generation of V O [15]. Similarly, Urs et al. reported FM at room temperature in SnO 2 nanoparticles doped with Cr 3+ at concentrations lower than 5 at.%, attributing its origin to exchange interaction through F-centers of V O and dopant magnetic impurities, explaining such interaction through the Bound Magnetic Polaron (BMP) model [16]. The BMP model in DMS considers the formation of bound states, known as magnetic polarons, where a charge carrier, such as electrons, becomes trapped near a magnetic impurity, such as chromium [17,18]. Therefore, a study to identify single electrons trapped in V O and another species present in SnO 2 :Cr, whose spins participate in the FM generation, can help to understand the FM origin in this semiconductor. In this work, we report the presence of single-ionized oxygen vacancies ( V ′ O ) in SnO 2 nanowires doped with different concentrations of chromium using the electron paramagnetic resonance (EPR) technique, confirming the presence of V O by the cathodoluminescence (CL) technique.

Experimental methodology
SnO 2 and chromium-doped SnO 2 nanostructures were synthesized by thermal evaporation of SnO 2 (Aldrich 99.99%) and CrO 3 powders (Aldrich 99.99%) onto SiO 2 /Si(100) substrates, in a homemade horizontal furnace operated at 130 mTorr and using Ar (Infra 99.999%) as carrier gas. A mechanical pump maintained low pressure in the furnace while Ar flow was regulated by a needle valve and measured by a mass flow meter (Omega type FMA-A2302). For the synthesis of the undoped sample, SnO 2 powder was placed in an alumina boat and maintained at 1300 °C. For the synthesis of Cr-doped samples, we prepared two samples, the first by maintaining SnO 2 and CrO 3 powders at 1300 and 1200 °C (sample 2), respectively; and the second by evaporating both SnO 2 and CrO 3 powders at 1300 °C (sample 3). This last to vary the evaporation rate of chromium in the furnace by considering that the melting temperature of the CrO 3 is 197 °C at atmospheric pressure. The SnO 2 and CrO 3 vapor diffused and condensed onto the SiO 2 /Si(100) substrates, placed at the downstream end of the quartz tube furnace at 620 °C. For the formation of Cr 3+ ions, we propose the decomposition of CrO 3 and SnO 2 precursors as follows: Thus, we propose chromium incorporation in SnO 2 following: Table 1 shows the growth temperature, argon flow, and chemical composition of samples measured by EDS.
The relative atomic composition of the samples was characterized by energy-dispersive spectroscopy (EDS) using an Oxford X-Max analyzed. The elemental quantification was calculated by Inca software (Oxford instruments) using a standard base sequence. The crystal structure was determined with a Phillips X'pert X-ray diffractometer using a CuKα (λ = 0.154 nm) line excitation source. Raman spectroscopy characterization was carried out with a Dimension-P2 λ s Raman system using an Nd 532 nm laser. X-ray photoelectron spectroscopy (XPS) and Auger electron spectroscopy (AES) were done in a PHI 535 system using an Al anode. High-resolution spectra were obtained using 500 scans during the XPS and AES measurements. AES measurements used an electron beam with an energy of 2 keV as the excitation source. All XPS spectra were calibrated using the C (1 s) signal (284.8 eV) as a reference and deconvoluted by software CASA XPS. Electronic paramagnetic resonance (EPR) measurements were carried out by using a JEOL, JES-TE300 system, operating at X-Band fashions at 100 kHz modulation frequency with a cylindrical cavity in the TE 011 mode. The external calibration of the magnetic field was conducted using a JEOL ES-FC5 precision gaussmeter. The acquisition and manipulation of spectra were performed using ES-IPRITS/TE. EPR spectra were recorded as the first derivative. The experimental acquisition frequency was 9.44 GHz at room temperature and a microwave power of 20 mW. For transmission electron microscopy (TEM) measurements a Jeol JEM-2010 operated at 200 keV was used. Samples morphology was studied using JEOL JSM-7800F. A Mono-CL4 Gatan system adapted to a JEOL FIB-4500 SEM system measured the cathodoluminescence (CL) signal from samples at room temperature and in the UV-Vis spectral range.
