Medium‐ and High‐Entropy Spinel Ferrite Nanoparticles via Low‐Temperature Synthesis for the Oxygen Evolution Reaction

High‐entropy oxides are a material class that is currently receiving rapidly increasing attention due to the large variety in composition and the adjustable properties. Cooperative effects between different metal cations in the crystal structure in addition to entropic phase stabilization have proven beneficial for electrocatalytic applications. Most synthesis methods, however, require high synthesis temperatures and long times, and additionally only yield selected samples in good phase‐purity. Furthermore, toxic or scarce elements are often present in large amounts. Herein, a non‐aqueous microwave‐assisted solvothermal synthesis is presented as a fast and low‐temperature alternative for the fabrication of a wide range of earth‐abundant ferrites (AFe2O4). Directly crystalline, phase‐pure spinel ferrites of various compositions ranging from one to seven different A‐ions are successfully obtained after only 30 min at 225 °C. A detailed characterization of their properties in relation to their composition is performed, and they are also employed for the alkaline oxygen evolution reaction. A partial replacement of Fe by Co moreover shows the high versatility of the synthesis that also allows for the simultaneous variation of the B‐ion.

Δ G mix = ΔH mix − TΔS mix (1)   Thermodynamically, the structure can be stabilized if the entropy is high enough to overcome the mixing enthalpy and result in an overall negative ΔG of formation.The entropy in turn increases with an increasing number of different ions (in this case cations) in the structure.It is furthermore dependent on the mole-fraction and is the highest for equal molar ratios. [4,23]According to the definition first introduced for alloys, materials can be classified as "high-entropy" if more than five different elements are present in concentrations above 5% for alloys, [24] or more than five cations for oxides, since the anion fraction is kept constant. [14,23]In this case the configurational entropy (S config ) equals to or exceeds 1.5 R (with R being the gas constant) -at least for alloys. [24,25]For oxides the situation is more complex, since more than one sublattice is present, and a variation of only the cation contribution will not be enough to exceed a S config of 1.5 R. The contribution of each ion in the oxide depends on the molar fraction and the number of different ions on one site and is approx.zero for oxygen. [23]hus, the entropic contribution to ΔG increases with an increasing M/O ratio and the number of different ions on each site. [26]or more than one cationic sublattice, the situation is even more complex, as the overall configuration entropy not only depends on the number of cations but also on the distribution over the different sites. [26]From Equation (1) it is additionally obvious, that the formation of high-entropy materials is favored at high temperatures, resulting in commonly high synthesis temperatures, which is impedimental for nanoparticle synthesis. [23,26]dditionally, enthalpy-driven phase segregation is a common problem encountered in the synthesis of HEOs, especially at lower temperatures. [14,27]he adjustable composition of HEOs is especially interesting for catalytic applications.Adsorption energies and activation barriers are of paramount importance in electrocatalysis, together with synergistic effects between adjacent metal centers -both parameters fundamentally depend on the elemental composition, especially on the surface. [4,28,13]Due to the large number of different elements in the structure, a large number of different surface atomic sites and chemical environments are available for specific adsorption and activation of reactants.Ideally, it can be adjusted through variations in the composition. [29]High entropy alloys have been explored as stable electrocatalysts for the hydrogen evolution reaction (HER), oxygen evolution reaction (OER) and oxygen reduction reaction (ORR) alike.By adjusting the elemental composition and thus electronic and surface properties, even multifunctional catalysts are attainable. [1,4]Apart from noble metals, oxides are frequently employed as electrocatalysts, especially in the field of earth-abundant electrocatalysis.][32] Conductivity and stability against corrosion and cycling are equally important, especially for application in batteries.Entropic stabilization was shown to have a significant influence on capacity retention in HEO electrodes for Liion batteries that were much higher compared to their mediumentropy counterparts. [33]pinel oxides of the general formula AB 2 O 4 are an interesting class of metal oxides, in which oxygen forms a cubic closepacked lattice, the A-ions occupy 1/8th of the tetrahedral sites and the B-ions occupy half of the octahedral sites.They also exhibit interesting magnetic, electronic, and optical properties that depend on the composition, but also on the cation distribution, described by the degree of inversion. [34,6][37][38][39][40] Most of the reported spinel electrocatalysts contain either iron or cobalt as the cation on the B-site.Compared to cobalt, iron is by far the more abundant, less toxic, and cheaper option.[46] The use of Fe as B-cation additionally allows for an exploitation of the synergy between Ni and Fe. [47]n additional interesting property of spinel oxides is the deviation from a normal cation distribution by partial or even complete inversion, in which case A-ions occupy octahedral sites that are exclusively occupied by B-ions in a normal structure and B-ions occupy tetrahedral sites. [34]This does not only affect the physical properties but also the degree of entropy in the system, as the contribution from B-site-occupation is no longer zero. [23][50] While this introduces another parameter for tailoring the properties via the adjustment of the structure, it makes a prediction of entropic stabilization effects highly challenging.
We have previously reported on the synthesis of several spinel oxides, namely NiFe 2 O 4 , [42] MgFe 2 O 4 , [51] and ZnFe 2 O 4 , [49] the synthesis of which is essentially an adaptation of the non-aqueous microwave synthesis of metal oxides reported by the group of Niederberger, employing metal acetylacetonates (acac) and an aromatic alcohol derivate. [52,53]Based on these syntheses that proceed under highly similar reaction conditions, we now herein delineate the versatility of this approach by introducing more than one M(acac) 2 precursor into the synthesis.A wide range of phase-pure earth-abundant spinel ferrites (AFe 2 O 4 ) with compositions ranging from one to seven different A-ions could successfully be prepared via this approach at a synthesis temperature of only 225 °C.All of these medium and high entropy oxides were obtained as nanocrystalline powders, with crystallite sizes of ≈10 nm.Since the crystallinity generally decreased with an increasing number of different A-ions, subsequent heat-treatment at 400 °C in air could increase the crystallinity without having a negative impact on phase-purity.A detailed characterization of their properties in relation to their composition was performed, and they were also employed for the alkaline oxygen evolution reaction (OER).A partial replacement of Fe by Co moreover shows the high versatility of the synthesis that also allows for the simultaneous variation of the B-ion.

