UST Research Online UST Research Online

: Incorporating radical ligands into metal complexes is one of the emerging trends in the design of single-molecule magnets (SMMs). While signi ﬁ cant e ﬀ ort has been expended to generate multinuclear transition metal-based SMMs with bridging radical ligands, less attention has been paid to mononuclear transition metal − radical SMMs. Herein, we describe the ﬁ rst α -diiminato radical-containing mononuclear transition metal SMM, namely, [ κ 2 -PhTt t Bu ] - Fe(AdNCHCHNAd) ( 1 ), and its analogue [ κ 2 -PhTt t Bu ]Fe-(CyNCHCHNCy) ( 2 ) (PhTt t Bu = phenyltris( tert-butylthiomethyl)- borate, Ad = adamantyl, and Cy = cyclohexyl). 1 and 2 feature nearly identical geometric and electronic structures, as shown by X-ray crystallography and electronic absorption spectroscopy. A more detailed description of the electronic structure of 1 was obtained through EPR and Mo ̈ ssbauer spectroscopies, SQUID magnetometry, and DFT, TD-DFT, and CAS calculations. 1 and 2 are best described as high-spin iron(II) complexes with antiferromagnetically coupled α -diiminato radical ligands. A strong magnetic exchange coupling between the iron(II) ion and the ligand radical was con ﬁ rmed in 1 , with an estimated coupling constant J < − 250 cm − 1 ( J = − 657 cm − 1 , DFT). Calibrated CAS calculations revealed that the ground-state Fe(II) − α -diiminato radical con ﬁ guration has signi ﬁ cant ionic contributions, which are weighted speci ﬁ cally toward the Fe(I)-neutral α -diimine species. Experimental data and theoretical calculations also suggest that 1 possesses an easy-axis anisotropy, with an axial zero-ﬁ eld splitting parameter D in the range from − 4 to − 1 cm − 1 . Finally, dynamic magnetic studies show that 1 exhibits slow magnetic relaxation behavior with an energy barrier close to the theoretical maximum, 2 | D | . These results demonstrate that incorporating strongly coupled α -diiminato radicals into mononuclear transition metal complexes can be an e ﬀ ective strategy to prepare SMMs.


■ INTRODUCTION
Due to the modern development of advanced physical and computational methods, in-depth studies of the electronic structures of metal complexes have become increasingly tractable.−16 In this context, a thorough understanding of their electronic structures through which structure−property and structure−reactivity relationships may be established is crucial for the rational design of catalysts and functional materials.
−19 During our interrogation of the electronic structures of the dioxolene complexes, we discovered the magnetic bistability of several five-coordinate Co(II)-semiquinonate radical complexes. 20These observations stimulated our interest in deducing the magnetic properties of the tris(thioether) ligand-containing transition metal complexes of redox-active ligands.Our attention was drawn to recent reports of radical-ligand-containing metal complexes that exhibit slow magnetic relaxation, a phenomenon characteristic of single-molecule magnets (SMMs). 12Due to potential applications in quantum information processing and high-density data storage, SMMs have received tremendous interest.Extraordinary effort has been put into synthesizing molecules with high spinrelaxation barriers.While the most recognized strategy to obtain SMMs focuses on maximizing the spin state (quantified as S) and the uniaxial magnetic anisotropy (quantified as the zero-field splitting parameter D), the design of the radicalligand-containing SMMs takes into consideration a third parameter that has received less attention: the exchange coupling constant J.When radical ligands are incorporated into metal complexes, direct-exchange coupling involving paramagnetic metal ion(s) and radical ligand(s) is present.When the exchange coupling is sufficiently strong, i.e., the J value is large, the energy level of the spin ground state will be well below the energy levels of the excited states, resulting in a suppression of the relaxation pathways that involve spin excited states.Meanwhile, quantum tunneling processes within a single spin-state manifold might also be suppressed by the strong exchange coupling.Thus, incorporating the direct-exchange coupling of metal ions with radical ligands can be an effective strategy to produce SMMs with higher relaxation barriers.
