Exploring the Potential of Al(III) Photosensitizers for Energy Transfer Reactions

Three homoleptic Al(III) complexes (Al1–Al3) with different degrees of methylation at the 2-pyridylpyrrolide ligand were systematically tested for their function as photosensitizers (PS) in two types of energy transfer reactions. First, in the generation of reactive singlet oxygen (1O2), and second, in the isomerization of (E)- to (Z)-stilbene. 1O2 was directly evidenced by its characteristic NIR emission at around 1276 nm and indirectly by the reaction with an organic substrate [e.g. 2,5-diphenylfuran (DPF)] using in situ UV/vis spectroscopy. In a previous study, the presence of additional methyl groups was found to be beneficial for the photocatalytic reduction of CO2 to CO, but here Al1 without any methyl groups exhibits superior performance. To rationalize this behavior, a combination of photophysical experiments (absorption, emission and excited state lifetimes) together with photostability measurements and scalar-relativistic time-dependent density functional theory calculations was applied. As a result, Al1 exhibited the highest emission quantum yield (64%), the longest emission lifetime (8.7 ns) and the best photostability under the reaction conditions required for the energy transfer reactions (e.g. in aerated chloroform). Moreover, Al1 provided the highest rate constant (0.043 min–1) for the photocatalytic oxygenation of DPF, outperforming even noble metal-based competitors such as [Ru(bpy)3]2+. Finally, its superior photostability enabled a long-term test (7 h), in which Al1 was successfully recycled seven times, underlining the high potential of this new class of earth-abundant PSs.

1 Further Experimental Details.Inert conditions.Inert conditions or deaerated solvents refer to experimental setups where freshly distilled and dried solvents were used in combination with Schlenk techniques and argon as an inert gas.This procedure is used to prevent any interaction with atmospheric oxygen or moisture.
To determine the 1 O 2 quantum yield Φ1 O 2 , a sufficiently intense and well resolved 1 O 2 emission signal is required, which in the case of these Al(III) complexes is only available in CDCl 3 (see Figure S2 & S3).However, commonly used reference compounds such as perinaphthenon (PN) either have no reported values in CDCl 3 or lack stability in this solvent.The heteroleptic Cu(I) complex [Cu(bcp)(xant)]PF 6 (CuPS, with bcp = bathocuproine and xant = xantphos, see Figure S14) could serve as another possible reference compound.However, both PN and CuPS decompose immediately or over time when irradiated in CDCl 3 , making them unsuitable for our needs.
Table S1.Excitation wavelengths λ exc of 1 O 2 emission measurements in various solvents (Figure S2).The experimental and theoretical spectra show the same features and strong similarities in shape.The red shift with higher degree of methylation is well described by the TDDFT calculations.The TDDFT simulations predict a switch of the electronic character from 3 ILCT in case of Al1 to 3 LLCT for Al2 and Al3.We account this observation to the gradually increased electronic density by the methyl groups at the pyrrolide rings.Therefore, it becomes energetically more favourable to delocalize the excited electron and hole at two different ligands than on the same 2-pyridylpyrrolide ligand.A similar behaviour was observed previously in oxidative and reductive spectroelectro-chemistry studies.S8.Spin densities of the optimized (unrestricted DFT; densities are show using an isovalue of 0.001) triplet ground states (T 1 ) of Al1, Al2 and Al3.Relative T 1 energies are provided with respect to the S 0 in the optimized S 0 equilibrium structure, respectively.

Photooxidation of 2,5-Diphenylfuran
The following experiment describe the oxidation of 2,5-diphenylfuran (DPF) by singlet oxygen, which was generated by a photosensitizer (PS) upon irradiation.The used PS is mentioned in the caption of respective spectra.The experiments were prepared in a way that there was a 1:10 ratio PS:DPF and after each hour, the same (initial) amount of DPF was added again to maintain an excess of DPF.For exact concentrations, check the experimental details (vide supra).
The marginal rise in the cis-1,2-dibenzoylethylene (DBE) signal at 260 nm following the addition of DPF can be attributed to the weak absorbance of DPF at 260 nm.Nevertheless, in comparison to the DBE, DPF has a lower absorptivity, leading to a noticeable increase in signal intensity upon conversion.

