Di‐Iron(II) [2+2] Helicates of Bis‐(Dipyrazolylpyridine) Ligands: The Influence of the Ligand Linker Group on Spin State Properties

Abstract Four bis[2‐{pyrazol‐1‐yl}‐6‐{pyrazol‐3‐yl}pyridine] ligands have been synthesized, with butane‐1,4‐diyl (L 1), pyrid‐2,6‐diyl (L 2), benzene‐1,2‐dimethylenyl (L 3) and propane‐1,3‐diyl (L 4) linkers between the tridentate metal‐binding domains. L 1 and L 2 form [Fe2(μ−L)2]X4 (X−=BF4 − or ClO4 −) helicate complexes when treated with the appropriate iron(II) precursor. Solvate crystals of [Fe2(μ−L 1)2][BF4]4 exhibit three different helicate conformations, which differ in the torsions of their butanediyl linker groups. The solvates exhibit gradual thermal spin‐crossover, with examples of stepwise switching and partial spin‐crossover to a low‐temperature mixed‐spin form. Salts of [Fe2(μ−L 2)2]4+ are high‐spin, which reflects their highly twisted iron coordination geometry. The composition and dynamics of assembly structures formed by iron(II) with L 1−L 3 vary with the ligand linker group, by mass spectrometry and 1H NMR spectroscopy. Gas‐phase DFT calculations imply the butanediyl linker conformation in [Fe2(μ−L 1)2]4+ influences its spin state properties, but show anomalies attributed to intramolecular electrostatic repulsion between the iron atoms.

Three helicate conformations are observed in the four structures, which differ in the torsions of their butanediyl linker groups. In 1[BF 4 ] 4 · 2MeNO 2 [ Figure 1, molecule (a)] both butanediyl linkers have two gauche torsions. Both its iron atoms are low-spin at the temperature of measurement (125 K; Table 1). In contrast, 1[BF 4 ] 4 · 2MeCN · Et 2 O [molecule (c)] has just one gauche torsion at each butanediyl group. One iron atom in that crystal is high-spin and the other is low-spin at 125 K. The complex in 1[BF 4 ] 4 · mMeNO 2 also adopts conformation (c), and is predominantly high-spin at that temperature ( Figure S17, Table 1). Lastly, the helicate in 1[BF 4 ] 4 · nMe 2 CO [molecule (b)] contains one of each butanediyl group conformation, and was measured at two temperatures. Both its iron atoms are highspin at 250 K. However, at 100 K Fe(1) adopts a mixed high/lowspin population, while Fe(2) has become fully low-spin (Table 1). Hence, the two iron centers in 1[BF 4 ] 4 · nMe 2 CO evidently undergo SCO at different temperatures on cooling.
The relative orientations of the two [Fe(bpp) 2 ] 2 + domains are quite similar in all these helicate conformations ( Figure 1). However, each additional butanediyl gauche torsion pushes the two iron atoms further apart, by 0.3-0.4 Å ( Table 1). The two solvates adopting conformation (c) exhibit almost identical Fe · · · Fe distances, implying this parameter may be only slightly perturbed by crystal packing effects. SCO in 1[BF 4 ] 4 · nMe 2 CO also has little effect on its Fe · · · Fe separation ( Table 1).
is 167.9(3)°; and, the least squares planes of the two bpp moieties bound to each metal atom (θ), which is 73.87(7)°. [45,46] An undistorted metal center of this type would have ϕ = 180 and θ = 90°. Crystalline, high-spin [Fe(bpp) 2 ] 2 + derivatives with comparable distortions to Fe(1) rarely exhibit thermal SCO, [47] and are kinetically trapped in their high-spin form upon cooling. [48] Hence, the distorted geometry at Fe(1) may explain the incomplete SCO in 1[BF 4 ] 4 · 2MeCN · Et 2 O ( Figure 2). The other iron atom in that structure, Fe (2), is low-spin at the temperature of measurement and adopts a more regular coordination geometry, as expected. The polycrystalline 1[BF 4 ] 4 solvatomorphs decompose to powders when exposed to air, leading to significant structural changes or loss of crystallinity by powder diffraction (Figur-es S20-S22). Elemental analysis implies some lattice solvent is retained, or replaced by atmospheric moisture, in the air-dried solids. Magnetic susceptibility data show the materials undergo gradual thermal SCO, although each is different in form ( Figure 3). All the magnetic data are reversible in cooling and warming temperature ramps, and so are not affected by in situ solvent loss. While no single crystals of 1[ClO 4 ] 4 were achieved, samples of that salt obtained from the same solvents show comparable X-ray powder patterns and SCO profiles to their BF 4 À analogues ( Figures S23-S25).
