Novel Fully Organic Water Oxidation Electrocatalysts: A Quest for Simplicity

Despite the growing need for readily available and inexpensive catalysts for the half-reactions involved in water splitting, water oxidation and reduction electrocatalysts are still traditionally based on noble metals. One long-standing challenge has been the development of an oxygen evolution reaction catalyzed by easily available, structurally simple, and purely organic compounds. Herein, we first generalize the performance of the known N-ethyl-flavinium ion to a number of derivatives. Furthermore, we demonstrate an unprecedented application of different pyridinium and related salts as very simple, inexpensive water oxidation organocatalysts consisting of earth-abundant elements (C, H, O, and N) exclusively. The results establish the prospects of heterocyclic aromatics for further design of new organic electrocatalysts for this challenging oxidation reaction.


INTRODUCTION
In the context of renewable energy transformation and storage, 1 the production of hydrogen through water electrolysis is of eminent importance. Although catalytic water reduction is already well investigated, 2 the water oxidation step has drawn less attention, despite representing the kinetically limiting halfreaction of the water splitting process. 3,4 To date, water oxidation catalysis is performed the most efficiently with molecular complexes and solids based on expensive transition metals, in particular Ru, 1−3,5 Ir, 6 Co, 7 or Os, 8 whereas on the technical scale electrolyzers mostly use Ni.
Recently, the N-ethyl-flavinium ion (EtFl + , Figure 1) was reported by Glusac and co-workers to function as a first organocatalyst for the four-electron electrocatalytic water oxidation. 9 This seminal piece of work represents an entry into an unexplored class of water oxidation catalysts based on potentially inexpensive and easily available organic heterocycles. 10,11 We envision that further developments within this class of compounds might yield improvements in catalytic activity and synthetic availability and perhaps thereby eventually provide an economical and environmentally friendly alternative for the water oxidation reaction. However, another cationic iminium derivative, N-methyl-9-phenyl-acridinium (Acr + , Figure 1), was found inactive toward the oxidation of water. 12 In the mechanism proposed, 9 the EtFl + catalyst adsorbed on the glassy carbon electrode surface is oxidized to a radical dication first, followed by the formation of a pseudo-base upon hydroxylation of the iminium carbon (position C4a) as the key step for oxygen evolution. The contrasting reactivities of EtFl + and Acr + suggest that a highly electrophilic iminium carbon (C4a in EtFl + ) is crucial to the catalytic activity. In the following, we assess to what extent the flavinium structure can be modified or simplified without loss of reactivity. In this paper, we will compare alternative catalysts and quantify their performance as the shift in water oxidation onset ΔE 0.5 , defined as the difference between the potentials needed to reach a current density of 0.5 mA cm −2 in the absence and presence of a catalyst. In essence, ΔE 0.5 represents the reduction of water oxidation overpotential (and concomitantly, of activation energy) achieved by the catalysts.

