Cobalt Amide Imidate Imidazolate Frameworks as Highly Active Oxygen Evolution Model Materials.

Two imidazolate-based Co-MOFs, IFP-5 and IFP-8 (imidazolate framework Potsdam), with a different peripheral group -R (-Me and -OMe, respectively) have been synthesized by a solvothermal method and tested toward the oxygen evolution reaction (OER). Remarkably, IFP-8 presents much lower overpotentials (319 mV at 10 mA/cm2 and 490 mV at 500 mA/cm2) than IFP-5 toward OER, as confirmed by online gas chromatography measurements (Faradaic yield of O2 > 99%). Moreover, the system is extraordinarily stable during 120 h, even when used as a catalyst toward the overall water splitting reaction without any sign of fatigue. An integrated ex situ spectroscopic study, based on powder X-ray diffraction, extended X-ray absorption fine structure, and attenuated total reflection, allows the identification of the active species and the factors that determine the catalytic activity. Indeed, it was found that the performances are highly affected by the nature of the -R group, because this small change strongly influences the conversion of the initial metal organic framework to the active species. As a consequence, the remarkable activity of IFP-8 can be ascribed to the formation of Co(O)OH phase with a particle size of a few nanometers (3-10 nm) during the electrocatalytic oxygen evolution.

Introduction

The oxygen evolution reaction (OER) is the ideal process for accessing a diverse number of renewable fuels and chemicals,[1,2] but its applicability is hampered by the difficulty to obtain stable and efficient catalysts.[3,4] Cobalt-based OER catalysts, both molecular[5] and heterogeneous,[6−10] have attracted enormous interest because of their relatively high activities and the metal’s abundance in the earth crust. Homogeneous molecular catalysts can be finely tuned to improve performances and allow for mechanistic studies, but the poor stability and recyclability of the reported systems hinder their applications.[11] In contrast, heterogeneous catalysts are more stable and cost-effective,[12] although the formation of the active centers on solid catalytic surfaces is difficult to control, identify, and quantity, making mechanistic studies very challenging.[13] Besides the well-investigated applications such as adsorption,[14] storage,[15] separation,[16] and conventional catalysis,[17−20] metal–organic frameworks (MOFs) are emerging materials for OER.[21] Among Co-based MOFs, ZIF-67 (ZIF = zeolitic imidazolate framework) is one of the benchmarks for oxygen evolution under several different working conditions. Indeed, it was demonstrated to be able to perform 21 TONs with a TOF of 0.035 s–1 under photocatalytic conditions, whereas if supported onto FTO (fluorine-doped tin oxide), ZIF-67 operates under a wide range of pH, requiring 1.73 V vs NHE to give a stable current of 1 mA/cm2 (at pH 9).[22] Nevertheless, the performances are quite far from those required for applications, because the poor conductivity of the framework generally hampers the direct use of MOFs in electrocatalysis.[23−31] Alternatively, MOFs can be subjected to framework degradation either by thermal decomposition,[32−35] chemical oxidation,[36] or in situ electrochemical aging,[37,38] in order to afford carbon/metal or metal oxide composites with improved performances.[39−42] As an example, An et al. demonstrated that thermal decomposition of ZIF-67 affords ultrathin Co3O4 nanomesh, rich in oxygen defects, with remarkable performances at pH 14 (η = 370 mV at 10 mA/cm2 and a Tafel slope of 74 mV/dec),[43] whereas Wang and co-workers improved the performances of Zn0.2Co0.8OOH via chemical oxidation of ZIF-67 in the presence of (NH4)2S2O8, lowering η10 to only 235 mV.[36]

