The Roles of Composition and Mesostructure of Cobalt‐Based Spinel Catalysts in Oxygen Evolution Reactions

Abstract By using the crystalline precursor decomposition approach and direct co‐precipitation the composition and mesostructure of cobalt‐based spinels can be controlled. A systematic substitution of cobalt with redox‐active iron and redox‐inactive magnesium and aluminum in a cobalt spinel with anisotropic particle morphology with a preferred 111 surface termination is presented, resulting in a substitution series including Co3O4, MgCo2O4, Co2FeO4, Co2AlO4 and CoFe2O4. The role of redox pairs in the spinels is investigated in chemical water oxidation by using ceric ammonium nitrate (CAN test), electrochemical oxygen evolution reaction (OER) and H2O2 decomposition. Studying the effect of dominant surface termination, isotropic Co3O4 and CoFe2O4 catalysts with more or less spherical particles are compared to their anisotropic analogues. For CAN‐test and OER, Co3+ plays the major role for high activity. In H2O2 decomposition, Co2+ reveals itself to be of major importance. Redox active cations in the structure enhance the catalytic activity in all reactions. A benefit of a predominant 111 surface termination depends on the cobalt oxidation state in the as‐prepared catalysts and the investigated reaction.

As reproducibility is of pivotal importance in catalyst synthesis, this aspect is being highlighted for the magnesium cobalt hydroxide precursor and the resulting spinel catalyst. Figure
Mössbauer spectra were recorded at low temperature (4.3 K) and high magnetic field (5 T) parallel to the -ray propagation direction. For the Fe 2+ 1/3Co 2+ 1/3Fe 3+ 1/3 LDH precursor material, the line intensity ratio A23 close to 2 reveals mainly antiferromagnetic behavior with random spin orientation, as it would be expected for LDH-based materials. [1] As a result, the evaluation of the experimental data points was achieved by means of an equidistant distribution of magnetic hyperfine fields, as no individual subspectra corresponding to different crystallographic sites could be discerned. With this method, we obtain a mean magnetic hyperfine field of ca. 46 T, with an isomer shift of 0.48 mm/s indicating that we are mainly dealing with Fe 3+ contributions. This is presumably caused by the extended measurement time required for Mössbauer spectroscopy, and the associated oxidation of the originally present Fe 2+ fraction, which was verified in XPS measurements. The sample preparation for XPS is comparatively fast and yet not the expected amount of 50 % of Fe 2+ is found. However, the contribution of the measured Mössbauer spectrum is predominantly from one phase and the X-ray diffraction pattern did not show any sign of a secondary phase. In our recently published paper, a premature spinel by-phase was detected by XRD and Mössbauer spectroscopy. [2] The synthesis, washing and drying procedure in the present work was improved and prevented this premature oxidation and consequential by-phase formation. Even though oxidation occurs during sample preparation and measurement of the X-ray diffraction data, the phase purity is potentially explained by a green rust -mössbauerite related situation. [3] Fe 2+ oxidizes to Fe 3+ , but through deprotonation of the hydroxyl groups the resulting additional excess charge in the brucite-like layers is compensated and the structure is stable.   In order to obtain a more detailed view on potential structural changes between precursor, resulting product, and possible by-phases, Mössbauer spectra were also recorded for a sample calcined at 400 °C and 600 °C. The results for the Fe 2+ 1/3Co 2+ 1/3Fe 3+ 1/3 LDH precursor material are discussed above.
Some changes can be observed for the sample heated to 400 °C, such as the increase of the magnetic hyperfine field to ca. 50 T, pointing towards a general phase change of the material, while the isomer shift is nearly unchanged. The magnetic hyperfine field distribution shows first signs of two spectral contributions, which, together with the reduction in A23 to ca. 1.8, representing beginning ferrimagnetic in-field alignment, indicates the starting formation of spinel material.
