Oxidative quenching within photosensitizer-acceptor dyads based on bis(bidentate) phosphine-connected osmium(II) bipyridyl light absorbers and reactive metal sites.

For the first time oxidative quenching of OsP2N4 chromophores by reactive PtII or PdII sites containing cis, trans, cis-1,2,3,4-tetrakis(diphenylphosphino)cyclobutane (dppcb) is directly observed despite the presence of a saturated cyclobutane backbone "bridge". This dramatic effect is measured as a sudden temperature-dependent onset of a reduction in phosphorescence lifetime in [Os(bpy)2(dppcb)MCl2](SbF6)2 (M = Pt, 1; Pd, 2). The appearance of this additional energy release is not detectable in [Os(bpy)2(dppcbO2)](PF6)2 (3), where dppcbO2 is cis, trans, cis-1,2-bis(diphenylphosphinoyl)-3,4-bis(diphenylphosphino)cyclobutane. Obviously, the square-planar metal centers in 1 and 2 are responsible for this effect. In line with these observations, the emission quantum yields at room temperature for 1 and 2 are drastically reduced compared with 3. Since this luminescence quenching implies strong intramolecular interaction between the OsII excited states and the acceptor sites and depends on the metal⋯metal distances, also the single crystal X-ray structures of 1-3 are given.

There is increasing interest in supramolecular devices consisting of luminescent, redox-active transition metal complexes for photoinduced H2-production via water splitting [1]. A lot of effort has been made with respect to RuII polypyridyl compounds, but OsII polypyridyl species also exhibit interesting photophysical and redox properties [2]. Both classes of compounds show triplet metal-to-ligand charge-transfer (3MLCT) states after photoexcitation. These 3MLCT states are known to be luminescent and long-lived and undergo oxidative quenching via charge separation [3]. In the case of polypyridine complexes of OsII oxidative quenching of the excited states by molecular dioxygen is a well-known process [4]. The OsII containing metal complex normally plays the role of a detector (light emission from the osmium center) for the arriving energy quantity [5]. Combined with electron acceptors (A) these OsII photosensitizers (P) produce so-called dyads and the disappearance of photoluminescence has been traced to an intramolecular electron transfer quenching process of the 3MLCT excited states [6].

The design of heterodimetallic compounds capable of performing charge separation leads to appropriate building blocks, fulfilling the D–P–A concept required for photochemical water splitting, where D is an electron donor [7]. Based on these insights the development of artificial photochemical molecular devices (PMDs) for photoinduced water splitting is feasible, in which a photoredox-active center is linked to a reaction center through a redox-active bridge [8]. However, the necessity of a truly redox-active bridge strongly depends on the metal⋯metal distances [9]. Furthermore, the use of OsII polypyridyl complexes shows advantages compared to their RuII analogs [9a]. Since in the OsII compounds the deactivation channel involving activated surface crossing to the upper lying d→d excited levels cannot enter into play [9a,9b,10], this leads to a photochemically inert behavior. Thus, with respect to photochemical water splitting no inactivation of the catalyst via photoreaction is possible.

In this work we present the first complexes of the type [Os(bpy)2(dppcb)MCl2](SbF6)2 (M = Pt, 1; Pd, 2; see Chart 1), where dppcb is cis, trans, cis-1,2,3,4-tetrakis(diphenylphosphino)cyclobutane [11]. The great effort, which is necessary to prepare these ideal molecular devices, will often be ignored [5]. Furthermore, for purposes of comparison also the new species [Os(bpy)2(dppcbO2)](PF6)2 (3) has been synthesized, where dppcbO2 is cis, trans, cis-1,2-bis(diphenylphosphinoyl)-3,4-bis(diphenyl-phosphino)cyclobutane. 1–3 are fully characterized1 including their single crystal X-ray structure analyses,2 which are a major clue to the understanding of their photophysical properties. Indeed, the small metal⋯metal distances in 1 and 2 (vide infra) allow an orbital mixing between the Os-chromophore and the reactive metal center [9a,12]. However, energy and/or electron transfer via the backbone can be neglected, since a region of saturated carbons disrupts significant electronic wave function mixing through the cyclobutane backbone “bridge” [9a,9b]. Though photoinduced intramolecular electron transfer is well-known in 1,2,4,5-tetrakis(diphenylphosphino)benzene-bridged di- and trimetallic complexes [9c,9d], we show for the first time that oxidative quenching of the OsP2N4 chromophores in 1 and 2 can be directly observed despite the presence of the saturated cyclobutane ring.