A clear advantage of those approaches consists in the possibility to control the formation of the new phases by changing the reaction condition. However, when the MOF undergoes electrochemical aging, the system becomes much more dynamic, and the identification of the active species is uncertain and sometimes controversial.[37,38,44] For example, Ai and co-workers proposed the conversion of UTSA-16 to CoO active phase under OER conditions,[38] whereas Zhang et al. suggested that electrochemical aging in O2 saturated electrolyte induces the exfoliation of a 3D pillared-layer structure to achieve a 2D amorphous ultrathin phase, excluding the possibility to form CoO because of its much lower activity.[37] Moreover, Fischer et al. proposed that the active phase of ZIF-9 consists of a nitrogen-coordinated cobalt oxyhydroxide N4COOH species.[44]

In order to understand how to control the transformation of the MOF to the active catalytic phase, we used two model systems named IFP (imidazolate framework Potsdam),[45,46] based on penta-coordinated Co secondary building units (SBU’s) connected by multidentate amido-imidate-imidazolate linkers. The models differ in the nature of the substituent at the imidazole C2 position, which can be either a −Me (IFP-5) or an −OMe (IFP-8) group. We have studied the OER activity of IFP-5 and IFP-8 under basic conditions (pH 14 and 13), demonstrating their high catalytic activity. More interestingly, the ex situ X-ray absorption spectra (XAS), powder X-ray diffraction (PXRD), and attenuated total reflection (ATR) experiments gave insight into how the transformation of the framework is able to tune the formation of the active species, which leads to excellent catalytic performances.

Result and Discussion

Synthesis and Characterization

The synthesis of Co-amide-imidate-imidazolate frameworks [Co(C5H3N4O2)-R] (with R = −Me[45] and −OMe[46]) were previously reported by Holdt et al. The synthesis of the MOFs proceeds by the reaction of 2-substituted-4,5-dicyanoimidazole under solvothermal conditions in the presence of Co(NO3)2·6H2O (see the Supporting Information (SI) for details). During the reaction, the partial hydrolysis of the ligand affords the imidazolate-4-amide-5-imidate (C5H3N4O2), which is the real linker of the IFP (Figure , top).

Figure 1

Top: general synthetic strategy of IFP frameworks, highlighting the in situ ligand hydrolysis. Bottom: 1D hexagonal channel structure of IFP, with an enlargement of the asymmetric unit of IFPs, clarifying the Co2+ coordination environment. The label R stands for −Me and −OMe in the case of IFP-5 and IFP-8, respectively.

For comparison, the canonical ZIF-67 was synthesized according to the reported literature procedure.[47] The experimental PXRD patterns, with the simulated ones, as well as XAS of the as synthesized IFPs are shown in Figure .

Figure 2

PXRD pattern (a) and Co K-edge X-ray absorption near edge structure (XANES) profiles (b) of IFP-5 (blue), IFP-8 (red), and ZIF-67 (magenta). Co3O4 (black) was used as a reference in the XANES.

The experimental diffraction patterns perfectly match the simulated ones, showing characteristic peaks at 7.4°, 9.9°, 15.9°, 17.3°, 23.3°, and 30.2°. The sharpness of the diffraction peaks indicates that the newly obtained IFP-5 is highly crystalline, whereas IFP-8 is less so.

Furthermore, the ATR spectrum (Figure S1) clearly shows the presence of characteristic bands around 1560 and 1660 cm–1, which are due to the formation of amide and imide groups. Furthermore, between 3320 and 3330 cm–1, a broad amide-imidate N–H band with a considerably fine structure was observed.

The Co K-edge of the extended X-ray absorption fine structure (EXAFS) spectra of IFPs show pre-edge 1s → 3d transitions centered around 7709.5 eV and rising edges at 7721.2 eV. These features are 0.4 and 1.5 eV lower in energy than the mixed Co2+/Co3+ centers in the Co3O4 reference, which is consistent with the Co2+ oxidation state for the IFP series. Moreover, EXAFS analysis (Figure S2) shows a 5-coordinated metal center having two first coordination sphere N/O scattering shells, with 2 N/O atoms at 2.00 Å and 3 N/O atoms at 2.11 Å. On the other hand, ZIF-67 has an intense pre-edge at 7709.9 eV that can be rationalized in terms of increased p–d mixing due to its tetrahedral geometry. Conversely to what is observed for the IFP model, ZIF-67 is better described by a single shell of N/O scattering atoms at 2.00 Å.

