Electrocatalyzed Oxygen Reduction at Manganese Oxide Nanoarchitectures: From Electroanalytical Characterization to Device-Relevant Performance in Composite Electrodes

,

Nanostructured manganese oxides (MnOx) are effective electrocatalysts in alkaline media for the oxygen-reduction reaction (ORR), thus serving a critical function in the air-breathing positive electrodes of zinc-air batteries and alkaline fuel cells. [1][2][3][4] The ORR activity of MnOx depends on multiple factors including: specific crystalline structure; 5-8 particle morphology; [9][10][11] manganese valence as influenced by surface defects; [12][13][14] and high specific surface area, either of the oxide itself or by incorporating it with high surface area carbon. [15][16][17][18][19][20] A survey of the literature shows that nanocrystalline MnOx with 2 × 2 tunnel structures (i.e., the polymorphic cousins α-MnO 2 , cryptomelane, hollandite) is the most effective electrocatalyst for ORR. 1,2,4,8,9,11 We previously reported the synthesis of cryptomelane MnOx in the form of sol-gel-derived nanoarchitectures with differing pore volume: aerogel≈ambigel>xerogel. 21 These materials exhibit high specific surface area (>130 m 2 g -1 ) whether expressed as a mesoporous xerogel obtained by ambient-pressure drying or as a low-density aerogel processed using supercritical-CO 2 extraction. 21 While our prior studies focused on lithium-ion insertion properties, the morphological characteristics exhibited by the sol-gel-derived aerogel-like MnOx nanoarchitectures-high surface area and through-connected pore networks for facile mass transport-should also enhance their performance as prospective electrocatalysts, as we recently reported with nickeliron oxide nanoarchitectures for the oxygen-evolution reaction. 22,23 Herein, we assess the ORR activity of MnOx nanoarchitectures as measured across three electrochemical characterization platforms: rotating-disk electrodes (RDEs), powder-composite electrodes in airbreathing half-cells, and device-relevant zinc-air button cells. Fundamental ORR measurements at RDEs show state-of-the-art activity for MnOx electrocatalysts, with comparable onset potential and reaction order for both xerogel and aerogel. Yet, the performance of these two MnOx forms diverge once they are fabricated as powder-composite electrodes and tested in air-breathing configurations, with the MnOx aerogel yielding superior ORR activity. To probe the interior microstructure of these powder composites, we use focused ion beam (FIB) milling in conjuction with scanning electron microscopy (SEM) and elemental mapping by energy-dispersive spectroscopy (EDS). We show that the MnOx xerogel-based powder-composite exhibits a relatively heterogenous structure, whereas MnOx domains are uniformly distributed in the aerogel-based composite. The combination of intimately dispersed MnOx catalytic sites and more porous interior of the composite gives rise to the higher ORR performance observed with the aerogel-based composite.

Experimental
Materials synthesis and characterization.-Manganese oxide (MnOx) xerogels and aerogels were synthesized via established solgel chemistry based on the reaction of potassium permanganate (KMnO 4 ) and fumaric acid, as described previously. 21 Xerogels were prepared by drying water-filled MnOx gels at 50 • C under flowing nitrogen at ambient pressure. Supercritical CO 2 extraction of wet MnOx gels (after exchanging H 2 O for acetone in the pores) generates the corresponding aerogel form. The MnOx xerogels and aerogels were heated for 2 h at 300 • C under static air, ramping to temperature at 2 • C min -1 . Powder X-ray diffraction was performed using a Rigaku SmartLab X-ray diffractometer with a Cu Kα (λ = 1.5406 Å) radiation source, scanning from 20-70 • 2θ with a 0.02 • step size and an integration time of 1 s per step. Nitrogen-sorption measurements were made using a Micromeritics ASAP 2020 porosimeter; samples were degassed for 12 h at 80 • C under vacuum prior to analysis. Scanning electron micrographs were taken using a Carl Zeiss Leo Supra MM microscope operating at 5 keV by affixing the MnOx powders and MnOx catalyst-carbon composite powders onto aluminum stubs with double-sided carbon tape. Elemental mapping was performed using energy-dispersive spectroscopy with the same instrument operated at 20 keV. Elemental analysis of the surface of xerogel and aerogel powder-composites was assessed using an X-ray photoelectron spectrometer (Thermo Scientific K-Alpha X-ray) equipped with a monochromatic Al Kα source (1486.68 eV) and a 400-μm elliptical spot size. High-resolution spectra over the C1s, O1s, Mn2p, F1s, and Zr3d regions were obtained at an energy step-size of 0.1 eV. The instrument was operated with a low-energy electron flood gun and the resulting spectra were not peak-shifted prior to quantitative analysis. The spectra were analyzed with Avantage software version 5. 35.