Results and Discussion
For the synthesis, the amount of Fe(acac) 3 and the total amount of M 2+ -precursors were kept constant with a targeted yield of 1 mmol MFe 2 O 4 per synthesis.Phase-pure ferrites of Co, Mn, Mg, Zn, and Ni were obtained in agreement to previously reported results. [42,54]While the crystallinity of NiFe 2 O 4 directly after the microwave-assisted synthesis is low and MgFe 2 O 4 is not yet fully condensed, subsequent calcination at 400 °C yields phase-pure, nanocrystalline ferrites with all different A-ions (Figure S1, Supporting Information).In the next step, ternary oxides were synthesized, containing two of these M 2+ -ions in equal amounts.Phase-pure ferrites could be obtained for different combinations of Co, Mn, Mg, Zn, and Ni.Only the addition of Cu as a second A-ion resulted in phase-separation, due to a reduction of copper (Figure S2, Supporting Information), especially if Zn and Mg are used as second M 2+ -ion.
Since NiFe 2 O 4 has shown promise as an efficient electrocatalyst for the OER, before, [42,41] Ni was also included in the synthesis of more complex spinel oxides.The ratio of Ni-precursor was progressively lowered, as Ni 2+ was partially replaced by more and more different M 2+ -precursors.All M 2+ -ions were employed in equal ratios.Spinel ferrites with three up to seven different M 2+ -ions could successfully be obtained, thus reaching compositions that can be considered as "high-entropy" based on the definition of more than five different cations in a sublattice, although true entropic-stabilization is not verified (Figure 1; Figures S3-S6, Supporting Information).Interestingly, the inclusion of Cu is possible for spinels with four or more different M 2+ions, although phase-separation was observed for the ternary oxides.This could be an indication of entropic stabilization.The obtained ferrites (NiCuCoMn)Fe 2 O 4 and (NiCuCoZn)Fe 2 O 4 appear completely phase-pure when characterized with Cu-X-ray diffraction (XRD).Very small reflections for Cu are observed when zooming in on the high-resolution Ag-XRD patterns, which might be an indication that entropic stabilization is not yet entirely sufficient for these oxides.No by-phases are observed after heat-treatment at 400 °C, which supports the assumption of entropy effects stabilizing the structure, although the presence of trace amounts of CuO cannot be excluded (Figure S4, Supporting Information).The synthesis temperature of 225 °C in the microwave is significantly lower, compared to other reported synthesis methods for high-entropy spinel oxides, in which commonly temperatures ≈1000 °C are required. [55,6,7]The assumption of entropic stabilization could be further cemented by reversible demixing upon calcination -an experiment wellknown for HEOs with a rocksalt structure. [5]Calcination of assynthesized (NiCoMnCuZnMg)Fe 2 O 4 at 750 °C for 5 h results in a demixing and the appearance of reflections for a Fe 2 O 3 sidephase.This transition is driven by enthalpic contributions since the mixing enthalpy favors phase segregation.Subsequent calcination for 5 h each at 1000 and 1200 °C progressively leads to a transition back to a phase-pure spinel structure (Figure 1c).This observation is an indication for entropic stabilization of the structure, while at the same time demonstrating that the microwaveassisted approach is able to yield a kinetically-controlled product that is not the thermodynamically most favored one at the given synthesis temperature of 225 °C.
Raman spectroscopy is a useful tool to further evaluate phasepurity, since the shift and intensity of observed vibrational modes depend on the structure, as does the sensitivity to Raman scattering.Thus, highly Raman-active phases might be observable, which may have too low concentrations to appear in powder XRD patterns.Additionally, the characteristic bands for Ramanactive vibrations in the spinel structure depend on the degree of inversion, since occupation of one crystal site by more than one different cation will result in a splitting of the observed modes. [34]Some of the herein-employed M 2+ -ions favor the inverse spinel structure, as in the case of NiFe 2 O 4 , while others, such as ZnFe 2 O 4 , almost exclusively occur in the normal spinel structure (Figure 2). [34]n many cases, the actual structure is somewhere between the ideally normal and the completely inverted structure.The degree of inversion significantly influences optical and electronic  and c) ferrites with increasingly more M 2+ -ions.d) The influence of calcination on the cation distribution is elucidated exemplary for ferrites containing four different M 2+ -ions.Approximate positions for a normal spinel structure are shown in blue, those for an inverse structure in grey. [34].
properties of the material.In the special case of many different metal cations in the system, the possibility for site-exchange with the Fe 3+ ions also affects the entropy, since a distribution of the M 2+ -ions over the B-sites significantly increases the configurational entropy, since it is no longer restricted to only one sublattice. [17]Such an effect has already been reported by Navrotsky et al. in the 1960s for binary spinels, in which partial inversion results in an increase in the configurational entropy that in turn affects the entropy of formation. [56,57]This can be seen by regarding the equations for configuration entropy in an oxide, not only considering the entropy in one sublattice, as frequently done, but the sum of all lattice contributions together.
In a binary oxide, S config can be calculated as [23] : x a ln x a Assuming an absence of oxygen defects and a spinel structure of AB 2 O 4 , this simplifies to: If the B-site was occupied solely by iron, this would further result in: If x b , that is, the relative mole fraction for each different ion on the B-site was no longer 1 -as in the case for only Fe on the B-position -the contribution of the B-sublattice to the configurational entropy would no longer be zero.Since the preferred site depends on the nature of the ion, the overall configurational entropy does not only depend on the number of different employed ions, but also on their nature.While an accurate determination of the degree of inversion based on Raman measurements is difficult due to small differences in the observed Raman shifts and significant absorption of the incident laser light (red) by the dark-brown ferrites, a qualitative assessment of the cation distribution might be attempted.