Putting together the pieces of information obtained, the chemical environment of the Co2+ ion in the IFP series is constituted by five donor atoms of three ligands to form a distorted trigonal-bipyramidal geometry (Figure , down). In this arrangement, the imidate N4 and O2 and the amide O1 reside in equatorial positions, with the two imidazolate N atoms (N1 and N2) occupying the axial positions. The multidentate imidazolate-based ligands, produced by the in situ hydration of the dicyano precursor, combine with Co2+ ions to form the neutral microporous imidazolate MOF with 1-D hexagonal channels. The Co2+ SBUs at the IFP structure and the bridging ligands act as 3-connected topological species forming a net with a uninodal topology, named etb. The functional groups at the imidazolic C2 position protrude into the channel opening, tuning the pore aperture (4.2–1.7 Å), polarity, and functionality of the channel walls (Figure ). As a consequence, the specific surface area of IFP is much higher for IFP-5 (666.2 m2g–1) than for IFP-8 (45.8 m2g–1) based on the narrowing the pore aperture window of the IFP structure (Figure S3). The scanning electron microscopy (SEM) analysis of IFP samples (Figure S4) revealed that IFP-5 displays hexagonal nanostructures, whereas IFP-8 forms spherical morphologies.

In order to investigate the chemical stability of the as-synthesized IFP materials, these were subjected to aging under very basic conditions. The final PXRD patterns revealed no obvious pronounced changes after 24 and 72 h at pH 13 (Figure S5), but some modifications took place when the pH was raised to 14. The new samples, labeled IFP-Xbt (where X identifies the corresponding IFP number and bt stands for “basic treatment”) were characterized via PXRD and ATR spectroscopy (Figures S6, S7). The broadening of the PXRD peaks clearly suggests the partial loss of crystallinity (Figures S6a, S7a). In the ATR spectrum, a very sharp stretching at 3630 cm–1 appears (Figures S6b, S7b) consistent with the formation of hydroxyl groups, indicating the occurrence of hydrolysis. Interestingly, the intensity of the signals at ca. 2800 and 1100 cm–1 that are attributable to CH3–O and C–O–C stretching, respectively,[48] decreases in IFP-8bt, whereas the intensities of all peaks are maintained for IFP-5 after the basic treatment, suggesting different lability of the corresponding R group toward hydrolysis.[46]

Along with the molecular precursors, we also synthesized the corresponding cobalt oxides Co3O4-X (when X is 5 or 8, stands for IFP-5 and IFP-8, respectively; whereas, when X is 67 or R, stands for the calcinated ZIF-67 and the reference Co3O4, respectively). The synthetic procedure to obtain Co3O4 spinels involves a simple thermal treatment at 700 °C of the corresponding IFP under N2 atmosphere (Figure S8), followed by acid etching in 0.5 M H2SO4 for 2 h, that affords Co3O4 phase as demonstrated by PXRD, ATR, XAS (Figure S9), SEM (Figure S10), and TEM (Figure S11 and S12) analysis, with an average particle size of 46 nm.

Electrocatalytic OER, Ex Situ Characterization of Catalytic Materials

As an initial step, we compared the electrochemical performances at pH 14 of IFPs and ZIF-67 (ZIF-67 has been used as a reference under exactly the same reaction conditions) by drop casting 200 μg of the corresponding material onto the nickel foam (NF).

