Efficient Photogeneration of Hydrogen Boosted by Long-Lived Dye-Modified Ir(III) Photosensitizers and Polyoxometalate Catalyst

Introduction

Since the Industrial Revolution in the 19th century, the continuous depletion of traditional fossil fuels has caused serious energy shortages and environmental challenges. To solve these problems, great efforts have been devoted to developing cost-effective, clean, and sustainable energy alternatives. As one of the widely studied renewable energy exploration strategies, solar-driven water splitting into H₂ and O₂ is a promising pathway to convert and store solar energy into renewable chemical bonds.^1–4 In the past few decades, considerable attention has been paid to establishing highly efficient photocatalytic water splitting systems.^1,3–8 For the catalytic hydrogen evolution half-reaction, the most widely adopted system consists of three components containing photosensitizer, a catalyst, and a sacrificial reagent.^4,9–11 Notwithstanding, key issues have been identified regarding the exploration of viable, efficient, and inexpensive water reduction catalysts (WRCs), as well as the design of tunable, stable, and strong visible-light-responsive photosensitizers.^12–16

As an emerging type of multielectron-transfer WRCs, transition-metal-substituted polyoxometalates (TMSPOMs) with tunable structural compositions, physicochemical, and photochemical properties, could combine the advantages of both heterogeneous metal oxide catalysts (e.g., stability) and homogeneous molecular catalysts (e.g., high activity, selectivity, and tunability).^2,17–21 After years of numerous efforts, a series of Mn, Ni, Co, and Cu-substituted POMs have been synthesized and studied for catalytic H₂ evolution when coupled with visible-light-absorbing photosensitizers and sacrificial reagents.^19,21–25 Although TMSPOMs have shown great potential applications in various H₂-evolving systems, the challenges facing these systems remain substantial. To date, the most widely used photosensitizers in POM-catalyzed H₂-evolving systems are limited to noble metal-based organometallic complexes such as [Ru(bpy)₃]²⁺, [Ir(ppy)₂(dtbbpy)]⁺ and their derivatives,^{21–24,26–29} as well as pure organic chromophores, including eosin Y, fluorescein, rhodamine, and others.^30–32 Such noble metal-based organometallic complexes usually exhibit a long-lived triplet state (³MLCT) to promote photocatalytic hydrogen evolution.^4,12,33–38 However, they often suffer from low molar absorption coefficients and narrow visible-light-absorbing range.^4,12,39 Meanwhile, pure organic chromophores generally show very high molar absorption coefficients. Nonetheless, the short-lived excited states and poor photostability hinder their practical application in photocatalysis considerably. To improve the solar-to-energy conversion efficiency, some pioneering research studies have reported the construction of strong visible-light-absorbing chromophores via covalently decorating organometallic photosensitizers with organic dyes.^38,40–47 For example, sensitization of colloidal Pt-TiO₂ with Bodipy or Rhodamine-modified platinum diimine dithiolate (PtN₂S₂) charge-transfer chromophore could enhance the photocatalytic hydrogen evolution efficiency remarkably with substantial longevity and high turnover numbers (TONs).^44,46 Also, the modification of [Ir(ppy)₂(bpy)]⁺ with two different organic dyes greatly increased the visible-light-absorbing ability and hydrogen evolution activity in the presence of [Co^III(dmgH)₂(py)Cl] catalyst.⁴⁷ Although many of these works reported extremely high TONs with respect to the photosensitizers, the TONs versus H₂-evolving catalysts are usually unsatisfactory. Moreover, among various dye-modified chromophores, none was applied as light-absorbers to drive catalytic reactions in POM-catalyzed H₂-evolving systems. We considered that employing the strong visible-light-absorbing ability of Coumarin 6 chromophore (ε < 10⁵ M⁻¹ cm⁻¹) and the good coordinating ability as the C^N ligand to iridium metal center would be a promising strategy to enhance the photocatalytic performance of POM-catalyzed H₂-evolving system, as this approach would combine the advantages of high visible-light-absorbing ability of Coumarin 6 and long-lasting triplet states of potential Ir(III) complexes.

Herein, in this context, we prepared different Coumarin-modified Ir(III) complexes (denoted as PS-2 and PS-3) and investigated them as chromophores to build a highly efficient photocatalytic H₂-evolving system using a combined Ni₃PW₁₀ as catalyst and triethanolamine (TEOA) as the sacrificial electron donor. Various spectroscopic and computational studies revealed that the factors of strong and broad visible-light-absorbing ability, long-lived triplet state, suitable redox potential versus POM catalyst, and large highest occupied molecular orbital (HOMO)–lowest unoccupied molecular orbital (LUMO) gap of PS-2 accounted for considerably enhanced photocatalytic activity.

