Spectroscopic capture of a low-spin Mn(IV)-oxo species in Ni–Mn3O4 nanoparticles during water oxidation catalysis

High-valent metal-oxo moieties have been implicated as key intermediates preceding various oxidation processes. The critical O–O bond formation step in the Kok cycle that is presumed to generate molecular oxygen occurs through the high-valent Mn-oxo species of the water oxidation complex, i.e., the Mn4Ca cluster in photosystem II. Here, we report the spectroscopic characterization of new intermediates during the water oxidation reaction of manganese-based heterogeneous catalysts and assign them as low-spin Mn(IV)-oxo species. Recently, the effects of the spin state in transition metal catalysts on catalytic reactivity have been intensely studied; however, no detailed characterization of a low-spin Mn(IV)-oxo intermediate species currently exists. We demonstrate that a low-spin configuration of Mn(IV), S = 1/2, is stably present in a heterogeneous electrocatalyst of Ni-doped monodisperse 10-nm Mn3O4 nanoparticles via oxo-ligand field engineering. An unprecedented signal (g = 1.83) is found to evolve in the electron paramagnetic resonance spectrum during the stepwise transition from the Jahn–Teller-distorted Mn(III). In-situ Raman analysis directly provides the evidence for Mn(IV)-oxo species as the active intermediate species. Computational analysis confirmed that the substituted nickel species induces the formation of a z-axis-compressed octahedral C4v crystal field that stabilizes the low-spin Mn(IV)-oxo intermediates.

T he identification of intermediate species produced during catalysis provides insight into reaction mechanisms and methods for controlling the underlying reaction pathways. Generally, high-valent transition metal-oxo species are involved in the rate-determining steps of various oxygen atom transfer (OAT) reactions, including C-H bond activation, hydroxylation, and O-O bond formation [1][2][3][4] . Characterization and manipulation of high-valent transition metal-oxo species are thus crucial to designing heterogeneous catalysts for water oxidation reaction, where O-O bond formation is the key elementary step. The proposed mechanistic cycle of water oxidation reaction starts with the adsorption of a water molecule on the surface of transition metal oxide catalysts. Then, the adsorbed water molecule undergoes multiple protons and electrons transfer with a redox state change of the transition metal site to generate high-valent transition metal-oxo species involved in the O-O bond formation step. Understanding each elementary step and developing catalysts under neutral pH conditions have been recently getting much interest due to the advantage of environmentally friendly, less corrosive electrolyte and the possibility of combining with CO 2 reduction or microbial hybrid electrolysis cell [5][6][7] . While many experimental and computational results imply the existence of a high-valent metal-oxo species on the surface of heterogeneous catalysts that serves as an active reaction intermediate during the water oxidation reaction, very few results offering direct evidence of the chemical identity of intermediates have been reported to date for heterogeneous catalysts.
Various in-situ/ex-situ spectroscopic analyses have been performed on several 1st-row transition metal-based catalysts (Mn, Fe, Co, and Ni) for the characterization of intermediate highvalent metal-oxo species. The presence of Co(IV) species with a low-spin configuration (S = 1/2) was demonstrated for cobalt phosphate catalysts during electrochemical water oxidation reaction by ex-situ electron paramagnetic resonance (EPR) analysis 8 . The mechanism of the O-O bond formation step with reactive Co(IV)-oxo (reformulated as a Co(III)-oxyl radical) species is suggested based on the above EPR results, in-situ X-ray absorption spectroscopy results 9,10 , and further isotope labeling studies 11 . Recently, rapid-scan attenuated total reflectance infrared (ATR-IR) spectroscopy was used to identify two distinct Co (IV)-oxo and Co(III)-superoxide species from Co 3 O 4 nanoparticles (NPs) as reaction intermediates of the photochemical water oxidation reaction 12 . The mechanism involving the Co(III)superoxide intermediate species leads to a more efficient water oxidation reaction. ATR-IR spectroscopy was also applied to detect the presence of an Fe(IV)-oxo species as an intermediate during photoelectrochemical water oxidation on hematite films 13 , although the precise electronic structure remains elusive. Although there is no experimental data on the direct characterization of the active high-valent metal-oxo species in NiFe oxides/ hydroxides at aqueous electrolyte, the results of density functional theory (DFT) calculations suggest that Fe(IV) (high-spin d 4 ) and Ni(IV) (low-spin d 6 ) species are active sites for facile oxo formation and favorable O-O bond formation respectively 14 . Interestingly, Gray and colleagues experimentally capture cisdioxo-Fe(VI) intermediate species from NiFe hydroxide catalyst in a nonaqueous electrolyte 15 . Limitation of the substrate (H 2 O and OH − ) in nonaqueous media permits the accumulation of active intermediate species, which allows the use of various in-situ spectroscopic measurements (infrared, UV-Vis, Raman, luminescence, and Mössbauer spectroelectrochemistry). They suggested that the O-O bond formed through the internal redox rearrangement mechanism of cis-dioxo-Fe(VI) to Fe(IV)-peroxide intermediate species.
