Towards Covalent Photosensitizer-Polyoxometalate Dyads-Bipyridyl-Functionalized Polyoxometalates and Their Transition Metal Complexes

A triol-functionalized 2,2′-bipyridine (bpy) derivative has been synthesized and used for the tris-alkoxylation of polyoxometalate (POM) precursors. The resultant POM-bpy conjugates of the Wells–Dawson- and Anderson-type feature a C–C bond as a linkage between the POM and bpy fragments. This structural motif is expected to increase the hydrolytic stability of the compounds. This is of particular relevance with respect to the application of POM-bpy metal complexes, as photocatalysts, in the hydrogen-evolution reaction (HER) in an aqueous environment. Accordingly, Rh(III) and Ir(III) complexes of the POM-bpy ligands have been prepared and characterized. These catalyst-photosensitizer dyads have been analyzed with respect to their electrochemical and photophysical properties. Cyclic and square-wave voltammetry, as well as UV/vis absorption and emission spectroscopy, indicated a negligible electronic interaction of the POM and metal-complex subunits in the ground state. However, emission–quenching experiments suggested an efficient intramolecular electron-transfer process from the photo-excited metal centers to the POM units to account for the non-emissive nature of the dyads (thus, suggesting a strong interaction of the subunits in the excited state). In-depth photophysical investigations, as well as a functional characterization, i.e., the applicability in the HER reaction, are currently ongoing.


Introduction
The molecular metal oxides, the so-called polyoxometalates (POMs), represent an important class of materials with high relevance for a range of potential applications. Their chemical and electronic properties can be modulated by their chemical composition, and, for example, the pronounced Brønsted acidity can be employed to catalyze a range of organic transformations (e.g., esterification, hydrolysis, etc.) [1]. Moreover, their ability to undergo redox processes in a fast and highly reversible manner enables their usage as a catalyst for the oxidation of inter alia alcohols and sulfides, alkenes or arenes. In addition, the reduction of organic substrates, in particular, the photoreduction of CO 2 , has attracted considerable interest [2]. Even more importantly, POMs have found to be active in

Synthesis and Characterization of the Organo-Functionalized POMs
To prepare the functionalized polyoxometalates 3-6 via the tris-alkoxylation method, the corresponding triols 1 and 2 had to be synthesized first. Following a literature procedure, 1 was obtained from phenylacetaldehyde in a Tollens-type condensation in a 37% yield [30,31]. The two-step synthesis of bpy-triol 2 is outlined in Scheme 1. The Kröhnke-type condensation of crotonaldehyde afforded 4-methyl-2,2 -bipyridine in a 61% yield [32]. This yield is much higher than reported in Constable's original protocol and is a result of the optimized purification procedure: The crude methanolic reaction mixture was extracted continuously with n-hexane for several days (the pure product was subsequently isolated from the hexane phase). The base-catalyzed hydroxymethylation of 4-methyl-bpy, inspired by Menozzi's hydroxymethylation of 4-picoline [33], was performed under highly forcing conditions (microwave irradiation, 130 • C, 10 h). After acidic hydrolysis of the intermediate hemi-and full acetals, 2 was isolated in moderate yield (37%).