Widely used in coordination chemistry and catalysis, 42−44 αdiimines are redox-active ligands that can access three different redox states: neutral α-diimines (L 0 ), monoanionic π-radicals (L −1 ), and dianionic diamides (L −2 ), where each state has distinct C−N and C−C bond distances (Scheme 1). 45In addition, α-diimine ligands are easily accessible, and their steric and electronic properties may be tuned by varying the ligand substituents, i.e., R 1 , R 2 , and R 3 (Scheme 1).Most importantly, the monoanionic radical form of α-diimine ligands is wellknown for delivering exceptionally strong exchange coupling, 15,16,46−48 making them attractive candidates for constructing mononuclear transition metal-radical SMMs.−52 Herein, we report an α-diiminato radical-containing mononuclear iron complex, [κ 2 -PhTt tBu ]Fe(AdNCHCHNAd) (1), that exhibits field-induced slow magnetic relaxation.We also report its analogue, [κ 2 -PhTt tBu ]Fe(CyNCHCHNCy) (2).Both complexes were subjected to structural and spectroscopic characterizations and were found to have nearly indistinguishable geometric structures and spectroscopic features.The electronic structure and magnetic properties of 1 were further interrogated using a combined experimental and computational approach.The results indicate that 1 contains an antiferromagnetically coupled α-diiminato π-radical ligand with an exchange coupling constant J < −250 cm −1 (J = −657 cm −1 , DFT).To the best of our knowledge, the magnitude of the exchange coupling in 1 is among the largest observed in radical-containing SMMs.The calibrated computational studies also revealed that the ground state Fe(II)−α-diiminato radical configuration has a significant ionic character, which is weighted specifically toward the Fe(I)-neutral α-diimine description.To our knowledge, this unique electronic property has not been previously reported in any α-diimine-containing transition metal complex.
■ EXPERIMENTAL AND COMPUTATIONAL METHODS 2.1.General Information.All air-and moisture-sensitive reactions were performed under N 2 using standard Schlenk techniques or under an Ar or N 2 atmosphere in a Vacuum Atmospheres glovebox equipped with a gas purification system.Unless otherwise noted, all reagents were purchased from commercial sources and used without further purification.Solvents were of reagent-grade or better and were dried by passing them through activated alumina, then stored over 4 Å molecular sieves prior to use.Deuterated solvents were purchased from Cambridge Isotope Laboratories and stored over 4 Å molecular sieves.[PhTt tBu ]FeCl, 53 AdNCHCHNAd, and CyNCHCHNCy 54 were prepared following published procedures.
2.2.Synthesis of 1 and 2. 2.2.1.Synthesis of 1.To the solution of [PhTt tBu ]FeCl (147 mg, 0.3 mmol) in 1 mL of THF and 30 mL of diethyl ether was added AdNCHCHNAd (97 mg, 0.3 mmol) in small portions.The reaction mixture was stirred for 1 h.A pink-purple precipitate formed.KC 8 (45 mg, 0.33 mmol) was then slowly added to the reaction mixture over about 10 min.The solution turned brown.After stirring for 3 h, the mixture was filtered through a pad of Celite, and the solvent was removed under vacuum.The residue was washed with pentane (2 × 4 mL), dissolved in 4 mL of benzene, and passed through a small plug of APTS-coated silica gel to remove impurities.After removing the benzene under vacuum, the residue was washed with pentane (2 × 2 mL) and dried under vacuum to obtain a yellow-brown powder (84 mg, 36%).Single crystals suitable for X-ray crystallography were obtained by the slow evaporation of the pentane/diethyl ether solution (v/v = 1:1) of the product at room temperature. 1 (188).LIFDI-MS: m/z calcd.for C 43 H 70 BFeN 2 S 3 (M) + 777.4144, found (M) + 777.4152 (100%).μ eff (C 6 D 6 ) = 3.9(1) μ B .
2.2.2.Synthesis of 2. To the solution of [PhTt tBu ]FeCl (147 mg, 0.3 mmol) in 8 mL of THF and 16 mL of diethyl ether was added CyNCHCHNCy (66 mg, 0.3 mmol) in small portions.The reaction mixture was stirred for 1 h.A dark blue solution was obtained.KC 8 (45 mg, 0.33 mmol) was then slowly added to the reaction mixture over about 10 min.The solution turned brown.After stirring for 3 h, the mixture was filtered through a pad of Celite, and the solvent was removed under vacuum.The residue was washed with pentane (2 × 4 mL), dissolved in 4 mL of benzene, and passed through a small plug of APTS-coated silica gel to remove impurities.After removing the benzene under vacuum, the residue was washed with pentane (2 × 2 mL) and dried under vacuum to obtain a yellow-brown powder (90 mg, 45%).Single crystals suitable for X-ray crystallography were obtained by the slow evaporation of a concentrated diethyl ether solution of the product at room temperature. 1 2.3.Physical Methods.NMR spectra were recorded on a Bruker AVIII 400 spectrometer.Chemical shifts (δ) were referenced to residual protons in the deuterated solvents.LIFDI-MS 55,56 was performed on a Waters GCT Premier mass spectrometer.Electronic absorption spectra were recorded on a Varian Cary 50 UV−vis spectrophotometer using screw-top quartz cuvettes with a 1 cm path length.−59 Continuous wave X-band EPR spectra were collected on a Bruker E580 at T = 10 K using a 0.2 mW (or 2 mW) power.The modulation amplitude was 10 G. EPR spectra were simulated in Matlab using EasySpin. 60ossbauer spectra were recorded on a closed-cycle refrigerator spectrometer (model CCR4K) equipped with a 0.04 T permanent magnet while maintaining temperatures between 5 and 300 K.The samples consisted of solid powders (or crystalline material) that were suspended in Nujol, placed in Delrin 1.00 mL cups, and frozen in liquid nitrogen.The isomer shifts are quoted at 5 K with respect to the iron metal spectrum recorded at 298 K. Mossbauer spectra were analyzed using WMOSS software (Thomas Kent, See Co., Edina, MN).