RESULTS AND DISCUSSION
For the reference compound EtFl + ClO 4 − , 9a we measure an onset potential shift of ΔE 0.5 = 0.385 V when the catalyst is added to the electrolytic mixture ( Figure S1 in the Supporting Information). Another quantification of the catalytic effect is provided by electrochemical impedance spectroscopy ( Figure  S2), with a 3-fold reduction of charge transfer resistance achieved by the catalyst at +1.94 V on a glassy carbon surface. In energy terms, this corresponds to a reduction of the activation free enthalpy by a modest 2.5 kJ mol −1 . Appending peripheral functional groups to the EtFl + backbone is tolerated. The carboxylic acid derivative EtFlCH 2 ΦCOOH + ClO 4 − yields an overpotential reduction value ΔE 0.5 = 0.277 V ( Figure S3).
On the basis of the accepted mechanistic hypothesis of water attack to the iminium carbon of EtFl + , 9a one would expect that a neutral flavin, Fl, in which the imine nitrogen is not quaternized should be electrocatalytically inert. Figure 2a shows that, unexpectedly, neutral Fl furnishes a species highly active for water oxidation as well: when water is added to the electrolytic mixture, the oxidation currents become significantly larger than their reductive counterparts. We measure ΔE 0.5 = 0.420 V, slightly larger than with EtFl + , and observe larger current densities as well. These results, however, are valid only for the second and subsequent voltammetric cycles. If we start with a fresh electrolytic solution of Fl (and water) near the open-circuit potential and sweep in the anodic direction (light blue curve of Figure 2b), no faradaic current (apart from the small pseudo-capacitive current due to glassy C electrode) is observed at all up to 2.50 V. After the initial voltammetric cycle, subsequent sweeps witness very clear faradaic peaks, which initially increase with each cycle and eventually stabilize. The reason for the appearance of two distinct peaks instead of one could possibly be clarified by spectroelectrochemical methods. If after such a set of voltammetric cycles, the electrolyte is stirred, then left to stand for a few seconds, and measured again in the same manner, the behavior shown on Figure 2b is reproducible. These observations contrast those observed for EtFl + , where the first and subsequent cycles are essentially identical to each other.
A plausible explanation for this behavior is furnished by recollecting the redox-based synthesis of EtFl + from Fl. 9a Indeed, alkylating of the heterocyclic N atom is performed in three steps: reduction by dithionite, electrophilic attack by an alkyl halide, and reoxidation. Thus, under our voltammetric conditions, each cathodic sweep must result in the formation of the intermediary compound, Fl •− , which is presumably protonated and subsequently oxidized to HFl + . This metastable species then features a reactivity much akin to that of the alkylated compound. The comparison of cyclic voltammograms recorded with two distinct switching potential values on the cathodic side is consistent with this model ( Figure S4). Finally,

ACS Omega
Article functionalized derivatives of neutral flavins may also be active. FlCH 2 ΦCOOH yields ΔE 0.5 = 0.284 V ( Figure S5) at the second and subsequent voltammetric cycles.
Let us now turn to simplified structures that still feature a highly electron-deficient heterocycle. Pyridinium salts, among them N-methyl-3,5-dinitropyridinium (MePy + ), have recently been reported by Cibulka and co-workers to successfully catalyze the sulfoxidation reaction. 13 This monooxidation, albeit a two-electron reaction only, is relevant to water oxidation if the proposed mechanism is correct, given the prominent intermediacy of an ortho-carbon-bound peroxide species. Despite their facile access, pyridinium compounds have not been investigated for water oxidation to date. Figure 3a shows the success of this simplified compound, which we prepared and investigated as a triflate salt, 13 and for which both the molecular weight and the synthetic effort are diminished very significantly. The shape of the voltammogram is qualitatively reminiscent of the flavinium family of compounds, with a peaking behavior characteristic of molecular, dissolved catalysts. 14 A quantitative comparison, however, reveals that MePy + is significantly more proficient than EtFl + . The water oxidation onset is further shifted to less positive potentials by approximately 0.2 V (ΔE 0.5 = 0.575 V). This corresponds to a reduction of the activation free enthalpy on the order of 18 kJ mol −1 .
The demonstration that water is indeed the substrate being oxidized electrocatalytically is provided in Figure 3b. Anodic current densities increase with water concentration, linearly at first, and reach a plateau in highly concentrated media (approximately 10 M). The anodic peak of interest also shifts from about 1.68 V to near 1.46 V (vs Ag/AgCl) due to the increasing polarity of the medium upon water addition, which stabilizes charged species. At the highest water concentration, dihydrogen evolution and reoxidation are also observable. Further, the product, O 2 , is directly measured by optical means (Figure 4) during bulk electrolysis at constant potentials. We compare the reference compound, EtFl + ClO 4 − , at relatively high potential E = 1.94 V versus Ag/AgCl (green dotted line) with our new MePy + OTf − at much more moderate value E = 1.34 V (continuous blue line). The gray curves represent the control measurements taken in the absence of a catalyst. The extreme electrocatalytic proficiency of MePy + toward water oxidation (as compared to that of EtFl + ) is materialized by a similar O 2 production rate obtained at 600 mV lesser overpotential, whereas most water oxidation catalysts increase the electrocatalytic current density by a factor 10 for every 60mV overpotential increase (Tafel plot).
Electrocatalytic current densities cannot be converted to turnover frequencies (TOFs) directly as in the case of homogeneous catalysis, given that the amount of molecular catalyst at the electrode surface is unknown. The Saveánt diffusion model for "foot-of-the-wave" analysis of the curve shape 14 is not applicable to our case due to the adsorption of the electrocatalyst demonstrated by Glusac et al. 9,12 To provide a value that can be compared to the homogeneous catalysts nevertheless, we therefore calculate a lower bound of the TOF

