Heavy-Atom Free spiro Organoboron Complexes As Triplet Excited States Photosensitizers for Singlet Oxygen Activation

Herein, we present a new strategy for the development of efficient heavy-atom free singlet oxygen photosensitizers based on rigid borafluorene scaffolds. Physicochemical properties of borafluorene complexes can be easily tuned through the choice of ligand, thus allowing exploration of numerous organoboron structures as potent 1O2 sensitizers. The singlet oxygen generation quantum yields of studied complexes vary in the range of 0.55–0.78. Theoretical calculations reveal that the introduction of the borafluorene moiety is crucial for the stabilization of a singlet charge transfer state, while intersystem crossing to a local triplet state is facilitated by orthogonal donor–acceptor molecular architecture. Our study shows that quantitative oxidation of selected organic substrates can be achieved in 20–120 min of irradiation with only 0.05 mol % loading of a photocatalyst.


Absorption and emission spectra
Absorption spectra were recorded using Hitachi U-2800 spectrophotometer. Emission and fluorescence quantum yields were recorded using spectrofluorometer Edinburgh FS5. The measurements were performed at room temperature, according to published procedures. 1,2 Suprasil quartz cuvettes (10.00 mm) were used. 1.5 nm slits were used for absorption and 2.5 nm slits were used for emission spectra. To eliminate any background emission, spectrum of pure solvent was subtracted from the samples' spectra. QY F were determined in diluted solutions (A < 0.1 for longest wavelength band) by comparison with known standardsrhodamine 6G (EtOH, QY F r = 0.94) 3 , fluorescein (0.1 M NaOH, QY F r =0.95) 4 , coumarin 153 (EtOH, QY F r = 0.54) 5 , quinine sulfate (0.1 M HClO4, QY F r = 0.60) 6 Concentrations were in the range of 0.5-2·10 -5 moldm -3 . Concentrations were adjusted to reach similar absorbance to absorbance of reference solution at the excitation wavelength. Fluorescence quantum yield was determined using reference substance using following formula: where: where F is the relative integrated photon flux of sample (x) and reference (r), A is the absorbance at the excitation wavelength, n is the refractive index of used solvents. Photon fluxes (F) were calculated by integration of corrected spectra (Ic), obtained by (I) division of intensity of emission spectra by the spectral responsivity (s) in corresponding wavelengths (λem). All measurements were carried out at room temperature.
= ∫ c em = ∫ ( em ) ( em ) em Spectroscopic data (measured and taken from literature) are stored in Table S1.

Photocatalytic setup
To ensure stable and repeatable conditions in all performed photocatalytic reactions we have designed our home made photoreactor. It is suitable for small scale reactions in 4 ml vials, however due to its modular design it can be easily used for reactions in larger vessels (up to 50 ml). The reactor comprises of main body housing the light source, coversample holder and a reflective base. The photoreactor can be placed on a magnetic stirrer. Main body of the reactor is made of aluminium tube (Φ150 mm, 4 mm wall) to ensure high reflection of light and proper heat transmission. LED strips connected in parallel were glued inside the tube to achieve even distribution of light. To maintain low temperature aluminium tubes is water cooled by a copper coiled tube heat exchanger sticked to the outside wall of the reactor. To the bottom of the aluminium tube three plastic legs were attached to ensure flow of air. The main body was placed on a basea square sheet of aluminium to ensure stable footing and reflection of stray light. As light source neutral white and UV (395 nm and 365 nm) LED strips were used. White LED strip SMD5630 (28.8 W/m, 3000 lumen/m) was purchased from www.akb-poland.com. 395 nm and 365 nm realUV TM LED strips (15 W/m) were purchased from www.waveformlighting.com. White light reactors were equipped with 0.9 m LED stripes (26 W, 54 diodes) and for UV reactors were equipped with 1.7 m LED stripes (26W, 204 diodes). The cover was made of round aluminium sheet (Φ141.5 mm) comprising 8 holes for reaction vials, a fan (Φ60 mm) with air diffusor and four adjustable handles. The fan with a diffusor directing air flow to the LED strips was mandatory to disperse heat radiating from the LEDs. The samples are located possibly close to the centre of the reactor to minimise convection of heat from the LEDs. The air cooling proved necessary, as without it temperature inside the photoreactor exceeded 40 °C. As reaction vessels we have used commercially available 4 ml vials. Vial were held by the plastic screw cap. Holes in the plastic screw caps were made to ensure access of air to the reaction mixture. To ensure vigorous mixing cross shaped stirrer bars were used. Hook type handles allow adjusting of the lid position inside main body. Temperature inside reactor was controlled by placing Pt-100 thermometer into one of the reaction vials filled with water. In such conditions temperature inside the reactor oscillates close to 25 °C. S7 Figure S7. Model of assembled photoreactor.

