The exploitation of new generation of sustainable, clean and efficient energy conversion and storage devices, such as metal-air batteries and alkaline fuel cells, has attracted significant attention in recent years [1-4]. However, due to a complex multistep four-electron process, the kinetically sluggish oxygen reduction reaction (ORR) at the cathode is an obstacle for the effective using of such electrochemical energy devices [4-6]. To date, the most efficient ORR electrocatalysts are Pt-based catalysts, but their high cost, scarce reserves, poor long-term stability and low methanol tolerance greatly limit their large-scale commercial application [4-9]. Accordingly, many researchers have devoted to searching for highly efficient and stable non-noble metal-based ORR electrocatalysts [4, 9-14].

Mixed valence oxides of transition metals, especially spinel-type oxides M₃O₄ (M = transition metal) consisting of multivalent metal cations occupying tetrahedral (four-coordination) and octahedral (six-coordination) sites, have emerged as an important and highly potential class of electrocatalytic material owing to their excellent activities, long-term stability and abundant reserves [15-20]. Nevertheless, the ORR performance of single metal spinel-type oxides still needs to be improved compared with that of Pt-based electrocatalysts. Researchers have found that constructing composite metal oxides consisting of two or more metallic elements is an effective way to improve the ORR activity [21-26]. Among the composited transition metal oxides, spinel manganese-cobalt oxides (Mn_xCo_3-xO₄), especially those combined with conductive materials, have shown large potential to replace Pt-based catalysts for ORR [24-25, 27-31]. For instance, Liang et al. [28] synthesized manganese-cobalt spinel MnCo₂O₄/graphene hybrid through a two-step solvothermal method, which exhibits boosted ORR performance compared with pure cobalt oxide hybrid in alkaline solutions. Ma et al. [30] prepared carbon nanotubes supported spinel cobalt-manganese oxide nanoparticles through an oxidation-precipitation and insertion-crystallization process. And the hybrid exhibits higher kinetic current and better stability than commercial Pt/C catalyst. Despite such considerable improvements, there is still a need for further enhancement of ORR performance of manganese-cobalt oxides for the purpose of practical application. On the other hand, in addition to the conventional high-temperature solid-state methods, Mn_xCo_3-xO₄ catalysts are usually synthesized through oxidation-precipitation under aqueous ammonia and then crystallization at high temperature (150–180 ℃) [25, 26, 28-30], which not only increases the energy consumption but also can induce the growth of oxide nanoparticles [25]. However, the electrocatalytic performance of nanosized metal oxides is generally higher than that of larger-sized ones, since nanosized materials can reduce diffusion resistance and have larger surface areas, thus exposing more active sites. Hence, it is highly desirable to develop a facile and low-temperature synthesis method to fabricate homogeneously dispersed and ultrasmall nanocrystalline Mn_xCo_3-xO₄ hybrids.

Herein, through a facile precipitation-dehydration process at low temperature (60 ℃), the hybrids of spinel manganese-cobalt oxide nanoparticles and conductive carbon black were successfully fabricated. Benefitting from the low temperature method, the particle size of metal oxides is ultrasmall (~5 nm) and the graphitic structure of carbon black is well maintained. The bimetallic composite oxide hybrid (Mn_1.5Co_1.5O₄/C) exhibits remarkable enhancement of ORR activity and stability compared to single metal oxides (both Mn₃O₄/C and Co₃O₄/C). Mn_1.5Co_1.5O₄/C is further integrated into a Zn-air battery as cathode catalyst and displays excellent discharge performance. And the essential reason for the enhancement of activity can be attributed to the presence of the mixed Mn³⁺ and Mn⁴⁺ cations.

Manganese(Ⅱ) acetate tetrahydrate (Mn(CH₃COO)₂·4H₂O) and ethanol were obtained from Tianjin Guangfu Fine Chemical Research Institute. Cobalt(Ⅱ) acetate tetrahydrate (Co(CH₃COO)₂·4H₂O) was purchased from Alfa Aesar. Potassium hydroxide was provided by Aladdin Industrial Corporation. Vulcan XC-72R was obtained from Carbot. 20 wt% Pt/C was supplied by Johnson Matthey. Nafion solution (5 wt%) was purchased from DuPont. Milli-Q ultrapure water (> 18.25 MΩ·cm^-1) was used in all experiments. All the chemicals were reagent grade and used as received without further purification.