Most interestingly, air-dried 1[BF 4 ] 4 · nMe 2 CO is high-spin at 300 K and undergoes stepwise SCO on cooling, with an abrupt discontinuity near 160 K corresponding to 50 % conversion. That is consistent with the crystallographic observation that the two iron sites in that material undergo SCO at different temperatures. The transition is ca. 80 % complete at 100 K in the magnetic data, but shows a residual paramagnetism below that temperature. That implies kinetic trapping of the remaining material in its high-spin state below 100 K, [50] which is often observed in [Fe(bpp) 2 ] 2 + derivatives whose SCO extends to such low temperature. [47,51] A mixture of the nitromethane solvates of 1[BF 4 ] 4 · mMeNO 2 undergoes significant structural changes during air-drying by powder diffraction. The dried material is low-spin below 100 K, and shows a very gradual continuous SCO above that temperature such that ca. 30 % of its iron atoms are high-spin at 300 K ( Figure 3). Interestingly, that is more consistent with the crystal structure of the minor solvatomorph 1[BF 4 ] 4 · 2MeNO 2 , which is low-spin at 125 K, than with the major crystal form 1[BF 4 ] 4 · mMeNO 2 . Conversely, air-dried 1[BF 4 ] 4 · 2MeCN · Et 2 O is poorly crystalline and is high-spin at room temperature, showing a gradual SCO on cooling which is ca. 30 % complete at 100 K.
The N atom of the central pyridyl ring of each ligand is oriented towards an open face of an iron atom, but at a distance of Fe · · · N = 3.1-3.2 Å which is too long for a significant covalent interaction. This coordination geometry implies the 2X 4 salts should remain high-spin on cooling ( Figure 2), which was confirmed by magnetic measurements ( Figure S32).
The helicate structure in [Fe 2 (μÀ L 2 ) 2 ] 4 + is further stabilized by one short and two longer intramolecular π · · · π interactions, between aromatic residues on each ligand (Figures S29-S30, Table S6). An additional intermolecular π · · · π interaction in both crystals associates the helicate cations into centrosymmetric dimers (the crystals are racemic, containing equal numbers of Λ and Δ helical molecules in their asymmetric unit).
Reaction of L 3 with 1 equiv. of the same iron(II) salts yielded glassy orange solids, which were perfectly amorphous by powder diffraction. These solids analyzed reasonably to the empirical formulae [Fe(L 3 )]X 2 (3X 2 ; X À =BF 4 À or ClO 4 À ), with some included lattice solvent. The amorphous materials are essentially high-spin at room temperature, and show very gradual SCO equilibria on cooling in ca. 15 [52,53] Weak higher mass peaks from pentameric and hexameric assemblies are also visible for 3[ClO 4 ] 2 . Hence, solutions of all three complexes are a mixture of assembly structures under these conditions. The spectrum of 1[ClO 4 ] 2 shows more fragmentation than the other complexes, including metal-free L 1 which is not present in the other spectra. That is consistent with the higher lability of 1[ClO 4 ] 2 observed by NMR (see below).
The 1 H NMR of 1[ClO 4 ] 2 in CD 3 CN shows just one paramagnetic, C 2 -symmetric L 1 environment ( Figure S38). The butanediyl CH 2 groups are diastereotopic in the spectrum, which is consistent with the chirality of the helicate structure. In contrast the 1 H NMR of 2[ClO 4 ] 2 in CD 3 CN contains one principal C 2 -symmetric L 2 environment, but with a second paramagnetic L 2 -containing species comprising 10-15 % of the sample by integration ( Figure S39). Neither spectrum has peaks in the diamagnetic region from uncoordinated ligand, or dangling bpp residues from mono-coordinated L 1 or L 2 . These data imply interconversion of the assembly structures detected by mass spectrometry occurs rapidly in solution when L=L 1 , giving a time-averaged NMR spectrum, but is slower than the NMR timescale when L=L 2 .

