Tuning spin-crossover transition temperatures in non-symmetrical homoleptic meridional/facial [Fe(didentate)₃]²⁺ complexes: what for and who cares about it?

Abstract

The [FeN₆] chromophores found in [Fe(didentate)₃]²⁺ complexes, where didentate is a non-symmetrical 2-(6-membered-heterocyclic ring)-benzimidazole ligand (Lk), exist as mixtures of two geometrical mer (C₁-symmetry) and fac (C₃-symmetry) isomers. Specific alkyl-substituted six-membered heterocyclic rings connected to the benzimidazole unit (pyridines in ligands L1–L3, pyrazines in L4–L5 and pyrimidines in L6–L7) control the ligand field strength and the electron delocalization so that [Fe^II(Lk)₃]²⁺ display tunable thermally-induced spin transitions in solution. Thermodynamic, spectroscopic (UV-Vis, NMR) and magnetic studies in solution demonstrate that [Fe(L6)₃]²⁺ (L6 = 1-methyl-2-(pyrimidin-2-yl)-1H-benzo[d]imidazole) exhibits a close to room temperature spin transition (T_1/2 = 273(3) K) combined with a high stability formation constant ( in acetonitrile), which makes this complex suitable for the potential modulation of lanthanide-based luminescence in polymetallic helicates. A novel method is proposed for assigning specific thermodynamic spin crossover parameters to fac-[Fe(L6)₃]²⁺ and mer-[Fe(L6)₃]²⁺ isomers in solution. The observed difference relies mainly on the entropic content ΔS^mer_SCO − ΔS^fac_SCO = 11(1) J mol⁻¹ K⁻¹, which favors the spin transition for the meridional isomer. Intermolecular interactions occurring in the crystalline state largely overcome minor thermodynamic trends operating in diluted solutions and a single configurational isomer is usually observed in the solid state. Among the thirteen solved crystal structures 1–13 containing the [M(Lk)₃]²⁺ cations (M = Fe, Ni, Zn, Lk = L6–L7), pure meridional isomers are observed six times, pure facial isomers also six times and a mixture (44% mer and 56% fac) is detected only once. Solid-state magnetic data recorded for the Fe^II complexes show the operation of slightly cooperative spin transitions in 7 (fac-[Fe(L6)₃]²⁺) and 12 (mer-[Fe(L7)₃]²⁺). For the meridional isomer in 6, a two-step spin state transition curve, associated with two phase transitions, is detected.

---

Introduction

Faced with the duality of paramagnetism measured for [Fe(H₂O)₆]²⁺ and diamagnetism for [Fe(CN)₆]⁴⁻, Pauling exploited valence bond theory in 1931 to develop the concept of ‘magnetic criterion of bond type’ (Scheme 1a).^1,2 He also recognized that two molecular systems with different spin states could be present simultaneously at a given temperature, according that the energy difference between them is comparable with thermal energy (mRT with 1 ≤ m ≤ 10).^3,4 The concomitant isolation by Cambi et al. of Fe^III complexes with dithiocarbamate ligands displaying thermal spin-state equilibria are still considered as the first experimental demonstration of what is known as the spin crossover (SCO) phenomenon.^5–7 The formulation of the crystal-field theory,⁸ further extended and improved by the ligand field theory,^9–11 provided a better description of chemical bonding for open-shell transition metal complexes (Scheme 1b). The SCO phenomenon exhibited by d⁶ Fe^II in an octahedral environment could be then quantified as twice the difference between the ligand field strength Δ_oct and the spin pairing energy P, the latter being modeled with the help of the Racah parameters B and C to give P = 2B + 4C ≈ 19B (Scheme 1b).¹² A complete ligand-field picture is provided by the relevant Tanabe–Sugano diagram, which displays the relative energies of all the Russell–Saunders multiplet terms arising from the d⁶ configuration as a function of the Δ_oct/B ratio (Scheme 1c).¹³

Scheme 1 a) Illustration of early Pauling's magnetic criterion of bond type; (b) crystal field/ligand field approach to spin crossover phenomenon (P is the electron spin pairing energy); (c) Tanabe–Sugano diagram for a d⁶ metal ion completed with a central red domain, which corresponds to non-accessible ligand-field strengths and green domains where low-spin and high-spin complexes co-exist at accessible temperatures (). Adapted from ref. 16 and 17.