Due to the relatively low crystallinity directly after the microwave synthesis, we initially focussed on the spectra of calcined samples.By regarding the A 1g (1) and A 1g (2) bands between 650 and 700 cm −1 and the T 2g bands between 600 and 300 cm −1 , it can be inferred from the spectra of binary spinels that CoFe 2 O 4 , NiFe 2 O 4 , and MgFe 2 O 4 are mostly inverse, whereas ZnFe 2 O 4 and MnFe 2 O 4 appear to be mostly normal, in good agreement to literature. [57,34,58,59]The A 1g bands are thereby associated with M─O stretching bonds in tetrahedral sites, whereas the T 2g bands arise from asymmetric stretching and bending M─O vibrations at octahedral sites. [34,59]Occupation of more than one cation at one site results in peak-splitting, since a different cation mass and different M─O bond strength result in a shift of the vibration frequency. [34]The spectra are fit with Gaussian-Lorentzian peak shapes (GL( 50)), assuming one peak for each, A 1g , T 2g (3), T 2g ( 2), E g and, if visible, one for T 2g (1), each for a normal spinel structure and peak splitting into two modes for the inverted structure, due to tetrahedral and octahedral sites both being occupied by two different ions (Figure S7). [34]From all samples, MnFe 2 O 4 shows the most normal spinel structure, with shoulders between 250 and 400 cm −1 , that is, around the region expected for the T 2g (2) peak in a normal spinel structure -indicating a low degree of inversion for ZnFe 2 O 4 .CoFe 2 O 4 is the most inverted spinel, according to the Raman fits, as indicated by the highest T 2g (3) to A 1g ratio -a similar observation has been made in literature. [59,60]he same general result is obtained by Rietveld refinement of the Ag-XRD patterns of the binary ferrites (further discussion in the following), which yields a degree of inversion of 0.91 for CoFe 2 O 4 , 0.95 for NiFe 2 O 4 , and 0.92 for MgFe 2 O 4 , while for ZnFe 2 O 4 a degree of inversion of 0.32 is obtained.No accurate fit of the cation distribution was possible for calcined MnFe 2 O 4 , likely due to the higher amount of amorphous phases that precede decomposition products, alongside the very low degree of inversion and highly similar atomic form factors of Mn and Fe.All values are given after calcination.When adding a second A-ion to the synthesis, still a significant dependence of cation distribution on the nature of the employed ion can be observed (Figure 2).While the spectra of (NiMn)Fe 2 O 4 and (NiZn)Fe 2 O 4 closely resemble that of NiFe 2 O 4 for an inverse spinel, those of (ZnMg)Fe 2 O 4 and (MnZn)Fe 2 O 4 are much more similar to that of ZnFe 2 O 4 , implying a normal spinel structure, with (MnMg)Fe 2 O 4 lying somewhere in between.This is an indication that Ni has a dominant influence in realizing an inverse spinel structure, whereas Zn pushes for a normal structure.A similar effect was reported by Jadhav et al, who observed a change from a normal towards an inverse spinel structure for (Ni x Zn 1-x )Fe 2 O 4 , with an increasing Niratio. [61]Generally, the bands are very broad, and even in the Nicontaining compositions, an additional shoulder slightly above 600 cm −1 is present.This is due to the fact that even in a normal spinel structure, two different cations now reside on the A-site, which will always result in a splitting of active modes at least for one site.This fact makes a differentiation between more normal or more inverse structures almost impossible for more extensively substituted compositions.As apparent in Figure 2c, bands become very broad with an increasing number of cations and almost all spectra have a prominent feature between 450 and 500 cm −1 , which would correspond to the T 2g (2) in an inverted binary spinel.This supports increasing cation disorder -also on the octahedral sites.
Another interesting difference between the spectra is the relative intensity of the T 2g (2)/T 2g (3) band between 400 and 500 cm −1 , which can be attributed to asymmetric bending and stretching of metal-oxygen bonds at octahedral (B)-sites. [34]The intensity is low for MnFe 2 O 4 with Mn 2+ in d 5 high-spin configuration and especially pronounced in CoFe 2 O 4 with Co 2+ in d 7 configuration, and thus depends on the occupation of d-orbitals with electrons.For Ni 2+ the intensity is slightly lower again, likely due to an equal distribution of electrons at least over the e g orbitals.From this observation it can be assumed that samples featuring a high intensity of this T 2g band likely contain larger amounts of Co 2+ and Cu 2+ on octahedral sites and are thus at least partially inverted.This is the case for almost all compositions shown in Figure 2c.As mentioned above, the partial inversion of the structure increases the disorder in the structure and thus the entropy.
Subsequent calcination increases the crystallinity and thus the intensity of the observed bands (Figure 2d; Figure S7, Supporting Information).Additionally, the relative intensities of the bands change, which is especially prominent for (NiMnZnMg)Fe 2 O 4 , for which the spectrum of the as-synthesized sample shows a closer resemblance to that of a normal spinel structure, whereas the one calcined at 400 °C is very similar to that of an inverted spinel.This could on the one hand be an indication of a change in the cation distribution, which is at first partly determined by kinetics in the synthesis, but on the other hand it could also elucidate the increasing influence of the entropic contribution to ΔG at higher temperatures and underlines an increased stabilization of the inverted structure compared to the normal one.Now having established macroscopic phase-purity, we wanted to confirm the uniform distribution of the elements in the structure.We performed scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDX) analysis of various samples (as-synthesized), including elemental mapping for several spinel ferrites with four or more different M 2+ -ions employed in the synthesis.All of the synthesized ferrites are nanoparticles that form larger agglomerates (Figure 3; Figures S8-S13, Supporting Information).
A slight decrease in the particle diameter from slightly above 20 nm for (NiCo)Fe 2 O 4 , down to 15-20 nm in the four M 2+ -ions containing spinel (NiCoMnCu)Fe 2 O 4 to ≈10 nm in the six M 2+ions containing (NiCoMnZnMgFe)Fe 2 O 4 is observable, but the size is very small in all cases.The relative element ratios were determined by analyzing at least two different point areas and are listed in Table S1 (Supporting Information).The relative ratios of the employed M 2+ -ions are in all cases highly similar.Only the concentration of Mg seems to be a little lower, which is in agreement to previous results for MgFe 2 O 4 , in which the employed Mg was not completely incorporated into the structure. [51]n some cases, the concentration of Cu appears to be a little higher compared to that of the other M 2+ -elements.This might, however, be partially an effect of overlap of the Cu-K  peak with the Pt-L  peak.Pt was used as a coating in the sample preparation to prevent charging effects.The M 2+ /Fe ratio is between 0.4 and 0.5 for most samples, independent of the number of employed M 2+ -ions.This is slightly lower than the ideal ratio of 0.5.A major problem in the quantification of elements is the rather similar energy of the respective K  -irradiation, which results in an overlap of peaks in the EDX due to insufficient resolution.This results in significant uncertainties in the absolute contributions, especially for combinations of Co and Ni, or worse for Co and Fe.In the latter case, it is especially difficult to distinguish between the Co K  and Fe K  lines.In the EDX maps of all ferrites with four or more different M 2+ -ions, a homogeneous distribution of the elements over the entirety of the analyzed area is observed (Figure 3, Figures S8-S13, Supporting Information).However, to eliminate any doubts about the apparent distribution due to peak overlap, we performed wavelength dispersive X-ray spectroscopy (WDX) analysis for (NiCoMnCuZnMg)Fe 2 O 4 .This allows for an accurate separation between the Co K  and Fe K  line (Figure S14, Supporting Information).
