Molecular Copper(I)‐Sensitized Photoanodes for Alcohol Oxidation under Ambient Conditions

Abstract

Dye‐sensitized photoelectrochemical cells can enable the production of molecules currently accessible through energetically demanding syntheses. Copper(I)‐based dyes represent electronically tunable charge transfer and separation systems. Herein, we report a Cu(I)‐bisdiimine donor‐chromophore‐acceptor dye with an absorbance in the visible part of the solar spectrum composed of a phenothiazine electron donor, and dipyrido[3,2‐a:2′,3′‐c]phenazine electron acceptor. This complex is incorporated onto a zinc oxide nanowire semiconductor surface effectively forming a photoanode that is characterized spectroscopically and electrochemically. We investigate the photo‐oxidation of hydroquinone, and the photosensitization of 2,2,6,6‐tetramethylpiperidine‐1‐oxyl and N‐hydroxyphthalimide for the oxidation of furfuryl alcohol to furfuraldehyde, resulting in near quantitative conversions, with poor selectivity to the alcohol.

A donor‐Cu(I)‐acceptor visible light photosensitizer is synthesized and characterized. Upon light irradiation, an electron/hole pair is formed first by metal‐to‐ligand charge transfer. The ligands are tailored such that the hole is delocalized into a phenothiazine donating moiety, and the electron is delocalized in a phenazine accepting moiety, leading to a charge separated state, capable of driving photoelectrochemical reactions.

Introduction

Fossil fuels and the associated refining processes provide useful chemical feedstocks to various industries. While there has been an increased focus on decarbonization and on sustainable H₂ production as a fuel (amongst others), mechanisms to bridge the cost of production and market value are needed.[ ¹ , ² ] By coupling proton reduction to high value‐added oxidative processes, it is possible to overcome this cost barrier, while also producing new streams of chemical feedstocks. ^[3] Dye‐sensitized photoelectrochemical cells (DS‐PECs) mimic the natural process of photosynthesis by storing solar energy in chemical bonds. It is possible to engage molecular donor‐chromophore‐acceptor systems as dyes in these devices to reach long‐lived charge‐separated states (CSSs) which are used to drive various transformations,[ ⁴ , ⁵ , ⁶ ] such as water oxidation, or the oxidation of alcohols to aldehydes via the stable radical (2,2,6,6,–tetramethylpiperidine‐1‐yl)oxyl (TEMPO) catalyst.[ ⁷ , ⁸ ] In the former case, in the absence of an appropriate water oxidation catalyst, the possibility to oxidize water can be approximated using hydroquinone (H₂Q) as a surrogate, also requiring a multiple electron oxidation (2 e⁻ and 2 H⁺) to form the oxidized benzoquinone (BQ), making it a better model for water oxidation than other common sacrificial reductants with single e⁻ oxidations.[ ⁹ , ¹⁰ ] The latter case of alcohol oxidation is typically performed at elevated temperatures, however through sensitization of TEMPO with a photoanode, the nitrosonium cation is generated, which can then drive the oxidation of alcohols photocatalytically. ^[11]

Herein, we report a photoelectrode consisting of a copper(I)‐bisdiimine donor‐chromophore‐acceptor triad D‐C‐A dye, with a heteroleptic copper(I) chromophore (C) containing a phenothiazine (PTZ) electron donating moiety on a 2,9‐dimethy‐1,10‐phenanothroline scaffold (D), and a dipyrido[3,2‐a:2′,3′‐c]phenazine (dppz) electron accepting moiety (A). Furthermore, the acceptor ligand A is modified to incorporate a carboxylic acid group, allowing for the dual purpose of anchoring the D‐C‐A system to a semiconductor surface, as well as accepting electrons to direct charge transport (Figure 1). ^[12]

Figure 1.

Proposed structures of the triad D−C‐A, and dyad C−A.

Results and Discussion

Synthesis and Ground State Characterization

D and A were synthesized using modifications of previously reported literature procedures (Scheme S1 and S2),[ ¹² , ¹³ ] and the copper(I)‐based D‐C‐A triad, and model C‐A dyad, were synthesized using a one‐pot, two‐step reaction via the sequential addition of ligands to a copper(I) containing solution, mimicking the HETPhen approach (Scheme S3).[ ¹⁴ , ¹⁵ ] The complexes were isolated as hexafluorophosphate salts and the UV‐Vis spectra exhibit broad metal‐to‐ligand‐charge‐transfer (¹MLCT) absorption band centered at ca. 460 nm (Figure 2A), as expected. ^[16] Both the D‐C‐A triad and C‐A dyad absorbance spectra are effectively a sum of the A and D ligands, indicating a weak ground‐state interaction between the ligands and the metal center.

Figure 2.

(A) UV‐Vis spectra of D, A, C−A, and D−C‐A in 9 : 1 v/v CH₂Cl₂:CH₃OH at room temperature. Inset: UV‐Vis transmittance spectra of ZnO NWs (black) and ZnO NWs|A‐C‐D (red) and UV‐Vis absorption spectrum of D−C‐A in 9 : 1 v/v CH₂Cl₂:CH₃OH at room temperature (blue). (B) Cyclic voltammograms of A, D, and D−C‐A in 0.1 M [n‐Bu₄N]PF₆ in CH₃CN (scan rate=100 mV/s, 2 mm Pt button working electrode, Pt mesh counter electrode, Ag wire pseudoreference electrode, and ferrocene as an internal standard; converted to NHE by adding +0.630 V).

