Valence tautomerism and spin crossover in pyridinophane-cobalt- dioxolene complexes: an experimental and computational study†

Compounds [Co(L-N4R2)(dbdiox)](BPh4) (L-N4R2 = N,N'-di-alkyl-2,11-diaza[3.3](2,6)pyridinophane, R = iPr (1a), Et (2a); dbdiox = 3,5-di-tert-butyldioxolene) and [M(LNiPr2)(dbdiox)](BPh4) (M = Mn (3a), Fe (4a)) have been synthesized and investigated with a view to possible valence tautomeric (VT) or spin crossover (SCO) interconversions. Single crystal Xray diffraction data for all compounds at 100 or 130 K indicate trivalent metal cations and di-tertbutylcatecholate (dbcat2-) dioxolene ligands. Variable temperature magnetic susceptibility data for all species between 2 and 340 K are consistent with these redox states, and low spin configurations for the cobalt(III) ions and high spin for the manganese(III) and iron(III) irons.


Introduction
Elucidation of the electronic lability in metal complexes that undergo spin crossover (SCO) or valence tautomeric (VT) transitions remains a fascinating challenge for inorganic chemists. These compounds can be reversibly interconverted between forms with different chemical and physical properties upon application of a stimulus, such as heat, which makes them of interest for future applications in displays, sensors, photoresponsive devices or molecular electronics/spintronics. [1][2][3] The switching in SCO complexes is due to a spin state transition at a single metal center from low spin (LS) to high spin (HS), which is well-established for complexes of Fe(II), Fe(III), Co(II) and Mn(III). 1,4,5 Very important for SCO is the ligand donor set around the metal center, as the ligand field splitting governs the thermodynamics of the transition. Typical ligand donor sets are specific to the switching metal center, with N-, O-and S-donor ligands the most common. More 3 complicated are VT transitions, which involve intramolecular electron transfer between a metal and a ligand, often combined with a spin state transition at the metal center. The main exemplar molecules for VT transitions are octahedral cobalt complexes with 3,5-di-tert-butyldioxolene (dbdiox) ligands, for which electron transfer from a di-tert-butylcatecholate (dbcat 2-) ligand to the LS-Co(III) ion is coupled with a spin transition, yielding a HS-Co(II) ion bound to a di-tertbutylsemiquinonate (dbsq -) ligand. [6][7][8] In these complexes, the LS-Co(III)-dbcat tautomer predominates at low temperature, interconverting to the HS-Co(II)-dbsq form at higher temperature. Cobalt-dioxolene VT complexes typically incorporate N-donor ancillary ligands.
The thermal induction of both SCO and VT transitions is largely entropically driven, with the entropy gain arising from the higher density of vibrational states of the high temperature species due to the longer metal-ligand bonds and the higher spin state degeneracy. Like SCO transitions of Fe(II) complexes, 9 VT transitions in cobalt-dioxolene systems can sometimes also be induced by irradiation with visible light. 10 Twelve-membered tetraazamacrocylic N,N'-dialkyl-2,11-diaza[3.3]-(2,6)pyridinophane (L-N4R2; Chart 1) ligands fold upon coordinating to an octahedral metal-center, leaving two cis sites vacant for other ligands. 11 In the case of cobalt complexes of such tetradentate N-donor ligands, coordination of a suitable dioxolene ligand in these cis-disposed sites might be considered likely to generate a VT transition. However, intriguingly, instead a thermally-induced SCO transition was reported for [Co(L-N4tBu2)(dbsq)](B(p-C6H4Cl)4, which remains the only example of a cobalt-dioxolene compound to exhibit SCO, rather than a VT transition. 12,13 Ferromagnetic coupling is evident between the LS-Co(II) ion and the semiquinonate ligand at low temperature, while the nature of the coupling with the HS-Co(II) ion at high temperature is difficult to ascertain.
In contrast, use of the pyridinophane azamacrocycle with methyl substitution at the amine position 4 affords the complex [Co(L-N4Me2)(dbcat)] + ; the BPh4 -salt of this complex contains a LS-Co(III) ion and does not exhibit either SCO or a VT transition. 12,13 The difference between the electronic lability of the complexes with methyl-and tert-butyl-substituted macrocyclic ligands is attributed to the steric interaction of the axial tert-butyl substituents with the equatorial ligands, which lengthens the bond between the cobalt and the amine nitrogen and reduces the s-donor capacity of these axial ligands.