DPF Conversion Rates for Al2, Al3, Alq 3 , PN and CuPS
The value of the rate constants k is defined in equation S2 and can be interpreted as the rate constant for the DPF conversion.The rates were calculated using a  = -  This approach assumes that the kinetic behavior during DPF conversion is similar for all complexes, which is the case for the first 50 min for most complexes used.The reference compounds CuPS as well as perinapthenone (PN, Figure S10), which is known to have a singlet oxygen yield close to unity, exhibits a lower photoactivity over time (Figure S19).In the case of CuPS and PN, the degradation process of the conversion happens so fast that a first order kinetic fit was only feasible for the first 15 min.

Figure S1 .
Figure S1.Emission spectrum of the power 405 nm LED light source.

Figure S6 .
Figure S6.Photostability measurements of Al2 (left) and Al3 (right) in aerated chloroform.A 150 W Xe lamp with a 400 nm long-pass filter was used as excitation source.

Figure S7 .
Figure S7.Photostability measurements of Al1 (top left), Al2 (top right) and Al3 (bottom) in degassed chloroform.A 150 W Xe lamp with a 400 nm long-pass filter was used as excitation source.

Figure S12 .
Figure S12.Electronic difference density plots of Al3.Electron density migrating from red to blue during the S 0 → S n transition (densities are show using an isovalue of 0.001).

Figure S19 .
Figure S19.Streak-camera measurements of Al1 (top left), Al2 (top right), Al3 (bottom left) and Alq 3 (bottom right).The measurement plots show (a) the streak camera signal, (b) the emission spectra, (c) the emission decay with corresponding emission lifetime and (d) the residuals of the decay fit.

Figure S20 .
Figure S20.Subsequent DPF catalysis measurements of Al1 after one (left) and two (right) DPF additions.Each graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).Irradiation time after addition of DPF was 1 h and the irradiated solutions were not removed from the cuvette.A 150 W Xe-lamp and a 400 nm long-pass filter was used.

Figure S21 .
Figure S21.DPF catalysis measurements of Al2 for the first cycle (top left), after one addition of DPF (second cycle, top right) and after a second addition of DPF (bottom).Irradiation time after addition of DPF was 1 h and the irradiated solutions were not removed from the cuvette.Each graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).A 150 W Xe-lamp and a 400 nm long-pass filter was used.

Figure S22 .
Figure S22.DPF catalysis measurements of Al3 for the first cycle (top left), after one addition of DPF (second cycle, top right) and after a second addition of DPF (bottom).Irradiation time after addition of DPF was 1 h and the irradiated solutions were not removed from the cuvette.Each graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).A 150 W Xe-lamp and a 400 nm long-pass filter was used.

Figure S23 .
Figure S23.DPF catalysis measurements of Alq 3 for the first cycle (top left), after one addition of DPF (second cycle, top right) and after a second addition of DPF (bottom).Irradiation time after addition of DPF was 1 h and the irradiated solutions were not removed from the cuvette.The graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).A decrease of DPF conversion rates is observable after only 2 additions.A 150 W Xe-lamp and a 400 nm long-pass filter was used.

Figure S24 .
Figure S24.DPF catalysis measurements of CuPS for the first cycle (left) and after one addition of DPF (second cycle, right).Irradiation time after addition of DPF was 1 h and the irradiated solutions were not removed from the cuvette.Each graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).A strong decrease of conversion was observable during the first cycle.A 150 W Xe-lamp and a 400 nm long-pass filter was used.

Figure S25 .
Figure S25.DPF catalysis measurements of RuPS.The graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).A 150 W Xelamp and a 400 nm long-pass filter was used.

