Oxidation of alkyl benzenes by a flavin photooxidation catalyst on nanostructured metal-oxide films

Significance

There are significant advantages in organic, interfacial photooxidation catalysis. Surface-bound catalysts on oxide interfaces offer highly reactive assemblies that minimize both catalyst and solution volume. We demonstrate here the oxygenation of alkyl benzenes by a flavin mononucleotide on nanoporous ZrO₂ or TiO₂ surfaces.

Abstract

We describe here a surface-bound, oxide-based procedure for the photooxidation of a family of aromatic hydrocarbons by a phosphate-bearing flavin mononucleotide (FMN) photocatalyst on high surface area metal-oxide films.

An important element in synthetic organic chemistry has been the development and application of organic excited states in solution, either by sensitization or electron transfer catalysis (1). Exploitation of organic excited states has evolved from the bench scale to the photochemical reactor scale. In a parallel effort, with a different focus, advances have also been made in exploiting nanoparticle oxide surfaces for energy conversion applications based on chemically bound molecular reactants (2). An advantage of the latter is maximization of the local microscopic surface volume for enhancing efficiencies. We describe here the integration of the two areas with the goal of creating stable photochemical environments that minimize reaction volumes in photooxidation reactions.

Our experiments exploit the excited states of the flavin cofactors found in riboflavin (vitamin B₂), flavin mononucleotide (FMN), and flavin adenine dinucleotide. They are potent photocatalysts and play important roles in biological reactions (3–6). In its redox properties, FMN is more flexible than nicotinamide adenine dinucleotide (NAD) as a donor because it can participate in either 1$e^{-}$ or 2$e^{-}$ transfer reactions (Eq. 1) as opposed to NAD, which undergoes 2$e^{-}$ hydride transfer (7–9).

Analogous to quinones, FMN can exist in three redox states––oxidized (flavin-quinone), one-electron reduced (flavin-semiquinone), and two-electron reduced (flavin-hydroquinone) (8). Riboflavin-based organocatalysis has been explored as a green and economic alternative to metal catalyzed reactions (10) in the photooxidation of alkyl benzenes, amino acids, phenols, saccharides, water, and other substrates (4, 11–17).

In the current study, we have explored an FMN photocatalyst bound to nanoparticle transparent metal-oxide surfaces in the C–H oxygenation of alkyl benzenes. The FMN derivative consists of the iso-alloxazine, ribityl, and phosphate groups as shown in Fig. 1. In the absence of external electron transfer donors, the ribityl group undergoes intramolecular photooxidation, reducing the iso-alloxazine and rendering the riboflavin photosensitive. In the current application, the phosphate group also serves to bind the FMN dye to oxide surfaces.

Fig. 1.

Structure of flavin mononucleotide (Left) and the surface-bound FMN assembly (Right) on the surface of ZrO₂ or TiO₂.

The photophysics and photooxidation of the flavins have been investigated earlier (18). Upon irradiation with blue light, the FMN chromophore forms a lowest-energy singlet-excited state (¹FMN) with high efficiency. It is a potent oxidant with an excited-state reduction potential (FMN*/FM${\text{N}}^{-•}$) of +1.77 V vs. saturated calomel electrode with the ability of oxidizing substrates with potentials <1.8 V (19). In the photooxidation of alkyl benzenes, Megerle et al. have reported that electron transfer (ET) occurs from the triplet state, ³FMN (20, 21), accessed after rapid intersystem crossing (7.8 ns in water) (15, 18). ET to ³FMN, or proton-coupled electron transfer (PCET) (22), oxidizes the alkyl aromatic substrate within 6 μs in 1:1 vol/vol H₂O/MeCN to give a dihydroflavin intermediate (FMNH₂) (23). FMNH₂ is subsequently reoxidized to the ground state with oxygen as the terminal oxidant (17, 24, 25). A mechanism analogous to methoxybenzyl alcohol oxidation by riboflavin tetraacetate in 1:1 vol/vol H₂O/MeCN by Megerle et al. (18) is shown in Eqs. 2–4. In an alternate pathway, formation of an alkyl benzene radical in the presence of oxygen has the capability to initiate radical chains, as the peroxyl radicals are very reactive with their corresponding alkyl benzene form. As pointed out earlier, there is a possibility of contributions in overall catalysis from competitive pathways (26). However, the mechanistic details for the reactions cited here remain to be established in detail, but there is evidence for the involvement of multiple pathways including a possible role for PCET (22, 27). A putative mechanism is shown in Scheme 1.