Chart 1 Tetradentate pyridinophane ligands and numbering scheme for metal complexes
Analogous iron complexes with the methyl-substituted macrocycle exist as [Fe(L-N4Me2)(dbsq)] 2+ , with a LS-Fe(III) center antiferromagnetically coupled to the semiquinonate radical, or as [Fe(L-N4Me2)(dbcat)] + with a HS-Fe(III) center. 14 In the absence of dioxolene coligands, SCO transitions have been observed for iron(II) and iron(III) complexes of L-N4Me2 and for iron(II) and cobalt(II) complexes of L-N4tBu2. 15 The only reported non-dioxolene cobalt(II) complex of L-N4Me2 exists only in the high spin state. 15 High spin manganese(II) and manganese(III) complexes of L-N4R2 (R = Me, iPr and tBu) ligands are known and do not exhibit electronic lability. [16][17][18][19] The L-N4R2 family of ligands display remarkable conformational flexibility in their metal complexes. 20 Clearly, very subtle electronic and steric factors govern the redox states and charge distributions in these metal complexes.
The crude product was dissolved in dichloromethane and layered with dioxane, yielding poorly crystalline purple square plates (95 mg, 76 %). Anal. Calcd (found) for C62H76N4O4BFe: C, 73.88 [Fe(L-N4iPr2)(dbdiox)](BPh4)×2MeOH (4a×2MeOH). The reaction followed the same procedure as describe for 4a.dioxane. The crude product was dissolved in dichloromethane and layered with methanol, yielding purple plate-shaped crystals. A suitable crystal was handpicked for structural analysis, and although a pure bulk sample was obtained, the product lost crystallinity quickly due to desolvation.

Single Crystal X-ray Data Collection and Structure Solution
Crystals were transferred directly from mother liquor to oil to prevent solvent loss. The crystallographic data (Table 1) for 2a×2dioxane and 3a×1.5dioxane, were collected at 130 K using a Rigaku Oxford Diffraction SuperNova Dual Wavelength single crystal X-ray diffractometer using Cu-Kα radiation (λ = 1.5418 Å). Data reduction was performed using CrysAlisPro software (Version 1.171.38.41) 32 using a numerical absorption correction based on Gaussian integration over a multifaceted crystal model. As crystals of 1a×dioxane and 4a×2MeOH were small and poorly diffracting data for these two compounds were collected at the MX1 beamline at the Australian Synchrotron, 33 with the wavelength tuned to Mo-Kα radiation (λ = 0.71073 Å). Using the Olex2 software, 34 the structures were solved using the intrinsic phasing routine in SHELXT 32 and refined using a full-matrix least squares procedure based on F 2 using SHELXL. 35,36 All non-hydrogen atoms were refined with anisotropic displacement parameters, while all hydrogen atoms were placed at geometrical estimates and refined using the riding model. For 3a×1.5dioxane one of the dioxane molecules was disordered over two orientations while for 4a×2MeOH one of the tert-butyl groups 9 was disordered over two orientations. For both structures the two disordered components were restrained to similar geometry; the final occupancy values were 0.874 (3)

Magnetic Measurements
Magnetic measurements in the solid state were carried out in the Unitat de Mesures Magnètiques Solution magnetic measurements employed the Evans NMR method. NMR samples for susceptibility measurements using the Evans method were prepared by dissolving a weighed amount of the metal compounds in a measured amount of deuterated dichloroethane or acetonitrile solvent. The concentration of the paramagnetic solute was in the range of 5-10 mg mL -1 . The temperature-dependent density changes of the solvent were corrected using equations and data from the International Critical Tables. [38][39][40][41] The complex solution was transferred into a coaxial 5 mm NMR tube containing a 1 mm capillary with the deuterated diamagnetic solvent and a small amount of tert-butanol as references. The residual 1 H NMR resonance of the tert-butanol methyl group was used for determining the susceptibility in solution.