The strict equality Δ_oct = P, , leads to Δ_oct/B ≈ 19 and an equal population of ¹A₁ and ⁵T₂ levels represented by a vertical dashed line (Scheme 1c). However, this approach is misleading and physically unsound since the vertical dashed line in the Tanabe–Sugano diagram (Scheme 1c) refers to a non-equilibrium geometry. Indeed, the population of the antibonding orbitals in high-spin Fe^II is accompanied by a 10% elongation of the Fe–N bond lengths¹⁴ together with minor geometrical distortions,¹⁵ which affect Δ_oct and P during the spin transition. Taking this into account, a domain in the Tanabe–Sugano diagram is no more accessible to coordination complexes (red surface in Scheme 1c), whereas two restricted domains (green surfaces in Scheme 1c) correspond to the limits 11 000 ≤ Δ^HS_oct ≤ 12 500 cm⁻¹ and 19 000 ≤ Δ^LS_oct ≤ 22 000 cm⁻¹ for which the phenomenon of a thermal spin transition ( with 1 ≤ m ≤ 10) can be expected in Fe^II coordination complexes with standard Racah parameters.^12,16 Consequently, an adequate modeling of SCO behaviors requires looking beyond ligand-field theory with the help of quantum chemistry for mapping quantitatively the energy surfaces at equilibrium geometries associated with the ¹A₁ and ⁵T₂ terms.²¹ Although significant progress is being made along this line,^18–23 reliable models for the prediction of molecular SCO systems are still limited because they involve open-shell species and small energy differences.²⁴ The systematic exploration of classes of ligands is thus the rule for designing molecular SCO complexes in coordination chemistry. Myriads of six-coordinate SCO Fe^II complexes have been thus actively synthesized, characterized and a posteriori rationalized. This global effort has been regularly and extensively reviewed during the last two decades in specialized journals^14,25–35 and in two comprehensive monographies.^36,37 There is no doubt that the pseudo-octahedral [FeN₆] chromophore, where N stands for a heterocyclic nitrogen donor atom, is one of the most studied and versatile unit for implementing Fe^II SCO properties in molecular complexes. Some ‘rules of thumb’ could be extracted from empirical analyses of structure–properties relationships. For instance, Shatruk and coworkers used the interatomic separation between the N-donor atoms in multidentate chelate ligands as a metric guide for programming SCO properties in Fe^II complexes,³⁸ whereas Brooker, Garden and coworkers reported on more sophisticated density functional theory (DFT) calculations for predicting the ¹⁵N NMR shifts in 1,2,4-triazole ligands, which can be then used as a criterion for rationalizing SCO behaviors.²⁴

Short-range and long-range orders produced by packing interactions in solid state samples drastically complicate the situation because intermolecular energies usually dominate the minor energy changes accompanying the spin state transition at the molecular level,^30,39–42 a phenomenon often at the origin of specific phase transitions accompanying the spin crossover process in the solid state.^39,43 Whereas spin-state transitions implying solvated complexes in solution can be satisfyingly modeled with the standard entropy of mixing according to the simple chemical equilibrium (1) and associated stability constant K_SCO,^44,45 intermolecular communications produce some cooperative effects in the solid state, which can be implemented either by adding an interaction term Γ(x_hs, T, P) = γ·x_hs·(1 − x_hs) to give eqn (2), where γ is an adjustable interaction parameter,^43,46 or by considering non-randomly distributed domains containing n molecules of like spin in eqn (3) and (4), where C_p are the heat capacities at constant pressure.^43,47

with

Since precise chemical programming of spin state transitions are not accessible in solution, there is much less chance to propose some reliable design in solids, and major interest in the SCO community moved toward the systematic structural exploration of classes of ligands and complexes for designing addressable (via temperature, pressure, light) switches implemented in various materials (solids, films, nanoparticles, liquid crystals, etc.).^37,48–53 The deliberate combination of SCO properties with other physical or chemical properties in a synergetic fashion in molecular entities, summarized under the term of multifunctionality,^54–56 remains in its infancy despite remarkable efforts focused on the modulation of magnetic communication^57,58 and luminescence.⁵⁹ For instance, the modulation of an emissive lanthanide by an adjacent SCO Fe^II metallic site in an isolated dimetallic complex has been recently attempted (Fig. 1), but it proved to depend crucially on some ultra-precise tuning of the spin state equilibrium, which is currently beyond rational chemical design.⁶⁰

