Ferrous and ferric complexes with cyclometalating N-heterocyclic carbene ligands: a case of dual emission revisited

Iron N-heterocyclic carbene (FeNHC) complexes with long-lived charge transfer states are emerging as a promising class of photoactive materials. We have synthesized [FeII(ImP)2] (ImP = bis(2,6-bis(3-methylimidazol-2-ylidene-1-yl)phenylene)) that combines carbene ligands with cyclometalation for additionally improved ligand field strength. The 9 ps lifetime of its 3MLCT (metal-to-ligand charge transfer) state however reveals no benefit from cyclometalation compared to Fe(ii) complexes with NHC/pyridine or pure NHC ligand sets. In acetonitrile solution, the Fe(ii) complex forms a photoproduct that features emission characteristics (450 nm, 5.1 ns) that were previously attributed to a higher (2MLCT) state of its Fe(iii) analogue [FeIII(ImP)2]+, which led to a claim of dual (MLCT and LMCT) emission. Revisiting the photophysics of [FeIII(ImP)2]+, we confirmed however that higher (2MLCT) states of [FeIII(ImP)2]+ are short-lived (<10 ps) and therefore, in contrast to the previous interpretation, cannot give rise to emission on the nanosecond timescale. Accordingly, pristine [FeIII(ImP)2]+ prepared by us only shows red emission from its lower 2LMCT state (740 nm, 240 ps). The long-lived, higher energy emission previously reported for [FeIII(ImP)2]+ is instead attributed to an impurity, most probably a photoproduct of the Fe(ii) precursor. The previously reported emission quenching on the nanosecond time scale hence does not support any excited state reactivity of [FeIII(ImP)2]+ itself.


Methods
Single crystal X-ray diffraction analysis All SC-XRD measurements were performed using graphite-monochromatized Mo Kα radiation (λ = 0.71073 Å) using the Agilent Xcalibur Sapphire3 diffractometer high-brilliance IμS radiation source. Data collections was performed at 150 K for [Fe III (ImP)2]PF6 and [Fe II (ImP)2]. Absorption was corrected using multi-scan empirical absorption correction with spherical harmonics as implemented in the SCALE3 ABSPACK scaling algorithm. 1 The structure was solved by direct methods and refined by fullmatrix least-squares techniques against F2 using all data (SHELXT, SHELXS). 2,3 All non-hydrogen atoms were refined with anisotropic displacement parameters if not stated otherwise. Hydrogen atoms were constrained in geometric positions to their parent atoms using OLEX2 software. 4 The crystallographic data for [Fe III (ImP)2]PF6 and [Fe II (ImP)2] have also been deposited at the Cambridge Crystallographic Data Centre, under deposition numbers 2254083 and 2254082 respectively. Copies of these data can be obtained free of charge from www.ccdc.cam.ac.uk/structures. Determination of magnetic susceptibility using Evans' NMR method 5,6 Solution state magnetic moments were determined by Evans' method using an NMR tube containing a solution of the paramagnetic complex and a capillary containing pure CD3CN. χM = 3000 Δδ / (4π c) μeff = (7.9933 χM T) ½ χM is the molar susceptibility, Δδ is the change in chemical shift between acetonitrile in the capillary and the NMR sample and c is the concentration in mol/L. μeff is the effective magnetic moment, and T is the temperature in K.

Magnetic susceptibility and magnetization measurements
The magnetic data were acquired on a Quantum-Design MPMS-XL SQUID magnetometer. Susceptibility data were acquired in a static field of 1.0 KOe. Magnetization data were obtained with selected fields from 1 to 50 KOe at T = 2 -10 K in 1 K intervals. The polycrystalline samples were measured on a compacted powder sample in a polycarbonate capsule. Data were corrected empirically for TIP and the diamagnetic contribution to the sample moment from the sample holder and sample was corrected through background measurements and Pascal constants, respectively.