Chemistry-A European Journal
Since a solvent correction could not be included in our calculations, [54] this anomaly was addressed by gas-phase minimizations of the isoelectronic charge-neutral molecules [Cr 2 (μÀ L) 2 ] 0 and [CrL] 0 (Tables S8-S9). The computed structures of the low-spin chromium complexes agree with expectation. However mixed-spin or high-spin chromium(0) minimizations yielded results that are more consistent with chromium(II) centers coordinated to [L * ] À ligand radicals. That was evidenced by their chromium coordination geometries, which are strongly Jahn-Teller-elongated (Tables S12-S13); and, by their α and β HOMO orbitals, which are ligand-centered in these chromium minimizations but metal-centered in their iron(II) counterparts ( Figures S46-S47). [55] Within that generalization, differences between the computed high-spin chromium centers suggest additional subtleties, which are beyond the scope of this study. [61] Because of these ambiguities, only the minimizations of the low-spin chromium complexes are analyzed in detail.
The three conformations of the iron complex, in a given spin state, are within 2 kcal mol À 1 of each other by this protocol. The difference is smaller for low-spin [Cr 2 (μÀ L 1 ) 2 ] 0 , where conformations (a)-(c) lie within 0.7 kcal mol À 1 (Table S9). Hence, all these conformations should be thermally accessible at room temperature, as observed. More detailed discussion of the minimized structures is not justified however, because of the ambiguities noted above.
The absolute spin state energies from a protocol like this are inaccurate, so the ΔE rel {HSÀ LS} energies in Tables 3 and S9 are scaled relative to conformation (a) of the relevant [M 2 (μÀ L 1 ) 2 ] 2z + molecule. A molecule with a positive ΔE rel {HSÀ LS} has a more stable low-spin state than conformation (a), and so should exhibit a higher T1 = 2 value. Similarly, a negative ΔE rel {HSÀ LS} implies T1 = 2 should be lower than conformation (a). By this measure, T1 = 2 for the conformations of [Fe 2 (μÀ L 1 ) 2 ] 4 + should follow the trend of (a) > (b) > (c). That is broadly consistent with the crystallographic and magnetic properties of the 1[BF 4 ] 2 solvates (Table 1, Figure 3).
The mixed-spin forms of [Fe 2 (μÀ L 1 ) 2 ] 4 + have almost identical energy to an equimolar mixture of high-spin and low-spin Table 3. Computed energies of the high-spin (HS, S = 4), mixed-spin (MS, S = 2) and low-spin (LS, S = 0) states of the iron complexes in this work. Energies of the corresponding chromium complex minimizations are listed in Table S9.

Chemistry-A European Journal
Research Article doi.org /10.1002/chem.202202578 molecules, to within 0.5 kcal mol À 1 (ΔE{HSÀ LS, MS}, However, all the spin states of [M 2 (μÀ L 2 ) 2 ] 2z + (M z + =Fe 2 + or Cr 0 ) minimized to a more symmetric L 2 ligand conformation than found crystallographically for 2[ClO 4 ] 2 . This places the metal atoms further apart, and with a less distorted coordination geometry than found experimentally. [55] Since the structure should reflect both the conformational preferences of L 2 and ligand field effects on the metal geometry, our DFT protocol may over-estimate the ligand field contribution to the structure of this molecule. Consistent with that, molecular mechanics minimizations of [M 2 (μÀ L 2 ) 2 ] 2z + , which exclude ligand field considerations, reproduced the experimental conformation of 2[ClO 4 ] 2 more accurately ( Figure 6).