Electrochemical Characterization

Cyclic voltammetry was performed on the complex and the ligands to establish ground‐state electrochemical properties (Figure 2B, Table 1). In the cyclic voltammogram of donor D, a reversible wave is observed at E _1/2(D^I/0)=+0.79 V [potentials recorded versus ferrocene internal standard and reported versus normal hydrogen electrode (NHE)], corresponding to a one electrode oxidation of the phenothiazine.[ ¹⁷ , ¹⁸ , ¹⁹ ] Two waves at E _p/2(A^0/−I)=−0.52 V and E _1/2(A^−I/−II)=−0.87 V for A were observed, corresponding to two one‐electron reductions of dppz. ^[12] In the D‐C‐A triad, a well‐defined reversible oxidation wave was observed at E _1/2=+1.11 V and was assigned to the Cu(I)/Cu(II) redox couple. The presence of the donor moiety (D) in the D‐C‐A triad is confirmed by the wave at E _1/2=+0.77 V, and the presence of the acceptor moiety (A) is observed by the two overlapping reductions which were deconvoluted by differential pulse voltammetry (DPV) with two reductions at E _p/2=−0.75 V and −1.11 V (Figure S1).

Table 1.

Oxidation and reduction potentials of A, D, and D‐C‐A.

Compound	E 1/2(DI/0)	E 1/2(CuII/I)	E 1/2(A0/−I) [E p/2(A0/−I)]	E 1/2(A−I/−II) [E p/2(A−I/−II)]	ΔGet	ΔGht
D	+0.79 V[a]
A			[−0.52 V][a]	−0.87 V[a]
D‐C‐A	+0.77 V[a]	+1.11 V[a]	[−0.75 V][b]	[−1.11 V][b]	−0.20 eV	−0.34 eV

[a] These values are calculated using cyclic voltammograms recorded in 0.1 M [n‐Bu₄N]PF₆ in CH₂Cl₂, scan rate: 100 mV/s, 2 mm Pt button working electrode, Pt mesh counter electrode, Ag wire pseudoreference electrode, and ferrocene as an internal standard and converted to NHE by adding +0.630 V. [b] These values are calculated using differential pulse voltammetry in 0.1 M [n‐Bu₄N]PF₆ in dimethylformamide with pulse height: 10 mV; width: 50 ms; period 200 ms; increment: 4 mV, sampling pre‐pulse width: 5 ms; post‐pulse width: 5 ms, 2 mm Pt button working electrode, Pt mesh counter electrode, Ag wire pseudoreference electrode, and ferrocene as an internal standard and converted to NHE by adding +0.630 V

The Gibbs free energies of the charge‐transfer processes between the copper center and the ligands have been determined using these data. First, the excited‐state oxidation potential for the MLCT of the copper(I) triad is estimated using the equation E _1/2(Cu^II*/I)=E _1/2 (Cu^II/I)−E ₀₀, where E ₀₀ is the zero‐zero energy of the ³MLCT excited state calculated from the wavelength at the intersection of the absorbance of the complex and photoluminescence spectra of the reference [Cu(dmp)₂][PF₆], dmp=2,9‐dimethyl‐1,10‐phenanthroline complex (602 nm, 2.06 eV, Figure S2) as the D‐C‐A complex is non‐emissive within the visible region of light. ^[20] Once E ₀₀ is determined, the thermodynamic driving force for photoinduced electron transfer to the acceptor ligand A can be approximated according to the equation ΔG_et=E _1/2(Cu^II*/I)−E _1/2(A^0/−I). Similarly, the photoinduced hole transfer to the donor ligand D can be calculated using the equation ΔG_ht=E _1/2(D^I/0)−E _1/2(Cu^II/I). From these calculations it was found that charge shift from the ³MLCT state of the chromophore to A is thermodynamically favourable (ΔG_et≈−0.20 eV). ^[21] The hole transfer from D to the chromophore, regenerating the ground state of copper, is also thermodynamically favourable (ΔG_ht≈−0.34 eV, Table 1). We hypothesize that upon photoexcitation and intersystem crossing, the ³MLCT state is formed at the Cu(I) chromophore followed by oxidative quenching of the excited state by the acceptor A dppz moiety (or reductive quenching by D) and regeneration of the ground state copper by a charge shift from the donor D phenothiazine moiety (or A), yielding a final ^+⋅ D‐C‐A^⋅− CSS with an approximate energy of 1.52 eV as calculated from the ground state electrochemical data (Figure 2B, Table 1) corresponding to a photoluminescence maximum of ca. 816 nm, outside of the visible window.

Density Functional Theory Calculations

The energetically favourable charge transfer from A and D potentially resulting in the ^+⋅ D‐C‐A^⋅− CSS was partially confirmed by density functional theory (DFT) calculations of the ground state D‐C‐A triad (computed frontier molecular orbitals depicted in Figure S3). The highest occupied molecular orbital (HOMO) is localized on the donor ligand D, while the lowest unoccupied molecular orbital (LUMO) is localized on the acceptor ligand A. The separation of the HOMO and LUMO confirms the success of our triad design for a long‐lived CSS and directional electron transfer. Upon excitation, we anticipate charge shift will lead to an electron localized on A which also contains a carboxylic acid anchoring group, allowing for charge injection into the semiconductor conduction band, leaving behind a hole on the donor, resulting in a photoanode capable of oxidative follow‐on chemistry.

