Activating Mn Sites by Ni Replacement in α-MnO2

Transition metal oxides are characterized by an acute structure and composition dependent electrocatalytic activity toward the oxygen evolution (OER) and oxygen reduction (ORR) reactions. For instance, Mn containing oxides are among the most active ORR catalysts, while Ni based compounds tend to show high activity toward the OER in alkaline solutions. In this study, we show that incorporation of Ni into α-MnO2, by adding Ni precursor into the Mn-containing hydrothermal solution, can generate distinctive sites with different electronic configurations and contrasting electrocatalytic activity. The structure and composition of the Ni modified hollandite α-MnO2 phase were investigated by X-ray absorption spectroscopy (XAS), X-ray diffraction (XRD), transmission electron microscopy coupled to energy-dispersive X-ray spectroscopy (TEM-EDX), inductively coupled plasma–optical emission spectroscopy (ICP-OES), and X-ray photoelectron spectroscopy (XPS). Our analysis suggests that Mn replacement by Ni into the α-MnO2 lattice (site A) occurs up to approximately 5% of the total Mn content, while further increasing Ni content promotes the nucleation of separate Ni phases (site B). XAS and XRD show that the introduction of sites A and B have a negligible effect on the overall Mn oxidation state and bonding characteristics, while very subtle changes in the XPS spectra appear to suggest changes in the electronic configuration upon Ni incorporation into the α-MnO2 lattice. On the other hand, changes in the electronic structure promoted by site A have a significant impact in the pseudocapacitive responses obtained by cyclic voltammetry in KOH solution at pH 13, revealing the appearance of Mn 3d orbitals at the energy (potential) range relevant to the ORR. The evolution of Mn 3d upon Ni replacement significantly increases the catalytic activity of α-MnO2 toward the ORR. Interestingly, the formation of segregated Ni phases (site B) leads to a decrease in the ORR activity while increasing the OER rate.


■ INTRODUCTION
The development of highly efficient oxygen electrocatalysts employing Earth-abundant elements is one of the key challenges in the development of scalable electrochemical energy conversion systems such as water electrolyzers, fuel cells, and metal air batteries. 1,2−15 Beyond aspects associated with binding energies of reactants and intermediates species, the whole electronic structure of these materials is also acutely sensitive to structure and composition, adding a significant level of complexity to the case of metallic phases. 11−22 It has been reported that electrocatalytic activity toward the oxygen reduction reaction (ORR) for the various MnO 2 phases follows the trend α-> β-> γ-MnO 2 . 13Other studies have shown that promoting oxygen vacancies in β-MnO 2 can significantly increase their electrocatalytic performance. 23−29 The work by Selvakumar et al. proposed that deposition of Co and Ni onto α-MnO 2 can lead to Mn sites with lower oxidation states, i.e., Mn(III) states, which are linked to a higher ORR activity. 30−33 However, studies by Celorrio et al. on LaMnO 3 , SrMnO 3 , CaMnO 3 , and YMnO 3 clearly demon-strated that there is no direct correlation between Mn oxidation state and ORR activity. 34n this work, we show for the first time that Ni incorporation into to the α-MnO 2 lattice leads to activation of Mn sites (site A) toward the ORR, as opposed to Ni sites nucleated at the surface of the oxide (site B).Systematic structural and composition analysis based on X-ray absorption spectroscopy (XAS), X-ray diffraction (XRD), transmission electron microscopy coupled to energy-dispersive X-ray spectroscopy (TEM-EDX), inductively coupled plasma−optical emission spectroscopy (ICP-OES), and X-ray photoelectron spectroscopy (XPS) shows little differentiation between both types of sites in α-MnO 2 .On the other hand, clear contrasts can be seen in the pseudocapacitive electrochemical fingerprints measured in alkaline solutions.Site A, primarily formed under a low Ni content, promotes Mn 3d orbitals at energies (potentials) relevant to the ORR, leading to an increase in ORR kinetics.By contrast, site B is dominated by Ni 3d orbitals, which catalyzes the OER kinetics.

