Designed intramolecular blocking of the spin crossover of an Fe ( II ) complex †

The exploitation of spin crossover (SCO) offers an opportunity for the implementation of a molecular-based switching property at the nano-scale. The phenomenon is displayed by transition metals that exhibit two possible d electron configurations, and a ligand field causing these to lie close in energy, so that they can be easily interconverted through external stimuli. The spin transition occurs along the changes to the structure, together with variations in physical properties such as colour, magnetic or electrical properties, mechanical responses, etc. The dynamics of spin transition, crucial for exploiting the switching behaviour, is governed by elastic interactions between active centres within the host crystal lattice. Thus, the presence of a dense network of intermolecular or covalent interactions often leads to cooperative SCO, which translates into abrupt transitions, sometimes accompanied by hysteresis. Therefore, great efforts are dedicated to unveil the subtle links between SCO and the associated changes to the crystal lattice. In this context, structural analyses indicate that complexes with the appropriate ligand field to undergo SCO may be kinetically trapped in the high spin (HS) state if the transition requires excessive crystal lattice rearrangement energy, which is usually gauged by a very distorted coordination geometry. This was also demonstrated by means of DFT+U calculations in the solid state. Complementary to this, by studying the Fe(II) complexes of a family of indazolylpyridine ligands, it was proposed that the relative stability of the HS and low spin (LS) states may be strongly modulated by intra-ligand steric interactions. In a theoretical investigation, it was suggested that the true reason for the drastic differences within that series of compounds is indeed inter-ligand (while intra-molecular) steric factors. Along the same lines, following the study of a series of HS Fe(II)/1bpp derivatives, it was suggested that steric intramolecular interactions could be playing a role in blocking the SCO. With the aim of elucidating these factors, we have prepared and used a ligand that serves to demonstrate that a complex expected to show SCO can be completely trapped into the HS state by virtue of purely intra-molecular, non-covalent interactions and not due to a lattice effect. A thorough DFT analysis provides an elegant rationalization of this behaviour. The new ligand Ph1,3bpp (2-(3-phenylpyrazol-1-yl)-6-(1Hpyrazol-3-yl)-pyridine; Scheme 1) was prepared in three steps, starting from the coupling of 3-phenylpyrazole with 2-acetyl-6bromopyridine, followed by the isolation of the corresponding propanone intermediate that results from the activation of the acetyl group with 1,1-dimethoxytrimethylamine, which upon ring-closure with hydrazine yields the desired product (ESI†).

The exploitation of spin crossover (SCO) offers an opportunity for the implementation of a molecular-based switching property at the nano-scale. [1][2][3] The phenomenon is displayed by transition metals that exhibit two possible d electron configurations, and a ligand field causing these to lie close in energy, so that they can be easily interconverted through external stimuli. The spin transition occurs along the changes to the structure, together with variations in physical properties such as colour, 4 magnetic 5-7 or electrical 8 properties, mechanical responses, 9 etc. The dynamics of spin transition, crucial for exploiting the switching behaviour, is governed by elastic interactions between active centres within the host crystal lattice. Thus, the presence of a dense network of intermolecular or covalent interactions often leads to cooperative SCO, which translates into abrupt transitions, sometimes accompanied by hysteresis. 10,11 Therefore, great efforts are dedicated to unveil the subtle links between SCO and the associated changes to the crystal lattice. [12][13][14] In this context, structural analyses indicate that complexes with the appropriate ligand field to undergo SCO may be kinetically trapped in the high spin (HS) state if the transition requires excessive crystal lattice rearrangement energy, which is usually gauged by a very distorted coordination geometry. 15,16 This was also demonstrated by means of DFT+U calculations in the solid state. 17 Complementary to this, by studying the Fe(II) complexes of a family of indazolylpyridine ligands, it was proposed that the relative stability of the HS and low spin (LS) states may be strongly modulated by intra-ligand steric interactions. 18 In a theoretical investigation, it was suggested that the true reason for the drastic differences within that series of compounds is indeed inter-ligand (while intra-molecular) steric factors. 19 Along the same lines, following the study of a series of HS Fe(II)/1bpp derivatives, it was suggested that steric intramolecular interactions could be playing a role in blocking the SCO. 20 With the aim of elucidating these factors, we have prepared and used a ligand that serves to demonstrate that a complex expected to show SCO can be completely trapped into the HS state by virtue of purely intra-molecular, non-covalent interactions and not due to a lattice effect. A thorough DFT analysis provides an elegant rationalization of this behaviour.
The new ligand Ph1,3bpp (2-(3-phenylpyrazol-1-yl)-6-(1Hpyrazol-3-yl)-pyridine; Scheme 1) was prepared in three steps, starting from the coupling of 3-phenylpyrazole with 2-acetyl-6bromopyridine, followed by the isolation of the corresponding propanone intermediate that results from the activation of the acetyl group with 1,1-dimethoxytrimethylamine, which upon ring-closure with hydrazine yields the desired product (ESI †). Scheme (2), respectively) is also witness to the HS state. 21 This strong distortion could also cause the HS of this compound to be kinetically trapped.