The synthetic procedure was adapted from the literature [17]. In a typical reaction, 0.05 g of conductive carbon black (VXC-72R C) was ultrasonically dispersed in 50 mL anhydrous ethanol. Then 0.32 g Mn(CH₃COO)₂·4H₂O and 0.32 g, Co(CH₃COO)₂·4H₂O were added to the carbon black suspension. After heating the mixture to 60 ℃, 2.6 mL KOH-C₂H₅OH solution (2 mol L^-1) was dropwise added under stirring and then maintained at 60 ℃ for 24 h. The product was separated and rinsed by repeated centrifugation with ethanol and deionized water, and dried overnight at 60 ℃. The sample was then used after grinding into powder, without any further size selection. The obtained sample was denoted as Mn_1.5Co_1.5O₄/C.

The synthesis of Mn₃O₄/C and Co₃O₄/C was similar to that of Mn_1.5Co_1.5O₄/C, except for the amount of metal salts added was 0.64 g manganese(Ⅱ) acetate tetrahydrate or 0.64 g cobalt(Ⅱ) acetate tetrahydrate, respectively.

X-ray diffraction (XRD) patterns were recorded on a Bruker AXS D8 focus (D8-S4) X-ray diffractometer equipped with a nickel-filtered Cu K_α radiation (λ = 1.5418 Å) at a scan rate of 5° min^-1. The Raman spectra were obtained by a Raman spectrometer (inVia reflex) using laser excitation at 532 nm. Transmission electron microscopy (TEM) and corresponding energy dispersive spectrum (EDS) characterization were performed with a JEM-2100F transmission electron microscope, working at 200 kV to analyze the morphology and composition of the prepared samples. X-ray photoelectron spectroscopy (XPS) analysis was performed on a PHI-1600 instrument equipped with Al K_α radiation and all the binding energies were calibrated by the C 1s peak at 284.8 eV.

All the electrochemical measurements were carried out on an IVIUMSTAT workstation (Ivium Technologies BV, Netherlands) equipped with a modulated speed rotator (MSR) (Pine Research Instrumentation, Inc.) in a standard three-electrode system (a modified glassy carbon electrode as the working electrode, a graphite rod as the counter electrode, and a Ag/AgCl (saturated KCl) electrode as the reference electrode by using 0.1 M KOH as the electrolyte. The working electrode modified with a thin catalyst film was prepared as follows: 5 mg catalyst was dispersed in 1 mL isopropanol-water mixture with the volume ratio of 3:1 with 50 μL 5 wt% Nafion solution by ultrasonic dispersion to form a homogeneous ink. Then 10 μL the catalyst ink was dropped onto a glassy-carbon electrode (5 mm in diameter) which was newly polished and washed before use. The carbon electrode was dried under room temperature. For comparison, the ink of the commercial Pt/C catalyst (20 wt%) was prepared with the same procedure. Before each measurement, the electrolyte was purged with O₂ for at least 30 min. All data were reported without iR compensation.

Zinc-air batteries were fabricated with a zinc foil as the anode and catalysts loaded on carbon paper as the cathode (catalyst load 1.0 mg cm^-2) by using 6 M KOH as the electrolyte. The constant-current discharge test was carried out with a LANHE battery testing station and the polarization curve test was carried out with an IVIUM electrochemical workstation.

The structures of the obtained materials were first characterized by XRD and Raman spectroscopy. As shown in Figure 1a, the XRD patterns of all the samples exhibit a broad peak at 2θ ≈ 25^o, which is assigned to the (002) planes of graphitic carbon. The diffraction peaks of the sample prepared with only Mn(CH₃COO)₂·4H₂O or Co(CH₃COO)₂·4H₂O as the metal source can be attributed to the characteristic peaks of tetragonal Mn₃O₄ (JCPDS No. 24-0734) or cubic Co₃O₄ (JCPDS No. 43-1003) respectively. The XRD pattern of bimetallic manganese-cobalt oxide can be assigned to the cubic spinel structure, which is consistent with the reported literature [29], and the peaks are slightly shifted to smaller angles compared with pure Co₃O₄, implying the incorporation of larger sized Mn cations into cubic Co₃O₄ lattice [28, 29]. Raman spectra of the prepared materials are shown in Fig. 1b. The peak at approximately 1350 cm^-1 is attributed to D band representing the structural defects on the graphitic plane and the peak at approximately 1586 cm^-1 is ascribed to G band reflecting the degree of graphitization of carbon. The peak intensity ratios of D to G band (I_D/I_G) of all the samples are approximately equal to 1 implying well graphitic degree of all the samples. This result indicates that the low-temperature synthetic process does not destroy the structure of carbon, which is beneficial to the enhancement of electrical conductivity and promoting charge transfer in the loaded manganese-cobalt oxides [32].

