Reversible Control of the Mn Oxidation State in SrTiO3 Bulk Powders

We demonstrate a low-temperature reduction method for exhibiting fine control over the oxidation state of substitutional Mn ions in strontium titanate (SrTiO3) bulk powder. We employ NaBH4 as the chemical reductant that causes significant changes in the oxidation state and oxygen vacancy complexation with Mn2+ dopants at temperatures <350°C where lattice reduction is negligible. At higher reduction temperatures, we also observe the formation of Ti3+ in the lattice by diffuse-reflectance and low-temperature electron paramagnetic resonance (EPR) spectroscopy. In addition to Mn2+, Mn4+, and the Mn2+ complex with an oxygen vacancy, we also observe a sharp resonance in the EPR spectrum of heavily reduced Mn-doped SrTiO3. This sharp signal is tentatively assigned to surface superoxide ion that is formed by the surface electron transfer reaction between Ti3+ and O2. The ability to control the relative amounts of various paramagnetic defects in SrTiO3 provides many possibilities to study in a model system the impact of tunable dopant-defect interactions for spin-based electronic applications or visible-light photocatalysis.


INTRODUCTION
The oxide SrTiO 3 is a classic perovskite-type member of the valuable ABO 3 semiconductor family. The promising properties such as a large tunable dielectric constant, structural phase transitions, superior charge storage capacity and tunable electronic structure have made SrTiO 3 an exciting candidate for a wide range of multifunctional applications (Weaver, 1959;Faughnan, 1971;Mattheiss, 1972;Müller and Burkard, 1979;Kamalasanan et al., 1993). Although in ambient conditions it exhibits a wide band gap and low electron mobility, introducing impurity dopants and intrinsic defects radically influence the conductivity and optical properties of the host material (Wild et al., 1973;Kozuka et al., 2010). The function of a semiconductor is intimately related to the chemistry and physics of native and targeted defects. The rich defect chemistry enabled by native oxygen vacancies (V O ) in semiconductors such as SrTiO 3 , PbTiO 3 , and BaTiO 3 has been correlated with numerous functions including ferroelectricity, visible-light photocatalysis, and multiferroics. These V O defects can donate up to two electrons to the host lattice. Transition metal dopants may also impart additional functionality that result from partially-filled d-orbitals. For example, Cr 3+ dopants, and Rh 3+ dopants in SrTiO 3 can reduce protons to generate H 2 gas using visible light that creates an oxidized dopant ion and a conduction band electron, e cb (Ishii et al., 2004;Sasaki et al., 2009;Kato et al., 2013). However, undesirable defects such as Cr 6+ can form to maintain charge neutrality, but limit the photochemical efficiency by serving as a trap for the e cb . These types of high-valent defects can be removed by either post-synthetic annealing under reducing atmospheres (Zuo et al., 2010;Tan et al., 2014;Lehuta and Kittilstved, 2016), co-doping the host lattice with additional n-type dopants (Chan et al., 1981;Kato and Kudo, 2002;Wang et al., 2014), irradiating with UV light (Wang et al., 2006), or applying a large electrical bias (La Mattina et al., 2008). Of these, the only potentially "green" reduction source could be UV irradiation from the sun. However, we note that the reported photoreduction step using a 400 W Hg-lamp in Cr:SrTiO 3 powders was of the order of tens of hours. The realization of a fast, low-energy method to modulate the oxidation state of transition-metal dopants in SrTiO 3 and related metal oxide semiconductors could impact various fields such as visible-light photocatalysis, sensing, and spin-based electronics. To this end, recent studies on the photodoping of colloidal Cr:SrTiO 3 nanocrystals show promise (Harrigan and Kittilstved, 2018).
We previously studied the effect of a relatively lowtemperature NaBH 4 reduction reaction on the oxidation state of Cr dopants in SrTiO 3 and related Sr 2 TiO 4 bulk powders Lehuta et al., 2017). In those studies, we observed an order of magnitude increase in the Cr 3+ concentration by EPR spectroscopy that we attributed to the reduction of EPR-silent high-valent Cr 4+ and Cr 6+ ions. The increase in the Cr 3+ concentration in n-type SrTiO 3 presents an interesting scenario where the Cr 3+ ion has a dual role of being an electron donor and a paramagnetic ion (S = 3/2). In addition, these observed changes are quantitatively reversible upon annealing the powders in air.