57
Fe Mössbauer spectra were measured on multiple samples of 4a×dioxane (~20 mg) at various temperatures from 295 K to 6 K. Spectra were measured on a Mössbauer spectrometer (Science Engineering & Education Co., MN) equipped with a closed-cycle refrigerator system from Janis Research Co. and SHI (Sumitomo Heavy Industries Ltd.) and a temperature controller from Lakeshore Cryotronics, Inc. Data were collected in constant acceleration mode in transmission geometry with an applied field of 47 mT parallel to the γ-rays.

Other Measurements
Elemental analyses were performed by the Campbell Microanalytical Laboratory, Chemistry Department, University of Otago, New Zealand. High resolution mass spectra (HR-MS) were performed on an Agilent 6520 Accurate-Mass Q-TOF spectrometer on acetonitrile solutions.
Variable temperature ultraviolet-visible (UV-Vis) absorption spectra were recorded on an Agilent Cary 60 UV-Vis spectrophotometer in the range 220-1000 nm. Infrared spectra (KBr disk) were recorded on a Bruker Tensor 27 FTIR spectrometer. Band intensities are described as strong (s), medium (m), weak (w) or shoulder (sh).

Density Functional Theory Calculations
The DFT calculations were performed using the Gaussian 09 program package 42 with the UTPSSh functional 43,44 and the 6-311++G(d,p) basis set. This methodology reproduces well the energy parameters of spin-state switching rearrangements. [45][46][47] The stationary points on the potential energy surfaces (PESs) were located by full geometry optimization with calculation of force constants and checked for the stabilities of DFT wave function. Exchange coupling of unpaired electrons in the paramagnetic centers was estimated using the "broken symmetry" (BS) approach. 48 The exchange coupling constants J (in cm −1; Ĥ = -2JS1×S2) were calculated with the use of Yamaguchi equation. 49 Structural visualizations were prepared using the ChemCraft software 50 with the calculated atomic coordinates as input parameters (Supplementary Information).

Syntheses
The synthesis of salts of complexes 1-4 was based on the previously reported synthesis of 5 and 6, 12 involving first binding the pyridinophane ancillary ligand to the divalent metal center, followed by addition of deprotonated 3,5-di-tert-butylcatechol. The first stages of the reaction are performed under inert atmosphere to ensure ligand binding prior to oxidation of the metal. The addition of an equivalent of ferrocenium tetraphenylborate to a suspension of the metal complex affords clean oxidation of the metal centers, allowing ready isolation of tetraphenylborate salts of the target complexes in good yield. This was an improvement over the previously reported method 13 of oxidation with ferrocenium tetrafluoroborate followed by metathesis to exchange the counteranion (for the literature complex 6, the dioxolene ligand oxidizes in preference to the cobalt). 12 Allowing the complexes to instead oxidize by exposure of the reaction mixture to air affords less tractable crude products. All compounds are soluble in a range of common organic solvents. The purity of the bulk samples is evident from elemental analysis.

Structure Description
The single crystal X-ray diffraction data for compounds 1a×dioxane, 2a×2dioxane, 3a×1.5dioxane and 4a×2MeOH are available in Table 1 Table 2.  Compounds 1a×dioxane and 4a×2MeOH crystallize in the monoclinic space group P21/n while compounds 2a×2dioxane and 3a×1.5dioxane crystallize in the triclinic space group P 1 # . In all cases, the asymmetric unit contains a mononuclear metal complex, two counteranions and solvent molecules. The metal complexes 1-4 exhibit the same molecular connectivity and in each case the metal center posseses a distorted coordination geometry that continuous symmetry measurements indicate is closest to octahedral (Table S1). 51 Cobalt complexes 1 and 2 are less distorted from octahedral geometry than the Fe and Mn analogues 3 and 4.
The macrocyclic ligands bind such that the two sites trans to the oxygen atoms of the dioxolene are occupied by the pyridyl nitrogen atoms. Detailed consideration of the metal and dioxolene-based structural parameters (Table 2) 19 Similarly the Fe-O/N distances in 4a×2MeOH are consistent with HS-Fe(III) and comparable to previously reported HS-Fe(III) complexes of L-N4R2 and catecholate ligands. 14,15,52 Empirical bond valence sum calculations indicate trivalent metal centers for all compounds.