Excited State Characterization

Femtosecond transient absorption (fsTA) measurements were carried out on dyad C‐A (Figure 3A–B) and triad D‐C‐A (Figure 3D–E) complexes to probe the evolution of the excited state manifold. The results were corrected for dispersion (chirp) and analyzed by a global fitting method based on four exponential components for which time constants were calculated (Figure 3C and F, Table 2).

Figure 3.

Femtosecond transient absorption difference spectra for (A, B, C) C‐A and (D, E, F) D‐C‐A at (A, D) short time delays (0.3‐10 ps), (B, E) long time delays (20‐8000 ps), and (C, F) kinetic analysis displaying the decay associated spectra. Spectra acquired in CH₂Cl₂, λ_ex=460 nm, 2 μJ/pulse, spectra are discontinuous between 460 and 475 nm due to the pump pulse.

Table 2.

Time constants of the excited‐state dynamics for C‐A and D‐C‐A from the global fitting analysis (λ_ex=460 nm, 2 μJ).

Compound	τ1	τ2	τ3	τ4
C‐A	1 ps	50 ps	328 ps	7634 ps
D‐C‐A	1 ps	42 ps	395 ps	6282 ps

Qualitatively, the fsTA spectra of the dyad C‐A, show a bleach associated with the ground‐state ¹MLCT absorption band as demonstrated by negative ΔA (centered around the pump pulse at ca. 460 nm). Positive excited‐state absorption (ESA) features at ca. 370 nm are consistent with radical anion 1,10‐phenanthroline absorptions, ^[22] and ESA features extending from 610 nm onwards are consistent with the spectral character previously attributed to the radical anion of the dppz acceptor A moiety.[ ¹² , ²³ , ²⁴ , ²⁵ , ²⁶ ] On global analysis, four separate transitions are proposed. Initial features that form rapidly (τ₁=1 ps, Figure 3C) correspond to pseudo‐Jahn‐Teller (pJT) distortions where the coordination sphere converts from tetrahedral to a square planar geometry [as copper(I) becomes copper(II) in the initial ¹MLCT state],[ ²² , ²⁷ , ²⁸ ] with spectral features that are qualitatively similar to related [Cu(Xantphos)(dppz)]⁺ [Xantphos=(9,9‐dimethyl‐9H‐xanthene‐4,5‐diyl)bis(diphenylphosphane)] complexes. ^[26] The negative signal observed is due to the growth of the following observed species, as is demonstrated by the mirrored ESA in τ₂. At τ₂=50 ps, (Figure 3C) the formation of the radical anion on the 1,10‐phenanthroline ligand is observed, and is attributed to ¹MLCT→³MLCT intersystem crossing (ISC).[ ²⁹ , ³⁰ ] The third time component may be associated with an electron density shift on the acceptor A ligand corresponding to a shift from a proximal (³MLCT_prox) to a distal (³MLCT_dist) dppz‐associated state (τ₃=328 ps), as it has a similar ESA profile (broad absorbance from 500–700 nm) and timescale as has been reported in analogous [Cu(Xantphos)(dppz)]⁺complexes.[ ¹² , ²⁶ ] The fourth component showing an absorbance ca. 570 nm (τ₄=7634 ps) is attributed to the long lived ESA of a homoleptic [Cu(dmp)₂][PF₆] or related impurity in the sample.[ ³¹ , ³² ]

In the fsTA spectra of the triad D‐C‐A, we observe similar excited state features to the dyad (Figure 3D−F), however an increase in the ESA feature observed at 410 nm is consistent with the ESA of PTZ,[ ³³ , ³⁴ ] with which the donor D is constructed. Other characteristic ESA of PTZ occur at ca. 460 nm,[ ³⁵ , ³⁶ , ³⁷ ] which is difficult to deconvolute from the ground state ¹MLCT bleach region and the pump pulse. Qualitatively, a lower intensity ESA is observed ca. 525 nm in D‐C‐A compared to C‐A. Furthermore, global analysis (Figure 3F) yields similar species with similar lifetimes (Table 2). In particular, the first three components exhibit very similar spectral characteristics, while the fourth component differs in ESA and lifetime between the two compounds. Instead of being due to potential [Cu(dmp)₂][PF₆] (the D‐C‐A contains no dmp), the relative increased magnitude of ESA ca. 400 nm is attributed to the PTZ radical cation. Taken in conjunction with orbital locations as determined by DFT calculations (Figure S3) and exergonic ΔGs for electron and hole transfer to the ligands in the excited state (Table 1) the validity of the chosen compound for potential applications as a photosensitizer for DS‐PECs is bolstered.

Photoelectrochemistry

To probe the utility of using these metal complex photosensitizers in a photoelectrochemical cell, a photoanode was constructed from solvothermally grown ZnO NWs on FTO glass (hydrothermal synthetic procedure for nanowire growth is described in the Experimental Section) followed by soaking in a solution containing the D‐C‐A complex as synthesized (10 mM in 9 : 1 v/v CH₂Cl₂:CH₃OH) whereby the D‐C‐A is anchored through the carboxylic acid anchoring group on the A ligand. The resulting ZnO|A‐C‐D films were characterized by UV‐Vis spectroscopy and display an ¹MLCT band comparable to that observed in the D‐C‐A in solution (Figure 2A inset). ^[12] Dissolving films in 1 M HCl, followed by extraction into 9 : 1 v/v CH₂Cl₂ : CH₃OH afforded an approximation of dye loading of 6.3±1.4 nmol per film (1 cm² working area, averaged across three films prepared under identical conditions).

