Oxygen reduction reaction at conducting polymer electrodes in a wider context: Insights from modelling concerning outer and inner sphere mechanisms

Abstract

Practical interest in oxygen reduction reaction (ORR) has traditionally been due to its application at fuel cells’ cathode following its complete 4e route to the water. In search of new electrode materials, it was discovered that conducting polymers (CPs) also are capable of driving ORR, though predominantly halting the process at 2e reduction leading to hydrogen peroxide generation. As alternative ways to produce this “green oxidant” are attracting increasing attention, a detailed study of the ORR mechanism at CP electrodes gains importance. Here, we summarize our recent theoretical work on the topic, which underscores the fundamental difference between CP and electrocatalytic metal ORR electrodes. Our insights also bring to us the attention of outer-sphere electron transfer, not unknown but somewhat ignored in the field. We also put the action of CP electrodes in a more general context of chemical ORR and redox mediation responsible for the electrocatalytic ORR mechanism.

1 OXYGEN REDUCTION REACTION AT METALLIC ELECTRODES: 2e AND 4e, INNER- AND OUTER-SPHERE

The electrochemistry of molecular oxygen O₂ (dioxygen) is attracting considerable attention due to the current transition into a green and fossil-free economy.^[1] Electrical energy storage devices such as fuel cells run the oxygen reduction reaction (ORR) at their cathode that transforms molecular oxygen into water in a 4e overall process, which, for acidic media, can be written as

$$$$\begin{eqnarray*} & & {\mathrm{O}}_{2}+4{\mathrm{H}}^{+}+4\mathrm{e}\to {\mathrm{H}}_{2}\mathrm{O}\\ & & {\mathrm{E}}^{\circ}=1.23\mathrm{V}\hspace*{0.28em}\left(\textit{SHE}\right) \end{eqnarray*}$$$$

The renewed interest in ORR is also due to the research for new ways of hydrogen peroxide electrolytic production in the framework of green chemistry^[2] and fuels. In this case, molecular oxygen undergoes a 2e reduction:

$$$$\begin{equation*}{{\mathop{\rm O}\nolimits} _2} + 2{{\mathop{\rm H}\nolimits} ^ + } + 2{\mathop{\rm e}\nolimits} \to {{\mathop{\rm H}\nolimits} _2}{{\mathop{\rm O}\nolimits} _2}\ \quad {\mathop{\rm E}\nolimits} ^\circ = 0.70{\mathop{\rm V}\nolimits} \left( {{\rm{SHE}}} \right)\end{equation*}$$$$

In the energy storage context, this process is an unwanted side reaction leading to energy losses due to smaller energy density per oxygen molecule and to the production of the byproduct H₂O₂ that typically is harmful to membranes.^[3] However, H₂O₂ is considered a ‘green’ oxidant and is also a possible energy carrier. This makes 2e ORR an important target process in itself, on which extensive literature is available.^{[4, 5]} Given the thermodynamic preference of the 4e reduction process, as follows from the redox potentials, searching for the electrode material ensuring desired 2e/4e ORR selectivity is a tricky issue that attracts particular attention. One successful electrocatalytic approach is at hand, which favors the OO bond cleavage of dioxygen. This process takes the use of dissociative chemisorption of both oxygen atoms of dioxygen, at two adjacent sites of the catalyst, where usually a platinum group metal is operated in acidic media, followed by inner-sphere electron transfer (Figure 1) that leads directly to water via 4e ORR.^[1] To suppress the 4e process, that is to decrease controllably the catalyst's activity, only one active metal atom should be made available for OOH chemisorption (thus without OO bond cleavage) by surrounding it with non-chemisorbing neighbors: PtHg₄ is one of the best inorganic 2e ORR catalytic alloys^{[3, 4]} for H₂O₂ production. Different modes of chemisorption of O₂ thus alter the energetics of subsequent electron transfers already at the thermodynamic level, so even without considering the kinetic barriers (transitions states) the relative efficiency of true electrocatalysts can be accounted for, leading to the theoretical volcano plots. This forms the basics of a successful theoretical methodology pioneered by Nørskov et al.,^[6] experiment and theory go here hand in hand, and the mechanism of true electrocatalysis involving chemisorbed intermediates is pretty well understood.

