Electrochemical Properties of a Rhodium(III) Mono-Terpyridyl Complex and Use as a Catalyst for Light-Driven Hydrogen Evolution in Water

Molecular hydrogen (H2) is considered one of the most promising fuels to decarbonize the industrial and transportation sectors, and its photocatalytic production from molecular catalysts is a research field that is still abounding. The search for new molecular catalysts for H2 production with simple and easily synthesized ligands is still ongoing, and the terpyridine ligand with its particular electronic and coordination properties, is a good candidate to design new catalysts meeting these requirements. Herein, we have isolated the new mono-terpyridyl rhodium complex, [RhIII(tpy)(CH3CN)Cl2](CF3SO3) (Rh-tpy), and shown that it can act as a catalyst for the light-induced proton reduction into H2 in water in the presence of the [Ru(bpy)3]Cl2 (Ru) photosensitizer and ascorbate as sacrificial electron donor. Under photocatalytic conditions, in acetate buffer at pH 4.5 with 0.1 M of ascorbate and 530 μM of Ru, the Rh-tpy catalyst produces H2 with turnover number versus catalyst (TONCat*) of 300 at a Rh concentration of 10 μM, and up to 1000 at a concentration of 1 μM. The photocatalytic performance of Ru/Rh-tpy/HA–/H2A has been also compared with that obtained with the bis-dimethyl-bipyridyl complex [RhIII(dmbpy)2Cl2]+ (Rh2) as a catalyst in the same experimental conditions. The investigation of the electrochemical properties of Rh-tpy in DMF solvent reveals that the two-electrons reduced state of the complex, the square-planar [RhI(tpy)Cl] (RhI-tpy), is quantitatively electrogenerated by bulk electrolysis. This complex is stable for hours under an inert atmosphere owing to the π-acceptor property of the terpyridine ligand that stabilizes the low oxidation states of the rhodium, making this catalyst less prone to degrade during photocatalysis. The π-acceptor property of terpyridine also confers to the Rh-tpy catalyst a moderately negative reduction potential (Epc(RhIII/RhI) = −0.83 V vs. SCE in DMF), making possible its reduction by the reduced state of Ru, [RuII(bpy)(bpy•−)]+ (Ru−) (E1/2(RuII/Ru−) = −1.50 V vs. SCE) generated by a reductive quenching of the Ru excited state (*Ru) by ascorbate during photocatalysis. A Stern–Volmer plot and transient absorption spectroscopy confirmed that the first step of the photocatalytic process is the reductive quenching of *Ru by ascorbate. The resulting reduced Ru species (Ru−) were then able to activate the RhIII-tpy H2-evolving catalyst by reduction generating RhI-tpy, which can react with a proton on a sub-nanosecond time scale to form a RhIII(H)-tpy hydride, the key intermediate for H2 evolution.


Introduction
Molecular hydrogen (H 2 ) is considered as one of the best alternatives to fossil fuels to speed up the transition towards a carbon-neutral future [1]. However, 96% of hydrogen is produced from fossil fuels via a variety of processes, such as the steam reforming of methane, generating carbon dioxide at the same time. Therefore, to reach carbon neutrality, hydrogen should be cleanly produced using a renewable energy, such as sunlight, with water as a proton source. The pioneer work of Lehn and Sauvage demonstrated in the late 1970s that green hydrogen could be produced by the photo-induced reduction of protons via homogeneous photocatalytic systems [2,3]. These systems are usually composed of a sacrificial electron donor (SD) providing electrons to the system, a H 2evolving catalyst (Cat), and a photosensitizer (PS) promoting the photo-induced electron transfers between the three compounds. For forty years, many photocatalytic systems have been investigated and numerous molecular catalysts have been employed, the latter being based on noble metal (Pd, Pt, Rh) or earth-abundant metal complexes (Co, Ni, Fe, Mo) [4,5]. Among the latter, the catalytic properties of rhodium complexes have been rarely explored in aqueous solutions (Rh1-10) [3,[6][7][8][9][10][11][12][13][14] but more largely in hydro-organic solvent (Rh11-24) [10,[15][16][17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33] (Scheme 1), although the use of water is a prerequisite for water-splitting applications [34]. Rhodium catalysts active in water for H 2 production have employed a small variety of ligands, such as bipyridyl (Rh1-3, Rh8-9) [3,8,9,[11][12][13], cyclopentadienyl (Rh3) [9], diphenylphosphinobenzene-sulphonate (Rh4) [6,7], acetate (Rh5-9) [8,10], and diisocyanopropane (Rh10) [8], under the form of mono-or bi-metallic complexes (Scheme 1a). The H 2 -evolution with such rhodium catalysts generally proceeds via the transient formation of rhodium hydride species [34][35][36] generated upon their reduction in presence of protons. Among these Rh catalysts, only Rh2, Rh3, and Rh4 exhibited turnover numbers per catalyst (TON Cat ) superior to 100 in the presence of a ruthenium or iridium photosensitizer and ascorbate as SD in water, and TON Cat for the others catalysts do not exceed 10 [34]. In addition, when the Rh2 bis-bipyridyl catalyst is covalently linked to two Ru photosensitizers, the performance of photocatalytic system SD/PS-Cat is clearly improved due to the faster kinetics of the electron transfer between PS and Cat [12]. In this context, exploring rhodium complexes with other polypyridyl ligands is important for identifying new efficient and robust H 2 -evolving catalysts.
Herein, we report on the synthesis and structural characterization of a new rhodium monoterpyridyl complex, namely [Rh III (tpy)(CH 3 CN)Cl 2 ](CF 3 SO 3 ) (tpy = 2,2 ;6 ,2 -terpyridine) (denoted Rh-tpy, Scheme 2). We also investigated the electrochemical properties of this complex in organic solvent and the electrochemical generation and UV-visible spectroscopic characterization of the corresponding two-electron Rh I reduced species. The catalytic activity of Rh-tpy towards the light-driven H 2 production was tested in purely aqueous solution, in association with [Ru(bpy) 3 ] 2+ (Ru) as photosensitizer and ascorbate (HA − ) as sacrificial electron donor. In 2012, the group of Ogo investigated the capacity of the chemically synthetized rhodium(I) terpyridyl [Rh I (tpy)(CH 3 CN)](CF 3 SO 3 ) complex to form a rhodium(III) monohydride species in the presence of protons in CH 3 4 , with the generation of H 2 by reductive elimination [37]. In this study, we demonstrate that the Rh-tpy can act as a H 2 -evolving catalyst under photocatalytic conditions in acidic water (pH 4.0-4.5) with an efficiency competing with the benchmark rhodium(III) bis-dimethyl-bipyridyl complex [Rh III (dmbpy) 2 Cl 2 ] + (Rh2) (Scheme 1) for a low catalyst concentration of 1 µM, this latter being among the most efficient rhodium catalysts in water [11][12][13]. Finally, the first photo-induced steps involved in the H 2 production with the molecular Ru/Rh-tpy/HA − /H 2 A system are investigated by means of time-resolved emission and absorption spectroscopies. Molecules 2022, 27, x FOR PEER REVIEW 3 of 21 Scheme 1. Rhodium catalysts previously reported for light-driven H2 production in water (a) and in hydro-organic solvent (b).

