Photophysical Integrity of the Iron(III) Scorpionate Framework in Iron(III)–NHC Complexes with Long-Lived 2LMCT Excited States

Fe(III) complexes with N-heterocyclic carbene (NHC) ligands belong to the rare examples of Earth-abundant transition metal complexes with long-lived luminescent charge-transfer excited states that enable applications as photosensitizers for charge separation reactions. We report three new hexa-NHC complexes of this class: [Fe(brphtmeimb)2]PF6 (brphtmeimb = [(4-bromophenyl)tris(3-methylimidazol-2-ylidene)borate]–, [Fe(meophtmeimb)2]PF6 (meophtmeimb = [(4-methoxyphenyl)tris(3-methylimidazol-2-ylidene)borate]–, and [Fe(coohphtmeimb)2]PF6 (coohphtmeimb = [(4-carboxyphenyl)tris(3-methylimidazol-2-ylidene)borate]–. These were derived from the parent complex [Fe(phtmeimb)2]PF6 (phtmeimb = [phenyltris(3-methylimidazol-2-ylidene)borate]– by modification with electron-withdrawing and electron-donating substituents, respectively, at the 4-phenyl position of the ligand framework. All three Fe(III) hexa-NHC complexes were characterized by NMR spectroscopy, high-resolution mass spectroscopy, elemental analysis, single crystal X-ray diffraction analysis, electrochemistry, Mößbauer spectroscopy, electronic spectroscopy, magnetic susceptibility measurements, and quantum chemical calculations. Their ligand-to-metal charge-transfer (2LMCT) excited states feature nanosecond lifetimes (1.6–1.7 ns) and sizable emission quantum yields (1.7–1.9%) through spin-allowed transition to the doublet ground state (2GS), completely in line with the parent complex [Fe(phtmeimb)2]PF6 (2.0 ns and 2.1%). The integrity of the favorable excited state characteristics upon substitution of the ligand framework demonstrates the robustness of the scorpionate motif that tolerates modifications in the 4-phenyl position for applications such as the attachment in molecular or hybrid assemblies.


■ INTRODUCTION
The development of photosensitizers based on Earthabundant, inexpensive, and nontoxic metals, with the goal of replacing the to-date widely used noble metals, has attracted a lot of interest in the field of coordination chemistry in recent years. 1 Such research is motivated by the desire to use solarenergy conversion processes on a large scale. Until recently, the field of solar energy conversion based on coordination compounds has to a large degree focused on octahedral metal complexes of noble metals with low-spin 4d 6 or 5d 6 electronic configurations using different second and third row transition metals (TMs) including Ru(II), Re(I), Os(II), and Ir(III). 2,3 The ligand field splitting in such transition metal complexes is inherently larger than that for the corresponding complexes containing first row TMs such as Cr, Mn, Fe, and Co. This shifts the metal-centered (MC) states to higher energies in the former case, which in turn results in slow deactivation of the photoactive charge-transfer (CT) states. 4 Together with the employment of π-accepting ligands such as 2,2-bipyridyl (bpy), in metal complexes involving 4d 6 or 5d 6 metal cations, this has led to the development of many metal complexes that have metal-to-ligand charge-transfer (MLCT) states lower than MC states in energy. 5,6 This implies that, for the MLCT state, a state with demonstrated importance for photofunctional applications for metal complexes based on 4d 6 and 5d 6 metal cations, the deactivation of the excited state via the MC states is a concern only at elevated temperatures. 5 Efficient intersystem crossing from 1 MLCT states usually populates 3 MLCT states that exhibit slow radiative and nonradiative relaxation to the ground state. Additionally, 4d 6 and 5d 6 metal complexes with π accepting ligands display a relatively wide visible light absorption window and favorable redox properties of the GS and the 3 MLCT state. As a result, they are heavily featured in photophysical applications. 2 In parallel, four-coordinate 5d 8 complexes involving Pt(II) and Au(III) have also been investigated and successfully used in photophysical applications thanks to the strong ligand field connected to these third-row transition metals. 5,7 There have been some reports about photoactive Earthabundant metal complexes, foremost from metal complexes containing TMs such as Cu(I), Cr(0), Mn(I/IV), and Co(III). 5 The problem with first row transition metal complexes for photophysical applications in general is that the weak ligand field results in their MC states being relatively low in energy, providing a fast deactivation pathway and reducing the efficiency of the photofunctional MLCT states. 4 Of the first row transition metals, iron is by far the most abundant. 