(2)  Table 1 lists the atomic composition of the samples measured by EDS, revealing that Sn concentration decreased by increasing the Cr amount in them, which suggests a substitutional incorporation of the magnetic impurity in the SnO 2 lattice. These results also reveal a notable low chromium concentration in samples, which we assign to their limited solubility in SnO 2 , as reported by other authors [14,16]. Furthermore, EDS results also show that the oxygen amount in samples decreased slightly due to Cr doping, corresponding with the absence of chromium oxide compounds and the possible generation of oxygen vacancies. XRD patterns obtained from samples revealed diffraction peaks corresponding with the rutile-type tetragonal structure of SnO 2 for all of them, as shown in Fig. 1, according to PDF card #88-0287. Such results did not reveal diffraction peaks associated with chromium oxide compounds, suggesting that Cr impurities were incorporated in the SnO 2 lattice. Further, XRD patterns of samples showed that (110), (101), and (211) peaks exhibited higher intensity, in agreement with other reports about the X-ray diffraction of SnO 2 nanowires [19,20]. Figure 1 also shows diffraction peaks corresponding to metallic Sn, which we attribute to the self-catalytic VLS growth mechanism of SnO 2 that we reported previously for undoped SnO 2 nanowires also obtained by the thermal evaporation method [21]. XRD patterns from sample 1 (curve 1), sample 2 (curve 2), and sample 3 (curve 3) with a rutile-type structure. The label (*) corresponds to metallic Sn signals generated by the self-catalyzed growth mechanism of nanowires, and the label (+) corresponds to the SiO 2 signal generated by the substrate  Figure 2 shows typical SEM images obtained from samples, revealing the formation of large SnO 2 nanowires with inhomogeneous diameters lower than 250 nm, besides the presence of irregular nanoparticles adhered to some Crdoped SnO 2 nanowires (Fig. 2b, c). EDS maps obtained from SnO 2 :Cr nanowires confirmed the presence of oxygen, tin, and chromium along them (Fig. 3). EDS measurements in SnO 2 :Cr nanoparticles adhered to nanowires revealed the same composition as Cr-doped SnO 2 nanowires, finding that their presence only represents a morphological variation of the semiconductor. Figure 4a shows a typical TEM image obtained from an undoped SnO 2 nanowire of sample 1, with a selected area electron diffraction (SAED) pattern corresponding with the [010] zone axis and revealing a growth direction along the [100] in agreement with other reports [19,22]. In addition, the zoom of the yellow square shown in Fig. 4a revealed that the sample exhibits a separation between the (101) planes of 2.6 Å (Fig. 4b), corresponding with the theoretically expected value for the rutile-type structure of SnO 2 . Figure 4c shows a typical TEM image of a nanowire of sample 2, also with a SAED corresponding with the [010]

Results and discussion
zone axis, although with extra diffraction points (arrows in inset) produced possibly by the presence of dislocations as the cause of crystal misorientation. The distance between the (101) planes on this nanowire recorded the same value obtained for sample 1, confirming a growth direction along the [100].
Raman measurements describe the chemical structure of crystals by evaluating their vibrational modes, and for rutile SnO 2 with a P 42 /mnm space group, the group theory indicates that its total vibrational modes at the Γ point of the Brillouin zone are given by [16,23]: Among them, only four A 1g , B 1g , B 2g , and E g modes are Raman active, and only O atoms vibrate while Sn atoms remain at rest. Figure 5, curve 1, shows three fundamental Raman peaks at 474, 632, and 774 cm −1 for the undoped SnO 2 sample, corresponding to the E g , A 1g , and B 2g vibration modes. The first two modes A 1g , B 2g , correspond with the O vibration in the perpendicular plane to the c-axis, while the E g mode arises from O-O vibration parallel to  [16,23,24]. In addition, the Raman spectra of this sample revealed a weak infrared (IR) active mode centered at 695 cm −1 , which corresponds to the A 2u longitudinal optical (LO) mode of the SnO 2 rutile structure [25,26]. Abello et al. have proposed that such IR mode becomes weakly active due to atomic disorder in the lattice structure [27], apparently by the presence of atomic defects in SnO 2 . Raman measurements of the chromium-doped samples revealed for sample 2 only the A 1g and B 2g signals (curve 2 in Fig. 5); in contrast, for sample 3, all vibrational modes found for the undoped SnO 2 sample were resolved, besides a peak at 552 cm −1 (curve 3 in Fig. 5). Urs et al. have reported that this mode is generated by the symmetric octahedral vibration Cr-O 6 present in the Cr 2 O 3 , suggesting that it can be assigned with the A 1g mode in the SnO 2 by considering the substitution of Sn 4+ by Cr 3+ ions, which also exhibit octahedral coordination with six oxygen ions in the rutile lattice [16].
Auger spectroscopy allowed us to determine the concentration of elements at the surface of SnO 2 nanowires since Auger electrons are generated in this technique via inelastic collision scattering with an average depth of 5.0 nm. Figure 6a shows the Auger spectra obtained from samples with their corresponding derivative curves (Fig. 6b-d). Sample 1 displays Auger peaks at about 421 and 429 eV that correspond to the Sn (MNN) transitions and a peak at 510 eV that corresponds to the O (KVV) transition [28]. Auger spectra of samples 2 and 3 show both Sn (MNN) and O (KVV) transitions besides an Auger peak centered at 528.9 eV corresponding with Cr (L3M23V) transition [29][30][31]. Table 2 shows the composition of samples calculated by AES measurements using the sensitivity factors reported in ref [32], revealing that the Cr-doping of SnO 2 nanowires increased its oxygen concentration at their surface. These results also exhibit higher chromium concentrations in samples than measured by the EDS technique, revealing that Cr 3+ predominantly incorporates at the surface of SnO 2 nanowires.