Scheme 2.
Structures of the photosensitizer (Ru), the H2-evolving catalyst (Rh-tpy), and the sacrificial electron donor (HA − ) used in this work.

Synthesis and Crystal Structure of the [Rh III (tpy)(CH 3 CN)Cl 2 ](CF 3 SO 3 ) Complex
The synthesis of the Rh-tpy complex was inspired from that of the [Rh III (tpy)(OH)(H 2 O) 2 ] (NO 3 ) 2 complex isolated by Ogo (Scheme 3) [38]. First, the [Rh III (tpy)Cl 3 ] complex is generated by reacting RhCl 3 •3H 2 O with one equivalent of terpyridine in ethanol at 90 • C [39]. Then, in order to solubilize the rhodium complex in water for the light-driven H 2 production experiments, a chloride ligand in axial position was exchanged by an acetonitrile molecule by reacting the tris-chloro complex with AgCF 3 SO 3 in acetonitrile at 90 • C, leading to the formation of the [Rh III (tpy)(CH 3 CN)Cl 2 ](CF 3 SO 3 ) complex (Rh-tpy) with a yield of 50%. The water solubility of Rh-tpy is also favored by the presence of the triflate counter-anion.

Synthesis and Crystal Structure of the [Rh III (tpy)(CH3CN)Cl2](CF3SO3) Complex
The synthesis of the Rh-tpy complex was inspired from that of the [Rh III (tpy)(OH)(H2O)2](NO3)2 complex isolated by Ogo (Scheme 3) [38]. First, the [Rh III (tpy)Cl3] complex is generated by reacting RhCl3•3H2O with one equivalent of terpyridine in ethanol at 90 °C [39]. Then, in order to solubilize the rhodium complex in water for the light-driven H2 production experiments, a chloride ligand in axial position was exchanged by an acetonitrile molecule by reacting the tris-chloro complex with AgCF3SO3 in acetonitrile at 90 °C, leading to the formation of the [Rh III (tpy)(CH3CN)Cl2](CF3SO3) complex (Rh-tpy) with a yield of 50%. The water solubility of Rh-tpy is also favored by the presence of the triflate counter-anion. Single crystals of Rh-tpy suitable for X-ray crystallography were obtained by diffusion of diethyl ether into an acetonitrile solution of the rhodium complex ( Figure 1, Tables S1-S4). Rh-tpy crystallizes in the P1 space group and displays a distorted octahedral geometry around the rhodium atom. The three nitrogen atoms of the terpyridine and one chloride ligand are in the equatorial plane, while the nitrogen atom of CH3CN and one chloride ligand are in an axial position. The terpyridine ligand is quasi-planar, with low torsion angles between the central pyridine of the terpyridine and the other two pyridines (3.90° between pyridines with N(1) and N(2) and 1.70° between pyridines with N(2) and N(3)). The nitrogen atoms of the terpyridine, the chloride atom, and the rhodium form a slightly distorted square plane, with N(1)-Rh-N(2) and N(2)-Rh-N(3) angles of 81.07(6)° and 81.00(6)°, respectively, which are slightly smaller than the ideal value of 90° for a perfect square plane geometry. This distortion is induced by the terpyridine, which imposes 5-atom rings comprising the rhodium, two nitrogen atoms (N(1)/N (2) or N(2)/N(3)), and two terpyridine carbon atoms. The distance between rhodium and the central pyridine nitrogen (Ncentral) of the terpyridine (N(2)) is shorter (1.9386 (14) Å) than that between rhodium and the two distal nitrogen (Ndistal) of the terpyridine (2.0258(14) and 2.0417(14) Å for Rh-N(1) and Rh-N(3), respectively). The rhodium atom is almost equidistant from the two chloride atoms (2.3530(5) vs. 2.3067(4) Å, respectively, for Rh-Cl(1) and Rh-Cl(2)) and the distance between the nitrogen of acetonitrile ligand and the rhodium atom (2.0220(14) Å for Rh-N(21)) is very similar to those between the rhodium and the terpyridine nitrogen atoms (Rh-Ntpy). The distances and angles between the Rh center and the terpyridine ligand in Rh-tpy are very similar to those found in the [Rh III (tpy)(OH)(OH2)2](NO3)2 [38]. Single crystals of Rh-tpy suitable for X-ray crystallography were obtained by diffusion of diethyl ether into an acetonitrile solution of the rhodium complex ( Figure 1, Tables S1-S4). Rh-tpy crystallizes in the P1 space group and displays a distorted octahedral geometry around the rhodium atom. The three nitrogen atoms of the terpyridine and one chloride ligand are in the equatorial plane, while the nitrogen atom of CH 3 CN and one chloride ligand are in an axial position. The terpyridine ligand is quasi-planar, with low torsion angles between the central pyridine of the terpyridine and the other two pyridines (3.90 • between pyridines with N(1) and N(2) and 1.70 • between pyridines with N(2) and N(3)). The nitrogen atoms of the terpyridine, the chloride atom, and the rhodium form a slightly distorted square plane, with N(1)-Rh-N(2) and N(2)-Rh-N(3) angles of 81.07(6) • and 81.00(6) • , respectively, which are slightly smaller than the ideal value of 90 • for a perfect square plane geometry. This distortion is induced by the terpyridine, which imposes 5-atom rings comprising the rhodium, two nitrogen atoms (N(1)/N(2) or N(2)/N(3)), and two terpyridine carbon atoms. The distance between rhodium and the central pyridine nitrogen (N central ) of the terpyridine (N(2)) is shorter (1.9386 (14) Å) than that between rhodium and the two distal nitrogen (N distal ) of the terpyridine (2.0258 (14) and 2.0417(14) Å for Rh-N(1) and Rh-N(3), respectively). The rhodium atom is almost equidistant from the two chloride atoms (2.3530(5) vs. 2.3067(4) Å, respectively, for Rh-Cl(1) and Rh-Cl(2)) and the distance between the nitrogen of acetonitrile ligand and the rhodium atom (2.0220(14) Å for Rh-N(21)) is very similar to those between the rhodium and the terpyridine nitrogen atoms (Rh-N tpy ). The distances and angles between the Rh center and the terpyridine ligand in Rh-tpy are very similar to those found in the [Rh III (tpy)(OH)(OH 2 ) 2 ](NO 3 ) 2 [38]. Slightly shorter valence angles were found in the dinuclear [Rh 2 II (tpy) 2 (CH 3 CN) 4 ](CF 3 SO 3 ) 4 complex where the two Rh centers are linked by a metal-metal bond, as a consequence of Rh III -N tpy lengths being slightly longer than those of Rh-tpy. Regarding the low-valent [Rh I (tpy)(CH 3 CN)](CF 3 SO 3 ) complex, which adopts a discrete structure of four [Rh I (tpy)(CH 3 CN)] + units stacked via tpy-tpy interaction at the solid state, this complex displays slightly shorter valence angles and Rh III -N tpy bond distances than those found in Rh-tpy [37].  [37].