X-ray photoelectron spectroscopy (XPS) survey scans of spinel ferrites with different compositions confirm the presence of all employed elements, not only in the bulk, but also on the surface (Figure 4; Figure S15, Supporting Information), which is important for, for example, catalytic applications, in which reagent adsorption, activation and conversion are all surface mediated pro-cesses.An accurate quantification of the elements present is difficult since a strong overlap with often more than one Auger signal is unavoidable when several transition metals are present in the structure.This, together with low concentrations and organic residues from the synthesis, seriously interferes with quantification attempts.Additionally, the Mg 1s and Mg 2s signals were so weak as to be hardly visible at low concentrations.Therefore, the Mg KLL line at 302 eV was used for quantification.Additionally, we measured high-resolution scans from 30 to 150 eV, which allows for a reasonable resolution of the M 3p and 3s peaks.These we used for the quantification since they do not overlap with Auger lines and in addition to that can be fitted with one peak instead of complex multiplets.Exemplary fits are shown in Figure 4c.The quantification results are summarized in Table S2 (Supporting Information).While the A/Fe ratio is ≈0.5 for most samples, some deviate considerably.This is most likely due to the low intensity of the M 3s and 3p peaks, possible uncertainties in the relative sensitivity factors (RSFs) for these rarely used signals and significant peak overlap.The M/O ratio is significantly lower than the ideal value of 0.75, probably due to large amounts of organic residues in the as-synthesized samples.Therefore, the XPS results have to be solely seen as a qualitative evaluation of the composition.
Using alcohols as a solvent, redox-reactions can occur during the synthesis. [53]M 2+ -ions might be further oxidized or reduced, while M 3+ -ions, in this case Fe 3+ , could be partially reduced, which is especially relevant for Ni, Mn, Co, Fe, and Cu, since they can occur in multiple oxidation states.Unfortunately, due to the interference of Auger lines and low signal to noise ratios, a quantification of redox species in compositions containing several 3d transition metals is not possible.However, high-resolution measurements of the Fe 2p signal for several ternary ferrites and (NiCoMnCuZnMg)Fe 2 O 4 are highly similar and can be fitted with the constraints for Fe 3 O 4 (Fe 3+ contribution) proposed by Biesinger et al. (Figure S15, Supporting Information), supporting the 3+ oxidation state of Fe. [62] Small differences in the peak shape between samples can be expected due to differences in the degree of inversion and different Auger contributions. [63,64]Additionally, the presence of the Co L 3 M 45 M 45 line at 713 eV is clearly visible especially for (NiCo)Fe 2 O 4 .
For Ni, high-resolution spectra of the Ni 2p signal of the binary oxides (NiCo)Fe 2 O 4 and (NiZn)Fe 2 O 4 were measured in a comparison to that of (NiCoMnCuZnMg)Fe 2 O 4 .All spectra are highly comparable and strongly resemble that reported for NiFe 2 O 4 .An exemplary fit of the Ni 2p 3/2 peak of (NiZn)Fe 2 O 4 , with the multiplet splitting reported by Biesinger et al. can be found in Figure S15 (Supporting Information). [62]Therefore, Ni predominantly occurs in the oxidation state of +2 also in the presence of various additional M 2+ ions.For Cu on the other hand, a partial reduction to Cu + and Cu 0 was observed, in agreement to literature observations of Cu being prone to reduction when alcohols are used as a solvent. [53,48]For Mn, the Mn 2p 3/2 spectra for (MnMg)Fe 2 O 4 and (MnZn)Fe 2 O 4 allow for a reasonably good fit with the multiplet parameters from Biesinger et al. for MnO. [62]Small amounts of a reduced manganese species appear to be present in (MnZn)Fe 2 O 4 .The Mn 2p peak shape of (NiCoMnCuZnMg)Fe 2 O 4 is generally similar, however, a strong Ni LMM and Cu LMM peak prevent accurate fitting.For Co the spectral overlap with the Fe LMM signal was too strong to allow for a meaningful fit.The Zn 2p 3/2 spectrum of (NiCoMnCuZnMg)Fe 2 O 4 is again highly similar to that of the binary ferrites.A slight shift toward lower binding energy is ob-served compared to (MnZn)Fe  [65] Reasons for the slight peak shifts might be the possible presence of low amounts of a reduced Zn species, or differences in the degree of inversion, since electrons in Zn 2+ in tetrahedral and octahedral environment might have slightly different binding energies. [63]From these analyses, it can be inferred that the oxidation state in the high-entropy spinel ferrites is very similar to the oxidation state in binary ferrites and thus not significantly influenced by the mixing with other M 2+ -ions.Some of the employed M 2+ -ions (especially Cu) in the structure might occur in a slightly reduced state, however, all ions are mostly in the target oxidation state also employed in the metal precursor.
