Two Excited State Collaboration of Heteroleptic Ir(III)-Coumarin Complexes for H2 Evolution Dye-Sensitized Photocatalysts

Interfacial electron injection from a photoexcited surface-immobilized dye to a semiconductor substrate is a key reaction for dye-sensitized photocatalysts. We previously reported that the molecular orientation of heteroleptic Ir(III) photosensitizer on the TiO2 nanoparticle surface was important for efficient interfacial electron injection. In this work, to overcome the weak light absorption ability of heteroleptic Ir(III) photosensitizer and to improve the photoinduced charge-separation efficiency at the dye–semiconductor interface, we synthesized two heteroleptic Ir(III) complexes with different coumarin dyes, [Ir(C6)2(H4CPbpy)]Cl and [Ir(C30)2(H4CPbpy)]Cl [Ir-CX; X = 6 or 30; HC6 = 3-(2-enzothiazolyl)-7-(diethylamino)coumarin, HC30 = 3-(2-N-methylbenzimidazolyl)-7-N,N-diethylaminocoumarin, H4CPbpy = 4,4′-bis(methylphosphonic acid)-2,2′-bipyridine], as the cyclometalated ligands and immobilized them on the surface of Pt-cocatalyst-loaded TiO2 nanoparticles. Ultraviolet-visible absorption and emission spectroscopy revealed that the singlet ligand-centered (1LC) absorption and triplet 3LC emission bands of Ir-C30 occurred at shorter wavelengths than those of Ir-C6, while time-dependent density-functional-theory data suggested that the ligand-to-ligand charge transfer (LLCT) excited states of the two complexes were comparable. The photocatalytic H2 evolution activity of the Ir-C6-sensitized Pt-TiO2 nanoparticles (Ir-C6@Pt-TiO2) under visible light irradiation (λ > 420 nm) was higher than that of Ir-C30@Pt-TiO2. In contrast, their activities were comparable under irradiation with monochromatic light (λ = 450 ± 10 nm), which is absorbed comparably by both Ir-CX complexes. These results suggest that the internal conversion from the higher-lying LC state to the LLCT state effectively occurs in both Ir-CX complexes to trigger electron injection to TiO2.


Syntheses
Caution! Although we did not encounter any difficulties, most of the chemicals used in this study are potentially harmful and should be used in small quantities and handled with care in a fume hood. All commercially available starting materials, including the coumarin dyes (HC6 and HC30), were used as received without further purification. The TiO2 nanoparticles (SSP-25, anatase, diameter: ~9 nm) were purchased from Sakai Chemical Industry Co. Ltd. Pt-TiO2 (Pt: 2.5 wt%, ~2.4 nm in diameter) was prepared using a previously reported photodeposition method [41] and characterized by powder X-ray diffraction (PXRD) and transmission electron microscope (TEM) measurements ( Figure S1). The dichloride-bridged Ir(III) complexes [Ir(C6)2Cl]2 and [Ir(C30)2Cl]2 and the H4CPbpy ligand were prepared according to methods reported in the literature [42,43] [Ir(C6)2Cl]2 (75.5 mg, 40.7 μmol) and H4CPbpy (28.8 mg, 83.7 μmol) were added to 18 mL THF/H2O mixture (1:1 v/v) in a capped vial. The mixture was heated at 130 °C for 20 min under microwave irradiation (Biotage Initiator + ). The reaction mixture was cooled naturally to room temperature (293 K) and then filtered. Approximately 130 mL of CH3CN was added to the filtrate, which was then incubated for 3 d at 298 K to afford the target complex Ir-C6 as yellow crystals. Yield: 50.3 mg (40.7 μmol, 49%). 1

Syntheses
Caution! Although we did not encounter any difficulties, most of the chemicals used in this study are potentially harmful and should be used in small quantities and handled with care in a fume hood. All commercially available starting materials, including the coumarin dyes (HC6 and HC30), were used as received without further purification. The TiO 2 nanoparticles (SSP-25, anatase, diameter:~9 nm) were purchased from Sakai Chemical Industry Co. Ltd., Osaka, Japan, Pt-TiO 2 (Pt: 2.5 wt%,~2.4 nm in diameter) was prepared using a previously reported photodeposition method [41] and characterized by powder X-ray diffraction (PXRD) and transmission electron microscope (TEM) measurements ( Figure S1). The dichloride-bridged Ir(III) complexes [Ir(C6) 2  The nanoparticle photocatalyst Ir-ppy@Pt-TiO 2 was prepared according to our previously reported method [25,44]. The other two photocatalysts, Ir-C6@Pt-TiO 2 and Ir-C30@Pt-TiO 2 , were prepared by a similar method with minor modifications as follows: Ir-C6 or Ir-C30 (40.4 or 41.5 µmol) was dissolved in a basic THF/H 2 O (73 mL) or MeOH/H 2 O (36 mL) mixed solvent (1:1 v/v for immobilization of Ir-C6 and Ir-C30, respectively) by adding a 1 M NaOH aqueous solution (pH 11). Next, Pt-TiO 2 nanoparticle powder (50.0 mg) was added to the Ir-C6 or Ir-C30 solution and stirred continuously for 3 d in the dark. The afforded Ir(III)-photosensitizer-immobilized Pt-TiO 2 nanoparticles (Ir-C6@Pt-TiO 2 or Ir-C30@Pt-TiO 2 , respectively) were collected by ultracentrifugation (50,000 rpm; 15 min) and the supernatant solution was manually removed by using a pipette. After washing twice with~6 mL of the same mixed solvent used for immobilization (1:1 v/v) without basification and deionized water, the product was dried in vacuo at 298 K for 1 d. The amount of Ir(III) photosensitizer immobilized on the Pt-TiO 2 nanoparticle surface was estimated by X-ray fluorescence (XRF) and ultraviolet-visible (UV-vis) absorption spectroscopy of the supernatant solution ( Figure S2, Table S1, and the "Calculation of the amount of immobilized Ir(III) complexes on the Pt-TiO 2 nanoparticles" section in the ESI).