■ RESULTS AND DISCUSSION
High aspect ratio α-MnO 2 nanostructures were prepared by an established hydrothermal method (details of experimental and synthesis methods are included in the Methods section), 30 while Ni modified α-MnO 2 nanostructures were prepared by adjusting the ratio of Ni:Mn introducing Ni(NO 3 ) 2 •6H 2 O in the precursor solution.It is important to differentiate the Ni content in the precursor solution (experimental variable) denoted as Ni pre , from the mean Ni content (Ni tot ) estimated by techniques such as ICP-OES and EDX, and from the surface Ni ratio (Ni surf ) obtained from XPS.The Ni content in all cases is defined as the molar ratio of Ni/(Mn + Ni).We will be using this notation throughout the whole paper.The precursor solution was heated in a Teflon-lined stainless-steel autoclave at 140 °C for 12 h.The product was washed with distilled water, filtered, and then dried in the air.
XRD patterns of the oxide with different Ni pre content varying from 0 (α-MnO 2 ) to 25% are displayed in Figure 1a.Materials prepared with up to 15% Ni pre show clear diffraction peaks corresponding to (110), ( 200), ( 220), (310), (400), (211), (420), (301), and (411) of hollandite α-MnO 2 (JCPDS file 00-44-0141) and a space group of 14/m.The significant broadening of the XRD features in 25% Ni pre suggests that the crystalline domain size of the oxide is decreased with respect to the lower Ni content.Figure 1b reveals interesting correlations between the Ni pre composition and the total Ni content measured in the nanostructures (Ni tot ) and at the surface (Ni surf ) as probed by ICP-OES and XPS.Ni tot changes only between 4.5% and 6% upon increasing Ni pre up to 15%.XPS composition analysis shows a similar trend, although the Ni surf values are slightly higher than Ni tot values.In principle, this could be rationalized as Ni concentration grading from the bulk toward the surface.However, α-MnO 2 is characterized by its high aspect ratio, as shown in Figure 1c; thus, the interplay of surface density of Mn sites and XPS penetration depth is far from trivial.Increasing Ni pre above 15% leads to a jump in Ni content in both sets of measurements.
Figure 1c show characteristic transmission electron microscopy images of the 15% Ni pre , illustrating the high aspect ratio characteristic of α-MnO 2 as well as lattice fringes with 0.44 nm d-spacing associated with the (200) planes.Figure S1 of the Supporting Information show additional TEM images of materials obtained across the Ni pre content investigated.All materials exhibit the characteristic high aspect ratio of the hollandite phase along with the lattice fringes associated with the (002) planes, which is also consistent with the phase purity of the material observed by XRD up to 15% Ni pre .In the case of 25% Ni pre , Figure S1 shows that the dimensions of the α-MnO 2 crystals are reduced along with the appearance of particles with different morphology.TEM-EDX images in Figure S2 show that the featureless particles are primarily composed of Ni.The apparent decrease in α-MnO 2 crystal size in Ni pre 25% is consistent with the broadening of the XRD features observed in Figure 1a, while the absence of clear diffraction features associated with Ni oxide phases suggests that the segregated material is amorphous.On the other hand, Figure S2 also shows that Ni is primarily found along the α-MnO 2 lattice up to Ni pre 15%.Tables S1 and S2 show that the   compositions extracted from EDX and ICP-OES are consistent, including the content of K + ions which are often linked to charge stabilization in the inside the [2 × 2] channel. 35-ray photoemission spectra (XPS) of Mn 2p and Ni 2p orbitals are shown for the different formulated electrocatalysts, shown in Figure 2a and b, respectively.Characteristic survey spectra of 5% and 25% Ni pre are shown in Figure S3.The Mn 2p spectra exhibit a broad emission at 642.6 eV corresponding to Mn 2p 3/2 , and even broader and asymmetric emission at 654.3 eV which is linked to Mn 2p 1/2 .The broadening of the latter originates from the strong overlap of Mn 3+ (641.9 eV) and Mn 4+ (642.2 eV) sites at the oxide surface, 36−38 which makes these contributions difficult to be accurately deconvoluted.However, it could be argued that the Mn 2p 1/2 photoemission peak is slightly shifted toward lower binding energies (BEs) in the case of Ni pre 5 and 10%, in comparison to Ni pre 0 and 25%.On the other hand, there is a more systematic shift in the maximum of the Ni 2p 3/2 photoemission line as the Ni content increases as shown in Figure 2b.These observations may suggest changes in the surface electronic configuration of Mn sites at a low Ni content, which effectively disappear once Ni nucleates separately from the α-MnO 2 lattice.