Mößbauer spectroscopy
Mößbauer measurements of [Fe III (ImP)2]PF6 were carried out in an Oxford Instrument flow cryostat at 85 K and at 295 K using a 57 CoRh source held at room temperature. The studied powder material was mixed with inert BN, pressed, and formed as pastille absorber with a concentration of about 30 mg/cm 2 S5 of studied substances. Calibration spectra were recorded from a natural iron metal foil held at 295 K. The resulting spectra were analyzed using a least square Mößbauer fitting program. Mößbauer measurements of [Fe II (ImP)2] were unsuccessful because of the rapid oxidation to [Fe III (ImP)2] + in air.

Cyclic Voltammetry and Spectroelectrochemistry
All electrochemical experiments were performed in acetonitrile (spectroscopic grade Uvasol ® , ≥99.9%, Merck; dried over 3Å molecular sieves activated at 300 °C for 15 hours) with 0.1 M tetrabutylammonium hexafluorophosphate (electrochemical grade, Sigma Aldrich; dried at 80 °C under vacuum) and purged with solvent-saturated argon. The samples were prepared with a concentration of 1 mM.
Cyclic voltammetry (CV) measurements were carried out in a three-electrode electrochemical cell, using an AUTOLAB potentiostat (PGSTAT302) controlled with GPES software (Version 4.9). The working electrode was a freshly-polished (with Buehler alumina paste) glassy carbon electrode (CH Instruments, 1 mm diameter); the reference electrode was a non-aqueous Ag/AgNO3 (CH Instruments; 10 mM of AgNO3 dissolved in dried acetonitrile; 0.08 V vs. ferrocene) and a Pt wire in a separate compartment was used as counter electrode.
UV-Vis spectroelectrochemistry measurements were performed in a diode array spectrophotometer (Agilent 8453) with an optically transparent thin-layer cell (1 mm optical path length) equipped with a platinum mesh working electrode and the same reference and counter electrodes used for voltammetry. Time-resolved spectra were recorded during controlled potential electrolysis using an AUTOLAB potentiostat (PGSTAT302).

Steady-State Spectroscopy measurements
Samples were prepared in either acetonitrile or tetrahydrofuran (both spectroscopic grade Uvasol ® , ≥99.9%, Merck). Samples of [Fe III (ImP)2] + were prepared under standard conditions, while those for [Fe II (ImP)2] were prepared with deaerated solvents in an argon-filled glove box due to its high oxygen sensitivity. UV-Vis absorption spectra were recorded on Varian Cary 50 or Cary 5000 spectrophotometers.
Steady-state emission and excitation measurements were performed on an FS5 Spectrofluorometer (Edinburgh Instruments) with 5 nm spectral resolution. Samples were prepared in 10 mm cuvettes and measurements were performed with right-angle geometry. Emission and excitation spectra were all background subtracted and corrected for detector response.
Time-Correlated Single Photon Counting measurements were completed with the same instrument with picosecond pulsed diode lasers (EPL Series) with excitation wavelengths of 340 and 375 nm. The fluorescence lifetime measurements were fitted with the instrument response function (IRF) using the in-built reconvolution fitting in the Fluoracle ® software.

Illumination measurements
The photoreaction of [Fe II (ImP)2] was studied by absorption and emission spectroscopy. Initial UV-Vis absorption and emission spectra of an oxygen free solution of [Fe II (ImP)2] in a gas tight cuvette were taken, with as little exposure to light in between as possible. The cuvette was then exposed to (i) ambient light and (ii) a white light LED lamp (ZENARO, SL-PAR38/B/P17/20/E50/TD/27/HAC, 50-60 Hz, ~ 50 mWcm -2 ). UV-Vis absorption and emission spectra were collected immediately after illumination (Shown in Fig. 3b).
To track the photodecomposition over time, solutions of [Fe II (ImP)2] were prepared in a glovebox under nitrogen and transferred to an air-tight quartz cuvette (10 mm, Hellma Analytics QS). The cuvette was irradiated using blue LEDs (λ = 450 nm), and the absorption spectrum was measured at regular time intervals using an Agilent Cary 60 spectrometer (Shown in Fig. 5).