8 For Fe(II) polypyridyl complexes, the most widely studied direct base metal analogues of the successful Ru-, Osand Ir-polypyridyl photosensitizers, the MLCT states are deactivated on the 100 fs time scale to the low-lying MC states. 6 McCusker and Heinze reported attempts to increase the MLCT excited state lifetime by employing ligands with increased bite angle and introducing π-accepting and/or push−pull moieties. 9,10 Recently, McCusker reported a cage compound involving the Fe(II)(bpy) 3 motif, exhibiting a 2.6 ps MLCT lifetime, the longest recorded to date for an iron polypyridine complex. 11 However, the introduction of strongly σ-donating N-heterocyclic carbene (NHC) ligands in the field of photoactive iron complexes 6,12 has significantly increased the excited state lifetime of Fe(II) MLCT states, reaching up to 528 ps. 13 By increasing the ligand field strength, the MC states increase in energy, thus slowing down the deactivation of the MLCT state. 6,14,15 The photophysical properties of Fe− NHC metal complexes have been further improved using different approaches. 16 23 This complex showed an LMCT excited-state lifetime of 2 ns and an intense fluorescence with a 2.1% quantum yield, 23 constituting the second example of room temperature photoluminescence from an iron complex. 24 Further, the LMCT excited state was oxidatively and reductively quenched in bimolecular reactions using standard electron donors and acceptors, which was the first example of such quenching involving an iron charge-transfer state being demonstrated. 23,25 Very recently, Therien synthesized an Fe− NHC−porphyrin conjugate that showed photoluminescence from a state with considerable MLCT contribution, 26 and Bauer found both MLCT and LMCT photoluminescence from photoexcited states of an Fe(III)−NHC-cyclometalated complex, as communicated in a preliminary report. 27 In fact, there are only three examples of iron complexes with a nanosecond excited CT state lifetime outside the class of NHC complexes, namely, the Fe(II) complexes reported by Herbert involving strongly electron-donating amide ligands 28 and the cyclometalated Fe(II) complex with a phenylphenanthroline framework reported by Berkefeld. 29 Given the few examples of complexes with iron-based photoluminescence and/or long-lived iron-based CT states, it is clearly a challenging task to generate new iron-based complexes possessing photophysical properties that allow for efficient applications. However, there is an interest in modifying existing, promising iron NHC complexes, as described by the increase in reported Fe−NHC complexes from 2013 to date. 14,15 Most new complexes are based on structural variations of the ligand framework of the first iron complex having a 3 MLCT lifetime above 1 ps, the Fe−NHC complex [Fe(pbmi) 2 ](PF 6 ) 2 (where pbmi = 1,1′-(pyridine-2,6diyl)bis(3-methylimidazol-2-ylidene)). 12 Here, we report the first series of modifications of [Fe(phtmeimb) 2 ]PF 6 with the purpose of finding possible structure−(photo)functional relationships based on the   (d). Thermal ellipsoids are shown at 50% probability. Hydrogen atoms, counterions, and solvent molecules are omitted for clarity. Fe = orange; B = purple; N = blue; C = gray; Br = brown; O = red.  6 suitable for scXRD analysis were grown from a saturated acetonitrile solution of the respective complex by slow diffusion of diethyl ether at room temperature. Purple single crystals of [Fe(coohphtmeimb) 2 ]PF 6 suitable for scXRD analysis were grown from a saturated acetonitrile:methanol  Figure S14, together with data for the parent [Fe(phtmeimb) 2 ]PF 6 complex. The magnetization (M) for all three systems shows the expected variation with the reduced field (BT −1 ) indicating a saturation not much above 1 Bohr magneton (μ β ). This agrees with the common S = 1 / 2 ground state for the expected low-spin electronic configuration (t 2g 5 ) imposed by the strong-field carbene ligands. For all three systems, the isothermally measured magnetization curves are superimposable in the (BT −1 ) plot, demonstrating the absence of zero field splitting as required for the low-spin (LS) t 2g 5 ground state. The temperature variations of the magnetic susceptibilities of the three systems are slightly different. In all three cases, the χT product shows a temperature dependence, reflecting an incompletely quenched orbital angular momentum. The near

Inorganic Chemistry
pubs.acs.org/IC Article orbital degeneracy of the t 2g orbitals expected for these closeto-octahedral structures will also make the electronic g-factor very sensitive to vibronic couplings. 34 This can contribute to the minor differences observed in the susceptibilities but, more importantly, will also provide a mechanism for very system dependent broadening of EPR signals. For temperature-coded data, see Supporting Information section S5 and Figures S15− S17.