The internal structure of the aerogel and xerogel powdercomposites were imaged and analyzed using an FEI Helios NanoLab Dualbeam fitted with an Oxford Instruments X-Max n 150 detector for energy-dispersive X-ray spectroscopy (EDS) measurements. Large-area focused-ion beam (FIB) milling into the sample was achieved with a Ga + ion beam operating at 30 kV using a beam current of 2.5 nA. The initial excavation sites were typically 30-μm wide by 25-μm long with a depth of 15-20 μm. Final 'cleaned' cross sections for SEM imaging and EDS analysis were prepared by FIB milling at lower beam current (0.23-0.46 nA) to minimize damage to the region of interest. At least four different locations were milled and imaged on each sample, with representative cross sections chosen for display in the Zinc-Air Full Cell Performance Evaluation section. Cross sections were imaged with the Dualbeam SEM operating at 20 kV using a beam current of 3.2 nA. Identical electron beam settings were used during collection of EDS elemental mapping data, which were analyzed using the Oxford Instruments Aztec 2.2 software. The EDS spectra were obtained in the range of 0-20 keV and equivalent X-ray collection settings were used for analysis of both aerogel and xerogel powder-composite electrode samples.
Rotating-disk electrode analysis.-Glassy-carbon rotating disk electrodes (RDE; 5-mm diameter, nominal area = 0.196 cm 2 , Pine Instruments) were mirror polished using 1.0-μm and then 0.05-μm alumina slurries followed by rinsing with and sonicating in 18 M cm water, with a final sonication in isopropanol. Catalyst-modified RDEs were prepared by suspending 5 mg of a ground mixture containing equal masses of MnOx xerogel or aerogel catalyst and Vulcan XC-72 carbon (Cabot) in 3 mL of a solution containing Nafion perfluorinated resin (5 wt% solution of Nafion in aliphatic alcohols and water; Aldrich), isopropanol, and water at a v:v ratio of 0.4 vol% Nafion, 20 vol% isopropanol, and 79.6 vol% water. This suspension was then sonicated for 1 h, stirred for 1 h, and sonicated for an additional 1 h; 10 μL of this suspension was subsequently deposited onto an inverted glassy-carbon RDE and dried under ambient conditions. 23 Current-potential trends for electrocatalytic O 2 reduction (ORR) at the MnOx catalysts were obtained under continuous bubbling of O 2 using linear-sweep voltammetry at 10 mV s -1 , a scan rate that is sufficiently slow to reflect steady-state conditions. 22,24 All potentials (applied and measured) were converted to overpotentials (η) relative to the thermodynamic potential using the equation: where 1.23 V vs. NHE is the potential for O 2 evolution at pH 7 and 0.138 V vs. NHE is the potential of the junction-isolated Hg/HgO reference electrode; the pH of the electrolyte (1 M KOH) was measured prior to each experiment using a Fisher Scientific AB15 pH meter. The counter electrode used for these measurements was a gold mesh. The potentials measured were compensated for iR loss based on resistance values determined from impedance at open-circuit potential.