An isoelectronic analog of Cr 3+ is Mn 4+ , which is known to also occupy the Ti 4+ -site in SrTiO 3 . Although additional defects are required to maintain charge neutrality in Cr 3+ :SrTiO 3 , Mn 4+ in the B-site of SrTiO 3 is an isovalent dopant. Amongst the transition-metal doped oxides, Mn:SrTiO 3 has recently received extensive attention due to its complex and unique behavior than intrinsic SrTiO 3. The concurrent doping of Mn and oxygen vacancies in SrTi 1−x Mn x O 3−δ is reported to promote ferromagnetic ordering, dielectric permittivity and possible metallic behavior (Savinov et al., 2008;Choudhury et al., 2011Choudhury et al., , 2013Middey et al., 2012;Thanh et al., 2014). These observations make nonstoichiometric Mn:SrTiO 3 a highly attractive candidate for spin-based electronics applications. Although not completely understood, the results are attributed to the interplay of redoxactive Mn ions and the intrinsic charge compensating defects. In this regard, quantitative research is challenging due to a lack of experimental control over the interactions, and the complexity of Mn ions present in multiple oxidation states. Herein we report on the nature of the oxidation state of Mn ions and associated defect centers in bulk Mn:SrTiO 3 powders. We utilized various dopantspecific spectroscopic probes to elucidate the Mn oxidation state including EPR and diffuse-reflectance spectroscopies. We extend the use of NaBH 4 as a solid-state reductant to monitor changes in the three, unique Mn-related species as well as oxygenrelated defects and "self-doped" Ti 3+ ions. Comparison to other studies of reduced Mn:SrTiO 3 and noticeable absences of certain EPR-active Mn-centers is also discussed. We also observe a new signal in reduced samples that we attribute to superoxide anions, O − 2 .

Synthesis of Bulk Mn-Doped SrTiO 3
Bulk powders of SrTi 1−x Mn x O 3−δ (abbreviated Mn:SrTiO 3 ) were synthesized by a conventional solid-state reaction method, where x is the nominal concentration of Mn (x = 0.001) and δ is the concentration of oxygen vacancies. Briefly, Sr(NO 3 ) 2 , Mn(NO 3 ) 2 ·4H 2 O, and TiO 2 were mixed in the desired stoichiometry and ground with a mortar and pestle for about 10 min. The mixture was then transferred to a porcelain combustion boat and placed in the center of a tube furnace inside a quartz insert. The reaction mixture was heated in air for 6 h at 1,000 • C, reground for 10 min, then heated again for an additional 16 h at 1,000 • C.

NaBH 4 Reductions and Reoxidation
Chemical reductions of the bulk powders were carried out using a modified version of reduction previously described by our group for Cr:SrTiO 3 . For each reduction, FIGURE 1 | Powder XRD patterns of 0.1% Mn:SrTiO 3 before (black) and after T red = 425 • C (red). The calculated pattern for SrTiO 3 (cubic phase) is shown in gray as sticks (Yamanaka et al., 2002). Inset: expanded region of the (110) diffraction peak clearly showing a shift to lower 2θ after T red = 425 • C.
an amount of powder was mixed in a 1:1 mole ratio with NaBH 4 using a mortar and pestle for 5 min and then placed in a porcelain combustion boat in the middle of a quartz insert in a tube furnace. The atmosphere in the quartz insert was continuously purged by a controlled flow of Ar gas monitored by a rotameter (Matheson 7300). The samples were heated at temperatures ranging from 300 to 425 • C in 25 • C increments under Ar flow for 30 min. After reducing, samples were cooled under Ar to room temperature, washed and centrifuged alternately with deionized water and ethanol to ensure complete removal of NaBH 4 . After washing, samples were dried in an oven at 100 • C for 2 h. Reoxidation was performed by aerobically annealing the reduced samples until the physical color of the sample reversed to Mn:SrTiO 3 as-prepared sample.