Synthesis and Characterization of the Organo-Functionalized POMs
To prepare the functionalized polyoxometalates 3-6 via the tris-alkoxylation method, the corresponding triols 1 and 2 had to be synthesized first. Following a literature procedure, 1 was obtained from phenylacetaldehyde in a Tollens-type condensation in a 37% yield [30,31]. The twostep synthesis of bpy-triol 2 is outlined in Scheme 1. The Kröhnke-type condensation of crotonaldehyde afforded 4-methyl-2,2′-bipyridine in a 61% yield [32]. This yield is much higher than reported in Constable's original protocol and is a result of the optimized purification procedure: The crude methanolic reaction mixture was extracted continuously with n-hexane for several days (the pure product was subsequently isolated from the hexane phase). The base-catalyzed hydroxymethylation of 4-methyl-bpy, inspired by Menozzi's hydroxymethylation of 4-picoline [33], was performed under highly forcing conditions (microwave irradiation, 130 °C, 10 h). After acidic hydrolysis of the intermediate hemi-and full acetals, 2 was isolated in moderate yield (37%). Scheme 1. Schematic representation of the two-step synthesis of 2,2′-bipyridine (bpy)-triol 2.
As outlined in Scheme 2, the triols 1 and 2 were reacted with the corresponding POM precursor materials [34,35] [36,37]. The hybrids were isolated in satisfying yields as their TBA salts with reasonable solubility in polar-aprotic solvents (e.g., DMSO, DMF, acetonitrile). All the functionalized POMs were characterized by NMR spectroscopy and mass spectrometry (all spectra are depicted in the Supplementary Material). In brief, the Dawson-type POMs 3 and 4 showed the expected ratio of integrals between the signals arising from the pending phenyl or bipyridyl moiety, the CH2 groups, and the accompanying counter cations in the 1 H NMR spectrum. Moreover, the characteristic signals in the 31 P NMR spectrum for the two different P atoms within the {V3}-capped cluster framework could also be observed (δ ≈ −7.9 ppm and −13.0 ppm) [36]. Due to the presence of the paramagnetic Mn(III) center in the Andersontype systems 5 and 6, the signals in the aromatic region of the 1 H NMR spectra appeared broadened but still could be related to those of the TBA cations. As a distinct feature, the signals of the CH2 groups experienced a pronounced paramagnetic downfield shift and could be identified as a broad unstructured signal at ca. 64 ppm [37]. The paramagnetic nature of the Mn(III) center also prevented the probing of 5 and 6 by 95 Mo NMR spectroscopy-a technique commonly used to analyze Mocontaining POMs (note, that the characteristic signal at 18.0 ppm was observed for the α-[Mo8O26] 4precursor when applying this method, Figure SM As outlined in Scheme 2, the triols 1 and 2 were reacted with the corresponding POM precursor materials [34,35], i.e., (TBA) 5 O 26 ] in the presence of Mn(OAc) 3 (TBA: tetra(n-butyl)ammonium) for the Dawson-(3 and 4) and Anderson-type materials (5 and 6), under the established conditions of POM alkoxylation in N,N-dimethylacetamide (DMAc) [36,37]. The hybrids were isolated in satisfying yields as their TBA salts with reasonable solubility in polar-aprotic solvents (e.g., DMSO, DMF, acetonitrile). All the functionalized POMs were characterized by NMR spectroscopy and mass spectrometry (all spectra are depicted in the Supplementary Material). In brief, the Dawson-type POMs 3 and 4 showed the expected ratio of integrals between the signals arising from the pending phenyl or bipyridyl moiety, the CH 2 groups, and the accompanying counter cations in the 1 H NMR spectrum. Moreover, the characteristic signals in the 31 P NMR spectrum for the two different P atoms within the {V 3 }-capped cluster framework could also be observed (δ ≈ −7.9 ppm and −13.0 ppm) [36]. Due to the presence of the paramagnetic Mn(III) center in the Anderson-type systems 5 and 6, the signals in the aromatic region of the 1 H NMR spectra appeared broadened but still could be related to those of the TBA cations. As a distinct feature, the signals of the CH 2 groups experienced a pronounced paramagnetic downfield shift and could be identified as a broad unstructured signal at ca. 64 ppm [37]. The paramagnetic nature of the Mn(III) center also prevented the probing of 5 and 6 by 95 Mo NMR spectroscopy-a technique commonly used to analyze Mo-containing POMs (note, that the characteristic signal at 18.0 ppm was observed for the α-[Mo 8 O 26 ] 4precursor when applying this method, Figure SM  In the case of the Anderson-type POMs 5 and 6, single crystals suited for X-ray structure analysis could be obtained by diffusion of diethyl ether into the DMAc solutions. The crystal structure parameters and the refinement details are given below (see the Experimental Section). The phenylbearing POM derivative 5 crystallized in a monoclinic space group C2/c with Z = 4, whereas the bpybearing derivative 6 yielded an orthorhombic space group Pca21 with Z = 4. As depicted in Figure 1, the typical structure for symmetrically disubstituted Anderson POMs was found (i.e., the so-called δ-form with almost D3 symmetry) [39]. The Mo-O bond distances, as well as the O-Mo-O bond angles, were in the normal dimensions for such clusters [37]. The central Mn(III) ion was surrounded by six O atoms carrying the two tripodal organic ligands, the Mn-O bond lengths were in the range of 1.926(8) to 2.039(8) Å; whereas the C-O bond lengths were in the range from 1.436(5) to 1.442(6) Å. Due to the more rigid-linear structure arising from the direct connection of the phenyl rings to the tripodal anchor groups, the molecular packing of 5 ( Figure 2) is significantly different to that reported for a related POM hybrid where two phenyl moieties have been attached to a bis-trisfunctionalized Anderson POM via imine linkages [40]. In the present case, the interplay of different supramolecular interactions results in an overall layer-like arrangement. In detail, chains of 5 are formed in which two TBA cations always separate the polyanions from each other. Thereby, the TBA cations interact with the POM cores via C-H⋅⋅⋅O hydrogen bonding [40]. These intermolecular interactions are mainly formed between the terminal oxo-moieties of the {MnMo6} core and various Scheme 2. Schematic representation of the synthesis of the phenyl and bpy-equipped POMs 3-6. Color legend for compounds 3 and 4: V yellow, P green, dark blue W; color legend for compounds 5-7: O red, Mo cyan, Mn violet.