Fig. 1 Molecular structure of the [EuFeL₃]⁵⁺ helicate in which red luminescence (620 nm) is modulated via energy transfer (ET) to the adjacent SCO Fe^II unit.⁶⁰

Starting from the didentate 2-(α-methyl-substituted pyridine)-benzimidazole ligands L1–L3 (Scheme 2), which exactly model the didentate units bound to Fe^II in the [EuFeL₃]⁵⁺ helicate, some systematic exploration of structurally related 2-(α-methyl-substituted pyrazine)-benzimidazole L4–L5 allowed some shifts of T_1/2 in the mononuclear [Fe(didentate)₃]²⁺ model complexes, but none of them improved the situation in the helicates.^16,61 Moreover, simple mononuclear [Fe(didentate)₃]²⁺ complexes exist as mixtures of meridional and facial isomers, for which only the latter one can be taken as a valuable model for the triple-stranded [EuFeL₃]⁵⁺ helicates.⁶¹

Scheme 2 Chemical structures of the didentate ligands L1–L7, together with the electronic properties of the associated pseudo-octahedral [Ni(Lk)₃]²⁺ complexes and thermodynamic SCO properties of the pseudo-octahedral [Fe(Lk)₃]²⁺ complexes in CD₃CN solutions.

We report here the connection of a 2-pyrimidine ring (L6) or a 4-pyrimidine ring (L7) to the benzimidazole ring, which eventually provides [Fe(Lk)₃]²⁺ complexes possessing lower transition temperatures compatible with luminescence modulation in triple-stranded helicates (Scheme 2). A novel approach for the estimation of specific thermodynamic SCO parameters for meridional and facial isomers in solution is described.

Results and discussion

Design, synthesis and characterization of the didentate pyrimidine–benzimidazole ligands L6 and L7

In an elegant contribution,²⁴ Brooker, Garden and co-workers suggested that the ¹⁵N NMR chemical shifts recorded for the binding nitrogen atom of the azine ring in a series of didentate triazole–azine ligands (L^Azine) can be taken as a valuable indicator for both the electronic density borne by the N-donor atoms and their capacity to control the critical temperature T_1/2 of spin transitions for [Fe(L^Azine)₂(NCBH₃)₂] in solution. The decreasing order found for T_1/2 (pyrazine > pyridine > pyrimidine) matches that reported in Scheme 2 for the two first related complexes [Fe(L4)₃]²⁺ (pyrazine) > [Fe(L1)₃]²⁺ (pyridine) and encourages us to prepare 2-pyrimidine (L6) and 4-pyrimidine (L7) analogues for (i) shifting T_1/2 toward lower temperatures in [Fe(Lk)₃]²⁺ and (ii) further exploring the effect of meridional ↔ facial isomerization on the SCO properties.

The ligands L6–L7 were thus synthesized (Scheme 3) using a well-established strategy based on modified Philips condensation reactions for the formation of benzimidazoles as key steps.⁶² Pyrimidine-carboxylic acids, contrary to pyrazine-2-carboxylic acid,^16,62 could be activated in situ to pyrimidine-2-carbonyl chloride through standard reactions with either thionyl chloride or oxalyl chloride. The ¹H NMR and ¹³C NMR spectra (Scheme 4a and Fig. S1–S12†) point to fast rotations around the inter-annular C–C bond leading to average C_s-symmetry on the NMR time scale. The lack of NOE effect between the protons of the methyl group of the benzimidazole and the pyrimidine proton at the 5-position of the pyrimidine in L7 suggests a preferred average planar conformation in solution with an anti location of the N-donor atoms in the free didentate ligand (shown in Schemes 3 and 4). These structural characteristics are supported by gas-phase energies computed as a function of the interplanar angle between the two aromatic rings (Fig. S13†). Finally, indirect ¹H–¹⁵N detection (Fig. S6 and S12†)⁶³ shows that the N-donor atom of 2-pyrimidine in L6 is less shielded than that in 4-pyridimidine in L7 (Scheme 4b), an experimental result which is confirmed by the larger negative charge computed for the azine donor group in L7 (Fig. S14†).