Remarkably, IFPs exhibit better performances than the well-known ZIF-67 benchmark. IFP-8 is the most active electrocatalyst (Figure a–d), with a low overpotential at 10 mA/cm2 (η10) of 319 and 257 mV at the onset, respectively (Figure S13). Also, the Tafel slope of 64 mV/dec is particularly low (Figure S14). The η10 for the other tested catalysts are 340 and 360 mV, whereas the onsets are 273 and 269 mV for IFP-5 and ZIF-67, respectively (Figure S13). The values of the Tafel slope (Figure S14) are slightly higher (ca. 70 mV/dec for both). Such values are considerably lower than previously reported cobalt-based MOF benchmarks, as those reported by and Pang and Qu.[30,31]

Figure 3

(a) Polarization curves and (b) long-term chronopotentiometry (CP) at J = 10 mA/cm2 for IFP-5 (blue), IFP-8 (red), and ZIF-67 (magenta); (c) LSV of IFP-8 before (black) and after (blue) the CP; (d) comparison in the overpotential at 10 mA/cm2 with the most common MOFs in the literature at pH 14; (e) evaluation of the nominal TOF for the material studied calculated at η = 400 mV and the catalyst loading taking account for the leaching (vide infra) and (f) performances of IFP materials at pH 13 (0.1 M KOH). For all the experiments, the current collector used was a 1 × 1 cm2 geometric area of NF.

In order to rationalize this remarkable activity, electrochemical impedance spectroscopy (EIS) in the potential range of 0.9–1.8 V vs RHE has been performed. The resistance associated with OER increases from IFP-8 (1.9 Ω at 1.6 V vs RHE) to IFP-5 (2.6 Ω at 1.6 V vs RHE), whereas the values of the pseudocapacitance for the different catalysts inversely fit this trend, since IFP-8 has the highest capacitance in the entire potential window investigated (Figure S15 and Table S1). This is an indication that IFP-8 exhibits a higher electrochemical surface area (ECSA) with respect to IFP-5 and, consequently, a higher number of active sites exposed. Indeed, the ECSA measured from the double layer capacitance is 11.2 and 2.7 cm2 for IFP-8 and IFP-5, respectively (see SI for calculation details).

Remarkably, chronopotentiometry (CP) at 10 mA/cm2 (Figure b) shows a slight decrease in the overpotential for both IPFs at the beginning of the CP, but then it remains stable during the successive 13 h, clearly indicating that an activation process occurs. Indeed, η10 of IFP-8 is 319 mV after the activation phase, and the η50 became as low as 390 mV (Figure c). Furthermore, a remarkable and sustained current of 480 mA/cm2 was observed at an overpotential as low as 490 mV. Such stability is uncommon for simply deposited MOF on the electrode material so this prompts us to push the stability tests toward a very long CP run during 180 h at 10 mA/cm2 (Figure S16). For comparison, the bioinspired cobalt-citrate MOF, UTSA-16, needs 408 mV of overpotential to deliver a stable current of 10 mA/cm2 during 7 h before deactivation,[38] whereas the Co2(OH)2BDC nanosheets recently reported by Pang et al., exhibit much lower η10 (273 mV) and η50 of ca. 360 mV, but such outstanding activity at pH 14 is maintained for only 3 h, before that current drops up to 60% of its initial value and η50 increases up to 410 mV (as estimated by LSV showed therein).[30] In general, the performances of IFPs are much better than the majority of Co-based MOFs and even with respect to many pure inorganic materials (Figure d and Table S4).[25,49]

Nevertheless, we performed an additional long-term chronopotentiometry experiment (4 h at 20 mA/cm2) aimed at monitoring the gas evolution by means of online gas-chromatography (see SI for calculation details). A quantitative Faradaic efficiency (>99%) was found during the entire experiment for both IFP-5 and IFP-8 (Figure S16), which ensures that the current measured is due to O2 evolution.