In the case of the Anderson-type POMs 5 and 6, single crystals suited for X-ray structure analysis could be obtained by diffusion of diethyl ether into the DMAc solutions. The crystal structure parameters and the refinement details are given below (see the Experimental Section). The phenyl-bearing POM derivative 5 crystallized in a monoclinic space group C2/c with Z = 4, whereas the bpy-bearing derivative 6 yielded an orthorhombic space group Pca2 1 with Z = 4. As depicted in Figure 1, the typical structure for symmetrically disubstituted Anderson POMs was found (i.e., the so-called δ-form with almost D 3 symmetry) [39]. The Mo-O bond distances, as well as the O-Mo-O bond angles, were in the normal dimensions for such clusters [37]. The central Mn(III) ion was surrounded by six O atoms carrying the two tripodal organic ligands, the Mn-O bond lengths were in the range of 1.926 (8)  In the case of the Anderson-type POMs 5 and 6, single crystals suited for X-ray structure analysis could be obtained by diffusion of diethyl ether into the DMAc solutions. The crystal structure parameters and the refinement details are given below (see the Experimental Section). The phenylbearing POM derivative 5 crystallized in a monoclinic space group C2/c with Z = 4, whereas the bpybearing derivative 6 yielded an orthorhombic space group Pca21 with Z = 4. As depicted in Figure 1, the typical structure for symmetrically disubstituted Anderson POMs was found (i.e., the so-called δ-form with almost D3 symmetry) [39].  Due to the more rigid-linear structure arising from the direct connection of the phenyl rings to the tripodal anchor groups, the molecular packing of 5 ( Figure 2) is significantly different to that reported for a related POM hybrid where two phenyl moieties have been attached to a bis-trisfunctionalized Anderson POM via imine linkages [40]. In the present case, the interplay of different supramolecular interactions results in an overall layer-like arrangement. In detail, chains of 5 are formed in which two TBA cations always separate the polyanions from each other. Thereby, the TBA cations interact with the POM cores via C-H⋅⋅⋅O hydrogen bonding [40]. These intermolecular interactions are mainly formed between the terminal oxo-moieties of the {MnMo6} core and various Due to the more rigid-linear structure arising from the direct connection of the phenyl rings to the tripodal anchor groups, the molecular packing of 5 ( Figure 2) is significantly different to that reported for a related POM hybrid where two phenyl moieties have been attached to a bis-tris-functionalized Anderson POM via imine linkages [40]. In the present case, the interplay of different supramolecular interactions results in an overall layer-like arrangement. In detail, chains of 5 are formed in which two TBA cations always separate the polyanions from each other. Thereby, the TBA cations interact with the POM cores via C-H···O hydrogen bonding [40]. These intermolecular interactions are mainly formed between the terminal oxo-moieties of the {MnMo 6 } core and various C-H moieties of the TBA cations (Mo-O-Mo fragments appear to be less involved in this). More important, the POM-TBA chains intercalate into each other via ππ-stacking interactions between the almost coplanar aromatic rings. Due to interference by the alkyl chains of the TBA cations, a strictly alternating order of close and more remote ππ-stacks could be observed (i.e., 4.430(4) vs. 7.885(4) Å). For 6, the packing in the crystalline state occurred in a different fashion (see the Supplementary Material). In particular, the ππ-stacking interactions between the bpy substituents was only expressed weakly. Instead, C-H···N hydrogen bonds between the organic residues could be identified. As in the previous case, a chain-like assembly motive of the POM unit was adopted with the bpy moieties tilted by almost 90 • with respect to those of the neighboring molecules. Moreover, close interchain hydrogen-bonding interactions involving the terminal oxo-moieties of the {MnMo 6 } core and the unsubstituted pyridine rings contributed to the overall packing mode. C-H moieties of the TBA cations (Mo-O-Mo fragments appear to be less involved in this). More important, the POM-TBA chains intercalate into each other via ππ-stacking interactions between the almost coplanar aromatic rings. Due to interference by the alkyl chains of the TBA cations, a strictly alternating order of close and more remote ππ-stacks could be observed (i.e., 4.430(4) vs. 7.885(4) Å). For 6, the packing in the crystalline state occurred in a different fashion (see the Supplementary Material). In particular, the ππ-stacking interactions between the bpy substituents was only expressed weakly. Instead, C-H⋅⋅⋅N hydrogen bonds between the organic residues could be identified. As in the previous case, a chain-like assembly motive of the POM unit was adopted with the bpy moieties tilted by almost 90° with respect to those of the neighboring molecules. Moreover, close interchain hydrogen-bonding interactions involving the terminal oxo-moieties of the {MnMo6} core and the unsubstituted pyridine rings contributed to the overall packing mode.