Chart 1

Structure types: heterodimetallic complexes 1 and 2 and monometallic species 3.

The absorption spectra are shown in Fig. S1 as supplementary material (page S1). For the new complexes 1–3 there are intense π, π* absorption bands in the UV region of the absorption spectra. They are due to intraligand (IL) transitions localized on the 2,2′-bipyridyl ligands of the [Os(bpy)2]2 + chromophore [9a,9b]. The maxima occur at 272 (1; ε = 45 600 dm3 mol− 1 cm− 1) and 277 nm (2; ε = 47 800; 3; ε = 30 700). The transitions centered at 315 (1; ε = 21 000; 2; ε = 22 500) and 307 nm (3; ε = 23 000) are attributed to the metal-centered (MC) d→d transitions of the Os center [9a]. Generally speaking, the MLCT absorption bands of OsP2N4 moieties like in 1–3 are blue-shifted from that observed for [Os(bpy)3](PF6)2 (λmax = 640 nm) because of the stabilization of the ground state by the enhanced d,π(Os)–σ*,π(P) back-bonding and the destabilization of the excited state by the poorer σ-donating phosphine ligands [9d]. Thus, the MLCT (Os→bpy) transitions are observed at 375 (1; ε = 9500; 2; ε = 9700) and 371 nm (3; ε = 7300). These MLCT states are predominantly singlet in character [9a,9d]. Further pronounced MLCT transitions concerning states, that are predominantly triplet in character, occur at 462 (1; ε = 2500; 2; ε = 1800) and 473 nm (3; ε = 1990). These UV–vis absorption data of 1–3 clearly indicate that they are dominated by a single [Os(bpy)2]2 + chromophore. Therefore, a simple visible light-driven energy transfer between “antenna” and “trap” sites, as measured in [Ru(bpy)(CH3CN)2(dppcb)Os(bpy)2]4 + and [Ru(bpy)2(dppcb)Os(bpy)2]4 + [9a,9b], is not possible in 1–3.

The emission spectra of 1–3 are shown in Figs. S2–S8 as supplementary material (pages S3–S6). The excitation and emission band maxima of 1–3 are summarized in Table 1. In the cases of 1 and 2 it is obvious that within a wide wavelength range, from about 300 to 550 nm, excitation occurs into the same emissive state (see Figs. S2–S5, pages S3–S5). Independent from this wide excitation wavelength range always the same shape of the emission spectrum is obtained, clearly indicating that no impurities are involved. Furthermore, the emission intensities exactly match the corresponding excitation intensities depending on the excitation wavelength. The emission band maxima at 298 K (Table 1) of 603 and 600 nm for 1 and 2, respectively, are typical of an emission from the MLCT (Os→bpy) states of the [Os(bpy)2]2 + chromophore combined with a bis(bidentate) phosphine [9]. These maxima show the expected blue shift at 77 K, occurring at 573 (1) and 570 nm (2). At this temperature the appearance of vibrational progressions is also in line with this kind of chromophore [9a,9b].

Table 1

Photophysical parameters for [Os(bpy)2(dppcb)MCl2](SbF6)2 (M = Pt, 1; Pd, 2) and [Os(bpy)2(dppcbO2)](PF6)2 (3).

Compound	Excitation band maxima, nm		Emission band maxima, nm		τa,b	τb,c	Φra
	298 Ka	77 Kc	298 Ka	77 Kc	298 K, ns	77 K, μs	298 K
1	375	480	603	573	219(11)	3.42(8)	0.0061
2	370	480	600	570	240(6)	3.35(14)	0.0077
3	520	480	605	580	214(6)	2.8(1)	0.024

In degassed spectrograde CH3CN.

The excitation/emission wavelengths (nm) are: 1. 355/600 (298 K), 480/573 (77 K). 2. 355/600 (298 K), 480/570 (77 K). 3. 520/600 (298 K), 480/580 (77 K).