Photosystems in nature adopt a Mn 4 Ca cluster to oxidize water into oxygen with high efficiency [16][17][18] . Interestingly, the geometric structure changes flexibly, and the spin state of the Mn 4 Ca cluster interconverts between low-spin (S = 1/2 with an open-cubane form) and high-spin (S = 5/2 with a closed-cubane form) forms during catalysis 19 . Inspired by this unique Mn 4 Ca cluster, various Mn-based heterogeneous catalysts have been developed [20][21][22][23][24][25] . However, the performance of man-made Mn-based heterogeneous catalysts has long suffered from the sluggish formation of Mn(III) species 26,27 . In general, the rigid lattice in Mn-based bulk inorganic crystals suppresses the generation of energetically favorable Jahn-Teller-distorted Mn(III) species, leading to slow kinetics of charge accumulation 27,28 . For this reason, considerable research effort has been dedicated to the generation and stabilization of Mn(III) intermediates. Previously, we demonstrated that Mn(III) species can be stably generated on the surface of manganese oxide NPs via a proton-coupled electron transfer pathway in a neutral phosphate electrolyte 29 and further suggested that the rate-determining step involves Mn(IV)-oxo species rather than Mn(III) based on the in-situ/ex-situ spectroscopy analysis 30 .
Here we show that the engineering of the local distortions via Ni atom substitution in Mn 3 O 4 NPs manipulates the spin state of the reaction intermediates during water oxidation reaction. Using a combination of experiment and computation, we explore the effect of Ni atom substitution on the intermediate species of nanoscale manganese oxide NPs. We find that the Ni substitution enables the compression of surface Mn octahedron, resulting in low-spin Mn(IV)-oxo intermediate species during the water oxidation reaction.

Results and discussion
Electrode preparation and electrochemical characterization. Ni atom substitution in Mn 3 O 4 NPs (Ni-Mn 3 O 4 NPs) was accomplished through thermal annealing after nano-junction formation between the Ni source and Mn 3 O 4 NPs. To create a nano-junction, a vertical-type Mn 3 O 4 NPs/Ni(OH) x layered structure electrode is designed (Fig. 1a). The detailed sample preparation of the Ni-Mn 3 O 4 NPs/NiO layered structure is as follows. First, an amorphous nickel hydroxide layer was electrodeposited on a fluorine doped tin oxide (FTO) glass electrode. Monodisperse sub-10-nm Mn 3 O 4 NPs, synthesized by thermal decomposition method (Fig. 1b), were then spin-coated onto the nickel hydroxide films. Calcination of the resulting electrode at 300°C in air for 5 h yielded the Ni-Mn 3 O 4 NPs/NiO/FTO electrode configuration (Fig. 1a). Approximately 400-500-nm-thick Ni-Mn 3 O 4 NPs film was formed (Fig. 1d). High-resolution transmission electron microscopy (HR-TEM) analysis confirmed that the Mn 3 O 4 NPs were not altered with respect to size and shape by the calcination treatment. Electron energy loss spectroscopy (EELS) line scans were used to examine the composition of the synthesized catalysts and revealed that the nickel species was uniformly distributed within the assembled NPs from the bottom interface between the Mn 3 O 4 NPs layer and the NiO layer to the top surface of the electrode (Fig. 1c).
The electrochemical properties of Ni-Mn 3 O 4 NPs/NiO were characterized in 500 mM phosphate electrolyte (pH 7). Cyclic voltammetry (CV) curves were obtained and corrected by polarization to minimize the contribution of non-faradaic current to oxygen evolution reaction (OER) performance (Fig. 2a) Supplementary Fig. 2. The electrochemical kinetic and redox behaviors of Ni-Mn 3 O 4 NPs/NiO upon water oxidation were further investigated in detail. As shown in Fig. 2b, c, the Tafel slopes and pH dependency were 70 mV dec −1 and -78 mV pH −1 , respectively, in neutral phosphate electrolyte. The transfer coefficient was derived as~1 from the Tafel slope, which indicates that the rate-determining process is a chemical step, preceded by a one-electron pre-equilibrium step 30,31 . The pH dependency indicates that log(j) has a first-order dependence on pH and that the current density has an inverse first-order dependence on proton activity. Thus, concerted one-proton and one-electron transfer reactions occur as a pre-equilibrium step, followed by the chemical rate-determining process 31 . Similar electrokinetic behavior was observed in manganese oxide NPs 30 .