ACS Omega
Article on the basis of the assumption that the electrode surface is covered with a compact monolayer of MePy + . 9 For this, we will use an area of the molecular cation of approximately 50 Å 2 based on its molecular structure, a value that is consistent with the integrated current of the cyclic voltammograms recorded in the absence of water (≈1 C m −2 ). On the basis of this value, the current density, J = 2 mA cm −2 (5 nmol O 2 s −1 cm −2 ), obtained at the voltammetric peak of Figure 3b yields TOF ≥ 16 s −1 or TOF ≥ 56 000 h −1 . Any possibly lower surface coverage would yield even larger TOF values.
Further derivatives MePyz + and MePym + give rise to an electrocatalytic proficiency less pronounced than MePy + and practically inexistent, respectively ( Figures S6 and S7). A rationale for this behavior can be derived from 1 H NMR spectroscopy. The chemical shifts exhibited by the ortho hydrogens of these three cations follow the same trend as the catalytic activities ( Figure S8): MePy + (9.92 ppm) > MePyz + (9.49 ppm) > MePym + (9.28 ppm). Given that these values furnish a measure of the electrophilic character of heterocycles, the contrasting reactivities of the three novel compounds provide strong credence to the crucial mechanistic importance of a highly electrophilic ortho carbon (C2). The combination of two nitro substituents ortho and para to the crucial C2 at which nucleophilic attack by water occurs (see Figure S9 for a sketch of the hypothetical mechanism, based on the published flavinium system) seems to be ideal.

CONCLUSIONS
The ethylflavinium salt (EtFl + ), which was initially published as an exceptional case of an organic water oxidation electrocatalyst, 9a now belongs to a broader group of heterocyclic cations, some of which are simplified, more easily accessible, and even more active ( Figure 5). Catalytic proficiency scales with the electrophilic character of the heterocycle. This property does not require the complex flavin structure. The much simpler pyridinium framework outperforms the reference compound, EtFl + , by a significant amount. The pyridinium compound, MePy + , is stable for weeks in solution and under conditions of electrocatalytic turnover and outperforms EtFl + by about 18 kJ mol −1 in activation energy reduction. These results open an avenue toward more insightful mechanistic investigation of the water oxidation mechanism, for example, by time-resolved and potential-dependent spectroelectrochemical experiments and, thereby, toward further improvements in the performance of organic electrocatalysis.

METHODS
4.1. Materials. All chemicals used for syntheses and electrochemistry were purchased from commercial sources and were used without further purification. Nitrogen served as protective gas. All solvents were purified by distillation using rotary evaporation or were purchased in HPLC quality. All products were dried in vacuum (10 −3 bar).
4.2. Electrochemistry Measurements. The glassy carbon working electrode was obtained from Bioanalytical Systems, Inc. (BASi), the Pt mesh counter electrode from ALS Co., Ltd., and the Ag wire as the pseudo-reference electrode from Alfa Aesar. The pseudo-reference was calibrated before and after measurements with respect to a Ag/AgCl (sat)/NaCl (
4.5. Mass Spectrometry. Matrix-assisted laser desorption ionization mass spectra were recorded with a Shimadzu Biotech AXIMA Confidence, electron ionization mass spectra with a Thermo Scientific Finnigan MAT 95 XP, and electrospray ionization mass spectra with a Bruker Daltonik maXis 4G or a Bruker Daltonik micrOTOF II focus. 4.6. Infrared Spectroscopy. IR spectra of the compounds were recorded as thin films on a Varian IR-660 apparatus. 4.7. Elemental Analysis. Elemental analyses (C, H, N) were carried out with an Elementar vario MICRO cube machine.

ACS Omega
Article