Figure S26 .
Figure S26.DPF catalysis measurements without PS for one cycle (top left) and with RuPS but without light (right).The graph contains in situ absorption spectra (top), difference absorption spectra (middle) and time dependent absorbance values (bottom) for the depletion of DPF (λ = 328 nm, black dots) and DBE generation (λ = 260 nm, grey crosses).DPF catalysis measurements without the presence of PS or light reveal no observable conversion even after 1 hour.This proves the essential role of a photosensitizer in generating 1 O 2 , which subsequently facilitates the oxidation of DPF.A 150 W Xelamp and a 400 nm long-pass filter was used.
fit (equation S2) logarithmic fit of c 0 /c t , where c 0 and c t are the initial DPF concentration (t = 0) and the concentration of DPF after time t.

Figure S27 .
Figure S27.DPF conversion rate of the first hour (crosses) and first order fit (black line) of ln(c t /c 0 ) per time.The slope of the fit is estimated to be the DPF conversion rate k and can be found in Table 2. Fits shown for DPF conversion experiments using Al2 (green, top left), Al3 (red, top right) Alq 3 (black, middle left), PN (magenta, middle right), CuPS (dark green, 2 nd row from bottom) and RuPS (orange, bottom).For CuPS, one fit for one hour (bottom left) and one for the first 15 min (bottom right) is given.

Figure S28 .
Figure S28.Photosensitized isomerization of E-stilbene with Al2 (top left), Al3 (top right), Alq3 (bottom left), CuPS (bottom right) in chloroform under inert conditions.The mole ratio of photosensitizer to Estilbene was 1:25.In situ UV/vis absorption spectra (top), the differential plot (middle) and the kinetic plots (bottom) of E-stilbene depletion at 296 nm (black dots) and Z-stilbene generation at 260 nm (grey crosses).A 405 nm power LED (~10.7 W) and a 400 nm long-pass filter was used.

Figure S29 .
Figure S29.Photosensitized isomerization of E-stilbene with Al1 but no light (top left), RuPS (top right)and no PS (bottom) in chloroform under inert conditions.The mole ratio of photosensitizer to E-stilbene was 1:25.In situ UV/vis absorption spectra (top), the differential plot (middle) and the kinetic plots (bottom) of E-stilbene depletion at 296 nm (black dots) and Z-stilbene generation at 260 nm (grey crosses).A 405 nm power LED (~10.7 W) and a 400 nm long-pass filter was used.

Table S2 .
Simulated excited state properties such as excitation energies, wavelengths oscillator strengths and electronic characters of prominent dipole-allowed transitions contributing to the electronic absorption spectrum of Al1 as obtained at (spin-free) TDDFT level of theory (B3LYP/def2-SVP) within the fully optimized singlet ground state (S 0 ) geometry.Implicit solvent effects were described by a polarizable continuum model (SMD, chloroform).

Table S5 .
Spin-orbit coupling elements (⟨  │ SOC │  ⟩ in cm -1 ) between prominent excited singlet and triplet states of Al1 (in chloroform).Excitation energies (E in eV) are given for all singlet and triplet excitations; oscillator strengths (f) are provided for singlet-singlet excitations only.The provided triplet states are selected to match the energy level of the highest considered singlet state (i.e. S 10 ).All results were obtained by TD-B3LYP as implemented in Orca 5.0.3.

Table S6 .
Spin-orbit coupling elements (⟨  │ SOC │  ⟩ in cm -1 ) between prominent excited singlet and triplet states of Al2 (in chloroform).Excitation energies (E in eV) are given for all singlet and triplet excitations; oscillator strengths (f) are provided for singlet-singlet excitations only.The provided triplet states are selected to match the energy level of the highest considered singlet state (i.e. S 10 ).All results were obtained by TD-B3LYP as implemented in Orca 5.0.3.

Table S7 .
Spin-orbit coupling elements (⟨  │ SOC │  ⟩ in cm -1 ) between prominent excited singlet and triplet states of Al3 (in chloroform).Excitation energies (E in eV) are given for all singlet and triplet excitations; oscillator strengths (f) are provided for singlet-singlet excitations only.The provided triplet states are selected to match the energy level of the highest considered singlet state (i.e. S 17 ).All results were obtained by TD-B3LYP as implemented in Orca 5.0.3.Table