Characterization
Powder X-ray diffraction (XRD) patterns were collected at room temperature using a Bragg-Brentano configuration with Cu K-α source (Rigaku SmartLab SE Diffraction System with cross-beam optics and D/Tex 250 Ultra 1D Si strip detector). X-band quantitative EPR spectra were collected at room temperature in 4 mm quartz EPR tubes (Wilmad-Glass) in a double rectangular resonator cavity (Bruker Elexsys E-500 with ER 4105DR cavity). Room temperature quantitative EPR spectra were collected consecutively on chemically perturbed samples (either oxidized or reduced) and an as-prepared sample using identical sample placement and instrument settings FIGURE 2 | (A) Scanned color images of Mn:SrTiO 3 powders as a function of T red . (B) Diffuse-reflectance spectra of Mn:SrTiO 3 as a function of reduction temperature (as-prepared = black; T red at 350 • C (blue), 375 • C (green), 400 • C (orange), and 425 • C (red), respectively). Data were normalized at 380 nm (3.26 eV; denoted by the circle) after setting the lowest y-value in the spectrum as the "zero." The T red = 400 • C spectrum was smoothed for presentation. (Eaton et al., 2010). The resonance field positions in the EPR spectra for each paramagnetic Mn center were simulated using the "resfields" function in EasySpin using the reported EPR parameters from literature and referenced below (Stoll and Schweiger, 2006). Low-temperature X-band EPR spectra were measured at 77 K on powders using the perpendicular mode of a dual-mode resonator cavity with a quartz finger dewar insert (Bruker Elexsys E-500 with ER-4116 cavity) ensuring the sample height exceeded the cavity height for quantitative analysis. Diffuse-reflectance spectra were collected with an integrating FIGURE 3 | Room temperature X-band EPR spectra of 0.1% Mn-doped SrTiO 3 bulk powder before (as-prepared, black) and after NaBH 4 reduction for 30 min at various temperatures (300-425 • C). The dotted lines in the 400 and 425 • C spectra are selected regions of the spectra that are multiplied (×) by 50 and 100, respectively. The calculated resonance fields for the hyperfine-split transitions of Mn 4+ (black), Mn 2+ -V ·· o (red), and Mn 2+ (blue) in SrTiO 3 are shown as vertical markers. The calculated resonance fields were simulated for each Mn center observed in the EPR with EasySpin (see characterization). Data were collected on the same day in a double-resonator cavity equipped with an as-prepared sample in the 2nd resonator and used to standardize sample intensities. The g-values and |A| obtained in this work are corroborated with previous reports in the literature. a An additional large axial component to the zero-field splitting was also estimated, D = 0.544 cm −1 .
FIGURE 4 | Relative EPR peak intensities plotted on a log scale as a function of NaBH 4 reduction temperature (T red ) for Mn:SrTiO 3 . The peak intensities are normalized vs. the intensity of the Mn 4+ EPR signal in the as-prepared sample. Quantitative measurements performed using a dual-cavity EPR resonator (see characterization section).
sphere (Ocean Optics ISP-REF) coupled by fiber optics to a CCD-based spectrophotometer (Ocean Optics USB2000+ VIS-NIR). The optical density between the absorption minimum and the absorption at 320 nm was adjusted by diluting the powders with MgO.

RESULTS AND DISCUSSION
The room temperature powder XRD patterns of as-prepared and reduced (T red = 425 • C) Mn:SrTiO 3 are shown in Figure 1. All samples designated Mn:SrTiO 3 contain nominally 0.1% Mn content. Both the as-prepared and reduced samples indicate the presence of the cubic phase of SrTiO 3 (Mitchell et al., 2000). However, a clear increase in the lattice parameter is observed after reduction. This result is consistent with other observations and has been attributed to both changes in ionic size and electronic effects after reduction of the lattice (Janotti et al., 2012). For example, the reduction of both Mn ions (Mn 4+ → Mn 2+ ) and lattice cations (Ti 4+ → Ti 3+ ) would result in larger ions leading FIGURE 5 | Quantitative reversibility of the EPR signal of Mn:SrTiO 3 bulk powder as-prepared (solid black line), after T red at 400 • C (solid orange line), and after T air at 500 • C for ∼1 h (gray dotted line). The weak Mn 4+ and Mn 2+ features after NaBH 4 reduction are still observed in the 50× scaled spectra (orange dotted lines).
to lattice expansion (Shannon, 1976). No appreciable secondary phases were observed after low-temperature chemical reduction despite clear spectroscopic changes in the samples (vide infra).