Cyclic voltammetry performed on H₂Q and pyridine (Py) confirms an oxidation onset potential of E(H₂Q)=+0.80 V vs NHE (Figure S4), suggesting that hole transfer from the donor moiety D to H₂Q is thermodynamically unfavourable [ΔG_ht=E(H₂Q)−E _1/2(D^I/0)≈+0.03 eV]. This suggests that the prepared D‐C‐A complex is on the cusp of being capable of producing enough driving force to oxidize H₂Q to BQ.

Photoelectrochemistry measurements were performed on the as‐fabricated photoelectrodes using a three‐electrode setup, composed of ZnO|A‐C‐D as the working electrode illuminated with a single white light Cree CXB3590 LED (540 mW/cm²) from the back of the film through the glass substrate, in 0.1 M [n‐Bu₄N]PF₆ (in CH₃CN). Linear sweep voltammetry (LSV) measurements were performed with a 10 mV/s sweep rate and were performed under light and dark conditions (Figure 4A). In the presence of only electrolyte, there is no noticeable increase in current with increasing potential, regardless of illumination. Following the addition of H₂Q, the current increases with increasing potential, indicating that with sufficient overpotential, our device is capable of H₂Q oxidation. Furthermore, irradiation by white light lowers the onset potential from ca. +800 mV to ca. 0 mV vs Ag wire (approximately +980 mV to +180 mV vs NHE), demonstrating the ability of the device to photocatalytically oxidize H₂Q with low overpotential, despite the unfavourable ΔG_ht. This onset is further shifted to a less oxidative potential upon the addition of pyridine (Py) which acts as a base, abstracting protons from H₂Q upon oxidation (Figure 4A, solid blue). The increased ease of oxidation in the presence of a base suggests a proton‐coupled electron transfer (PCET) mechanism for the oxidation of H₂Q.[ ³⁸ , ³⁹ ]

Figure 4.

(A) Linear sweep voltammetry of a ZnO|A‐C‐D working electrode in electrolyte solution (black), with the addition of 5 mM hydroquinone (H₂Q) (red), and 5 mM pyridine (Py) (blue), in the dark (dashed lines) and under illumination (solid lines, scanning potentials from −0.2 V to 1 V vs Ag wire, with ferrocene as an internal standard; converted to NHE by adding +0.630 V, scan rate=10 mV/s). (B) Chopped light chronoamperometry measurements of a ZnO|A‐C‐D working electrode (applied bias of 400 mV vs Ag wire, 580 mV vs NHE) with chopped illumination by a single LED (540 mW/cm² white light with 400 nm long pass filter) from the back of the film. (black) with the addition of H₂Q (red), and Py (blue). All measurements carried out in 0.1 M [n‐Bu₄N]PF₆ in CH₃CN.

Chopped light chronoamperometry were performed with the same three electrode setup with an applied bias of 400 mV vs Ag wire (580 mV vs NHE), and films were cycled between light and dark periods of 20 seconds each (40 second cycle total, Figure 4B, S5). Negligible photocurrent is observed upon light cycling of a bare ZnO film (Figure S5, black), while ZnO|A‐C‐D is capable of producing substantial photocurrents (ca. 12 μA/cm² Figure S5, red), demonstrating the need for a visible light absorbing photosensitizer to produce photocurrent.

An increase in the magnitude of the photocurrent density in the presence of H₂Q is observed (ca. 26 μA/cm², Figure 4B, red) and is indicative of a higher yield of electron injection into the semiconductor. ^[40] Addition of Py to the solution results in further increased photocurrent densities (ca. 45 μA/cm², Figure 4B, blue) mirroring the increased currents and decreased oxidation onset seen in the LSV experiments. The presence of Py also reduces the capacitive currents observed, indicating increasingly rapid D‐C‐A regeneration kinetics to the ground state. The photoanodes were also tested under aqueous conditions in pH 7, 0.1 M KPi buffer (Figure S6) achieving photocurrents of an order of magnitude lower than in organic conditions (ca. 4.5 μA/cm² for ZnO|A‐C‐D in the presence of H₂Q). Similar experiments performed on a dyad photoanode (ZnO|A‐C) show reduced photocurrent compared to the triad containing photoanode (Figure S7‐8), demonstrating the importance of having both an accepting moiety A and donating moiety D to reduce charge recombination rates by increasing the spatial charge separation of the electron‐hole pair in the CSS.