Since SEM with EDX only gives insights into the element distribution over a larger area and XPS is a surface sensitive technique, the homogenous distribution of all employed elements in the structure was verified with high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) analysis of (NiCoMnCuZnMg)Fe 2 O 4 (Figure 5).The crystallization in a spinel structure was confirmed.Additionally, all elements are present in each single particle, proving that they are truly incorporated into the structure.The observed element ratios are with 61.5:23.9:0.9:1.5:2.3:2.1:5.5:2.3 for O:Fe:Mg:Mn:Co:Ni:Cu:Zn in very good agreement to the results obtained from EDX and XPS, further elucidating homogeneity on a nanometre scale as well as across multiple particles.The amount of copper is slightly overestimated, due to EDX signals originating from Cu parts of the STEM holder.The same homogeneity was observed if the mapping was performed over more particles together (Figure S16, Supporting Information).Having examplified the homogeneous elemental distribution in (NiCoMnCuZnMg)Fe 2 O 4 , we further demonstrated the  After having established phase-purity of the synthesized oxides, we took a closer look at their properties, especially with regards to structure-property relationships.Rietveld refinement of the Ag-XRD patterns was performed in order to gain insights into influences of the elemental composition on crystallite size, lattice parameter/ cell volume and the presence of microstrain.While the first one is expected to influence the surface area and the number of active catalytic sites, the latter two affect the electronic properties.Especially strain is a factor commonly observed in high-entropy alloys, [66,3] but also reported for metal chalcogenides. [26,21]In mixed transition metal highentropy spinel oxides it was observed that the lattice parameters decreased upon replacement of either Ni or Co by Zn, due to the weaker electronegativity. [6]irst, crystallite sizes were determined from the (400) reflection in Cu-and Ag-XRD patterns by the integral breadth method, that is, by evaluating the height of a reflection in relation to its area.Thus, a first impression of crystallite sizes could be gained.Due to the broadness of the reflections, only the (400) reflection was used for the first size approximation.The determined sizes from both patterns for each composition were in good agreement, with the one derived from the Ag-XRD pattern being slightly smaller, due to pronounced asymmetry in the reflections that stem from the measurement configuration, in addition to a lower signal to noise ratio in the Cu-XRD patterns.Instrument broadening was not subtracted for this first estimation of crystallite sizes.For all synthesized spinel oxides, the determined crystallite sizes from Cu-XRD patterns are ≈6 nm for as-synthesized and ≈7 nm for calcined spinel oxides.From Ag-XRD patterns, the calculated sizes are ≈5 nm for as-synthesized and ≈6 nm for calcined spinel oxides (Figure S19, Supporting Information).Crystallite sizes determined by Rietveld refinement are in generally good agreement to those calculated with the integral breadth method (Figure 6).Apart from MgFe 2 O 4 , which is mostly amorphous directly after the synthesis and still not very crystalline after calcination at 400 °C, all binary spinel oxides have similar crystallite sizes of ≈6 to 7 nm.Especially for the as-synthesized samples, spinels with more than one M 2+ion have slightly lower crystallite sizes of ≈5 nm.There is, however, also a noticeable dependence of the size on the composition.While spinels that do not contain Cu show an almost constant decrease with the number of different M 2+ -ions, the crystallite sizes of those spinels with Cu are generally larger, probably due to local heating upon microwave absorption and partial reduction of Cu.As-synthesized spinel oxides with high relative amounts of nickel are also smaller, in agreement to the small-crystallite size of NiFe 2 O 4 directly after the synthesis (Figure 6b).From a comparison with SEM and STEM images, it can be inferred that the nanoparticles are mostly single-crystalline.No significant difference in the fitted strain is observed, although a slight increase from spinels with four different M 2+ -ions to seven M 2+ -ions is observable (Figure S20, Supporting Information).
The lattice parameter is ≈8.4 Å for all ferrites (Figure 6).It slightly decreases upon calcination, possibly due to cation redistribution toward a thermodynamically more favored structure, as observed in Raman measurements, in addition to a further condensation of the structural network, and increase in the crystallinity.Furthermore, re-oxidation of partially reduced species might play a part.The determined lattice parameters for ferrites containing only one M 2+ -ion fit well to literature reports (compare ICDD reference files employed for phase identification).The trend in determined lattice parameters moreover follows the trend in ionic radii, with the values for sixfold coordination reported by Shannon used for the correlation (Figure S21, Supporting Information). [67]For ternary and quarternary ferrites, those containing more Zn and/ or Mn possess a larger lattice parameter, in agreement to the larger ionic radius, whereas those containing more Co and Ni possess smaller lattice parameters.A decrease in the lattice parameter has also been observed by Jadhav et al. upon Zn substitution by Ni in mixed Zn x Ni 1-x Fe 2 O 4 , [61] while Ansari et al. reported on an increasing lattice parameter upon Co substitution by Mn. [68] All in all, the largest lattice parameter is obtained for calcined ZnFe 2 O 4 , the smallest for calcined MnFe 2 O 4 and NiFe 2 O 4 , with all medium and high entropy spinels having intermediate lattice parameters.This is reflected by a convergence of the maximum of the main reflection (311) ≈12.62°2  for high-entropy spinel ferrites (Figure S23, Supporting Information).The fits are depicted in Figures S22-S25 (Supporting Information).
In agreement to the highly similar crystallite size, with a small reduction in size upon the addition of a multitude of different ions, the BET surface area is highly similar for all spinel ferrites of different compositions (Figure 6; Figure S26 and Table S3, Supporting Information).Interestingly, the BET surface area decreases almost consistently for all spinel ferrites upon calcination at 400 °C, although the crystallite size decreases in some cases.The BET area is, however, also influenced by the organic residues adhering to the particle surface.Those amounts are often non-negligible, as observed by XPS and DRIFT analysis and significantly reduced upon calcination (Figure S27, Supporting Information).
The band gap is another important material property that is very sensitive to the composition.Optical absorption in spinels is generally complex, since several transitions might occur, such as ligand to metal charge transfer (LMCT), intersublattice charge transfer (ISCT) between ions at octahedral and tetrahedral sites, intervalent charge transfer (IVCT) between M 2+ and Fe 3+ and crystal field (CF) transitions between d-orbitals of one type of ion at the same site. [34,69]For the latter, relatively small excitation energies are sufficient.Therefore, CF transitions commonly occur in the NIR range, whereas the other three types of optical transitions fall into the UV and visible light range.The main absorption feature can be ascribed to LMCT transitions in normal spinels, while for inverse spinels ISCT and IVCT additionally occur.Absorption features between 2 and 3 eV and those above thus commonly involve mixtures of LMCT, ISCT, and IVCT, the character and ratio of those depends on the M 2+ -ion. [69][72] In this case, the indirect band gap transition below 2 eV (correlated to an absorption feature at ≈2.3 eV) is ascribed to excitation in the minority spin channel, whereas the optical band gaps at 2.29 and 2.77 eV in NiFe 2 O 4 are ascribed to transitions in both minority and majority channels. [70,73]The observed spectra are likely a combination between localized optical transitions and fundamental conduction to valence band transitions.
Many of the examined ferrites exhibit an optical absorption feature at ≈2.5 eV (Figure 7; Figures S28-S30, Supporting Information), which corresponds to Fe-based ISCT from Fe 3+ in T d to Fe 3+ in O h sites, or to M 2+ to Fe 3+ IVCT in inverse spinels.Whether the M 2+ (O h ) to Fe 3+ (O h ) IVCT transition, or the Fe 3+ (T d ), Fe 3+ (O h ) ISCT is of lower energy, depends on the M 2+ cation. [73]For normal spinels LMCT to Fe 3+ in O h sites can occur in this energy range. [74]For the binary spinel ferrites absorption in this range is the weakest for ZnFe 2 O 4 , in agreement to the mostly normal structure that prevents IVCT and ISCT involving either Fe 3+ in T d , or M 2+ in O h configuration (Figure 7).A second ISCT transition from Fe 3+ (T d ) to Fe 3+ (O h ) appears above 3 eV.The two separate bands are due to crystal field splitting of Fe 3d orbitals and transitions into t 2g and e g orbitals, respectively, and show a characteristic separation of ≈1.4 eV. [73,69]The features below 1 eV and at ≈2.5 eV are most prominent in the indirect Tauc plot and thus correspond to indirect transitions, while the one between 3.7 and 4 eV is most dominant in the direct Tauc plot, in agreement to literature. [70,71]The possibility of IVCT at octahedral sites for inverted spinels is probably the reason for the larger apparent direct optical band gap of ZnFe 2 O 4 .Specifically, the direct band gap decreases from 2.9 eV in ZnFe  [61] The d-d transitions below 1.4 eV (1000 nm) are much more pronounced in the as-synthesized samples compared to the calcined ones, possibly due to more Fe 3+ with five unpaired d-electrons in octahedral sites (lower degree of inversion).Additionally, symmetry distortions might be healed to some extent during calcination, which would result in less d-d transitions being possible, since they are normally Laporte forbidden.The same effect is observed in medium and high-entropy ferrites (Figure 7).
