Catalyst and electrode preparation
The overall synthetic and processing protocol is summarized in Figure 1. We synthesized carbon-coated Co (Co@C) nanoparticles, starting from the thermal treatment of Co(bIm)2 (bIm = 2-benzimidazolate), a metal-organic framework (MOF) precursor (ZIF-9)37. Then Co@C was oxidized at low-temperature to achieve its full transformation into cobalt oxide nanoparticles, covered by an amorphous, nitrogen doped-carbon coating derived from the organic skeleton (Co3O4@C, Figure 1). Powder X-ray diffraction (PXRD) patterns and Raman spectra confirmed the presence of a Co3O4 phase and the carbon support (Figures S1–S4). High-resolution transmission electron microscopy (HRTEM) the presence of graphitic-like nanostructures all around the sample, embedding the Co3O4 nanoparticles. Some of these C-nanostructures had a nanosheet-like morphology (Figure 2), while some others were folded forming onion-like rings around the Co3O4 nanoparticles (Figure S5). Electron energy loss spectroscopy in scanning TEM mode (EELS-STEM) confirmed the chemical composition of the nanoparticles and surrounding nanostructures (Figure S6). The Co3O4@C composition was determined as (Co3O4)(H2O)0.30(OH)0.85C2.00N0.05 by thermogravimetry elemental analysis(Figure S7 and Table S1).
For the preparation of the electrode composites, Co3O4@C was mixed with graphite (G) and paraffin oil (PO) in the desired ratio (see Methods section) to prepare a homogeneous composite (Co3O4@C/GPO) with the desired Co3O4@C content up to 40% (40- Co3O4@C). Composites above 40% were mechanically too fragile for further processing into the working electrode pocket. HRTEM images and EELS-STEM maps showed the similar nanostructures within Co3O4@C/GPO and close contact between Co3O4@C and GPO (Figure 2 and Figure S8). X-ray photoemission spectroscopy (XPS) analysis was employed to further identify the surface chemical composition and the mixed oxidation state Co2+/3+ (Figure S9) consistent with the presence of the Co3O4, as confirmed by PXRD and HRTEM data38. XPS spectra from the Co3O4@C/GPO composite show no differences respect to the Co3O4@C precursor, demonstrating the absence of chemical modification during composite preparation.
OER electrocatalytic activity in 1 M H2SO4
The x-Co3O4@C/GPO composites (x corresponds to the % in weight for the metal oxide) were inserted into the pocket of a working electrode and used as anode during electrochemical water oxidation in 1 M H2SO4 (pH ≈ 0.3). The cyclic voltammetry (CV) showed the appearance of a catalytic current density on the Co3O4@C/GPO electrode at relatively low overpotentials, which was sustained after successive cycling curves (Figure S10). Comparative linear sweep voltammetry (LSV) showed an enhanced electrochemical activity upon increasing Co3O4@C content, reaching a very low onset overpotential (ηonset = 190 mV) for the 40-Co3O4@C/GPO electrode (Figure 3). These electrodes reach 10 mA cm−2 currents at just ≥ 356 mV overpotential. Interestingly, no sign of a transport-limited regime appeared in the studied potential range, reaching over 20 mA cm−2 at η ≥ 393 mV.
We prepared analogous IrO2/GPO working electrodes to benchmark our results in the same conditions with the state-of-the-art IrO2. The IrO2/GPO anodes delivered higher overpotentials, \({{\eta }}_{\text{j}=10 \text{m}\text{A} {\text{c}\text{m}}^{-2}}\) = 368 mV at 10 mA cm−2 and \({{\eta }}_{\text{j}=20 \text{m}\text{A} {\text{c}\text{m}}^{-2}}\) = 396 mV at 20 mA cm−2, slightly above those obtained for the Co3O4@C-based electrode (Table S2). This competitive activity becomes even more significant if we normalize current density per gram of catalyst (Figure 3c).
Tafel analyses of the LSV data yielded slopes of 139 mV dec−1 for Co3O4@C and 83 mV dec−1 for IrO2 (Figure 3d), suggesting a different reaction mechanism (rate-limiting step) for these two catalysts, and indicating a faster increment of current density with the applied potential for IrO239, 40. Interestingly, this is compensated by the lower onset potential of Co3O4@C/GPO. The electrochemical double-layer capacitance (EDLC) of Co3O4@C/GPO and IrO2/GPO were calculated as 25 and 2 mF cm−2, respectively, with 0.03 mF cm−2 for the blank GPO (Figure S11), respectively. This indicates a greater electrochemical active surface area for Co3O4@C/GPO, due to its higher density of active sites in Co3O4@C/GPO, thanks to is nano-structuration, favouring the higher current densities observed in the potential range studied41, 42.
