Photoinduced host-to-guest electron transfer in a self-assembled coordination cage

A [Pd2L4] coordination cage, assembled from electron-rich phenothiazine-based ligands and encapsulating an electron-deficient anthraquinone-based disulfonate guest, is reported. Upon excitation at 400 nm, transient absorption spectroscopy unveils photoinduced electron transfer from the host's chromophores to the guest, as indicated by characteristic spectral features assigned to the oxidized donor and reduced acceptor. The structure of the host–guest complex was characterized by NMR spectroscopy, mass spectrometry and single-crystal X-ray analysis. Spectroelectrochemical experiments and DFT calculations both agree with the proposed light-induced charge separation. A kinetic analysis of the involved charge transfer channels reveals, besides a guest-independent LMCT path, 44% efficiency for the host–guest charge transfer (HGCT).

Single-crystal X-ray diffraction data of ligand L were collected on a Bruker d8 venture diffractometer an Incoatec Iµs 3.0 microfocussed CuK source. Due to very thin plate-shaped crystals of supramolecular coordination cage [G1@1] 2+ , the analysis was hampered by the limited scattering power of the samples not allowing us to reach the desired atomic resolution using in-house equipment. Gaining detailed structural insight required cryogenic crystal handling and highly brilliant synchrotron radiation. Hence, diffraction data of most of supramolecular assembly [G1@1] 2+ was collected at macromolecular synchrotron beamline P11, PETRA III, DESY. Disorder in ligand side chains, counterions and solvent molecules required carefully adapted macromolecular refinement protocols employing geometrical restraint dictionaries, similarity restraints and restraints for anisotropic displacement parameters (ADPs).
Geometry optimized models of structures were constructed using Wavefunction SPARTAN′18 and first optimized on semiempirical PM6 level of theory without constraints. Further, they were optimized on B3LYP/def2-SVP level, then the energy calculations were run at B3LYP/def2-TZVP.
The transient UV-Vis -pump-probe setup was described before. 1,2 Briefly, a a Ti:sapphire based oscillator/regenerative amplifier system (Solstice Ace, Spectra Physics) producing 35 fs laser pulses at 800 nm was used to create ~0.5 µJ pump pulses at 400 nm by 2nd harmonic generation in a BBO crystal. A probe white light continuum (WLC) was generated by focusing a small portion of the 800 nm beam (pulse energy 3 µJ) into a 4 mm CaF2 crystal. The pump-probe time delay was adjusted with a computer-controlled translation stage (M-415.DG, Physik Instrumente). For every laser shot about 50% of the WLC energy was used to record a reference spectrum. The other half was for probing pump pulse-induced changes in the spectrum by superimposing both beams at the center of the sample cell (UV quartz cuvette, optical path 2 mm, Starna). A synchronized chopper blocked every second pump pulse to determine difference spectra with and without the pump pulse. The relative plane of polarization of pump and probe light was adjusted to 54.7°. Both probe and reference spectra were recorded by individual spectrographs each equipped with a 256-element linear diode array. The covered spectral range was 350-730 nm. All measurements were performed with stirred DMSO solutions at ligand concentrations of 0.35 mM. The data were corrected for shifts of the baseline (determined at negative pump probe delays) and wavelength-dependent temporal shifts due to group delay differences.

Experimental procedures
Where necessary, experiments were performed under nitrogen atmosphere using standard Schlenk techniques. Chemicals and standard solvents were purchased from Sigma Aldrich, Acros Organics, Carl Roth, TCI Europe, VWR, ABCR and used as received, if not mentioned differently. Dry solvents were purchased or purified and dried over absorbent-filled columns on a GS-Systems solvent purification system (SPS). Reactions were monitored with thin layer chromatography (TLC) using silica coated aluminium plates (Merck, silica 60, fluorescence indicator F254, thickness 0.25 mm). For column chromatography, silica (Merck, silica 60, 0.02-0.063 mesh ASTM) was used as the stationary phase.

Host-guest supramolecular donor-acceptor systems:
Figure S11: Schematic representation of guest encapsulation.
[G1@1] 2+ was formed in quantitative yield by adding G1 (0.35 μmol, 35 μL of 10 mm in DMSO-d6, 1 equiv.) to a solution of cage 1 (0.35 μmol, 500 μL of a 0.70 mM solution in DMSO-d6, 1 equiv.) at room temperature for 30 min to give a solution of G1@1. Although G2 was added in similar way, only peak broadening without any change in the peak position was observed. This is might be due to larger agglomerates formed by the positively charged cage and negatively charged sulfonate anion G2 that is obviously too large to fit inside the cavity. [

Binding constant
Binding constants were calculated form online bindfit software 4 using 1 H NMR titration experiments data. Figure S21: Binding constant calculation for G1 (1.5 equiv.) with cage 1 (1.0 equiv., 0.7 mM). Further increase in the concentration of guest led to form agglomeration, followed by precipitation. From the bindfit software, the obtained binding constant K is ~1.16*10 5 M⁻¹ (± 60000).