FIGURE 1

Dioxygen chemisorption at an electrocatalyst followed by inner-sphere electron transfer

At intrinsically inert and additionally blocked by ion adsorption electrodes when chemisorption of dioxygen is impossible, a mechanism markedly different from the electrocatalysis introduced above and also leading to hydrogen peroxide, which does not require OO bond cleavage, takes place^[7-9]: outer-sphere electron transfer, as recognized rather early.^[10] Observed insensitivity of ORR to the nature of some metal cathodes in alkaline media was the first indication toward outer-sphere electron transfer to solvated O₂. The further analysis considered such factors as oxygen chemisorption energy, the potential of zero charges of the metal, and the structure of the double layer on it, including its composition and the orientation of water dipoles, which can even make this mechanism rather efficient.^[11] This leads to the conclusion that the outer sphere mechanism can be favorable only in alkaline solutions, while in acidic media an inner-sphere mechanism competes with the former successfully. This may be indeed true for metals, but not necessarily for carbon materials, which we are going to discuss ahead. Also, it should be borne in mind that a strong argument in favor of the outer-sphere mechanism is the detection of superoxide radical anion O₂^• in alkaline solutions.^[12] However this species becomes particularly unstable toward disproportionation with its protonated neutral counterpart, peroxyl radical HO₂^• (pKa 4.8) in acidic media,^[13] which does not imply that it is not formed at all.

2 ORR AT NON-METALLIC ELECTRODES AND ITS RELATION TO CHEMICAL ORR

Metal-free electrodes are also being explored for use in H₂O₂ production for economic and environmental, in addition to fundamental, reasons. Moreover, inexpensive composite carbon chips made the first electrode material for the industrial Huron-Dow process, but the reaction mechanism is much less clear, in spite of a high level of practical optimization.^{[11, 14]} Indeed, on basal plane graphite ORR is very sluggish,^[15] so the clue to activity clearly lies in the defects, chemical or structural. Two ORR waves were thus observed, one with a lower overpotential due to quinoid mediation and the other, more cathodic, due to the uncatalyzed two-electron reduction of O₂ in alkaline media.^[16] Noteworthy, only the uncatalyzed, that is purely outer-sphere, the reduction was observed in the same work in an acidic solution. Specifically engineered graphene sheets may contain well-defined radical sites favoring chemisorption of dioxygen^{[17, 18]}; numerous theoretical studies of ORR on graphene materials were aimed at rationalizing their action, to cite just a recent one.^[19] An interesting experimental approach consisted also in the controlled elimination of defects in a carbon material thus suggesting that ORR activity was related to phenolic oxygen.^[20]

The involvement of quinone/phenol transformation in the above electrochemical catalysis establishes a bridge to the old and dominant chemical industrial anthraquinone process of H₂O₂ production. Its mechanism, unclear until recently, can be due to efficient hydrogen atom transfer to molecular oxygen from reduced anthraquinone (anthraquinol).^[21] This chemistry, in a wider context, belongs to “autooxidation” of organic compounds defined as “apparently uncatalyzed oxidation of a substance-exposed to the oxygen of the air”, implying that it takes place in the absence of transition metals, in particular.^[22] The crucial catalytic role of the latter is due to spin restrictions in the reactions involving dioxygen,^[23] whose ground state is triplet ³O₂, with singlet (closed-shell) organic molecules. While total spin in chemical reactions should normally be conserved, which severely impacts the kinetics of even thermodynamically favored transformations, transition metals introduce spin-orbit coupling that lift spin restrictions to some extent. The exact mechanism of autooxidation is still under debate,^[24] but it is certainly a radical chain mechanism, as well as that of the anthraquinone process, precisely due to the spin rules that allow open-shell particles, such as radicals and triplet dioxygen, to react.