Air-breathing half-cell electrode assessment.-Carboncomposite electrodes were prepared by first ballmilling an ethanol-dispersed mixture of MnOx xerogel or aerogel with XC-72 and PTFE polymer binder in a mass ratio of 20/65/15 for 16 h using a U.S. Stoneware Jar Mill with zirconia milling media at a variable speed of 200 rpm. A 1 2 " nickel mesh was then sandwiched between evenly distributed ball-milled carbon-composite powders (∼160 mg) and pressed to 1,750 psi for 1 min. The carbon-composite electrode was affixed to nickel foil perforated with a 0.26 cm 2 hole using carbon glue (DAG EB-020A) painted along the outside of the hole to adhere the composite electrode. The outside face of the electrode was then epoxied to a Viton O-ring to avoid any leaks during testing. To complete the assembly of the air-breathing half-cell, 25 another Viton O-ring was vacuum-greased on one side with the nickel flag aligned in the middle. The electrolyte reserve was filled with 6 M KOH and a platinum wire and zinc wire (Alfa Aesar) were used as the counter and quasi-reference electrodes, respectively. Chronoamperometric measurements were made from 1.425-1.295 V vs. Zn/Zn 2+ at (i) 5-mV intervals with a 10-min hold at each potential to obtain a steady-state current and at (ii) 25-mV intervals with a 25-min hold at each potential from 1.295-0.900 V vs. Zn/Zn 2+ .
Zinc-air full cell assembly.-The unmounted MnOx xerogel and aerogel carbon-composite electrodes prepared for the air-breathing half-cell tests were used directly as the cathode in prototype zincair cells. Monolithic zinc sponges were used as the anodes for the cells. 26,27 A nylon button-cap cell (Hillman Group #881032) was used as the casing for the zinc-air cell. For cell construction, a zinc sponge anode (∼1 cm 2 ; ∼0.1 g) was first placed on top of a tin foil (Alfa Aesar) current collector, the metal of choice for zinc-based batteries as opposed to copper, stainless steel or nickel, all of which accelerate gassing at the zinc-current collector interface. 28,29 A separator system composed of a Celgard 3501 microporous separator and a Freudenberg 700/28 K nonwoven separator pre-infiltrated with 6 M KOH were placed on top of the anode. The MnOx xerogel or aerogel carbon-composite cathode was placed on top of the separator system followed by a nickel-mesh current collector and a platinum wire for the positive terminal lead. Voltage-capacity trends were measured at constant-current densities spanning 5-125 mA cm -2 and discharging to capacities that sum to a total depth of discharge of 20% of the theoretical capacity of zinc (819.73 mAh g Zn -1 ), highlighting the stability of the ORR response in continuous-operation mode, as opposed to short-duration pulses that are sometimes used for reports on air-cathode performance.

Results and Discussion
Materials characterization.-The structural characteristics of the MnOx xerogels and aerogels were first determined prior to electrochemical testing. The MnOx xerogel and aerogel nanoarchitectures exhibit the cryptomelane 2 × 2 tunnel structure as verified by Xray diffraction (α-MnO 2 ; ICDD #00-044-1386; crystal structure is shown in the inset of Figure 1a). The broad X-ray reflections indicate nanoscale features; Scherrer analysis yields average crystallite sizes of ∼7 nm in both cases. Prior transmission electron microscopy of related cryptomelane-MnOx nanoarchitectures showed a nanorod-like morphology, ∼5 nm wide by 10-30 nm long. 21 When examined by field-emission scanning electron microscopy (SEM), xerogels present a more dense structure of networked MnOx nanoparticles (Figure 1b), whereas the corresponding aerogel has a highly textured, ultraporous structure ( Figure 1c). Nitrogen-sorption analysis (nitrogen-sorption porosimetry isotherms and pore size distributions of MnOx xerogel and aerogel can be found in Figures S1a,b) reveals that both calcined MnOx forms exhibit high specific surface area, though the aerogel is superior at 181 m 2 g -1 vs. 137 m 2 g -1 for the xerogel (Table I). Xerogel and aerogel forms are further differentiated by their total pore volume, which is 1.5 cm 3 g -1 for the low-density aerogel compared to 0.6 cm 3 g -1 for the xerogel.