Scheme 3 Multistep synthesis of the didentate ligands L6 (2-pyrimidine) and L7 (4-pyrimidine).

Scheme 4 a) ¹H NMR chemical shifts (with respect to tetramethylsilane at 0 ppm in CD₃CN) and (b) ¹⁵N NMR chemical shifts (with respect to nitromethane at 381.7 ppm in CD₂Cl₂) observed for ligands L6 and L7.

Slow diffusion of n-hexane into concentrate solutions of L6 and L7 in dichloromethane yielded crystals suitable for X-ray diffraction analysis (Tables S1–S6 and Fig. S15–S20†). Each unit cell contains two slightly different molecules of the ligands, both confirming the formation of almost planar didentate aromatic units adopting anti arrangements of the N-donor atoms (Fig. 2).

Fig. 2 Molecular structures of the two different molecules in the unit cells of the crystal structures of (a) L6 (interplanar angles 5.93(5)° and 9.71(5)°) and (b) L7 (interplanar angles 7.65(8)° and 12.73(8)°). The N-donor atoms are highlighted as blue spheres.

Complexation of the didentate pyrimidine–benzimidazole ligands L6 and L7 with Zn^II, Ni^II and Fe^II in solution

Reactions of L6 or L7 with M(CF₃SO₃)₂ (M = Zn, Fe) or Ni(BF₄)₂ in acetonitrile is characterized by a red-shift of the ligand-centered π* ← n,π electronic transitions (Fig. S21a–S24a†). Similar shifts were originally reported for 2,2′-bipyridine,⁶⁴ and could be theoretically assigned to the anti → syn conformational change of the α,α′-diimine binding unit accompanying its chelate complexation to the metal.^65,66 The spectrophotometric titrations of submillimolar concentrations of the ligands L6 and L7 in dry acetonitrile with Zn(CF₃SO₃)₂ were therefore exploited for the quantitative analysis of the successive formation of three complexes, characterized by smooth endpoints at Zn/Lk = 1 : 3, Zn/Lk = 1 : 2, Zn/Lk = 1 : 1 (Fig. S21b–S22b†). This trend was further confirmed by factor analysis (Fig. S21c–S22c†), which suggested the existence of four individual absorbing eigenvectors Lk, ZnLk, Zn(Lk)₂ and Zn(Lk)₃ with satisfying re-constructed absorption spectra (Fig. S21d–S22d†).^67–70 Non-linear least-squares fit to equilibria (5)–(7) provided satisfying estimates for the formation macroconstants gathered in Table 1 (column 1, entries 1 and 2).^71–73

Table 1 Stability constants for the successive formation of [M(Lk)_n]²⁺ (M = Zn, Ni, Fe, Lk = L6, L7, n = 1, 2, 3, non-coordinating anions: ClO₄⁻, CF₃SO₃⁻ or BF₄⁻) in acetonitrile (293 K), metal–ligand interaction parameter (f^M,Lk) and inter-ligand interaction parameter (u^Lk,Lk), and associated energies ΔG_affinity = −RTln(f^M,Lk) and ΔE_interaction = −RTln(u^Lk,Lk)

		log(f^M,L)	ΔGaffinity/kJ mol^−1	log(u^L,L)	ΔEinteraction/kJ mol^−1
[Zn(L6)n]^2+	log(β1,1) = 7.02(4)	6.0(4)	−34.4(2.4)	0.7(5)	−3.6(2.8)
	log(β1,2) = 15.2(1)
	log(β1,3) = 21.8(1)
[Zn(L7)n]^2+	log(β1,1) = 5.10(1)	3.9(2)	−22.1(9)	0.2(2)	−1.3(1.1)
	log(β1,2) = 10.22(2)
	log(β1,3) = 14.07(4)
[Fe(L6)n]^2+	log(β1,1) = 5.0(1)	4.1(5)	−23.2(2.7)	0.3(6)	−1.6(3.2)
	log(β1,2) = 10.95(1)
	log(β1,3) = 15.70(1)
[Ni(L7)n]^2+	log(β1,1) = 4.11(1)	3.0(2)	−16.8(1.3)	1.0(3)	−5.8(1.5)
	log(β1,2) = 9.22(1)
	log(β1,3) = 13.65(1)