Characterization
Elemental analysis was performed using a CE-440 element analyzer at the Analysis Center, Hokkaido University. 1 H NMR spectra were recorded on a JEOL JNM-ECZ400S NMR spectrometer at room temperature, while energy-dispersive XRF spectra were recorded on a Bruker S2 PUMA analyzer using a Pd target. Dynamic light scattering (DLS) analysis was conducted using an OTSUKA ELSZ-1000SCI analyzer. IR spectra were recorded on a JASCO FT-IR 660 spectrophotometer using the KBr method. PXRD measurements were conducted using a Bruker D8 Advance diffractometer equipped with a graphite monochromator using Cu-Kα radiation. The single-crystal X-ray diffraction (SXRD) data of Ir-C30 were collected using a Rigaku XtaLAB synergy diffractometer equipped with Cu-Kα radiation [PhotonJet(Cu)]. A suitable crystal was mounted on a MicroMount using paraffin oil. The crystal was then cooled using an N 2 -flow-type temperature controller, and the diffraction data were processed using CrysAlisPRO software [45]. The structures were solved by the direct method and refined by full-matrix least-squares refinement using SHELXL [46]. The non-hydrogen atoms were refined anisotropically, while the hydrogen atoms were refined using the riding model. All calculations were performed using the Olex2 software package [47]. The crystallographic data and selected bond lengths and angles are listed in Tables S2 and S3 in the Supporting Information. Full crystallographic data were deposited with the Cambridge Crystallographic Data Center (CCDC-2062083).

Photophysical and Electrochemical Measurements
UV-vis absorption and luminescence spectra were recorded on a Shimadzu UV-2400PC spectrophotometer and JASCO FP-8600 spectrofluorometer, respectively. Quartz cells with a 1 cm optical path length were used for both spectroscopic analyses in the solution state. Prior to the luminescence measurements, the sample solutions were degassed by N 2 bubbling for 15 min. The emission quantum yields (Φ em ) were measured using a Hamamatsu C9920-02 absolute photoluminescence quantum yield measurement system equipped with an integrating sphere apparatus and a 150 W continuous-wave xenon light source. Emission lifetime measurements were conducted using a Hamamatsu Photonics C4334 system equipped with a streak camera as the photodetector and a nitrogen laser as the excitation light source (λ ex = 337 nm).
Cyclic voltammetry (CV) was performed using a HOKUTO DENKO HZ-3000 electrochemical measurement system equipped with Pt wire and Ag/Ag + electrodes as the counter and reference electrodes, respectively. An Ir(III) complex-modified indium tin oxide (ITO) electrode was used as the working electrode. The ITO electrode was prepared according to a slightly modified literature method [48] as follows: A bare ITO electrode was cleaned by heating in basic hydrogen peroxide solution for 30 min at 80 • C and then washed with deionized water. Next, the pre-cleaned ITO electrode was immersed in a

Photocatalytic Hydrogen Evolution Reactions
For the photochemical H 2 evolution reactions, each sample was prepared using a hand-made Schlenk-flask-equipped quartz cell (volume: 257 mL). A typical sample preparation was as follows: Ir(III)-photosensitizer-immobilized Pt-TiO 2 nanoparticles and L-ascorbic acid (AA) were added to the quartz cell as the photocatalyst and sacrificial electron donor (SED), respectively. The total sample volume was adjusted to 10 mL by adding a CH 3 CN/deionized water (1:1 v/v) mixture. Each sample flask was doubly sealed using rubber septa and degassed by Ar bubbling for 1 h. Prior to irradiation, the gas (0.3 mL) was collected from the headspace using a gas-tight syringe (Hamilton 1001LTN) and analyzed by gas chromatography (GC, Shimadzu GC-14B) to confirm Ar purging. Subsequently, the samples were irradiated using a 300 W xenon lamp (MAX-303, ASAHI Spectra) in a water bath (293 K) combined with a visible-light-passed mirror module (385 nm < λ < 740 nm) and a longpass (SCF-50S-42L, λ > 420 nm) or band-pass (HMX0450, λ = 450 ± 10 nm) filter at room temperature. The gas samples (0.3 mL) were collected from the headspace at each analysis time (15 or 60 min) to determine the amount of H 2 evolved as a function of irradiation time. The AQY (Φ) was calculated using the equation: where N e represents the number of reacted electrons, N H2 is the number of evolved H 2 molecules, and N p is the number of incident photons.