Experimental Methods

Materials

2-Phenylpyridine, 4,4′-di-tert-butyl-2,2′-bipyridine, 4,4′-dibromo-2,2′-bipyridine, 3-(2-benzothiazolyl)-7-(diethylamino)coumarin (Coumarin 6), 2-tributylstannylthiophene ligands, and Iridium(III) chloride trihydrate were used as received without further purification. Catalyst K₆Na[Ni₃(H₂O)₃PW₁₀O₃₉H₂O]•12H₂O (K₆Na- Ni₃PW₁₀)⁴⁸ was prepared according to the modified literature methods and confirmed by single-crystal X-ray diffraction, as well as Fourier transform infrared (FT-IR) spectroscopy. TBA- Ni₃PW₁₀ was obtained by the exchange of counter cations with tetrabutylammonium (TBA⁺) salt from K₆Na- Ni₃PW₁₀. The spy ligand was prepared using reported methods.⁴⁹ Iridium complexes were prepared according to a modified method described by Takizawa et al.⁵⁰ All solvents and other reagents were used as received.

Synthesis of 4,4′-di(thien-2-yl)-2,2′-bipyridine (spy)

4,4′-Dibromo-2,2′-bipyridine (0.5 g, 1.6 mmol, 1.0 equiv), 2-tributylstannylthiophene (1.5 g, 4.0 mmol, 2.5 equiv), [Pd (PPh)₃Cl₂] (67 mg, 95.5 μmol, 0.06 equiv), and CsF (1.0 g, 6.4 μmol, 4.0 equiv) were suspended in a 12 mL degassed solution mixture of toluene/tetrahydrofuran (THF) (2/1). After several degassing and argon bubbling times, the mixture was stirred and heated at 110 °C for 6 h. The resulting mixture was poured into MeOH that afforded a dark grey precipitate, which was filtered off, washed with MeOH, a small quantity of dichloromethane (DCM), and diethyl ether to produce a pure light grey solid product. Yield: 90%; proton nuclear magnetic resonance (¹H NMR; 400 MHz, CDCl₃, 298 K, δ): 8.70–8.67 (m, 2H), 7.67 (dd, 1H, J = 3.64, 0.84 Hz), 7.53 (dd, 1H, J = 5.04, 1.96 Hz), 7.44 (dd, 1H, J = 5.04, 0.88 Hz), 7.16 (dd, 1H, J = 5.00, 3.72 Hz). High-resolution mass spectrometry (HRMS) (electrospray ionization mass spectrometry [ESI-MS]) (m/z): [M]⁺ calcd for [C₁₈H₁₃N₂S₂], 321.0520; found, 321.0500.

Synthetic procedure for iridium complexes

Synthesis of PS-1

Iridium(III) chloride (450 mg, 1.28 mmol, 1.0 equiv) and 2-phenylpyridine (440 mg, 2.82 mmol, 2.2 equiv) were suspended in 30 mL 2-methoxyethanol/water (3:1 v/v) solution, and the mixture was heated at 110 °C for 48 h under argon conditions. After completion, the solution was cooled to room temperature, then the precipitate was filtered off, washed with water and EtOH to afford a solid orange Ir(III) μ-chloro-bridged dimer product. The resulting Ir(III) μ-chloro-bridged dimer complex (200 mg, 0.19 mmol, 1.0 equiv) and 4,4′-di-tert-butyl-2,2′-bipyridine (110 mg, 0.41 mmol, 2.2 equiv) were suspended in 10 mL of 2-methoxyethanol solution, and the mixture was heated at 110 °C for 12 h under argon conditions. After cooling to room temperature, an excess NH₄PF₆ aqueous solution was added to the above solution and stirred for 4 h at room temperature. The resulting precipitate was filtered off, washed with water and EtOH, and the crude product was purified by silica gel column chromatography using dichloromethane/MeOH (v/v = 1/0 to 200/1) as the eluent, affording the pure product PS-1 as an orange solid. Yield: 60%. ¹H NMR (400 MHz, CDCl₃, 298 K, δ): 8.37 (d, 2H, J = 1.56 Hz), 7.88 (d, 2H, J = 8.08 Hz), 7.82 (d, 2H, J = 5.84 Hz), 7.74 (m, 2H), 7.66 (d, 2H, J = 7.68 Hz), 7.60 (d, 2H, J = 5.72 Hz), 7.38 (dd, 2H, J = 5.84, 1.80 Hz), 7.08 (m, 2H), 7.00 (m, 2H), 6.89 (m, 2H), 6.29 (d, 2H, J = 7.52 Hz), 1.42 (s, 18 H). HRMS (ESI-MS) (m/z): [M-PF₆]⁺ calcd for [C₄₀H₄₀IrN₄]⁺, 769.2879; found, 769.2877.