---

Taking the experimental ¹⁵N HMR chemical shifts (Scheme 4b) as a body of evidence for optimizing M-σL binding and stability upon coordination to spherical and ionic Zn²⁺, one can reasonably predict . However, the reverse situation holds with a cumulative constant log() = 14.07(4), found for [Zn(L7)₃]²⁺, being almost 8 orders of magnitude smaller than log() = 21.8(1) (Table 1). Translated into ligand speciation,⁷³ these thermodynamic constants imply that the target saturated complex [Zn(L6)₃]²⁺ is quantitatively formed (>98%) in solution for a stoichiometric ratio Zn : L6 = 1 : 3 and a total ligand concentration of 10 mM (Fig. 3a). For [Zn(L7)₃]²⁺, a maximum of 82% of this complex can be detected in acetonitrile for a total ligand concentration of 10 mM (Fig. 3b). Its quantitative formation (≥97%) requires a ten times larger total concentration of ligand L7 (100 mM, Fig. 3c).

Fig. 3 Concentration profiles obtained using HySS2009 simulation⁷³ and the formation constants collected in Table 1 for the complex species (a) [Zn(L6)_n]²⁺ at |L6|_tot = 10 mM, and (b and c) [Zn(L7)_n]²⁺ at (b) |L7|_tot = 10 mM and (c) |L7|_tot = 100 mM (acetonitrile, 293 K).

We conclude that σ-donation from the bound ligand to Zn²⁺ is not the only important thermodynamic driving force. Comparison with the previous azine found in [Zn(L4)_n]²⁺ (Table S7†)⁷⁴ shows that L4 (2-pyrazine, pK_a = 0.65), which is a much weaker σ-donor but a stronger π-acceptor than L7, indeed provides stability constants up to three orders of magnitude larger than those observed for [Zn(L7)_n]²⁺ (4-pyrimidine, pK_a = 1.23). Surprisingly, the stability constants of [Zn(L6)_n]²⁺ (2-pyrimidine, pK_a = 1.23) largely overcome those of its azine analogues with L4 and L7, and can be only compared with that produced by the more basic 2-pyridine ring (pK_a = 5.2) in [Zn(L1)_n]²⁺ (Table S7†). In fact, the complexation of L1, L4 and L7 with the entering cation is limited by some considerable energetic penalties accompanying the anti → syn conformational change of the chelate α,α′-diimine unit. On the contrary, L6 is ‘symmetrically preorganized’ and no ligand reorganization is required. This beneficial effect is reminiscent to the gain of four orders of magnitude in stability observed in going from [Zn(2,2′-bipyridine)₃]²⁺ (log(β_1,3) = 13.2(2)) to [Zn(1,10-phenanthroline)₃]²⁺ (log(β_1,3) = 17.1(1)) in water.⁷⁵

A deeper and quantitative analysis relies on the site binding model (eqn (8)),^76–78 which combines a statistical factor K^stat_1,n (related to the contribution produced by the change in rotational entropy between the reactants and products and to the entropy of mixing of enantiomers),^79–81 with a chemical contribution K^chem_1,n made up of two microscopic chemical parameters and . The first term stands for the intermolecular microscopic affinities (including desolvation) characterizing the free energy of inner sphere connection of the didentate binding site in Lk to M^II () and accounts for the free energy of interligand interactions (), which operate when two ligands are bound to the same metallic centre.

Application of eqn (8) to all possible geometric isomers of pseudo-octahedral [Zn(Lk)(S)₄]²⁺ (one isomer, S = solvent), [Zn(Lk)₂(S)₂]²⁺ (five stereoisomers) and [Zn(Lk)₃]²⁺ (two stereoisomers) has been previously detailed (Fig. S25†)⁷⁴ and provides eight micro-constants which can be combined to give a global model (eqn (9)–(11)) for the three macro-constants characterizing equilibria (5)–(7).

Assuming the operation of a single average intermetallic interaction reduces the number microscopic descriptors to f^Zn,Lk and u^Lk,Lk for modelling the three macroscopic constants obtained by spectrophotometry (eqn (9), (12) and (13)).