Theoretical Calculations
TD-DFT calculations were performed using the CAM-B3LYP functional with SDD(Ir) and 6-31+G** (other atoms) basis sets using Gaussian 09W [49][50][51][52]. Geometry optimization was achieved using the B3LYP functional and the same basis sets [53]. In these calculations, the Et 2 N-group of the coumarin ligand was replaced by the Me 2 N-group and the two methylenephosphonate groups attached to the H 4 CPbpy ligand were replaced by H atoms to reduce the calculation cost. The Cartesian coordinates of the optimized structures are listed in Table S4 in the supporting information. Molecular orbital (MO) diagrams for all the complexes were reproduced using GaussView 5.0 (Wallingford, CT, USA) [54]. Figure 2 shows the molecular structure of Ir-C30 determined by SXRD analysis. The Ir(III) center was surrounded by two cyclometalated C30 ligands and one CPbpy ligand to form the octahedral IrC 2 N 4 coordination geometry. The coordinating C atoms of the C30 ligands were located at the trans positions to the N atoms of the CPbpy ligand, as observed for the heteroleptic Ir(III) cyclometalated complexes. The N-Ir-C bond angles of the C30 ligand in Ir-C30 (~79 • ) were near-identical to those of the C6 ligand in [Ir(C6) 2 (vacac)] (~80 • , vacac = allylacetoacetate) reported by DeRosa et al. [55], suggesting that the effect of functional group substitution, from the thiazole to methylimidazole ring, on the Ir(III) coordination environment was negligible (Table S3). In contrast, the torsion angle between the coumarin ring and thiazole/methyl-imidazole ring was notable. Indeed, these two rings were located in almost the same plane in [Ir(C6) 2 (vacac)], while the methylimidazole ring was largely twisted against the coumarin ring with a torsion angle ranging between 23 and 27 • , probably because of the steric repulsion between the methyl group attached to the imidazole ring and the carbonyl group of the coumarin ring. The Ir(III) center was surrounded by two cyclometalated C30 ligands and one CPbpy ligand to form the octahedral IrC2N4 coordination geometry. The coordinating C atoms of the C30 ligands were located at the trans positions to the N atoms of the CPbpy ligand, as observed for the heteroleptic Ir(III) cyclometalated complexes. The N-Ir-C bond angles of the C30 ligand in Ir-C30 (~79°) were near-identical to those of the C6 ligand in [Ir(C6)2(vacac)] (~80°, vacac = allylacetoacetate) reported by DeRosa et al. [55], suggesting that the effect of functional group substitution, from the thiazole to methylimidazole ring, on the Ir(III) coordination environment was negligible (Table S3). In contrast, the torsion angle between the coumarin ring and thiazole/methyl-imidazole ring was notable. Indeed, these two rings were located in almost the same plane in [Ir(C6)2(vacac)], while the methyl-imidazole ring was largely twisted against the coumarin ring with a torsion angle ranging between 23 and 27°, probably because of the steric repulsion between the methyl group attached to the imidazole ring and the carbonyl group of the coumarin ring.

Photophysical and Electrochemical Properties
To reveal the role of the coumarin ligand on the photophysical properties of the Ir(III) cyclometalated complexes, the UV-vis absorption and luminescence spectra of Ir-C6 and Ir-C30 were obtained in the solution state at 298 K ( Figure 3). Brown, light blue, red, and pale pink ellipsoids represent C, N, O, and P atoms, respectively, and H atoms are omitted for clarity. Displacement parameters are drawn at the 50% probability level.