Synthesis of PS-2

Iridium(III) chloride (450 mg, 1.28 mmol, 1.0 equiv) and Coumarin 6 (990 mg, 2.82 mmol, 2.2 equiv) were suspended in 30 mL 2-methoxyethanol/water (3:1 v/v) solution; the mixture was heated at 110 °C for 48 h under argon conditions. After completion, the solution was cooled to room temperature, then the precipitate was filtered off, washed with water and EtOH, affording Ir(III) μ-chloro-bridged dimer product as orange solid. Subsequently, the resulting Ir(III) μ-chloro-bridged dimer complex (200 mg, 0.11 mmol, 1.0 equiv) and 4,4′-di-tert-butyl-2,2′-bipyridine (65 mg, 0.24 mmol, 2.2 equiv) were suspended in 10 mL of 2-methoxyethanol solution, and the mixture was heated at 110 °C for 12 h under argon conditions. After cooling to room temperature, an excess NH₄PF₆ aqueous solution was added to the above solution and stirred for 4 h at room temperature. The resulting precipitate was filtered off, washed with water and EtOH, the crude product was purified by silica gel column chromatography using dichloromethane/MeOH (v/v = 1/0 to 200/1) as the eluent, affording the pure product PS-2 as an orange solid. Yield: 56%. ¹H NMR (400 MHz, CDCl₃, 298 K, δ): 8.41 (d, 2H, J = 5.96 Hz), 8.24 (d, 2H, J = 1.52 Hz), 7.82 (d, 2H, J = 7.84 Hz), 7.63 (dd, 2H, J = 6.00, 1.68 Hz), 7.29 (m, 2H), 6.91 (m, 2H), 6.38 (d, 2H, J = 2,60 Hz), 6.03 (d, 2H, J = 9.40 Hz), 5.86 (dd, 2H, J = 9.44, 2,60 Hz), 5.72 (d, 2H, J = 8.52 Hz), 3.29 (m, 8H), 1.41 (s, 18H), 1.09 (t, 12H, J = 7.00 Hz). HRMS (ESI-MS) (m/z): [M-PF₆]⁺ calcd for [C₅₈H₅₈IrN₆O₄S₂]⁺, 1159.3590; found, 1159.3598.

Synthesis of PS-3

To synthesize PS-3, the 4,4′-di(thien-2-yl)-2,2′-bipyridine (spy) (77 mg, 0.24 mmol, 2.2 equiv) ligand was used to react with the above Ir(III) μ-chloro-bridged dimer complex (200 mg, 0.11 mmol, 1.0 equiv) in 10 mL of 2-methoxyethanol solution, the mixture was heated at 110 °C for 12 h under argon condition. After cooling to room temperature, an excessive NH₄PF₆ aqueous solution was added to the above solution and stirred for 4 h at room temperature. The resulting precipitate was filtered off, washed with water and EtOH, and the crude product was purified by silica gel column chromatography using dichloromethane/Hexane (v/v = 4/1 to 1/0) as the eluent, affording the pure product PS-3 as an orange solid. Yield: 40%. ¹H NMR (400 MHz, d⁶-DMSO, 298 K, δ): 8.93 (d, 2H, J = 1.76 Hz), 8.68 (d, 2H, J = 6.08 Hz), 8.20 (dd, 2H, J = 4.04, 0.80 Hz), 8.11 (d, 2H, J = 8.00 Hz), 8.03 (dd, 2H, J = 6.00, 1.80 Hz), 7.97 (dd, 2H, J = 4.96, 0.84 Hz), 7.34 (dd, 2H, J = 4.84, 3.84 Hz), 7.25 (t, 2H, J = 7.40 Hz), 7.05 (m, 2H), 6.49 (s, 2H), 6.12 (d, 2H, J = 8.44 Hz), 6.03 (s, 4H,), 3.30 (overlap by solvent peak, m, 8H), 0.97 (t, 12H, J = 6.88 Hz). HRMS (ESI-MS) (m/z): [M-PF₆]⁺ calcd for [C₅₈H₄₆IrN₆O₄S₄]⁺, 1211.2093; found, 1211.2097.