Infrared Spectroscopy
Recent detailed spectroscopic studies of cobalt-dioxolene complexes with tetradentate N-donor ancillary ligands have facilitated the assignment of the catecholate-based bands in the midinfrared, although bands that can be definitively attributed to the semiquinonate state are less clear. 13,54 The infrared spectra for 1a×dioxane, 2a×2dioxane, 3a×1.5dioxane and 4a×dioxane, measured at room temperature as pressed KBr disks (Fig. S3),  11,14,16 In summary the room temperature infrared spectra are consistent with the presence of catecholate ligands, although some semiquinonate character cannot be ruled out.

Electronic Absorption Spectroscopy
Following confirmation of the presence of complexes 1-4 in acetonitrile solutions of compounds 1a-4a by mass-spectrometry studies, room temperature electronic absorption spectra (Fig. 2) in 18 the UV-Visible range (280-1000 nm) were measured in acetonitrile, butyronitrile and 1,2dichloroethane. For compounds 1a-3a, the spectra remained unchanged over a period of at least several hours under ambient conditions, consistent with solution stability and insensitivity to oxygen. In contrast, the spectra for 4a change rapidly in air, but not under an nitrogen atmosphere, consistent with the catalytic reaction of the coordinated dbcat 2-ligand with oxygen reported previously for the L-N4Me2 analogue of this complex. 52 The data are tabulated in Table S2, with band assignments where possible. 19 The spectra obtained for cobalt compounds 1a in acetonitrile and butyronitrile and 2a in all solvents are characteristic of LS-Co(III)-catecholate chromophores. 24,55 A band between 400 and 420 nm is attributed to a d-d transition for six-coordinate Co(III) and a low intensity broad feature between 600 and 1000 nm has been assigned as due to ligand to metal charge transfer (LMCT). 55 The spectrum for 1a in 1,2-dichloroethane exhibits some additional features, a more intense band around 400 nm, several sharp bands between 500 and 700 nm, a broad feature between 700 and 800 nm as well as the onset of a transition into the near infrared. These features are all suggestive of a HS-Co(II)-semiquinonate chromophore. [11][12][13]24,55,56 The band at ~ 400 nm is assigned to a Co(II)  The room temperature electronic absorption spectra for manganese compound 3a vary little across the three solvents used (Fig. 2) and are very similar to the spectra of 1a in acetonitrile and butyronitrile. This is consistent with a Mn(III)-catecholate chromophore. 16,[57][58][59][60][61] Variable temperature spectra were also measured for 3a in butyronitrile (293-333 K) and show little temperature dependence, consistent with no VT or SCO in these temperature ranges (Fig. S4). The room temperature spectra for iron compound 4a in the different solvents (Fig. 2) are very similar to the 21 spectrum reported for [Fe III (L-N4Me2)(dbcat)], 52 exhibiting a pair of broad bands in the range 500-1000 nm that are more intense than analogous bands for the cobalt and manganese analogues.
Similar spectral features are evident for other Fe(III)-catecholate complexes and have been assigned as LMCT transitions. 62 Variable temperature spectra were also measured for 4a in dichloroethane (293-333 K), but little temperature dependence is evident (Fig. S4), consistent with no VT or SCO in this temperature range.