Synthesis and Characterization of the Photosensitizer-Polyoxometalate Dyads
As described previously by the groups of Hasenknopf [26] and Streb [10], the bpy-equipped POMs represent versatile platforms for the coordination of transition metal complexes, such as biscyclometalated Rh(III) and Ir(III) or Re(I) tricarbonyl fragments. In line with these studies, the coordination of appropriate precursor complexes to the POM-bpy conjugates was investigated. Accordingly, the reaction of 4 with the μ-dichloro-bridged dimeric [M(ppy)2Cl]2 precursor complexes (M = Ir or Rh) in dichloromethane/methanol mixtures (5:2 v:v) yielded the dyads 8 and 9 (Scheme 3). The complexation of the Anderson-type POM-bpy conjugate 6 had to be performed at a higher temperature and, thus, acetonitrile was used as a solvent in these cases. This protocol yielded the dinuclear assemblies 10 and 11 (Scheme 4).

Synthesis and Characterization of the Photosensitizer-Polyoxometalate Dyads
As described previously by the groups of Hasenknopf [26] and Streb [10], the bpy-equipped POMs represent versatile platforms for the coordination of transition metal complexes, such as biscyclometalated Rh(III) and Ir(III) or Re(I) tricarbonyl fragments. In line with these studies, the coordination of appropriate precursor complexes to the POM-bpy conjugates was investigated. Accordingly, the reaction of 4 with the µ-dichloro-bridged dimeric [M(ppy) 2 Cl] 2 precursor complexes (M = Ir or Rh) in dichloromethane/methanol mixtures (5:2 v:v) yielded the dyads 8 and 9 (Scheme 3). The complexation of the Anderson-type POM-bpy conjugate 6 had to be performed at a higher temperature and, thus, acetonitrile was used as a solvent in these cases. This protocol yielded the dinuclear assemblies 10 and 11 (Scheme 4). The identity and purity of all photosensitizer-POM dyads were concluded from NMR and X-ray photoelectron spectroscopy (XPS) as well as mass spectrometry (see the Supplementary Material). Representatively, the 1 H NMR and MALDI-TOF spectra of the dinuclear POM-Ir(III) dyad 11 are depicted in Figures 3 and 4, respectively. In the case of the Anderson-type PS-POM dyads 10 and 11, the characteristic broad signal at ca. 64 ppm, assigned to the CH2 groups of the trialkoxy linker, was observed. Single crystals suited for X-ray structure determination could, however, not yet be obtained in any case. The elemental composition of the PS-POM dyads was determined by XPS. The overview spectra, as summarized in Figure SM-26, revealed the expected elements of the dyads 8-11. A closer inspection of the high-resolution spectra revealed the characteristic signals of the heavy-metal atoms, e.g., the Rh 3d as well as the Ir 4f and Mo 3d signals in the dyads 8 and 11, respectively (Figures SM-27 and SM-28): The spin-orbit splitting of the d and f photoelectrons gave rise to a signal splitting. The quantitative analysis of these spectra enabled to determine the elemental composition. For example, Figure SM-28 was used to determine the Ir-to-Mo ratio of 1:(3.1 ± 0.2), which was in a very good agreement with the expected 1:3 stoichiometry. For all dyads, the relevant elemental ratios were determined in the same way; the data are summarized in Table 1. The results matched the calculated ratios and, thus, verified the chemical structure of the dyads. In short, the observed W-to-P rations of The identity and purity of all photosensitizer-POM dyads were concluded from NMR and X-ray photoelectron spectroscopy (XPS) as well as mass spectrometry (see the Supplementary Material). Representatively, the 1 H NMR and MALDI-TOF spectra of the dinuclear POM-Ir(III) dyad 11 are depicted in Figures 3 and 4, respectively. In the case of the Anderson-type PS-POM dyads 10 and 11, the characteristic broad signal at ca. 64 ppm, assigned to the CH2 groups of the trialkoxy linker, was observed. Single crystals suited for X-ray structure determination could, however, not yet be obtained in any case. The elemental composition of the PS-POM dyads was determined by XPS. The overview spectra, as summarized in Figure SM The quantitative analysis of these spectra enabled to determine the elemental composition. For example, Figure SM-28 was used to determine the Ir-to-Mo ratio of 1:(3.1 ± 0.2), which was in a very good agreement with the expected 1:3 stoichiometry. For all dyads, the relevant elemental ratios were determined in the same way; the data are summarized in Table 1. The results matched the calculated ratios and, thus, verified the chemical structure of the dyads. In short, the observed W-to-P rations of The identity and purity of all photosensitizer-POM dyads were concluded from NMR and X-ray photoelectron spectroscopy (XPS) as well as mass spectrometry (see the Supplementary Material). Representatively, the 1 H NMR and MALDI-TOF spectra of the dinuclear POM-Ir(III) dyad 11 are depicted in Figures 3 and 4, respectively. In the case of the Anderson-type PS-POM dyads 10 and 11, the characteristic broad signal at ca. 64 ppm, assigned to the CH 2 groups of the trialkoxy linker, was observed. Single crystals suited for X-ray structure determination could, however, not yet be obtained in any case. The elemental composition of the PS-POM dyads was determined by XPS. The overview spectra, as summarized in Figure SM-26, revealed the expected elements of the dyads 8-11. A closer inspection of the high-resolution spectra revealed the characteristic signals of the heavy-metal atoms, e.g., the Rh 3d as well as the Ir 4f and Mo 3d signals in the dyads 8 and 11, respectively (Figures SM-27 and SM-28): The spin-orbit splitting of the d and f photoelectrons gave rise to a signal splitting. The quantitative analysis of these spectra enabled to determine the elemental composition. For example, Figure SM-28 was used to determine the Ir-to-Mo ratio of 1:(3.1 ± 0.2), which was in a very good agreement with the expected 1:3 stoichiometry. For all dyads, the relevant elemental ratios were determined in the same way; the data are summarized in Table 1. The results matched the calculated ratios and, thus, verified the chemical structure of the dyads. In short, the observed W-to-P ratios of (5.9 ± 0.2):1 and (6.4 ± 0.2):1 in the case of 8 and 9, respectively, confirmed the integrity of the POM subunit after complexation (the theoretical W-to-P ratio is 6:1). In addition, the derived Rh-to-P and Ir-to-P ratios corroborated the successful coordination of the [(ppy) 2 M] + fragments to 4. The same also holds true for the dyads 10 and 11: The experimental Mo-to-Mn ratios of (6 ± 0.3):1 fully matched with the theoretical ones; the Rh-to-Mo and Ir-to-Mo ratios [1:(2.9 ± 0.2) and 1:(3.1 ± 0.2)] were also very close of the expected 1:3 ratio. Thus, the twofold complexation of 5 by [(ppy) 2 M] + fragments could be confirmed based on the XPS results. The summary of the analyzed XP spectra and more detailed information about the quantitative analysis are presented in the Supplementary Material (Table SM-1). very close of the expected 1:3 ratio. Thus, the twofold complexation of 5 by [(ppy)2M] + fragments could be confirmed based on the XPS results. The summary of the analyzed XP spectra and more detailed information about the quantitative analysis are presented in the Supplementary Material (Table SM-1).