Photoanodes were also tested in a similar manner for their potential to photosensitize TEMPO. Hole transfer from the donor D to TEMPO is thermodynamically favourable (ΔG_ht≈−0.16 eV) and we theorized that photooxidation of TEMPO would form the nitrosonium radical, which has been shown to act as a co‐catalyst for alcohol oxidation.[ ⁴¹ , ⁴² , ⁴³ ] LSV using ZnO|A‐C‐D as a working electrode demonstrates that upon photoirradiation the onset potential for TEMPO oxidation is lowered from ca. 730 mV to ca. 530 mV vs NHE. Similarly, the onset of the oxidation of benzyl alcohol (BA, Figure S7) is lowered upon illumination. ^[11] Chopped light chronoamperometry measurements appear promising for TEMPO sensitization towards the oxidation of benzyl alcohol (Figure S7), however upon utilizing the photoanodes under previously reported benzyl alcohol oxidation conditions, ^[11] no appreciable yield of the corresponding aldehyde was observed. Complementary experiments with a +400 mV applied bias vs Ag wire demonstrate a decrease in measured photocurrent in the presence of TEMPO, suggesting system complexity needing analyses beyond simple redox leveling (Figure S8). Similarly, N‐hydroxyphthalimide (NHPI) can be oxidized to the phthalmide N‐oxyl radical to perform the oxidative transformations, and has been previously sensitized with Cu(I) sensitized photoelectrodes. ^[12] When applied to furfuryl alcohol oxidation, little to no yield (up to 2 % for FA oxidation) of the corresponding aldehyde is observed, however the corresponding alcohol signals were not observable by ¹H NMR using an internal standard, suggesting the photoanode is capable of driving the transformation of the chosen alcohols to an (in our hands) indeterminate product.

Conclusions

Many systems have been explored for alcohol oxidation, most of which either require elevated temperature, or the use various oxidants to drive the process,[ ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹ ] which drive up the cost of such experiments and complicate product isolation. Although the photocurrent densities achieved by the reported photoanode are low compared to many previously reported photoelectrodes (>150 μA/cm²),[ ⁷ , ⁸ , ¹¹ , ⁵⁰ ] we are able to successfully oxidize furfuryl alcohol albeit with poor selectivity towards the corresponding aldehyde (2 % yield of furfuraldehyde), without the use of extra sacrificial oxidants or expensive metal catalysts. Experiments are ongoing to determine the potential products of the alcohol oxidation studies. Furthermore, the modest photocurrents achieved by these photoanodes towards H₂Q oxidation in both organic and aqueous media, demonstrates a potential for a versatile, non‐specific, photoanode device capable of photosensitizing a variety of oxidation reactions, provided the oxidation potential lines up favourably with the excited state of the D‐C‐A.

Experimental Section

Materials

Furfuryl alcohol (FA) was purchased from Alfa Aesar. 1,3,5‐Trimethoxybenzene, dichloromethane, glacial acetic acid, nitric acid, palladium(II) acetate, potassium bromide, potassium chloride, potassium phosphate monobasic, potassium phosphate tribasic, pyridine, sulfuric acid, tetrahydrofuran, and zinc nitrate hexahydrate were purchased from Fisher Scientific. 2,9‐Dimethyl‐1,10‐phenanthroline (dmp), 4‐bromobenzaldehyde, acetonitrile, ceasium carbonate, fluorine doped tin oxide (10 Ω) on glass substrates (FTO), hexamethylenetetramine (HMTA), phenothiazine (PTZ), (2,2,6,6‐tetramethylpiperidin‐1‐yl)oxyl (TEMPO), tetrakis(acetonitrile)copper(I) hexafluorphosphate, and zinc chloride were purchased from Sigma‐Aldrich. 2,2‐Bis(diphenylphosphino)‐1,1‐binaphthyl (BINAP+), 3,4‐diaminobenzoic acid, ammonium acetate, and hydroquinone were purchased from TCI. Tetrabutylammonium hexafluorophosphate [n‐Bu₄N]PF₆ was purchased from AmBeed and was recrystallized three (3) times from ethanol prior to use. Ammonium acetate was sublimated using a sand bath and room temperature cold finger at 100 °C and 100 mbar for 16 hours. All other chemicals were used as received.

Synthesis

Synthesis of A (Scheme S1): The synthesis of acceptor ligand A was performed as follows. First, dmp (1.00 g, 4.8 mmol) and concentrated sulfuric acid (16 mL, 18.3 M) were added to a 50 mL round bottom flask and stirred at 0 °C for 15 minutes. Potassium bromide (5.60 g, 47 mmol) was then slowly added, followed by the dropwise addition of concentrated nitric acid (7 mL, 15.7 M). A condenser was added with an aqueous sodium sulfite trap and the solution was heated at 80 °C for 3 hrs. The reaction was allowed to cool to room temperature, acid was neutralized with saturated aqueous sodium bicarbonate solution, following which 2,9‐dimethyl‐1,10‐phenanthroline‐5,6‐dione was extracted with three washes of 20 mL of CH₂Cl₂. The organic layer was washed with brine three times, dried over sodium sulfate, filtered, and the solvent was evaporated, leaving a brown product. ^[13] 2,9‐Dimethyl‐1,10‐phenanthroline‐5,6‐dione (150 mg, 0.63 mmol) and 3,4‐diaminobenzoic acid (96 mg, 0.63 mmol) were added to a 50 mL round bottom flask and dissolved in ethanol (10 mL). The solution was refluxed for 4 hours, and the resulting solid was filtered, washed, and dried under vacuo yielding acceptor (A) the identity of which was confirmed by comparison to literature. ^[12]