The band gap values derived from Kubelka-Munk plots decrease slightly upon calcination for some of the ferrites, while for others a slight increase is observed (Figure 8; Table S4,

Supporting Information).
][77] On the one hand, an increasing crystallinity has an effect, as especially apparent in the spectra of NiFe 2 O 4 .On the other hand, cation redistribution during calcination might affect the relative intensity of the CT-bands.The differences in apparent band gaps of as-synthesized and calcined samples are very small in all cases.For those ferrites with strong d-transitions in the form of ISCT and IVCT, mostly the transition at ≈2.5 eV (overlapping with LMCT transitions) determines the apparent band edge, while for others, such as ZnFe 2 O 4 , the main contribution arises from LMCT, resulting in the significant difference in the optical band gap.
All medium-and high-entropy ferrites show strong contributions from d-d transitions, especially after calcination, which dominate the UV/vis spectra.They do not necessarily represent electronic transitions from valence to conduction band, though.
The difference between the direct band gap at ≈2.2 eV and the indirect band gap at ≈1 eV is with 1.1-1.2eV (Figure 8) slightly lower, but still close to the expected 1.3 eV for crystal field splitting in Fe 3+ O h . [73]This confirms that d-d charge transfer transitions dominate the optical properties of the simple and the highentropy ferrites.The low energy of CF transitions in all ferrites is the reason for the dark brown to black color of the nanoparticulate powders.The deviation from 1.3 eV is probably due to an overlap between ISCT and IVCT and depends on the cations.Since Fe 3+ is predominant, variations are mainly caused by the different M 2+ -ions.
NiFe 2 O 4 and CoFe 2 O 4 are both very efficient earth-abundant electrocatalysts for the OER. [41,42,45,46]Both are very similar in structure, but vary in regard to the M 2+ -ion.In order to explore the influence of the nature and diversity of the M 2+ -ions, the OER performance of various high-entropy ferrites was tested and compared to that of NiFe 2 O 4 and CoFe 2 O 4 (Figure 9).When combining Co and Ni as M 2+ -ions, the activity could be significantly improved and the overpotential decreased, especially at higher currents (Table S5, Supporting Information), in agreement to the findings of Chakrapani et al. for Ni-substituted CoFe 2 O 4 , [78] but in contradiction to Maruthapandian et al. for Co-substituted NiFe 2 O 4 . [79]An improved conductivity has been reported for mixed Co-Ni ferrites, that might contribute to the improved OER activity. [80]The combination of two different M 2+ -ions can thus already improve the activity of some highly promising earth abundant electrocatalyts for the OER.
The addition of further M 2+ -ions, however, resulted in a decreased activity.The same trend is observed for the ferrites with four different M 2+ -ions (Figure 9).Reducing the content of Ni and Co further in the ferrites containing five and more different M 2+ -ions, leads to a further decrease in the activity.Noticeably, the high-entropy ferrites with six and seven M 2+ -ions exhibit a slightly improved performance again, although the concentration of Ni and Co are reduced even further.Still, such entropy effects are insufficient to match the high-activity of ferrites with a higher ratio of Ni and Co.Our results show that the OER activity is more dependent on the M 2+ -ion species than of the number of different ions.
The OER activity was tested for both as-synthesized and calcined spinel ferrites.Interestingly, the as-synthesized spinels ex-hibit a higher activity compared to the calcined ones, which is in contrast to previous experiments with NiFe 2 O 4 , [42] but demonstrates that no calcination is required in order to obtain active electrocatalysts, which reduces the energy requirement for the synthesis significantly.The higher activity in as-synthesized ferrites might on the one hand be an effect of the small particle size and the presence of organic residues, both of which improve the dispersibility and ink homogeneity.Furthermore changes in the coordination environment of active elements might influence the performance.The lower activity compared to literature results are likely due to the employed carbon paper electrode, which resulted in lower currents, compared to glassy carbon, or nickel foam (Figure S31, Supporting Information).However, the carbon paper has the advantage of not being active in the OER, in contrast to Ni-foam, which was found to be very active, and of the ferrite ink adhering well to the electrode substrate -much better compared to glassy carbon.A decrease in the performance was still observed with increasing linear sweep voltammetry (LSV) scan number (Figure S31, Supporting Information), which is likely due to a gradual removal of organic residues, but to some extent perhaps also due to powder detachment.It is also apparent for the high-entropy ferrites (Figure S32, Supporting Information) which initially show a better performance.The relative activity depending on the composition, remains, however, the same.
To test the hypothesis of too little concentration of active elements in the high-entropy spinel oxides, four additional ferrites were synthesized, in which half of the Fe 3+ was replaced by Co 3+ .Some Cu-byphases were present in ((NiCoCuZn  [81] This is likely due to redox reactions that are known to occur under the employed conditions, that is, heating in the presence of an oxidizable solvent. [53]Additionally, the presence of Co in a more oxidized state was verified by a shift of the Co L 3 -edge toward a higher loss energy. [82]s these AFe 1 Co 1 O 4 spinel oxides increase the Co content by 33%, they clearly elucidate the reason for the improved OER performance.In addition, the presence of Co 3+ can significantly improve the performance also compared to CoFe 2 O 4 , where Co occurs as Co 2+ .Therefore, for high-entropy spinels, higher OER activity can be achieved at the price of more Co and Ni contents -and thus also higher material costs, lower sustainability, and abundance.The Fe content in these spinels (67% in AFe 2 O 4 , 33% in AFe 1 Co 1 O 4 ) is still higher than those employed in literature with equimolar ratios of all cations (20% in (MnFeNiMgCr) 3 O 4 ), for which an improved OER activity was reported. [55]

Conclusion
High-entropy spinel ferrites with a large variety of different M 2+ions were successfully synthesized.A synthesis temperature of 225 °C and a time of 30 min in the microwave was sufficient for the preparation of phase pure high-entropy spinel oxides.These conditions are significantly milder compared to others reported in the literature.Additionally, the large variety of different ferrites synthesized elucidates the versatility of this approach.Due to the low relative ratio of active elements, such as Co and Ni in the structure, the high-entropy spinel ferrites (MFe 2 O 4 ) were less active compared to NiFe 2 O 4 or CoFe 2 O 4 in the OER.The most active ferrite was found to be (NiCo)Fe 2 O 4 , which is thus a highly promising earth-abundant OER electrocatalyst.Partial replacement of Fe 3+ by Co 3+ resulted in a significantly improved activity exceeding that of either NiFe 2 O 4 or CoFe 2 O 4 .This result not only proves that the synthesis allows for further variations also on the M 3+ -ion site, but also paves the way for the design and synthesis of highly active novel high-entropy spinel oxide electrocatalysts under mild conditions.

