Density Functional Theory Mechanistic Study on H2O2 Production Using an Organic Semiconductor Epindolidione

Organic semiconductors have recently emerged as promising catalytic materials for oxygen reduction to hydrogen peroxide, H2O2, a chemical of great importance in industry as well as biology. While examples of organic semiconductor-mediated photocatalytic and electrocatalytic processes for H2O2 production become more numerous and improve in performance, fundamental understanding of the reaction mechanisms at play have been explored far less. The aim of the present work is to computationally test hypotheses of how selective oxygen reduction to H2O2 generally occurs on carbonyl dyes and pigments. As an example material, we consider epindolidione (EPI), an industrial pigment with demonstrated semiconductor properties, which photocatalytic activity in oxygen reduction reaction (ORR) and thereby producing hydrogen peroxide (H2O2) in low pH environment has been recently experimentally demonstrated. In this work, the ability of the reduced form of EPI, viz. EPI-2H (which was formed after a photoinduced 2e–/2H+ process), to reduce molecular triplet oxygen to peroxide and the possible mechanism of this reaction are computationally investigated using density functional theory. In the main reaction pathway, the reduction of O2 to H2O2 reaction occurs via abstraction of one of the hydrogen atoms of EPI-2H by triplet dioxygen to produce an intermediate complex consisting of the radicals of hydrogen peroxide (HOO•) and EPI-H• at the initial stage. HOO• thus released can abstract another hydrogen atom from EPI-H• to produce H2O2 and regenerates EPI; otherwise, it can enter another pathway to abstract hydrogen from a neighboring EPI-2H to form EPI-H• and H2O2. EPI, after reduction, thus plays in ORR the role of hydrogen atom transfer (HAT) agent via its OH group, similar to anthraquinone in the industrial process, while HAT from its amino hydrogen is found unfavorable.


■ INTRODUCTION
Hydrogen peroxide (H 2 O 2 ) is utilized extensively for water treatment, 1,2 chemical synthesis, and in the pulp and paper industry. 3 An emerging application of H 2 O 2 is energy storage as a high energy density liquid fuel that can be used to drive fuel cells. 4−6 At present, most H 2 O 2 is generated at an industrial scale through the oxidation process involving anthraquinone. 7 This complex process requires large plants and transportation of hazardous materials, significant energy input, and generates waste. The three key disadvantages are that this method requires input of H 2 gas, large volumes of organic solvents, and noble metal catalysts. The inherent complexity motivates researchers toward developing new concepts for direct production of H 2 O 2 , among which includes solar-to-fuel conversion via photochemical generation. For instance, generation of H 2 O 2 from O 2 and water is possible using graphitic carbon nitride (g-C 3 N 4 ) by photocatalytic reductions, 8,9 though catalytic stability of these systems has not been tested.
In recent studies from our laboratory, H 2 O 2 was produced by photocatalytic oxygen reduction reaction (ORR) at ambipolar organic semiconductors: low-cost hydrogen-bonded crystals of epindolidione (EPI) and quinacridone (QNC), 10 which are nontoxic and ecofriendly. 11,12 Of these two studied semiconductors, EPI was proven to be a more promising candidate for ORR, providing unique catalytic properties which exhibit performance that far exceeds that from previously reported inorganic semiconductors such as CdS and ZnO. Furthermore, it was subsequently shown that water-soluble analogues of these and other organic semiconductors are ORR photoactive in solution in the absence of aggregation. 13 Therefore, the roots of performance of these materials are to be sought at the molecular level. In the following, we concentrate on the role played by the EPI molecule in photochemical ORR.
EPI is a quinone, just like anthraquinone, and this chemical similarity invites us to compare the reactions of both these molecules leading to peroxide. The anthraquinone process comprises two distinct stages, whose mechanism was attentively studied theoretically only quite recently, though the process itself is quite old. First, anthraquinone is catalytically hydrogenated at a noble metal (Pd); 14 next, the anthrahydroquinone thus obtained is oxidized by molecular oxygen, yielding the target hydrogen peroxide and regenerating anthraquinone. 15 One can legitimately wonder why such a complex procedure was necessary and what is actually the role of anthraquinone. As discussed in a recent and very thorough theoretical study, 16 an enol or a phenol is among rare molecules that can ensure efficient hydrogen atom transfer (HAT) from the hydroxy group to molecular oxygen. Note that the formation of oxygen adduct intermediates has been considered and discarded in both refs 14 and 16. Indeed, the total spin conservation requirement imposes severe constraints on the mechanisms of reactions of triplet dioxygen with singlet molecules in particular (see ref 17 for a discussion). Specifically, direct dioxygen attachment to unsaturated bonds is actually spin-forbidden, as it would lead to very high-energy triplet products, intersystem crossing from a triplet to a singlet pathway being problematic. Therefore, the initial ORR step leading to hydrogen peroxide in the anthraquinone process consists in abstraction of an H atom by O 2 .