The EPR spectra of [Fe(brphtmeimb) 2 ]PF 6 , [Fe-(meophtmeimb) 2 ]PF 6 , and [Fe(coohphtmeimb) 2 ]PF 6 do not show any EPR signal at X-band frequencies, in either perpendicular or parallel mode (for details, see Supporting Information section S6), similar to their parent complex [Fe(phtmeimb) 2 ]PF 6 . 23 The low-temperature (80 K) Moßbauer spectra of [Fe-(brphtmeimb) 2 ]PF 6 , [Fe(meophtmeimb) 2 ]PF 6 , and [Fe-(coohphtmeimb) 2 ]PF 6 reveal a quadrupole split doublet with almost the same center shift (CS) and magnitude of the electric quadrupole splitting (QS) indicating low-spin Fe(III), just as the parent [Fe(phtmeimb) 2 ]PF 6 complex at 80 K (Supporting Information Figure S18). 23 Furthermore, the 295 K spectrum of [Fe(meophtmeimb) 2 ]PF 6 also shows an extra doublet (spectral intensity 25(5)%) with CS and QS representative of low-spin Fe(II) (Supporting Information Figure S19). The Lorentzian line width Γ for the main Fe(III) component in the three spectra was only about 0.27(1) mm/s at 295 K, unveiling a narrow distribution in crystal environments in all samples. The spectra at low temperature show an asymmetric doublet structure with broad lines ( Figure S18). The fitting results are presented in Table S6 which also includes the result from [Fe(phtmeimb) 2 ]PF 6 . 23 The Fe(II) doublet seen in the room temperature spectrum for [Fe-(meophtmeimb) 2 ]PF 6 is hardly detected in the low-temperature (80 K) spectrum ( Figure S18). An analysis shows a maximum spectral intensity of less than 3% for this Fe(II) component at around 80 K in the [Fe(meophtmeimb) 2 ]PF 6 sample. This could be due to different Moßbauer recoil free factors (f) for Fe(III) and Fe(II) in this complex. The f-factors become more equal at lower temperature, which is why a maximum limit of the ratio Fe(II)/Fe(III) to less than 3% can be determined. One explanation to the origin of the Fe(II) impurity(ies) is given in the caption of Supporting Information Figure S19. The center shift and electric quadrupole splitting for all of the doublets above fall in the range of reported values for low-spin S = 1 / 2 , Fe(III) ions. 23,24 In Supporting Information Figure S20, areas of different Fe valencies are presented in a |QS| vs CS (at 80 K) diagram for other Fecarbenes. The asymmetries of the Fe(III) doublet found at 80 K can furthermore be explained on the basis of magnetic relaxation effects 35 and a negative sign of QS. The relaxation time of the magnetic moment (in fact, the Moßbauer effect detects the magnetic hyperfine field acting at the Fe nucleus) of the Fe(III) ion at 80 K is comparable to the observation time τ of the Moßbauer effect. The observation time τ corresponds to the mean lifetime of the nuclear excited level, in the case of 57 Fe spectroscopy to ∼70 ns. Magnetization, EPR, and Moßbauer results used to assign the spin state of the investigated compounds are summarized in Table 1 (Figure 3). No significant trend in the absorption maxima shift can be discerned with the given experimental accuracy. These only minor differences suggest that the para-substitution of the phenyl moiety has little impact on the excited state of [Fe(phtmeimb) 2 ]PF 6 . In fact, one can suspect that phenyl rings are not involved in the transitions in the lower energy manifold. This notion is corroborated by the electrochemical data (see below). Therefore, the transition peaking at around 500 nm is assigned to an 2 LMCT-band for   Figure 5 and Table 3) revealed two reversible one-electron waves that can attributed to the Fe(III/II) and Fe(IV/III) couples in analogy to the parent complex [Fe(phtmeimb) 2 ]-PF 6 . 23 In comparison to the latter, the potentials of the metalcentered couples show only very moderate shifts of about 30 mV toward higher potentials for the electron-withdrawing bromide and carboxylic acid substituents and similar shifts in the opposite direction for the electron-donating methoxy substituent. Analogous but more pronounced substituent effects were found for the potential of the first ligand oxidation that is shifted relative to the parent complex by +80 and −280 mV in [Fe(brphtmeimb) 2 ]PF 6 and [Fe(meophtmeimb) 2 ]PF 6 , respectively. Further reduction of the Fe(II) state was observed for [Fe(brphtmeimb) 2 ]PF 6 and [Fe(coohphtmeimb) 2 ]PF 6 . The peaks at −2.3 and −2.7 V, respectively, can be tentatively attributed to the reduction of the functionalized aryl moieties rather than the actual carbene ligands. The latter are not reduced within the available potential window in the case of the parent complex and [Fe(meophtmeimb) 2 ]PF 6 , and spectroscopic data confirms that the same situation also applies to [Fe(brphtmeimb) 2 ]PF 6 and [Fe(coohphtmeimb) 2 ]-PF 6 .
