Fundamental Characterization, Photophysics and Photocatalysis of a Base Metal Iron(II)‐Cobalt(III) Dyad

Abstract A new base metal iron‐cobalt dyad has been obtained by connection between a heteroleptic tetra‐NHC iron(II) photosensitizer combining a 2,6‐bis[3‐(2,6‐diisopropylphenyl)imidazol‐2‐ylidene]pyridine with 2,6‐bis(3‐methyl‐imidazol‐2‐ylidene)‐4,4′‐bipyridine ligand, and a cobaloxime catalyst. This novel iron(II)‐cobalt(III) assembly has been extensively characterized by ground‐ and excited‐state methods like X‐ray crystallography, X‐ray absorption spectroscopy, (spectro‐)electrochemistry, and steady‐state and time‐resolved optical absorption spectroscopy, with a particular focus on the stability of the molecular assembly in solution and determination of the excited‐state landscape. NMR and UV/Vis spectroscopy reveal dissociation of the dyad in acetonitrile at concentrations below 1 mM and high photostability. Transient absorption spectroscopy after excitation into the metal‐to‐ligand charge transfer absorption band suggests a relaxation cascade originating from hot singlet and triplet MLCT states, leading to the population of the 3MLCT state that exhibits the longest lifetime. Finally, decay into the ground state involves a 3MC state. Attachment of cobaloxime to the iron photosensitizer increases the 3MLCT lifetime at the iron centre. Together with the directing effect of the linker, this potentially makes the dyad more active in photocatalytic proton reduction experiments than the analogous two‐component system, consisting of the iron photosensitizer and Co(dmgH)2(py)Cl. This work thus sheds new light on the functionality of base metal dyads, which are important for more efficient and sustainable future proton reduction systems.


XAS spectroscopy
XANES and EXAFS measurements. The EXAFS measurements were conducted at the P65 beamline, DESY, Hamburg. Energy selection was done with the use of the Si(111) Double Crystal Monochromator (DCM) and photon flux on the sample was approximately 10 11 ph/s. The experiment was performed in Total Fluorescence Mode with the use of the PIPS detector in 45° geometry. Energy calibration was set to the first inflection point in Fe foil XANES spectrum with an energy of 7112.1 eV. Samples were in form of powder pressed into boron nitride pellet and data collection was performed at room temperature. Data reduction, normalization, and EXAFS fitting were performed in the Demeter package. 1 Prior to the main analysis an EXAFS analysis of the Fe foil was conducted in order to obtain SO2 value which is listed in Table S5. The RBKG parameter for all compounds was set to 1.

XANES analysis and FEFF calculations
Fe K-edge XANES spectra of [Fe-BL] and [Fe-BL-Co] with respect to Fe 0 are presented in Figure S4a. The spectra of [Fe-BL] and [Fe-BL-Co] are very similar with only a slight redshift of the unsymmetrical first white line feature by around 1 eV. According to comparison with well-known references with highly defined oxidation states, the iron centers in [Fe-BL] and [Fe-BL-Co] are numerically determined to be in a +1.7 oxidation state in agreement with Fe II ( Figure S4c). This conclusion is supported by the single pre-peak feature around 7114 eV, together with a well-pronounced near-edge shoulder around 7120 eV that are well-known features in Fe II tetra-NHC complexes. 6 Both complexes show a well-pronounced double-peak white-line feature, which is known for similar complexes coordinated by a 4,4'-bipyridine bridging ligand motif. 7 The identical energy position and intensity of the pre-edge and near-edge is indicating only negligible changes in the electronic structure around the iron atom and ligand. FEFF calculations conducted for crystal structures of [Fe-BL] and [Fe-BL-Co] show, that the 7114 eV pre-edge peak arises due to quadrupole-like transitions allowed by unoccupied Fe 3d-states mixing with a very small fraction of N p-states. 8,9 Contrary to that, a pre-peak at 6-7 eV is an overlap of pre-edge peaks which raised due to strong mixing of Fe dDOS with p-states both from C and N (Figures S5). According to a standard electronic energy level distribution of a distorted octahedral symmetry of the Fe II center (formally 3d 6 ) the presence of two dDOS peaks in the pre-edge region indicates not fully occupied t 2g states and mostly empty e g states. This effect can be induced by changes in a degeneration scheme of the d-states by the significantly distorted ligand field. Moreover, the e g states are involved in bond creation as they strongly overlap with ligand p-states. There is also no significant difference when compared XAS spectra of [Fe-BL] and [Fe-BL-Co] in solid and solution which indicates marginal solvation effects in MeCN (cf. Figure S4b).
In [Fe-BL] 3d DOS there is a peak at -1 eV -it is from residual occupied 3d states. Its relative position to the first pre-edge peak is ~2.7 eV and can be interpreted as a value of bandgap of the compound. 10 Analogically for the [Fe-BL-Co], the bandgap was of 2.4 eV, as the residual occupied 3d peak is at Fermi level, while the centre of mass for the first pre peak is at 2.4 eV. The small differences in white line shape can be explained by a slightly different contribution of ligand p-states to Fe pDOS function as well as a subsequent change of empty Fe p-states.  Figure S6 and the extracted structural parameters are given in Table S4. Corresponding bond lengths are included in Table 1. Crystal structure analysis showed that both [Fe-BL] and [Fe-BL-Co] are characterized by similar Fe-ligand bond length and angles in the solid. The spatial resolution of EXAFS is limited to approximately 0.01 Å, 11 therefore it serves as an indicator of significant geometric differences between both studied structures. All observed values (coordination number and atomic positions) are equal in the range of uncertainties for both complexes, thus no significant difference is observed in the coordination sphere of both Fe centres. It is worth to mention, that all fitted values, which include atomic distances, are averaged over 4 C and 2 N atoms (cf. Table 1). EXAFSderived values of Fe-N bond lengths are however different from values obtained in the XRD experiment by a factor of 0.1 Å. On the other hand, the difference between EXAFS Fe-C distance and XRD values is by a factor of 0.01 Å. These discrepancies can be accounted to the differences between the crystal structure environment and powder environment as well as to the lower EXAFS spatial resolution.   1.945 (8) 2.112 (24) 2.363 (7) 2.879 (11) 3.076 (21) 3.602 (15) 3.670 (1) 3.975 (11) 4.118 (9) 4.382 (27) 0.0013 (2) 0.0019 (4) 0.0010 (4) 0.0009 (3) 0.0019 (6) 0.0029 (8) 0.0010 (4) 0.0027 (6) 0.0037 (8) 0.0027 (6) [   (3) 0.0015 (4) 0.0008 (5) 0.0007 (4) 0.0016 (7) 0.0023 (9) 0.0008 (5) 0.0022 (6) 0.0030 (8) 0.0022 (6) a) Adapted from ref. 12