The chemical state of the elements in the samples was determined by XPS. Figure 7a displays the XPS spectra of the Sn 3d 5/2 and 3d 3/2 signals from samples centered at 486.3 and 494.7 eV, respectively, revealing an energy difference of 8.4 eV that corresponds with the spin-orbit splitting reported for the bound of Sn 4+ ions with O 2− ions in the SnO 2 lattice [7,33,34]. Figure 7a also shows that the Sn 3d 5/2 peak of samples exhibits only one symmetrical component, without shoulders, confirming the absence of Sn 2+ ions in all samples. Figure 7b shows the XPS spectra of the O 1s signal from samples centered at 531 eV and composed with a shoulder at about 533 eV. The deconvolutions of this XPS signal for each sample are shown in Fig. 8, which were calculated using two Gaussian curves centered at 530.8 ± 0.5, and 532.5 ± 0.5 eV. The component of 530.8 eV corresponds to the O-Sn bonds [21,34], and the component of 532.5 eV to single bonded oxygen atoms with different hydrocarbon molecules adsorbed onto the SnO 2 surface [35,36]. Figure 9 displays the Cr 2p 3/2 and 2p 1/2 signals obtained from sample 3 with binding energies of 576.5 and 586.5 eV, respectively, revealing an energy difference of 10 eV that corresponds with the value reported to Cr 3+ ions [37][38][39]. To confirm the chemical state 3 + of chromium we calculated the multiplet components of the 2p 3/2 signal reported by Biesinger et al. [40], which are shown in Fig. 9b. These components calculated with Gaussian curves were centered at 575.7, 576.7, 577.5, 578.5, and 578.9 obtaining a value for coefficient of determination R 2 of 0.9106.
CL measurements allow us to determine the luminescence of the point defects present in SnO 2 samples. Figure 11a displays the broad CL spectrum obtained from sample 1, revealing a shoulder at about 2.5 eV that suggests the presence of several components, which we calculated by deconvoluting the spectrum using Gaussian curves centered at 2.11, 2.50, and 2.79 eV with the same FWHM value of 0.5 eV, and values for the coefficient R 2 between 0.9983 and 0.9990. For the yellow component (2.11 eV), Zhou et al. reported in a time-resolved X-ray excited optical luminescence study that the SnO 2 yellow emission corresponds to radiative transitions involving oxygen vacancy-related states with intrinsic surface states [41]. However, Prades et al. demonstrated by density functional theory (DFT) and CL measurements that the SnO 2 yellow emission corresponds with the formation of surface oxygen vacancies 100° coordinated with Sn atoms (bridge oxygen vacancies), also proposing that the SnO 2 blue emission (2.79 eV) corresponds with the surface-V O 130° coordinated (in-plane oxygen vacancies) [42]. Similarly, other authors proposed that this blue emission corresponds with the radiative transition between the conduction band and surface states associated with V O [41][42][43][44]. The green component centered at 2.50 eV (Fig. 11a), corresponds with the SnO 2 green emission reported by several authors, centered at 2.58 eV, attributing an origin to shallow levels involving surface defects [14,22,45]. CL spectra from Crdoped samples exhibited the same components found for the undoped SnO 2 nanowires (Fig. 11b, c), although with a notorious increase in the CL intensity for Cr-doped nanowires of sample 3.