Back to electrochemistry, controlled modification of graphite and carbon with quinones as a way to boost ORR but also uncover its mechanism at non-metallic electrodes attracted a lot of attention since the 80s,^{[25, 26]} with more recent important contributions from quite a few groups,^[27-32] including a theoretical study.^[33] Noteworthy, the reaction does not start here by chemisorption of dioxygen prior to electron transfer, as on catalytic metals. Instead, chemi- or even physisorbed quinone first undergoes 1e reduction to semiquinone radical serving as an efficient mediator of electron transfer to dioxygen. The huge biogeochemical role of quinones^[34] is largely due to their role as mediators in reactions with oxygen. The details of this redox mediation are still under study and may involve either chemical bonding between the semiquinone radical and triplet (diradical) dioxygen or electron and proton transfer to unbound dioxygen, without OO bond breaking in both cases and therefore leading to H₂O₂. In other words, this reaction can be inner- or outer-sphere: that is, respectively, involving or not chemical bond formation between the oxygen intermediates and the electrode material. In any case, it is the redox mediation by specific functional groups at carbon materials that makes ORR at carbon and quinone-modified electrodes fundamentally different from that at metal electrodes. The semiquinone reactive intermediate efficiently transfers the first electron to dioxygen by outer- or inner-sphere mechanism resulting in a clear electrocatalytic effect. However, the exact reasons for such efficiency are yet to be understood. In a recent work can be mentioned^[35] where theoretical treatment of ORR at oxygenated carbon materials comprised the comparison of anthraquinone-like (outer-sphere) and chemisorption pathways, in favor of the latter.

A redox mediator is a molecule assuring an efficient chemical reaction with the target, molecular oxygen in the case of ORR. The active form of an ORR redox mediator is thus usually a radical. Importantly, the electronic structure of a mediator, that is energetics and spatial distribution of its molecular orbitals is independent of the electrode potential, which can influence only the rate of its recovery, but not directly its reaction with the target. On the contrary, the energy of the electrons exchanged with a metal electrode is directly defined by the applied potential. Anthraquinone, as a soluble redox mediator, was proposed as an efficient mediator also in photochemical ORR at a dye-sensitized semiconductor.^[36]

3 ORR AT CONDUCTING POLYMER ELECTRODES

In this laboratory, we are interested in electrochemical reactions, ORR in particular, at conducting polymer (CP) electrodes, combining experimental and theoretical studies. A CP is a molecular material that combines the functions of an electron conductor (depending on its state of doping) and an ion conductor (due to the presence of compensating ions).^{[37, 38]} This combination of properties suggests the utilization of CP in the role of an electrically and ionically conducting matrix in a composite with an established electrocatalyst as the main active component.^[39] Ion-selective properties of CP electrodes allowed us recently to demonstrate a promising behavior of ion-selective electrocatalysis (ISEC).^[40] However, the performance of CP as active electrode materials themselves is well documented.

A p-type polymer (poly[3,4-ethylenedioxythiophene], PEDOT) (Figure 2) is commercialized and widely used due to its stability and well-established manufacturing technology.^[41-44] The current status of its theoretical understanding was recently reviewed by us.^[45] Winther-Jensen et al.^[46] were the first to demonstrate that PEDOT electrode catalyzes the ORR. The activity of PEDOT free of any metal impurities in driving predominantly 2e reduction to hydrogen peroxide has been confirmed since then in a number of studies.^[47-51] At the same time, one must admit that the role of PEDOT in electrochemical ORR remains a puzzle. Indeed, ORR is detected at cathodic potentials where PEDOT is predominantly in its de-doped neutral form PEDOT⁰, neither electrically conducting, nor containing radicals (polarons).

FIGURE 2

Oxygen reduction reaction (ORR) activity of poly(benzimidazole-benzophenanthroline) (BBL) and poly(3,4-ethylenedioxythiophene) (PEDOT) (linear sweep voltammograms in oxygen-saturated 0.1 M KCl electrolyte^[53]) in comparison to glassy carbon electrode (GCE), their chemical formulae and schematic band structures (VB and CB, valence and conducting bands, respectively).

Poly(benzimidazole-benzophenanthroline) (BBL) is a ladder-type redox polymer (Figure 2) synthesized already in the late 1960s featuring a rigid linear pi-conjugated backbone unit with integrated redox sites and a high n-doping level. BBL appeared to us as an attractive CP candidate electrode for ORR because, unlike p-type PEDOT, it is electrically conductive in the cathodic region and at the same time highly stable in water.^[52] Our group was the first to show recently that rigorously metal-free BBL indeed drives not only 2e ORR to H₂O₂, but also hydrogen peroxide reduction, thus allowing further conversion to water via a complete (2+2)e ORR.^[53]

Here, we describe our application of computational chemistry helping to uncover some basic features of ORR at a p- and n-type CP.