In addition, two distinct oxidation peaks were observed at 0.6 and 1.0 V (Fig. 2d). According to a previous study 30 , the oxidation peak at 0.6 V corresponds to the transition from Mn (II)-H 2 O to Mn(III)-OH. We thus speculated that high-valent Mn(IV) species are generated as reaction intermediates in the Ni-Mn 3 O 4 NPs/NiO system during the water oxidation reaction. In-situ X-ray absorption near edge structure (XANES) analysis ( Supplementary Fig. 3) further supported the observed Mn redox behavior, as the average Mn valence in Ni-Mn 3 O 4 NPs/NiO increased from 2.83 to 3.45 with increasing applied potential, whereas the valence state of Ni is only slightly increased.
In-situ and ex-situ spectroscopic analysis. To capture the generated intermediate Mn species on the NPs surface, potentialdependent in-situ UV-Vis spectra for the Mn 3 O 4 NPs and Ni-Mn 3 O 4 NPs/NiO electrodes were monitored at pH 7.0 ( Fig. 2e, f). Absorbance spectra at each potential were obtained by subtracting the spectra obtained at 0.8 V, i.e., before the onset potential for water oxidation. In accordance with the results of previous studies, ligand-to-metal charge transfer (LMCT,~400 nm) and d-d transition (~600 nm) bands, which are characteristic features of high-spin Mn(IV)-oxo species, were detected in the spectra of Mn 3 O 4 NPs (Fig. 2e) 30 . Clearly, Ni-Mn 3 O 4 NPs/NiO catalyst exhibits a shifted peak at 420 nm (denoted as # in Fig. 2f) and an intense absorption change between 550 and 800 nm region (denoted as an arrow in Fig. 2f). The shifted peak at 420 nm is originated from new intermediate species from Ni-Mn 3 O 4 NPs layer. The calculation results, which will be discussed later, support that the origin of the red-shifted peak at 420 nm is a πtype LMCT band of the new intermediate species. It is turned out that the broad peak between 550 and 800 nm is due to the overlapping between the new intermediate's another LMCT band and the absorption of nickel oxyhydroxide formed by the oxidation of underlying NiO layer 32,33 .
Potential-dependent ex-situ electron paramagnetic resonance (EPR) measurements were conducted by freeze-quenching the samples to understand the electronic structure of the intermediates. The catalysts on the FTO substrates were collected immediately after electrolysis and subsequently quenched in liquid nitrogen. Dual-mode (perpendicular and parallel mode) continuous-wave EPR (CW-EPR) spectra of Ni-Mn 3 36 . We consider the possibility that the newly detected EPR signal (g~1.83) can be attributed to a low-spin Mn (IV)-oxo species in octahedral geometry (S = 1/2). Direct evidence for low-spin Mn(IV) is also obtained using potential-dependent superconducting quantum interference device (SQUID) analysis. Since the saturated magnetization value (M s ) is directly proportional to the total spin number, the spin Although only a few studies of low-spin d 3 systems have been reported, one can consider the pseudo Jahn-Teller (PJT) effect of the Mn(IV) species where spin-orbit coupling competes with the interactions that split the orbital degeneracy, thus lowering the molecular symmetry to C 4v symmetry 38 (Fig. 5). The gvalues can be described in terms of the ratio of the sum of the vibronic coupling energy at the equilibrium distortion (V vib ) and the solvent-dependent environmental energies affecting the orbital splitting (V L ) to the spin-orbit coupling constant (λ) 39 , through the fictitious angle 2θ: where k is the d-electron delocalization factor (0 (fully delocalized) ≤ k ≤ 1 (free metal ion)) [40][41][42][43] . As shown in Fig. 5a, Eqs. (3) and (4) 3 ] configuration represents the tetragonally elongated octahedral geometry (Fig. 5b). From the simulated values of g ∥ = 1.82 and g ⊥ = 1.84, k is estimated to be 0.23 and these g-values for the low-spin Mn (IV) species indicate the [b 2 e 1 ] configuration, which represents a tetragonally compressed octahedral geometry. In addition, the PJT splitting energy (E PJT ), which indicates the stabilization resulted from the breaking of the orbital degeneracy, can be estimated from the r value and g-values as E PJT = λ[1 + tan 2 2θ] 1/2 = 1000 cm -1 by using Eqs. (1)-(3) 40,42 . This PJT splitting is characteristic of the distorted C 4v symmetry of the Mn(IV) species, where