Figure 2c and d show the Mn-edge of the X-ray absorption spectra (XAS) and the FT of the k 3 -weighted EXAFS spectra for the various materials, respectively.The position of the Mn K-edge does not show any systematic change with Ni content, indicating that there is no notable change in oxidation state on Mn. Figure 2d shows the comparison between the Fourier transforms (FTs) of the EXAFS data for the different compositions, in comparison with spectra of α-MnO 2 (simulated from reference data for COD ID 1514116 downloaded from the Crystallography Open Database).It can be observed that the data collected for the synthesized samples strongly resemble that of the theoretical α-MnO 2 structure.The amplitude reduction factor (S 0 2 ), bond length, Debye−Waller factor (σ 2 ), and energy shift parameter (ΔE 0 ) were refined.The best-fit parameters are summarized in Table S3, and data and fits are provided in Figure S4.The data confirm that there are no extended structural changes in the Mn coordination even at high Ni content, which is expected given their similar ionic radii.However, as demonstrated below, Ni replacement does introduce discrete changes in the structural/electronic configuration which cannot be detected in XAS spectra.
Figure 3a contrasts cyclic voltammograms of α-MnO 2 , obtained with various Ni pre content, supported on a mesoporous carbon layer with a 398 μg cm −2 oxide loading, 50 μg cm −2 Vulcan, and 50 μg cm −2 Nafion (see Methods section) in argon-saturated 0.1 M KOH solutions at 10 mV s −1 .α-MnO 2 is characterized by a broad pseudocapacitive reduction peak centered at potentials close to 0.6 V vs RHE, and a sharper oxidation peak in the reversed scan at 1.0 V vs RHE.−42 We correlate these signals with the density and distribution of Mn 3d states across the potential (energy) region relevant to oxygen electrocatalysis. 6,33Interestingly, 5 and 10% Ni pre are characterized by strong pseudocapacitive responses, leading to a broad cathodic current with a peak centered at 0.4 V and an anodic peak at about 1.3 V in the reverse scan.Smaller features can be observed between 1 and 1.2 V in the forward (negative) scan, but the main observation is the substantial increase in the density of Mn 3d states upon introducing Ni into the lattice.This observation reveals the emergence of a new local electronic configuration, which we will refer to as site A, in which Mn displays a higher local density of 3d states at a potential relevant to the ORR.A dampening of these responses is observed in 15% and 25% Ni pre , along with the emergence of a different redox transition between 1.2 and 1.4 V vs RHE.This new redox transition is associated with changes in the oxidation state of Ni, 30,34,35,40,43,44 which coincides with the onset of Ni phase segregation as discussed previously.These observations suggest that Ni segregation generates a different site (Site B) which has a negligible effect on the electronic properties α-MnO 2 .
Figure 3b illustrates the ORR current responses on a rotating-ring disk electrode (RRDE) at 1600 rpm in an O 2staurated 0.1 M KOH solution, with the Pt ring potential fixed at 1.2 V vs RHE (generation-collection mode).The results clearly show that the introduction of Ni decreases the onset potential for the ORR and increases the overall current with respect to α-MnO 2 .The Pt ring current decreases as Ni pre increases to 15%, which is associated with the rate of HO 2 − generated at the disk electrode.Interestingly, the rate of HO 2 − generation increases in the case of 25% Ni pre .Figure 3c contrasts linear sweep voltammetry (LSV) in the OER potential range, showing a systematic shift of the onset potential of the Ni in the Ni content increases.The IR drop in our electrode configuration is independent of the catalyst composition with an average of 10 ± 2 Ω. Figure S6a shows that IR compensation has a minor effect on the current− potential dependence.Tafel plots (Figure S6b) also show that the effective transfer coefficient is little dependent on the Ni content.Stability studies under OER conditions, as those reported in other works, 45,46 are outside of the scope of this work, which primarily focus on the contrasting nature of the Ni generated active sites.Potential induced surface reconstruction is another important aspect that is not explicitly considered in our analysis, which would require carefully conducted experiments at single crystal electrodes.