Synthesis of 4‐(10H‐phenothiazin‐10‐yl)benzaldehyde (Scheme S2): PTZ (800 mg, 4 mmol), 4‐bromobenzaldehyde (820 mg, 4.4 mmol), and cesium carbonate (400 mg, 1.2 mmol) were added to a 50 mL round bottom flask and a vacuum/nitrogen cycle was performed three times. Toluene (20 mL) was added and sparged with nitrogen for 5 minutes, following which palladium(II) acetate (225 mg, 0.4 mmol) and BINAP+ (500 mg, 0.8 mmol) were added under positive nitrogen pressure. The flask was covered with aluminum foil and the solution refluxed for 24 hours. Toluene was evaporated using rotary‐evaporation, and the product was dissolved in minimal CH₂Cl₂ and passed through a silica column with a CH₂Cl₂ mobile phase. The second dark yellow/orange band contains the product which fluoresces yellow on silica TLC plates under long wave UV light in 44 % yield.

¹H NMR (500 MHz, CDCl₃) δ 9.85 (s, 1H), 7.78–7.71 (m, 2H), 7.47–7.39 (m, 2H), 7.33–7.22 (m, 4H), 7.21–7.13 (m, 4H) ppm.

¹³C NMR (126 MHz, CDCl₃) δ 190.4, 150.2, 141.1, 132.4, 131.6, 130.4, 128.7, 127.2, 125.9, 125.3, 116.9, 77.3, 77.0, 76.7 ppm.

HR‐MS (ESI+): m/z=303.0720, calcd for C₁₉H₁₃NOS m/z=303.0718.

Synthesis of D (Scheme S2): 2,9‐Dimethyl‐1,10‐phenanthroline‐5,6‐dione (400 mg, 1.7 mmol), 4‐(10H‐phenothiazin‐10‐yl)benzaldehyde (600 mg, 2.0 mmol), and ammonium acetate (3.4 g, 44 mmol) were loaded into a 100 mL round bottom flask and a vacuum/nitrogen cycle was performed three times. Glacial acetic acid (7.5 mL) and CH₂Cl₂ (50 mL) were added and the resulting solution was refluxed under nitrogen atmosphere for 24 hours. The solvents were evaporated under reduced pressure and the product was dissolved in minimal amounts of CH₂Cl₂ and precipitated with 10× volume of diethyl ether. The donor D was isolated by filtration, rinsed with diethyl ether, and dried under vacuo with a 37 % yield. ^[13]

¹H NMR (500 MHz, (CD₃)₂SO) δ 8.80 (t, J=8.4 Hz, 2H), 8.62 (d, J=8.1 Hz, 1H), 8.48–8.43 (m, 1H), 8.08 (dd, J=7.7, 1.6 Hz, 1H), 7.75–7.68 (m, 3H), 7.63–7.54 (m, 2H), 7.33 (t, J=7.4 Hz, 1H), 7.17 (dd, J=7.7, 1.6 Hz, 1H), 7.04 (td, J=7.8, 1.6 Hz, 1H), 6.95 (td, J=7.5, 1.2 Hz, 1H), 6.82 (d, J=8.6 Hz, 1H), 6.49 (dd, J=8.2, 1.2 Hz, 1H), 2.80 (d, J=3.5 Hz, 6H) ppm.

¹³C NMR (126 MHz, (CD₃)₂SO) δ 156.46, 149.98, 149.57, 143.59, 142.41, 142.16, 139.93, 138.72, 133.43, 132.19, 131.48, 131.14, 129.51, 129.00, 127.91, 127.50, 124.20, 123.84, 123.05, 122.72, 121.85, 118.06, 117.41, 24.96 ppm.

HR‐MS (ESI+): m/z=522.1744, calcd for [C₃₃H₂₃N₅S]⁺ m/z=522.1752.

Synthesis of D‐C‐A (Scheme S3): A (35.4 mg, 105 mmol) was added to a CH₂Cl₂:CH₃OH mixture (10 mL, 9 : 1 v/v) which was then sparged with nitrogen for 10 minutes. Following which tetrakis(acetonitrile)copper(I) hexafluorophosphate (37.2 mg, 100 mmol) was added under positive Ar pressure, and stirred for 30 minutes before adding D (52.2 mg, 100 mmol) and stirring for an additional 30 minutes at room temperature. The solution was filtered with a 0.2 μm syringe filter to remove excess insoluble A. Half (5 mL) of this solution was diluted to 10 mL with CH₂Cl₂ and used for device fabrication. The solvent was removed from the other half (5 mL) of solution, yielding a dark red solid D‐C‐A, and was used for characterization.

¹H NMR (500 MHz, (CD₃)₂CO) δ 9.66 (ddd, J=12.1, 8.2, 1.6 Hz, 2H), 9.08 (s, 2H), 8.97 (d, J=5.1 Hz, 1H), 8.60–8.45 (m, 4H), 8.15 (ddd, J=10.5, 8.3, 4.8 Hz, 2H), 8.02 (t, J=8.7 Hz, 2H), 7.58 (d, J=8.0 Hz, 2H), 7.20 (dd, J=7.6, 1.5 Hz, 2H), 7.09 (t, J=7.9 Hz, 2H), 7.01 (d, J=7.8 Hz, 2H), 6.67 (d, J=8.2 Hz, 2H), 2.09–2.08 (m, 6H) ppm.