Femtosecond transient absorption measurements
Femtosecond transient absorption spectroscopy (fs-TAS) measurements were performed probing in the UV-Vis region on a Newport TAS system with a Coherent Libra Ti:sapphire amplifier (800 nm, 1.5 mJ, 3 kHz repetition rate, FWHM 45 fs). Different excitation wavelengths were generated by optical parametric amplifiers (TOPAS-Prime and NIRUVVIS, Light Conversion) and then focused and centered on the 1 mm cuvette with corresponding pump powers. The white light supercontinuum probe light was generated using a CaF2 crystal (Crystran) and was detected by a silicon diode array (Newport custom made). A mechanical chopper blocked every other pump pulse, and the transient absorption at each time point was calculated for an average of 1000 ms chopped/un-chopped pulse pairs. To record the transient absorption spectra at different time points, an optical delay line was used to scan the delay of the probe beam relative to the pump beam (maximum -5 ps to 8 ns). A total of 8 scans were collected and averaged for each sample. Prior to analysis, the data was corrected for the spectral chirp using Surface Xplorer v4, where single wavelength fits were also performed.

Global Analysis
All fs-TAS data were further analyzed with Global Analysis using the software Glotaran (Version 1. 5. 1), which is a Java-based graphical user interface to the R package TIMP. 7 All data were fitted with necessary number of components and the decay-associated spectra (DAS) were exported from the software and analyzed in Origin.

Computational Details
Density functional theory (DFT) calculations utilizing the B3LYP functional 8 with Grimme's D2 dispersion correction 9 were employed to optimize structure of the singlet ground state ( 1 GS), as well S7 as the triplet and quintet metal-centered states ( 3,5 MC) and triplet metal-to-ligand charge transfer ( 3 MLCT) states of [Fe II (ImP)2] 0 complex, and the doublet ( 2 GS), quartet ( 4 MC) and sextet ( 6 MC) states of [Fe III (ImP)2] + complex. The 6-311G* basis set was used for all atoms (H, C, N) 10, 11 except for Fe, where the SDD basis sets and its accompanying pseudopotential 12 were employed in all calculations. Solvent effects (acetonitrile) were included in all the calculations via the polarizable continuum model (PCM). 13 Vibrational frequency analysis was performed to ensure that the optimized structures correspond to minima on their respective potential energy surfaces. Fragment molecular orbital analysis (FMOA) based on the Mulliken population analysis was carried out with the AOMix software 14,15 in order to obtain the percent contributions of each fragment. A two-fragment scheme was employed: Fragment 1 (Fe), Fragment 2 (ligands). All calculations were carried out using the Gaussian 16, Revision A.03 software package. 16 Time-dependent DFT (TD-DFT) 17 calculations at the same level of theory as the structure optimizations were employed to simulate the UV-Visible absorption spectra in acetonitrile. The stick spectra were broadened using Lorentzian functions with a half-width-at-half-maximum (HWHM) of 0.12 eV. The hole and electron density pairs of each transition responsible for the absorption spectra were characterized by means of natural transition orbitals (NTO). 18 Potential energy curves (PECs) versus the average Fe-C bonds were constructed for the various spin states of Fe(II) and Fe(III) complexes to understand the excited states dynamics. DFT single point energy calculations were carried out on the initial fully optimized minima, to obtain their corresponding 2 GS, 4,6 MC states for Fe(III), and 1 GS, 3,5 MC states for Fe(II). Tamm-Dancoff approximation (TDA) 19 was then employed to calculate the vertical excitations on the produced reference wavefunctions: 2 GS for Fe(III) and 1 GS for Fe(II), in order to obtain the first 2 LMCT and 2 MLCT states in Fe(III), and the first 1 MLCT, 3 MC, and 3 MLCT states in Fe(II). TDA rather than TD-DFT was used for these calculations due to the triplet instabilities in TD-DFT calculations that occurred at 3,5 MC geometries. 20 PECs from a relaxed potential energy surface scan along the partially detached Fe-C bond (shown in red in Fig. S29) were also carried out to compare the reaction pathway of the ligand detachment for the 1 GS and 3 MC states of the van der Waals solvent coordinated complex [Fe(ImP)2(CH3CN)]. The Fe-C bonds were constrained to values between 4.6 Å and 2.0 Å with 0.1 Å step, starting from the conformation in which the Fe-C bonds are set to 4.6 Å.