Spectroelectrochemical Properties of the [Rh III (tpy)(CH3CN)Cl2](CF3SO3) Complex (Rh-tpy) in N,N-Dimethylformamide (DMF)
The redox properties of Rh-tpy were investigated in DMF because of the wider electrochemical window of this solvent compared to water, facilitating the observation of the reduced state of the rhodium complex ( Figure 2 and Table S5). By analogy with the previously reported electrochemical behavior of the rhodium bis-bipyridyl [11][12][13]36,40] and bis-terpyridyl [41] complexes in organic solvent, the irreversible redox process at Epa = −1.13 V vs. Ag/AgNO3 observed on the cyclic voltammogram of Rh-tpy is attributed to the two-electron reduction of the metal center (Rh III → Rh I ), while the poorly reversible process located at a lower potential (Epc = −1.95 V) is ascribed to the one-electron reduction of the terpyridine ligand ( Figure 2A). The irreversibility of the Rh III /Rh I system is the result of the release of the two axial ligands upon reduction to form a Rh I species with a squareplanar geometry. On the reverse scan, the irreversible oxidation peak observed at Epa = −0.70 V, followed by another less defined at ca. −0.3 V, is related to the reoxidation of the metal center (Rh I → Rh III ) along with the recoordination of two exogeneous ligands, which can be the solvent (DMF), CH3CN, and/or a chloride ion.

Spectroelectrochemical Properties of the [Rh III (tpy)(CH 3 CN)Cl 2 ](CF 3 SO 3 ) Complex (Rh-tpy) in N,N-Dimethylformamide (DMF)
The redox properties of Rh-tpy were investigated in DMF because of the wider electrochemical window of this solvent compared to water, facilitating the observation of the reduced state of the rhodium complex ( Figure 2 and Table S5). By analogy with the previously reported electrochemical behavior of the rhodium bis-bipyridyl [11][12][13]36,40] and bis-terpyridyl [41] complexes in organic solvent, the irreversible redox process at Ep a = −1.13 V vs. Ag/AgNO 3 observed on the cyclic voltammogram of Rh-tpy is attributed to the two-electron reduction of the metal center (Rh III → Rh I ), while the poorly reversible process located at a lower potential (Ep c = −1.95 V) is ascribed to the one-electron reduction of the terpyridine ligand ( Figure 2A). The irreversibility of the Rh III /Rh I system is the result of the release of the two axial ligands upon reduction to form a Rh I species with a square-planar geometry. On the reverse scan, the irreversible oxidation peak observed at Ep a = −0.70 V, followed by another less defined at ca. −0.3 V, is related to the reoxidation of the metal center (Rh I → Rh III ) along with the recoordination of two exogeneous ligands, which can be the solvent (DMF), CH 3 CN, and/or a chloride ion.
In order to further characterize and evaluate the stability of the reduced Rh I state of Rh-tpy, the initial species involved in the catalytic reduction of protons to H 2 , an electrolysis was performed at −1.25 V. In accordance with the generation of a Rh I species, the exhaustive electrolysis required the exchange of two electrons per initial Rh III complex. The formation of this species was attested by the UV-Visible absorption spectrum ( Figure 3) and cyclic voltammograms ( Figure 2B) of the resulting electrolyzed solution. The initial pale yellow Rh-tpy solution displays mainly intense absorption bands below 350 nm. After complete reduction, the deep blue solution obtained from Rh I exhibits two strong absorption bands at 279 and 327 nm, and four less intense absorption bands in the visible region at 384, 515, 607, and 665 nm ( Figure 3). Very similar absorption spectra were reported by the group of Hartl in ethanol for the square-planar rhodium(I) terpyridyl halide complexes, [Rh I (tpy)(X)] (X = Cl or Br), chemically synthesized from [Rh(X)(COD)] 2 precursors (X = Cl or Br; COD = cycloocta-1,5-diene) under inert atmosphere [42]. The similarity of the absorption spectra strongly suggests the [Rh I (tpy)Cl] structure for the electrogenerated Rh I species as a result of the release of one CH 3 CN and one chloride ligand. In order to further characterize and evaluate the stability of the reduced Rh I state of Rh-tpy, the initial species involved in the catalytic reduction of protons to H2, an electrolysis was performed at −1.25 V. In accordance with the generation of a Rh I species, the exhaustive electrolysis required the exchange of two electrons per initial Rh III complex. The formation of this species was attested by the UV-Visible absorption spectrum ( Figure 3) and cyclic voltammograms ( Figure 2B) of the resulting electrolyzed solution. The initial pale yellow Rh-tpy solution displays mainly intense absorption bands below 350 nm. After complete reduction, the deep blue solution obtained from Rh I exhibits two strong absorption bands at 279 and 327 nm, and four less intense absorption bands in the visible region at 384, 515, 607, and 665 nm ( Figure 3). Very similar absorption spectra were reported by the group of Hartl in ethanol for the square-planar rhodium(I) terpyridyl halide complexes, [Rh I (tpy)(X)] (X = Cl or Br), chemically synthesized from [Rh(X)(COD)]2 precursors (X = Cl or Br; COD = cycloocta-1,5-diene) under inert atmosphere [42]. The similarity of the absorption spectra strongly suggests the [Rh I (tpy)Cl] structure for the electrogenerated Rh I species as a result of the release of one CH3CN and one chloride ligand.