Compound
Rh-to-P ratio 1 Ir-to-P ratio 1 Rh-to-Mo ratio 1 Ir-to-Mo ratio 1 8 1      To investigate the electrochemical properties of the materials possessing a Wells-Dawson-type POM, cyclic voltammetry (CV) and square-wave voltammetry (SWV) measurements were performed in dry acetonitrile (containing 0.1 M TBAF as electrolyte) to circumvent the reported adsorption phenomena on the electrode upon reduction when using DMF as a solvent. The observed redox potentials are summarized in Table 2. The Ph-POM conjugate 3 was used as reference material. Due to the absence of any redox-active organic moiety, all reduction and oxidation processes were expected to occur on the POM fragment. The CV of 3 showed three signals in the regime from 0 to −2 V ( Figure SM-29a). From these, only the signal at −1.34 V might be considered as reversible. To gain further insight into the redox properties of 3, SWV was applied additionally (Figure SM-29b). At positive voltages, three signals were identified (0.69, 0.43, and 0.09 V), which were correlated with oxidation processes involving the {V 3 } tris-vanadium cap of the POM fragment. These results agree with the data reported by Auvray et al. elsewhere for related organo-functionalized Dawson-type POMs [26]. The same also holds true for the four reduction processes which were assigned to the cluster; the first two are expected to occur on the V(V) centers. According to the available literature report, these reductions might be observed either as three individual one-electron or two separated processes, in which a one-electron reduction is followed by a two-electron reduction [41]. The latter is in line with the experimental data. The additional reduction processes involve the rest of the cluster, but no clear assignment can yet be made. Similar signal patterns have been reported in the literature. However, no statement about their exact origin has been made [26,41]. Presumably, the intense signal at −1.34 V is due to a reduction of the tungsten centers.
The electrochemical analysis of the Rh(III)-containing dyad 8 basically yielded the same results as for the model compound 3 (though the CV and SWV were poorly resolved, Figure SM-30). In addition to the oxidation of the {V 3 } cap, the irreversible signal at ca. 1.4 V was assigned to the oxidation of the Rh(III) center. At negative potentials, two prominent signals were found which were assigned to reductions of the bpy and ppy ligands of the attached complex. The multiple-electron reduction of the cluster appeared as a broad, unstructured signal at ca. −1 V. In addition, for the Ir(III)-containing counterpart 9, an irreversible metal-based oxidation at 1.0 V became apparent in the CV curve ( Figure SM-31). The SWV analysis of 9 yielded basically the same features as for 8, though the overlapping of the signals prevents any detailed discussion.

