Report

Luminescence and reactivity of a charge-transfer excited iron complex with nanosecond lifetime

See allHide authors and affiliations

Science  18 Jan 2019:
Vol. 363, Issue 6424, pp. 249-253
DOI: 10.1126/science.aau7160

Orange-glowing iron at room temperature

Many photoactive coordination compounds contain precious metals. Replacing ruthenium with more–earth-abundant iron has been a long-sought goal, but iron compounds generally relax too rapidly after light absorption to channel the energy productively. Kjær et al. prepared an iron compound with an excited state stable enough to emit light for nanoseconds, or that could engage in bimolecular electron transfer (see the Perspective by Young and Oldacre). Targeting a ligand-to-metal rather than metal-to-ligand charge-transfer state was key to the achievement, as was the octahedral coordination environment rigidly enforced by two tridentate carbene ligands.

Science, this issue p. 249; see also p. 225

Abstract

Iron’s abundance and rich coordination chemistry are potentially appealing features for photochemical applications. However, the photoexcitable charge-transfer states of most iron complexes are limited by picosecond or subpicosecond deactivation through low-lying metal-centered states, resulting in inefficient electron-transfer reactivity and complete lack of photoluminescence. In this study, we show that octahedral coordination of iron(III) by two mono-anionic facial tris-carbene ligands can markedly suppress such deactivation. The resulting complex [Fe(phtmeimb)2]+, where phtmeimb is {phenyl[tris(3-methylimidazol-1-ylidene)]borate}, exhibits strong, visible, room temperature photoluminescence with a 2.0-nanosecond lifetime and 2% quantum yield via spin-allowed transition from a doublet ligand-to-metal charge-transfer (2LMCT) state to the doublet ground state. Reductive and oxidative electron-transfer reactions were observed for the 2LMCT state of [Fe(phtmeimb)2]+ in bimolecular quenching studies with methylviologen and diphenylamine.

Photoactive transition metal complexes play an important role in processes ranging from solar light harvesting (13) and light-emitting technology (4) to photocatalysis (5) and photodynamic therapy (6). Such applications almost always rely on charge-transfer (CT) excited states with sufficient lifetime and energy to drive electron transfer and visible light emission. Iron complexes provide an earth-abundant and environmentally benign alternative to noble metal systems (7) but have until recently been limited by subpicosecond deactivation of their CT states (8) to low-energy metal-centered (MC) states (912). These dynamics arise from the moderate ligand field splitting of Fe complexes with commonly used oligopyridyl ligands (8, 11, 12). Early work on Fe-centered oligopyridyl systems suggested the involvement of MC states with nanosecond lifetime in electron-transfer reactions (13). However, these results were later shown to be incompatible with the excited-state dynamics of the systems (14); the MC states are now generally considered too low in energy to participate in photochemistry of interest. Efforts to develop Fe-centered photofunctional systems have therefore focused on inhibiting the ultrafast CT → MC transitions (8).

We recently showed that strongly electron donating N-heterocyclic carbene (NHC) ligands raise the energy of MC states relative to CT states of iron complexes, thereby increasing the lifetime of the excited CT states (15). For FeII complexes with four NHC moieties and two pyridine moieties, we and others have recently demonstrated triplet metal-to-ligand charge-transfer (3MLCT) state lifetimes of a few tens of picoseconds (1517), thereby crossing the threshold for efficient interfacial electron injection from surface-bound Fe photosensitizers to a TiO2 electrode (17).

To further increase the lifetime of the CT states in iron complexes, we very recently saturated the iron center with six coordinating NHC moieties, leading to the [Fe(btz)3]2+/3+ complex [btz, 3,3′-dimethyl-1,1′-bis(p-tolyl)-4,4′-bis(1,2,3-triazol-5-ylidene)]. This complex featured order-of-magnitude-higher charge-transfer lifetimes in both its FeIII [100-ps ligand-to-metal charge-transfer (2LMCT)] and FeII (528-ps 3MLCT) states (18, 19). Moreover [Fe(btz)3]3+ exhibited room temperature photoluminescence (PL) from a CT state in the visible regime, albeit with an extremely low quantum yield (0.03%). The picosecond CT lifetimes still preclude most light-harvesting and light-emitting applications, but these results suggested that further improvements of FeII and FeIII complexes are notable in the broader context of development of photoactive and photoluminescent 3d6 and 3d5 complexes, respectively (20, 21).