Regarding the cyclic voltammograms of the Rh I solution, the initial irreversible reduction peak (Rh III → Rh I ) has fully disappeared and only the redox process centered on the tpy ligand is now present on the cyclic voltammogram, with an intensity similar to that of the initial solution of [Rh III (tpy)(MeCN)Cl2] + (Figure 2A,B), attesting to the stability of the Rh I species under inert atmosphere. In addition, the irreversible oxidation peaks (Rh I → Rh III ) observed by scanning from −1.1 to 0 V are accompanied, on the reverse scan, by the appearance of a Rh III → Rh I reduction peak in accordance with the regeneration of Regarding the cyclic voltammograms of the Rh I solution, the initial irreversible reduction peak (Rh III → Rh I ) has fully disappeared and only the redox process centered on the tpy ligand is now present on the cyclic voltammogram, with an intensity similar to that of the initial solution of [Rh III (tpy)(MeCN)Cl 2 ] + (Figure 2A,B), attesting to the stability of the Rh I species under inert atmosphere. In addition, the irreversible oxidation peaks (Rh I → Rh III ) observed by scanning from −1.1 to 0 V are accompanied, on the reverse scan, by the appearance of a Rh III → Rh I reduction peak in accordance with the regeneration of a Rh III species. Besides, the UV-vis. spectrum and the cyclic voltammograms of the electrolyzed solution do not show any changes after several hours when the latter is stored under inert atmosphere free of oxygen (glove box), showing the excellent stability of the Rh I terpyridyl complex. By contrast, under similar experimental conditions, solutions of [Rh I (Rbpy) 2 ] + electrogenerated from rhodium(III) bis-bipyridyl complexes [Rh III (Rbpy) 2 Cl 2 ] + (R = dimethyl or ditertio-butyl) are less stable and start to decompose after 1.5 h [36]. This is likely due to the higher π-acceptor capacity of terpyridine compared to that of bipyridine, which discharges the excess of electron density of Rh I to the ligand, thus stabilizing the low oxidation states of the metal [42]. The rigid and planar geometry of the tpy ligand can also contribute to the stability of the square planar Rh I species. A back exhaustive electrolysis at E = 0 V restores a Rh III complex. However, the two-electron reduction of the metal center (Rh III → Rh I ) is now located at a slightly more negative potential (Ep a = −1.27 V vs. Ag/AgNO 3 ) compared to the initial solution of Rh-tpy, suggesting a slightly different coordination sphere for the Rh III center, which might be the substitution of the initial CH 3 CN axial ligand by DMF. Indeed, the two chloride ligands remain coordinated to the Rh III center, as judged by the absence of the characteristic electroactivity of free chlorides (irreversible oxidation peak at ca. 0.7 V vs. Ag/AgNO 3 ) in the positive potential region from Rh-tpy [36].
after 1.5 h [36]. This is likely due to the higher π-acceptor capacity of terpyridine compared to that of bipyridine, which discharges the excess of electron density of Rh I to the ligand, thus stabilizing the low oxidation states of the metal [42]. The rigid and planar geometry of the tpy ligand can also contribute to the stability of the square planar Rh I species. A back exhaustive electrolysis at E = 0 V restores a Rh III complex. However, the two-electron reduction of the metal center (Rh III → Rh I ) is now located at a slightly more negative potential (Epa = −1.27 V vs. Ag/AgNO3) compared to the initial solution of Rh-tpy, suggesting a slightly different coordination sphere for the Rh III center, which might be the substitution of the initial CH3CN axial ligand by DMF. Indeed, the two chloride ligands remain coordinated to the Rh III center, as judged by the absence of the characteristic electroactivity of free chlorides (irreversible oxidation peak at ca. 0.7 V vs. Ag/AgNO3) in the positive potential region from Rh-tpy [36]. This electrochemical study shows that the activation of the Rh-tpy complex for photo-induced H2 production is possible using Ru as photosensitizer. Indeed, Rh-tpy can be reduced either by the reduced [Ru II (bpy)(bpy •− )] + (denoted Ru − ) or by the excited states of Ru (denoted *Ru II ), since both reactions are favored from a thermodynamic point of view. Considering the reduction potential of Rh-tpy (Epc (Rh III /Rh I ) = −1.13 V vs. Ag/AgNO3 (−0.83 V vs. SCE)) in DMF, and the redox potential of Ru at reduced Ru − and excited *Ru II states in water (E1/2 (Ru II /Ru − ) = −1.50 V and E1/2 (Ru III /*Ru II ) = −1.07 V vs. SCE [3], see Table S5), the driving forces (ΔG0) are clearly exergonic with values of −0.67 and −0.24 eV, respectively. Photophysical experiments (see below) allowed for deciphering the photocatalytic mechanism occurring with the system Ru/Rh-tpy/HA − /H2A. This electrochemical study shows that the activation of the Rh-tpy complex for photoinduced H 2 production is possible using Ru as photosensitizer. Indeed, Rh-tpy can be reduced either by the reduced [Ru II (bpy)(bpy •− )] + (denoted Ru − ) or by the excited states of Ru (denoted *Ru II ), since both reactions are favored from a thermodynamic point of view. Considering the reduction potential of Rh-tpy (Ep c (Rh III /Rh I ) = −1.13 V vs. Ag/AgNO 3 (−0.83 V vs. SCE)) in DMF, and the redox potential of Ru at reduced Ru − and excited *Ru II states in water (E 1/2 (Ru II /Ru − ) = −1.50 V and E 1/2 (Ru III /*Ru II ) = −1.07 V vs. SCE [3], see Table S5), the driving forces (∆G 0 ) are clearly exergonic with values of −0.67 and −0.24 eV, respectively. Photophysical experiments (see below) allowed for deciphering the photocatalytic mechanism occurring with the system Ru/Rh-tpy/HA − /H 2 A.

Photocatalytic Hydrogen Production
The catalytic activity of Rh-tpy for the light-driven production of H 2 was evaluated in the presence of the Ru photosensitizer and ascorbate as a sacrificial electron donor in deaerated aqueous solutions (5 mL) at 298 K under visible light irradiation (400-700 nm). The light-driven evolution of H 2 was determined by gas chromatography in the course of the photocatalysis, and these data were used to calculate the TON (turnover number) and initial TOF values (turnover frequency) per molecule of catalyst (respectively denoted as TON Cat and TOF Cat ), which characterize the performance of the photocatalytic system (Table S7).