Anthraquinone is also known to serve as an electrocatalyst in dark ORR to H 2 O 2 ; when it is grafted onto a glassy carbon electrode, 18 we can reasonably suppose that HAT chemistry from the reduced (hydroquinone) form is at the origin of this electrocatalytic phenomenon.
The photochemical cycle with the participation of EPI is also separated in two distinct stages, though both take place in the same environment: (i) the photoelectrochemical stage of reductive hydrogenation of the EPI-quinone to EPI-biphenol: EPI + 2H + + 2e → EPI-2H, (ii) the chemical reduction of molecular oxygen to hydrogen peroxide with regeneration of the quinone: EPI-2H + O 2 → EPI + H 2 O 2 . 10 The reaction sequence is apparently most similar to the anthraquinone process, and the advantage is the absence of the specific and costly heterogeneous catalyst for the first stage, allowing for the one-pot reaction. Taking for the time being the production of EPI-2H in the first stage for granted, we direct our attention in this work to the second, chemical ORR stage and use density functional theory (DFT) calculations to study whether its mechanism can be similar to that generally adopted for the anthraquinone process.

■ COMPUTATIONAL DETAILS
All calculations were performed using the Gaussian 09 package 19 with a range separated hybrid functional ωB97XD, 20 which comprises 100% long-range Hartree−Fock (HF) exact exchange, a small fraction (∼22%) of short-range HF exact exchange, a modified B97 exchange density functional for shortrange interaction, the B97 correlation density functional, and a model of Grimme's D2 dispersion corrections. 21 The 6-31G(d) basis set was used for all atoms. Here, we employed the SMD implicit aqueous solvation model for obtaining accurate solvation thermochemical properties. Transition states are located using the Berny algorithm. 22,23 All transition-state conformers manifested a single imaginary frequency. To arrive at free energies, the thermochemistry calculations herein were performed at a temperature of 298.15 K and pressure of 1 atm.
Reaction Schemes of ORR by the Reduced EPI-2H. As discussed above, we envisage HAT to triplet oxygen as the initial  ■ RESULTS AND DISCUSSION Reaction Pathway A (= A1 + A2). The free energy profile with the optimized geometries during the oxidation process of EPI-2H for the formation of hydrogen peroxide (H 2 O 2 ) is shown in Figure 2. At the initial stage A1 (of the scheme in Figure 1), 1 EPI-2H is allowed to react with 3 O 2 on a triplet potential energy surface, and we termed such individual reactants as "ind.react" for the respective reaction herein. At first, O 2 forms a weakly bound equilibrated reactant complex with EPI-2H 1 in a common solvent cage. As the reaction proceeds, O 2 directly abstracts a hydrogen atom from one of the enol groups of EPI-2H, thus forming HOȮand EPI-Ḣspecies of the intermediate 2.
In the produced EPI-Ḣ, an unpaired electron is found to be delocalized over the entire aromatic ring with negative charge on oxygen, whereas nitrogen nearby to it acquires positive charge in equal amount due to bond rearrangements, maintaining neutrality of this complex. It is noteworthy that the reaction (ind.react) → 1 → 2 is 0.25 eV exothermic and proceeds with the activation energy of only E act = E TS 1−2 − E ind.react = 0.29 eV (transition state TS 1−2 ). Therefore, it is remarkably facile, which indicates a high reductive capability of EPI-2H toward triplet dioxygen.
We note that intermediate 2 is practically degenerate with completely separated EPI-H • and HOO • (−0.27 eV), which indicates that the interaction between these is weak. EPI-H • and HOO • can thus easily separate. On the other hand, and for the same reason, a rearrangement of intermediate 2 viz. 3 in such a way that HOO • just formed can abstract another hydrogen from the second enol unit of the same organic molecule, without leaving the solvent cage, should be easy. Such a rearrangement is required to enter stage A2 with the second HAT restoring EPI along with the generation of a H 2 O 2 as the final products 4. This process, still on the triplet potential surface, is found to be 0.22 eV exothermic and takes place via a transition state with the activation energy of only E act = E TS 3−4 − E 3 = 0.09 eV. As A1, A2 is thus favorable and facile. However, it should be understood that this entire triplet pathway is doomed, from some point on, to describe the electronically excited state of the system, as it does not lead to the products in the ground singlet state.