The electronic structure of Mn:SrTiO 3 is dependent on the nature of the Mn-ion speciation (i.e., oxidation state(s) and firstcoordination sphere). Mn 4+ has a d 3 electronic configuration yielding a 4 A 2g ground state when substituted at the Ti 4+ -site in SrTiO 3 . The physical appearance of the Mn:SrTiO 3 as-prepared (oxidized) powders is off-white and gradually turns to black with increased reduction temperature as shown in Figure 2A. The black appearance of SrTiO 3 has been previously observed and indicates reduction in the SrTiO 3 lattice resulting in self-trapped electrons localized at Ti 3+ sites (Tan et al., 2014;Lehuta and Kittilstved, 2016). The diffuse-reflectance spectra corroborates the assignment of the black color to excitations from Ti 3+ to conduction band states also referred to as a metal-to-metal charge transfer (MMCT) transition in the near-IR region (Khomenko et al., 1998). In the Mn:SrTiO 3 powder this MMCT appears at T red between 350 and 375 • C. The sub-bandgap tailing absorption at ∼2.9 eV has been assigned to excitations from the valence band to V O 's with different charge states (Mitra et al., 2012). With increasing T red , the relative intensity of the V O -related transitions decreases and disappears at T red = 375 • C and is consistent with electron accumulation in the V O states. Both spectral changes observed here for Mn:SrTiO 3 with increasing T red are similar to our recent study on chemically-reduced Cr:SrTiO 3 .
We also do not observe Mn-centered transitions from the diffuse-reflectance spectra which we attribute to either (1) low concentrations of Mn 3+ or Mn 4+ , which have spin-allowed transitions in the visible, or (2) the Mn ions are primarily in their +2 oxidation state, which only has spin-forbidden transitions when the d-electrons order in the high-spin configuration (S = 5/2, 6 A 1g ground state).
EPR-active species involving Mn ions in the +2, +3, and +4 oxidation states in SrTiO 3 have previously been reported (Müller, 1959;Serway et al., 1977;Azamat et al., 2012). However, Mn 3+ exhibits large zero-field splitting due to the S = 2 electronic spin state and thus, it is EPR-silent at conventional X-band frequencies (Azamat et al., 2012). The room temperature quantitative X-band EPR spectra of the Mn-doped SrTiO 3 samples are shown in Figure 3 as a function of reduction temperatures ranging from T red = 300-425 • C. The as-prepared sample consists of two sets of sextet peaks. In accordance with the reported g-value and hyperfine splitting constant (A) of Mn 4+ in SrTiO 3 , the main sextet in the as-prepared sample is assigned to Mn 4+ substituting for Ti 4+ with an isotropic g = 1.996 and |A| = 69.4 × 10 −4 cm −1 (Müller, 1959). The second and much weaker set of sextets is somewhat occluded by the Mn 4+ transitions, but the low-field resonances agree well with substitutional Mn 2+ at the Ti 4+ site in SrTiO 3 with g = 2.004 and |A| = 82.30 × 10 −4 cm −1 (Azzoni et al., 2000;Choudhury et al., 2013). Despite reports of both axial Mn 2+ -V ·· o and Mn 3+ -V · o complexes in Mn:SrTiO 3 , we do not observe these complexes in the as-prepared Mn:SrTiO 3 sample. Hence, only the substitutional Mn 4+ and Mn 2+ species in an octahedral oxide crystal field co-exist in the as-prepared Mn:SrTiO 3 powders.