Instrumentation

¹H NMR spectra were recorded on a Bruker Ascend 400M (Avance IIIHD 400 MHz) Fourier transform NMR spectrometer (Bruker, Germany) with chemical shifts (δ, ppm) relative to tetramethylsilane (Me₄Si). High-resolution ESI-MS was performed using AGILENT Q-TOF 6520 mass spectrometer (Agilent, United States). Infrared spectra were acquired on a Bruker TENSOR II FT-IR spectrometer by preparing dry solid samples mixed with KBr pellets. The X-ray diffraction data were obtained using a Bruker APEXII DUO CMOS detector with monochromated Mo-Kα radiation. Cell refinement and data reduction were processed with the Protenum2 program package.⁵¹ Using olex2, the molecular structure was solved with the ShelXS structure solution program using direct methods and refined using the XL refinement package with least-squares minimization.⁵² Then all non-H atoms of the complexes were refined with anisotropic thermal parameters. Subsequently, the hydrogen atoms were included in idealized positions and refined using fixed geometry relative to their carrier atoms. Crystal structure graphics and their corresponding packing diagrams were acquired using software Mercury 2.4.5 (Cambridge Crystallographic Data Centre, Cambridge). UV–vis absorption spectra following recording on a UV 2600 UV–vis spectrophotometer and all emission spectra were performed on spectrofluorometer FS5 (Edinburgh Instruments Ltd., Edinburgh); errors for λ values (±1 nm) were estimated. The emission lifetime was measured using an EPL-450 picosecond pulsed diode (Edinburgh Instruments Ltd.) laser system (pulse output 450 nm). These long-lived triplet decay kinetic curves were fitted by a single-exponential decay function. Cyclic voltammetry was recorded on a CHI660E Electrochemical Workstation (Shanghai Chenhua Company, Shanghai, China). The experimental data were collected at a scan rate of 100 mV s⁻¹. The working electrode was polished with 0.05 μm alumina, sonicated with deionized water, rinsed with EtOH, and then with deionized water before use. The reference and the counter electrodes were Ag/Ag⁺ (0.1 M TBAPF₆ and 0.01M AgNO₃ in acetonitrile) and a platinum wire, respectively. Solutions were degassed for 10 min by N₂ before experiments. The E_1/2 of the ferrocenium/ferrocene couple (Fc⁺/Fc) was measured in the same solution as samples and was used as an internal reference.

Photocatalytic hydrogen production experiments

The photocatalytic hydrogen production was performed in deaerated CH₃CN-DMF (v/v = 3/2) solution with TEOA (0.25 M) as a sacrificial electron donor, H₂O (2 M) as a proton source, iridium complexes (200 μM) as photosensitizers, and Ni-substituted polyoxometalate ( Ni₃PW₁₀; 20 μM) as the catalyst. The reaction solution was degassed with mixed argon/methane gases for 15 min before irritation. Xe lamp (λ < 400 nm, 300 W, Beijing Perfectlight Technology Co. Ltd., Beijing, China) was employed as the irradiation light. All TONs were calculated based on the Ni₃PW₁₀ catalyst.

Density functional theory calculations

Density functional theory (DFT) calculations were performed using B3LYP functional with Grimme’s D3 dispersion correction. The 6-31G(d) basis set was applied on nonmetal atoms, and the Lanl2DZ basis set was used on the Ir atom. The continuum solvation model based on density (SMD) with solvent CH₃CN was used to address the solvent effect. The calculations were carried out using the Gaussian09 program (refer to the Supporting Information for details).

Results and Discussion

Syntheses and characterization

The structures of bis-cyclometalated Ir(III) complexes (denoted as PS-1, PS-2, and PS-3) and Ni₃PW₁₀ catalyst are illustrated in Figures 1a and 1b. The polyoxoanion Ni₃PW₁₀ catalyst was synthesized and confirmed by single-crystal X-ray diffraction and FT-IR spectrum (see Supporting Information). To achieve good solubility in the photocatalytic system, counter cations Na⁺ and K⁺ of polyoxoanion Ni₃PW₁₀ were exchanged by TBA⁺ salt; the exchange of counter cations did not affect the geometrical structure of the polyoxoanion Ni₃PW₁₀ ( Supporting Information Figure S1). The TBA- Ni₃PW₁₀ complex was used for all subsequent experiments unless otherwise stated. The bis-cyclometalated Iridium(III) complexes were prepared using a classical two-step method (Scheme 1). Typically, treatment of IrCl₃ with 2-phenylpyridine or Coumarin 6 (2.2 equiv) gave Ir(III) μ-chloro-bridged dimers, which further reacted with a coordinated bipyridine ligand (1.5 equiv) to afford ensembles PS-1, PS-2, and PS-3, respectively (40−60% yields). The ¹H NMR spectra of the complexes ( Supporting Information Figures S2–S4) agreed well with their structures and the diamagnetic nature of Ir(III) species. The high-resolution ESI mass spectra of all three complexes are in agreement with the simulated patterns ( Supporting Information Figures S5–S7). Detailed preparation and characterization of these complexes are included in the Supporting Information.

• Download figure
• Download PowerPoint

Needle-shaped single crystals of PS-2 considered suitable for X-ray diffraction were obtained by diffusion of diethyl ether into acetonitrile solution of the ensemble. After multiple attempts, crystals of PS-3 were afforded using diverse solvent systems (ether/acetonitrile or dichloromethane/hexane); however, the quality of these crystals was too low for suitable X-ray diffraction. PS-2 crystallized in the monoclinic space group P2₁ (Figures 1c and 1d). Ir(III) metal ions adopted an octahedral six-coordinated geometry with two C^N (Coumarin 6) and one closely surrounded N^N (bipyridine) ligands. The Ir–N (on Coumarin 6) distances of PS-2 complex are 2.056(8) and 2.063(8) Å were slightly shorter than Ir–N (on bipyridine) distances [2.147(9) and 2.130(7) Å]. The Ir–C distance ranged from 2.045(9) to 2.053(9) Å, with the N1–Ir1–C50, N3–Ir1–C50, and C30–Ir1–C50 bond angles being 169.8(3)°, 101.5(3)°, and 91.8(4)°, respectively. All the Ir center involved bond distances and angles were consistent with other bis-cyclometalated Iridium(III) complexes.