Scheme 1.

Generalized mechanism for the photooxidation of benzyl alcohol as a representative substrate by an FTO|ZrO₂-FMN surface in the presence of O₂.

The reactivity of flavins, incorporating oxygen as both a reactant and terminal oxidant, makes them good catalysts in aerobic environments. Paradoxically, the use of O₂ as an oxidant in these reactions poses a problem because H₂O₂ is an intermediate (17, 28) and is known to cause degradation of the flavin cofactor resulting in low reaction yields and poor selectivities (15).

To circumvent H₂O₂ as an intermediate, a number of strategies have been explored including the use of transition-metal disproportionation cocatalysts (15, 29), surfactants (30), modification of the flavin (7), and others (11, 31, 32). Mühldorf and Wolf (15) recently added a nonheme iron complex, [Fe(TPA)(MeCN)₂](ClO₄)₂ (TPA = Tris(2-pyridylmethyl)amine) as the cocatalyst (29) for the disproportionation of H₂O₂ ($2{\text{H}}_2{\text{O}}_2\to {\text{O}}_2+2{\text{H}}_2\text{O}$) with notable increases in efficiencies for the photooxidation of alkyl benzenes to ketones and carboxylic acids. Experimental evidence suggested that the increased activity could be attributed to efficient disproportionation of H₂O₂ with a contribution from higher oxidation states of Fe(TPA) in the C–H photooxygenation of the alkyl benzenes.

Extension of the flavin photochemistry to inert surfaces remains largely unexplored. There are examples of flavins immobilized on silica (4) and attached to a polystyrene copolymer (33). However, systematic exploitation of surface binding and flow-through reactors could provide significant synthetic advantages (34). Surface binding to inert, mesoporous oxides minimizes catalyst loading and reaction volumes and subsequently simplifies the reaction conditions and product workup. With transparent metal-oxide supports, a variety of surface-coated reactor designs are available which would be difficult or impossible to achieve with more conventional supports based on silica or polymer beads. In homogeneous catalysis, solubility can also play a major role in the efficiency of a reaction. The majority of the results in the literature on flavin-mediated photooxidations use riboflavin tetraacetate rather than FMN as the photocatalyst because of a gain in catalyst solubility. In heterogeneous catalysis, solubility is no longer a primary factor, allowing the surface-bound FMN to be used with good efficiency.

We report here data on surface binding, stabilization, and reactivity of a phosphate-bearing FMN on high surface area, nanoparticle metal-oxide surfaces. As an example, research in this area has resulted in oxide surfaces that have found applications in photoelectrochemical water splitting (2). Techniques are available for preparing optically transparent, nanoparticle surfaces with dimensions in the tens of micrometers with absorptivities sufficient to drive photochemical reactions with high efficiencies even with intense light sources (2). In the experiments described here, 4-μm films of 10-nm nanoparticles of ZrO₂ or 15–20 nm of TiO₂ nanoparticles on fluorine-doped tin oxide (FTO) were added to coated glass substrates (1.0 cm²). Subsequent addition of the dye to the surface gave films of FTO|ZrO₂|-FMN or FTO|TiO₂|-FMN.

A series of experiments were undertaken to explore the surface properties of the bound dye (Table 1 and SI Methods). Catalyst loading used techniques developed for related photoelectrodes (2). Oxide slides were loaded in 0.5 mM methanol solutions of FMN for up to ∼6 h to achieve the highest surface loadings. The highest surface coverages under these conditions were Γ = 5.7 × 10⁻¹¹ mol cm⁻² (35). By way of comparison, loading of the metal-to-ligand charge-transfer chromophores [M(bpy)₂((4,4′-PO₃H₂)₂bpy)]²⁺ (M = Ru, Os, bpy = 2,2′-bipyridine, and 4,4′-(PO₃H₂)₂ bpy = 4,4′-bis-(phosphonic acid)-2,2′-bipyridine) under the same conditions result in coverages of ∼10⁻¹⁰ mol/cm². Fig. S1 shows the absorption spectrum of a fully loaded FTO|ZrO₂|FMN film with a loading of Γ = 5.7 × 10⁻¹¹ mol cm⁻² in 1:1 vol/vol MeCN/H₂O.

Table 1.