As shown in the TEM images (Fig. 2), all the Mn_xCo_3-xO₄ nanoparticles are uniformly dispersed on conductive carbon black, and the particle sizes are all around 5 nm. It is worth noting that the particle sizes do not increase significantly with the change of composition. Recently, Zhao and her co-workers reported the synthesis of manganese-cobalt oxide/nitrogen-doped multiwalled carbon nanotube hybrids through a solvothermal method (150 ℃) and concluded that the particle size increased notably with increasing the Mn/Co ratio [25]. While the facile and low-temperature precipitation-dehydration method applied here not only the energy consumption but also prevents the increase of particle size with the change of Mn/Co ratio. The HR-TEM images (Figs. 2b, d, f and Fig. S1) and the selected area electron diffraction patterns (inset of Figs. 2b, d, f) indicate that these oxides particles are all well crystallized (partially highlighted by yellow circles). The corresponding elemental mapping images of Mn_1.5Co_1.5O₄/C (Figs. 2g-j) prove that Mn and Co are homogeneously dispersed on the carbon matrix. Mn/Co atomic ratio of Mn_1.5Co_1.5O₄/C measured by TEM-EDS (Fig. S2) and XPS (Fig. S3) is shown in Table S1, which is close to 1 and consistent with that of Mn_1.5Co_1.5O₄ reported in the literature [29]. According to the reported literature [25, 28], the crystal structure of Mn_xCo_3-xO₄ is highly related to the Mn/Co ratio. Generally, low Mn content (0 ≤ x ≤ 1.3) tends to form the cubic phase, while high proportions of Mn favours the tetragonal phase. Hence, the most familiar and studied bimetallic manganese-cobalt oxide with cubic structure is MnCo₂O₄ [24, 25, 27, 28, 33]. Whereas, herein, the bimetallic manganese-cobalt oxide with cubic spinel structure and high Mn content (x = 1.5), which are both beneficial for ORR [25, 29], is successfully synthesized by the facile precipitation-dehydration method at low temperature.

The ORR catalytic activities of as-prepared Mn_xCo_3-xO₄/C were first evaluated by linear sweep voltammetry (LSV) on a rotating disk electrode (RDE) in O₂-saturated 0.1 M KOH with a scan rate of 10 mV s^-1, which is commonly used in the literature [25, 27, 34]. As shown in Figure 3a and Table S2, Mn_1.5Co_1.5O₄/C has much better ORR activity than Mn₃O₄/C and Co₃O₄/C. The half-wave potential (E_1/2) and diffusion limiting current density (j_L) of Mn_1.5Co_1.5O₄/C are 0.820 V vs RHE and 5.6 mA cm^-2, respectively, which are comparable to those of commercial 20 wt% Pt/C (0.814 V vs RHE, 5.6 mA cm^-2) used here and in the literature [25, 34, 35]. It is worth noting that the polarization curve of Mn₃O₄/C is steeper than that of Co₃O₄/C in the kinetic control zone (1.0-0.85 V vs RHE). The initial activity of Mn₃O₄/C is better, that is, the current response increases faster with the change of voltage, suggesting that the intrinsic ORR activity of manganese is superior to that of cobalt. While, the ORR activity of Mn_1.5Co_1.5O₄/C in the kinetic region is better than that of Co₃O₄/C and even Mn₃O₄/C. As shown in Table S2, kinetic current density (j_K) (at 0.90 V vs RHE in 0.1 M KOH) of Mn_1.5Co_1.5O₄/C (1.34 mA cm^-2) is much higher than that of Mn₃O₄/C (0.65 mA cm^-2) and Co₃O₄/C (0.15 mA cm^-2), indicating that constructing bimetal oxides can enhance ORR performance, which is consistent with reported literature [21, 25, 28]. The kinetic performance was further quantificationally evaluated by mass transfer corrective Tafel plots. As shown in Fig. 3b, Tafel slopes are in the order of Mn_1.5Co_1.5O₄/C (98 mV dec^-1) < Mn₃O₄/C (107 mV dec^-1) < Co₃O₄/C (163 mV dec^-1), which is consistent with the result of ORR polarization curves.