Crystal structure of L (sg6)
Yellow plate-shaped crystals of L (sg6) were grown by slow evaporation of a saturated solution of L in DMSO at room temperature. A single crystal in mother liquor was mounted onto a 0.1 mm nylon loop using NVH oil. Single crystal X-ray diffraction data was collected on a Bruker D8 venture equipped with an Incoatec microfocus source (Iμs 3.0) using Cukα radiation on a four axis κgoniometer, equipped with an Oxford Cryostream 800 and a Photon II detector. All data was integrated with SAINT V8.40A and a multi-scan absorption correction using SADABS-2016/2 was applied. The space group was determined using XPREP. 5,6 The structure was solved by intrinsic phasing/direct methods using SHELXT 7 and refined with SHELXL 8 for full-matrix least-squares routines on F 2 and ShelXle 9 as a graphical user interface. All non-hydrogen atoms were refined with anisotropic displacement parameters. The hydrogen atoms were refined isotropically on calculated positions using a riding model with their Uiso values constrained to 1.5 times the Ueq of their pivot atoms for terminal sp3 carbon atoms and 1.2 times for all other carbon atoms.
Crystallographic data (including structure factors) for the structures reported in this paper have been deposited with the Cambridge Crystallographic Data Centre. CCDC 2114030 contains the supplementary crystallographic data for this structure. Copies of the data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/structures. Figure 29. Crystal structure of ligand L. Anisotropic displacement parameters at 50% probability level.

Crystal structure of [G1@1] 2+ (sg88y_8)
Extremely thin plate-shaped single crystals of [G1@1] 2+ were grown by slow diffusion of 1,4 dioxane into solution of [G1@1] 2+ in DMSO over a period of four weeks. Single crystals of [G1@1] 2+ in mother liquor was pipetted onto a glass slide containing NVH oil. To avoid cracking of the crystal, the crystal was quickly mounted onto a 200 µm nylon loop and immediately flash cooled in liquid nitrogen. Crystals were stored at cryogenic temperature in dry shippers, in which they were safely transported to macromolecular beamline P11 at Petra III, 5 DESY, Germany. A wavelength of λ = 0.6888 Å was chosen using a liquid N2 cooled double crystal monochromator. Single crystal X-ray diffraction data was collected at 80(2) K on a single axis goniometer, equipped with an Oxford Cryostream 800 open flow cooling device and a Pilatus 6M fast detector. 3600 diffraction images were collected in a 360° φ sweep at a detector distance of 154 mm, 40,80% filter transmission, 0.1° step width and 0.1 seconds exposure time per image. Data integration and reduction were undertaken using XDS. 6 The structure was solved by intrinsic phasing/direct methods using SHELXT 7 and refined with SHELXL 8 using 22 cpu cores for full-matrix least-squares routines on F 2 and ShelXle 9 as a graphical user interface and the DSR program plugin was employed for modeling. 10 The asymmetric unit contains two supramolecular coordination cages, four G1 guest molecules as well as 14 DMSO solvents molecules. The subatomic resolution of 0.76 Å revealed disorder in four out of eight sulfonate groups of G1 guest molecules (residue class ASO), three of the hexyl chains at ligand (residue class PPH) and as well as one of the DMSO solvents molecules (residue class DMS).
In each case, disorder was modelled with two discrete positions refining their occupancy factor using a free variable and ensuring sensible geometry by employing stereochemical restraints. Despite reaching subatomic resolution of 0.76 Å, disorder and poor crystal quality required stereochemical restraints to be employed for ensuring a sensible geometry of the organic part of the structure.
Stereochemical restraints for the ligands (residue class PPH), guest (residue class ASO) and DMSO solvents molecules (residue class DMS) were generated by the GRADE program using the GRADE Web Server (http://grade.globalphasing.org) and applied in the refinement. A GRADE dictionary for SHELXL contains target values and standard deviations for 1,2-distances (DFIX) and 1,3-distances (DANG), as well as restraints for planar groups (FLAT). All displacements for non-hydrogen atoms were refined anisotropically. The refinement of ADP's for carbon, nitrogen and oxygen atoms was enabled by a combination of similarity restraints (SIMU) and rigid bond restraints (RIGU). 11 The contribution of the electron density from disordered counterions and solvent molecules, which could not be modeled with discrete atomic positions were handled using the SQUEEZE 12 routine in PLATON. 13

Experimental setup
For the spectroelectrochemical measurements a thin layer quartz glass spectroelectrochemical cell was used with an optical path length of 1.0 mm. Instead of a glassy carbon working electrode a Pt mesh electrode was applied. As counter electrode a Pt wire was used. As reference electrode the aforementioned Ag/AgNO3 electrode was used. All spectra were recorded at room temperature. An AvaLight Deuterium-Halogen light source (200 nm -1000 nm) was used for the UV/VIS measurements. The light was conducted via fiber optic cable (200 μm diameter) to the spectroelectrochemical cell and further to a DAD AVA-SPEC 2048 spectrometer. With every voltage step with a scan rate of 0.1 V/s, the potentiostat PGSTAT101 triggered the measurement of a full UV-Vis spectrum. The data was recorded and processed using the AVASOFT 7.5 software.     . The inset c) shows no influence of G3 on the time traces at the probe wavelength of 613 nm (note that the scaling of time axis changes at 1.0 ps). The reason for this is, that G3 only features a low binding constant of about 140 M -1 , not leading to the formation of significant amounts of the host-guest complex at these low concentrations while G1 binds stronger (K ≈ 900 M -1 ) and hence leads to population of the host-guest complex even at low concentrations, resulting in the observed charge-transfer phenomena discussed in the main text (compare Fig. S49).

Kinetic model
The reaction scheme of Figure 8 in the main text results in four coupled differential equations for the relative concentrations in the ground state G, the S1 state of photo-excited PTZ 1 , the LMCT state, and the HGCT state.  Table S4 are in fair agreement with the spectra shown in Figure S45, and Figure 6 of the main paper, respectively.