4 MODELLING ORR AT PEDOT

As at cathodic potentials required for ORR, PEDOT is predominantly in its de-doped, neutral, spin-singlet state ¹PEDOT⁰, in our density-functional theory (DFT) modelling^[54] it is the reactivity of this state toward spin-triplet dioxygen ³O₂ that was considered first, taking care of spin conservation and hydration effect. Stable peroxo-adducts to neutral PEDOT could be located at the potential surface, in agreement with spectroscopic evidence, though the formation of such structures was rather endergonic. On the other hand, electron transfer from ¹PEDOT⁰ to molecular oxygen without prior formation of chemical bonds between these two reactants, that is, “outer sphere” (Figure 3) was found favorable, provided proton participation leading to hydroperoxyl HOO^• rather than superoxide O₂^•:

$$$$\begin{eqnarray} {\rm{PEDO}}{{\rm{T}}^0} + {{\mathop{\rm O}\nolimits} _2} + {{\mathop{\rm H}\nolimits} _3}{{\mathop{\rm O}\nolimits} ^ + } \to {\rm{PEDO}}{{\rm{T}}^{ + \bullet }} + {\rm{HO}}{{\rm{O}}^ \bullet } + {{\mathop{\rm H}\nolimits} _2}{\mathop{\rm O}\nolimits} \nonumber\\ \end{eqnarray}$$$$(1)

Indeed, PEDOT, in its reduced (de-doped, neutral) state in the cathodic ORR region, undergoes chemical oxidative re-doping by molecular oxygen,^[47] which should produce superoxide radical anion O₂^•, or its protonated counterpart hydroperoxyl radical HO₂^• though it was not identified.

FIGURE 3

Outer sphere electron transfer as the first stage of oxygen reduction reaction (ORR) at poly(3,4-ethylenedioxythiophene) (PEDOT)

Redox mediation of electron transfer to triplet dioxygen from a singlet organic molecule is reminiscent not so of quinones but of another unique family of molecules, namely flavins. In biological systems, metal-free flavin-dependent enzymes, molecular oxygen activation consists in electron transfer to O₂ from a closed-shell reduced flavin anion, chemisorption prior to charge transfer being clearly spin forbidden.^[55] Quantum chemical modelling^{[56, 57]} further shows that coupled proton and electron transfer (CPET) is an important factor in making these transformations possible.

The oxidized form ²PEDOT^+• is certainly also present as a necessary component according to the Nernst equation, moreover, as a result of electron transfer to dioxygen as expressed above. Note that it is this form that ensures electron conduction of the electrode material. The reactivity of this oxidized form was, therefore, to be considered as well. Dioxygen attachment to ²PEDOT^+• led to endergonic products and was thus found unfavorable, in spite of the radical character of both species. This is in agreement with the well-known air stability of the doped form of PEDOT.

The outer-sphere mechanism whose thermodynamic relevance for a CP was first suggested in our DFT modelling^[54] explains the exclusivity of the 2e ORR pathway followed at PEDOT. Its kinetic feasibility was tackled by us in a subsequent Born-Oppenheimer Molecular Dynamics (BOMD) study that can be considered as a computational reaction chamber, where the system itself selects the appropriate reaction paths, thus following more closely the course of the reaction between a PEDOT oligomer and dioxygen, including all the stages toward H₂O₂, in the presence of explicit water and proton.^[58]

In a system (O₂ + PEDOT⁰ + H₃O⁺ + 11 H₂O), the electron transfer from PEDOT to dioxygen takes place spontaneously in the course of simulation but only when the excessive proton, initially not in the immediate vicinity to PEDOT and dioxygen, travels by Grotthus mechanism from one water to the next until its binding to oxygen. Note that the speed of the reaction in a simulation at room temperature indicates that it is practically barrierless, which we could not infer from the DFT study. The outer sphere reaction (1) thus finds its BOMD confirmation as a proton-coupled electron transfer. We are aware that the small ensembles we were obliged to use do not correspond to the concentrations in the bulk solution. Nevertheless, we believe that they reflect reasonably what may happen locally at the reaction site. Electrode potential was not explicitly taken into account in our simulations, supposing that its effect can be, in the first approximation, limited to changing the ratio between the reduced and oxidized forms of the polymer, according to the Nernst equation.