The static magnetic properties of complex 1 were measured on samples of ground crystals using a Quantum Design MPMS-XL SQUID magnetometer operating over the temperature range 1.8−330 K at a 1000 Oe direct-current field.The data were corrected for diamagnetic contributions using Pascal constants.AC magnetic susceptibility measurements were performed on the same samples with an oscillating field of 5 Oe in the frequency range from 1 Hz to 1.5 kHz.
Single-crystal XRD data were obtained by mounting crystals onto plastic mesh using a viscous oil and cooling them to the data collection temperature.Data were collected on a Bruker-AXS APEX II CCD diffractometer with graphite-monochromated Mo Kα radiation (λ = 0.71073 Å).Unit cell parameters were obtained from 36 data frames, 0.3°ω, from three different sections of the Ewald sphere.The data sets were treated with absorption corrections based on redundant multiscan data.For 1, the systematic absences in the diffraction data are consistent with the triclinic space group, P1.For 2, the systematic absences in the diffraction data are consistent with the monoclinic space group, P2 1 /c, and the asymmetric unit contains two symmetryindependent molecules.All non-hydrogen atoms were refined with anisotropic displacement parameters.All hydrogen atoms were treated as idealized contributions.Atomic scattering factors are contained in the SHELXL 2013−2014 program libraries.The CIF filess have been deposited with the Cambridge Crystallographic Database as CCDC 1974684 for 1 and 1974685 for 2.
2.4.Computational Methods.Initial density functional theory (DFT) calculations were performed in Gaussian 16 rev.C.01. 61 A model of 1 was built from crystallographic coordinates and optimized using the BP86 functional 62,63 with definition two of Ahlrich's triple-ζ with polarization (def2-TZVP) basis 64 on all atoms in the gas phase on the broken symmetry S = 3/2 surface.An analytical frequency calculation was performed to ensure that a stationary point on the potential-energy surface had been reached and provide thermal corrections to the Gibbs free energy.No imaginary frequencies were obtained.Molecular orbitals were visualized with Lumo. 65 truncated model of 1 called 1* was built using the optimized coordinates of 1 by replacing the phenyl ring with a hydrogen, replacing the noncoordinated thioether arm at the S position with a proton, and replacing the adamantane with a methyl that had the C located at the bridgehead carbon position.This truncated model was partially optimized by freezing the positions of the heavy atoms and allowing the hydrogens to freely optimize with the BP86 functional on the broken-symmetry S = 3/2 spin surface.Time-dependent (TD) DFT calculations were performed with the TPSSh functional with the polarizable continuum model in THF to yield the first 200 excited states.
The truncated model 1* was subsequently used for property calculations in ORCA ver.4.2.1. 66,67All calculations used the same basis sets as above.A broken-symmetry DFT solution was initially computed and equivalent to the Gaussian computation.The fractional orbital occupation density (FOD) was computed on both 1 and 1*. 68raphical FOD plots were generated from the electron density to define the delocalization of correlated electrons.Mossbauer parameters were calculated by determining the field gradient (V) and the electron density at the nucleus (ρ 0 ).The value of δ iso was derived using ρ 0 and the TZVP correction parameters α, β, and C for the TPSSh functional 69 by δ iso = α(ρ 0 − C) + β.EPR parameters were calculated using the gauge invariant origin.The D tensor was calculated by including spin−spin and spin−orbit contributions.The spin−orbit component included the 1e − terms, semi-numerical Coulomb terms, and exchange via one-center exact integrals.