The conformation of the Ph1,3bpp ligands within the complex must be emphasized; the phenyl substituents are in all the cases twisted with respect to their adjacent pyrazolyl rings with the angles of 47.71/46.56°and 49.24/47.75°, and establish intramolecular π⋯π interactions with their opposite 1,3bpp fragments in the complex (Fig. S2 †). These π⋯π interactions are slightly different in their strength and position. Their strength may be gauged by the distance between the Ph ring centroid and the idealized 1,3bpp plane (here 3.412 Å in average). The above two factors likely play a role in the pronounced mutual rotation of both 1,3bpp moieties away from orthogonality (angles between the idealized planes of 69.79 and 71.53°). The N-H groups of the complex interact through strong hydrogen bonds with the acetone molecule and one perchlorate anion, respectively. The other ClO 4 − group in the formula is subjected to weaker interactions within the lattice. This happens for both complexes of the asymmetric unit. The complexes are disposed in sheets that contain both possible enantiomers, where they interact with each other only via C-H⋯π interactions (Fig. S3 †). Thus, the organization within these sheets does not follow the compact network of C-H⋯π and π⋯π contacts known as the "terpy embrace" usually encountered in complexes with bpp type ligands. 22 Acetone and ClO 4 − anions are located in between the sheets of complexes, yielding a lattice that alternates essentially hydrophobic with rather hydrophilic layers (Fig. S4 †). The sheets of metal complexes feature two significantly different separations (amounting to 9.368 and 12.190 Å, respectively). Half of the ClO 4 − anions in the compound and all the molecules of acetone are located in the space between the layers with the largest separation, whereas the remaining anions lie in the other inter-sheet domain (Fig. 2). In contact with air, compound 1a exchanges all its acetone content with half equivalent of water molecules, in a single-crystal-to-single-crystal (SCSC) manner. This transformation does not lead to any apparent changes while single-crystal X-ray diffraction allows the determination of the molecular structure of the new product, [Fe(Ph1,3bpp) 2 ] (ClO 4 ) 2 ·0.5H 2 O (1b). Thermogravimetric analyses (TGA) also confirm the process (Fig. S5 †). This system joins a growing family of bpp-based Fe(II) complexes that experience SCSC solvent exchange processes allowing the use of SCXRD to study the initial and the final forms, [23][24][25][26] and sometimes even intermediate phases. 27 Compound 1b is found in the orthorhombic space group Pna2 1 , its asymmetric unit (Fig. S6 †) (Fig. 2). It is thus clear that the space located in between the largest inter-layer separation mediates the diffusion of molecules transiting in and out of the lattice. The exchange process causes a 9% contraction of the lattice. Contrary to most SCSC small-molecule exchanges reported for the family of bpp/Fe(II) complexes, 23-26 the 1a → 1b transformation does not produce a spin switching (at least down to 100 K) from the HS state. This was corroborated by means of magnetic susceptibility measurements down to 5 K (Fig. S7 †). Thus, under a constant magnetic field of 0.1 T, compounds 1a and 1b exhibit values of the χ M T product (χ M is molar paramagnetic susceptibility) of 4.04 and 3.94 cm 3 K mol −1 at 350 and 301 K, respectively, consistent with both systems being completely in the HS state (S = 2). The Curie Law behaviour holds in both cases down to below 50 K, from where the χ M T vs. T curve plots exhibit the sharp expected decline resulting from zero field splitting effects on HS Fe(II).