To comprehend better the role of protonation, it is instructive to consider the “square scheme”^{[59, 60]} for 2e, 2H⁺ ORR in aquo (Figure 4), in the absence of chemisorption, constructed with the thermodynamic data available.^[61-63]

FIGURE 4

Thermodynamics of 2e oxygen reduction reaction (ORR): the square scheme (redox horizontal, protonations vertical). Only relevant redox potentials (vs. SHE) and protonation constants (with their (RT/F) lnKa voltage counterparts) are shown. The values calculated here from those tabulated are marked with a star

Note that the difference in the standard potentials for the first pure 1e reduction of dioxygen to superoxide, -0.35 V (SHE), and its proton-coupled counterpart to hydroperoxyl, -0.07 V (SHE), is due to protonation bringing about (RT/F) lnKa = 59 mV × 4.7 = 0.28 V. This makes the route to hydroperoxyl less cathodic. It can be argued that though hydroperoxyl HO₂ is dominant over superoxide O₂^– in the bulk only at pH < 4.7, it is hydroperoxyl that tends to be formed in the first reduction step in a wide range of pH.

Further spontaneous steps observed in our BOMD simulations^[64] (Figure 5) include:

1. formation of peroxo-adduct PEDOT-OOH, by recombination of hydroperoxyl and oxidized PEDOT on the singlet potential surface (following the spin-flip from the initial triplet)

   $$$$\begin{equation}{\rm{PEDO}}{{\rm{T}}^{ + \bullet }} + {\rm{H}}{{\rm{O}}_2}^ \bullet \to {\rm{PEDOT}} - {\rm{OO}}{{\rm{H}}^ + }\end{equation}$$$$(2)

2. its protonolysis producing the target hydrogen peroxide:

   $$$$\begin{eqnarray} {{ \def\eqcellsep{&}\begin{array}{l} {\rm{PEDOT}}-{\rm{OO}}{{\rm{H}}^ + } + {{\mathop{\rm H}\nolimits} _3}{{\mathop{\rm O}\nolimits} ^ + } \to {\rm{PEDO}}{{\rm{T}}^{2 + }} + {{\mathop{\rm H}\nolimits} _2}{{\mathop{\rm O}\nolimits} _2} + {{\mathop{\rm H}\nolimits} _2}{\mathop{\rm O}\nolimits} \end{array} }}\nonumber\hspace*{-10pt}\\ \end{eqnarray}$$$$(3)

3. proton-coupled outer-sphere reduction of hydroperoxyl to the target hydrogen peroxide by reduced or singly oxidized PEDOT:

   $$$$\begin{eqnarray} {{ \def\eqcellsep{&}\begin{array}{l} {\rm{PEDO}}{{\rm{T}}^0} + {\rm{H}}{{\rm{O}}_2}^ \bullet + {{\mathop{\rm H}\nolimits} _3}{{\mathop{\rm O}\nolimits} ^ + } \to {\rm{PEDO}}{{\rm{T}}^{ + \bullet }} + {{\mathop{\rm H}\nolimits} _2}{{\mathop{\rm O}\nolimits} _2} + {{\mathop{\rm H}\nolimits} _2}{\mathop{\rm O}\nolimits} \end{array} }}\nonumber\hspace*{-10pt}\\ \end{eqnarray}$$$$(4)
   $$$$\begin{eqnarray} {{ \def\eqcellsep{&}\begin{array}{l} {\rm{PEDO}}{{\rm{T}}^{ + \bullet }} + {\rm{H}}{{\rm{O}}_2}^ \bullet + {{\mathop{\rm H}\nolimits} _3}{{\mathop{\rm O}\nolimits} ^ + } \to {\rm{PEDO}}{{\rm{T}}^{2 + }} + {{\mathop{\rm H}\nolimits} _2}{{\mathop{\rm O}\nolimits} _2} + {{\mathop{\rm H}\nolimits} _2}{\mathop{\rm O}\nolimits} \end{array} }}\nonumber\hspace*{-10pt}\\ \end{eqnarray}$$$$(5)

FIGURE 5

Summary of the pathways of oxygen reduction reaction for the peroxide production on poly(3,4-ethylenedioxythiophene) (PEDOT). Reprinted from Copyright (Year), with permission from Elsevier

To close the redox mediation cycle, it is supposed that the singly or doubly oxidized forms of PEDOT are reduced back to neutral by electrons supplied at a necessary cathodic potential.