the degeneracy of the d xz and d yz orbitals is broken 38 . Thus, the estimation of this d-orbital configuration provides valuable information about the local geometry and the electronic structure of the Mn(IV) center. We conceive that the atomic origin of the (slightly) broken degeneracy of the d xz and d yz orbitals is due to the disrupted rhombicity of the equatorial ligand field near the Mn(IV) center. Furthermore, potential-dependent in-situ Raman spectroscopy analysis was conducted to get direct evidence for oxo species (Fig. 6 and Supplementary Fig. 11). The sharp and intense Raman peak at 660 cm −1 , corresponding to the A 1g vibrational mode of spinel crystal structure 44 , is visible at the open circuit potential (Fig. 6a). Note that the Raman peak of soluble MnO 4 − is observed at 837 cm −1 before the onset potential (1.0 V vs. NHE) 45 . When the anodic potential is above the onset potential for water oxidation reaction (1.1 V vs. NHE), a broad Raman peak near 744 cm −1 is developed. This broad Raman peak reversibly disappears when the applied potential is below the onset potential, which indicates that the origin of this Raman peak is from the active intermediate species. From the previously studied molecular complexes, reported Raman peak positions for stretching vibrational mode of Mn(IV)-oxo species are between 707 and 803 cm −1 (refs. [46][47][48], while those of Mn(V)-oxo species are between 957 and 997 cm −1 (refs. [49][50][51]. Therefore, it is reasonable to conclude that the broad Raman peak is attributed to the stretching vibrational mode of Mn(IV)-oxo species.
To further confirm the nature of the Mn(IV)-oxo species, insitu Raman spectra are also obtained in the H 2 18 O and D 2 16 O electrolyte (Fig. 6b). The broad Raman peak at 744 cm −1 is redshifted to 716 cm −1 in the H 2 18 O electrolyte. The oxygen isotope shift of 28 cm −1 with 18 O substitution is well matched with the expected value of 33 cm −1 from a diatomic oscillator calculation of Mn(IV)-oxo species. Besides, the positions of Raman peaks remain unchanged in the D 2 16 O electrolyte compared to the spectra obtained in the H 2   DFT calculations. Using DFT, we attempted to elucidate the origin of the low-spin Mn(IV) species and identify the spectroscopic features of Mn(IV) species that can be compared with our experimental results. From periodic DFT calculations with a Hubbard-type correction (DFT+U) method, we optimized the structure of the Ni-Mn 3 O 4 NPs (001) and (231) surfaces that are representative low-and high-index surfaces found in our NPs ( Fig. 7a and Supplementary Fig. 12). On such facets, we generally found that Ni substitution into the Mn 3 O 4 substantially reduced the axial Mn-O distance (a compression of d Mn-O (axial)), and disrupted the rhombicity (an increase in Max-Min d Mn-O (equatorial)) of the truncated MnO 6 octahedron at the surface (Fig. 7b, c, Supplementary Fig. 13, and Supplementary Table 2). This finding suggests that the Ni substitution into the Mn-O network around the surface active site increases the axial crystal field and slightly break the degeneracy of the d xz and d yz energy levels ( Supplementary Fig. 14). Thus, the axial crystal field becomes stronger after Ni substitution due to the shortened d Mn-O (axial); therefore, the d z 2 and d xz/yz orbitals become destabilized, yielding an increased energy gap between d xz/yz and d xy . This increased gap provides an energetic driving force for the high-spin to low-spin transition, and yields an electronic configuration of [b 2 e 1 ] as shown in Fig. 5.
To mimic the ligand field effect on the Mn(IV) center imposed by the Mn 3 O 4 lattice, we built a cluster model with a Mn(IV) center coordinated by five OHligands, the oxygen positions of which were fixed at the same as in the periodic DFT-optimized structure of Ni-doped Mn 3 O 4 (inset of Fig. 7d and Supplementary  Fig. 15). Then, an additional oxygen atom was adsorbed to the Mn center to form the Mn(IV)=O bond. We note that the atomic positions of H and the adsorbate O were fully relaxed using a non-periodic DFT calculation including the spin-orbit effect. For comparison, we also performed a non-periodic DFT calculation for the high-spin Mn(IV)-oxo species (S = 3/2) using the same cluster model.