Photophysical and Electrochemical Properties
To reveal the role of the coumarin ligand on the photophysical properties of the Ir(III) cyclometalated complexes, the UV-vis absorption and luminescence spectra of Ir-C6 and Ir-C30 were obtained in the solution state at 298 K ( Figure 3).  As already reported by several groups, the Ir(III) complex [Ir(C6)2(bpy)] + without the methylenephosphonic acid linkers exhibited a strong absorption band at 483 nm, which originated from the LC transition in the C6 ligand coupled with the metal-ligand-to-ligand charge transfer (MLLCT) transition [32,40]. The two Ir(III) complexes Ir-C6 and Ir-C30, with two phosphonic acid linkers, also exhibited similar absorption and emission bands in the visible region. The absorption and emission bands observed for Ir-C6 were slightly shifted to shorter wavelengths (~8 nm) than those of [Ir(C6)2(bpy)] + (Table 1), attributed to the electron-donating effect of the -CH2-PO3H2 linker attached to the bpy ligand. The absorption and emission bands observed for Ir-C30 were located at a shorter (by ~20 nm) wavelength than those of Ir-C6. Considering that the thiazole ring in C6 was replaced by the methyl imidazole ring in the C30 ligand, the spectral difference between Ir-C6 and Ir- As already reported by several groups, the Ir(III) complex [Ir(C6) 2 (bpy)] + without the methylenephosphonic acid linkers exhibited a strong absorption band at 483 nm, which originated from the LC transition in the C6 ligand coupled with the metal-ligandto-ligand charge transfer (MLLCT) transition [32,40]. The two Ir(III) complexes Ir-C6 and Ir-C30, with two phosphonic acid linkers, also exhibited similar absorption and emission bands in the visible region. The absorption and emission bands observed for Ir-C6 were slightly shifted to shorter wavelengths (~8 nm) than those of [Ir(C6) 2 (bpy)] + (Table 1), attributed to the electron-donating effect of the -CH 2 -PO 3 H 2 linker attached to the bpy ligand. The absorption and emission bands observed for Ir-C30 were located at a shorter (by~20 nm) wavelength than those of Ir-C6. Considering that the thiazole ring in C6 was replaced by the methyl imidazole ring in the C30 ligand, the spectral difference between Ir-C6 and Ir-C30 was assigned to the difference in the electron-donating ability of these functional groups. Indeed, the methyl-imidazole ring is generally electron-poorer than the thiazole group because of the higher electronegativity of the N atom over that of S, leading to a higher LC transition energy for Ir-C30 over that for Ir-C6. Table 1. Comparison of the photophysical and electrochemical data of Ir-C30, Ir-C6, and Ir-ppy.
Complex  Figure S5). g E* ox was estimated using the equation E* ox = E ox − E 00 , where E 00 was approximated as λ em .
This trend agrees well with that observed for the UV-vis absorption spectra of the C6 and C30 ligands in solution ( Figure S3 and Table S5). The lowest-energy absorption band of C30 was located at a shorter (by~44 nm) wavelength than that of C6. Moreover, a marked difference was also observed in the emission lifetime and quantum yield: The emission lifetime of Ir-C30 was less than half that of Ir-C6 ( Figure S4), while the emission quantum yield of Ir-C30 at 298 K in the solution state was approximately one-fourth that of Ir-C6 (Table 1). As a result, the non-radiative rate constant (k nr ) of Ir-C30 was estimated to be approximately four times higher than that of Ir-C6, while the radiative rate constants (k r ) of the two complexes were comparable. These results indicate that the functional group difference in the coumarin ligand exerted considerable effects not only on the emission wavelength but also on the deactivation process from the photo-excited state. The faster non-radiative deactivation of Ir-C30 over that of Ir-C6 was attributed to the structural distortion around the C30 ligand, as discussed in Section 3.1. This nonplanar structure of the C30 ligand may stabilize the non-radiative metal-centered (d-d) excited state. This is also consistent with the electrochemical measurements, wherein the quasi-reversible Ir(IV)/Ir(III) redox wave was observed at +1.34 V (vs. NHE, Figure S5) for Ir-C30, which was slightly less positive than that of Ir-C6 (+1.41 V). This was probably due to the weaker ligand field splitting of the C30 ligand over that of C6. On the other hand, the emission lifetimes of both Ir(III)-coumarin complexes were at least one order of magnitude longer than that of the simple cyclometalated complex Ir-ppy. Furthermore, the redox potentials at the photo-excited state Ir(IV)/Ir(III)* of Ir-CX were estimated to be slightly more negative than that of Ir-ppy (E* ox in Table 1), suggesting the potential of these two Ir-CX complexes as photosensitizers for the H 2 evolution reaction.
To obtain further insight into the differences between Ir-C6 and Ir-C30, TD-DFT calculations were conducted for the simplified Ir-C6 and Ir-C30 complexes, where the -CH 2 PO 3 H 2 groups attached to the bpy ligand and the diethylamino group of the coumarin (C6 and C30) ligands were replaced by H atoms and the dimethylamino group, respectively. The bond lengths around the central Ir(III) cation in the optimized Ir-C6 and Ir-C30 structures are reasonably comparable to those observed in the X-ray structure of Ir-C30 (Table S6). In the optimized structures, the planarity of the coumarin ligands was notably different, as discussed in Section 3.1. The dihedral angle between the coumarin and methyl imidazole rings was estimated to be 18.4 • for Ir-C30 and near-zero for Ir-C6 . Considering that the C6 ligand in the [Ir(C6) 2 (vacac)] complex was also near-planar, the non-planarity of the C30 ligand was attributed to the steric effect of the methyl group attached to the imidazole ring. A schematic energy diagram estimated by TD-DFT calculations is shown in Figure 4, while the estimated transition energy and oscillator strength of each transition are listed in Tables S7 and S8. The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of Ir-C6 were localized on the π and π*orbitals of C6 and the bpy ligands, respectively. The second and third HOMOs (H-1 and H-2, respectively) and second and third LU-MOs (L+1 and L+2, respectively) were assigned to the π and π* orbitals of the C6 ligand. These results are consistent with the results reported by several groups [37,40]. Although a similar trend was confirmed for Ir-C30′, the orbital energies were positively shifted, probably due to the poorer electron-donating ability of C30 over that of C6. The positive shift of the HOMO is consistent with the negative shift of the Ir(IV)/Ir(III) redox potential observed in the CV measurements (Table 1). TD-DFT calculations also suggest a notable difference between these two complexes. The spin-allowed lowest-energy electronic (S0 to S1) transition was suggested to be an LLCT (from CX to bpy) transition from the HOMO to LUMO in both complexes (as indicated by the red arrows in Figure 4). The estimated transition energy was comparable (~400 nm) for both complexes and the oscillator strength was commonly small (f < 0.03, Tables S7 and S8), probably because of the small orbital overlap of these MOs. The second lowest-energy (S0 to S2) transition for Ir-C6′ was assignable to the LC transition in the C6 ligand (HOMO to L+1, 396 nm) with a large oscillator strength (f = 0.48), and the transition energy difference between the LC and LLCT transitions was negligible. In contrast, the same LC transition for Ir-C30′ appeared as the third-lowest energy transition with a higher transition energy (S3, HOMO to L+2, 376 nm), resulting in an energy difference between the LLCT and LC transition states. The esti- The second and third HOMOs (H-1 and H-2, respectively) and second and third LUMOs (L+1 and L+2, respectively) were assigned to the π and π* orbitals of the C6 ligand. These results are consistent with the results reported by several groups [37,40]. Although a similar trend was confirmed for Ir-C30 , the orbital energies were positively shifted, probably due to the poorer electron-donating ability of C30 over that of C6. The positive shift of the HOMO is consistent with the negative shift of the Ir(IV)/Ir(III) redox potential observed in the CV measurements (Table 1). TD-DFT calculations also suggest a notable difference between these two complexes. The spin-allowed lowest-energy electronic (S 0 to S 1 ) transition was suggested to be an LLCT (from CX to bpy) transition from the HOMO to LUMO in both complexes (as indicated by the red arrows in Figure 4). The estimated transition energy was comparable (~400 nm) for both complexes and the oscillator strength was commonly small (f < 0.03, Tables S7 and S8), probably because of the small orbital overlap of these MOs. The second lowest-energy (S 0 to S 2 ) transition for Ir-C6 was assignable to the LC transition in the C6 ligand (HOMO to L+1, 396 nm) with a large oscillator strength (f = 0.48), and the transition energy difference between the LC and LLCT transitions was negligible. In contrast, the same LC transition for Ir-C30 appeared as the third-lowest energy transition with a higher transition energy (S 3 , HOMO to L+2, 376 nm), resulting in an energy difference between the LLCT and LC transition states. The estimated LC transition energy difference between Ir-C30 and Ir-C6 (~20 nm) qualitatively agreed with the positions of the LC transition bands observed in the UV-vis absorption and emission spectra of Ir-CX ( Figure 3). Notably, the lowest excited triplet state in both complexes was suggested to be in the 3 LC state for both complexes, and the 3 LLCT was found to be the fourth-lowest (T 4 ) excited state with an excitation energy similar to that of the 1 LLCT state (Tables S7 and S8). Thus, these TD-DFT calculations suggest that the coumarin ligand replacement, from C6 to C30, does not affect the LLCT excited state of the Ir-CX complex but affects the LC excited state.