Synthesis
All reactions were carried out using oven-dried glassware under an atmosphere of nitrogen. All solvents for synthesis were used as received and were of synthesis grade, unless otherwise stated. Tetrahydrofuran (THF) was dried over Na/benzophenone and subsequently distilled under nitrogen before use. Acetonitrile was dried over molecular sieves and degassed by freeze-pump-thaw cycles. Anhydrous dichloromethane was obtained from a Braun SPS-800 system. Reagents were obtained from commercially available sources and used as received unless stated otherwise. Commercially available starting materials were purchased from Acros, Merck, or Fischer Scientific. [Bis(2,6-bis(3-methylimidazol-2-ylidene-1-yl))phenylene)iron(III)] hexafluorophosphate, [Fe III (ImP) 2 ]PF 6 [1,1'-(1,3-Phenylene)bis(3-methyl-1-imidazolium)] dibromide (803 mg, 2.01 mmol) and tetrakis(dimethylamido)zirconium (642 mg, 2.40 mmol) were charged into a Schlenk flask in a glovebox under nitrogen. Dry THF (10 mL) was added to the Schlenk flask, and the yellow suspension was stirred for 2 h under nitrogen at room temperature. Iron(II)bromide (240 mg, 1.10 mmol) was sonicated in dry THF (24 mL) under nitrogen to give a brown solution that was added in one portion to the reaction. The reaction turned into a brown suspension which was stirred in the dark under nitrogen for another 16 h. The brown suspension was then subjected to air, upon which it immediately turned blue. To the suspension was added methanol (1 mL) and the reaction mixture was stirred for another 3 h, after S9 which it was filtered through a glass frit filter (porosity 3). The filter was washed with acetonitrile until the filtrate was colourless, and the combined filtrates were concentrated in vacuo to give a dark blue solid. The solid was redissolved in methanol (30 mL), filtered and reprecipitated by pouring it into an aqueous solution of ammonium hexafluorophosphate (855 mg in 50 mL water). The dark blue precipitate was filtered off, redissolved in dichloromethane and reprecipitated by the addition of diethyl ether. The resulting dark blue product was purified by size-exclusion chromatography on BioBeads S-X1 (3 × 120 cm) using acetonitrile/toluene (50/50) as eluent. The resulting product was recrystallized from acetonitrile/diethyl ether to give [Fe III (ImP)2]PF6 as dark blue crystals (171 mg, 26%). [Bis(2,6-bis(3-methylimidazol-2-ylidene-1-yl)phenylene)iron(II)], [Fe II (ImP) 2 ] To a suspension of [bis(2,6-bis(3-methylimidazol-1-ylidene)phenyl)iron(III)] hexafluorophosphate (82.9 mg, 0.123 mmol) in dry THF (5 mL) was added lithium aluminium hydride (1 M in THF, 0.6 mL, 5 equiv.) under nitrogen at room temperature. The dark blue suspension turned into a clear orange solution. The solvent was evaporated with a stream of nitrogen gas, and the resulting orange solid was redissolved in dichloromethane (8 mL), filtered through a Schlenk filter (porosity 3) and the dichloromethane was evaporated with a stream of nitrogen gas to give [Fe II (ImP)2] as an orange solid (35 mg, 54%).