The most dramatic change takes place after heating to 600 °C, with the resulting spectrum revealing two clearly separated sextet subspectra. Such a spectral structure is typical for ferrimagnetic spinel systems, making it possible to calculate the degree of inversion from the relative spectral areas, as these are proportional to the number of Fe ions on the tetrahedral and octahedral sites. We obtain an inversion parameter of 0.76(3), with this material thus being closer to the inverse (1) than the regular spinel (0), as one would expect for phase pure CoFe2O4. [5] The mean spin canting angle, derived from A23, decreased to ca. 40°, which is still substantial, usually caused by the high magnetocrystalline anisotropy and canted surface spins. Latter could here be connected to the enhanced surface fraction and the strongly porous nature of the material.
As Mössbauer spectra displayed a considerable change in magnetic structure, the formation of spinel material from the LDH precursor was also investigated by magnetometry. The M(H) sweep of the LDH precursor sample displays a quasi-paramagnetic behavior at room temperature, while a slow rise in magnetization is seen at 4.3 K, indicative of the strongly canted state and resulting lack of full saturation. A very minute saturation effect is observable at very small fields at 300 K, which could represent a miniscule parasitic phase. After heating to 400 °C, the 9 T magnetization at 4.3 K has decreased from 48 Am 2 /kg to ca. 19 Am 2 /kg, while the sample now also shows a significant remanence and coercive field that were almost completely absent in the the untreated sample. The most dramatic change takes place after heating to 600 °C, where we observe a very broad hysteresis curve at 4.3 K, with a high magnetization of ca. 73 Am 2 /kg at 9 T. This value is only slightly below the literature value of 80 Am 2 /kg [6] for pure, well-ordered CoFe2O4 nanoparticles, which can be explained by the still substantial degree of spin canting deduced from the 5 T Mössbauer spectrum for this sample. We can therefore assume that this slightly lowered magnetization is caused by unordered surface spins, and not by an impurity of the material at hand. Regarding the M(T) sweeps, we can of course observe the same changes in magnetization, but as the materials are not in saturation at the applied field of 1 T, we chose to normalize the curves to the end points of the FC branches. Thanks to this, we can directly compare the qualitative temperature dependent progression of each sample's magnetization. We can observe that the LDH precursor does not display any distinct features, with the magnetization continuously dropping towards higher temperatures after attaining the paramagnetic state. In contrast, both the 400 °C and the 600 °C heated sample display typical relaxation behavior known from magnetic nanoparticles, with the inflection points of ca. 86 K for the 400 °C and 130 K for the 600 °C sample clearly showing a rise in blocking temperatures.
In conclusion the magnetic investigations underline the phase purity of the sample calcined at 600 °C, indicating a possible minor by-phase or incomplete transformation for the sample calcined at 400 °C.
Isotropic Co3O4 and CoFe2O4 was synthesized to further investigate the impact of morphology on the catalytic activity. In Figure S9 to S12 characterization data of the before mentioned can be found.      To analyze the preferred surface termination of the anisotropic samples and to show a more statistical facet distribution for the isotropic analogs, TEM measurements and electron diffraction were performed. Due to the intergrown nature of the anisotropic particles, it was not possible to achieve electron diffraction patterns of a single platelet in most cases. The aggregates make a surface termination determination nearly impossible. Nevertheless, a 111 surface termination is considered for all samples of the anisotropic substitution series. For Co3O4 several reports can be found in literature, confirming the topotactic spinel formation. [7] The related dehydroxylation of Mg(OH)2 to MgO has been investigated as well. [8] For CoFe2O4 we recently showed the same transformation and for Co2FeO4 it was shown in the present work (See. Figure 5 in the main text). [2] The 111 planes of the spinel and the 001 planes of the brucite type hydroxides exhibit a very similar d-spacing and therefore transformation occurs along them. This topotactic transformation leads to the preservation of the morphology. As all anisotropic spinels clearly exhibit the hexagonal platelet morphology of their (layered double) hydroxide precursor, this is additional prove for the suggested predominant surface termination.