Finally, we measured anodic oxygen evolution during chronopotentiometry experiments with Co3O4@C/GPO electrodes (Figure S12). We found over >96% Faradaic efficiency, confirming that OER is the dominant process at these electrodes’ surface, and confirming no significant oxidation of the carbon-based matrix is taking place in these conditions.
OER electrocatalytic stability in 1 M H2SO4
As mentioned before, stability is a critical issue for earth abundant OER catalysts in acidic media43–47. To determine the stability of our Co3O4@C/GPO electrodes, we take advantage of the benchmarking protocol designed by Jaramillo’s group6,48 that uses as figure of merit the overpotential required to achieve and maintain a 10 mA cm−2 current density per geometric area at ambient temperature after two hours of continuous water electrolysis. The corresponding chronopotentiometry data (Figure 4a-c) show very good stability for all electrodes, independently of their Co3O4@C content. In all cases, \({{\eta }}_{\text{j}=10 \text{m}\text{A} {\text{c}\text{m}}^{-2}}\) after 2 hours shows just a small increment. In the case of our best electrodes, the 40-Co3O4@C/GPO, this increment is of just 3 mV, and the stability is maintained for long times. After 43 h of continuous electrolysis, the overpotential is essentially identical to the starting value (Figure 4a).
The benchmarking of these electrodes with previous literature is highlighted in the \({{\eta }}_{j=10 \text{m}\text{A} {\text{c}\text{m}}^{-2}, t=2 \text{h}}\) vs \({{\eta }}_{j=10 \text{m}\text{A} {\text{c}\text{m}}^{-2}, t=0 \text{h}}\) plot (Figure 4d). This comparative plot illustrates the high activity and stability of our electrodes. The three of them appear at the diagonal of the graph, as expected for sustainable performance, and very close and competitive to the results obtained with noble metal counterparts. For the first time, earth abundant anodes successfully pass this benchmarking protocol for OER performance in acidic media.
Stability number (S-number) and Activity-Stability Factor (ASF) were also proposed as key metrics for estimating lifetime and long-term stability for electrocatalysts49–51. Thus, we analyzed the electrolyte after stability tests to check for Co leaching (Table S3). We found the presence of Co but at the ppb level, corresponding to just ≈ 0.4% of the total. Based on this number, we can estimate a 25 S-number, an ASF of 101 and a lifetime of 462 h. These estimations are comparable even to Ir-based catalysts such as SrIrO3 in analogous conditions, and confirm the promising performance/stability of these electrodes. It is worthy to mention that this small Co loss does not significantly affect performance. So we assign it essentially to catalytically non-active areas.
Post-electrolysis Co3O4@C/GPO characterization
To further confirm the stability of Co3O4@C as a genuine OER catalyst, we characterized the structural and chemical evolution of the electrodes after these 2 h electrolysis at 10 mA cm−2 in 1 M H2SO4. The powder XRD patterns did not show any significant change nor shift in the observed peaks, still typical of Co3O4@C and graphite (Figure S13). This suggests no major structural changes are occurring to the bulk of the material Co3O4@C.