According to the measures of the inductively coupled plasma optical emission spectrometry (ICP-OES), IFP-8 also has the highest nominal TOF value (0.53 s–1 at η = 0.4 V) over IFP-5 and ZIF-67 (Figure e). However, it should be noted that these TOF values are an underestimation of the real performances because they are calculated under the assumption that all the catalytic centers equally participate in catalysis.[50] However, even the nominal TOF value reported for IFP-8 largely exceeds the most common benchmarks reported so far; for example, that of Co-MOF nanosheet array directly grown on NF (0.18 s–1 at η = 0.4 V) recently reported by Qu et al.[31]

Since the chemical stability tests revealed that IFPs at pH 13 do not suffer of partial hydrolysis, the same measurements were performed at this lower pH (Figure f). Remarkably, the activity of IFPs is competitive even under these conditions, and indeed, η10 are 360 and 410 mV for IFP-8 and IFP-5, respectively, which are comparable with similar single Co-based MOF MCF-12 nanosheets recently reported by Zhang et al. (η10 = 310 mV)[37] and largely exceeding the performances of other benchmarks as MAF-X27-OH (η10 = 461 mV),[25] MOF(Fe1–Co3)550N (η10 = 390 mV),[51] including inorganic materials.[52−54] Interestingly, at this pH, the η10 drifts after 12 h of long-term chronopotentiometry experiments (15 h at 10 mA/cm2), approaching 500 and 680 mV (Figure S17).

The remarkably low overpotentials at 10 mA/cm2 prompt us to study IFP-8 as a bifunctional catalyst for both OER and hydrogen evolution reaction (HER) and therefore as a water splitting catalyst. HER tests (Figure S18) were performed in a three-electrode cell with the same configuration as for OER, but this time, the electrolyte solution was continuously degassed by N2 flow, in order to remove O2. Furthermore, the potential at the working electrode was varied between 0.9 V to −0.5 V vs RHE at a scan rate of 1 mV/s. The η10 of 227 mV is quite lower if compared to other Co-based catalysts[10,55−58] and competitive with the most recent Co-based hybrid systems.[59−61] As indicated in Figure , the overall water splitting reaction can be performed at 10 mA/cm2 by applying ca. 1.89 V. A long CP experiment highlights the extreme durability of IFP-8 under total water splitting conditions (for 120 h at 10 mA/cm2), as the activity remains rather constant for the entire duration of the experiment without suffering apparent fatigue (Figure b). The overall water splitting is confirmed by the online measured H2/O2 ratio of ca. 1.9, with quantitative faraday efficiency during the first 3 h of reaction (Figure S19). In total, after 120 h of electrolysis, minimum TONs of 26 000 and 13 000 are calculated for H2 and O2, respectively, considering all the deposited material over the electrode.

Figure 4

(a) Overall water splitting LSV (in red, HER region and in blue OER region). The potentials at 10 mA/cm2 are highlighted in the picture, whereas thermodynamic potential for HER and OER are marked as E0 in the figure. (b) Represents a CP experiment carried out at 10 mA/cm2 during 120 h for the overall water splitting reaction (inset shows the H2/O2 ratio recorded during the measurement). All the measurements were performed at pH 14 by KOH, supporting IFP-8 onto NF.

In order to disclose the origin of the difference in the activity among those catalysts and the mechanism of activation, we performed several experiments. First of all, IFPs were compared to their related calcinated derivatives Co3O4-X samples and Co3O4-R, the latter prepared independently via solution combustion synthesis (see SI for further details). IFP-8 exhibits much higher performances than both Co3O4-8 (η10 = 380 mV) or the Co3O4 reference (η10 = 390 mV), and the same trend is observed for IFP-5 (Figure S20). The difference in the catalytic activity between IFPs and Co3O4-X is even more evident when TOF values are compared (Figure S20c), suggesting that the active species of the two families should be quite different in nature. In other experiments, the comparison was focused between IFP and IFP-bt, finding nearly identical activity (Figure S20). More interestingly, the ECSA of the materials treated with base and IFP-8 are very close (ca. 12 cm2), and it is much higher than that of IFP-5 (ca. 2 cm2), in agreement with the fact that the loss of crystallinity might induce higher exposure of the active sites. This suggests that the pH-induced degradation of the MOF might be one of the factors that drives the formation of the active species.[37]

Encouraged by these results, PXRD, ATR, and EXAFS experiments were performed ex situ before and after the catalysis, by plating the materials onto FTO as the electrode material (IFP@FTO).