We identified maximal ligand field strength and an optically allowed lowest CT excited state (18, 19) as key design elements for extending lifetimes and further increasing PL. For superior ligand field strength, we targeted anions for even more pronounced σ-donor ability. Near-perfect octahedral coordination capability (2224) was another factor that drew us to the tridentate facial NHC ligand {phenyl[tris(3-methylimidazol-1-ylidene)]borate} (phtmeimb) (25). This ligand has very recently been shown to support weak low-temperature solid-state LMCT and d-d emission in the d3 complex [MnIV(phtmeimb)2](OTf)2 (OTf, triflate) (26).

In this study, we demonstrate that the combination of a 2LMCT lowest excited state with the exceptional electronic and steric properties of the phtmeimb ligand results in a [FeIII(phtmeimb)2]PF6 complex featuring a CT state with nanosecond lifetime. [FeIII(phtmeimb)2]PF6 was efficiently synthesized from FeIIBr2 and in situ generated tris-NHC-carbene (phtmeimb) as illustrated in Fig. 1A. During the workup procedure in air, FeII is spontaneously oxidized to FeIII, resulting in an analytically pure product (see supplementary materials). Counterion metathesis with NaBPh4 provided access to the corresponding BPh4 salt (Ph, phenyl). The oxidation state of iron in both complexes was confirmed by single-crystal x-ray diffraction analysis. The cation in both [FeIII(phtmeimb)2]X [X = PF6 (Fig. 1B) or BPh4] salts displays a near-perfect octahedral geometry (table S4) in contrast to [Fe(btz)3](PF6)3 (19).

Fig. 1 Synthesis and structure of [Fe(phtmeimb)2]PF6.

(A) Synthetic route: 1, precipitation with tetra-n-butyl-ammonium bromide in acetone; 2, dissolution in water and precipitation with ammonium hexafluorophosphate; 3, dissolution in tetrahydrofuran under N2, then cooling to –78°C and addition of tert-butoxide; 4, addition of FeBr2, stirring under N2 at room temperature for 24 hours. (B) X-ray crystal structure. Thermal ellipsoids are shown at 50% probability with the six Fe–C bonds highlighted in black and gray stripes. Hydrogen atoms, counterions, and solvent molecules are omitted for clarity. Fe, orange; B, purple; N, blue; C, black.

Mößbauer spectroscopy and magnetometry identified the ground state of the isolated sample of [FeIII(phtmeimb)2]PF6 as low spin (S = ½), containing <1% FeII (figs. S12 to S14). Surprisingly, but similar to some low-spin FeIII-porphyrin complexes (27), the 1H nuclear magnetic resonance (NMR) spectrum of the paramagnetic [FeIII(phtmeimb)2]PF6 in CD3CN shows narrow peaks (<15 Hz at 298 K) (table S1), whereas the X-band electron paramagnetic resonance (EPR) spectrum in frozen solvent glasses at T = 4 to 20 K displays no distinct assignable bands (see the supplementary materials section).