• Download figure
• Download PowerPoint

Photophysical properties

The photophysical properties of the PS-1, PS-2, and PS-3 were examined in acetonitrile solutions at 298 K. The data of UV–vis absorption maxima (λ_max) with molar extinction coefficients, emission maxima (λ_em), and triplet emission lifetimes (τ) are summarized in Supporting Information Table S1. As shown in Figure 2a, PS-1 displayed absorption maxima at 263 nm with shoulders around 310 and 340 nm, whereas PS-2 and PS-3 exhibited intense absorption in the range of 400–550 nm with a maximum molar absorption coefficient up to 110,620 and 91,430 M⁻¹ cm⁻¹, respectively. These results demonstrated that the visible-light absorption ability of cyclometalated Iridium(III) complexes were remarkably enhanced by modification of the Coumarin 6 dye groups. Upon varying the groups on bipyridine ligand from tert-butyl to thiophene, the absorption spectra of PS-3 showed a slight change in the range of 400–550 nm with respect to PS-2, indicating there was negligible electronic interaction between bipyridine and Coumarin 6 units. Therefore, the absorption bands of these Ir(III) complexes in the range of 400–550 nm corresponded to intraligand charge transfer (ILCT), while those in the range of 250–400 nm were dominated by metal-perturbed intraligand transitions.⁴¹

• Download figure
• Download PowerPoint

The emission spectra of PS-2 and PS-3 under air (Figure 2b) and N₂ (Figure 2c and Supporting Information Figure S8) conditions showed vibronic structured bands, indicating that the emissive excited states were ligand-centered. Under N₂ condition, the emission maxima appeared at 570, 506/587, and 506/575 nm for PS-1, PS-2, and PS-3, respectively. The emission lifetimes of the three complexes in deaerated CH₃CN solution (Figures 2d–2f) were determined as 259 ns (λ_em = 570 m), 1559 ns (λ_em = 587 nm), and 1703 ns (λ_em = 575 nm) for PS-1, PS-2, and PS-3, respectively, suggesting the phosphorescence nature of emission that originates from triplet-excited states. Besides, the emission spectra of PS-2 (Figure 2c) and PS-3 ( Supporting Information Figure S8b) at longer wavelengths (550–700 nm) showed a noticeable quenching effect by air, which was in agreement with their long phosphorescent lifetimes, while their emission spectra at shorter wavelengths (450–530 nm) presented negligible or a very slight quenching effect by air. By considering their absorption spectra (lower energy level absorption assigned as S0 → S1 transition), the shorter wavelength bands (emission maxima at 506 nm) of PS-2 and PS-3 could be assigned as fluorescence (S1 → S0 transition), which was further verified by the corresponding extremely short-lived fluorescent lifetimes ( Supporting Information Figure S9). In addition, the Coumarin 6 ligand displayed intense fluorescence in the range of 460–650 nm ( Supporting Information Figure S8c); however, its emission was dramatically quenched after coordination to an iridium center, indicating the existence of efficient intersystem crossing (ISC) process⁴⁷ (singlet to triplet excited-states transition) in PS-2 and PS-3.

Visible-light-driven hydrogen evolution

The photocatalytic hydrogen production was conducted in deaerated CH₃CN-DMF (v/v = 3/2) solution with TEOA as sacrificial electron donor, H₂O as proton source, Ir(III) complexes as photosensitizers, and Ni₃PW₁₀ as H₂-evolving catalyst. In view of the broad and intense absorption spectra of PS-2 and PS-3, Xe lamp (λ < 400 nm, 300 W) was employed as the irradiation light source. All TONs were calculated relative to the Ni₃PW₁₀ catalyst. After an identical photocatalytic reaction had proceeded for 6 h, the TONs of PS-2- and PS-3-containing catalytic systems reached up to 1295 (∼155 μmol H₂ gas) and 204.4 (∼24.5 μmol H₂ gas), which corresponded to 13.1 and 2.1 times, respectively, relative to the PS-1-containing system (TON ∼ 98.9) (Figure 3a). These experimental results illustrated that the modification of Ir(III) complex with Coumarin 6 dyes accelerated the photocatalytic process significantly, thereby leading to efficient generation of H₂ gas. The long-term catalytic activity of the system containing PS-2 or PS-3 photosensitizer and Ni₃PW₁₀ catalyst was also evaluated. We observed that TONs of 4026 or 807 were achieved for PS-2 or PS-3-containing catalytic systems after 120 h irradiation, respectively (Figure 3b), indicating the robustness of the catalytic system and high photostability of Coumarin-modified Ir(III) chromophores under catalytic conditions. Given the best performance of complex PS-2, additional experiments were conducted to optimize the PS-2-containing catalytic system, as described below.