In support of the stages following the initial outer-sphere electron transfer, it has been observed that, as ORR starts, a new IR band, which is assigned to OOH covalently bound to the beta-carbon of the thiophene ring of PEDOT appears and then disappears as hydrogen peroxide, the final product, is detected; XPS spectra of reduced indicate the formation of carbonyl and/or sulfone group, thus to oxygen attachment to the backbone.^[48] ORR at such materials as polythiophene^[65] or lignin,^[66] apparently incapable of chemisorbing O₂, is believed to proceed via 1e outer sphere reduction followed by disproportionation of O₂^• or its protonated counterpart HO₂^• yielding H₂O₂. In our modelling of ORR at PEDOT, the participation of protons was necessary for the outer-sphere ET to become favorable and happen spontaneously in our MD simulation. Experimentally,^[48] free PEDOT:PSS was producing H₂O₂ at high Faradaic yield and without loss of activity for at least 24 h in acidic and basic solutions alike. This discrepancy needs further studies to be resolved.

5 MODELLING ORR AT BBL

Our DFT study in the joint experimental and theoretical work^[53] was limited to the evaluation of the initial stage of ORR. Note that reduced BBL is assumed to be predominantly in its radical-anion (polaron) spin-doublet state, ²BBL ^•, for which bonding to ³O₂ is not forbidden by spin. One might expect that such radical recombination would be easy and barrierless. Nevertheless, we found that the attachment of dioxygen was endergonic to any site of the ²BBL ^• oligomer. One can rationalize this surprising finding by the delocalized and nonreactive nature of polaronic radicals in conjugated polymers. For n-doped BBL, in particular, this theoretical evidence is certainly in agreement with its well-known atmospheric chemical stability, the absence of its irreversible degradation. On the other hand, computed free energies indicate the reduction of molecular oxygen by n-doped BBL:

$$$$\begin{equation*}{\rm{BB}}{{\rm{L}}^ \bullet } + {{\mathop{\rm O}\nolimits} _2} \to {\rm{BB}}{{\rm{L}}^0} + {{\mathop{\rm O}\nolimits} _2}^ \bullet ,\end{equation*}$$$$

is thermodynamically favorable. Therefore, the initial stage of ORR at BBL ^• apparently follows an outer-sphere pathway, as in the case of PEDOT⁰. Furthermore, this is in agreement with our observation in the same work that exposure to oxygen increases the resistance of BBL indicating partial de-doping of this n-type polymer; in the same vein as PEDOT was partially p-doped (oxidized) in the same conditions. Noteworthy, n-doped BBL is predicted to reduce dioxygen even without protonation of superoxide, which we found crucial for PEDOT in our modelling. Certainly, n-doped BBL is a stronger reductant than neutral PEDOT.

Still, BBL is a more complex molecule than PEDOT, containing imide and carbonyl functional groups and many distinct redox states whose nature is probably not yet fully understood^{[67, 68]} in spite of recent theoretical studies.^{[69, 70]} Consequently, its behavior in ORR is also more complex. In particular, the mechanism of the observed hydrogen peroxide reduction to water, which allowed us to study a model fuel cell with BBL cathode, has not yet been addressed in detail. ORR at BBL was experimentally studied by us in a wide range of pH. The onset of stationary kinetic current (1 mA/cm²) was practically pH independent (which manifested as the shift close to the theoretical 59 mV × pH when plotted vs. reversible hydrogen electrode), and at these low current densities the number of transferred electrons was close to 2 for all pH. This agrees with our modelling results indicating that the starting point for this reaction can be outer-sphere electron transfer to O₂ not involving protons. In acidic media, at more cathodic potentials and consequently higher current densities, the number of participating electrons increases and almost attains 4 at an intermediate pH 3.8. This change might be due to the onset of sequential (2+2)e made possible by the presence of non-dissociated H₂O₂ (pKa 11.7), while its anion HO₂ at pH 12.5 was repelled by negative BBL.