UV/vis Absorption and Emission Studies
The UV/vis absorption spectra of the compounds were recorded in CH 3 CN; the absorption bands and the corresponding molar extinction coefficients are summarized in Table 3. Both compounds 3 and 4 revealed intense absorption of UV light due to π→π* transitions in the organic unit (Figure 5a). These were accompanied by ligand-to-metal charge-transfer (LMCT) bands, which tailed from the UV regime into the high-energy visible light part of the spectrum; the bands were assigned to the characteristic O→W(VI) and O→V(V) transitions of the Wells-Dawson-type POM structure.  The [Ir(ppy)2(bpy)] + complex is brightly luminescent at room temperature; upon excitation in its MLCT band, intense emission at 592 nm can be observed [43]. However, the dyad 9 did not exhibit any luminescence. On the other hand, a stoichiometric mixture of 3 and [Ir(ppy)2(bpy)] + revealed a residual emission with approximately 25% of the intensity of the small-molecule reference complex (Figure 5b). Detailed photophysical studies on the Ir(III)-containing dyads 9 and 11 with respect to the quenching mechanism are currently ongoing. However, the results line up with the observations already made for related POM-Re dyads, for which intramolecular electron-transfer processes were proposed [26]. Auvray et al. pointed out that the luminescence quenching via such a pathway would be significantly more efficient in molecular dyads comprising a covalent linkage of the components; for the intermolecular electron transfer, as in the mixture of 3 and [Ir(ppy)2(bpy)] + , the diffusion rate will limit the quenching efficiency. Due to the non-emissive nature of [Rh(ppy)2(bpy)] + under the experimental conditions, a similar quenching study could not be performed with compound 8. The [Ir(ppy) 2 (bpy)] + model complex showed typical absorption bands at 258, 310, and 380 nm, which correspond to π→π*, metal-to-ligand charge-transfer (MLCT) and metal-centered (MC) transitions, respectively [42]. As depicted in Figure 5a, these bands could also be observed for the dyad 9; the only marginal shift of the bands suggests a negligible electronic interaction between the metal complex and the POM in the ground state. The increased molar absorption of 9 compared to the model complex is due to the contributions of the POM in the UV to the high-energy visible regime. Basically, the same photophysical behavior was found for the Rh(III)-containing dyad 8, i.e., the characteristic absorption bands known from the [Rh(ppy) 2 (bpy)] + complex (λ abs = 258, 303, and 370 nm) were also present in 8 and electronic ground-state interactions between the subunits were again excluded ( Figure SM-35).
The [Ir(ppy) 2 (bpy)] + complex is brightly luminescent at room temperature; upon excitation in its MLCT band, intense emission at 592 nm can be observed [43]. However, the dyad 9 did not exhibit any luminescence. On the other hand, a stoichiometric mixture of 3 and [Ir(ppy) 2 (bpy)] + revealed a residual emission with approximately 25% of the intensity of the small-molecule reference complex (Figure 5b). Detailed photophysical studies on the Ir(III)-containing dyads 9 and 11 with respect to the quenching mechanism are currently ongoing. However, the results line up with the observations already made for related POM-Re dyads, for which intramolecular electron-transfer processes were proposed [26]. Auvray et al. pointed out that the luminescence quenching via such a pathway would be significantly more efficient in molecular dyads comprising a covalent linkage of the components; for the intermolecular electron transfer, as in the mixture of 3 and [Ir(ppy) 2 (bpy)] + , the diffusion rate will limit the quenching efficiency. Due to the non-emissive nature of [Rh(ppy) 2 (bpy)] + under the experimental conditions, a similar quenching study could not be performed with compound 8.

DFT Studies
As pointed out above, the linkage of the photosensitizer moieties to the POMs via C-C bonds is expected to offer increased chemical stability in comparison to the related dyads, e.g., featuring hydrolytically more labile imine linkages. However, the electronic structure of the polyanions should not (or only marginally) be affected by the nature of the covalent bond to the PS. To verify this assumption, the frontier molecular orbitals (FMOs), i.e., the highest occupied and lowest unoccupied molecular orbitals (HOMO and LUMO, respectively), of the phenyl-equipped Anderson-type POM 5, were calculated. The DFT calculation was carried out on the geometry-optimized structure of 5 using the B3LYP functional (details on the computational methodology are given in the Experimental Section). The derived values were compared to the HOMO and LUMO energies reported recently by Schönweiz et al. for some TRIS-functionalized Anderson-type POMs [10]. It was proposed that the first one-electron reduction of these clusters would occur at the central metal ion, i.e., reduction from the +III to the +II state. This process, arising from a photoinduced electron transfer from the PS to the POM, is believed to contribute to the overall HER activity found for these PS-POM dyads. The calculated frontier molecular geometries and the HOMO/LUMO energies of 5 in comparison to the Mn(III)-centered TRIS-functionalized Anderson-type POM (7, Scheme 2) are depicted in Figure 6. The computed closed-shell ground-state configurations of 5 and 7 feature basically the same localization of the FMOs, which is in full agreement with the results presented by Streb and co-workers (the deviation of the calculated HOMO/LUMO energies is likely due to differences in the applied computational methods) [10]. Based on these findings, it is concluded that the electronic structure of the Anderson-type cluster unit is hardly affected by the lateral substituent. applied computational methods) [10]. Based on these findings, it is concluded that the electronic structure of the Anderson-type cluster unit is hardly affected by the lateral substituent.