The electrochemical studies of IFPs@FTO were carried out analogously to the characterization in NF. Initial LSVs confirm the reactivity order found in NF (Figure S21a,b). At the same time, during the 5 h CP experiments performed at 5 mA/cm2, the reduction of the potential and the increased number of exposed active sites (as indicated by the integrated charge of Co redox features and by ECSA, Table S5) indicate that the catalysts undergo an activation process, although with different rate with respect to that observed in NF. Nevertheless, the main kinetic descriptors for the intrinsic activity, that is, TOF and ECSA-normalized current density, decrease. In particular, TOF values (at η = 300 mV) were observed to almost half from 0.29 s–1 to 0.16 s–1 for IFP-8 and reduce from 0.35 s–1 to 0.25 s–1 for IFP-5. The decrease of the TOF value might be symptomatic for the loss of molecular behavior in favor of an aggregation process, which would likely be stronger in the case of IFP-8 than for IFP-5 because of the higher drop of the TOF.

In order to better describe the activation process, after the 5 h CP, the samples were investigated spectroscopically. The PXRD pattern and the ATR spectrum of IFP-5@FTO presented important attenuation of the signals (Figure ). In particular, in the diffraction pattern, only the main peak at 10° is visible, whereas in the infrared spectrum, all the features from the framework (1000–1700 cm–1) appear, but with reduced intensity. In the TEM images, it is apparent that the surface remains intact with minor degradation and roughness, suggesting that the initial Co distribution could be maintained or undergoes minor aggregation (Figure S22). These indications, together with minor metal leaching detected by ICP-OES measures (accounting for only 5% of the total metal deposited), point toward the retention of the original structure of IFP-5.

Figure 5

Characterization of IFP materials plated on FTO. Panels (a) and (b) offer a comparison between PXRD pattern and ATR before (black) and after (red) OER for IFP-5 (upper) and IFP-8 (lower), respectively, whereas TEM images (scale bars = 50 nm) are depicted for IFP-8 before (c) and after (d) OER. (e) Co K-edge XANES profiles for the IFP-8 measured as a pellet (red line) and supported in FTO before (dark red line) and after catalysis (blue line) versus the Co3O4 reference (black line). Fourier transformed EXAFS data and corresponding fits within the inset showing k-space spectra of IFP-8 supported in FTO after catalysis (f).

A different scenario can be depicted for IFP-8@FTO because the attenuation of the PXRD pattern, the disappearance of ATR signals in the organic region, and the metal leaching of ca. 35% clearly indicate that the framework undergoes major degradation. Moreover, TEM images reveal the formation of amorphous and very small particles (3–10 nm) embedded in a thin and plain surface (Figure c,d and Figure S23).