After chemical reduction with NaBH 4 under Ar(g) at T red = 300 • C, a new, third set of transitions are detected near the Mn 2+ lines. Concomitant with the appearance of this new set of peaks is a decrease in the intensity of Mn 4+ lines and an increase in the relative intensity of Mn 2+ . The new set of lines agree well with the report of a substitutional Mn 2+ center at the Ti 4+ -site coupled to a doubly ionized oxygen vacancy (Serway et al., 1977). The reported EPR parameters of this Mn 2+ -V ·· o complex includes a large axial component to the zero-field splitting (D = 0.544 cm −1 ), g || = 2.003, and |A| = 76 × 10 −4 cm −1 . We were unable to detect any transitions at lower or higher magnetic fields likely from the low relative concentration and low nominal concentration of Mn in the lattice. The Mn 2+ -V ·· o complex forms at low temperatures before reduction of the SrTiO 3 lattice at T red < 375 • C (see Figure 2). One possible mechanism to explain the formation of this complex could be that oxygen vacancies may diffuse through the lattice and localize in the vicinity of Mn 4+ substitutional sites at low temperatures. This work demonstrates that mild reduction at only T red = 300-325 • C is sufficient to form the Mn 2+ -V ·· o complex in bulk powders. This result contrasts with the high temperature reductions above 825 • C previously used to create these centers in Mn:SrTiO 3 (Blazey et al., 1983;Kutty et al., 1986). In addition, we observe the coexistence of Mn 4+ and the Mn 2+ -V ·· o complex in the same EPR spectra at every T red . This does not agree with previous single crystal studies where the Mn 2+ -V ·· o complex was only observed when the Mn 4+ lines were fully removed upon reduction in 5% hydrogen for 3 h at 1,000 • C (Serway et al., 1977).
A new single feature centered at B 0 ∼ 350 mT (g ∼ 2.003) with no associated hyperfine structure was also observed in the EPR spectra after reduction. This feature increases in spectral intensity and also narrows with increasing T red . This feature is similar to a feature observed in Cr:SrTiO 3 at T red > 375 • C , but has a significantly larger relative intensity compared to the dopant EPR signal in the Mn:SrTiO 3 sample with the same nominal dopant concentration. This feature is tentatively assigned as the EPR-active superoxide FIGURE 7 | EPR spectra of Mn:SrTiO 3 bulk powders reduced at T red = 375 • C measured at 300 K (top) and 77 K (bottom). Sharp features in the 77 K spectra are attributed to artifacts resulting from bubbles that arise in the liquid nitrogen finger dewar.

anion (O −
2 ) adsorbed on the surface of SrTiO 3 and could form via a surface reaction between Ti 3+ and oxygen (Bykov et al., 2013;Harrigan et al., 2016). Table 1 below summarizes the EPR spectral parameters for previously reported Mn species in SrTiO 3 in the as-prepared and reduced Mn:SrTiO 3 .
The quantitative EPR spectra measured using the double resonator cavity shown in Figure 3 was analyzed and the relative intensity of the observed EPR centers is shown in Figure 4 as a function of T red . Compared to the as-prepared Mn 4+ signal intensity (defined as 1), a gradual decrease in the Mn 4+ and Mn 2+ EPR signals is observed with similar correlations in their temperature dependences. The signal for Mn 2+ is not detected for T red ≥ 375 • C. In contrast, the EPR intensity of the Mn 2+ -V ·· o complex shows little change and is more intense than the Mn 4+ and Mn 2+ EPR signals for T red ≥ 375 • C. The intensity of the Mn 2+ -V ·· o complex drops by an order of magnitude after increasing T red from 375 to 400 • C. At the highest temperature, T red = 425 • C, the EPR intensity of the O − 2 ion is nearly 3 orders of magnitude more intense than the Mn 2+ -V ·· o complex, indicative of substantial surface defects. Studies to identify the nature of this defect center are currently underway.
The EPR spectra of the Mn:SrTiO 3 powders after preparation, after T red = 400 • C, and after aerobic reoxidation at T air = 500 • C for ∼1 h are shown in Figure 5. The observed changes in the EPR spectra of the reduced samples revert to the asprepared EPR spectrum by aerobically annealing the sample. The process of forming Mn 2+ -V ·· o complex in the reduced samples is thus reversible. However, elevated temperatures and longer reoxidation times were required in contrast with the chemical reductions. Since the Mn 2+ -V ·· o complex is a charge-neutral complex in the lattice, it is expected to be at least metastable. The apparently slower reoxidation kinetics compared to reduction kinetics suggest a metastable complex.