Minimizations of the [M 2 (μÀ L 3 ) 2 ] 2z + helicate were also investigated. [55] These also revealed conformational flexibility in the xylyl linker group, placing the metal atoms in the low-spin iron complex 0.9 Å further apart than in the chromium compound ( Figure S48). MM2 minimizations reproduced the chromium complex conformation, so the extended conformation of [Fe 2 (μÀ L 3 ) 2 ] 4 + could again reflect electrostatic repulsion between its Fe 2 + ions. The high-spin state of [Fe 2 (μÀ L 3 ) 2 ] 4 + is significantly stabilized compared to [Fe 2 (μÀ L 1 ) 2 ] 4 + (Table 3), which is consistent with the high-spin nature of 3X 2 at room temperature. However, since no crystallographic data are available for the Fe/L 3 complex, the relevance of these results to its experimental properties is unclear.

Conclusion
Four new ditopic ligands have been synthesized, by connecting two 1,3-bpp metal-binding moieties with different spacers via their distal NÀ H groups. Two of these, L 1 and L 2 , cleanly afford [2 + 2] helicate complexes when complexed to iron(II). Three helical conformations of [Fe 2 (μÀ L 1 ) 2 ] 4 + were observed in different solvate crystals of 1[BF 4 ] 4 (Figure 1), which exhibit a range of spin state properties. These include a clear, unusual stepwise SCO of the two iron centers in 1[BF 4 ] 4 · nMe 2 CO, via a mixedspin intermediate which was detected crystallographically (Figure 3). [62] The DFT calculations imply conformations (a)-(c) of [Fe 2 (μÀ L 1 ) 2 ] 4 + should all exist in solution, while NMR showed they are in rapid chemical exchange at room temperature. Hence, the observation of different helicate conformations in crystals of 1[BF 4 ] 4 should simply reflect the crystal packing in each solvate.
In contrast, salts of [Fe 2 (μÀ L 2 ) 2 ] 4 + remain high-spin at all temperatures. That is explained by the molecular conformation shown by 2[ClO 4 ] 4 , which leads to the most distorted sixcoordinate geometries yet observed in the extended family of [Fe(bpp) 2 ] 2 + SCO materials (Figures 2 and 4). Analytically pure iron(II) complexes of L 3 were also obtained, which are however completely amorphous in the solid state.  4 ] 8 + (L=L 1 À L 3 ) by mass spectrometry, with higher nuclearity species also being present in some cases ( Figures 5 and S35-S37). Hence, the helicate complexes exist in equilibrium with other assembly structures in solution. That being the case, 1 H NMR implies those assemblies interconvert in solution more rapidly when L=L 1 than for the more rigid L=L 2 or L 3 . Hence, the identity of the spacer group strongly influences the composition and dynamics of the supramolecular assemblies formed by L 1 À L 3 .
Gas phase DFT calculations confirm conformations (a)-(c) of [Fe 2 (μÀ L 1 ) 2 ] 4 + have almost identical energies, but should show detectably different spin state properties. These appear consistent with experiment, in that the computed spin state energies mirror the observed trend in solid state SCO temperatures, of conformation (a) > (b) > (c) (Figure 3, Table 3). The calculations also reproduce the high-spin nature of [Fe 2 (μÀ L 2 ) 2 ] 4 + . However in other respects the calculations present anomalies, which are consistent with the influence of intramolecular electrostatic repulsion between the positively charged iron ions. Such effects would be compensated in condensed phases, by ion pairing and weaker intermolecular Figure 6. Computed structures of [Fe 2 (μÀ L 2 ) 2 ] 4 + . Top: DFT energy minimization of the high-spin (S = 4) iron complex (Fe · · · Fe = 7.418 Å). Bottom: molecular mechanics geometry minimization (M · · · M = 5.356 Å). The molecular mechanics calculation is closer to the experimental structure (Table 2, Figure 4). Color code: C, gray; H, white; N, blue; Fe or M, green. dipolar interactions with the surrounding medium. That could lead to significant discrepancies between the results of our single point calculations and experiment, as observed. Further calculations of isoelectronic [Cr 2 (μÀ L 1 ) 2 ] 0 and [Cr 2 (μÀ L 2 ) 2 ] 0 support that view, but were themselves only partly successful because the high-spin chromium(0) centers undergo valence tautomeric oxidation to chromium(II) in silico.