However, all the samples clearly show a new dominant phase in the 400–650 cm–1 region of the ATR, corresponding to Co–O stretching (Figure ).[38,62−65] Such a broad band was proposed to be a convolution of LCoO (Co4+), LCoOH (Co3+), and LCoOOH (Co3+) species.[44,66,67] In order to probe the chemical nature of the new phase, XAS measurements were performed. The spectrum of IFP-8 after catalysis exhibits a rising edge comparable to that of Co3O4 but with a clear shift to higher energy, having a pre-edge at 7710.4 eV and a rising edge at 7725.6 eV. Such a shift, together with the structure of the XANES and the Fourier transformed EXAFS spectrum are compatible with the previously reported Co-Pi or CoO(OH).[68−70] Indeed, in the FT data, two dominant peaks can be observed, corresponding to (Co–O at 1.90 Å) and Co–Co scattering paths with distance of 2.85 Å (Figure e,f). Those signals, which are symptomatic of an octahedral environment, correspond to a single first shell of 6 N/O scattering atoms, whereas the number of scattering atoms (N) for the latter shell averages to 4.3 ± 0.4, which is lower than the 6 expected for CoO(OH). Subsequently, in analogy with the previously reported EXAFS analysis by Nocera et al., the model requires the inclusion of four more contributions: Co–O (3.43 Å), Co–Co (4.94 Å), multiple scattering Co–Co–Co (5.77 Å), as intralayer paths and, finally, also the interlayer scattering shell Co–O (3.72 Å). Among them, the features at R(Å) at 4.94 and 5.77 Å are still in agreement with the CoO(OH) and bulk Co-Pi spectra and definitely rule out the presence of Co3O4 that does not have the path at 4.94 Å. However, it is important to notice that the coordination numbers for the paths at 4.94 and 5.77 are 4.3 and 8.6, instead of 6 and 12, respectively. Such a loss indicates that the forming Co(O)OH exhibits significant lattice imperfections, in agreement with the traces of degradation observed in the TEM.

The kinetic of the degradation of the framework and the formation of CoOOH species was further studied by collecting ATR spectra at different times during the CP experiment (Figure ).

Figure 6

Comparison of the CP activation (blue line) of IFP-5 (a) and IFP-8 (b) with the formation of the catalytic active species (red dots) and the disappearance of the organic framework (black dots). The values of the ATR areas of the organic region are normalized to the initial values calculated at t = 0. For those experiments, IFPs were supported onto the FTO electrode and employed in 1 M KOH solution.

Figure suggests a correlation between the increment of the catalytic activity during time (the η5 lowers), the formation of active species (Co–O bond signal), and the framework degradation (Figure S24). Moreover, full evolution/activation of IFP-8 occurs in only 3 h; however, for IFP-5, the η5 stabilizes after 5 h, although the organic framework is still largely present. Furthermore, the ATR signal for the Co–O stretching of IFP-5 is 3 times lower in intensity than for IFP-8 (Figure S25).

A plausible explanation for different activation rates relies on the robustness of the organic linker, which should be imparted by the −R group for two main aspects. The first is that the −Me moiety from IFP-5 infers hydrophobic character to the MOF pores, repelling water and preventing the structural degradation.[71] As a second reason, the degradation might depend on the different rate in the hydrolysis of the −R group in the C2 position of the imidazolium moiety, which is expected to be easier for the −OMe group than for the −Me one.[46] An indirect proof for this was already found for the basic treated samples (Figure S7, vide supra). Once the hydrolysis occurs, the successive oxidative imidazolium ring-opening mechanism might be the most favored event under those conditions even at room temperature, as well described by Elabd et al.[46,72]

In this sense, the −R group modulates the extent of conversion of the MOF into the active species. For IFP-8, the fast and complete degradation of the framework leads to a higher amount of the active species, and discrete particles became visible in TEM. During electrochemical tests, IFP-8 exhibited higher capacitance over IFP-5, indicating that the same active species forms more active sites (Figure S25c) with respect to its −Me analogue.

Conclusions

IFP materials are demonstrated to be competent electrocatalysts toward OER with quantitative Faradaic yield, and in particular, IFP-8 exhibits performances that exceed those of the majority of other Co-based MOFs, especially for its durability. The ex situ characterization indicates that the catalytic activity can be correlated with the formation of the active phase, suggesting that a key feature to achieve high performances consists in controlling the rate of the hydrolysis of the framework, which drives the conversion of the MOF into catalytically active Co(O)OH phase. Such a control can be achieved by combining the design of suitable ligands with labile groups that are able to optimize the growth, size, and electric connection of catalytic efficient Co(O)OH nanoparticles. We envision that the in-depth analysis reported would help to bridge the gap between the single-site molecular well-structured materials and aggregated systems, guiding the improvement of catalytic materials under highly demanding conditions such as the OER.