The ligand Ph1,3bpp was designed to prove that intramolecular interactions could be exploited to block the SCO and trap a complex in the HS state. The magnetic response of 1a and the fact that the exchange of acetone/water does not affect this response point in this direction. For comparison, the related complex [Fe(1,3bpp) 2 ](ClO 4 ) 2 (2), 28 which features the same core as 1a and 1b without the Ph substituent, exhibits a complete SCO close to 275 K (Fig. S7 †), which also supports this hypothesis. An elegant explanation for the SCO blocking of 1a and 1b was given by means of computational tools. Thus, the optimized geometries associated with the HS and hypothetical LS states of the complex [Fe(Ph1,3bpp) 2 ] 2+ (1) under gas-phase conditions were determined, starting from the crystal coordinates (Fig. S8 †). The calculated HS structure (1 HS ) is very similar to the experimental one, reproducing the Fe-N distances, and most notably, the distortion of the coordination geometry (average Fe-N distance, 2.191 Å; Σ = 165.04; Θ = 530). The structure obtained for the hypothetical LS state (1 LS ) features the expected compression of the FeN 6 core (average Fe-N value, 1.963 Å), which also becomes closer to an ideal octahedron (Σ = 99.26; Θ = 308). Interestingly, in both cases, the phenyl substituents of the ligands are twisted with respect to their respective 1,3bpp planes (Fig. S8 †), as also observed experimentally. This could be ascribed to a tendency for establishing favourable inter-ligand π⋯π interactions (see below). The calculated energies of both forms predict higher electronic enthalpy for 1 HS (ΔH elec = −1.7 kJ mol −1 ), which would indeed preclude the thermal SCO, since the latter is an entropy driven process. In order to compare with 2, single point energy evaluations were performed on 1 HS and 1 LS without the phenyl substituents. This results in a dramatic inversion of the relative HS vs. LS electronic enthalpies, now giving ΔH elec = +10.8 kJ mol −1 and restoring the thermodynamic stability of the LS state, consistent with the experimental results. 28 Both sets of calculations confirm that the blocking of the SCO in 1 is entirely related to the phenyl groups. The influence of the intramolecular π⋯π interaction was examined using several models. It was found that this interaction causes a difference in stability of only −0.1 to −0.5 kJ mol −1 in favor of the HS state. Such contribution is nearly two orders of magnitude smaller than the overall effect of ca. 12 kJ mol −1 caused by the Ph rings (by adding ΔH elec in systems 1 and 2, i.e. −1.7 and 10.8 kJ mol −1 ). Another difference between 1 HS and 1 LS , also related to the phenyl substituent of Ph1,3bpp, is the torsion angle between this substituent and their carrier pyrazolyl rings, with average values of 35.58°( 1 HS ) and 51.26°(1 LS ). The theoretical analysis of the energy of a phenylpyrazole moiety as a function of the torsion angle unveils a significant destabilization of the system as it is removed from planarity by rotation about the CC bond between both rings (Fig. 3), which indicates that this factor would favour the HS state. From this calculation, a change from 35°to 50°, as observed in the optimized structures, corresponds to a loss of ca. 4 kJ mol −1 .
Considering that 1 contains two such units, the effective stabilization of 1 HS on this account would be ca. 8 kJ mol −1 , close to the calculated ca. 12 kJ mol −1 HS vs. LS energy difference related to the Ph ring. The reason for an increased planarity loss upon HS to LS transition is the added steric hindrance between the ligands generated with the compression of the FeN 6 core (Fig. S8 †). The absence of this additional rotation would force, in the LS state, the presence of a 2.3 Å C⋯H contact, in addition to other C⋯C, C⋯H or H⋯H contacts near 2.5 Å, which is extremely unfavorable energetically. This analysis clearly indicates that the reason why the HS state is blocked for compounds 1a and 1b is the energy penalty that would signify the additional rotation of the Ph group upon SCO, as the only mechanism to avoid intramolecular, interligand steric hindrance. It must be mentioned that the torsion angles observed in 1a and 1b are larger than the angles predicted by the calculations for 1 HS , which would bring the former structures closer to the calculated 1 LS one than the latter. Comparing 1a and 1b with 1 LS would decrease the size of the "torque" energy that causes the trapping of the HS state. However, in the absence of any experimental LS structure, using both computed configurations to estimate the change in torsion that would occur upon SCO is the most reasonable option.
In light of the above results, it becomes important to identify in the literature other possible examples affected by the same effect. Indeed, two Fe(II)/R-1bpp complexes (R-1bpp = 2,6-bis-( pyrazol-1-yl)-pyridine with a substituent R in the position 3 of each pyrazole; R = Ph, CO 2 Et) 20 were found to exhibit the same behaviour as 1a and 1b, with a HS state up to 300 K. The species with the phenyl substituent is clearly analogous to the complex reported here. The ethoxycarbonyl group also probably requires energy to deviate from coplanarity with the pyridine ligand, thus blocking the system in the HS state as well. Interestingly, the related compound with R = mesityl 29 shows the opposite behaviour and it is found in the LS state. Here, the aromatic substituent of pyrazole is perpendicular to the latter, forced by the steric hindrance of the methyl groups, thus deactivating the blocking of the HS state. These two behaviours are therefore consistent with the findings reported here.
In summary, the ligand Ph1,3bpp, featuring a Ph substituent at the 1,3bpp core, was prepared in order to show that the SCO of an Fe(II) chromophore could be blocked by only intramolecular forces, with no effect from the crystal field. in the gas phase show that indeed its optimized structure in the LS state is less stable than in the HS state because the Ph group needs to rotate to alleviate the steric hindrance caused by the neighboring ligand. The rotation implies an energetic penalty associated with the loss of planarity in the phenylpyrazol-3-yl group. This would provide an elegant rationalization of the SCO inhibition, which here would be thermodynamic and not kinetic, as rationalized theoretically 17 Fig. 3 Computed energy variation of a phenylpyrazole moiety as a function of the torsion angle between both aromatic rings, as gauged by the torsion angle featured by the four atoms highlighted in yellow in the inset.