The 300 K (room temperature) and 77 K (liquid N 2 ) EPR spectra of the Mn centers in the as-prepared and T red = 300 • C powders are shown in Figure 6. Two things are revealed from the EPR spectra of both as-prepared and lightly-reduced Mn:SrTiO 3 samples: (1) there is no evidence of self-trapped electrons at Ti 3+ sites in the lattice based on the 77 K spectra, and (2) the EPR intensity of Mn 2+ completely disappears at 77 K. At low temperature, the intensity of Mn 4+ is pronounced following the typical Boltzmann statistics. In contrast, the Mn 2+ EPR signal completely disappears at 77 K in these two samples. These results agree with a previous magnetic susceptibility and EPR study of the Mn 2+ signal vanishing, where the behavior was attributed to increased antiferromagnetic interactions between adjacent Mn 2+ ions with decreasing temperature (Azzoni et al., 2000). This explanation cannot be extended to describe the EPR signal of the Mn 2+ -V ·· o complex, which does not disappear at 77 K in the T red = 300 • C sample. To confirm the behavior of EPR signals as a function of temperature, the EPR spectrum of the reduced sample at 300 K was repeated after cooling it to 77 K and the entire spectrum is nearly identical.
The cryogenic EPR measurements were also collected for samples reduced above 350 • C to reveal the effect of Ti 3+ defects on the EPR spectra that are observed in the diffusereflectance spectra shown in Figure 2. Figure 7 shows the 300 K and 77 K EPR spectra of T red = 375 • C. The paramagnetic Ti 3+ defects are not observed in the EPR performed at 300 K due to fast spin-lattice relaxation but are promptly observed at 77 K Harrigan and Kittilstved, 2018). At 77 K, the T red = 375 • C sample is dominated by a broad and intense asymmetric Ti 3+ lattice defect centered at g = 1.94. The appearance of this fast-relaxing defect, however, has no apparent effect on the linewidth of the Mn-centers nor the single line that we tentatively assign to surface-adsorbed O − 2 ions. We recently showed that linewidth and relaxation-dynamics of substitutional Cr 3+ ions in SrTiO 3 powders and colloidal nanocrystals can be significantly altered when Ti 3+ defects are present in the lattice through a near-resonant cross-relaxation process Harrigan and Kittilstved, 2018). This same behavior is not observed for any of the Mn-centers in the reduced SrTiO 3 powder.

CONCLUSIONS
A low-temperature chemical reduction technique has been implemented for tunability of the Mn dopant oxidation states and the related intrinsic defects in bulk Mn:SrTiO 3 . We employed a myriad of structural and spectroscopic techniques on samples subjected to a systematic chemical reduction. Both isotropic Mn 4+ and Mn 2+ species were identified in the as-prepared powders. Following the thermal reduction, the samples exhibited a continuous decrease in Mn 4+ EPR signal and an increase in the Mn 2+ intensity, accompanied by the introduction of a Mn 2+ -V ·· o complex. We demonstrate that our chemical treatment at merely T red = 300-325 • C generates sufficient driving force to significantly reduce the intensity of the octahedral Mn 4+ and Mn 2+ dopants and form the Mn 2+ -V ·· o complex. All the Mn peaks showed distinctive changes at low-temperature in the EPR that are readily reversible upon warming back the samples. Reductions at 375 • C and above generated significant concentrations of Ti 3+ defects that were confirmed by diffusereflectance and low-temperature EPR spectroscopy. All the observed perturbations in the reduced samples are entirely reversible by aerobic annealing at elevated temperatures. We also observe an intense spectral feature in the EPR spectrum in heavily-reduced Mn:SrTiO 3 powders that we attribute to O − 2 ions at the surface. This fast and effective strategy offers a general lowtemperature reduction process that allows tunability and control over the rich dopant-defect chemistry in transition-metal doped SrTiO 3 materials.

DATA AVAILABILITY
The datasets generated for this study are available on request to the corresponding author.

AUTHOR CONTRIBUTIONS
HM and KL carried out the experiments, data analysis and interpretation, and edited the manuscript. WH contributed to the interpretation of the results and edited the manuscript. HM and KK contributed to the data analysis and interpretation and wrote the manuscript.