Figure 3d shows the dependence of the kinetically limiting current (i k ) at 0.70 V vs RHE, as estimated from Koutecky− Levich analysis, on the total Ni molar ratio in the nanostructures (Ni tot ).Details of this analysis are shown in the Methods section and are exemplified in Figure S5.We can see a sharp increase in the i k value, indicating an increase in ORR activity up to Ni tot of 4.5%, followed by a decrease at higher content.Analysis of the effective number of electrons in the ORR process (Table S4) based on the generationcollection RRDE configuration (see Methods) shows that α-MnO 2 primarily reduces oxygen through the two-electron process, generating HO 2 − under alkaline conditions.Introducing 5−10% Ni to the precursor solution, which leads to primarily site A, swiftly changes the mechanism to the fourelectron process.Further increasing the Ni content in the precursor, promoting site B instead, switches the reaction pathway back to the two-electron process.This behavior indicates that the ORR kinetics (i k ) and pathways are dictated by the emergence of site A. On the other hand, the OER current measured at 1.85 V vs RHE exhibits a monotonic increase with increasing Ni content.
Our analysis provides unambiguous evidence that Ni insertion into the α-MnO 2 lattice (site A) leads to a distortion of the electronic structure which manifests itself by an increase of Mn 3d states at potentials relevant to the ORR.This can be clearly seen by the emergence of the pseudocapacitive responses for Ni tot in the range of 5%.Although we have not been able to determine the changes in the electronic configuration associated with site A, we do not see this as just changes in the local oxidation state.As demonstrated in our previous works, Mn oxides with a nominal +3 oxidation state such us YMnO 3 are rather inactive, 34 while compounds such CaMnO 3 (Mn 4+ ) are active toward the ORR. 47ttempting to increase the Ni content by increasing Ni pre leads to the formation of surface segregated Ni phases (site B) which are less active toward the ORR.It is interesting to contrast these observations with our previous analysis of LaMn x Ni 1−x O 3 , 6 where we see a systematic decrease in ORR activity and increase in OER activity with increasing Ni content.In the case of the mixed perovskite, replacing Ni by Mn leads to minimal structural changes, while electronic interactions between Ni and Mn are also relatively weak.

■ CONCLUSIONS
This study reveals that insertion of Ni into α-MnO 2 can lead to an increase in the activity toward the ORR only if the electronic structure of the oxide is distorted such that Mn 3d states are generated at the relevant potential (energy) scale.By introducing Ni into the hydrothermal precursor solution, we were able to incorporate Ni up to approximately 5% of the total cation content in the oxide lattice.Further increase of the Ni precursors leads to the nucleation of Ni phases at the surface of α-MnO 2 , which are less active toward the ORR.The emergence of active Mn sites (site A), promoted by the insertion Ni, is observed by the evolution of pseudocapacitive responses between 0.2 and 1.2 V vs RHE.On the other hand, the OER kinetics monotonically increase with increasing overall Ni content in the catalysts.This is an important observation, demonstrating that ORR is extremely sensitive to structure and coordination of Mn sites, while OER is primarily determined by number density of Ni sites.Further studies are required to identify the exact nature of the ORR active sites promoted by Ni replacement, which could not be clearly identified from XAS, XPS, or TEM analysis.In any case, these studies further confirm the presence of pseudocapacitive responses associated with d-states as an effective descriptor of electrocatalytic activity in these complex materials.In more molecular terms, the oxidation state is not the determining factor as such but rather changes in oxidation states which arises from populating/depopulating d-states in the potential range relevant for catalysis.