Cyclic voltammetry of [FeIII(phtmeimb)2]+ (Fig. 2A) illustrates that both reduction to FeII and oxidation to FeIV are reversible, with half-wave potentials of E½ = −1.16 and 0.25 V versus Ferrocene (Fc), respectively. The pronounced shift toward negative potentials compared with potentials of previously reported Fe-NHC complexes (15, 17, 19) illustrates the exceptionally strong electron donor properties of the negatively charged tris-NHC ligands. Further oxidation of the FeIV complex at 1.67 V is irreversible (fig. S17). Previous observation of irreversible oxidation of [Fe(btz)3]3+ (19) at very similar potential indicates the assignment of this process to ligand oxidation as the potentials of metal-centered couples are lowered substantially by the phtmeimb ligand. With this assignment, the electrochemical potentials agree consistently with the energies of the LMCT transitions found for [Fe(btz)3]3+ (19) and the FeIII and FeIV states of [FeIII(phtmeimb)2] (see fig. S18 and associated discussion). Reduction of the FeII complex does not occur within the solvent-electrolyte potential window, demonstrating that the ligand reduction potential is below −3.3 V, which is again consistent with the interpretation of the electronic spectra (see supplementary materials).

Fig. 2 Electrochemistry and spectroscopy of [Fe(phtmeimb)2]+ in dry, air-saturated acetonitrile at room temperature.

(A) Cyclic and differential pulse voltammetry. (B) Optical absorption (left black curve), normalized PL (right black curve), and normalized excitation spectra (red circles). (C) Visible orange PL of 50 μM [Fe(phtmeimb)2]+ in dry, air-saturated acetonitrile upon 532-nm excitation.

The visible absorption spectrum of [Fe(phtmeimb)2]+ in acetonitrile (Fig. 2B, left curve) is dominated by a single band peaking at 502 nm (molar decadic absorption coefficient εmax = 2950 M−1 cm−1) with a minor shoulder around 545 nm. This band is bleached upon oxidation and reduction of the metal center, and its energy matches relatively well the difference in electrochemical potential between the FeIII-FeII couple and ligand oxidation. The lowest-energy absorption band of [Fe(phtmeimb)2]+ is therefore assigned to a LMCT transition.

Excitation of [Fe(phtmeimb)2]+ in acetonitrile with visible light below 600 nm results in strong orange PL (Fig. 2C). The spectral profile of the PL shown in Fig. 2B, right, peaks at 655 nm and has no appreciable structure besides a broad shoulder around 620 nm, which mirrors the LMCT absorption band of the complex. The PL intensity tracks the absorption cross section throughout the visible region, as illustrated by the superimposable absorption and excitation spectra (red circles in Fig. 2B).

The measured emission quantum yield of [Fe(phtmeimb)2]+ in air-saturated dry acetonitrile at room temperature was 2.1% (Φe = 0.021 ± 0.002). Notably, this value is a factor of 70 higher than the quantum yield measured for [Fe(btz)3]3+ (19) and is even slightly higher than the 1.8% quantum yield of the prototypical transition metal photosensitizer [Ru(bpy)3]2+ (bpy, 2,2′-bipyridine) under air-saturated conditions, used as reference for the quantification (28). The emission decay kinetics measured by time-correlated single-photon counting (TCSPC) (Fig. 3A) show a single exponential with a lifetime of τ = 1.96 ± 0.04 ns, a factor of 20 longer than τ observed in [Fe(btz)3]3+ (19). The photophysical properties of [Fe(phtmeimb)2]+, [Fe(btz)3]3+, and [Ru(bpy)3]2+ are compared in Table 1.

Fig. 3 Excited-state dynamics and computational analysis of [Fe(phtmeimb)2]+ photophysics.

(A) Time-correlated single-photon counting data (black, left y axis), transient absorption data at 390 nm (green circles, right y axis), and monoexponential fit of 1.96 ± 0.04 ns (red). a.u., arbitrary units; O.D., optical density. (B) Potential energies and Fe–C equilibrium bond lengths (Q) for relevant electronic states of [Fe(phtmeimb)2]+ (triangles) and potential surfaces (lines) drawn through the energy of each state at the geometry of the two other states (circles). The 2LMCT surface was extrapolated from the ground-state (GS) shape and experimental energy as described in the supplementary materials section.

Table 1 Photophysical parameters of [Fe(phtmeimb)2]+, [Fe(btz)3]3+, and [Ru(bpy)3]2+ in air-saturated acetonitrile at room temperature.