Electrochemistry
Following confirmation of solution stability via electronic absorption spectroscopy, cyclic and rotating disk electrode (RDE) (Fig. 4) voltammograms were measured for 1 mM solutions of 1a, 2a, 3a and 4a in acetonitrile using a glassy carbon electrode. All potentials are referenced to the ferrocene/ferrocenium couple using ferrocene as an internal standard or measured immediately afterwards.
22 Cyclic voltammetry for compounds 1a, 2a, 3a and 4a reveals between two and four resolved oxidation processes (labelled ox1 to ox4) and two reduction processes (labelled red1 and red2) in the accessible potential window. In each case the initial potential was selected to coincide with the measured rest potential and the position of zero current in the RDE voltammograms confirms the nature of the processes as oxidations or reductions. The mid-point potentials (Em) tabulated in Table 3  complexes. 14,16,19 In all cases, the limiting currents in the RDE voltammograms for ox1 and red1 are very similar (Table 3), consistent with both processes involving the transfer of one electron. A second irreversible (1a) or quasi-reversible (3a, 4a) process, red2, is evident at much more negative potentials and is assigned as originating from the pyridinophane ligands, the RDE limiting currents suggest that this may also be a 1e-process. 19 For all compounds, the second oxidation, ox2, is irreversible and assigned to the oxidation of the tetraphenylborate counterion. 63 The remaining oxidations, ox3 and ox4, show some degree of reversibility in terms of current response on the reverse sweep for compounds 1a and 3a. Processes ox2 and ox3 are merged in the RDE voltammograms. Definitive assignment of processes ox3 and ox4 is not possible, but it is likely that one of these processes is due to oxidation of the dbsqligand to the neutral benzoquinone form, while another process may be attributed to the 1e-oxidation from the trivalent to the tetravalent metal ion. 24

25
For metal-dioxolene complexes that exhibit VT transitions, the separation in redox potential between the catecholate oxidation and metal-based reduction correlates with the accessibility and transition temperature of a VT interconversion. 55 The differences between the Em values for ox1 and red2 are tabulated in Table 3. We have previously suggested that a D(ox1-red1) value of less than ~ 0.74 V is a good indicator for an accessible VT transition. 28

Solid State Magnetic Measurements
Magnetic susceptibility data were collected on crushed crystalline samples of 1a×dioxane, 2a×2dioxane, 3a×1.5dioxane and 4a×dioxane with an applied field of 0.1 T. For 1a×dioxane and 2a×2dioxane data were collected upon cooling from 200 K to 2 K, then heating from 2 K to 400 K and finally cooling from 400 K back down to 2 K; for 3a×1.5dioxane and 4a×dioxane data were collected upon cooling from 200 K to 2 K and then upon heating up to 340 K. The data 26 obtained upon first cooling and heating overlay within error in all cases. Magnetization data for 3a×1.5dioxane and 4a×dioxane were measured at 2 and 4 K with magnetic fields up to 5 T.
Compound 1a×dioxane is diamagnetic between 2 K and 200 K, with a small cMT product value of 0.1-0.2 cm 3 K mol -1 (Fig. 5), which is consistent with a LS-Co(III)-dbcat formulation. As the temperature increases above 200 K there is sharp increase in the cMT product, indicating the formation of a paramagnetic species and consistent with the onset of a VT interconversion. At the highest measured temperature of 400 K the transition is incomplete and the cMT product is 0.55 cm 3 K mol -1 , much less than the value of 3.0-3.8 cm 3 K mol -1 expected for the HS-Co(II)-dbsq complex that would arise from a complete VT transition. 55,64-66 Upon cooling from 400 K back down to 2 K, the cMT profile is similar to the initial data set although the values are around 0.04 cm 3 mol -1 K higher. The discrepancy in profiles on heating and cooling above room temperature is likely due to desolvation, but is still suggestive of reversibility of the VT interconversion. For compound 2a×2dioxane, the cMT product value of 0 cm 3 K mol -1 at 2 K increases gradually to 0.15 cm 3 K mol -1 at 400 K. This is consistent with a diamagnetic LS-Co(III)-dbcat formulation across the entire temperature range with the presence of a temperature-independent-paramagnetism contribution, that is well-known for Co(III). 67 While VT interconversions in cobalt-dioxolene complexes can also be induced by irradiation with visible light at low temperatures, 10 equivalent photo-induced SCO has yet to be reported for cobalt complexes that exhibit thermally-induced SCO. 68 In the present case, the high temperature of the onset of the thermally-induced VT interconversion for 1a renders photo-induced VT unlikely and photomagnetic studies were not pursued. 26,69   For 4a×dioxane, the thermal-dependence of cMT is qualitatively similar to that observed for 3a×1.5dioxane (Fig. 6). The cMT value of 4.7 cm 3 K mol -1 between 25 and 340 K is consistent with the HS-Fe(III)-dbcat (S = 5/2) ground state suggested by the 100 K structural data measured for 4a×2MeOH. There is no sign of a SCO or VT transition, despite the fact that the {N4O2} donor set sometimes affording SCO for Fe(III). [74][75][76] Below 25 K, cMT decreases to a value of 3.3 cm 3 K mol -1 at 2 K. As for 3a, the 2 and 4 K isotherms in the reduced magnetization plots 29 for 4a do not superimpose (Fig. S5), again suggesting considerable single ion anisotropy.
Simultaneously fitting the susceptibility and magnetization data required only the inclusion of axial ZFS, affording g = 2.07(1) and D = 3.62(2) cm -1 , with this D value consistent with values reported previously for significantly distorted octahedral HS-Fe(III) complexes. 77,78 Efforts to fit the data with a negative D value afforded a poorer fit. Definitive determination of the sign of D might be possible with low temperature EPR spectroscopy. 79 The g value obtained from fitting the magnetic data is higher than the value of 2.0 expected for HS-Fe(III), which might be due to mixing of the HS-Fe(III)-dbcat with some HS-Fe(II)-dbsq component, as has been observed previously for iron catecholate systems. 80