Catalytic reactions
We have found that oxidation might be inhibited by solvent stabilizer. The oxidation of FA with Bf-A3 in HPLC grade CHCl3 containing amylene as stabilizer delivers unsatisfactory conversion of 16% after 1.5 h. However, with ACS or technical grade CHCl3 containing EtOH, the conversion rose to 33%. Further addition of EtOH accelerated oxidation, with conversion reaching 49% after 1 h in 1:2 CHCl3:EtOH mixture. Nonetheless, markedly increased conversion rates of FA were gained with more polar MeCN solvent. This process was, however, unaffected by addition of EtOH.
The control experiments show that reaction does not proceed neither in the absence of light nor photocatalyst. Since, oxidation of thioanisole and triphenylphosphine is considered to proceed either through singlet oxygen or electron transfer followed by formation of radicals, [13][14][15] we have performed two additional sets of photooxidation experiments with Bf-A3 in the presence of TEMPO (1.5 eqv. with respect to substrate) as the radical inhibitor and DABCO (1.5 eqv.) as the singlet oxygen scavenger. We have found that oxidations were supressed in the presence of both DABCO and TEMPO indicating that the process might be mediated by 1 O2 and O2 ‧radical generated in the reaction.

Photocatalytic stability
Photocatalytic stability under reaction conditions for all analysed photosensitizers was determined by UV-Vis spectroscopy using Hitachi U-2800 spectrophotometer. Additionally, corresponding measurements were carried out without irradiation, on samples held in darkroom. Referential dyes, Rose Bengal (RB) and tetraphenylporphyrin (TPP), were measured in the same manner for comparison. Experimental conditions were retained from photocatalytic test reactions. The parallel experiments were performed without light to determine hydrolytic stability.

Crystal structures
Single crystals suitable for single crystal diffraction measurements of BPh2-A1, BF2-A1, Bf-A2 and Bf-A1 were obtained by slow solvent evaporation from corresponding CHCl3 solutions. X-ray diffraction data were collected on a SuperNova diffractometer equipped with Atlas detector using Cu-Kα radiation (λ = 1.5418 Å). Data reduction and analysis were carried out with the CrysAlisPro program. 16 All structures were solved by intrinsic phasing using SHELXT 17 and refined using SHELXL-2014 18 with Olex2 suite. 19 All non-hydrogen atoms were refined anisotropically. Selected crystal data are summarized in Table S3. The crystal structure of BPh2-A1 is similar to already published structure of related BODIPY system, namely 2,6-Diethyl-4,4-difluoro-1,3,5,7-tetramethyl-8-(4-methylphenyl)-4-bora-3a,4a-diazas-indacene. The latter system comprises (4-methylphenyl) substituent at meso position of BODIPY frame. 20 The geometrical parameters of both molecules are in general very similar, however the crystal packing is different (BPh2-A1 crystalizes in monoclinic P21/c space group, while its methyl-substituted derivative in triclinic P-1 space group with two molecules in the asymmetric part of the unit cell