The DFT broken-symmetry S = 3/2 model was used as an initial guess for a second-order Møller−Plesset perturbation theory (MP2) single-point calculation. 70This calculation was subsequently used as the initial guess for a domain-based local pair-natural orbital (DLPNO) coupled-cluster (CC) calculation. 71,72The CC calculation included single and double excitations and perturbative triples (CCSD(T)).CCSD(T) single-point calculations were performed on both the quartet and sextet states.The electric field gradient for the Mossbauer parameters was calculated analogously to that described above.Electron paramagnetic resonance properties were computed analogously to that described above, except the origin was defined as the center of charge and the resolution of the identity (RI) approximation 73 was used with associated auxiliary basis sets.
The DFT broken-symmetry S = 3/2 model was used to generate quasi-restricted orbitals (QRO) as the initial guess for complete active-space self-consistent field (CASSCF) computations.Preliminary CASSCF calculations were converged using a smaller basis set (def2-SVP), followed by a tighter convergence with the full basis as described above.These preliminary calculations employed the iterative-configuration expansion (ICE-SCF) 74 to dramatically speed up the computations.Orbitals were selected for inclusion in the active space using occupation number N (0.005 > N > 1.995).The final CASSCF calculations employed 11 electrons in 14 orbitals (CAS- (11,14)) utilizing the same basis sets as those in the DFT and CCSD(T) calculations.These calculations also required RI and appropriate auxiliary basis sets.Roots for state-averaged (SA) CASSCF were determined by an initial computation with frozen orbitals.Quartet and sextet roots within 10 000 cm −1 of the ground state were chosen for the SA-CASSCF.n-Electron valence second order perturbation theory (NEVPT2) 75 was used to determine the dynamic correlation energy correction to the ground-state CAS solutions, and the RI approximation was used.The strongly contracted (SC) variant of this method was used.Mossbauer parameters were calculated analogously to the CC computations, and the EPR parameters were calculated within the CASSCF module.
■ RESULTS AND DISCUSSION 3.1.Syntheses and Crystal Structures.Complexes 1 and 2 were synthesized in 36% and 45% yields, respectively, by the addition of 1 equiv of potassium graphite (KC 8 ) to a 1:1 mixture of [PhTt tBu ]FeCl and the corresponding α-diimine in THF/diethyl ether (Scheme 2).Yellow-brown crystals suitable for X-ray diffraction were obtained for 1 and 2 by the slow evaporation of concentrated pentane/diethyl ether and diethyl ether solutions, respectively.Thermal ellipsoid plots of the molecular structures and selected metric parameters of 1 and 2 are shown in Figure 1.Complex 2 crystallizes with two symmetry-independent molecules, 2-Fe1 and 2-Fe2, and only 2-Fe1 is shown in Figure 1.Both 1 and 2 contain a pseudotetrahedral iron center coordinated by two thioether sulfur atoms and two nitrogen atoms from the α-diimine ligand.The third sulfur is uncoordinated, likely due to the steric bulk of the R groups.The κ 2 -coordination mode of the tris(thioether) ligand was also found in a dialkyltetraazadiene iron complex, namely [κ 2 -PhTt tBu ]Fe(AdNNNNAd), 53 which is an isoelectronic analogue of 1 and 2.
3.2.Spectroscopic and Static Magnetic Properties.In addition to their metric parameters, 1 and 2 share striking similarities in their electronic absorption spectra (Figure 2), which point to a common electronic structure.While the detailed assignments of the electronic transitions will be discussed later, it is informative to compare 1 and 2 with [iPrNC(Ph)CHNiPr] 2 Fe, an iron(II) complex containing αdiiminato radical ligands, and [iPrNC(Ph)CHNiPr]FeCl 2 , an iron(II) complex containing a neutral α-diimine ligand. 451 exhibits strong charge-transfer (CT) transitions at 402 (7983 M −1 cm −1 ) and 456 (sh) nm, and 2 exhibits similar CT bands at 401 (6816) and 458 (sh) nm.While analogous CT transitions in the 400−500 nm region, i.e., 441 (6000 M −1 cm −1 ) and 515 nm (2000), were also observed for [iPrNC-(Ph)CHNiPr] 2 Fe, [iPrNC(Ph)CHNiPr]FeCl 2 lacks intense CT bands in the same region.Thus, the CT bands in the 400− 500 nm region are suggestive of an α-diiminato radical ligand.Given that 1 and 2 share a common electronic structure, only complex 1, whose pure crystalline form was more readily obtainable in sufficient quantities, was subjected to further spectroscopic and magnetic property interrogations.While the C−N and C−C bond lengths are informative of the redox states of the α-diimine ligands, 57 Fe Mossbauer spectroscopy provides insight into the oxidation and spin states of the iron center.Figure 3 shows the Mossbauer spectrum of 1 at 80 K in a magnetic field of 0.04 T. Table S2 contains the Mossbauer parameters of 1 and other relevant iron complexes for comparison.