[Fe(btz)3]3+ data are from (19); [Ru(bpy)3]2+ data are from (28, 34). λmax,abs, maximum absorption wavelength; λmax,emiss, maximum emission wavelength; τ, CT excited-state lifetime.

View this table:

Taken together, the emission quantum yield and excited-state lifetime provide a radiative rate constant of kr = Φe/τ = 1.1 ± 0.2 × 107 s−1. The good agreement of this value with the approximate radiative rate constant kr = 1.5 × 107 s−1 estimated from the integrated extinction coefficient A (Table 1) of the LMCT band via the Strickler-Berg relationship strongly suggests that the emission occurs directly from the 2LMCT state. Thus, the intersection of the normalized absorption and emission bands at 582 nm provides the energy of the 2LMCT excited state, E0-0 = 2.13 eV (17,200 cm−1). From the photophysical parameters, the 70-fold increase in emission quantum yield from 0.03% for [Fe(btz)3]3+ to 2.1% for [Fe(phtmeimb)2]+ can be rationalized in terms of a 20-fold slower nonradiative decay (knr = [1 − Φe]/τ) and a 3.5-fold faster radiative rate constant (kr = Φe/τ).

The transient absorption (TA) spectra of [Fe(phtmeimb)2]+ recorded after excitation of the LMCT band at 500 nm are dominated by excited-state absorption (ESA) below 450 nm and exhibit a clear stimulated emission band between 600 and 800 nm. The ESA can be attributed to the transiently reduced iron center of the 2LMCT state. The stimulated emission indicates that the ground-state recovery is spin allowed, offering further support for the 2LMCT assignment of the excited state (see supplementary materials). The TA decay kinetics of [Fe(phtmeimb)2]+ (Fig. 3A, green) are in perfect agreement with the TCSPC results.

The long-term photostability of [Fe(phtmeimb)2]+ was measured and compared with that of [Ru(bpy)3]2+ by irradiating aerated acetonitrile solutions of both complexes with an 11-W compact fluorescent lamp for a total of 156 hours and measuring the absorption and emission spectra at intervals during this time period. Whereas clear signs of degradation set in for [Ru(bpy)3]2+ after 48 hours, the [Fe(phtmeimb)2]+ sample was virtually unchanged throughout the 156-hour experiment (figs. S21 and S22).

Density functional theory (DFT) revealed the minimum energies of the 4MC and 6MC states of [Fe(phtmeimb)2]+ (Fig. 3B) that are destabilized by 13 and 23% with respect to the ground-state minimum as compared with the previously reported [Fe(btz)3]3+ (19). The increased energies of the MC states together with the fact that the 2LMCT states are isoenergetic within the experimental uncertainty for the two systems suggest that the increase in the experimentally observed lifetime is related to an effective increase of the activation barrier for the decay of the 2LMCT state into the 4MC state.

Temperature-dependent emission lifetime measurements (fig. S23) show that the excited-state lifetime of [Fe(phtmeimb)2]+ increases by a factor of 4 (from 2.0 to 7.8 ns) upon decreasing the temperature to 100 K, in near-perfect agreement with the behavior of [Fe(btz)3]3+. Fitting the temperature-dependent lifetimes by an Arrhenius model retrieves an activation barrier of 3 kJ mol−1 and preexponential factor of 1 × 109 s−1, suggesting that for the decay channels that dominate this temperature dependence, the energy barriers for deactivation of [Fe(phtmeimb)2]+ and [Fe(btz)3]3+ are almost identical but that the preexponential factor for [Fe(phtmeimb)2]+ is lower. It is thus the decreased preexponential factor for the transition that results in an increased lifetime, and we tentatively ascribe this to an effective reduction of the crossing frequency from the 2LMCT state to the 4MC state, owing to the combined effect of several structural factors, including the higher symmetry and tighter spatial confinement of the [Fe(phtmeimb)2]+ ligand system.