The catalytic activity of Rh-tpy was first investigated in a HA -/H 2 A buffer (total concentration 1.1 M, see Table S6) at pH 4.0 with 530 µM of PS, similar to our previous experiments with Rh2 [11,12] and cobalt catalysts [43][44][45][46][47]. In these conditions, the Ru/Rh-tpy/HA -/H 2 A system displays 220 TON Cat after 22 h of irradiation with a catalyst concentration of 10 µM (Figure 4). However, the stability of the photocatalytic system did not exceed 7-8 h and about 90% of the total amount of H 2 was achieved after 6 h. The UV-vis. absorption spectrum of the photocatalytic solution recorded at the end of the photocatalysis indicates the complete transformation of the Ru PS into the bis-bipyridyl Ru(II) derivative, [Ru(bpy) 2 (X) 2 ] n+ , with the disappearance of the initial absorption band at 450 nm of Ru and the appearance of a new shoulder at 473 nm with a lower intensity ( Figure 5A). This is due to the well-known poor stability of the reduced state Ru − in acidic water, generated upon the reductive quenching of *Ru by HA − (see below), which easily undergoes a ligand substitution by solvent molecules and/or anions such as ascorbate [48,49]. In view of slowing down the degradation of Ru and thus improving the stability of the photocatalytic system, a lower HA − /H 2 A concentration of 0.1 M was used. To keep the pH constant during the photocatalytic experiment, the aqueous solution was buffered with an acetate buffer (1 M) at pH 4.0. Such conditions have been already employed for several photocatalytic systems for H 2 production involving cobalt catalysts [48,[50][51][52][53]. In acetate buffer containing 0.1 M of HA − /H 2 A at pH 4.0, the Ru/Rh-tpy photocatalytic system is clearly more stable, since the H 2 production is still effective after 22 h of irradiation. At this stage, the TON Cat value has reached 380, which is almost twice as high as the TON Cat value obtained in 1.1 M HA − /H 2 A buffer, and the Ru PS is less degraded at the end of photocatalysis ( Figure 5B). However, when the concentration of the sacrificial electron donor is divided by 10, the initial TOF Cat value is lower than that obtained in 1.1 M HA − /H 2 A (38 vs. 95 TOF Cat , respectively, see Table S7). The pH was further optimized using the acetate buffer with 0.1 M HA − /H 2 A medium. The highest TON Cat value of 423 was obtained at pH 4.5 after 22 h of irradiation. A further increase of the pH to 4.8 leads to a decrease of the TON Cat value ( Figure 4). Basically, the pH at which photocatalysis is optimal is governed by a delicate balance between the reactivity of the catalyst (generation of the Rh III hydride species) and the efficiency of the photoinduced electron transfer process (concentration of both protonated and non-protonated form of the SD) [34,54]. More acidic conditions make the protonation of the catalyst, i.e., the H 2 -evolution catalysis, faster, but the generation of the reductant Ru − is slowed down because the concentration of the ascorbate electron donor is smaller. In other words, in less acidic conditions, the higher concentration of ascorbate makes the quenching or regeneration of PS more efficient, but the low concentration in protons decreases the H 2 -evolving activity of the catalyst.     In order to rule out the formation of Rh 0 colloids from Rh-tpy during the photocatalysis, an experiment was carried out with 530 µM of Ru, 10 µM of Rh-tpy in acetate buffer (0.1 M) containing HA − /H 2 A (0.1 M) at pH 4.5, in the presence of an excess of mercury ( Figure 6). Rh 0 colloids are known to catalyze the reduction of protons to H 2 and can be deactivated by generating an amalgam with mercury [55,56]. The absence of effect on the photocatalytic behavior of the Ru/Rh-tpy/HA − /H 2 A system due to the Hg addition confirms that no colloidal rhodium is formed during photocatalysis and the molecular Rh-tpy catalyst is involved in the H 2 evolution. deactivated by generating an amalgam with mercury [55,56]. The absence of effect on t photocatalytic behavior of the Ru/Rh-tpy/HA − /H2A system due to the Hg addition co firms that no colloidal rhodium is formed during photocatalysis and the molecular R tpy catalyst is involved in the H2 evolution. The photocatalytic performances of the Ru/Rh-tpy/HA − /H2A system were also inve tigated at lower concentrations of Rh-tpy in acetate buffer at pH 4.5 by keeping the co centration of Ru constant in view to promote the reduction of the catalyst (Figure 7). A and 1 µM, the TONCat values increase drastically to reach 804 and 2242, respectively, aft 22-25 h of irradiation. This strong increase of TONCat values when the PS/Cat ratio i creases has been previously observed in many photocatalytic molecular systems usin rhodium [11,15] or cobalt catalysts [45][46][47][48][49]53,[57][58][59][60][61][62][63]. However, these values do not ta into account the amount of H2 that comes only from the Ru photosensitizer in the absen of a catalyst. Indeed, while control experiments in the absence of Ru or HA − do not pr duce any H2, some H2 is generated from solutions containing only Ru and HA − /H2A (Tab S7, Figure 7B). This amount is not negligible, as it represents about 30% of the H2 produc for the 10 and 5 µM concentrations and more than half (55%) for the lowest concentratio of 1 µM. Consequently, the actual TONCat values (denoted as TONCat*), calculated by su tracting the amount of H2 stemming from Ru, are thus of 300, 558, and 1012 at 10, 5, an 1 µM, respectively (Table S7). The photocatalytic performances of the Ru/Rh-tpy/HA − /H 2 A system were also investigated at lower concentrations of Rh-tpy in acetate buffer at pH 4.5 by keeping the concentration of Ru constant in view to promote the reduction of the catalyst (Figure 7). At 5 and 1 µM, the TON Cat values increase drastically to reach 804 and 2242, respectively, after 22-25 h of irradiation. This strong increase of TON Cat values when the PS/Cat ratio increases has been previously observed in many photocatalytic molecular systems using rhodium [11,15] or cobalt catalysts [45][46][47][48][49]53,[57][58][59][60][61][62][63]. However, these values do not take into account the amount of H 2 that comes only from the Ru photosensitizer in the absence of a catalyst. Indeed, while control experiments in the absence of Ru or HA − do not produce any H 2 , some H 2 is generated from solutions containing only Ru and HA − /H 2 A (Table S7, Figure 7B). This amount is not negligible, as it represents about 30% of the H 2 produced for the 10 and 5 µM concentrations and more than half (55%) for the lowest concentration of 1 µM. Consequently, the actual TON Cat values (denoted as TON Cat *), calculated by subtracting the amount of H 2 stemming from Ru, are thus of 300, 558, and 1012 at 10, 5, and 1 µM, respectively (Table S7).
UV-vis absorption spectra recorded at the end of the photocatalysis show that the Ru PS is partially transformed in Ru-bis-bipyridine species and that this transformation is more pronounced as the concentration of the catalyst is low (Figure 8) [53]. The degradation of the PS is most probably the main limiting factor for the H 2 production after 22 h [54].