• Download figure
• Download PowerPoint

Various control experiments have been performed to study the effect of each component in the PS-2-containing photocatalytic system. Figure 3c shows that in the absence of any essential components such as the catalyst Ni₃PW₁₀, PS-2, or TEOA, no yield or negligible H₂ generation was noted. Moreover, replacing Ni₃PW₁₀ with three molar equivalents of NiCl₂ resulted in an obvious decrease of TON from 1295 to 106, revealing an essential role of an intact skeleton of the catalyst Ni₃PW₁₀. To verify the molecular integrity of the Ni-POM catalyst, two distinct sets of experiments were conducted: First, the mercury-poison test (Figure 3c) using up to 200 mg Hg in the photocatalytic solution showed no apparent decrease of H₂ production, which excluded the formation of Ni-based nanoparticles and proved the molecular stability of Ni₃PW₁₀ during photocatalysis. Second, we collected the FT-IR spectra of the isolated Ni₃PW₁₀ catalyst before and after a photocatalytic reaction had proceeded for 12 h. Given the anionic nature of the Ni₃PW₁₀ polyoxoanion, it is easy to precipitate the Ni₃PW₁₀ catalyst with cationic PS-2 after removing the reaction solvent using rotary evaporation. As shown in Supporting Information Figure S10, the FT-IR spectra of the isolated PS-2- Ni₃PW₁₀ adducts remained largely unchanged before and after 12 h of photocatalysis, further confirming the molecular integrity of Ni₃PW₁₀ catalyst after the photocatalytic reaction. In addition, varying the concentration of the three reaction components mentioned above affected the catalytic rate considerably. Under optimized concentration of the catalyst Ni₃PW₁₀, a TON as high as 19,739 was obtained (Figure 3d and Supporting Information Figure S11), which, to the best of our knowledge, was the highest value recorded, thus far, among all known POM-catalyzed H₂-evolving systems (data listed in Supporting Information Table S2). Moreover, adjusting the concentration of both PS-2 (from 0.1 to 0.3 mM) and TEOA (from 0.05 to 0.25 M) led to increased H₂ production, corresponding to the enhancement of TONs from 364 to 2104 (Figure 3e) and from 606 to 1295 (Figure 3f), respectively. Collectively, these results indicated that the three components ( Ni₃PW₁₀, PS-2, and TEOA) were crucial for highly efficient photocatalytic H₂ production. Further, the photostability of PS-2 and PS-3 was examined by monitoring their UV–vis absorption spectra in CH₃CN-DMF solvent over time course under strong visible light illumination. As shown in Supporting Information Figure S12, although PS-2 and PS-3 decomposed slightly with prolonged reaction time, the absorption peaks at both 250–400 nm (metal-perturbed intraligand transitions) and 400–500 nm (ILCT of the Coumarin ligand) remained largely unchanged after strong visible-light illumination, indicating a relatively good photostability of PS-2 and PS-3 chromophores.

To better understand the excited-state characteristics of different complexes and the origin of the high performance of PS-2 photosensitizer, we conducted both nanosecond transient absorption (ns-TA) and femtosecond TA (fs-TA) measurements for the titled complexes and the Coumarin 6 ligand under N₂ atmosphere. The ns-TA spectra of PS-1 and PS-2 in CH₃CN/DMF solution are displayed in Supporting Information Figure S13. Upon excitation at 267 nm, PS-1 generated two positive absorption bands around 380 and 500 nm representing the excited-state absorption (ESA) from triplet state, which was in accordance with a similarly reported complex.⁴² The transient absorption for modified complex PS-2 showed strong bleaching peaks (negative band) around 450 and 480 nm upon irradiation at 420 nm, consistent with the ground-state absorption of Coumarin 6 ligand. Accordingly, the triplet state (positive band) of PS-2 was distributed on the Coumarin part, which agreed with the estimation from the above emission studies. This result was also in line with that of a similarly reported Coumarin-modified complex.⁴⁷

To uncover the intramolecular energy transfer process, the fs-TA measurements of Coumarin 6 ligand and PS-2 were performed at various time delays upon 420 nm laser excitation ( Supporting Information Figures S14 and S15). For Coumarin 6 ligand ( Supporting Information Figure S14), the positive band was found around 460, representing its singlet ESA. The negative bands of Coumarin 6 ligand were located in the range of 475–550 nm, corresponding to its ground-state bleach (GSB) and stimulated emission (SE).⁵³ The redshift of the SE band was ascribed to a decrease in the excited-state potential energy surface induced by the solvation process. For Coumarin 6 ligand, no transformation from singlet to triplet was observed. The fs-TA spectra of PS-2 ( Supporting Information Figure S15) displayed intense GSB around 480 nm, which was quite different from that of Coumarin 6 ligand. Obviously, the SE in PS-2 was much weaker than that in Coumarin 6, and it disappeared rapidly, accompanied by the rising of a long-lived excited band above 500 nm, assigned to the triplet-excited state of PS-2 generated by the ISC process. Both the ns-TA and fs-TA spectra exhibited major contributions from triplet state absorption of PS-2, resulting in a better photocatalytic performance under the experimental conditions.