Spectroelectrochemistry data featuring the Fe(III) ground state together with the corresponding Fe(II) and Fe(IV) states obtained from controlled potential bulk electrolysis is shown in  Table 3. For the Fe(III) and Fe(IV) complexes, the energies of their lowest energy LMCT bands are within error margins indistinguishable from those of the parent complex (Table 3). 23 At a first glance, this appears to be at odds with the trends in electrochemical potentials, in particular the significantly lowered potential for ligand        Figures S24 and S25). The excited state decay can be accurately described by a single exponential model and a global fit 36 to the data resulting in a universal lifetime of ∼2 ns at all observed features and for all [Fe(phtmeimb) 2 ]PF 6 derivatives; see Table 4. These results are in good agreement with emission lifetimes (of 1.9 ns) determined by time-correlated single photon counting (TC-SPC, in Supporting Information section S10, Figure S27, and Table S7). The excited state lifetimes of all three derivatives   2 ] + congeners were calculated by using unrestricted density functional theory (DFT). For brevity and due to the close similarity between the different studied molecules, only the [Fe(brphtmeimb) 2 ] + energy profile has been plotted in Figure 10. The quantum chemical results reveal a doublet ground state ( 2 GS) and quartet ( 4 MC) and hextet states ( 6 MC) stable under phenyl group functionalization following the same energy trend as previously reported for [Fe(phtmeimb) 2 ] + . 23 The calculated spin density for the doublet ground state of all the investigated complexes is found to be mainly located on the metal and carbene lone pairs, as shown for [Fe(brphtmeimb) 2 ] + in Figure 10. Overall, the results from the quantum chemical calculations highlight the similarity of the electronic structure properties across the full series of complexes, including the lack of involvement of the phenyl-based moieties, which is consistent with the overall observed lack of electronic communication between the metal center and the side groups. The spin density on the metal in the relaxed quartet and hextet states indicates the same metal center nature of these states for the three iron carbene derivatives. All spin densities for [Fe(phtmeimb) 2 ] + and congeners are displayed in Supporting Information Table S8. The relaxed ground state geometries of the three iron complexes are in good agreement with the reported X-ray structures. The average iron−carbene distances are also reported in Supporting Information Tables S8−S12 for all complexes and suggest unremarkable structural changes due to addition to bromide, methoxy, or carboxylic groups in the 4position of the phenyl.  6 containing the scorpionate ligand [phtmeimb] − , the substitution of the latter in the 4-phenyl position with either −Br, −OMe, or −COOH substituents did not result in any significant changes of the ground state properties such as geometry and magnetic properties, however adding three new iron complexes with ns lifetime and visible photoluminescence to the existing very small library of such complexes. Electrochemistry and quantum chemistry calculations indicate weak electronic communication between the phenyl moiety of the scorpionate ligand and the iron center, leading to only marginal electrochemical shifts between the complexes. The essentially identical charge-transfer absorption bands of the three complexes in their Fe(II), Fe(III), and Fe(IV) states, further suggest that the spectroscopically relevant ligand orbitals do not extend over the phenyl moieties. Importantly, the 2 LMCT excited state of the substituted Fe(III) complexes not only retains the excited state energy but also shows only modestly reduced emission quantum yields and excited state lifetimes relative to the parent complex. This demonstrates that the favorable photophysical properties, characteristic of the parent complex, could be exploited in prospective photoactive assemblies with the 4-phenyl position as an attachment point. Our results reveal remarkably small effects of both electronwithdrawing and -donating substituents on the ground and excited state properties, thereby demonstrating that the [Fe(phtmeimb) 2 ]PF 6 motif should tolerate a wide range of modification for the above purposes without loss of the favorable photofunctionalities. ■ ASSOCIATED CONTENT
Synthesis, 1 H and 13 C NMR spectra, HR-MS spectra, single crystal X-ray diffraction, magnetic susceptibility and magnetization measurements, Moßbauer spectroscopy, electron paramagnetic resonance measurements, steady state spectroscopy, steady state absorption, steady state emission, transient absorption spectroscopy,