Linear least-squares fits of the logarithmic forms provide the affinity parameters and interligand interactions gathered in Table 1 (columns 3–6), which satisfyingly reproduce the experimental data. Altogether, this rough thermodynamic model establishes that (i) the affinity of the selected didentate ligands for spherical Zn²⁺ () follows the trend L1 (2-pyridine) ≈ L6 (2-pyrimidine) > L4 (2-pyrazine) ≫ L7 (4-pyrimidine) and (ii) the successive binding of ligands leading to [Zn(Lk)₃]²⁺ follows a weakly cooperative procedure ().

Repeating the spectrophotometric titrations for submillimolar concentrations of the ligands L6 and L7 in dry acetonitrile with Fe(CF₃SO₃)₂ lead to smoother endpoints for L6 (Fig. S23†) and to only poor drifts for L7. Satisfying factor analysis and non-linear least-squares fit using equilibria (5)–(7) could be obtained for L6. The associated macro-constants are however drastically reduced by two to five orders of magnitude in going from Zn²⁺/L6 (Table 1, entry 1) to Fe²⁺/L6 (Table 1, entry 3). Though less pronounced, the same trend was previously reported for 2-pyrazine-benzimidazole L4 ligands in going from Zn²⁺ to Fe²⁺ (Table S7†).⁶¹ The replacement of ligand L6 with L7 leads to , a trend which strictly mirrors that previously observed for Zn²⁺. Finally, titrations of L7 with analogous Ni²⁺ cations (Fig. S24†) additionally fixes in line with the well-known Irving–Williams series (Table 1, entry 4).^16,82,83 Consequently, the speciation profiles computed for Fe²⁺/L6 (Fig. S26†) mirror those highlighted for Zn²⁺/L7 (Fig. 4b and c), which possess similar stability constants, but the Fe²⁺/L7 system is expected to produce much smaller quantities of [Fe(L7)₃]²⁺, which are overcome by those computed for [Ni(L7)₃]²⁺ (Fig. S27†).

Fig. 4 Meridional to facial isomerization process operating in [M(Lk)₃]²⁺ complexes (eqn (14)) together with its modeling with the help of the site binding model. The molecular structures are those found in the crystal structures of [Zn(L6)₃](ClO₄)₂ (CH₃CN)_2.5 (1).

Meridional to facial isomerization process operating for [Zn(Lk)₃]²⁺ (Lk = L6 and L7) in solution

A focus on the target [M(Lk)₃]²⁺ complexes requires the variable-temperature analysis of the thermodynamic equilibrium (14) linking the meridional and facial isomers (Fig. 4). This challenge can be addressed for diamagnetic closed-shell [Zn(Lk)₃]²⁺ cations with the help of variable-temperature ¹H NMR spectra.

At room temperature in CD₃CN and CD₃OD, both [Zn(L6)₃]²⁺ (10 mM) and [Zn(L7)₃]²⁺ (0.1 M) are quantitatively (>95%) formed and display dynamically average C₃-symmetrical symmetries on the NMR time scale (Fig. S28 and S29†). Upon stepwise decrease of the temperature, the dynamic exchange process slows down so that the ¹H NMR spectra display two series of resolved signals, which can be assigned to a mixture of mer-[Zn(Lk)₃]²⁺ (C₁ symmetry, three signals for a given proton) and fac-[Zn(Lk)₃]²⁺ (C₃ symmetry, one signal for a given proton, Fig. S28 and S29†). Integration of low temperature ¹H NMR data provides the experimental ratio of the concentrations mer-[Zn(Lk)₃]²⁺/fac-[Zn(Lk)₃]²⁺, from which can be estimated at different temperatures. Van't Hoff plots display linear correlations (Fig. S30†) and lead to enthalpic and entropic contributions collected in Table 2 together with speciations highlighted in Fig. 5. As previously reported for mer-[Zn(Lk)₃]²⁺ → fac-[Zn(Lk)₃]²⁺ equilibria with L1 (2-pyridine) and L4 (2 pyrazine, Table S8†),⁷⁴ the latter mer → fac isomerization process for L6 and L7 is solvent dependent and deviates from a pure statistical behavior ( and = −RT·ln(1/3) = 2.7 kJ mol⁻¹ at 298 K).

Fig. 5 Reconstruction of the mole fractions of the meridional and facial isomers for the complex [Zn(L6)₃]²⁺ in (a) CD₃CN and (b) CD₃OD, and (c) for the complex [Zn(L7)₃]²⁺ in CD₃OD as a function of temperature. The circles represent the accessible experimental data for which the dynamic of mer ↔ fac isomerization is slow on the ¹H NMR time scale.