Synthesis and Characterization of the Ir-PS-Immobilized Pt-TiO 2 Nanoparticles
The immobilization of the Ir-C6 and Ir-C30 complexes on the Pt-TiO 2 nanoparticle surface was conducted in basified THF/H 2 O and MeOH/H 2 O (1:1 v/v), respectively (Experimental Section). Almost the same immobilization procedure used for Ir-ppy was applicable to Ir-C30 [25] but not to Ir-C6, probably because of the low solubility of the latter. The UV-vis absorption spectra of the supernatant solutions isolated from the immobilization reaction enabled us to estimate the amount of these immobilized Ir(III) complexes (ESI and Figure S2). The obtained results are summarized and compared to those of Ir-ppy in Table 2. The amounts of immobilized Ir-C6 and Ir-C30 were comparable (~130 nmol/1 mg TiO 2 ) and slightly lower than our previously reported value for Ir-ppy [25]. The surface coverage and molecular footprint (i.e., area occupied by one Ir(III) complex molecule) were calculated assuming that the TiO 2 nanoparticles were spherical with a diameter of 9 nm (ESI). The Ir-CX molecular footprint was estimated to be~2.1 nm 2 , which is slightly larger than that of the Ir-ppy complex. These results are reasonable because Ir-CX complexes have larger coumarin dyes as the cyclometalated ligands than the ppy ligand in Ir-ppy ( Figure 1). Notably, the estimated molecular footprint based on the UV-vis absorption spectra of the supernatant solution qualitatively agreed with the value estimated from the X-ray structure of Ir-C30 (~2.2 nm 2 , Figure S6), suggesting that the Pt-TiO 2 nanoparticle surface was almost fully covered by the Ir-CX complexes. Further evidence of Ir-CX immobilization on the Pt-TiO 2 nanoparticle surface was obtained by XRF spectroscopy ( Figure 5). The characteristic peaks assigned to the Ir-L and Pt-L radiation originating from the surface-immobilized Ir-CX and Pt cocatalyst were observed clearly for both nanoparticles. The comparable peak intensities of the Ir L radiation of Ir-C30@Pt-TiO 2 and Ir-C6@Pt-TiO 2 provide further evidence that the amount of Ir-CX immobilized on the Pt-TiO 2 surface of the two nanoparticles was near-identical. The comparable intensity of the Pt-L radiation for both nanoparticles also indicates that the Pt cocatalyst loaded on the TiO 2 surface was barely detached during the Ir-CX immobilization reaction. The Ir-CX photoluminescence was negligible for both Ir-CX@Pt-TiO 2 nanoparticles in the solid state at 293 K, suggesting electron transfer quenching from the photoexcited Ir-CX* to TiO 2 . a Estimated based on the absorbance in the UV-vis spectra of each supernatant solution ( Figure S2). b Ref. [25].