S10
NMR spectra NMR spectra were recorded at ambient temperature on a BrukerAvance II 400 MHz NMR spectrometer (400/101 MHz 1 H/ 13 C) or a Bruker Avance Neo 600 MHz spectrometer (600/151 MHz 1 H/ 13 C), equipped with a QCI CryoProbe. The initial 13 C-spectra of [Fe III (ImP)2]PF6 showed splitting of the peaks at 474 and -202 ppm. However, the splitting disappeared when the carrier frequency for the 1 H-decoupling was shifted to get proper decoupling. The 2D 13 C-HSQC experiment of [Fe III (ImP)2]PF6 was recorded in subsections, using different carrier frequencies for 1 H and 13 C in order to get proper excitation profile and 13 C-decoupling. Chemical shifts (d) for 1 H and 13 C NMR spectra are reported in parts per million (ppm), relative to the residual solvent peak of the respective NMR solvent: CD3CN (δH = 1.94 and δC = 118.26 ppm), DMSO-d6 (δH = 2.50 and δC = 39.52 ppm), and THF-d8 (δH = 3.58 and δC = 67.21 ppm). 22 Coupling constants (J) are given in Hertz (Hz), with the multiplicities being denoted as follows: singlet (s), doublet (d), triplet (t), quartet(q), quintet(qi), multiplet (m), broad (br). NMR spectra for 13 C were recorded with decoupling of 1 H. For the assignment of spectra, see below.      HRMS Electrospray ionization-high resolution mass spectrometry (ESI-HRMS) was recorded on a Waters Micromass Q-Tof micro mass spectrometer. HRMS using MALDI-TOF and ESI-TOF was attempted for [Fe II (ImP)2], but we were only able to obtain the mass of the oxidized species [Fe III (ImP)2] + . Elemental analyses were performed by Mikroanalytisches Laboratorium KOLBE (Mülheim an der Ruhr, Germany).

Compound Bond lengths (Å) Bond angles (°)
[Fe II (ImP)2] Fe(1)-C(6)#1: 1.938 (8) Fe (1) Table S5. Results of the fitting procedure of the 295 K and 85 K Mößbauer spectra. CS is the center shift relative natural Fe held at 295 K, |QS| is the magnitude of the electric quadrupole splitting , + is the Lorentzian line width for the high velocity peak and -/ + is the ratio between the low and high velocity peak, respectively.

Complex
Temperature     The differential pulse and cyclic voltammograms are in good agreement with the recently reported electrochemistry measurements. 23 The report also mentions a suspected irreversible reduction of the ligand at around -2.7 V, which is however not observed in our measurements. Exhaustive, controlled potential electrolysis resulted in the clean conversion of [Fe III (ImP)2] + to its Fe(II) and Fe(IV) oxidation state, respectively. The resulting absorption spectra differ significantly from the reported spectra. 23