Luminescence quenching studies

Next, we unveiled the photocatalytic mechanism and electron-transfer dynamics by employing both steady-state and time-resolved fluorescence spectroscopy to investigate the luminescence quenching of PS-1, PS-2, and PS-3 using either Ni₃PW₁₀ or TEOA in CH₃CN/DMF (3/2) solution under air and N₂ atmosphere. We showed that the presence of both Ni₃PW₁₀ and TEOA indeed quenched the emission intensity of all three photosensitizers (Figure 4a and Supporting Information Figures S16–S18) under the simulated photocatalytic conditions; such quenching effects of PS-1 and PS-2 by Ni₃PW₁₀ and TEOA was much more efficient in N₂ atmosphere than in air ( Supporting Information Figures S16 and S17), further confirming the critical role of the phosphorescence quenching process. The emission intensity of all three photosensitizers decreased progressively at increasing concentrations of Ni₃PW₁₀ (0–50 μM) and TEOA (0–0.25 M). Analyses of Stern−Volmer plots (Figures 4c and 4d and Supporting Information Figures S19–S21) yielded fluorescence and the phosphorescence quenching rate constants listed in Table 1. The oxidative phosphorescence quenching rate constants (k_oq) by Ni₃PW₁₀ were in the order of PS-1 (3.6 × 10¹⁰ M⁻¹ s⁻¹) < PS-2 (2.8 × 10⁹ M⁻¹ s⁻¹) < PS-3 (2.0 × 10⁹ M⁻¹ s⁻¹) (Table 1). Meanwhile, the reductive phosphorescence quenching rate constants (k_rq) of PS-1, PS-2, and PS-3 by TEOA were 4.7 × 10⁷ M⁻¹ s⁻¹, 1.1 × 10⁶ M⁻¹ s⁻¹, and 1.4 × 10⁶ M⁻¹s⁻¹, respectively. Although the quenching rate constants of PS-1 by both Ni₃PW₁₀ and TEOA displayed were 10 times higher than that of PS-2 and PS-3, a very weak absorption of PS-1 in the visible-light region led to its poor performance of hydrogen production. Moreover, though the oxidative quenching rate constant outperforms the reductive quenching process by about three orders of magnitude, the reductive quenching still played the predominant role in the catalytic processes due to the much higher concentration of TEOA (0.25 M), compared with Ni₃PW₁₀ (20 μM). Consistent with the steady-state emission quenching results, the presence of Ni₃PW₁₀ or TEOA also accelerated the luminescence decay kinetics of all three photosensitizers ( Supporting Information Table S3). For example, the single-exponential fitting of the decay kinetics of PS-2 in N₂ gave a lifetime of 1063.1 ns, which decreased to 1012.7 ns in the presence of Ni₃PW₁₀, and 887.1 ns in the presence of TEOA, respectively (Figure 4b), further confirming the existence of both oxidative and reductive quenching processes. Notably, the luminescence decay effect was more pronounced in N₂ than in air ( Supporting Information Table S3), revealing that the long-lived triplet states of these photosensitizers were preferable for efficient electron transfer.

• Download figure
• Download PowerPoint

Electrochemical studies

Further, we employed electrochemical studies to investigate the electron-transfer thermodynamics and elucidate the distinguished activities of different photosensitizers for catalytic hydrogen production. Cyclic voltammograms (CVs) of components PS-1, PS-2, PS-3, Ni₃PW₁₀, and TEOA were determined in CH₃CN/DMF (3/2) solution ( Supporting Information Figure S22 and Table 2). We established that the oxidation potentials [vs reference saturated calomel electrode (SCE)] of PS-1, PS-2, and PS-3 were located at 1.42, 1.08, and 1.01, respectively, corresponding to Ir³⁺/Ir⁴⁺ processes. Compared with PS-1, PS-2 exhibited three reduction potentials, ascribed to Coumarin 6⁰/⁻¹ (–1.26 V), Coumarin 6⁻¹/⁻² (–1.53 V), and bpy⁰/⁻¹ (–1.80 V), respectively.⁴⁷ Only two E_red values for PS-3 were detected at −1.20 and –1.44 V vs SCE. Obviously, PS-3 exhibited less negative reductive potential and a close E_ox value, compared with that of PS-2, which led to a narrow HOMO–LUMO energy gap due to the presence of thiophene groups on bipyridine ligand of PS-3. These experimental results also explained the lower hydrogen evolution activity of PS-3-containing system (Figure 3a), despite its comparable absorption and triplet lifetimes to that of PS-2 ( Supporting Information Tables S1 and S3). Furthermore, the redox potentials of Ni₃PW₁₀ POM and TEOA were measured as –0.13 and 0.73 V, corresponding to TEOA/TEOA⁺ and POM/POM^–, respectively. Based on the potential values of all catalytic components, the Gibbs free energy changes (ΔG_CS)³⁸ were also calculated to gain an understanding of the feasibility of the electron transfer. The calculated ΔG_CS values of these important redox electron-transfer processes (e.g., from TEOA to excited photosensitizers, from excited photosensitizers to Ni₃PW₁₀, and from reduced photosensitizers to Ni₃PW₁₀) were all negative, indicating the thermodynamic feasibility of these processes (Table 2). The driving forces (ΔG_CS) for the electron transfer from excited photosensitizer state to Ni₃PW₁₀ (–0.96 for PS-1, –1.06 for PS-2, and –1.31 for PS-3) outperformed that of TEOA to excited photosensitizer state (–0.14 for PS-1, –0.28 for PS-2, and –0.52 for PS-3). However, given the excess amount of TEOA, compared with that of Ni₃PW₁₀, the dominant redox process was still subjected to electron transfer from TEOA to excited photosensitizer, consistent with the above experimental emission quenching results. Such mechanism could be further substantiated by the more negative ΔG_CS value of reduced PS-2 to Ni₃PW₁₀ (–1.39), compared with that of PS-3 (–1.07), which also explained the higher photocatalytic hydrogen evolution activity of the PS-2-containing system (Figure 3a).