¹³C NMR (126 MHz, (CD₃)₂CO) δ 166.73, 161.93, 161.76, 161.57, 157.03, 156.83, 151.95, 146.87, 146.80, 146.63, 146.57, 144.85, 144.61, 144.35, 142.42, 141.97, 135.68, 135.52, 135.47, 135.32, 133.83, 132.76, 131.43, 130.92, 130.82, 129.50, 129.13, 129.11, 128.76, 128.18, 128.13, 127.91, 127.87, 127.83, 127.79, 127.41, 127.35, 126.61, 124.64, 119.78, 30.10 ppm.

HR‐MS (ESI+): m/z=938.2079, calcd for [C₅₄H₃₇CuN₉O₂S]⁺ m/z=938.2087.

Synthesis of C‐A (Scheme S3): A (35.4 mg, 105 mmol) was added to a CH₂Cl₂:CH₃OH mixture (10 mL, 9 : 1 v/v) which was then sparged with Ar for 10 minutes. Following which tetrakis(acetonitrile)copper(I) hexafluorophosphate (37.4 mg, 100 mmol) was added under positive Ar pressure, and stirred for 30 minutes before adding dmp (20.4 mg, 100 mmol) and stirring for an additional 30 minutes at room temperature. The solution was filtered with a 0.2 μm syringe filter to remove excess insoluble A. Half (5 mL) of this solution was diluted to 10 mL with CH₂Cl₂ and used for device fabrication. The solvent was removed from the other half (5 mL) of solution, yielding a dark red solid C‐A, and was used for characterization and compared to literature. ^[12]

Device Fabrication

Synthesis of Zinc Oxide Nanoparticles (ZnO NPs): Zinc chloride (5.00 g, 37.5 mmol) and sodium hydroxide (3.00 g, 75 mmol) were dissolved separately in CH₃OH (45 mL) each. Under rapid stirring, the sodium hydroxide was added dropwise to the zinc chloride solution via addition funnel and allowed to stir overnight (15 hours). The ZnO NPs were collected by centrifugation for 10 minutes at 2377 RCF followed by washing and centrifuging 3× with water to remove sodium chloride, and 3× with ethanol, followed by drying in vacuo.

Synthesis of Zinc Oxide Nanowires (ZnO NWs): FTO on glass substrates were cut into 1 cm×2 cm pieces and washed by sonication separately in soapy water (15 min), 18 MΩ water (15 min), and ethanol (15 min). Clean FTO was then spin‐coated (2×, 1000 rpm, 10 s) with 2 drops of 1 % wt/wt ZnO NP suspension in 95 % ethanol. 8 spin coated FTO slides were then placed horizontally with the FTO|ZnO NPs facing down. Zinc nitrate hexahydrate (1.00 g, 5.3 mmol) and HMTA (400 mg, 2.9 mmol) were separately dissolved in 15 mL of water each. These solutions were mixed and transferred to the centrifuge tube containing the seeded FTO films. The tube was left in a 90 °C oven for 1 hour for the NWs to grow. The films were thoroughly rinsed with water and acetone and left to air dry. These films were vertically dipped halfway into 1 M HCl, and rinsed with water and acetone, leaving a 1 cm² NW surface.

Device Fabrication: ZnO NW electrodes were submerged in the dyad or triad containing solutions prepared above for 24 hours, following which they were rinsed with acetone and allowed to air dry before being characterized and used for photocatalysis. ^[12]

General Characterization Procedures

¹H NMR and ¹³C NMR were recorded on a 300 MHz Bruker and a 500 MHz Varian spectrometer, respectively, and the chemical shifts were referenced to the residual solvent peaks.

High‐resolution electrospray mass spectrometry (HR‐ESI‐MS) was performed using a Waters model QTOF Ultima instrument. Samples were prepared as 0.1 mM solutions in 9 : 1 v/v CH₂Cl₂ : CH₃OH.

UV‐vis spectra were collected in 9 : 1 v/v CH₂Cl₂ : CH₃OH using quartz cuvettes with a 1 cm path length using a dual beam Varian Cary 5000 UV‐vis near‐infrared (NIR) spectrophotometer with a wavelength changeover at 350 nm and a scan rate of 10 nm s⁻¹. Samples were blanked and zeroed against the 9 : 1 v/v CH₂Cl₂ : CH₃OH solvent mixture.