Photocatalytic H2 Evolution Reactions
The results of the photocatalytic H2 evolution reactions driven by the Ir(III)-coumarin-dye-immobilized Pt-TiO2 nanoparticles, Ir-C6@Pt-TiO2 or Ir-C30@Pt-TiO2, in the presence of a sacrificial AA donor aqueous solution under visible light irradiation (λ > 420 nm) are shown in Figure 6a. The amount of evolved H2, turnover number (TON), and turnover frequency (TOF) per Ir(III) dye are listed in Table 3. In 0.2 M AA solution, both nanoparticle photocatalysts were highly active, with estimated TOFs of ~1299 and 699 for Ir-C6@Pt-TiO2 and Ir-C30@Pt-TiO2, respectively, which are at least two orders of magnitude larger than that of the previously reported Ir-ppy@Pt-TiO2 under near-identical conditions.

Photocatalytic H 2 Evolution Reactions
The results of the photocatalytic H 2 evolution reactions driven by the Ir(III)-coumarindye-immobilized Pt-TiO 2 nanoparticles, Ir-C6@Pt-TiO 2 or Ir-C30@Pt-TiO 2 , in the presence of a sacrificial AA donor aqueous solution under visible light irradiation (λ > 420 nm) are shown in Figure 6a. The amount of evolved H 2 , turnover number (TON), and turnover frequency (TOF) per Ir(III) dye are listed in Table 3. In 0.2 M AA solution, both nanoparticle photocatalysts were highly active, with estimated TOFs of~1299 and 699 for Ir-C6@Pt-TiO 2 and Ir-C30@Pt-TiO 2 , respectively, which are at least two orders of magnitude larger than that of the previously reported Ir-ppy@Pt-TiO 2 under near-identical conditions.