Then, the ORR catalytic performance of Mn_xCo_3-xO₄/C was investigated by cyclic voltammetry (CV), Mn_1.5Co_1.5O₄/C exhibits a bigger closed-cycle than the others in the CV curves (Fig. 3c). A larger closed-loop indicates a larger current density response at a certain overpotential, implying a higher ORR activity. The result of CV curves that Mn_1.5Co_1.5O₄/C has higher ORR activity is well consistent with that of ORR polarization curves.

ORR efficiency of prepared catalysts was further evaluated by measuring the electron transfer number (n) and yield of peroxide species. The electron transfer number can be obtained from polarization curves at different rotating speeds on rotating disk electrode (RDE) (Figs. 4a-d and Fig. S4) and the ring and disk current recorded on rotating ring-disk electrode (RRDE) (Figs. 4e, f). Although the Koutecky-Levich (K-L) method is not suitable for calculating the exact n value of ORR [36, 37], it can be evidently seen form K-L plots (Fig. 4d) that the n values of the hybrids of spinel oxides and carbon black remarkably increase compared with carbon black. As shown in Fig. 4e, the disk current represents the oxygen reduction current and the ring current represents the oxidation current of peroxide species. The ring current of Mn₃O₄/C and Mn_1.5Co_1.5O₄/C is lower than that of Co₃O₄/C and close to 0, implying that the production of peroxides is very little. As shown in Figure 4f, the n values of Mn₃O₄/C and Mn_1.5Co_1.5O₄/C are calculated to be above 3.95 in the range of 0.4-08 V (vs RHE) and the percentages of generated peroxide species are below 5%. In comparison, Co₃O₄/C has a lower electron transfer number and obvious peroxide species are produced during oxygen reduction.

For practical application, long-term durability is an important parameter of catalytic activity. The stability of prepared catalysts was evaluated by chronoamperometric measurement at 1600 rpm in O₂-saturated 0.1 M KOH. As shown in Fig. 5a, the relative current density of Co₃O₄/C decreases to 81.2% after 12 h test, while that of Mn₃O₄/C decreases even more and only remains 62.4%. Whereas, the relative current density of Mn_1.5Co_1.5O₄/C only decreases 1.4% after 12 h test, showing much better stability than Co₃O₄/C, Mn₃O₄/C and commercial Pt/C (decreasing 12% only after 20000 s test) (Fig. S5). Constructing of bimetallic oxide greatly improves the long-term durability, which may be directly attributed to the ascending of valence state of Mn and the formation of new metal-ion redox couples [25].

To further investigate the performance of Mn_1.5Co_1.5O₄/C catalyst in practical application, the primary Zn-air batteries were constructed using a polished zinc plate as the anode and Mn_1.5Co_1.5O₄/C or Pt/C catalyst as the air cathode. As shown in Figure 5b, although the voltage drop of Mn_1.5Co_1.5O₄/C catalyst is larger than that of Pt/C, the initial galvanostatic discharge voltage with Mn_1.5Co_1.5O₄/C catalyst is ~1.3 V, which is a little higher than that with the benchmark Pt/C (~1.2 V). As displayed in the polarization and corresponding power density curves (Fig. 5c), the peak discharge power density with Mn_1.5Co_1.5O₄/C catalyst is 173 mW cm^-2 at 276.7 mA cm^-2, superior to those with Pt/C catalyst (142 mW cm^-2 at 198.3 mA cm^-2), indicating an excellent performance as a cathode catalyst in Zn-air battery and a potential for future application.

All the above results indicate that Mn_1.5Co_1.5O₄/C catalyst exhibits remarkable enhancement of ORR activity and stability, and is one of the most effective and stable manganese-cobalt oxide electrocatalysts for ORR compared with the reported ones listed in Table S3.

In order to explore the essential reasons for the enhancement of activity of spinel manganese-cobalt oxides, XPS measurement was performed to analyze the differences in surface electronic structures of prepared catalysts. As Mn in manganese-cobalt oxides has been identified as an active site for ORR [38] and the initial activity of Mn₃O₄/C is better than that of Co₃O₄/C in the kinetic region as shown in Fig. 3a, only the differences in Mn electronic states of Mn₃O₄/C and Mn_1.5Co_1.5O₄/C were investigated.