Table 2 Thermodynamic enthalpic () and entropic () contributions to equilibrium (14) in solution

Complex	Solvent	ε r ^a	/kJ mol^−1	/J mol^−1 K^−1	^b/kJ mol^−1	T 1/2/K
a Relative dielectric constants. b are calculated at T = 298 K.
[Zn(L6)3]^2+	CD3CN	37.5	−2.8(1)	−16.2(4)	2.1(2)	171(8)
[Zn(L6)3]^2+	CD3OD	33.0	−6.9(5)	−30(2)	2.2(5)	227(26)
[Zn(L7)3]^2+	CD3OD	33.0	−7.3(9)	−38(4)	3.8(1.5)	195(32)

---

The enthalpic preference for the facial isomer (−7.3 ≤ ≤ −2.8 kJ mol⁻¹, Table 2), in which each benzimidazole N-donor atom is bound trans to the N-atom of the pyridimidine ring, is reminiscent of the thermodynamic trans influence.^74,84 The opposite entropic preference for the meridional isomer (4.8 ≤ ≤ 11.3 kJ mol⁻¹ at 298 K, Table 2), which is significantly larger than the pure statistical contribution of 2.7 kJ mol⁻¹ at 298 K, has been previously attributed to the larger relaxation of the solvent organization around the meridional isomer which possesses a smaller dipole moment.⁷⁴

Due to the remarkable stability found for [Zn(L6)₃]²⁺ in solution, the kinetic rate constant of the mer ↔ fac isomerization process (eqn (14)) is small enough to give resolved ¹H NMR spectra for each isomer on the NMR time scale in acetonitrile prior to reach its freezing point (233 ≤ T ≤ 258 K, Fig. 6a and S28†). The situation changes with the less stable [Zn(L7)₃]²⁺ complex, and the coalescence temperature for the ¹H NMR signals lie below the freezing point of CD₃CN (228 K). In deuterated methanol, some resolved ¹H NMR signals can be obtained for T < 228 K (Fig. S29†) for both [Zn(L6)₃]²⁺ (Fig. 5b) and [Zn(L7)₃]²⁺ (Fig. 5c).

Fig. 6 Plot of molar paramagnetic magnetic susceptibility (χ^para_MT) versus temperature (T) for a 10.2 mM solution of the complex [Fe(L6)₃]²⁺ in CD₃CN (green), and a 21.3 mM solution of the complex [Fe(L7)₃]²⁺ in CD₃CN (orange). The black dashed curves represent the best fit obtained by using eqn (20) and the thermodynamic and magnetic parameters collected in Table 3.

Electronic and spin crossover properties of [Fe(L6)₃]²⁺ and [Fe(L7)₃]²⁺ in solution

According to Busch and co-workers,⁸⁵ the easily accessible ligand field strengths measured for the pseudo-octahedral [Ni(Lk)₃]²⁺ complexes can be used as a reliable benchmark (11 200 ≤ Δ_oct(Ni^II) ≤ 12 400 cm⁻¹) for predicting and rationalizing spin-crossover operative in the related Fe^II complexes. The absorption spectra recorded for 0.1 M acetonitrile solution of [Ni(Lk)₃]²⁺ (Lk = L6, L7, Fig. S31 and S32†) display Ni(³T₂ ← ³A₂) transitions at 11 079 cm⁻¹ for L6 and 11 237 cm⁻¹ for L7, which can be used as estimates for Δ_oct (Table S9†).^61,86–88 Altogether, the four [Ni(Lk)₃]²⁺ complexes (Lk = L1, L4, L6 and L7) possess very similar electronic properties with an average value of Δ_oct/B = 13.2(2) compatible with SCO behavior around room temperature for the related [Fe(Lk)₃]²⁺ complexes (Scheme 2).^61,85 Variable temperature ¹H NMR spectra recorded for 0.1 mM solution of [Fe(L6)₃]²⁺ in acetonitrile (233–298 K) indeed unambiguously demonstrate the existence of temperature-dependent paramagnetic shifts which are diagnostic for the formation of increasing amounts of fast-relaxing high-spin Fe^II (S = 2) at high temperature (Fig. S33†). However, the broadening of the signals due to the combination of low-spin to high-spin transformation with thermally-activated mer ↔ fac isomerization prevents the change in chemical shifts to be exploited for extracting reliable speciation in solution and associated spin state equilibria.^89,90