We explored potential changes on the chemical composition of the catalyst due to OER process by XPS characterization of the fresh electrode and after water electrolysis under different conditions. The Co 2p XPS spectrum of the fresh Co3O4@C/GPO electrode (Figure 5a) shows two peaks located at 794.9 eV (Co 2p 1/2) and 779.7 eV (Co 2p3/2), corresponding to the spin-orbit splitting of the 2p orbital. Both components contain equivalent chemical information. The deconvoluted analysis of the peaks reveals the presence of two different chemical components, which we attribute to the Co3+ (blue) and Co2+ (green) states, in agreement with the presence of Co3O438. In addition, we observe two doubled satellite peaks arising from charge transfer and final states effects from Co2+ (satellite A, yellow) and Co3+ (satellite B, red),52 again characteristic of Co3O4. We also analyzed the O 1s peak (Figure 5a). In addition to the Co-O component (brown) related to the Co3O4, we observe a higher binding energy component attributed to residual OH/H2O (pink). Quantitative analysis of the Co 2p and O 1s core levels of Co3O4@C/GPO after chronopotentiometry at 10 mA cm−2 for 2 h (Figure 5b) and 5 mA cm−2 for 24 h (Figure 5c) revealed no shifts in the binding energies of the components respect to the fresh sample. Crucially, the main spectral features attributed to a Co3O4@C catalyst remain unaltered after water electrolysis, which demonstrates the preservation of the oxidation state of the catalyst. Changes in the intensity of the OH/H2O component on the O 1s can be fairly attributed to the different environmental conditions of the emersed electrode (see Experimental section). XPS analysis of the C component does not evidence significant changes in the oxidation state of the C 1s peak, supporting the preservation of the carbon-based matrix (Figure S14) in agreement to the obtained Faradaic efficiencies. Nitrogen detection in the system is below our resolution limit; therefore, no discussion is referred to this element. Based on the current analysis, the most important finding is that the oxide film is stable and no cobalt oxide is lost nor further oxidized during the electrolysis process. In summary, XRD and XPS strongly support the bulk and surface stability of Co3O4@C during acidic OER electro catalysis, and its genuine catalytic activity.
We also investigated the Co3O4@C/GPO composite after 2 h electrolysis at 10 mA cm−2 by means of HR-TEM (Figure 6). The images and power spectra (FFT) analyses also confirm a high structural and chemical stability. Neither crystallinity nor particle size are affected by the electrochemical process.
Critical role of GPO
To investigate the actual role of the carbon paste in the stability of the electrodes, we carried out additional alternative experiments. First, we directly deposited Co3O4@C on a glassy carbon (GC) electrode. This electrode showed a significantly lower electrocatalytic activity when compared with the Co3O4@C/GPO (Figure 3-4 vs Figure S15). More importantly, after 30-min of the benchmarking test in 1 M H2SO4, the Co3O4@C/GC electrode is apparently deactivated. This suggests that the GPO binder is fundamental to confer the acidic stability and activity of the Co3O4@C component.
This effect of the GPO binder could be due to a modified local pH at the electrode/electrolyte interface53–55. To check this hypothesis, we decided to investigate the effect of the GPO binder on the local pH through the reversible H+/H2 pair as catalyzed with commercial Pt/C56. The reversible potential for this model reaction differs when the Pt/C is directly deposited on a graphite electrode, or when incorporated into a GPO electrode as observed in their CV plots in a hydrogen saturated 1 M H2SO4 electrolyte (Figure S16). An average value of + 0.001 V vs ERHE was estimated for the Pt/C catalyst, in good agreement with the theoretical +0.0 V value. A +0.06 V vs ERHE was found for the (Pt/C)/GPO electrode. If we associate this potential difference to the local pH, ∆E = 0.059 ∆pH, we can estimate a pH difference of 1 unit between both electrodes, which does not immediately explain the higher stability obtained for the GPO electrodes under acidic water oxidation conditions. Therefore, we associate the protective function essentially to the hydrophobicity environment, which avoids proper solvation of the oxides, precluding its dissolution.
We also compared the activity/stability of Co3O4@C vs Co3O4 (Figure S15). The corresponding x-Co3O4/GPO electrodes showed good stability during preliminary CV cycles and chronopotentiometry measurement, but at higher overpotentials. A 5 mF cm−2 EDLC was determined, just 1/5 that of Co3O4@C/GPO (Figure S11). Specific surface area from N2 sorption isotherm curves for Co3O4@C was also about 5 times greater than that of Co3O4 (Figure S17). These results suggest that the role of the carbon coating is to improve the nanostructuration of the active Co3O4 material.
In addition, the Co3O4@C/GPO, Co3O4/GPO and IrO2/GPO electrodes were studied by Electrochemical Impedance Spectroscopy (EIS) at different applied potentials. Figure S18 shows the obtained Nyquist plots, which systematically feature two arcs (or distorted arc for IrO2), consistent with two simultaneous/consecutive charge-transfer channels57. Fitting the experimental data to a suitable equivalent circuit model (Figure S18) revealed that the best ohmic contact (reflected by the series resistance, RS) is obtained for the Co3O4@C/GPO (Figure S19a). On the other hand, the charge transfer resistance (Rct), scales inversely with the electrocatalytic activity of the different electrodes, being the lowest one for the Co3O4@C/GPO electrode (Figure S19b). This is consistent with the estimated surface capacitance, which scales with electrode performance, as a result of higher surface area and hence, higher density of catalytic sites (Figure S19c).