3.2
Dyad association study    Figure S10, A and B) which can be assigned to a terminal pyridine ring due to its characteristic coupling pattern and are comparable to H A and H B in [Fe-BL-Co]. Furthermore, an additional singlet proton signal is detected at δ = 7.09 ppm similar to H C which has the only singlet proton signal in the aromatic region of the iron part. Most likely, the crucial pyridine-cobalt bond is destabilized or a twist in the 4,4'-bipyridine moiety occurs by irradiation causing shifts of characteristic proton signals H A -H C to A-C.      Photosensitizer fraction at a certain concentration was deduced from NMR dissociation study. Grey, vertical reference line at 510 nm helps to see that after photosensitizer subtraction the band maximum is around 510 nm for [Fe-BL-Co] instead of 486 nm for each concentration.

Analysis of [Fe-BL]
A glassy carbon electrode (2mm diameter) was used to measure cyclic and square-wave voltammograms.        scan rate v 0.5 [(mV/s) 0.5 ] Figure S22: Linear dependence of peak current I pa (blue) and I pc (orange) versus square-root of scan rate v for quasireversible redox processes at E 0 1/2 = -1.46 V (linear fits in corresponding color).  Table S12: XYZ files of all optimized structures.

Optical transient absorption spectroscopy
Difference spectra of oxidized species ΔA ox are used for comparison with DAS of transient optical absorption spectroscopy. Due to lower concentrations in the spectroelectrochemical measurements (c = 0.1 mM) dyad dissociation was taken into (cf. NMR dissociation study). This led to modified difference spectrum of oxidized [Fe-BL-Co] suitable for TA discussion ( Figure S26, red curve). Standard fitting procedure with the convolution of the instrument response function (IRF) was applied using the multi-exponential function given below:         Aqueous TEOA solution (10 %) was adjusted with concentrated HCl to pH = 7 and mixed 1:1 with MeCN. Solvent mixture was freeze-pumped to remove all gases. The reactor was prepared by evacuating and filling with argon for five times. All solid compounds were degassed in one sealed vial with argon. In this vial 20 ml of solvent mixture were added and the complete solution was transferred to the reactor. The solution was stirred for 5 minutes to allow the solution to equilibrate. While the light source was turned on the measurement was started simultaneously. Solvent mixture measurements were used as blind curves to exclude any solvent effects and subtracted from each measurement. Then any influence of temperature changes during the experiment was removed by a temperature correction approach. 13 Finally, the measurements were averaged.

Photocatalytic proton reduction
Proof of hydrogen was conducted by headspace analysis via gas chromatography (micro-GC, cf. Figure  S33 and S34). First broad peak belongs to a pressure burst caused by the GC apparatus. Second signal is assigned to hydrogen, third signal to oxygen and fourth to nitrogen. Oxygen and nitrogen originated from the experimental setup. Connection to the GC has to be switched between two proton reduction apparatus leading to O 2 and N 2 from the air. Produced hydrogen was quantified by the automatic gas burette connected to the proton reduction apparatus. 13