Solution State Magnetic Measurements
Solution magnetic susceptibility data were measured by the Evans NMR method on solutions of 1a×dioxane in 1,2-dichloroethane-d4, 1,2-dibromoethane-d4 and acetonitrile-d3 (Fig. 5). Data were measured across different temperatures ranges depending on the liquid range of the solvent and the solubility of the compound. In each case an increase in cMT is observed with increasing temperature, consistent with the onset of a VT interconversion in solution, however the transition is incomplete in all three solvents, with a largest measured cMT value of 2.6 cm 3 K mol -1 , compared to a value of around 3.2 cm 3 K mol -1 expected for a HS-Co(II)-dbsq species. 28 The transition temperature increases in the order 1,2-dichloroethane < 1,2-dibromoethane < acetonitrile. This solvent-dependence is consistent with the variable temperature solution electronic absorption spectra. Previous studies of both VT and SCO systems in solution have found that the transition temperature dependence does not simply follow the variation in dielectric constant, but instead can depend on specific solvent-complex interactions, including hydrogen bonding, that differ between the two electromers or spin states. [81][82][83][84] In the present case, the origin of the trend is not clear, but the lower transition temperature in chlorinated solvents versus acetonitrile is consistent with observation for VT interconversions in other cobaltdioxolene complexes. 28,82 Mössbauer Spectroscopy 57 Fe Mössbauer spectra were measured for compound 4a×dioxane as a function of temperature.
The spectra show a single broad quadrupole doublet with parameters (d = 0.43, DEQ = 2.08 mm/s) consistent with HS-Fe(III) and therefore in agreement with the magnetic data. However, the asymmetry of the spectra is temperature dependent (Fig. 7) with the lower energy absorption broadened significantly at low temperatures (< 30 K) and the higher energy absorption broadened significantly at higher temperatures (> 30 K). Different samples and different sample preparation produced consistent and reproducible data. The origin of this effect is currently unknown, although it could be caused by an unusual Goldanskii-Karyagin effect and this will be investigated in future studies. 85 Similar asymmetry has been observed previously for iron(III)-catecholate complexes. 80 closed-shell LS-Co(III)-dbcat isomer on the singlet PES is the ground state, 47 consistent with experimental data measured for the tetraphenylborate compound 5a×0.8MeCN×0.2Et2O. 12 In contrast, calculations for 6a indicate a LS-Co(II)-dbsq isomer in the ground state capable of a SCO transition to the HS-Co(II)-dbsq electromer, 47 which was observed experimentally. 12,46 Comparison of the energies of the three tautomers for the four cobalt complexes is interesting (Fig. 8). As the alkyl substituents on the pyridinophane ligand increases in bulk from methyl (5) to ethyl (2) to iso-propyl (1), the LS-Co(II)-dbsq and HS-Co(II)-dbsq both decrease in energy relative to the LS-Co(III)-dbcat, with a VT interconversion thus most favorable for 1.
Upon further increase to the tert-butyl substituents in 6, there is a reversal in the energies, with the LS-Co(II)-dbsq electromer stabilized by 4.7 kcal mol -1 over the LS-Co(III)-dbcat form, establishing the conditions for SCO for 6 rather than the VT transition that is evident for 1.