Experimental Section
Synthesis: For the synthesis of 0.5 mmol of AFe 2 O 4 , 353.2 mg (1 mmol) of Fe(acac) 3 (Acros Organics) was dissolved in 15 mL of rac-1-phenylethanol (Sigma-Aldrich) together with equal molar ratios of M(acac) 2 , with the sum of all M(acac) 2 equalling to 0.5 mmol.The solution was transferred to a 30 mL borosilicate microwave vial and either heated as fast as possible -or, if high amounts of Cu(acac) 2 were employed, with reduced power -to 225 °C under stirring in a microwave reactor (Anton Paar Monowave 400) equipped with a MAS24 autosampler.The solution was kept at the target temperature for 30 min, before being cooled as fast as possible to 55 °C, using compressed air.The obtained product was precipitated with n-pentane and subsequently washed thrice with acetone/water and once with diethyl ether, before finally being dried at 80 °C overnight.A portion of the as-synthesized product was further calcined at 400 °C for 5 h in air (Nabertherm furnace) to improve crystallinity and remove organic residues.
Characterization: Powder XRD was performed on a Malvern PANalytical Empyrean diffractometer, using Cu K  irradiation.A spinning sample holder in Bragg-Brentano geometry was used.Additionally, selected samples were measured on a STOE STADI P Mythen2 4 K diffractometer, using Ag K  irradiation and Hilgenberg capillaries (0.5 mm outer diameter).The instrument was equipped with a Ge(111) monochromator and four Dectris MYTHEN2 R 1 K strip detectors. [83]Rietveld refinements were performed using FullProf. [84]A Thompson-Cox-Hastings pseudo-Voigt function was used for peak shape modeling. [85]Refinements were based on the structure reported by Mahmood et al. for cubic CuFe 2 O 4 , changing the A ion to either Co, Ni, or Mn, as appropriate and calculated absorption correction for NiFe 2 O 4 . [86]For ferrites containing only one A-ion, lattice parameters in the starting conditions and absorption correction were adapted.Refined parameters include instrumental zero shift, background, FWHM parameters, mainly U and Y, lattice parameters, isotropic B, scale, source width/detector distance, size anisotropy (spherical harmonics) and, if required, asymmetry parameters.Occupation refinement was only attempted for structures with one A-ion.Instrumental broadening was determined with LaB 6 (NIST SRM 660c) as a standard.Size was calculated by FullProf from the size parameters contributing to the FWHM, that is, IG for the Gaussian contribution, as well as Y and Sz for the Lorentzian contribution.Strain was calculated from the strain parameters, that is, mainly U for the Gaussian contribution and X for the Lorentzian contribution.
UV/vis/NIR spectra were obtained on a Perkin Elmer Lambda 750 spectrometer, equipped with a Praying Mantis (Harrick) and using spectralon as a white standard.DRIFT (Diffuse Reflectance Infrared Fourier Transformed) spectra were measured on a Bruker Alpha II spectrometer.Raman measurements were conducted on a Horiba Yvon Raman microscope, equipped with a 11.5 W He-Ne laser ( = 633 nm).The laser intensity was reduced if required to avoid extensive sample heating and undesired interference from fluorescence.Casa XPS 2.3.25 was used for spectra fitting, assuming Gaussian-Lorentzian peak shapes (GL( 50)).Nitrogen physisorption measurements were performed on a Quadrasorb Evo and a Nova 800 device from Anton Paar QuantaTec at 77 K. Samples were degassed for 12 h at 120 °C prior to the measurements.The surface area was evaluated with the software ASiQwin using the Brunauer-Emmet-Teller (BET) model.XPS was performed on a VersaProbe III Scanning XPS Microprobe (Physical Electronics PHI), using monochromated Al K  irradiation.Beam voltage and X-ray power were set to 15 kV and 25 W, respectively, and the spot size was 100 μm.Wide-range survey scans were measured with a step size of 0.4 eV, a pass energy of 224 eV and a time per step of 50 ms.High-resolution short-range survey scans and conventional high-resolution spectra were measured with a step size of 0.1 eV, a pass energy of 26 eV and a time per step of 50 ms.Samples were continuously flooded with electrons and Ar + ions at low energy.CasaXPS 2.3.25 was used for data evaluation, using Shirley or linear backgrounds and LA(1643) (survey) or GL(30) (high-resolution) line shapes. [87]Binding energies were corrected using the C 1s (C-C, C-H) peak as reference at 284.8 eV.
SEM and EDX were performed on a Zeiss Leo 1530 device additionally equipped with an ultra-dry EDX detector from Thermo Fisher Scientific NS7.Samples were sputter coated with platinum (Cressington Sputter Coater 208 HR).An acceleration voltage of 3 and 20 kV was used for SEM and EDX, respectively.For the WDX analysis, a Zeiss Ultra plus electron microscope and a Magna Ray WDX-spectrometer from Thermo Fisher Scientific NS7 were used with the acceleration voltage set to 15 kV.