The photocatalytic performance of Ru/Rh-tpy/HA − /H 2 A has also been compared with that obtained with Rh2 as catalyst in the same experimental conditions (Table S7). In combination with a molecular photosensitizer, the Rh2 complex is among the most efficient H 2 evolving catalysts based on rhodium that operate in water [11,12,15,34]. At pH 4.5 in 1 M acetate buffer containing 0.1 M of HA − /H 2 A, the Rh2 catalyst appears significantly more active than Rh-tpy for the higher catalyst concentration of 10 µM, with corrected TON Cat * of 1140 vs. 300 for Rh-tpy. However, in contrast to what was observed for Rh-tpy and, in a previous reported study for Rh2 in the 1.1 M HA − /H 2 A buffer [11], if the catalyst concentration is decreased, the TON Cat * also decreases to about 780 at 5 µM and 770 at 1 µM ( Figure 7). Thus, Rh-tpy is clearly less efficient than Rh2 at the higher catalyst concentrations of 10 and 5 µM (TON Cat * of 300 and 558 for Rh-tpy vs. 1140 and 778 for Rh2), but challenges Rh2 in more diluted conditions (for 1 µM, TON Cat * of 1012 for Rh-tpy vs. 772 for Rh2). In fact, the higher stability of the Rh I species for Rh-tpy compared to that of Rh2, as revealed in organic solvent, could explain the photocatalytic performance of Ru/Rh-tpy/HA − /H 2 A competing with that of Ru/Rh2/HA − /H 2 A at very low concentrations of catalyst. Therefore, the use of tridentate tpy ligand compared to bidentate bpy ligands leads to a less active but potentially more stable H 2 -evolving catalyst. However, the instability of Ru PS does not allow for comparing the long-term stability of Rh-tpy and Rh2 catalysts. A more stable PS in water such as the triazatriangulenium TATA + organic dye should be employed, but the reduction potential of this PS, which is less negative than that of Ru [53], will not allow for an efficient reduction of the Rh catalysts.  The photocatalytic performance of Ru/Rh-tpy/HA − /H2A has also b with that obtained with Rh2 as catalyst in the same experimental condition combination with a molecular photosensitizer, the Rh2 complex is among cient H2 evolving catalysts based on rhodium that operate in water [11,12 4.5 in 1 M acetate buffer containing 0.1 M of HA − /H2A, the Rh2 catalyst a cantly more active than Rh-tpy for the higher catalyst concentration of 10 rected TONCat* of 1140 vs. 300 for Rh-tpy. However, in contrast to what wa Rh-tpy and, in a previous reported study for Rh2 in the 1.1 M HA − /H2A bu catalyst concentration is decreased, the TONCat* also decreases to about 78 770 at 1 µM ( Figure 7). Thus, Rh-tpy is clearly less efficient than Rh2 at the concentrations of 10 and 5 µM (TONCat* of 300 and 558 for Rh-tpy vs. 11 Rh2), but challenges Rh2 in more diluted conditions (for 1 µM, TONCat* o tpy vs. 772 for Rh2). In fact, the higher stability of the Rh I species for Rh-tp that of Rh2, as revealed in organic solvent, could explain the photocatalyt of Ru/Rh-tpy/HA − /H2A competing with that of Ru/Rh2/HA − /H2A at very tions of catalyst. Therefore, the use of tridentate tpy ligand compared to ligands leads to a less active but potentially more stable H2-evolving cata the instability of Ru PS does not allow for comparing the long-term stab and Rh2 catalysts. A more stable PS in water such as the triazatrianguleni ganic dye should be employed, but the reduction potential of this PS, whi tive than that of Ru [53], will not allow for an efficient reduction of the Rh We also summarized the performances of the various three-compone lytic systems for H2 production in homogeneous solution in Tables S8 and rhodium complex (Rh1-24, Scheme 1) as a H2-evolving catalyst previous We also summarized the performances of the various three-components photocatalytic systems for H 2 production in homogeneous solution in Tables S8 and S9, involving a rhodium complex (Rh1-24, Scheme 1) as a H 2 -evolving catalyst previously reported in view to compare the efficiency of these systems with those of Rh2 and Rh-tpy as catalysts. Tables S8 and S9 summarize the performances in hydro-organic (Rh11-24, Scheme 1b) and in purely aqueous (Rh1-10, Scheme 1a) solution, respectively. The structures of the various photosensitizers employed in these photocatalytic systems are gathered in Scheme S1. In aqueous organic media, PSs mainly rely on heteroleptic cyclometaled iridium(III) complexes of the type [Ir(CˆN) 2  Among the Rh catalysts employed in water (Rh1-10, Scheme 1a), only Rh2, Rh3, and Rh4 exhibited turnover numbers per catalyst (TON Cat ) higher than 100 in the presence of a ruthenium or iridium photosensitizer and ascorbate as SD [11]; TON Cat for the others catalysts, Rh5-10, do not exceed 10 [3,8] (Table S9). However, these later systems used significantly higher catalyst concentration ranging from 50 µM to 1.56 mM, and the catalyst concentration is generally higher than that of the PS, resulting in lower TON Cat . Concerning the Wilkinson catalyst Rh4, although up to 2000 TONs Cat were reported [6,7], further studies by our group have shown that this catalyst is much less efficient than Rh2 [11].
These examples show that even with quite similar families of catalysts (for instance, Rh5-9 [8] and Rh5, Rh15-16 [28,29]) and PS, the TON Cat values can be very different depending on the experimental conditions. The most relevant comparisons are therefore those made under similar experimental conditions. We can thus compare the efficiency of our systems with rhodium with those of the cobalt(III) tetraazamacrocyclic [Co III (CR14)Cl 2 ] + (Co) complex, which is one of the most efficient H 2 -evolving catalyst in acidic water [43,45,47,53]. The catalytic performances of this complex have been recorded under similar experimental conditions with 500 µM Ru and 0.1 M of HA − /H 2 A in 1 M acetate buffer at pH 4.5 for catalyst concentrations of 10 and 5 µM (Table S7) [53]. TON Cat * of 1086 and 1822 were obtained for Co compared to 300 and 558 for Rh-tpy for 10 and 5 µM, respectively. At both catalysts' concentrations, Rh-tpy is thus much less efficient than Co, with more than three times less hydrogen produced.

Mechanistic Insight for the Ru/Rh-tpy/HA − /H 2 A System from Photophysical Measurements
The light-driven H 2 production with the Ru/Rh-tpy/HA − /H 2 A three-component system is initiated by the generation of the excited state of the Ru, *Ru II , under light absorption (Scheme 4). *Ru could be then quenched by an electron transfer to the sacrificial electron donor (HA − ) to generate the reduced form of PS, Ru − , and the oxidized form of ascorbate (HA • ) (reductive quenching). Ru − is able, in turn, to reduce the catalyst (Rh III -tpy to Rh I -tpy), reforming the ground state Ru. HA • can lose a proton to form A •-, which can disproportionate generating the dehydroascorbic acid (DHA) [34,64,65], a very good electron acceptor able to withdraw electrons from Ru − and/or Rh I -tpy. Therefore, a back electron transfer takes place from Ru − to HA • or DHA (BET process) to restore Ru II and HA -, or from Rh I -tpy to HA • or DHA (BETC process) to regenerate Rh III -tpy.
Molecules 2022, 27, x FOR PEER REVIEW 14 of 21 Scheme 4. Proposed catalytic mechanism for the light-driven H2 production with the system Ru/Rhtpy/HA − /H2A. For reasons of simplicity, only the pathway showing the heterolytic mechanism from Rh III (H)-tpy is shown. *Ru II represents the triplet excited state of Ru II , and Ru − , its one-electron reduced state.
Photophysical measurements allow us to identify which one of these two mechanisms is the most favorable. By using a Stern-Volmer plot, a rate constant (kQ1) of 1.0 × 10 7 M −1 s −1 was determined by the Schmehl's group for the reductive quenching of the *Ru II luminescence by HA − in an aqueous acetate buffered at pH 4.5. In the same medium, we determined a rate constant of 6.87 × 10 8 M −1 s −1 for the oxidative quenching (kQ2) of *Ru by Rh-tpy (Figure 9), which is about 70 times higher than that of the reductive quenching of *Ru by HA -. Although the oxidative quenching is kinetically more favored than the reductive quenching, the reductive pathway dominates over the oxidative one with pseudofirst-order kinetics of 1.0 × 10 6 s −1 and 0.069-3.4 × 10 4 s −1 , respectively, considering that the HA − concentration (0.1 M) is much higher than that of Rh-tpy (1-10 µM) under photocatalytic conditions. Noteworthily, the rate constant of the oxidative quenching of *Ru by Rh III -tpy is very similar to that previously determined by our group between *Ru and Rh2 The *Ru can also be quenched by the catalyst leading to the oxidized form of the PS, Ru III , and the reduced state of catalyst, Rh I -tpy (oxidative quenching), this process being unfavorable (see below).