According to the calculation results of the BS states, the variation of alkyl subsitituents on the pyridinophane ligands does not significantly impact the nature of the exchange interactions in the isomers of cobalt complexes 1, 2, 5 and 6: the HS-Co(II)-dbsq electromer is characterized by moderate antiferromagnetic exchange in all cases compared to very strong ferromagnetic exchange for the LS-Co(II)-dbsq tautomers (Table S4).
For the manganese compound 3a, three redox isomers were found by DFT. The Thus, a VT transition is possible in the manganese complex above room temperature, but it will not be accompanied by large changes of the magnetic characteristics in view of strong antiferromagnetic exchange in the isomer HS-Mn(II)-dbsq. 6 80 In the present case, the lowest energy electromer was found to be the LS-Fe(III)-dbcat with the IS-Fe(III)-dbcat next highest in energy. However, the experimental bond lengths determined at 100 K for 4a×MeOH (Table 2) are in closest agreement with the values calculated (optimized geometry) for HS-Fe(III)-dbcat, which is also consistent with the magnetic susceptibility up to 340 K (Fig. 6), low temperature magnetization measurements (Fig. S5) and variable temperature Mössbauer spectroscopic data measured for 4a×dioxane (Fig. 7).
For the assignment of the electrochemical processes, the geometry optimization of the cationic complexes in the reduced (red) and oxidized (ox) forms has been performed. In  (Fig. S12). Thus, analysis of spin density distributions and the calculated bond lengths of the dioxolene fragments unambiguously indicate that for all four complexes, the first 1ereduction is metal-centered, while the first 1e-oxidation occurs at the ligand. The calculated (Table S6) differences between the redox potential (D(ox1-red1)) for couples 1 0 /1 + (2 0 /2 + ) and 1 + /1 2+ (2 + /2 2+ ) are equal to 0.76 V and 0.87 V for 1a and 2a, 93,94 respectively, which are in excellent agreement with the measured values of 0.73 and 0.86 V (Table 3). Solution voltammetric measurements indicate rich redox chemistry for all four complexes. Notably the two cobalt complexes exhibit a reversible first oxidation that is ligandbased and first reduction that is metal-based. The potential separation between these processes is less for [Co(L-N4iPr2)(dbdiox)] + than for [Co(L-N4Et2)(dbdiox)] + . The relative values for the two complexes are consistent with a proposition previously made by some of us, that VT transitions can be expected when the potential separation between the "frontier" ligand-and metal-based redox processes is less than around 740 mV. 28 38 Density functional theory calculations have provided important insights in this work. This work represents the first systematic study of a series of members of a structural family of cobalt-dioxolene complexes that can exhibit thermally-induced SCO or VT transitions. By combining experimental investigations with DFT calculations, we have been able to elucidate how the relative energies of the different electomeric forms vary with the steric bulk of the substituents on the non-dioxolene co-ligand. Key to the unusual behavior for this family of cobalt-dioxolene complexes is the highly constrained nature of the 12-membered pyridinophane azamacrocyclic ligands, which can facilitate SCO as well as the more commonly observed VT interconversions.