The non-periodic DFT calculation for the low-spin Mn(IV)oxo species resulted in a large and fairly isotropic value of hyperfine coupling constants |A| = [158, 143, 135] G, which is in good agreement with the experimental values of |A| = [121, 121, 132] G. Notably, the DFT calculation for the high-spin Mn(IV)oxo species yielded a much smaller value of the hyperfine tensor |A| = [44,43,49] G, which was intuitively anticipated for a highvalent metal center in a high-spin state. Thus, the experimentally observed large value of |A| supports the existence of a low-spin Mn(IV)-oxo species. Furthermore, the isotropic |A| values of the low-spin Mn(IV)-oxo species can be explained by the fairly isotropic shape of the spin density ( Supplementary Fig. 16), which originated from the electronic configuration of [b 2 e 1 ], where the singly occupied molecular orbital (SOMO) has d xz /d yz character.
Using the same cluster models, we further calculated the UV-Vis spectrum of Mn(IV)-oxo species using time-dependent DFT ( Fig. 7d and Supplementary Fig. 17). As shown in Fig. 7d, the π-type LMCT band for the high-spin Mn(IV)-oxo is found at 370.5 nm, while that for the low-spin Mn(IV)-oxo is found to be red-shifted to 375.2 nm, which agrees with our experimentally measured UV-Vis spectrum. In addition, the high-spin Mn(IV)oxo exhibits a d-d transition band at 534.5 nm, while the lowspin Mn(IV)-oxo exhibits another LMCT band at 554.1 nm due to the broken degeneracy of d xz and d yz orbital. Therefore, lowspin Mn(IV)-oxo shows a relatively intense absorption peak at 554.1 nm compared to high-spin Mn(IV)-oxo. Manganese oxide and nickel oxide NPs were synthesized by thermal decomposition and hot-injection methods. To prepare sub-10-nm monodispersed metal oxide NPs, separate reaction pots were used. One pot contained 1 mmol of M (ac) 3 (M = Mn and Ni) and 2 mmol of myristic acid dissolved in 20 mL of octadecene, and the other pot contained 3 mmol of decanol dissolved in 1 mL of octadecene. The two separate mixtures were degassed at 110°C for 1 h with vigorous stirring, and, then, the carboxylate mixture was heated above 295°C in an argon atmosphere. When the temperature reached 295°C, the decanol solution was injected rapidly into the carboxylate solution to induce burst nucleation. The reaction mixture was incubated at 295°C for 1 h followed by cooling to room temperature. A 1:1:1 ratio of the solution, acetone and toluene was mixed and centrifuged to remove organic residues from the solution. After repeating the purification step, the metal oxide NPs were re-dispersed in nonpolar solvents, such as hexane and cyclohexane. Materials characterization. The morphology of Ni-Mn 3 O 4 NPs/NiO was characterized using a high-resolution scanning electron microscope (Supra 55VP, Carl Zeiss, Germany). Pt coating was carried out using a Pt sputter coater (BAL-TEC/ SCD 005). Images were obtained using an acceleration voltage of 2 kV, and energydispersive X-ray (EDX) spectra were obtained at 15 kV. The sample positions coincided with the illuminated area. HR-TEM images and selected area electron diffraction (SAED) patterns were obtained using a high-resolution transmission electron microscope (JEM-2100F, JEOL, Japan) with an acceleration voltage of 300 kV. The TEM samples were collected from FTO glass and dispersed in ethanol by sonication for~1 min.
Electrochemical analysis. All electrochemical experiments were conducted using a three-electrode electrochemical cell system. A BASi Ag/AgCl/3 M NaCl reference electrode and Pt foil (2 cm × 2 cm × 0.1 mm, 99.997% purity, Alfa Aesar) were used as the reference and counter electrodes, respectively. Electrochemical tests were conducted at ambient temperature (21 ± 1°C) using a potentiostat system (CHI 760E, CH Instruments, Inc.). The electrode potential was converted to the NHE scale using the following equation: E(NHE) = E(Ag/AgCl) + 0.210 V. Additionally, overpotential values were calculated from the difference between the iR corrected potential (V = V applied -iR) and the thermodynamic point for water oxidation at a specified pH. The electrolyte was 500 mM phosphate buffer at pH 7. The working electrode was cycled to obtain a CV curve without pauses at a scan rate of 10 mV s -1 . Prior to conducting the electrochemical experiments, the solution resistance was measured in an electrolysis bath. All of the data were iR-compensated. The Tafel slope was calculated from the steady-state current density that is measured by bulk electrolysis with various applied potentials. The pH dependence was measured by chronopotentiometric analysis at a constant current density (1 mA cm -2 ). Then, the pH dependency on the current density was calculated with the following formula (Eq. (1)) combined with the Tafel slope results: According to the formula, we can determine the dependence of log(j) on pH.