Electrochemical Properties
As for the compounds possessing the Wells-Dawson-type POM unit, the electrochemical properties of the Anderson-type POM derivatives were also studied by CV and SWV. However, for solubility reasons, DMF (containing 0.1 M TBAF as supporting electrolyte) had to be used as a solvent in all cases. The phenyl-decorated POM 5 exhibited two reversible redox processes at 0.24 V and −1.34 V which correspond to the Mn IV /Mn III and Mn III /Mn II redox couple, respectively (for the latter, there is some inconsistency in the literature-this redox process has also been assigned to the Mo VI /Mo V couple, Figure SM-32) [10,18]. The bpy-functionalized counterpart 6 revealed basically the same electrochemical behavior, but with the redox potentials being shifted to 0.39 and −1.17 V, respectively (attributed to the different electronic nature of the aromatic substituent, Figure SM-33a/c). At more negative potentials, an additional reduction signal appeared for 6 (at −2.71 V). This signal decreased in intensity with the increasing number of CV cycles; thus, degradation of the cluster due to the reduction of the Mo(VI) centers is proposed under these conditions ( Figure SM-33b). However, the redox process associated with the bpy ⋅ -/bpy couple could not be identified. The redox potentials of the Rh(III)-and Ir(III)-containing dyads (10 and 11) are summarized in Table 2. In addition to the aforementioned redox processes arising from the POM unit, the typical oxidation and reduction processes associated with the attached metal complexes could also be observed. The Ir(III)-containing

Electrochemical Properties
As for the compounds possessing the Wells-Dawson-type POM unit, the electrochemical properties of the Anderson-type POM derivatives were also studied by CV and SWV. However, for solubility reasons, DMF (containing 0.1 M TBAF as supporting electrolyte) had to be used as a solvent in all cases. The phenyl-decorated POM 5 exhibited two reversible redox processes at 0.24 V and −1.34 V which correspond to the Mn IV /Mn III and Mn III /Mn II redox couple, respectively (for the latter, there is some inconsistency in the literature-this redox process has also been assigned to the Mo VI /Mo V couple, Figure SM-32) [10,18]. The bpy-functionalized counterpart 6 revealed basically the same electrochemical behavior, but with the redox potentials being shifted to 0.39 and −1.17 V, respectively (attributed to the different electronic nature of the aromatic substituent, Figure SM-33a/c). At more negative potentials, an additional reduction signal appeared for 6 (at −2.71 V). This signal decreased in intensity with the increasing number of CV cycles; thus, degradation of the cluster due to the reduction of the Mo(VI) centers is proposed under these conditions ( Figure SM-33b). However, the redox process associated with the bpy · -/bpy couple could not be identified. The redox potentials of the Rh(III)-and Ir(III)-containing dyads (10 and 11) are summarized in Table 2. In addition to the aforementioned redox processes arising from the POM unit, the typical oxidation and reduction processes associated with the attached metal complexes could also be observed. The Ir(III)-containing dyad 11, for example, showed the typical Ir IV /Ir III redox couple at 0.83 V; moreover, the ligand-centered reductions (ppy/ppy − bpy/bpy − ) and were identified at negative potentials (Figure 7).

UV/vis Absorption and Emission Properties
UV/vis absorption and emission spectroscopy were employed to obtain a first insight into the photophysical properties of the dyads based on the bpy-functionalized Anderson-type POM 6. As already pointed out in the context of the dyads 8 and 9 (see Section 2.4.2), the UV/vis absorption spectra of 10 and 11 reflected the absorption behavior of the subunits, i.e., the central POM and the attached Rh(III) or Ir(III) complexes. The only marginal difference in the position and shape of the bands pointed to a negligible electronic interaction in the ground state ( Figure 8a). As for the Ir(III)containing dyad 9, no room-temperature emission was detected for 11. This behavior was rationalized by an intramolecular electron-transfer from the photo-excited Ir(III) centers to the POM moieties [26]. As shown by the control experiment, the intermolecular emission quenching was by far less efficient: The mixture of 5 and [Ir(ppy)2(bpy)] + , in a 1:2 ratio, revealed an emission at 592 nm with a relative intensity of ca. 70% compared to that of the [Ir(ppy)2(bpy)] + reference (Figure 8b). Currently, in-depth photophysical studies of compound 11, including time-resolved measurements, are ongoing. With respect to a potential application of 11 as photocatalyst in the HER, a fundamental understanding of the photophysical (and electrochemical) properties is required. This comparative study, also taking the analogous Ir(III)-containing dyad reported by Streb et al. into account [10,18], will be published elsewhere. As pointed out above, quenching studies using the Rh(III)-containing dyad 10 were obsolete, due to the non-emissive character of [Rh(ppy)2(bpy)] + at room temperature.