1 exhibited an isomer shift at δ 0.60 mm/s and quadrupole splitting ΔE Q = −4.03mm/s (Table S2).The isomer shift of 1 is much larger than those of the all the S-and N,S-ligated iron(III) compounds in the literature (one example 77 is given in Table S2) but somewhat closer to those of the the high-spin iron(II) dimer {[PhTt tBu ]FeCl} 2 (δ 0.96(3) mm/s, Table S2) 53 and a bis(benzimidazolato) ligated diferrous [2Fe-2S] cluster (δ 0.79 mm/s, Table S2). 78Interestingly, some sulfurligated high-spin Fe(I) complexes such as [PhTt tBu ]Fe(PMe 3 ) and [PhTt tBu ]Fe(PhCCPh) also exhibit similar isomer shifts (Table S2). 53Nevertheless, when the Mossbauer data were considered with the crystallographic and electron absorption spectroscopic data, it is clear that 1 is most consistent with the description of a high-spin iron(II) complex with an αdiiminato radical ligand.
To study the magnetic interaction between the high-spin iron(II) and the α-diiminato radical ligand, static magnetic properties were measured on samples of crushed crystals of 1.
The data were corrected for the diamagnetic contribution using Pascal constants.As shown in Figure 4, the χT value of 1.91 cm 3 •K/mol at 300 K (μ eff = 3.91 μ B ) comports with the solution magnetic moment at room temperature, μ eff = 3.9(1) μ B , and corresponds to an S = 3/2 ground state (spin-only, g = 2.02) derived from the antiferromagnetic coupling between the high spin iron(II) center (S = 2) and the ligand radical (S = 1/ 2).The χT values remain nearly constant throughout the temperature range 25−300 K, indicating very strong magneticexchange coupling.Below 25 K, the χT values dropped gradually to 1.38 cm 3 •K/mol −1 at 2 K, which may be attributed to zero-field splitting (ZFS) or intermolecular interactions.The field-dependent magnetization data collected at 1.8 K slowly increased at lower fields (Figure S1), which is indicative of antiferromagnetic intermolecular interactions.At fields above 5 T, the magnetization approaches saturation at 2.24 μ B , which is well below the expected value for the S = 3/2 ground state and is also indicative of ZFS or intermolecular interactions.The field-dependent magnetization curves at different temperatures (1.8−4.5 K) are nonsuperimposable (Figure 4 inset), which is further indicative of ZFS.In order to obtain an acceptable estimation of the magnitude of the coupling constant J, the susceptibility data were simulated using PHI software 79 under the Hamiltonian The first term corresponds to Zeeman splitting, whereas the second and third terms are the zero-field splitting in the metal ion and Heisenberg−Dirac−van Vleck magnetic exchange, respectively, where S 1 is the metal spin (S = 2) and S 2 is the radical spin (S = 1/2).Satisfactory simulations were obtained for coupling constant J values < −250 cm −1 (Figure 4).Simultaneous fitting of χT and the reduced magnetization data using the Hamiltonian in eq 1 resulted in |D ion |(E) = 2.24 (0.55) cm −1 with J = −660 cm −1 (g 1 = 2.19, g 2 = 1.94, and g 3 = 1.84).Notably, a similar tetrahedral iron(II)−radical complex also exhibits very strong antiferromagnetic coupling, which leads to an effective S = 3/2 ground state (LMeFe-(AdNNNNAd), L = HC[C(Me)N(2,6-iPr2C 6 H 3 )] 2 , J = −850 cm −1 ). 80Alternatively, the system was treated as a fully coupled S = 3/2 system, and the data were fit according to the Hamiltonian Where S T is the fully coupled 3/2 spin state resulted in D T (E) = −3.52 (−1.2) cm −1 (g av = 1.99 given g 1 = 2.19, g 2 = 1.94, and g 3 = 1.84,EPR).While static magnetometry measurements were used to estimate the ZFS parameters, this technique could not unequivocally determine the sign of the axial ZFS parameter D (i.e., D T in eq 2).In this context, continuous-wave X-band EPR spectroscopy was performed to provide a high-resolution evaluation of the electronic structure of 1 (Figure 5).Experiments were conducted on dilute solutions (1 mM) in THF at 10 K at 2.0 and 0.2 mW power levels.The spectra are dominated by broad effective S = 1/2 features with effective gvalues of 4.83, 3.27, and 1.92, which are consistent with highspin iron.Given the model of a high-spin iron(II) strongly antiferromagnetically coupled to a α-diiminato ligand radical that developed above (|J| > 250 cm −1 , Figure 4), the spectrum of 1 was modeled in a J ≫ D regime as an effective S = 3/2 system, meaning that the shape of the spectrum is dominated by ZFS.Consequently, fits to both positive and negative D  were explored, which yielded the following two equivalently good fits: (i) g = 2.11, 2.04, and 2.02 (g av = 2.06); D = +2.19cm −1 ; and E/D = 0.146 (Figure S5); and (ii) g = 2.19, 1.94, and 1.84 (g av = 1.99);D = −2.65 cm −1 ; and E/D = 0.073 (Figure 5).While these fits are indistinguishable, no satisfactory fit of the static magnetometry data can be accomplished with g av > 2.02, suggesting that the latter fit with a negative D value is more consistent with the magnetometry experiments.In addition, computational studies with various methods further suggest that the sign of D is negative (vide infra).