EPR study. All EPR measurements were carried out at the Korea Basic Science Institute (KBSI) in Seoul, Korea. The simulations of the EPR spectra were carried out by EasySpin 52 . CW X-band (9.6 GHz) EPR spectra were collected using a Bruker EMX plus 6/1 spectrometer equipped with an Oxford Instruments ESR900 liquid-He cryostat using an Oxford ITC 503 temperature controller. Low temperatures were achieved and controlled using an Oxford Instruments ESR900 liquid-He quartz cryostat with an Oxford Instruments ITC 503 temperature and gas flow controller. The experimental conditions were as follows: microwave frequencies of 9.64 GHz (perpendicular mode) and 9.40 GHz (parallel mode), a modulation amplitude of 10 G, a modulation frequency of 100 kHz, microwave powers of 0.94 mW (perpendicular mode) and 5.0 mW (parallel mode), and a temperature of 5.7 K. Five scans were combined for each spectrum. The bulk electrolysis was conducted using a potentiostat (CHI 600D CH Instruments, Inc.) at pH 7 in 500 mM phosphate buffer solution. Set potentials were applied to each sample for 15 min. After carrying out the bulk electrolysis, the samples were rinsed gently with deionized water and rapidly transferred to an EPR tube using a blade. The EPR tubes were then immediately frozen and stored at 77 K in liquid nitrogen.
In-situ Raman study. All the in-situ Raman spectroscopy analyses were performed with a Raman microscope (LabRAM HR Evolution, Horiba) with a ×50 long working distance (LWD) visible objective lens. Raman spectra were collected with a 532-nm excitation laser source with a laser power of 1.6 mW. The individual Raman spectrum was obtained with an acquisition time of 300 s with 600 g mm −1 grating. To ensure the reliability of in-situ measurements, the Raman shift was calibrated using a silicon standard sample (520.6 cm −1 ) before the initial acquisition of Raman spectra. The in-situ measurements were implemented in a homemade electrochemical cell with a three-electrode configuration at room temperature. To eliminate the possible overlapping of the target Mn-oxo peak to the peaks of the phosphate electrolyte, here we utilized 1 M KHCO 3 electrolyte (pH 8.2) for the in-situ Raman measurement. The Raman spectrum at each applied potential was obtained via bulk electrolysis using a potentiostat (CHI 600E CH Instruments, Inc.) and the acquisition was initiated after the current was attained to steady-state.
Computational study. DFT calculations were carried out using the Vienna Ab Initio Software Package (VASP) 53,54 to understand the characteristics of the Mn 3 O 4 surface system with Ni substitution. All calculations were carried out using the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional 55 and the electron-ion interaction was considered in the form used by the projectoraugmented-wave (PAW) method 56 with a plane wave cutoff energy of 400 eV. To adjust the electron correlation contribution for Mn and Ni 3d electrons, a PBE+U scheme was adopted with an effective U value of 5.0 eV. The Mn 3 O 4 (001) and (231) surface models, with an additional 20 Å vacuum layer along the z-axis, were prepared using the optimized Mn 3 O 4 bulk structure. Gamma-centered k-point grids of (6 × 6 × 1) and (5 × 2 × 1) were employed for the (001) and (231) surfaces, respectively. A dipole correction was applied along the z-direction to properly compensate for the artificial dipole interactions across the periodic boundary. Nonperiodic DFT calculations for the cluster models were carried out using Orca 4.1 57 , to compute the EPR hyperfine coupling constant and the UV-Vis spectra. We used the Becke three-parameter functional (B3) combined with the correlation functional of Lee, Yang, and Parr (LYP) 58 , which is known as the proper choice for modeling spin crossover phenomena, and the ZORA-def2-TZVP SARC/J 59 basis set functions. Relativistic spin-orbit coupling was included using a zeroth-order regular approximation (ZORA) 60 for both the structural optimizations and the time-dependent DFT (TD-DFT) calculations. UV-Vis spectra were calculated using TD-DFT with the Tamm-Dancoff approximation 61 .