Theoretical calculations
Ab initio calculations were performed using Gaussian16 programme package. 22 In the first step the molecules were optimised in their ground states using B3LYP (DFT) 23,24 method with 6-31+G(d) 25 basis set. Starting geometries were adopted from crystal structures. Structures BPh2-A1, BF2-A1, Bf-A1 and Bf-A2 were measured by us in the framework of this work, Bf-A3, 21 Bf-B1, Bf-B2, Bf-B3, Bf-C, Bf-D, Bf-E 12 are already published. Crystal structures of BPh2-A2 and BF2-A2 are unknown, thus they were constructed in GausView 6.0 program based on closely related BPh2-A1 and BF2-A1 analogues. After geometry optimization, the vibrational frequencies were calculated and the results showed that optimized structures are stable geometric structures (no imaginary frequencies). Calculated molecular orbitals for representative complexes (BF2-A1, BPh2-A1, Bf-A1, Bf-A3, Bf-B1 and Bf-C) in their ground and excited states are presented on Figures S39-S44. They were visualized with Avogadro programme. 26 In the next step, excited singlet and triplet state geometries were obtained with TD-DFT methods. DFT and TD-DFT calculations were performed in the presence of the solvent (MeCN) with the polarizable continuum model (PCM). 27 The energy values of two lowest energy excited singlet and five lowest energy triplet states with respect to ground state energy are provided in Table S4. The calculations show that molecular geometries of singlet and triplet excited states are generally preserved from corresponding ground states ( Figure S45). The estimated values for emission maxima were compared to experimental values (Table S5). Since the calculated values are typically underestimated with respect to experimental values by 10-60 nm, it is expected that in case of borafluorene complexes the second lowest energy singlet excited state would correspond to experimentally observed transition, while excitation to lowest energy singlet excited state would be inefficient. This statement is supported by low oscillator strength value for CT transition, and much higher oscillator strength value for LE transition.
In order to determine the nature of singlet and triplet excited states Natural Transition Orbitals 28 were calculated. The NTO calculations were performed for BF2-A1, BPh2-A1 and Bf-A1 series in order to elucidate the effect of boron atom substitution, and for Bf-A3, Bf-B1 and Bf-C complexes to show differences between various ligands ( Figure S48-S53). The energy diagrams are presented on Figure S46 and S47. In case of Bf complexes 1 CT character of the lowest energy singlet excited state is confirmed by the location of highest occupied (HONTO) and lowest unoccupied natural transient orbitals (LUNTO). The location of HONTO and LUNTO for second lowest energy singlet excited state indicates that it possesses local character ( 1 LE-Lig). Spin density isosurfaces generated for all considered molecules in their optimised lowest energy triplet states ( Figure S54, S55) indicate that the lowest energy triplet state is located on ligand ( 3 LE-Lig). The energy of this state is similar comparing the corresponding BF2, BPh2 and Bf complexes from A1 and A2 series. This suggests that boron substitution has marginal influence on the energy of lowest energy triplet state. In turn, the observed differences between Bf complexes (Bf-BODIPYs, Bf-B1, Bf-B2, Bf-B3, Bf-C, Bf-D, Bf-E) originate from different ligand electronic features. Table S4. Calculated energy values of two lowest energy singlet (Sn) and five lowest energy triplet (Tn) states for all studied complexes. The values are given with respect to the energy of ground state in its optimised geometry. The nature of excited states in BF2-A1, BPh2-A1, Bf-A1, Bf-A3, Bf-B1 and Bf-C were evaluated with NTO analysis.       Figure S54. Spin density (iso = 0.0004) for all studied BODIPYs in theirs optimised triplet states. Hydrogen atoms were omitted for clarity. Figure S55. Spin density (iso = 0.0004) for borafluorene complexes (except BODIPYs) in theirs optimised triplet states. Hydrogen atoms were omitted for clarity.  Table S9. Atomic coordinates for optimised structure of Bf-A1 in its S1 singlet excited state.  Table S36. Atomic coordinates for optimised structure of Bf-D in its S1 excited state.  Table S51. Atomic coordinates for optimised structure of BF2-A1 in its S1 singlet excited state.  Table S54. Atomic coordinates for optimised structure of BF2-A2 in its S1 singlet excited state.