In a 4:1:2 (v/v, degassed spectrograde quality) EtOH/MeOH/CH3CN mixture.

In the case of 3 the single MLCT (Os→bpy) emission also fulfills these conditions of a typical [Os(bpy)2]2 + chromophore (see Figs. S6–S8, pages S5, S6 and Table 1). Moreover, Figs. S3, S5, and S8 (pages S4–S6) clearly reveal that also the excitation spectra of 1–3 at 77 K are nearly identical. This means that at this temperature the photophysically relevant electronic properties of 1–3 are dominated by the presence of the [Os(bpy)2]2 + chromophore. However, at 298 K in contrast to 1 and 2 Fig. S7 (page S6) shows that for 3 the red tail of the MLCT states, that are predominantly triplet in character, leads to its excitation maximum at 520 nm. The combination of the [Os(bpy)2]2 + chromophore with reactive square-planar PtII or PdII centers in 1 and 2, respectively, leads to systems that possess further stabilized MLCT (Os→bpy) excited states (predominantly triplet in character) relative to the monometallic analog 3 [9a,13]. Obviously, these states are emissive, since only in the case of 3 the MLCT (Os→bpy) states centered at 371 nm, that are predominantly singlet in character, are nearly no more emissive (see Fig. S7, page S6). Since the “steric pressure” in 1–3 is comparable (vide infra), steric effects only play a minor role.

The photophysical data related to the excited state decay of 1–3 are summarized in Table 1. At 77 K 1–3 show very long luminescence lifetimes typical for 3MLCT states [9a,9b]. They are nearly identical within statistical significance. The values of 3.42(8), 3.35(14), and 2.8(1) μs for 1–3, respectively, are completely in line with the energy gap law [9a], since the corresponding emission centers occur at 573, 570, and 580 nm. At 298 K the luminescence lifetimes of 1–3 are the same within statistical significance and also identical with the corresponding parameter of 243(8) ns for the homodimetallic analog [Os2(dppcb)(bpy)4]4 + [9b]. The values of 219(11), 240(6), and 214(6) ns for 1–3, respectively, are again in agreement with the energy gap law [9a], showing the emission centers at 603, 600, and 605 nm. However, Figs. S9 and S10 (page S8) clearly reveal that the temperature dependencies of the luminescence lifetimes of 1–3 show different features. Fig. S9 shows that in both compounds 1 and 2 there is an additional effect of energy release starting above about 170 K. Since the upper lying d→d excited levels are not populated in the case of OsII compounds, this effect is attributed to oxidative quenching of the chromophore by the reactive metal centers. This is confirmed by the fact that for compound 3 only the typical glass transitions of the cryogenic glass are observed as shown in Fig. S10 [10]. Furthermore, the lack of oxidative quenching for 3 is also obvious regarding the emission quantum yields of 0.0061 (1), 0.0077 (2), and 0.024 (3).

The single crystal X-ray structure analyses of 1–3 are also in line with these considerations. Thus, the Os⋯Pt and Os⋯Pd distances are 7.300(1) and 7.282(1) Å in 1 and 2, respectively, leading to orbital overlap between the metal centers [9a]. This means that despite the presence of a saturated cyclobutane backbone electron transfer is directly possible between the metal centers. To the best of our knowledge the crystal structures of 1 and 2 are the first structures of heterodimetallic complexes, where octahedral [Os(bpy)2]2 + chromophores and square-planar metal centers are connected by a tetraphosphine (Fig. 1). The luminescence quantum yields of 1–3 (Table 1) clearly indicate that the energy loss of their excited states is dominated by non-radiative decay. This energy loss occurs into a series of medium-frequency ring-stretching vibrations with energy spacings between 1000 and 1600 cm− 1 [9a]. However, the availability of these vibrations mainly depends on the “steric pressure” within the excited molecules [9a,9b].