3.3.Dynamic Magnetic Properties.Recently, metal− radical complexes have been extensively pursued as a viable route for improved magnetically coupled systems with SMM behavior. 12,21In 1, the strong exchange coupling between the α-diiminato radical and the high-spin iron(II) leads to an effective S = 3/2 magnetic center in an axially distorted tetrahedral coordination environment, which is typically expected to give rise to SMM behavior. 81,82The large magnitude of the exchange coupling constant, |J| > 250 cm −1 , ensures a large separation between the low-lying spin excited states and the ground state.Such a large separation of the excited states is expected to eliminate a possible fast relaxation pathway that could otherwise be detrimental to the thermal energy barrier.In addition, strongly coupled radical discrete molecules were shown to exhibit a significant enhancement of the thermal relaxation barriers up to the theoretical maximum due to the suppression of quantum tunneling pathways. 26Thus, despite of the relatively small easy-axis anisotropy (−4 cm −1 < D < 0), a small effective energy barrier to the spin reversal was expected (maximum relaxation barrier U = (S 2 −1/4) |D| for half-integer spin systems when S = 3/2 and U = 2|D|).Indeed, the AC susceptibility measurements of compound 1 revealed a fieldinduced out-of-phase AC signal up to 2.5 K, as shown in Figure 6, which is indicative of possible SMM behavior.To determine the optimum applied DC field, the frequency-dependent AC measurements of 1 were scanned under different applied DC fields to give the best signal under 1000 Oe (Figure S2).The frequency-dependent measurements at different temperatures in the range of ν (1−1500 Hz) were measured for the pure sample over the temperature range 1.8−2.5 K under a 1000 Oe DC field (Figure 7).The data were fit using a Debye model 83 to give an estimated energy barrier U eff = 5.3 cm −1 and a preexponential factor τ 0 = 7.5 × 10 −6 s (1000 Oe) (Figure S3).Notably, the effective relaxation barrier exactly matches the theoretical maximum barrier, U = 2|D| (|D| = 2.65 cm −1 , EPR).Thus, these results demonstrate the potential for using strongly exchange-coupled nonbridging radical ligands to achieve slow magnetic relaxation in mononuclear transition metal complexes even when the magnetic anisotropy is relatively small.While a number of mononuclear iron complexes, including iron(I), 84,85 iron(II), 86−95 and iron(III) complexes, 96−98 have  been reported to exhibit SMM behavior, only one example 41 besides complex 1 contains a radical ligand.In that case, the relaxation time was not reported due to the lack of a clear maximum in the out-of-phase AC susceptibility.