List of control experiments with benzylalcohol to benzaldehyde as the model reaction

Entry	Condition	t, h	Yield, %
1	FTO|ZrO2-FMN/ O2/MnO2/blue LED	2.5	64
2	FMN/O2/MnO2/blue LED	2.5	31
3	No ZrO2-FMN	2.5	1
4	No O2	2.5	3
5	Dark	2.5	0
6	Bare ZrO2	2.5	4
7	No cocatalyst (MnO2)	2.5	16

Fig. S1.

UV-vis of a fully loaded (Γ = 5.7 × 10⁻¹¹ mol cm⁻²) FTO|ZrO₂-FMN in oxygen-saturated MeCN/H₂O (1/1 vol/vol) at room temperature.

In the 4-μm films of the oxides used here, the absorption spectrum of the adsorbed dye at its λ_max of 450 nm is 0.9. For these high surface area oxide films, the ratio of internal to geometric surface area was ∼1,000 (32). The highest catalyst loading was calculated to be ∼5.7 × 10⁻⁸ mol (<0.1 mol%).

Following surface binding, a major challenge in photocatalytic applications was the stability of the oxide-bound catalyst (36). Upon irradiation with visible light (3-W blue LED, 440 nm) in 1:1 vol/vol MeCN/H₂O, the FMN catalyst desorbed from the surface (Fig. 2, red trace) (37), leading to diminished reactivities. To address issues of surface stability, we performed a series of photostability measurements on modified ZrO₂ and TiO₂ metal-oxide surfaces by time-dependent spectral monitoring of the FMN catalyst at 450 nm (Fig. S2).

Fig. 2.

Spectral changes with time at 450 nm following irradiation of 1.0-cm² films loaded with 4-μm-thick FTO|ZrO₂|-FMN films either stabilized with ALD with a 1.0-nm overlayer of Al₂O₃ (black) or unstabilized (red). The data were obtained in 1:1 MeCN/H₂O (vol/vol).

Fig. S2.

UV-vis of FTO|ZrO₂-FMN, interval of 5 min in oxygen-saturated MeCN/H₂O (1/1 vol/vol) (A) with continuous blue LED irradiation and (B) same experiment in 1,2-dichloroethane at room temperature.

Stabilities were assessed by varying solvent polarity in both nitrogen and oxygen atmospheres under LED irradiation. Under most conditions, the catalyst showed full desorption in ∼30 min (Fig. S2). To address the surface stability issue, we attempted to stabilize surface binding on FTO|ZrO₂|-FMN or FTO|TiO₂|-FMN by using a thin coating of a poly(methyl methacrylate) (PMMA) overlayer as described elsewhere for a Ru(II) polypyridyl dye (38). Addition of PMMA did result in greatly increased FMN stability on the oxide surface. Unfortunately, the PMMA layer was unstable in organic solvents and its application was limited to water.

To stabilize FMN on the surface, we used atomic layer deposition (ALD) to coat the derivatized surfaces with a thin 1.0-nm overlayer of Al₂O₃. The ALD procedure resulted in significant enhancements in stability, for periods of up to ∼3 h (Fig. 2, black). ALD has been used extensively for stabilizing chromophores and chromophore–catalyst assemblies on the surfaces of dye-sensitized photoelectrosynthesis cells for water splitting (2, 39).

With the ALD stabilization of FMN established, a platform was available for exploring C–H photooxidation of alkyl benzenes on the oxide films. Selecting the conversion of benzyl alcohol to benzaldehyde as a model reaction (Eq. 5), we explored a variety of conditions with results shown in Table 1; in a series of control experiments, we explored the impact of reaction conditions in 1:1 H₂O:MeCN. Variations with light intensity, O₂, and an added cocatalyst were all explored, which ruled out adventitious photooxidation. Based on these experiments, in the absence of FTO|ZrO₂|-FMN, a ZrO₂ surface was unreactive. In homogeneous solutions with added FMN, conversions were lower by a factor of ∼2 compared with the heterogeneous reactions in part due to the limited solubility of the dye, an additional advantage of the surface-bound reactant. (37)

The reactivity of the adsorbed dye on FTO|ZrO₂|-FMN was explored for the substrates shown in Table 2 using the blue LED with 0.06 mmol of substrate and 8 mol % of added MnO₂ or [Fe(TPA)(MeCN)₂](ClO₄)₂ (Table S1), in 1:1 H₂O:MeCN (vol/vol). Further experimental details are provided in SI Methods.

Table 2.