Computational study

To uncover the nature of the frontier molecular orbitals and the difference in the HOMO–LUMO gap between PS-2 and PS-3, DFT calculations were performed by referring to the crystallographic data of PS-2 ( Supporting Information Tables S4 and S5). Geometry optimizations were applied to the singlet ground states of complexes PS-2 and PS-3 (see Supporting Information for details). Observed from the frontier molecular orbitals in Figure 5a, HOMOs of both PS-2 and PS-3 were distributed over the C^N (Coumarin 6) ligands with little Ir-based d orbital characteristics, whereas the spin density on the LUMOs is localized on N^N (bipyridine) ligands. More optimized orbitals are depicted in Supporting Information Figure S23 and Table S6. The calculated energy levels of both HOMOs were located at −5.23 eV, but the energy level of the LUMO lowered from −2.09 eV ( PS-2) to −2.45 eV ( PS-3), thereby making the HOMO–LUMO energy gap of PS-2 being 0.36 eV larger than that of PS-3. Thus, the calculations revealed that the substitution of the electron-donating tert-butyl group ( PS-2) to the electron-withdrawing thiophene group had a minor effect on the HOMO but stabilized the LUMO. In brief, the calculations were consistent with the experimental results of UV–vis and CV measurements, also supporting the outperformance of PS-2 over PS-3 in the photocatalytic systems.

• Download figure
• Download PowerPoint

Proposed photocatalytic mechanism

Based on the above studies, a mechanism for the photocatalytic H₂ generation using Coumarin-modified PS was proposed (Figure 5b), as follows: Upon illumination by strong visible light, the photosensitizer is excited from the singlet ground state to singlet excited state, which can further transfer to triplet excited state through efficient ISC. The resulting triplet ³[Coumarin-Ir]^+* could be either reductively quenched by TEOA or oxidatively quenched by Ni₃PW₁₀ catalyst with the reductive quenching pathway as the dominant process. Then the reduced Ni₃PW₁₀ catalyst generated by either receiving electrons from the reduced [Coumarin-Ir] or triplet ³[Coumarin-Ir]^+* could further reduce water to H₂ effectively, which is finally accompanied by the generation of ground-state PS.

Conclusion

Coumarin 6 dye was incorporated into bis-cyclometalated Ir(III) complexes to enhance their photosensitivity to visible light. The obtained Coumarin-modified Ir(III) chromophores ( PS-2 and PS-3) and Ni₃PW₁₀ catalyst worked cooperatively to establish efficient and robust photocatalytic H₂ evolution systems, exhibiting approximately 13 and 2 times H₂ production activity, respectively, compared with that of a commercially available photosensitizer, PS-1, under identical photocatalytic conditions. Our experimental results have revealed that the distinguishing features of broad and intense absorption in visible-light region, appropriate redox potentials toward POM catalyst, and the displayed larger HOMO–LUMO energy gap contribute to the remarkable enhancement of photocatalytic activity, supporting the proposed catalytic mechanism. Our present research does not only provides insightful guidelines for designing effective light-absorbers but also enriches the POM-based catalytic chemistry for solar energy conversion.

Supporting Information

Supporting Information is available and includes other detailed experimental procedures, data, X-ray crystallographic data for PS-2 and theoretical calculation details, Figures S1–S23, and Tables S1–S6.

Conflict of Interest

There is no conflict of interest to report.

Funding Information

The authors gratefully acknowledge the financial support from the National Natural Science Foundation of China (nos. 21871025 and 21831001), the Recruitment Program of Global Experts (Young Talents), and the BIT Teli Young Fellow Program.