EPR measurements were used to identify the different paramagnetic radicals, or species, associated with point defects in SnO 2 and SnO 2 :Cr samples. We used the EasySpin MATLAB toolbox (supplementary information) to simulate the experimental EPR spectra and to calculate the g tensor values for the paramagnetic species present in samples. Figure 10a shows the EPR spectrum obtained from sample 1 in the range 335-340 mT (curve 1) with its corresponding simulated signal (curve 2) obtained for a g value of 2.0037, as a singlet with a line width peak-to-peak, ∆H p-p , of 0.34 mT assigned to the presence of S = 1/2, which match with the reported for single ionized oxygen vacancies ( V ′ O ) defects in SnO 2 [46][47][48]. A single ionized oxygen vacancy is an oxygen vacancy point defect with one trapped electron that acts as a ½ spin system with an associated magnetic moment, , which align along the direction of an external magnetic field, B, minimizing the energy E = − ⋅ B . Figure 10b and c show the presence of this same EPR signal for the Cr-doped samples 1 and 2, both with a g value of 2.0035 and ∆H p-p of 0.32 and 0.35 mT, respectively. This g value of 2.0035 matches with the g value of sample 1, considering that the uncertainty of the EPR system used is ± 0.0003 mT. Additionally, Fig. 10c line 1 reveals a notable decrease in the EPR intensity for sample 3. Figure 12a shows the magnetization vs. magnetic field (M-H) curve obtained from undoped SnO 2 nanowires (sample 1) after subtracting the diamagnetic behavior of the Si substrate [49], revealing a ferromagnetic behavior with magnetization saturation (M S ) of ± 1 × 10 -3 emu/gr in agreement with other authors [50], besides a small magnetic coercivity (H C ) of 180 Oe (inset in Fig. 12a) in agreement with the reported value by Zhang et al. for similar undoped SnO 2 nanowires [7]. Some authors have proposed that the intrinsic FM of undoped SnO 2 is due to the exchange interaction between unpaired electron spins arising from oxygen vacancies (V O ) [9,13], which we confirmed their presence by EPR Fig. 6 Auger spectra of the a samples 1-3 (labels 1-3), and b-d their corresponding derivative curves, dNE/dE, showing signals for Sn, O, and Cr  (Figs. 10a, 11a). In addition, the M-H curve of SnO 2 nanowires doped with chromium at 0.12 at.% (sample 2) also revealed an FM behavior, with M S and H C values of ± 1.6 × 10 -3 emu/g and 200 Oe, respectively, resulting be higher than observed to undoped SnO 2 (sample 1) due to the chromium incorporation in the rutile structure. The enhancement of magnetic saturation by incorporating magnetic impurities in other DMS has been reported previously [7,9], and particularly for sample 2, it could be explained in terms of the formation of V ′ O during chromium incorporation. Therefore, we propose that the formation of this paramagnetic specie is generated by a charge unbalance generated by incorporating Cr 3+ ions in the lattice, involving native-neutral oxygen vacancies. Using the Kröger-Vink notation [51,52], we suggest the following reaction to describe this effect during the substitution of Cr 3+ by Sn 4+  where × , ·, and ′ refer to neutral, positive, and negatively charged point defects. Thus, we assign the FM in SnO 2 :Cr to the alignment of the magnetic moment of Cr 3+ ions and spin ½ system single ionized V ′ O . Figure 12c shows the M-H curve acquired from sample 3, revealing a quenching for their magnetic coercivity and a decrease in the magnetic  In summary, we propose that the FM of Cr-doped SnO 2 nanowires is influenced by competitive electric charge transfer processes at their surface between oxygen vacancies and adsorbed oxygen, finding that V ′ O defects, acting as a spin ½ system, dominate the FM generation.

Conclusions
Undoped and Cr-doped SnO 2 nanowires were synthesized by thermal evaporation method varying the chromium concentration to evaluate the role of point defects in the generation of ferromagnetism (FM). XRD measurements revealed the formation of the rutile-type tetragonal structure in samples.
SEM images show the formation of large SnO 2 and SnO 2 :Cr nanowires with diameters lower than 250 nm. TEM images revealed that undoped and doped SnO 2 nanowires grew along the [100] direction. Raman spectroscopy revealed the vibrational modes E g , A 1g , A 2u (LO), and B 2g centered at 474, 632, 695, and 774 cm −1 , respectively, corresponding with the SnO 2 rutile structure, besides a mode centered at 552 cm −1 associated with the presence of Cr 3+ ions with octahedral coordination with oxygen ions in the sample with higher Cr concentration. Auger spectroscopy measurements from samples revealed the Sn (MNN) doublet signal at 421 and 429 eV, and the O (KVV) signal at 510 eV, while the Cr-doped samples the Cr (L3M23V) signal at 528.9 eV. Furthermore, XPS measurements revealed that chromium ions incorporated in the SnO 2 structure with chemical state 3 + . Electron paramagnetic resonance (EPR) results revealed a signal associated with a spin ½ system corresponding with single ionized oxygen vacancies ( V ′ O ), with g values of 2.0037 and 2.0035 for undoped and Cr-doped samples, respectively. Magnetic measurements revealed FM behavior from the undoped SnO 2 (sample 1) and SnO 2 :Cr doped with 0.12 at.% (sample 2), with magnetization saturation values (M S ) of ± 1 × 10 -3 and ± 1.6 × 10 -3 emu/g and magnetic coercivity values (H C ) of 180 and 200 Oe, respectively. We assigned the FM response of SnO 2 :Cr to the alignment of both the magnetic moment of Cr 3+ ions and spin ½ system single ionized V ′ O . Finally, we found a quenching in the FM signal from SnO 2 :Cr nanowires doped with Cr at 0.52 at.% (sample 3), which correlated with a decrease in the EPR Fig. 11 Deconvolution of the CL spectra obtained from a sample 1, b sample 2, and c sample 3 signal of V ′ O defects, revealing that such spin ½ system in SnO 2 dominates the FM generation.