Substrates for photooxidation by FTO|ZrO₂|-FMN with the added MnO₂ cocatalyst, their corresponding products, reaction times, and product yields in 1/1 H₂O:MeCN (vol/vol)

^*Standard conditions: 0.06 mmol substrate, 8 mol % MnO₂, 1.5 mL 1:1 H₂O:MeCN vol/vol, O₂ saturated, LED (3 W, 450 nm).

^†Yields were determined using NMR with hexamethyldisiloxane (HMDSO) as an internal standard.

^‡Obtained with added 30 mol % HClO₄.

Table S1.

Substrates for photooxidation by FTO|ZrO₂|-FMN with added [Fe(TPA)(MeCN)₂](ClO₄)₂ cocatalyst, their corresponding products, reaction times, and product yields in 1:1 H₂O/MeCN vol/vol

Standard conditions: 0.06 mmol substrate, 8 mol % FeTPA, 1.5 mL 1:1 H₂O:MeCN vol/vol, oxygen-saturated, LED (3 W, 450 nm). The yields were determined using NMR with HMDSO as an internal standard.

Benzyl alcohol and 1-phenylethanol were converted to the respective aldehyde and ketone in good yields (Table 2, entry 1). The substituted toluene and ethylbenzene derivatives gave the corresponding aldehydes and acetophenones in good yield (entries 2 and 3). Isochroman, indane, and fluorene (entries 4–6) were all converted into their ketones in moderate to good yields. Xanthene and thioxanthene generated xanthone and thioxanthone in excellent yields (entries 7 and 8) and, for thioxanthene, very short reaction times. Substrate 4-methoxybenzyl methyl ether yielded the benzyl ester in high yield (entry 9). The conversion of tolane to benzil occurs (entry 10), but in reduced yield similar to the corresponding homogeneous reaction (13). Additionally, trans-stilbene (entry 11) underwent photooxygenation to yield benzaldehyde in good yield. A representative set of NMR data is presented in Figs. S3–S5.

Fig. S3.

Plots of product conversion percent vs. reaction time for (A) ethylbenzene to acetophenone and (B) stilbene to benzaldehyde with FTO|ZrO₂-FMN and MnO₂ in oxygen-saturated MeCN/H₂O 1/1 vol/vol under blue light irradiation (440 nm) at room temperature.

Fig. S5.

NMR spectra showing the appearance of acetophenone upon oxidation of ethylbenzene in the presence of FTO|ZrO₂-FMN, MnO₂, and 30 mol% of HClO₄ in oxygen-saturated MeCN/H₂O 1/1 vol/vol under blue light irradiation (440 nm) at room temperature.

Fig. S4.

NMR spectra showing the appearance of benzaldehyde upon oxidation of stilbene in the presence of FTO|ZrO₂-FMN and MnO₂ in oxygen-saturated MeCN/H₂O 1/1 vol/vol under blue light irradiation (440 nm) at room temperature.

We also investigated FTO|TiO₂|-FMN surfaces to explore the role of the oxide surface on reactivity. In the comparative experiments, the product yield was largely the same between oxides. Although no attempt was made in the reaction sequence to exploit the known redox reactivity of TiO₂, given its accessible conduction band (CB) at E_CB ∼0 V vs. normal hydrogen electrode, there was no evidence for its involvement in surface reactivity of the dye (37). The latter could be an issue with other reactions, but is of less importance for films of FTO|ZrO₂|-FMN given its 0.5-V-higher CB potential for ZrO₂.

We examined the effect of the deuterated version of the solvent on product yield to rule out participation by singlet oxygen (40). Given its known reactivity, ¹O₂*, formed by energy transfer to ³O₂,${}^3\text{FMN}^{*}+{}^3\text{O}_2\to {}^1\text{O}_2^{*}+\text{FMN}$, could play a significant role under our conditions (41). Since the lifetime of singlet oxygen is extended in the deuterated version of the solvent (42), we investigated the oxidation of ethylbenzene, FTO|ZrO₂-FMN, with added MnO₂ for 1 h in 1/1 D₂O/CD₃CN (vol/vol). Analysis of the products gave a conversion yield of ∼56%, compared with 67% for the nondeuterated solvent 67% (Table 2). In a second experiment, the known singlet oxygen sensitizer, Rose Bengal, was used (10 mol %, 0.01 mmol, 9.7 mg) with benzyl alcohol used (0.1 mmol, 10.4 μL) in 7.5 mL of 1:1 vol/vol H₂O/MeCN. Oxygen was continuously bubbled through the solution while irradiating with green light (520-nm LEDs) for 2.5 h. There was no evidence for the hydrocarbon oxidation product under these conditions.