With excited-state redox potentials of Eo(III*/II) = 1.0 V and Eo(IV/III*) = −1.9 V versus Fc [1.6 V and −1.3 V versus NHE (normal hydrogen electrode)], the 2LMCT state should be potent as both a photo-oxidant and photoreductant; furthermore, its nanosecond lifetime should enable efficient bimolecular electron transfer as long as rate constants are not too far from their diffusion-controlled limit. The reactivity of the 2LMCT state toward both electron donors and acceptors was studied by monitoring the impact of the quenchers on both steady-state emission intensity as well as emission lifetime (Fig. 4, A and B). The methylviologen dication (MV2+) and arylamines such as diphenylamine (DPA) are widely used electron-transfer quenching agents with suitable redox properties. The observed emission quenching results were attributed to oxidative and reductive electron transfer, respectively, generating [FeIV(phtmeimb)2]2+ and MV•+ (change in Gibbs free energy ΔGo = −1.08 eV) in the former case and [FeII(phtmeimb)2] and DPA•+Go = −0.55 eV) in the latter.

Fig. 4 Reactivity of the 2LMCT excited state of [Fe(phtmeimb)2]+ toward electron donors and acceptors.

Emission quenching was monitored by emission lifetime (TCSPC traces with exponential fits) and steady-state emission spectra (insets) for increasing concentrations (black to cyan) of (A) diphenylamine donor (0, 0.005, 0.01, 0.02, 0.05, 0.1, and 0.2 M) and (B) methylviologen acceptor (0, 0.01, 0.025, 0.05, 0.1, 0.25, and 0.5 M) in acetonitrile. (C) Stern-Volmer plots for steady-state intensity (open symbols) and lifetime data (solid symbols) from quenching experiments with diphenylamine (triangles) and methylviologen (circles). (D) Transient absorption spectra after laser flash excitation (465 nm) of [Fe(phtmeimb)2]+, monitoring products of oxidative quenching by methylviologen (0.25 M) 500 ns after excitation (black) and of reductive quenching by diphenylamine (0.2 M) 100 ns after excitation (red).

Bimolecular quenching rate constants in acetonitrile for dynamic quenching were determined from Stern-Volmer plots of emission lifetimes τ0/τ (Fig. 4C); rates were diffusion controlled for DPA (kq = 1.4 × 1010 M−1 s−1) and only somewhat lower even for MV2+ (kq = 2.7 × 109 M−1 s−1). Although no indications of ground-state complexation were observed with MV2+, additional static quenching by DPA was evident from the curved Stern-Volmer plot of steady-state intensities I0/I. With both quenchers, the formation of electron transfer products was unambiguously confirmed by transient absorption spectroscopy (Fig. 4D). Spectra after quenching by DPA show characteristic absorption of the donor cation radical peaking at 680 nm (29) and of the FeII state rising toward the ultraviolet region (<420 nm) (see also fig. S18). Also quenching by MV2+ resulted in transient absorption spectra that display the well-known absorption features of the acceptor radical MV•+ (at 396 and 606 nm) (30) together with the broad 700-nm band of the Fe(IV) complex (see fig. S18). In the flash photolysis experiments, the excited state at an initial concentration of ~1.5 × 105 M (determined by actinometry with [Ru(bpy)3]2+) was quenched with efficiencies of ~0.7 by 0.25 M MV2+ and nearly unity by 0.2 M DPA. From the initial concentrations of MV•+ (Δε396 = 41,800 M−1 cm−1) (30) and DPA•+ (Δε680 = 19,200 M−1 cm−1) (29), we estimate that in both quenching reactions about 5% of the charge-separated products escape geminate recombination in the solvent cage. The diffusional recombination of the separated products occurs on the time scale of 100 μs and, in case of DPA, proceeds via oxidation of the FeIII ground state by the donor radical (see supplementary materials). Although the cage escape yields values three to five times lower than those typically observed in the quenching of the 3MLCT state of [Ru(bpy)3]2+, that latter process benefits from spin restrictions to back electron transfer in the triplet radical pair (31); the yields compare very favorably to the negligible cage escape encountered in other cases of spin-allowed back electron transfer in the quenching of, for instance, singlet excited states of porphyrins (32).