Electrochemical measurements were performed using a WaveDriver 20 Integrated Bipotentiostat/Galvanostat workstation (Pine Research Instrumentation, Inc.) in a conventional three‐electrode cell. Molecular species were analyzed using a glassy carbon button (2 mm) working electrode (WE), Pt mesh as the reference electrode (RE), and Ag wire as the pseudoreference electrode. Copper complexes (1 mM), or ligand (1 mM) and [n‐Bu₄N]PF₆ (100 mM) were dissolved in deaerated CH₃CN. Cyclic voltammetry (CV) measurements were done under argon atmosphere with a scan rate of 100 mV s⁻¹. A 0.5 molar ratio of ZnCl₂ was added to A to increase solubility via complexation. Potentials were converted and reported versus NHE by first converting to potentials versus ferrocene/ferrocenium (Fc⁰/Fc⁺) redox couple as an internal standard and then adding +0.630 V. Differential pulse voltammetry (DPV) was performed with the same cell setup as CV with the exception of dimethylformamide as the solvent. DPV measurements were performed with pulse height: 10 mV; width: 50 ms; period 200 ms; increment: 4 mV, sampling pre‐pulse width: 5 ms; post‐pulse width: 5 ms with potentials also reported versus NHE. For film characterization, a similar set up is used, however, the WE was ZnO NW|A‐Cu(I)‐D or ZnO NW|A‐Cu(I). Chronoamperometry (CA) studies were performed with no applied bias in 100 mM KCl_(aq) solution, 200 mV applied bias versus Ag/AgCl reference electrode in pH 7 100 mM KPi solution, and 400 mV applied bias versus Ag wire pseudoreference electrode in 100 mM [n‐Bu₄N]PF₆ CH₃CN solution. The working electrode was illuminated by a homemade photoreactor with backside illumination (single Cree CXB3590 LED on a 140 mm pin heatsink, 540 mW/cm² with 400 nm long pass filter, as measured with a PM16‐405 thermal sensor power meter from ThorLabs). Linear sweep voltammetry (LSV) experiments were performed in degassed CH₃CN solution containing 100 mM [n‐Bu₄N]PF₆, 5 mM TEMPO, 25 mM FA, and 100 mM pyridine, at a scan rate of 10 mV s⁻¹. ^[11]

DFT calculations were carried out with Gaussian 16 ^[51] using the B3PW91 functional ^[52] with LANL2DZ as basis set. ^[53] The optimizations were conducted without symmetry constraint, and frequency calculations were made to confirm reaching the energy minima. GaussView5 ^[54] and IQmol were used for data analysis, visualization and plots. All calculations were conducted for gaseous phase complexes.

Femtosecond transient absorption (fsTA) measurements were performed using a pump‐probe femtosecond transient absorption spectroscopy system (TAS, Newport). This system utilizes an amplified Ti:sapphire femtosecond laser (Legend, Coherent) with an output at 800 nm, a 120 fs pulse duration, a 2 mJ pulse energy, and a 1 kHz repetition rate. The output of the laser is used, in part, to pump an optical parametric amplifier (TOPAS, Light Conversion) to generate excitation (pump) pulses at 460 nm. The probe pulses were produced by focusing a small portion of the 800 nm laser beam onto a 2 mm thick CaF₂ disk to generate a white light continuum spanning from 350 to 700 nm. The pump and probe beams were focused onto the sample with their relative polarization set at magic angle, while a delay stage (Newport, DL325) controlled the time delay of the probe pulses. The pump pulses were modulated at 500 Hz by an optical chopper, and the probe spectra were recorded with a spectrograph (Oriel, MS260) and a CCD camera (S10453‐1024Q, Hamamatsu). The open source software Glotaran was used to perform global analysis of the TA data.

Catalysis

TEMPO sensitization – Photocatalysis: In a representative procedure, TEMPO (4.9 mg, 0.025 mmol, 5 mM), furfuryl alcohol (FA, 12.3 mg, 0.125 mmol, 25 mM) or benzyl alcohol (BA, 13.5 mg, 0.125 mmol), and pyridine (39.5 mg, 0.50 mmol, 100 mM), were dissolved in CH₃CN (5 mL) in a 20 mL scintillation vial.¹ Subsequently, a ZnO|A‐C‐D device was placed in the vial, and the vial was sealed and degassed with argon, bubbled with O₂ or left open under ambient conditions and irradiated with white light in the above described photoreactor for 4 hours. A fan was placed on top of the reactor to maintain the reaction temperature below 25 °C. Once complete, 1,3,5‐trimethoxybenzene (TMB, ca. 20 mg, 0.12 mmol) was added as an internal standard and a crude yield was obtained by diluting ca. 0.2 mL the reaction solution with ca. 0.4 mL of CDCl₃ and integrating the aliphatic proton on FA versus the aromatic protons the TMB standard. The yield is taken as the ratio of the aldehyde peak (9.64 ppm, 1H) to the aromatic protons in TMB (6.04 ppm, 3H) from a ¹H‐NMR, multiplied by the number of moles of TMB.

TEMPO Sensitization – Photoelectrocatalysis: The reaction solution is prepared as above with the addition of [n‐Bu₄N]PF₆ (190 mg, 0.50 mmol, 100 mM) supporting electrolyte in the reaction solution. Subsequently, the solution was transferred to a photoelectrochemical cell using a three‐electrode setup with ZnO NWs|A‐C‐D WE, Pt mesh CE, and Ag wire pseudoreference electrode. The cell was then placed in the previously described photoreactor with an applied bias of 750 mV versus Ag wire (930 mV vs NHE). A fan was placed on top of the reactor to maintain the reaction temperature below 25 °C. Once complete, a crude yield was determined as above.

NHPI Sensitization: An equimolar solution (0.1 mmol in 5 mL of CH₃CN) of NHPI, 2,6‐lutidine, and FA is prepared, to which 5 μmol D‐C‐A (5 mol%) is added. The solution is bubbled with O₂ gas and irradiated for 4 or 24 hrs and a crude yield is determined as above.

Supporting Information

Differential pulse voltammetry, DFT calculation details, cyclic voltammetry, chopped light chronoamperometry measurements, NMR characterization data.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Ricardo-Noordberg J. F., Kamal S., Majewski M. B., ChemSusChem 2024, 17, e202400611. 10.1002/cssc.202400611

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.