STEM was performed on a Titan Themis microscope operated at 300 kV.Nanoparticle samples were drop cast onto a gold grid with lacey carbon.STEM-EDX spectrum imaging was acquired using a SuperX detector covering 0.7 sr collection angle.STEM-EELS spectrum imaging was collected using a Quantum ERS spectrometer.Multivariate statistical analysis was conducted to denoise the spectrum imaging datasets and detect for potential inhomogeneity in the elemental distribution. [88] three electrode H-cell setup was used for electrochemical measurements.A platinum counter electrode, a reversible hydrogen electrode reference electrode (Gaskatel), a Selemion AMV-N anion-exchange membrane (AGC group) and either a Parstat 3000A-DX potentiostat (Princton Applied Research) and the software VersaStudios, or a Gamry Reference 3000 potentiostat and the software Gamry Frameworks were used, respectively.iR compensation for measurements with the Parstat potentiostat was done via EIS measurements, while the current interrupt method was used for correction with the Gamry potentiostat.The internal resistance was measured prior to the electrochemical characterizations and was very small.All reported potentials are compensated, if not otherwise noted. 1 m KOH was employed as electrolyte that was continuously purged with Ar.For the preparation of the working electrode, 10 mg of the respective sample were dispersed in 300 μL of i-propanol (p.a.) and 20 μL of a 5 wt.%Nafion solution (Alfa Aesar) using ultrasonication.50 μL of the dispersion were drop-cast onto carbon paper (Freudenberg H2315-C2) with the exposed surface area being restricted to 1 cm 2 using Kapton tape.25 LSV scans were recorded at a sweep rate of 5 mV s −1 after three initial CV scans.

Figure 2 .
Figure 2. a) Raman measurements for calcined binary ferrites, b) ternary ferrites,and c) ferrites with increasingly more M 2+ -ions.d) The influence of calcination on the cation distribution is elucidated exemplary for ferrites containing four different M 2+ -ions.Approximate positions for a normal spinel structure are shown in blue, those for an inverse structure in grey.[34].

Figure 3 .
Figure 3. SEM images of four exemplary spinel ferrites with 4 to 7 different M 2+ ion and EDX mapping of (NiCoMnCuZn)Fe 2 O 4 , elucidating the homogeneous distribution of all elements in the structure.

Figure 4 .
Figure 4. a) XPS survey scans of medium-and high-entropy ferrites, b) high resolution survey scans of the lower binding energy region used for quantification, and c) exemplary fit of such spectra.

Figure 5 .
Figure 5. HAADF-STEM of (NiCoMnCuZnMg)Fe 2 O 4 together with EDX, highlighting the homogeneous distribution of elements on the nanometre scale across multiple nanoparticles.The corresponding EDX spectrum is shown in Figure S16 (Supporting Information).

Figure 6 .
Figure 6.a) Crystallite size depending on the composition determined by the integral breadth method and Rietveld refinement from Ag-XRD patterns, b) BET surface area of selected medium-and high-entropy ferrites, c) crystallite size and composition for selected samples showing the general decrease in crystallite size of compositionally related ferrites with a larger number of M 2+ -ions in the structure, and d) refined parameters for selected samples depending on the composition.As-synthesized samples are depicted in a darker color, whereas calcined ones are shown in paler color.

Figure 7 .
Figure 7. a) Kubelka-Munk plots, b) direct Tauc plots and c) indirect Tauc plots of as-synthesized spinel ferrites (top) and d) Kubelka-Munk, e) direct Tauc and f) indirect Tauc plots for calcined ones (bottom).Sharp features at 319 nm (3.9 eV) result from a measurement artefact of the used spectrometer.
2 O 4 and 2.4 eV in MgFe 2 O 4 to 2.0-2.1 eV in the other three binary ferrites (Figure S30, Supporting Information).Since Mg 2+ does not possess 3d electrons, d-d transitions are weakened compared to ferrites with Mn 2+ , Ni 2+ , or Co 2+ .Indirect band gaps likewise decrease from 2.1 eV in ZnFe 2 O 4 over 1.6 eV in MgFe 2 O 4 to around or below 1 eV in the others.The contribution of Zn in ternary and multinary ferrites shows in larger direct and indirect band gaps in those ferrites containing larger amounts of Zn, e.g.(NiZn)Fe 2 O 4 , or (ZnMg)Fe 2 O 4 , in comparison to the others, which is in agreement to reports on (Ni x Zn 1-x )Fe 2 O 4 .

Figure 8 .
Figure 8. a) Bandgap values derived from Kubelka-Munk plots, b) from direct and indirect Tauc plots and c) the difference between the bandgap values determined from direct and indirect Tauc plots.
This can be attributed to low concentrations of the actually active elements, especially Co and Ni.When comparing the binary ferrites, NiFe 2 O 4 and CoFe 2 O 4 are the only ones exhibiting good OER activity (Figure S31, Supporting Information), whereas the other ferrites are mostly inactive.Especially Ni seems to play a crucial role in the OER activity, since (NiZn)Fe 2 O 4 is still very active, although less so than NiFe 2 O 4 and CoFe 2 O 4 .Since only the M 2+ -ion is varied, the relative ratio of Ni or Co in the spinel is with 33% (based on the total metal content) already comparably low.A further replacement by inactive elements leads to a decreased activity, since (NiCoMn)Fe 2 O 4 (40% Ni+Co) is considerably more active than (NiZnMn)Fe 2 O 4 (20% Ni+Co)(Figure 9; Figure S32, Supporting Information).
) 1 Fe 1 Co 1 O 4 and (NiCoMnCuZnMg) 1 Fe 1 Co 1 O 4 directly after the synthesis, which disappeared after calcination (Figure S33, Supporting Information).The activity was significantly improved, compared to the ferrites, with (NiCoMnCu) 1 Fe 1 Co 1 O 4 easily outperforming both NiFe 2 O 4 and CoFe 2 O 4 .A low overpotential of 0.46 V was realized for this spinel at 10 mA.When Cu was replaced by inactive Zn, the activity was significantly reduced.This effect is even more pronounced, when Mg is additionally used.Electron energy loss spectroscopy (EELS) confirmed the oxidation state of Fe 3+ after partial replacement by Co in (NiCoMnCu) 1 Fe 1 Co 1 O 4 (Figure S34, Supporting Information).Interestingly, a comparison with the Fe L-edge in Fe 3 O 4 and Fe 2 O 3 revealed that in both as-synthesized CoFe 2 O 4 and (NiCoMnCu)Fe 1 Co 1 O 4 , a low amount of the iron ions (<20%) occur as Fe 2+ .

Figure 9 .
Figure 9. LSV measurements for medium and high-entropy ferrites in comparison to NiFe 2 O 4 and CoFe 2 O 4 , both a) as-syn and b) after calcination.c) LSV scans for ferrites with four different M 2+ -ions are shown in (c), with both the initial sweep (after three previous cyclovoltammetry scans) and the final sweep shown.LSV scans for spinel oxides with Co as a partial replacement for Fe are depicted in (d).A comparison of the determined overpotential at 10 mA (if applicable) and at 1 mA depending on the composition is shown in (e).