XPS spectra of Mn₃O₄/C and Mn_1.5Co_1.5O₄/C are shown in Fig. 6, which all have been calibrated by using the carbon peak at 284.8 eV (Fig. S6). As shown in Mn 2p spectra (Fig. 6a), there are two peaks at 642.1 eV and 653.7 eV and the space between two peaks is about 11.6 eV corresponded to Mn 2p spin-orbital splits 2p_3/2 and 2p_1/2, which are the characteristic peaks of Mn₃O₄ [39, 40]. Compared with Mn₃O₄/C, the two peaks of Mn_1.5Co_1.5O₄/C shift toward higher binding energy, indicating the presence of more Mn cations with the high valence state in bimetallic oxides. The deconvoluted Mn 2p spectra can be fitted with three pairs of spin-orbit triplet peaks, representing the oxidation states of Mn²⁺, Mn³⁺ and Mn⁴⁺, respectively [30, 41]. However, the differences among the oxidation states of Mn²⁺, Mn³⁺ and Mn⁴⁺ are very little and difficult to distinguish, hence the further differentiation-imitating analysis will engender fairly great errors. Fortunately, Mn 3s orbital can split into two peaks due to the coupling between the nonionizing electron and valence electron. And the space between the two peaks (ΔE) in Mn 3s spectra is sensitive to the variation of the oxidation state of Mn, which can be utilized to identify the exact valence state of Mn in the oxides [40, 42, 43]. As shown in Figure 6b, the energy separation of Mn 3s for Mn₃O₄/C is 5.55 eV, indicating that the Mn oxidation state is 2-3, and Mn exists as a mixture of Mn²⁺ and Mn³⁺. Whereas, the energy separation for Mn_1.5Co_1.5O₄/C is 4.95 eV, which significantly decreases compared with Mn₃O₄/C, implying the ascending of valence state of Mn and the presence of Mn³⁺ and Mn⁴⁺ in the bimetallic oxide. Mn⁴⁺ cations derive from partial electron transfer between Mn³⁺ and Co³⁺ in the octahedral sites, namely, Mn³⁺ + Co³⁺ → Mn⁴⁺ + Co²⁺ [44, 45]. As shown in Fig. S7, Co 2p_1/2 of Mn_1.5Co_1.5O₄/C shifts toward lower binding energy compared with that of Co₃O₄, indicating the decrease in valence of Co, which is consistent with the results of Mn 2p and 3s spectra. Mn cations with high valence states, especially the mixed Mn³⁺/Mn⁴⁺ valence states, are more beneficial for ORR, hence Mn_1.5Co_1.5O₄/C exhibits enhancement of ORR activity compared with Mn₃O₄/C. And the high stability of Mn_1.5Co_1.5O₄/C could also be ascribed to the intrinsic Mn cations with high valence state, as well as Mn⁴⁺ generated during the process of testing.

In summary, we provide a facile and low-temperature precipitation-dehydration method to fabricate homogeneously dispersed and ultrasmall nanocrystalline Mn_xCo_3-xO₄ hybrids. And the bimetallic composite oxide (Mn_1.5Co_1.5O₄/C) hybrid, which possesses ultrasmall particle size (~ 5 nm), cubic spinel structure and high Mn content, exhibits remarkable enhancement of activity and stability for ORR compared with single metal oxide (both Mn₃O₄/C and Co₃O₄/C). In addition, Mn_1.5Co_1.5O₄/C displays an excellent discharge performance in the Zn-air battery as a cathode catalyst. We also explore the differences in Mn electronic states of Mn₃O₄/C and Mn_1.5Co_1.5O₄/C and conclude that the enhancement of ORR activity of manganese-cobalt bimetallic oxides can be ascribed to the presence of the mixed Mn³⁺ and Mn⁴⁺ cations.

The energy-dispersive spectroscopy (EDS) and XPS survey spectrum of Mn_1.5Co_1.5O₄/C; ORR polarization curves of VXC-72R C at different rotating speeds; chronoamperometric response of Pt/C; C 1s XPS spectra of Mn₃O₄/C and Mn_1.5Co_1.5O₄/C; Mn/Co atomic ratio of Mn_1.5Co_1.5O₄/C characterized by TEM and XPS; ORR activities of Mn_xCo_3-xO₄/C (x = 0, 1.5, 3) and Pt/C catalysts; comparison of ORR activities with the reported Mn_xCo_3-xO₄ catalysts in alkaline media.