Materials Synthesis
High purity KMnO 4 , MnCl 2 •4H 2 O, Ni(NO 3 ) 2 •6H 2 O, Vulcan XC 72R, and Nafion were purchased from Sigma-Aldrich.α-MnO 2 was synthesized by a hydrothermal method following reports published elsewhere. 19,23Amounts of 0.5 g of KMnO 4 and 0.2 g of MnCl 2 •4H 2 O were added to 15 mL of Milli-Q water under magnetic stirring for 10 min.The solution was transferred to a Teflon-lined stainless-steel autoclave, which was sealed and heated at 140 C for 12 h.The Teflonlined stainless-steel autoclave was naturally cooled down to the room temperature, and then the product was centrifuged three times with Milli-Q water.After that, the precipitate was dried in a vacuum oven at 80 C for 12 h.Ni-modified α-MnO 2 was synthesized through the same hydrothermal method, adjusting the Ni/(Ni + Mn) molar ratio in the precursor (Ni pre ) by adding Ni(NO 3 ) 2 •6H 2 O.

Electrocatalyst Catalyst Layer
The first step in the preparation of the electrocatalyst ink involved mixing 16 mg of Vulcan XC 72R, 1 mL of Nafion solution as a perfluorinated resin binder, and 15 mL of deionized water.The mixture was dispersed in an ultrasound bath for 40 min.Then 2.5 mg of the oxide electrocatalyst was mixed with 500 μL of the ink and sonicated for 30 min to ensure a homogeneous suspension.Prior to the deposition of the catalysts, the rotating ring-disk electrode (RRDE) was polished with alumina powder and Milli-Q water.Immediately after sonication of the precursor solution, 10 μL of the ink was carefully drop casted onto the freshly polished glassy carbon disk, minimizing contamination of the Pt ring.The catalyst layer was left to dry for 5 min with a hot air gun.The catalyst loading at the electrode was 398 μg cm −2 .

Material Characterization and Electrochemical Studies
X-ray diffraction (XRD) patterns were recorded utilizing a Bruker AXS D8 Advance diffractometer with a θ−θ configuration, utilizing Cu Kα radiation (λ = 0.154 nm).The diffraction patterns were recorded at 25 °C with a step size of 0.02°and a time per step of 2 s over an angular range of 10−55°.A JEOL JEM 2010 transmission electron microscope fitted with a Gatan Orius digital camera was used in these studies, coupled to an energy-dispersive X-ray spectroscopy (EDX) detector.Samples were prepared for TEM analysis by dispersing the powder in ethanol utilizing ultrasonication and pipetting 1000 μL drops of ethanol dispersed onto 3 mm diameter copper grids.
Inductively coupled plasma atomic emission spectroscopy (ICP-OES) was conducted on an Agilent 710 simultaneous spectrometer.The sample preparation was 5 mg of each sample dissolved in 1 cm −3 nitric acid (1 wt %) which was diluted to reach a final volume of 10 cm −3 by adding Milli-Q water.The standard solutions contain Mn, K, and Ni, respectively, all dissolved in a 1 wt % nitric acid (supra pure).The wavelengths used for quantification of Mn were 257.610, 259.372, 260.568, 294.921 nm, while 216.555, 221.648, 230.299, and 231.604 nm lines were used for Ni quantification.
XAS spectra were acquired at the B18 beamline of the Diamond Light Source, featuring a Canberra 35-element monolithic planar Ge pixel array detector in fluorescence mode at the Mn K-edge (6539 eV).The samples were made into pellets (1.32 cm 2 pellet area) by combining the ground sample with cellulose (80 mg) to obtain a homogeneous mixture, which was then compressed (5 tonnes) by using pellet press.The Mn K-edge range was established by using Mn foil.Data analysis was performed with Athena and Artemis software.
X-ray photoelectron spectroscopy (XPS) analysis was conducted in a NanoESCA II instrument at ambient temperature and under ultrahigh vacuum (4 × 10 −11 mbar).The photoemission spectra of C 1s, O 1s, Mn 2p, and Ni 2p were recorded at room temperature.Spectra were recorded using a step size of 0.1 eV, a collection time of 0.5 s, and a pass energy of 20 eV.Binding energies (BEs) were calibrated using the C 1s peak (284.6 eV) as reference.Surface composition of the materials was measured from high-resolution Ni 2p and Mn 2p spectra.