Electronic absorption spectrum of [Fe III (ImP) 2 ] +
The UV-Vis absorption spectra of the Fe(III) complex is shown in Fig. S19. The complex displays three main absorption bands. The band at higher energy (< 270 nm) is assigned to ligand-centered (LC) -* transition, while the lowest energy band (370-425 nm) corresponds to a ligand-to-metal charge transfer (LMCT). The final band with a shoulder between 350 and 280 nm comprises of transitions with mixed states: metal-to-ligand CT (MLCT), metal-centered (MC), LC, LMCT. The intense peak around 284 nm (f=0.049) shows a transition of ~ 50% MLCT character with LC and LMCT contributions (see Fig S19).  Table S8), the doublet ground state ( 2 GS) is the most stable electronic state at all conformations investigated. The metal-centered states, 4,6 MC, are calculated to be high in energy at the optimized doublet geometry, with the relaxed minima states displaced to much longer Fe-C bond lengths. This is indicative of the strong σ-donating capability of the ligand, hence the destabilization of the metal-centered states. Excitation of the 2 GS leads to excited states with transitions characterized as 2 LMCT and 2 MLCT' ( 2 MLCT) states. Deactivation from the 2 LMCT state leads back to the 2 GS leads to a radiative decay since an ISC into 4 MC is spin-forbidden. However, upon excitation into the higher 2 MLCT states various decay pathways are likely to occur: (1) 2 MLCT → 2 LMCT → 2 GS or (2) 2 MLCT → 2 MLCT' → 2 GS which were observed in the experiment.   S21). The computed average Fe-C bond distances display an increase from 1.96 Å in the 1 GS to 2.20 Å in the 3 MC state and 2.30 Å in the 5 MC state. The 3 MC state shows a deviation of 0.23 Å from the 1 GS with a greater deviation observed from three Fe-C bonds, in which one of the Fe-C (carbene) bonds pops out (see Fig. S22 and Table S6). The 5 MC also shows a very substantial deviation from the 1 GS with an average of 0.34 Å extension of Fe-C bond: all the six Fe-C bonds are elongated. This suggest that the 5 MC geometry is significantly different from the 1 GS, thus, the 5 MC state is further away and inaccessible for deactivation.  S47 grey). Knowledge about the ground state frontier orbitals may provide insights into the type of transitions for the excited states. In this case, the first three HOMOs being metal t2g based and the LUMOs (LUMO to LUMO+4) being ligand π* orbitals suggest that the lowest energy transitions would be metal-to-ligand charge transfer (MLCT), which are of interest for photophysical applications.

Electronic absorption spectrum of [Fe II (ImP) 2 ]
The simulated absorption spectrum is similar to the experimentally observed spectrum of the Fe(II) complex. For the Fe(II) complex, the calculated absorption shows the lowest absorption band consists of almost degenerate energy levels at 428 nm (f=0.0197) and 424 nm (f=0.0182) whose transitions originates from HOMO to LUMO+1 (83%) and HOMO-1 to LUMO+1 (75%) respectively. The assignment of these transitions, HOMO (and HOMO-1) to LUMO+1, is metal-to-ligand charge transfer (MLCT). The strongest absorption peak which occurs at 345 nm (f=0.211) is also assigned to a MLCT transition. This transition comprises of contributions from HOMO-1 to LUMO+4 (41%), HOMO to LUMO+3 (27%), and HOMO-2 to LUMO+1/ LUMO+4 (14%). Several MLCT transitions with significant absorptivity are also observed. The presence of several low-lying MLCT states is consistent with the MO energy level diagram, which in general can be used as an approximation to describe the type of transitions in the excited states.    From the electronic structure calculations, we found that the ligand detaches in the 3 MC of the Fe(II) complex as shown in Fig. S22, which led us to explore whether the solvent (CH3CN) can bind to the complex at the various spin states ( 1 GS, 3    A relaxed potential energy curve (PEC) scan along the Fe-C bond (the carbon bond of the partially detached ligand) was carried out to describe the reaction pathway of the ligand detachment for the 1 GS and 3 MC states of the van der Waals solvent coordinated complex, [Fe II (ImP)2(CH3CN)]. For each spin state, a scan was performed along the Fe-C bond length from 4.6 to 2.0 Å in 0.10 Å decrements (see Fig. S13-15). From the PEC, we observe that in the ground state, the thermodynamically favorable structure occurs at Fe-C bond length of 1.9 Å which is ~ 23.0 kcal/mol favorable compared to the detached ligand. In contrast, for the 3 MC state, we observe a very small change in energy along the reaction coordinate, in which the thermodynamically favorable structure occurs when the ligand is detached: Fe-C bond length of 4.6 Å. These data confirm that for the Fe(II) complex, ligand detachment is likely to occur in the metal-centered excited states.  Potential energy curves of [Fe II (ImP) 2

(CH 3 CN)]
A similar potential energy curve as described in Fig. S27 was carried out for the van der Waals coordinated complex [Fe II (ImP)2(CH3CN)]. Overall, the ordering of the electronic states is very similar to those of the non-coordinated complex, [Fe II (ImP)2] (see Fig. S27).