Fig. 1

ORTEP diagram of the cation of 1 with 30% probability ellipsoids. The hydrogen atoms are omitted for clarity and only the ipso carbon atoms of the phenyl units are shown. The molecular structure of 2 is analogous to 1. Selected bond lengths (Å) and angles (°): 1. Os1⋯Pt1 7.300(1), Os1–P1 2.342(3), Os1–P2 2.254(3), Pt1–P3 2.238(3), Pt1–P4 2.164(3), P1–Os1–P2 84.24(11), P3–Pt1–P4 89.08(11); 2. Os1⋯Pd1 7.282(1), Os1–P1 2.306(4), Os1–P2 2.323(4), Pd1–P3 2.236(3), Pd1–P4 2.234(3), P1–Os1–P2 86.44(12), P3–Pd1–P4 84.20(11).

Thus, the shortest intramolecular contacts between a bpy ligand and a phenyl ring are 2.425 Å in 1, 2.105 Å in 2, and 2.297 Å in 3. The analogous contacts between the phenyl rings along a trans axis of the cyclobutane rings are 2.939 Å in 1, 2.608 Å in 2, and 2.534 Å in 3. These contact approaches are indicative of high “steric pressure” [9a,9b], being present in the heterodimetallic species 1 and 2 as well as in the monometallic species 3. Fig. S11 (page S10) clearly reveals, how this “steric pressure” is partially released via “envelope” folding of the five-membered metallacycles in 1 and 2. The corresponding folding angles are 158.4(5)° and 164.7(5)° in 1 and are the dihedral angles between the least-squares planes through the atoms P1, C1, C2, P2 and P1, Os1, P2 and through P3, C3, C4, P4 and P3, Pt1, P4, respectively (Fig. S11, below). Both foldings occur toward the cyclobutane ring. The analogous folding angles are 159.1(3)° and 151.7(2)° in 2. However, only the latter folding occurs again toward the cyclobutane ring (Fig. S11, above). The former folding angle corresponds to the Os(II) moiety and this folding is orientated away from the cyclobutane ring being typical of high “steric pressure” [9a]. Furthermore, the cyclobutane ring in 1 is planar within statistical significance, whereas in 2 the deviations from a least-squares plane through the cyclobutane ring are 0.026(5) Å for C(1) and C(3), and − 0.026(5) Å for C(2) and C(4). The deviations of the platinum and palladium atoms in 1 or 2 from the coordination planes, defined by P(3), P(4), Cl(1), and Cl(2) (Fig. S11), are 0.074(1) Å for Pt(1) and 0.126(1) Å for Pd(1). They occur toward the cyclobutane rings and are also significantly different, being in line with a reduction of the square-planar stabilization energy in Pd(II) compared with Pt(II). At this point it is important to note, that these subtle conformational differences in 1 and 2 play a minor role for their photophysical parameters (see Table 1).

Due to the monometallic character of 3 the folding of the cyclobutane ring is more pronounced and slightly asymmetric (Fig. 2). Thus, the deviations from a least-squares plane through the cyclobutane ring are 0.170(3), − 0.169(3), 0.167(3), and − 0.168(3) Å for C(1)–C(4), respectively. This corresponds to a very effective partial release of “steric pressure” and therefore the remaining “envelope” folding angle of the five-membered metallacycle in 3, defined as above, is only 176.2(3)°. Nevertheless, the overall “steric pressure” measured via the intramolecular contact distances given above is comparable in 1–3. Therefore, electronic reasons must be responsible for the dramatic reduction of the emission quantum yields of 1 and 2 compared with 3 (Table 1). The same is true for the additional effect of energy release starting above about 170 K, being only present in 1 and 2. Since d→d excited levels are not available in 1–3, this possibility can also be ruled out. Therefore, oxidative quenching of the chromophores in 1 and 2 by their reactive metal centers is held responsible for the photophysical differences in 1 and 2 compared with 3. The accessibility of reduced states for Pt(II) in 1 and Pd(II) in 2 is evident from their cyclic voltammograms (see footnote 1). Furthermore, derivatives of 2 are active water splitting precatalysts, producing H2 during photolysis with visible light [9a]. This is only possible, if electron transfer occurs.

Fig. 2

ORTEP diagram of the cation of 3 with 30% probability ellipsoids. The hydrogen atoms are omitted for clarity. Selected bond lengths (Å) and angles (°): Os1–P1 2.3233(15), Os1–P2 2.3185(16), P3–O1 1.488(5), P4–O2 1.484(5), P1–Os1–P2 85.02(5).