3.4.Computational Studies.Density functional theory calculations were performed on 1 to evaluate its electronic structure in greater detail.Using the molecular structure derived from crystallographically determined coordinates, initial single-point energies that used the BP86 functional revealed that the quartet state was lower in energy than the sextet, which is consistent with the experimental finding.Consequently, the quartet wave function was used for the optimization.DFT (BP86) predicts a tetrahedral geometric structure, which is in close agreement with the structure derived from the X-ray crystallography of 1. Specifically, two of the thioether arms are coordinated to the iron at Fe−S distances of 2. The inspection of the frontier molecular orbitals from a single-point calculation (TPSSh) of 1 further supports a highspin ferrous and L π-radical assignment (Figure 8).Using the unfilled orbitals, which reflect the uncompensated spin of their filled counterparts, four spin-down, predominantly iron, 3d orbitals were found, which is reflective of a high-spin ferrous (d 6 ) electronic structure.A spin-up π* orbital of the α-diimine ligand was also identified (213α), indicating the presence of an α-diimine radical ligand.The ⟨S 2 ⟩ value of this quartet was 4.45 and became 3.77 after the removal of the first spin contaminant, which is close to the expected value of 3.75 for an S = 3/2 system.Thus, J can be estimated using the formalism of Yamaguchi by where LS E and HS E are the SCF energies of the S = 3/2 and S = 5/2 states, respectively, and ⟨S 2 ⟩ is the spin expectation value of the respective calculation.This yielded a sizable J value of −657 cm −1 .When interpreted in the context of the molecular orbitals, there exists a single ef ficient pathway for magnetic exchange out of four possible pathways for this antiferromagnetic coupling, wherein the ligand radical strongly interacts with the Fe 3d xy (213β) orbital to strongly stabilize the S = 3/2 state.
TD-DFT (TPSSh) calculations were performed to aid in the assignment of the electronic excited states.For computational efficiency, a truncated model of 1 was created (1*) where the heavy-atom positions were frozen to the larger model positions (see the Computational Methods section).This choice was   justified based on the nearly identical quartet−sextet splitting and g-tensor calculations (see below).Vacuum and solvated calculations were performed on 1* and both yielded qualitatively similar results compared to those of the experimental spectrum, lending confidence to the band assignments (Figure S6 and Table S7) and the overall electronic structure model.The prominent feature at 402 nm in the experimental spectrum was determined to be two closely spaced transitions at 402 and 404 nm, which are composed of coordinated thioether S p σ → Fe 3d yz and thioether S p σ → L π 3 transitions in nature.The shoulder on the low-energy side of this transition, which was experimentally observed at 456 nm, is correlated to a TD-DFT transition at 496 nm, L π 3 → Fe d xy .The dominant higher-energy transition was predicted to be L π → L π* overlapped with less-intense thioether S p σ → Fe d transitions.
The close spacing of the LMCT transitions and the significant spin contamination in the quartet from the sextet led us to consider the electron correlation with a benchmark CCSD(T) methodology and the multireference character with a CAS methodology.These calculations used the truncated model 1* for computational feasibility.CCSD(T) incorrectly orders the sextet state as lower in energy than the quartet.Moreover, the calculation of the fractional occupation numberweighted electron density (Table S10) demonstrated significant Fermi smearing (ρ FOD = 1.68) across the Fe and diimine radical−ligand π-system.The large delocalized ρ FOD value indicates a true multireference wave function and suggests that a multiconfigurational methodology (i.e., CAS) should be employed at minimum.
The broken-symmetry DFT wave function of 1* was used to generate quasi-restricted orbitals (QROs) for the initial CASSCF guess.The active space was systematically expanded by the addition of occupied and virtual orbitals to include orbitals with occupancy N between 0.005 and 1.995.Guided by the plot of the FOD, the π-system of the alpha-diimine, S p σ orbitals, and the Fe 3d set were considered, but the final 11 electrons in the 14 orbitals solution also included the "doubleshell" unoccupied Fe 4d orbitals 99 and not the S p σ orbitals.This calculation yielded a strongly multiconfigurational ground state spanning the following three configurations (Table S11): 48% (3d xy ) 2 (π 3 ) 0 , 29% (3d xy ) 1 (π 3 ) 1 , and 18% (3d xy ) 0 (π 3 ) 2 (invariant active orbitals were omitted).The (3d xy ) 1 (π 3 ) 1 configuration is equivalent to the broken-symmetry Kohn− Sham determinant.These three configurations mirror the common resonance forms of the α-diimine ligand and suggest a significant ionic contribution to bonding, which is weighted toward the Fe(I) and neutral α-diimine description (−5J = 1706 [1992] cm −1 (square brackets indicate the NEVPT2 corrected value)).An investigation of quartet and sextet excited states of the final CAS (11,14) wave function revealed several low-lying states on each spin manifold.Thus, final molecular orbitals were optimized by the systematic inclusion of quartet and sextet roots; ultimately, five of each spin were equally weighted to generate a state-averaged (SA) CAS solution.The SA-CAS ground state configurations indicate a less ionic ground state, and inclusion of the four low-lying quartet excited states shift the configurational weighting toward the Fe(II)−α-diimine radical description (Figure 9 and Table S13).The first sextet excited state is composed of a single dominant configuration, (3d xy ) 1 (π 3 ) 1 , regardless of if the orbitals are state averaged, again reflecting that this state is well-described by a single determinant.The state-averaged quartet−sextet energy gap (−5J) is 1301 [1715] cm −1 (the NEVPT2 is indicated by square brackets), which is indicative of the strong antiferromagnetic coupling.