Therefore, we also investigated selected points of the singlet reaction pathway, starting from intermediate 2 toward the final products. Note that we have no computational access to the singlet counterparts of 2 and 3, as these are open-shell singlets with essentially multideterminant wave function, not reliably described by conventional DFT. 24 Aware of this, we, nevertheless present the singlet intermediate 2 as an estimate. Indeed, The Journal of Physical Chemistry A pubs.acs.org/JPCA Article the energy of isolated EPI-H • and HOO • (obviously the same for presumed singlet or triplet spin coupling between them at infinite separation) can be a good estimate of their singlet pair in a common solvent cage, as it is quasi-degenerate with their triplet complex. Also, this quasi-degeneracy suggests that the interconversion of weakly bound EPI-H • and HOO • can be presumed feasible, as it only requires the unpaired electrons of separated EPI-H • and HOO • to be recoupled with opposite spins. When the products 4 are in their ground singlet electronic state, their energy is much lower (−2.06 eV, relative to very initial reactants) than that achieved for a very similar geometry on the triplet potential surface. By comparing geometries and energies at the initial triplet and the final singlet pathways, it becomes clear that somewhere between 2 and 3 there lies their point of degeneracy, at which intersystem crossing becomes possible, though we are not in a position to estimate the probability of this event.
Reaction Pathway B. The energy profile in Figure 3 investigates HAT from the enol group of EPI-Ḣintermediate, that can be produced at the stage A1, to 3 O 2 on the doublet potential energy surface. As both reactants meet in a common solvent cage 5, enol hydrogen abstraction takes place without any activation barrier, leading to the formation of products viz. EPI and HOO • 6, and the overall process is exothermic by 1.81 eV as the products are separated.
Reaction Pathway C. As we have demonstrated that 2 HOO • species is generated via A1 or B, it is reasonable to inquire if this radical can abstract H from a neighboring 1 EPI-2H molecule, on the doublet potential surface, via the pathway C (see Figure 4). The energy profile C is found to resemble A2: HOO • species abstracts hydrogen from a enol group of EPI-2H with activation energy of E act = E TS 1−2 − E ind.react = 0.17 eV, and this process is highly exothermic by −1.37 eV, with 2

■ CONCLUSION
Intrigued by the chemical similarity between EPI employed in a recently discovered photochemical production of hydrogen peroxide and anthraquinone, the workhorse of a well-established chemical industrial process, we were wondering if the    16 but reduced EPI was not among them. In the DFT calculations (including implicit water as solvent) presented above, we studied the interaction between molecular oxygen and EPI-2H as well as between the intermediates resulting after the first HAT. The overall mechanism presented herein is found analogous to the one proposed computationally for the industrial production of H 2 O 2 in the anthraquinone reaction. 15 HAT is preceded by the formation of an intermediate complex between the reactants in a common solvent cage, thus preparing them for the target reaction. All of the possible HAT reactions with these are exothermic and, importantly, proceed with negligible activation barrier, being therefore thermodynamically and kinetically facile. These are free radical reactions starting with the triplet diradical molecular oxygen and passing via reactive radical intermediates. Such reactions usually involve a few pathways in parallel, and our results also suggest that there are a few possibilities to arrive at the target products. Importantly, the hydrogen atoms transferred to build up H 2 O 2 originate only from the OH groups of the organic molecule. Indeed, we have observed that hydrogen attached to nitrogen could not participate in HAT, in agreement with the experimental result observed earlier. 10 EPI-2H thus belongs to the family of enols capable to ensure efficient transfer of hydrogen atoms to molecular oxygen via a radical mechanism. The role of the photochemical stage consists in generating reactive EPI-2H from the stable colorant EPI. Understanding these basics will hopefully be useful in designing efficient means to generate hydrogen peroxide from molecular oxygen. Knowing the significance of HAT is a key mechanistic insight which can allow screening of organic dyes and pigments to maximize this feature while balancing good optical and electronic properties.