Photocatalytic H2 Evolution Reactions
The results of the photocatalytic H2 evolution reactions driven by the Ir(III)-coumarin-dye-immobilized Pt-TiO2 nanoparticles, Ir-C6@Pt-TiO2 or Ir-C30@Pt-TiO2, in the presence of a sacrificial AA donor aqueous solution under visible light irradiation (λ > 420 nm) are shown in Figure 6a. The amount of evolved H2, turnover number (TON), and turnover frequency (TOF) per Ir(III) dye are listed in Table 3. In 0.2 M AA solution, both nanoparticle photocatalysts were highly active, with estimated TOFs of ~1299 and 699 for Ir-C6@Pt-TiO2 and Ir-C30@Pt-TiO2, respectively, which are at least two orders of magnitude larger than that of the previously reported Ir-ppy@Pt-TiO2 under near-identical conditions.  After the first reaction for 20 h, the AA solution was added to adjust the concentration to 0.02 M, after which the reaction solution was degassed by Ar bubbling for 1 h before the second irradiation. A 300 W xenon lamp (0.25 W) with a longpass filter (λ > 420 nm) was used as the irradiation source. The higher activities of Ir-CX@Pt-TiO 2 indicate that the strong light adsorption ability originating from the coumarin ligand contributes to the photocatalytic H 2 evolution reaction. The TON after 6 h irradiation of Ir-C6@Pt-TiO 2 was estimated to be 1.85 times higher than that of Ir-C30@Pt-TiO 2 . Considering that a constant amount of Ir(III)-coumarin complex ([Ir] = 10 µM) was used in these reactions, the higher photocatalytic activity of Ir-C6@Pt-TiO 2 under these conditions were ascribed to the wider and stronger absorption bands of Ir-C6 compared to those of Ir-C30 ( Figure 3). In fact, the integral of the molar absorption coefficient of Ir-C6 in the irradiated wavelength range (>420 nm) was 1.62 times larger than that of Ir-C30. Further, the photocatalytic H 2 evolution activities of these two nanoparticles under monochromatic light irradiation (λ = 450 ± 10 nm) were near-similar ( Figure S7) because of the comparable molar absorption coefficients at 450 nm (ε at 450 nm = 7.13 × 10 4 and 6.72 × 10 4 for Ir-C6 and Ir-C30, respectively). Although the evolved amount of H 2 under the monochromatic light irradiation was significantly lower because of the lower number of irradiated photons, the AQYs for H 2 evolution under these conditions were estimated to be 2.3 and 2.1% for Ir-C6@Pt-TiO 2 and Ir-C30@Pt-TiO 2 , respectively. These values are nearly double that of RuCP 2 @Pt-TiO 2 (1.4%, RuCP 2 = [Ru(bpy) 2 (CPbpy)] 2− , bpy = 2,2 -bipyridine), which is composed of well-known tris-bipyridyl-type Ru(II) dyes and Pt-TiO 2 nanoparticles [44]. These results suggest a superior performance of the Ir(III) heteroleptic photosensitizer for photoinduced charge separation. The comparable AQY values for Ir-CX@Pt-TiO 2 are reasonable because of the comparable driving force for electron injection to TiO 2 from the LLCT excited state, as suggested by the TD-DFT data. On the other hand, the difference in the H 2 evolution photocatalytic activity between these two nanoparticles became more pronounced in the more diluted (20 mM) AA aqueous solution. Indeed, the TON of Ir-C6@Pt-TiO 2 in 0.02 M AA solution was 856, which is 2.65 times higher than that of Ir-C30@Pt-TiO 2 . This different dependence on the AA concentration suggests that the reactivity between the surface-immobilized Ir-C30 and AA is lower than that of Ir-C6. As discussed in Section 3.2, the HOMOs of both complexes are almost localized on the coumarin ligands ( Figure 4). The two coumarin ligands, C6 and C30, have very similar molecular structures; however, the methyl group attached to the N atom of the C30 ligand may suppress electron donation from AA by the steric effect. In addition, as shown in Figure 7, the Ir(IV)/Ir(III) redox potential of Ir-C30 (1.34 V vs. NHE) was slightly more negative than that of Ir-C6 (1.41 V vs. NHE), suggesting a smaller driving force for electron donation from AA to the photooxidized Ir(IV) species. It should be noted that the Ir(IV)/Ir(III)* redox potentials estimated by the combination of cyclic voltammograms and emission spectra (E* ox in Table 1) may be underestimated because the higher-lying and non-emissive 3 LLCT state than the emissive 3 LC state could contribute to the photo-induced electron injection. Unfortunately, the emission quenching experiments to evaluate the reactivity between the photoexcited Ir-CX* and AA failed because of the very weak emission intensity at pH 4.5 condition ( Figure S8). The activity of Ir-C6@Pt-TiO 2 was slightly lower than that of the simple mixture Pt-TiO 2 nanoparticles and Ir-C6 without immobilization ( Figure S9), probably because the electron donation from AA to the surface-immobilized Ir-C6 would occur more slowly than that to free Ir-C6 in the solution. Notably, the high activity of Ir-C6@Pt-TiO 2 was retained during two cycle reactions (Figure 6b), and the amounts of H 2 produced in both cycles were~200 µmol. These results suggest that all the AA (0.02 M) was consumed as the two-electron donor for the H 2 evolution reaction. The absorption band derived from the surface-detached Ir-C6 was almost negligible in the UV-vis absorption spectra of the supernatant solution ( Figure S10). Similar results were also obtained for Ir-C30@Pt-TiO 2 ( Figure S11). These results clearly indicate that the performance of the Ir-CX@Pt-TiO 2 nanoparticle photocatalyst is superior. activity of Ir-C6@Pt-TiO2 was slightly lower than that of the simple mixture Pt-TiO2 nanoparticles and Ir-C6 without immobilization ( Figure S9), probably because the electron donation from AA to the surface-immobilized Ir-C6 would occur more slowly than that to free Ir-C6 in the solution. Notably, the high activity of Ir-C6@Pt-TiO2 was retained during two cycle reactions (Figure 6b), and the amounts of H2 produced in both cycles were ~200 μmol. These results suggest that all the AA (0.02 M) was consumed as the two-electron donor for the H2 evolution reaction. The absorption band derived from the surface-detached Ir-C6 was almost negligible in the UV-vis absorption spectra of the supernatant solution ( Figure S10). Similar results were also obtained for Ir-C30@Pt-TiO2 ( Figure S11). These results clearly indicate that the performance of the Ir-CX@Pt-TiO2 nanoparticle photocatalyst is superior.

Conclusions
In this work, two Ir(III)-coumarin complexes, Ir-C6 and Ir-C30, were synthesized as visible light photosensitizers, for the H2 evolution reaction. UV-vis absorption and emission spectroscopy revealed that the LC absorption and emission bands of Ir-CX were blueshifted by the replacement of the coumarin ligand from C6 to C30, whereas the LLCT transition energies estimated by TD-DFT calculations were comparable. These Ir-CX complexes were successfully immobilized on the surface of Pt-cocatalyst-loaded TiO2 nanoparticles to form H2-evolving Ir(III)-dye-sensitized photocatalysts with the general formula Ir-CX@Pt-TiO2. Both photocatalysts were highly active toward H2 evolution under irradiation of visible light (λ > 420 nm, TON: ~1300 for Ir-C6@Pt-TiO2 and ~700 for Ir-C30@Pt-TiO2 after 6 h irradiation) in the presence of 0.2 M L-AA electron donor aqueous solution. The AQYs of the H2 evolution reaction under monochromatic light irradiation (λ = 450 ± 10 nm) were comparable (2.3 and 2.1% for Ir-C6@Pt-TiO2 and Ir-C30@Pt-TiO2, respectively), owing to the comparable LLCT excited states, which are the key states for electron injection to TiO2 nanoparticles. Thus, the superior performance of Ir-C6@Pt-TiO2 for H2 evolution over that of Ir-C30@Pt-TiO2 under visible light originated from the wider and stronger light absorption ability of Ir-C6 (ε475 = 10.3 × 10 4 ) compared to that of Ir-C30 (ε452 = 6.76 × 10 4 ). Additionally, the energy difference between the LC and LLCT states of Ir-C6 may be sufficient for rapid internal conversion from the higher-lying LC to LLCT state. Further study on heteroleptic Ir(III) photosensitizers with intramolecular chargeseparated LLCT states for solar water splitting is now in progress.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1, Calculation of the amount of immobilized Ir(III) complexes on the Pt-TiO2 nanoparticles and the surface