Photophysical measurements allow us to identify which one of these two mechanisms is the most favorable. By using a Stern-Volmer plot, a rate constant (k Q1 ) of 1.0 × 10 7 M −1 s −1 was determined by the Schmehl's group for the reductive quenching of the *Ru II luminescence by HA − in an aqueous acetate buffered at pH 4.5. In the same medium, we determined a rate constant of 6.87 × 10 8 M −1 s −1 for the oxidative quenching (k Q2 ) of *Ru by Rh-tpy (Figure 9), which is about 70 times higher than that of the reductive quenching of *Ru by HA -. Although the oxidative quenching is kinetically more favored than the reductive quenching, the reductive pathway dominates over the oxidative one with pseudo-first-order kinetics of 1.0 × 10 6 s −1 and 0.069-3.4 × 10 4 s −1 , respectively, considering that the HA − concentration (0.1 M) is much higher than that of Rh-tpy (1-10 µM) under photocatalytic conditions. Noteworthily, the rate constant of the oxidative quenching of *Ru by Rh III -tpy is very similar to that previously determined by our group between *Ru and Rh2 (3.2 × 10 8 M −1 s −1 ) [11]. This similarity of rate constants could be correlated to their akin driving forces ∆G 0 of −0.24 eV for *Ru/Rh-tpy and −0.28 eV for *Ru/Rh2, considering the reduction potentials (Rh III /Rh I ) of Rh-tpy and Rh2, and the oxidation potential of *Ru (see Table S5). Nanosecond flash photolysis experiments have also been performed to characterize the photoinduced electron transfer process occurring in the system Ru/Rh-tpy/HA − /H2A at pH 4.5. In the absence of Rh-tpy, the transient absorption spectra recorded after excitation at 455 nm ( Figure S1) show the formation of the Ru − species with positive absorption bands at 360 and 510 nm [66]. The formation of Ru − occurs from an electron transfer between *Ru and HA - [9,11]. Ru − growth follows a pseudo first-order kinetics with an estimated rate constant of 4.3 × 10 6 s −1 (τ = 235 ns, Figure S2). The decay of the transient absorption trace at 510 nm, due to the back electron transfer (BET) between Ru − and the oxidized forms of ascorbate (HA • and DHA), can be fitted according to a second order kinetics law ( Figure S3). The rate constant (kBET) is estimated to be 7.4 × 10 9 M −1 s −1 .
In the presence of Rh-tpy, flash photolysis experiments also show the formation of the Ru − species, evidenced by an increase of absorption at 360 and 510 nm ( Figure 10). The growth of the signal occurs in a similar time scale (τ = 215 ns, corresponding to a constant at 4.6 × 10 6 s −1 , Figure S4) to that observed without Rh-tpy ( Figure S2). This confirms that the photocatalytic cycle is initiated by a reductive quenching of *Ru by HA − and that the presence of Rh-tpy does not interfere with this first electron transfer process. However, the decay of the transient absorption trace at 510 nm is accelerated in the presence of Rhtpy ( Figure 11). This is a consequence of an electron transfer process between the transient Ru − and Rh III -tpy leading to the initial Ru II and Rh I -tpy, which efficiently competes with the BET process between Ru − and the oxidized ascorbate (e.g., HA • and DHA). Nanosecond flash photolysis experiments have also been performed to characterize the photoinduced electron transfer process occurring in the system Ru/Rh-tpy/HA − /H 2 A at pH 4.5. In the absence of Rh-tpy, the transient absorption spectra recorded after excitation at 455 nm ( Figure S1) show the formation of the Ru − species with positive absorption bands at 360 and 510 nm [66]. The formation of Ru − occurs from an electron transfer between *Ru and HA - [9,11]. Ru − growth follows a pseudo first-order kinetics with an estimated rate constant of 4.3 × 10 6 s −1 (τ = 235 ns, Figure S2). The decay of the transient absorption trace at 510 nm, due to the back electron transfer (BET) between Ru − and the oxidized forms of ascorbate (HA • and DHA), can be fitted according to a second order kinetics law ( Figure S3). The rate constant (k BET ) is estimated to be 7.4 × 10 9 M −1 s −1 .
In the presence of Rh-tpy, flash photolysis experiments also show the formation of the Ru − species, evidenced by an increase of absorption at 360 and 510 nm ( Figure 10). The growth of the signal occurs in a similar time scale (τ = 215 ns, corresponding to a constant at 4.6 × 10 6 s −1 , Figure S4) to that observed without Rh-tpy ( Figure S2). This confirms that the photocatalytic cycle is initiated by a reductive quenching of *Ru by HA − and that the presence of Rh-tpy does not interfere with this first electron transfer process. However, the decay of the transient absorption trace at 510 nm is accelerated in the presence of Rh-tpy ( Figure 11). This is a consequence of an electron transfer process between the transient Ru − and Rh III -tpy leading to the initial Ru II and Rh I -tpy, which efficiently competes with the BET process between Ru − and the oxidized ascorbate (e.g., HA • and DHA).  In the presence of Rh-tpy, the decay of the transient Ru − species is best fitted by a pseudo-first-order kinetics law, leading to a rate constant for the ET process of 2.4 × 10 4 s −1 ( Figure S5). Considering the concentration of the catalyst in solution (200 µM), this would correspond to a bimolecular rate constant (kET) of 1.2 × 10 8 M −1 s −1 , which is 62 times lower than the kBET value (7.4 × 10 9 M −1 s −1 ). The difference between these two kinetics should contradict the faster decay of Ru − in presence of Rh III -tpy observed in the transient absorption trace in Figure 11. Nevertheless, we have estimated the initial concentration of Ru − (2.27 × 10 −7 M), calculated from the concentration of *Ru II (9.1 × 10 −7 M) just after the excitation pulse (ΔA@450 nm = −0.01 and Δε = −11,000 M −1 cm −1 ) and considering a quantum yield of 25% for the Ru − formation from reaction between *Ru and HA - [45]. This has allowed us to determine the real rates of the ET and BET processes (vET and vBET) at 5.45 × 10 −3 and 3.81 × 10 −4 M s −1 , respectively (see ESI for the calculation details). In other words, under our experimental conditions, the ET process dominates over the BET process with  In the presence of Rh-tpy, the decay of the transient Ru − species is best fitted by a pseudo-first-order kinetics law, leading to a rate constant for the ET process of 2.4 × 10 4 s −1 ( Figure S5). Considering the concentration of the catalyst in solution (200 µM), this would correspond to a bimolecular rate constant (kET) of 1.2 × 10 8 M −1 s −1 , which is 62 times lower than the