The extended CT lifetimes in [Fe(phtmeimb)2]+ were accomplished without substantial loss of the >2-eV excited-state energy, providing the 2LMCT state with a superior combination of oxidative and reductive power exceeding the corresponding values of the archetypal [Ru(bpy)3]2+ sensitizer (fig. S26). Thermodynamically, the 2LMCT state should be capable of oxidizing or reducing a wide range of molecular donors and acceptors and p- or n-type semiconductor materials and of driving demanding photocatalytic reactions such as water oxidation or carbon dioxide reduction, which could further benefit from the complex’s intrinsic stability (fig. S26).

The 2% PL quantum yield also raises the prospect of applying Fe-NHC systems to biosensors and organic light-emitting diodes (33). These applications would benefit from the intrinsic low toxicity and earth abundance of Fe complexes, as well as the insensitivity of the 2LMCT excited state of [Fe(phtmeimb)2]+ to oxygen. Moreover, because both the ground and LMCT excited states of the FeIII light-emitting complex are doublets, they will not suffer from the endemic singlet-versus-triplet formation problem (33) of typical rare-earth light emitting complexes. Taken together, our results suggest that the 2LMCT state deserves more attention as a photofunctional state for iron and other transition metals as well.

Supplementary Materials

www.sciencemag.org/content/363/6424/249/suppl/DC1

Materials and Methods

Figs. S1 to S26

Tables S1 to S5

References (3556)

References and Notes

Acknowledgments: Funding: K.S.K. and J.B. acknowledge the Danish Council for Independent Research (5051-00095A and 8021-00410B) and the Carlsberg Foundation. N.W.R. acknowledges the Alexander von Humboldt Foundation within the Feodor-Lynen Fellowship program. P.P., R.L., and K.W. acknowledge the Swedish Foundation for Strategic Research. P.P. and K.W. acknowledge the Swedish Research Council. P.P. acknowledges the Swedish Energy Agency, the Knut and Alice Wallenberg Foundation, and the Swedish National Supercomputing Centers LUNARC and NSC via SNIC. K.W. acknowledges the LMK Foundation, Stiftelsen Olle Engkvist Byggmästare, the Carl Trygger Foundation, the Wenner-Gren Foundation, the Crafoord Foundation, Sten K Johnsons Stiftelse, and the Royal Physiographic Society. Author contributions: K.S.K., P.C., A.H., O.G., N.W.R., J.U., V.S., A.Y., and L.L. conducted the femtosecond-to-nanosecond TA and PL measurements. N.K. conducted all steady-state and time-resolved spectroscopy related to the excited-state quenching studies and analyzed the data. O.P. conducted the synthesis. L.A.F. and P.P. conducted the DFT calculations and theoretical analysis. K.-E.B. conducted NMR spectroscopy. L.H. and T.E. conducted Mößbauer spectroscopy. S.S., P.H., and J.B. conducted EPR spectroscopy. J.B. conducted magnetic susceptibility and magnetization measurements. D.S. conducted x-ray crystallography. K.S.K., V.S., and P.P. conceived of and interpreted the photophysics and excited-state cascade. R.L. conducted electro- and spectroelectrochemistry and emission spectroscopy; interpreted the electrochemical, spectroscopic, and photophysical properties; and conceived of and interpreted the excited-state quenching studies. K.W. conceived of the design and the synthesis of the ligand and the metal complex. K.S.K., O.P., V.S., P.P., R.L., and K.W. wrote the paper with co-writing input from K.-E.B., T.E., L.H., P.H., J.B., D.S., and N.W.R. All authors read and commented on the paper. Competing interests: The authors declare no competing interests. Data and materials availability: Crystallographic data are available free of charge from the Cambridge Crystallographic Data Centre under reference numbers CCDC-1842079 {[Fe(phtmeimb)2]PF6} and CCDC-1842084 {[Fe(phtmeimb)2]BPh4}. All other data are available in the main text or the supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article