SI Methods

Materials.

All of the substrates were purchased from Sigma-Aldrich and used as received. FMN was purchased from TCI Co. Ltd. in >98% purity and used as received. The cocatalyst MnO₂ was purchased from Fluka and used as received. The synthesis of [Fe(TPA)(MeCN)₂](ClO₄)₂ (TPA = Tris(2-pyridylmethyl)amine) is described in the following sections. A 3-W continuous LED light was purchased from Thor Laboratories provided with a manual light intensity controller.

Preparation of FTO|Oxide-FMN.

FTO was purchased from Hartford Glass with a sheet resistance of 15 Ω/sq. Preparation of nanocrystalline ZrO₂ or TiO₂ over layers on a bare FTO glass was described previously (2). Catalyst loading on the surfaces was achieved by techniques developed for related photoelectrodes (2) by immersing the oxide slides in a 0.5 mM methanol solutions of FMN for up to ∼6 h to achieve the highest surface loadings. FMN adsorption isotherms on FTO were measured by soaking the slides for 6 h in methanol solutions of FMN with concentrations of 0.0125, 0.025, 0.05, 0.25, 0.5, and 0.75 mM. The results of a surface-binding study on FTO as a function of FMN concentration in methanol were analyzed by using the Langmuir adsorption relation in the form Г = Г_max((K[FMN])/(1 + K[FMN])), where Г_max is the maximum surface loading in mol.cm⁻² and K is the surface loading constant (40).

Experimental Setup.

In general, a substrate (0.06 mmol) and cocatalyst (8 mo l%) were directly added to 1.5 mL (acetonitrile/water 1/1 vol/vol) mixture in a 1-dram vial equipped with a magnetic stir bar. FTO|ZrO₂-FMN was immersed in the solution with the setup that allowed full exposure of the visible light from LED light source and the vial was capped with a rubber septum. The mixture was then saturated with oxygen and placed on a stir plate to allow continuous stirring during the course of the reaction.

After the photolysis periods in Table 2, the reaction vials were removed with sample analysis by NMR (unless otherwise noted). A UV-vis spectrum of the solution was acquired after each reaction to test for residual FMN in the reaction mixture. No bands arising from FMN were observed. The products were analyzed by NMR with hexamethyldisiloxane (HMDSO) as an internal standard.

Preparation of Fe(TPA) Complex.

The Fe(TPA) complex was prepared similar to a literature procedure (41). Iron(II)triflate (122 mg, 0.34 mmol, 1 eq) and lithium perchlorate (365 mg, 3.4 mmol, 10 eq) were dissolved in acetonitrile (5 mL). To this solution was added a solution of TPA (100 mg, 0.34 mmol, 1 eq) in acetonitrile (1 mL). The reaction was stirred at room temperature for 3 h and then precipitated in ether. The solid was taken up again in acetonitrile and reprecipitated in ether to give the product as a microcrystalline red solid. A ¹⁹F NMR confirmed that there was no remaining triflate. The ¹H NMR data are in agreement with the previous literature report: (CD₃CN, 400 MHz) δ 10.93 (s, 3H), 8.49 (m, 6H), 7.37 (t, 3H), 6.35 (s, 6H) ppm.

UV-Vis Spectra.

All of the absorption spectroscopy was performed using an Agilent 8453 UV-visible spectroscopy system. The resulting data were exported into OriginPro 8.0 for further processing.

Conclusions

We have demonstrated here a procedure for a minimum-volume, high-efficiency, organic photochemical reactor. It is based on a surface-bound, flavin-based chromophore on high surface area oxide films. This configuration provides a basis for stable photoredox catalysis in an ultrathin, high surface area nanoparticle reactor.

In the example investigated here, a phosphate-bearing flavin mononucleotide dye was bound to high surface area ZrO₂ or TiO₂ films with surface stabilization by ALD. The modified metal-oxide films demonstrated excellent catalytic activity following irradiation with a 440-nm blue LED. The experiments demonstrate the viability of the approach, one that could be of general value in organic photochemical reactions under conditions that minimize reaction volumes and the time for reaction workup.

For the FMN reactions, the reactivity of the dye on the high surface area oxide films was maintained with high yields obtained for a series of hydrocarbon oxidations. The results presented here provide a basis for extension to larger scales and to catalyst–oxide assemblies with potential applications for larger-scale organic reactions.

Methods

Detailed spectral data and analysis are included in SI Methods.