UV/vis Absorption and Emission Properties
UV/vis absorption and emission spectroscopy were employed to obtain a first insight into the photophysical properties of the dyads based on the bpy-functionalized Anderson-type POM 6. As already pointed out in the context of the dyads 8 and 9 (see Section 2.4.2), the UV/vis absorption spectra of 10 and 11 reflected the absorption behavior of the subunits, i.e., the central POM and the attached Rh(III) or Ir(III) complexes. The only marginal difference in the position and shape of the bands pointed to a negligible electronic interaction in the ground state ( Figure 8a). As for the Ir(III)-containing dyad 9, no room-temperature emission was detected for 11. This behavior was rationalized by an intramolecular electron-transfer from the photo-excited Ir(III) centers to the POM moieties [26]. As shown by the control experiment, the intermolecular emission quenching was by far less efficient: The mixture of 5 and [Ir(ppy) 2 (bpy)] + , in a 1:2 ratio, revealed an emission at 592 nm with a relative intensity of ca. 70% compared to that of the [Ir(ppy) 2 (bpy)] + reference (Figure 8b). Currently, in-depth photophysical studies of compound 11, including time-resolved measurements, are ongoing. With respect to a potential application of 11 as photocatalyst in the HER, a fundamental understanding of the photophysical (and electrochemical) properties is required. This comparative study, also taking the analogous Ir(III)-containing dyad reported by Streb et al. into account [10,18], will be published elsewhere. As pointed out above, quenching studies using the Rh(III)-containing dyad 10 were obsolete, due to the non-emissive character of [Rh(ppy) 2 (bpy)] + at room temperature.

2-([2,2 -Bipyridin]-4-yl)-2-(hydroxymethyl)
propane-1,3-diol (bpy-triol, 2). 4-Methyl-2,2 -bipyridine (1 g, 5.87 mmol) and KOH (20 mg, 0.36 mmol) were filled into a microwave vial. Formalin solution (37 wt-%, 10 mL) was added, and the mixture was stirred for 10 h at 130 • C under microwave irradiation. In the course of the reaction, the pressure inside the vial was raised to 14 bar. After cooling, the reaction mixture was evaporated to dryness. Methanol (30 mL) was added and evaporated to remove the remaining formaldehyde (this step of the work-up procedure was repeated). The solid was redissolved in methanol (30 mL); concentrated HCl was added until the solution showed a pH value of 1 (ca. 20 drops), and the solution was stirred overnight under reflux. Afterward, the solvent was evaporated to yield a red solid, which was suspended in ethanol (5 mL). The white hydrochloride precipitate was filtered off and was subsequently washed with ethanol and diethyl ether (each 5 mL). The filter residue was redissolved in water (10 mL), and solid Na 2 CO 3 was added until the solution became basic (pH value of ca. 10). The white precipitate was isolated and washed with ethanol and diethyl ether (each 5 mL) to yield 2 as a white solid. 1 13  Crystal Structure Determination. The intensity data of 5 and 6 were collected on a Nonius KappaCCD diffractometer using graphite-monochromated Mo-K α radiation (Bruker AXS, Karlsruhe, Germany). Data were corrected for Lorentz and polarization effects; absorption was considered on a semi-empirical basis using multiple-scans [46][47][48]. The structures were solved by direct methods (SHELXS) and refined by full-matrix least-squares techniques against Fo 2 (SHELXL-97) [49]. All hydrogen atoms were included at calculated positions with fixed thermal parameters. All non-disordered, non-hydrogen atoms were refined anisotropically [49]. Mercury was used for structure representations [50]. Computational methodology. The theoretical calculations based on density functional theory (DFT) were performed with the Gaussian-16 program package (version A.03) [51]. The standard B3LYP functional was selected [52,53]. The 6-31G** basis set was used for all atoms (C, H, O, N) except for Mo and Mn, which were described by an effective core potential and the associated orbitals (LANL2DZ). For all calculations, the solvent environment was modeled for DMF using the implemented polarization continuum model (PCM) [54,55]. The corresponding geometries of the singlet ground states were optimized from reasonable initial estimates. In cases of difficult SCF convergence, additional extra quadratic (xqc) functions were used. The true nature of all minima structures was confirmed by vibrational analysis showing no imaginary frequencies. The graphical visualizations of the 3D representations were generated by GaussView (version 6.0.16).

Conclusions
The tris-alkoxylation of two appropriate POM precursors with a triol-functionalized bpy derivative has yielded the corresponding organo-functionalized POM hybrids. The pending bpy units have subsequently been used for the binding of transition metal complexes, namely a [Rh(ppy) 2 ] + and an [Ir(ppy) 2 ] + fragment. These dyads have been characterized by NMR and XPS spectroscopy, as well as mass spectrometry. Basic photophysical and electrochemical studies have been conducted and indicated a negligible electronic interaction of the subunits, i.e., the POM and the metal complex, in the ground state. Moreover, emission quenching studies on the Ir(III)-containing systems have suggested an efficient intramolecular electron transfer from the photo-excited Ir(III) centers to the POM units. In-depth photophysical studies of these assemblies, in comparison to Streb's Ir(III) PS-POM conjugate [10,18], are currently ongoing. Fundamental knowledge of the photophysical processes in such dyads is of significant importance to understand their photocatalytic activity in the HER. In this respect, the chemically more robust dyad 11 is expected to be superior to the related imine-based dyad [10,18], in particular under hydrolytic conditions (e.g., using DMF/water mixtures, as solvents, and/or AcOH, as proton source). A comparative study taking these aspects into account is also the subject of current research.
Supplementary Materials: NMR and mass spectra of all synthesized compounds, the XPS data of all POM-containing dyads, the output files of the computational studies as well as additional electrochemical and photophysical spectra are available online.