Other gas phase DFT studies on dinuclear [63] or higher nuclearity [14,53,64] SCO complexes have investigated more conformationally rigid molecules, many of which are electroneutral. The influence of intramolecular electrostatic effects on the calculations should be less apparent in those cases. However, a recent gas phase DFT study of Fe 4 grid complexes noted that the high-spin states of more highly charged molecules in that study were over-stabilized computationally, compared to their uncharged analogues. [14] Electrostatic interactions between the iron atoms in those molecules were proposed to contribute to that discrepancy. In contrast, the low-spin state of the dinuclear complexes appears to be overstabilized in this work (Tables 3  and S9), although we base that observation on different criteria from those in Ref. [14]. In other respects, our results are consistent with the conclusions from that earlier study. [65] Experimental Section Instrumentation: Solid state magnetic susceptibility measurements were performed with freshly isolated, unground polycrystalline samples, using a Quantum Design MPMS-3 SQUID/VSM magnetometer in an applied field of 5000 G. Samples were protected against solvent loss by saturating the tightly sealed MPMS-3 powder capsules with diethyl ether vapor, Unless otherwise specified, the measurements employed a temperature ramp of 5 Kmin À 1 . Diamagnetic corrections for the samples were estimated from Pascal's constants; [66] a previously measured diamagnetic correction for the sample holder was also applied to the data.
Elemental microanalyses were performed by the microanalytical services at London Metropolitan University School of Human Sciences. Electrospray mass spectra were recorded on a Bruker MicroTOF-q instrument from chloroform (organic ligands) or acetonitrile (metal complexes) solution. The peak simulations in Figures 5, S36 and S37 were plotted with ORIGIN, [67] starting from simulations of the individual component species produced by Bruker Compass. [68] Diamagnetic NMR spectra employed a Bruker AV3HD spectrometer operating at 400.1 ( 1 H) or 100.6 MHz ( 13 C), while paramagnetic 1 H NMR spectra were obtained with a Bruker AV3 spectrometer operating at 300.1 MHz. X-ray powder diffraction measurements were obtained at room temperature from a Bruker D2 Phaser diffractometer, using Cu-K α radiation (λ = 1.5419 Å).
All calculations were performed by using SPARTAN'18. [69] DFT calculations employed the B86PW91 functional and def2-SVP basis set. Low-spin systems were treated as spin restricted and high-spin systems as spin unrestricted. The calculations were performed in the gas phase, since a solvent gradient for iron is not implemented in SPARTAN'18. Crystallographic atomic coordinates for the different conformations of 1 2 + , and for 2 2 + , were used as a starting point for those geometry minimizations. Otherwise, initial models were constructed de novo in the program, then subjected to a preliminary molecular mechanics minimization before the full DFT energy minimization was undertaken.

Synthesis of {[Fe(L 3 )][ClO
Crystal Structure Analyses: Crystals of 1,3-bpp were grown by slow evaporation of an NMR sample of the compound in CDCl 3 . The solvent-free crystals 2[ClO 4 ] 4 were obtained by slow diffusion of diethyl ether vapor into a filtered solution of the complex in acetone. The other solvated crystals were grown similarly, by diethyl ether vapor diffusion in the appropriate solvent. Diffraction data for 1[BF 4 ] 4 · nMe 2 CO were recorded at station I19 of the Diamond synchrotron (λ = 0.6889 Å). All other diffraction data were measured with an Agilent Supernova dual-source diffractometer using monochromated Cu-K α (λ = 1.5418 Å) radiation. The diffractometer was fitted with an Oxford Cryostream low-temperature device.
Crystallographic experimental details and refinement protocols are given in the Supporting Information. All the structures were solved by direct methods (SHELXS [70] ), and developed by full least-squares refinement on F 2 (SHELXL-2018 [70] ). Crystallographic figures were prepared using XSEED, [71] while calculation of structural indices and preparation of publication materials was performed with Olex2. [72] Deposition Number ( 4 ) and 2169637 (2[ClO 4 ] 4 · 3MeNO 2 · 0.75H 2 O) contain(s) the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.