Electroanalytical assessment at rotating-disk electrodes.-The fundamental oxygen-reduction activity of the MnOx xerogel and aerogel was assessed at RDEs (Figure 2a), a common method used to screen MnOx electrocatalysts. 1,2,4 Linear-sweep voltammograms of ink-cast carbon + MnOx films at glassy-carbon RDEs in O 2 -saturated 1 M KOH show the characteristic sigmoidal shape for O 2 reduction, with comparable onset potential of 0.9 V vs. RHE, which translates to an overpotential of η ≈ 310 mV (complementary plots referenced to Hg/HgO and overpotentials are shown in Figure S2). Previous studies on tunneled α-MnO 2 reported similar values of onset potential and limiting current density. 1,2,4,9 The effectiveness of manganese oxides as alkaline ORR catalysts is further confirmed here, with onset potentials that are only ∼70 mV more negative than for carbon-supported Pt nanoparticles. 30 These RDEs were subjected to a range of rotation rates (ω), with the resulting trends in current density values (J) used to calculate the number of electrons passed (n e ), according to the Koutecky-Levich theory: 31 where J K is the kinetics-limited current density (J K = n e FkC O ), and J L is the diffusion-limited current density. The B term is further defined as: where F is the Faraday constant (96,485 C mol -1 ), C O is the bulk concentration of O 2 (7.8 × 10 -7 mol cm -3 ), D O is the diffusion coefficient of O 2 (1.8 × 10 -5 cm 2 s -1 ), and ν is the kinematic viscosity  Air-breathing half-cell assessment.-Rotating-disk electrode measurements assess the intrinsic activity of prospective electrocatalysts, but the activity and η observed with RDEs may not always translate to the performance trends exhibited by the practical air-breathing electrode structures used in metal-air batteries and fuel cells. 34 Such electrodes typically comprise complex physical mixtures of electrocatalyst, conductive carbon, and polymer binder, in which factors such as porosity, hydrophobicity, electronic conductivity, and dispersion of catalyst determine overall performance. Prior to evaluation in full-cell devices, air-breathing composite electrodes can and should be tested in an air-breathing half-cell configuration that includes independent reference and counter electrodes, such that the working electrode performance is isolated and not limited by an opposing electrode. 6,25,35,36 In order to fabricate practical composite electrode structures for this study, MnOx xerogel and aerogel powders were first ball-milled with Vulcan carbon powder and PTFE binder; a Ni-mesh current collector was then sandwiched between the MnOx/carbon/PTFE mixtures by pressing at 1750 psi. The resulting electrodes were interfaced with an air-breathing electroanalytical cell (Figure 3a), 25 with one side contacting a battery-relevant electrolyte (aqueous 6 M KOH) and the other side exposed to static air. Electrocatalytic activity for ORR was assessed with potential-step chronoamperometry within a potential window of 1.4 to 0.9 V vs. Zn/Zn 2+ , a range typical of a zinc-air cell. Potential-dependent steady-state current density trends obtained with the MnOx xerogel and aerogel-containing composite electrodes are shown in Figure 4b from 1.4 to 1.1 V. The ORR performance of MnOx xerogel and aerogel immediately begins to diverge as the potential is stepped down from 1.40 V to 1.35 V and below. For example, at 1.35 V the aerogel-based electrode provides a current density of -20 mA cm -2 , while the xerogel-based analog yields only -7 mA cm -2 . To match the -20 mA cm -2 activity of the aerogel at 1.35 V, the xerogel must operate at 1.28 V, a 70-mV penalty in overpotential (Table II). The MnOx xerogel-based composite still provides enhanced performance relative to a MnOxfree carbon-based electrode, which only reaches -20 mA cm -2 at 1.18 V. 25 The aerogel-based electrode maintains superior performance before ultimately converging with that of the xerogel at ∼1.0 V (Figure S4), where concentration polarization arises because of long-range transport limitations.
The performance of the MnOx xerogel-based electrode is consistent with that previously reported for powder composites incorporating nanoscale α-MnO 2 as the active catalyst, for example the work of Cao et al. 6 The enhanced performance of the aerogel-based air cathode at Table II   moderate overpotentials may arise from improved short-range transport of O 2 to catalytically active sites at the oxide surface, facilitated by the aerogel's high pore volume 37 and hierarchical dispersity of pore size.
Zinc-air full cell performance evaluation.-Two-terminal zincair full cells were assembled to validate the distinctions observed between the MnOx xerogel and aerogel carbon-composite electrodes in the air-breathing half-cell, but now, under technologically relevant, limited-electrolyte conditions. Zinc sponge electrodes, recently developed in our laboratory, were chosen for full-cell demonstrations, because the interconnecting, co-continuous networks of solid metal and electrolyte-filled void in the sponge confer higher capacity and rates relative to commercial, powder-bed counterparts; 26,27,38 thus, the sponge electrode form-factor is less likely to limit performance at the opposing air cathode. Galvanostatic discharge measurements were applied to the resulting button cells, with current densities ranging from 5-125 mA cm -2 to monitor the ability of the zinc-air full cell to sustain steady-state discharge voltages at these moderate to challenging rates.