Electrochemical measurements were conducted in a three-electrode cell by using a rotating ring-disk electrode (RRDE) attached to an ALS rotation control system and linked to a CompactStat bipotentiostat (Ivium).The RRDE electrode was made up of a catalyst casted onto a glassy carbon disk 4 mm diameter (0.126 cm 2 surface area) surrounded by a Pt ring 7 mm inner diameter, both of which served as the working electrode.Hg/HgO was utilized as the reference electrode, and a carbon rod was used as the counter electrode.In this work, all of the potentials have been converted in reference to an RHE scale.Pseudocapacitive responses were measured by cyclic voltammetry (CV) in argon-saturated 0.1 M KOH electrolyte at 10 mV scan rates.ORR kinetics were obtained from linear sweep voltammetry studies at 10 mV s −1 in O 2 saturated 0.1 M KOH solution employing RRDE electrodes.The effective number of electrons transferred (n) can be calculated from the currents at the disk (i D ) and ring (i R ) electrodes based on = + ( ) where N c is the collection efficiency.The Pt ring electrode is held constant at 1.2 V vs RHE while the disk electrode is swept across the potential range.The kinetically limited current (i k ) for the ORR was calculated from the Koutecky−Levich expression where c, D, ω, and ν correspond the bulk oxygen concentration (1.2 × 10 −6 mol cm −3 ), the oxygen diffusion coefficient (1.9 × 10 −5 cm 2 s −1 ), the angular rate, and the kinematic viscosity of water (0.01 cm 2 s −1 ), respectively.

Figure 1 .
Figure 1.Structure and composition analysis of α-MnO 2 with various Ni content: (a) XRD patterns of the materials obtained from precursors containing various Ni/(Ni + Mn) molar ratios (Ni pre ).(b) Correlations between Ni pre and the Ni−Mn molar ratio of the nanostructures estimated by ICP-OES (Ni tot ) and the surface molar ration (Ni surf ) obtained by XPS analysis.(c) Characteristic TEM images of 15% Ni pre illustrating the high aspect ratio of the hollandite phase and lattice fringes, with 0.44 ± 0.2 nm d-spacing, associated with the (002) planes; scale bar corresponds to 1 μm.(d) TEM-EDX images showing distribution of Ni, Mn, and O across the 15% Nipre sample.

Figure 2 .
Figure 2. Surface composition and bonding analysis: (a) XPS spectra of Mn 2p showing that the oxidation state of Mn appears to be affected little by the introduction of Ni.(b) XPS spectra of Ni 2p, in which the Ni 2p 3/2 slightly shifts to higher binding energies as the Ni pre content increases.(c) Normalized XAS spectra contrasting the edge of Mn(II) and Mn (IV) standard with those of α-MnO 2 obtained with different Ni pre .(d) FT of the k 3 -weighted EXAFS spectra showing the first and second coordination shell for the various Ni pre samples.

Figure 3 .
Figure 3. Electrochemical responses and electrocatalytic activity: (a) Pseudocapacitance responses of Ni doped MnO 2 catalyst layers at a glassy carbon (GC) electrode in Ar-saturated 0.1 M KOH at scan rate 10 mV s −1 .(b) Rotating ring-disk electrode measurements at 1600 rpm for the various catalyst compositions in O 2 -saturated 0.1 M KOH with a sweep rate of 10 mV s −1 , with the Pt ring electrode held constant at 1.2 vs RHE.(c) Linear sweep voltammetry of α-MnO 2 obtained with different Ni pre content in 0.1 M KOH with a scan rate of 10 mV s −1 and 1600 rpm.(d) Kinetically limited current (i k ) for the ORR at 0.70 V vs RHE and current density (normalized by geometric area) for the OER at 1.85 V vs RHE as a function of the actual Ni content in the nanostructure (Ni tot ).