6 CONCLUSIONS AND OUTLOOK

The framework for the current understanding of ORR is largely dominated by its mechanism on electrocatalytic metals, in which dioxygen chemisorption is the first essential step. This is understandable, as formidable experimental and theoretical efforts were deployed to explain and boost the activity of these cathodes for driving 4e ORR in fuel cells. Cleverly engineered metal electrocatalysts can also enforce 2e ORR leading to H₂O₂. Activation of molecular oxygen at these catalysts is achieved by chemisorption prior to electron transfer, and 2e/4e selectivity is an important problem to solve.

In theoretical modelling, uncritical following the pattern of ORR at electrocatalytic metals, where chemisorption of dioxygen precedes electron transfer, leads to imposing various sorts of dioxygen binding to the substrate. While designing such intermediates, care should be taken to conserve spin, given that the ground state of dioxygen is spin-triplet (diradical). It should be noted that while spin conservation is at the center of attention of researchers molecularly, in particular biologically, driven ORR, it is rarely recalled in electrochemical context, apparently because it is indeed irrelevant for metal electrodes.

However, when dioxygen chemisorption is unfavorable due to electrode intrinsic inertness, blocking its surface by ion adsorption, or spin selection rules, ORR can still proceed via outer sphere electron transfer, at least as the initial stage. This ensures that the reaction does not follow direct 4e ORR to water, as this latter pathway requires dissociative (that is OO bond breaking) chemisorption of O2. Rather 2e ORR to H₂O₂ is favored by an outer-sphere mechanism, though subsequent reduction to H₂O via a sequential (2+2)e mechanism cannot be completely excluded. Another practical advantage of the outer-sphere mechanism is to avoid electrode material degradation, passivation, or poisoning. Redox mediation at carbonaceous electrodes oxidized or functionalized with quinones in ORR can be considered as occupying a median position between electrocatalysis at oxygen chemisorbing electrodes and outer-sphere electron transfer at inactive electrodes.

According to our DFT and BOMD modelling, the representatives of both p- and n-type CPs, PEDOT, and BBL drive 2e ORR following a mechanism similar to redox mediation by quinones, in which outer sphere transfer can play an important role. For neutral PEDOT, a weaker oxidant than negative BBL, CPET appears essential. Experimentally, the kinetic signature of CPET in the rate-determining stage is the hydrogen/deuterium isotope effect, which has indeed been observed at molecular ORR catalysts^[71] but not at dispersed Pt.^{[72, 73]} It would therefore be instructive to study isotope effect in ORR at CPs. The formation of intermediates with reactive oxygen species chemically bound to the polymer was also observed in our simulations. The role of CP is, therefore, to be compared to that of molecular materials, including those in living matter, such as flavins, rather than metals. However, a small-molecule redox mediator has its fixed redox potential at which mediation takes place, while electrochemistry of a CP is more complex, possibly opening broader possibilities for electrochemical reactions it drives. Also, molecular porosity and ion exchangeability add to the functionality of CP as electrode materials, such as ISEC. Though special stability studies were not carried out, no degradation in the course of ORR at CP could be observed. Such stability in the presence of reactive oxygen species for an organic unsaturated polymer, not electrochemically inert because undergoing reversible redox transformations, might be due to the metal-free organic electrochemical interface. We believe that CP-based electrode materials have not yet revealed all their possibilities.

ACKNOWLEDGMENTS

We thank the Knut and Alice Wallenberg Foundation through the project “H₂O₂”, Wallenberg Wood Science Center (program 1, 3), the Swedish Research Council (VR 2019-05577, Flexible metal-air primary batteries; VR 2016-05990, Paper fuel cell), and the Swedish Government Strategic Research Area in Materials Science on Advanced Functional Materials at Linköping University (Faculty Grant SFO-Mat-LiU No. 2009-00971).

CONFLICT OF INTEREST

The authors declare that they have no conflict of interest.

Open Research

DATA AVAILABILITY STATEMENT

Not applicable.