At moderate rates (5-50 mA cm -2 ), the MnOx aerogel carboncomposite electrode confers higher discharge voltages than its xerogel counterpart (by ∼60 mV; Figure 4). At more challenging rates (>50 mA cm -2 ), the separation in discharge voltage expands to ∼100 mV (Table II). By 125 mA cm -2 , the discharge voltage of the xerogel-based cell falls below the conventional lower voltage limit used for zinc-air cells (0.90 V), 39 whereas the aerogel analog maintains voltage at 0.93 V, within practical operating conditions. The maximum specific power observed for the MnOx aerogelbased zinc-air cell, normalized to the mass of the zinc electrode, is 1,250 W kg Zn -1 . Projecting this value to a fully packaged cell where zinc commonly occupies 46% of the battery 40 translates to a specific power of 580 W kg -1 and a corresponding area-normalized power of 112 mW cm -2 . The discharge voltage and area-normalized power with xerogel-based electrodes are commensurate with other device-level zinc-air cells using MnOx-type catalysts, [41][42][43][44] while MnOx aerogelbased air cathodes boost the ORR performance beyond typical.
With aerogel-based composite electrodes providing clear performance advantages, we turned to SEM and EDS to characterize the nano-and micro-scale structure of the composite electrodes. In both cases, micrographs show a well-dispersed matrix containing all three components-nanofibers of PTFE interspersed among agglomerates of ∼50-nm carbon particles and nanostructured MnOx (Figures 5a,  5b)-with no obvious micrographic distinctions between aerogel-and xerogel-based powder-composites. We used EDS in elemental mapping mode to examine for Mn-and O-containing domains that would be associated with the MnOx phase (Figures 5c, 5d). When the xerogel powder-composite is imaged in multiple locations on the composite surface ( Figures S5 and S6 for xerogel and aerogel, respectively), the Mn and O EDS maps show that localized MnOx domains, ranging in size from ∼0.5-1.5 μm, are inhomogeneously distributed within the electrode matrix (Figure 5c). Few such MnOx-rich features are observed in the elemental map of the aerogel-based powder-composite (Figures 5d), which exhibits more homogeneous Mn and O distribution at this size scale.
To probe the internal structure of these powder composites, cross sections of both xerogel and aerogel powder-composite electrodes were prepared by Ga + ion-beam milling and imaged by SEM (Figures 6a, 6b). The bright spots in the moderate-magnification SEM indicate a few zirconia particles introduced during the ballmilling process; XPS analysis shows that zirconium content is less than 1% (see Supporting Information). At finer magnification, the aerogelbased composite shows markedly higher void space compared to the xerogel (Figures 6c, 6d), which correlates with the results of the nitrogen-sorption porosimetry data presented in Table I for the MnOx aerogel itself. Elemental mapping of the cross-sections yields similar results to those obtained at the outer surface, with the xerogel-based composite showing evidence of some imhomogenieties (regions with high Mn and O concentration). On the basis of the SEM/EDS results, we posit that the low density and friable nature of the MnOx aerogel yields a more uniform dispersion within the carbon/PTFE matrix during the ballmilling process, which leads to enhanced ORR activity in a technologically relevant electrode structure.

Conclusions
The results reported herein highlight the importance of assessing the translation of electrocatalytic activity from fundamental measurements to technologically relevant electrode structures. Fundamental studies derived from rotating-disk electrodes yield key information such as electrocatalytic mechanisms, yet do not provide insight into overall catalytic activy when expressed in a microhetergeneous structure that includes conductive carbon and polymer binder. Configurations that test powder-composite electrodes in an air-breathing mode show clear distinctions in ORR activity for xerogel versus aerogel-based compositions, and the performance evaluation is a better predictor of real full cell performance than the fundamental RDE study. The performance trends reported in this study-lower overpotentials and higher discharge voltages for MnOx aerogel-based composite electrodes-presents an exciting direction for establishing devicerelevant metrics of ORR catalyst materials.