Conclusions
In this work, two Ir(III)-coumarin complexes, Ir-C6 and Ir-C30, were synthesized as visible light photosensitizers, for the H 2 evolution reaction. UV-vis absorption and emission spectroscopy revealed that the LC absorption and emission bands of Ir-CX were blue-shifted by the replacement of the coumarin ligand from C6 to C30, whereas the LLCT transition energies estimated by TD-DFT calculations were comparable. These Ir-CX complexes were successfully immobilized on the surface of Pt-cocatalyst-loaded TiO 2 nanoparticles to form H 2 -evolving Ir(III)-dye-sensitized photocatalysts with the general formula Ir-CX@Pt-TiO 2 . Both photocatalysts were highly active toward H 2 evolution under irradiation of visible light (λ > 420 nm, TON:~1300 for Ir-C6@Pt-TiO 2 and~700 for Ir-C30@Pt-TiO 2 after 6 h irradiation) in the presence of 0.2 M L-AA electron donor aqueous solution. The AQYs of the H 2 evolution reaction under monochromatic light irradiation (λ = 450 ± 10 nm) were comparable (2.3 and 2.1% for Ir-C6@Pt-TiO 2 and Ir-C30@Pt-TiO 2 , respectively), owing to the comparable LLCT excited states, which are the key states for electron injection to TiO 2 nanoparticles. Thus, the superior performance of Ir-C6@Pt-TiO 2 for H 2 evolution over that of Ir-C30@Pt-TiO 2 under visible light originated from the wider and stronger light absorption ability of Ir-C6 (ε 475 = 10.3 × 10 4 ) compared to that of Ir-C30 (ε 452 = 6.76 × 10 4 ). Additionally, the energy difference between the LC and LLCT states of Ir-C6 may be sufficient for rapid internal conversion from the higher-lying LC to LLCT state. Further study on heteroleptic Ir(III) photosensitizers with intramolecular charge-separated LLCT states for solar water splitting is now in progress.

Supplementary Materials:
The following are available online at https://www.mdpi.com/article/10 .3390/en14092425/s1, Calculation of the amount of immobilized Ir(III) complexes on the Pt-TiO 2 nanoparticles and the surface coverage of Ir(III) complexes per unit area of TiO 2 , Figure S1: PXRD pattern and TEM image of Pt-TiO 2 nanoparticles, Figure S2: UV-vis absorption spectra of the supernatant solutions obtained from the immobilization reaction for Ir-C6@Pt-TiO 2 and Ir-C30@Pt-TiO 2 at 298 k, Figure S3: UV-vis absorption and emission spectra of C6 and C30, Figure S4: Emission decays of Ir-C6 and Ir-C30 in 12 µM MeOH/H 2 O solution, Figure S5: Cyclic voltammograms of Ir-C30 and Ir-C6 immobilized on ITO working electrode, Figure S6: Estimated area occupied by one Ir-C30 molecule based on the X-ray structure, Figure S7: Photocatalytic H 2 evolution reaction under monochromatic light irradiation driven by the Pt-cocatalyst-loaded TiO 2 nanoparticles Ir-C6@Pt-TiO 2 and Ir-C30@Pt-TiO 2 , Figure S8: Changes of UV-vis absorption and emission spectra of Ir-C30 by lowering pH from 10 to 4.5 in MeOH/H 2 O mixed solvent at 298 K, Figure S9: Photocatalytic H 2 evolution reaction driven by the mixture of homogeneous Ir-C6 photosensitizer and Pt-TiO 2 nanoparticle, Figure S10: UV-vis absorption spectra of the supernatant solution obtained by centrifugation of the reaction solution of Ir-C6@Pt-TiO 2 after two cycles of 20 h photocatalytic H 2 evolution reactions, Figure S11: Two-cycle photocatalytic H 2 evolution reaction driven by Ir-C30@Pt-TiO 2 , Table S1: Absorbance of each supernatant solution and the calculated C B and M i values, Table S2: Crystal parameters and refinement data for Ir-C30, Table S3: Selected bond lengths and angles of Ir-C30, Table S4: Cartesian coordinates of the optimized Ir-C6 and Ir-C30 complexes, Table S5: Photophysical data of C6 and C30 in comparison with Ir-C30 and Ir-C6 in the solution state, Table S6: Structural comparison of the X-ray structure of Ir-C30 and optimized structures Ir-C30 and Ir-C6 by DFT calculations, Tables S7 and S8: Energy, oscillator strength and major contribution of spin-allowed transitions for Ir-C30 and Ir-C6 .