With the above computational methods, we further investigated the spectroscopic properties using the truncated model 1* to compare them with the experimental values (Tables 1 and S14).Mossbauer parameters for the quadrupole splitting ΔE Q (−2.995 mm/s) and the isomer shift δ (0.523 mm/s) were computed by DFT(TPSSh) and match reasonably well with the experimental values (δ 0.60 mm/s and ΔE Q = −4.03mm/s), further indicating that the DFT model effectively describes the electronic structure of 1.The calculation of the isomer shift requires fitting constants that are not available for CAS or CCSD(T).Nonetheless, the quadrupole splitting values can be directly calculated with these methods (CAS ΔE Q = −3.572mm/s CCSD(T) ΔE Q = −3.560mm/s) and are in good agreement with the experimental value.Computational EPR parameters (g 11 , g 22 , g 33 , D, and E/D) were also computed.The computational values were in agreement with the experimental values, with all methods yielding negative D values; however, the CAS approaches overestimate the rhombicity, leading to E/D ratios significantly larger than the experimental values.

■ CONCLUSIONS
The idea of incorporating a radical ligand in the design of SMMs has received increasing attention in recent years.In this context, we described the first α-diiminato radical-containing mononuclear transition metal SMM, namely, [κ 2 -PhTt tBu ]Fe-(AdNCHCHNAd) (1), and its analogue [κ 2 -PhTt tBu ]Fe-(CyNCHCHNCy) (2).X-ray crystallography; electronic, EPR and Mossbauer spectroscopies; SQUID magnetometry; and DFT and CAS calculations were employed to fully elucidate the geometric and electronic structures. 1 and 2 feature nearly identical geometric and electronic structures, both of which contain pseudotetrahedral high-spin iron(II) ions antiferromagnetically coupled to the α-diiminato radicals.The variable-temperature magnetic susceptibility data of 1 suggest strong exchange coupling between the high-spin ferrous ion and the ligand radical, with an estimated coupling constant J < −250 cm −1 .Such a strong magnetic coupling is supported by theoretical calculations.In particular, exper- imentally gauged DFT calculations predict an exceptionally large coupling constant of J = −657 cm −1 .
EPR spectroscopy, SQUID magnetometry, and computational analysis suggest that 1 possesses an easy-axis anisotropy, with axial ZFS −4 cm −1 < D < −1 cm −1 .Consistent with a negative D value and a large |J| value, which likely suppress quantum tunneling processes, 1 exhibits slow magnetic relaxation behavior with an energy barrier close to the theoretical maximum 2|D|.These results demonstrate that incorporating a strongly coupled radical ligand, such as the αdiiminato radical, into mononuclear transition metal complexes can be an effective route to SMMs.
Further insight into the electronic structure of 1 and 2 was provided by calibrated CAS calculations, which revealed that the ground state Fe(II)−α-diiminato radical configuration has significant ionic contributions that are weighted specifically toward the Fe(I)-neutral α-diimine formalism.While it is unknown how the dynamics of the magnetic relaxation correlate with the electronic structure, it will be of interest to interrogate the structure−property relationship in the future by switching to more electron-accepting α-diimine ligands so that the Fe(II)−α-diiminato radical state is more dominant.Preparing new Fe(II)−α-diiminato radical complexes with larger |D| values and thus higher energy barriers to magnetic relaxation is also of interest.

Figure 3 .
Figure 3. Mossbauer spectrum of a solid microcrystalline sample of 1 in Nujol at 80 K in a magnetic field of 0.04 T. The hashed lines represent the raw data, and the solid line is a result of the least-squares fit of the data.

Figure 6 .
Figure 6.Temperature dependence of the AC signal in 1.

Figure 7 .
Figure 7. Frequency dependence of the AC signal in 1.

Table 1 .
Comparison of Computationally Derived Spectral Parameters with Experiment aValues in square brackets are SC-NEVPT2 corrected values.b CCSD(T) with the domain local-pair natural orbital approximation.c Used the ground-state orbitals with five quartet roots and five sextet roots.d Used state-averaged orbitals with five quartet roots and five sextet roots.e CCSD and CAS calibration tables for the calculation of δ are not available; f Zero-field splitting parameters from the second-order spin−orbit coupling contribution. a