kBET value (7.4 × 10 9 M −1 s −1 ). The difference between these two kinetics should contradict the faster decay of Ru − in presence of Rh III -tpy observed in the transient absorption trace in Figure 11. Nevertheless, we have estimated the initial concentration of Ru − (2.27 × 10 −7 M), calculated from the concentration of *Ru II (9.1 × 10 −7 M) just after the excitation pulse (ΔA@450 nm = −0.01 and Δε = −11,000 M −1 cm −1 ) and considering a quantum yield of 25% for the Ru − formation from reaction between *Ru and HA - [45]. This has allowed us to determine the real rates of the ET and BET processes (vET and vBET) at 5.45 × 10 −3 and 3.81 × 10 −4 M s −1 , respectively (see ESI for the calculation details). In other words, under our experimental conditions, the ET process dominates over the BET process with In the presence of Rh-tpy, the decay of the transient Ru − species is best fitted by a pseudo-first-order kinetics law, leading to a rate constant for the ET process of 2.4 × 10 4 s −1 ( Figure S5). Considering the concentration of the catalyst in solution (200 µM), this would correspond to a bimolecular rate constant (k ET ) of 1.2 × 10 8 M −1 s −1 , which is 62 times lower than the k BET value (7.4 × 10 9 M −1 s −1 ). The difference between these two kinetics should contradict the faster decay of Ru − in presence of Rh III -tpy observed in the transient absorption trace in Figure 11. Nevertheless, we have estimated the initial concentration of Ru − (2.27 × 10 −7 M), calculated from the concentration of *Ru II (9.1 × 10 −7 M) just after the excitation pulse (∆A@450 nm = −0.01 and ∆ε = −11,000 M −1 cm −1 ) and considering a quantum yield of 25% for the Ru − formation from reaction between *Ru and HA - [45]. This has allowed us to determine the real rates of the ET and BET processes (v ET and v BET ) at 5.45 × 10 −3 and 3.81 × 10 −4 M s −1 , respectively (see ESI for the calculation details).
In other words, under our experimental conditions, the ET process dominates over the BET process with a v ET /v BET ratio of 14.3. Finally, although the two-electron reduced catalyst (Rh I -tpy) exhibits a large absorption band between 550 and 700 nm (Figure 3), its typical spectroscopic signature is not observed in the transient absorption spectra. This is attributed to the fast reactivity of the Rh I -tpy species (i.e., [Rh I (tpy)X] n+ , X = Cl − or H 2 O with n = 0 or 1, respectively) with protons, below the nanosecond time-scale, to form a Rh III (H)-tpy hydride species (i.e., [Rh III (H)(tpy)(H 2 O)X] n+ , X = Cl − or H 2 O with n = 1 or 2, respectively), the key intermediate to reduce protons into H 2 . Indeed, from this species, different pathways can lead to the production of H 2 via homolytic (reaction with another Rh III (H)-tpy species to generate H 2 and two Rh(II) species) or heterolytic (reaction with a proton to form H 2 and a Rh(III) species) route (Scheme 4). Rh III (H)-tpy could be also further reduced by Ru − to a Rh(II) hydride species, Rh II (H)-tpy, from which H 2 can be released via similar homolytic and heterolytic pathways, generating Rh(I) and Rh(II) species, respectively. The group of Ogo has shown that in CH 3 CN, the hydride complex [Rh III (H)(tpy)(CH 3 CN) 2 ] 2+ can slowly generate H 2 via reductive elimination leading to the Rh(II) dimer species (homolytic route) [37]. However, we cannot rule out that in acidic water, other pathways can proceed in parallel to this homolytic mechanism. For instance, according to theoretical calculations, we have shown that, for the Rh2 catalyst in acidic water, H 2 is preferentially released through a heterolytic mechanism from the Rh III (H) species and that both homo-and heterolytic mechanisms are thermodynamically favorable to generate H 2 via the Rh II (H) species [13]. Furthermore, although the formation of the Rh III (H)-tpy hydride species is very fast, the catalysis of proton reduction generally remains the rate-determining step of the photocatalytic mechanism, occurring within threecomponent systems for photocatalytic evolution of H 2 in homogeneous media [44,54]. 300 and up to 1000 were obtained for H 2 production for a Rh-tpy catalyst concentration at 10 and 1 µM, respectively, after subtraction of the amount of H 2 stemming from the Ru only. The photocatalytic performance of Ru/Rh-tpy/HA − /H 2 A has also been compared with that obtained with Rh2 as a catalyst in the same experimental conditions. It appears that Rh-tpy is clearly less efficient than Rh2 at the higher catalyst concentrations of 10 and 5 µM (TON Cat * of 300 and 558 for Rh-tpy vs 1140 and 778 for Rh2), but challenges Rh2 in more diluted conditions (for 1 µM, TON Cat * of 1012 for Rh-tpy vs. 772 for Rh2). Therefore, the use of tridentate tpy ligand compared to bidentate bpy ligands leads to a less active but potentially more stable H 2 -evolving catalyst. However, Rh-tpy, is much less efficient than the cobalt(III) tetraazamacrocyclic [Co III (CR14)Cl 2 ] + (Co) complex, ione of the most efficient H 2 -evolving catalysts in acidic water, since with this catalyst, TON Cat * of 1086 and 1822 were reached at 10 and 5 µM, respectively.
The electrochemical study in DMF reveals that the reduced state of the rhodium catalyst, identified as [Rh I (tpy)Cl] (Rh I -tpy) from its UV-visible signature, is stable for several hours under an inert atmosphere owing to the π-acceptor property of the terpyridine ligand that stabilizes the low oxidation states of the rhodium. A good stability of the lowvalent form of the rhodium catalyst makes it less prone to degradation in the course of photocatalysis. The π-acceptor property of terpyridine also confers moderately negative reduction potential to the Rh-tpy catalyst (about −0.8 V vs. SCE), making its reduction by the reduced state of Ru effective (E 1/2 (Ru II /Ru I ) = −1.50 V vs. SCE). A Stern-Volmer plot and transient absorption spectroscopy have shown that the first step of the photocatalytic process is a reductive quenching of the Ru excited state by ascorbate. The resulting Ru − species is then able to reduce the Rh III -tpy catalyst generating Rh I -tpy, which reacts with a proton on a sub-nanosecond time scale to form a Rh III (H)-tpy hydride, the key intermediate for H 2 evolution. The search for new molecular catalysts for H 2 production with simple and easily synthesized ligands is still ongoing, and the terpyridine ligand, with its particular electronic and coordination properties, is a good candidate for designing new catalysts to meet these requirements.