Alternatively, the total magnetic susceptibility of a paramagnetic solute χ^para_M (cm³ mol⁻¹) can be obtained with the help of Evans’ method,^45,91–93 adapted for the temperature dependence of solvent density⁹⁴ and for an adequate treatment of the various diamagnetic corrections (eqn (15); S_f = 4π/3 is the shape factor of the superconducting magnet, ν₀ is the Larmor frequency of the spectrometer in Hz, Δv = v_ref − v_int is the difference in Hz between the frequency of the standard (tert-butanol) in contact with the complex of interest (v_ref = δ_ref·v_o) and that of the standard (v_int = δ_int·v_o) placed in the coaxial tube in absence of complex, M is the molecular weight in g·mol⁻¹ of the paramagnetic complex or of its diamagnetic analogue when Fe^II is replaced with Zn^II, and m is the concentration of the pertinent complex in g cm⁻³, Fig. S34†).^95,96

Application of eqn (15) for the variable-temperature shifts of the ¹H-NMR signals recorded for tert-butanol in acetonitrile solutions of [Fe(L6)₃]²⁺ (10.2 mM, Fig. S34†) and [Fe(L7)₃]²⁺ (21.3 mM), while using [Zn(L6)₃]²⁺ (13 mM) as the diamagnetic reference, eventually gave plots of χ^para_MT versus T, which are diagnostic for the operation of spin crossover in solution (Fig. 6).

The general expression of the Curie law, corrected for diamagnetism, for a mixture containing high-spin (mole fraction x_hs) and low-spin (mole fraction x_ls) complexes is given in eqn (16), where C_hs and C_ls are the Curie constants for the high-spin and low-spin forms (C = (g²/8)S(S + 1) within the frame of the spin-only approximation) and TIP_hs and TIP_ls are the temperature-independent paramagnetic contributions.^39,45

χ^para_MT = x_hs·(C_hs + T·TIP_hs) + x_ls·(C_ls + T·TIP_ls) (16)

The introduction of mass balance (x_ls + x_hs = 1) gives a straightforward access to the speciation (eqn (17) and Fig. 7). The consideration of spin-state eqn (1) eventually transforms eqn (16) into eqn (18) (see ESI Appendix 1†).⁶¹

Fig. 7 Plot of high-spin mole fraction (estimated from χ^para_MT by using eqn (17)) versus temperature (T) for a 10.2 mM solution of the complex [Fe(L6)₃]²⁺ in CD₃CN (green circles) and its analysis using eqn (20) as originating from fac-[Fe(L6)₃]²⁺ (blue trace) and mer-[Fe(L6)₃]²⁺ (red trace) controlled by the isomerization constant .

The non-linear least-square fits of the experimental paramagnetic data depicted in Fig. 6 for [Fe(L6)₃]²⁺ and [Fe(L7)₃]²⁺ by using eqn (18) provides Curie constants, TIP and thermodynamic parameters ΔH_SCO and ΔS_SCO for the spin crossover equilibrium gathered in Table 3 (see ESI Appendix 1 for details†).

Table 3 Magnetic and thermodynamic parameters obtained from the non-linear least square fits of χ^para_MT vs. T plots recorded for [Fe(Lk)₃]²⁺ complexes in solution (CD₃CN)

Complex	Anion	C hs/cm^3 K mol^−1	TIPhs× 10^−6/cm^3 mol^−1	C ls/cm^3 K mol^−1	TIPls× 10^−6/cm^3 mol^−1	ΔHsco/kJ mol^−1	ΔSsco/J mol^−1 K^−1	T 1/2/K	Ref.
a 100% low-spin at all studied temperatures. b 100% high-spin at all studied temperatures. c C ls and TIPls found for [Fe(bipy)3]^2+ and Chs and TIPhs found for [Fe(L2)3]^2+ have been used in ref. 65 for fitting the SCO behaviors.
[Fe(bipy)3]^2+ ^a	CF3SO3^−	—	—	0.00(8)	351(55)	—	—	—	61
[Fe(L2)3]^2+ ^b	PF6^−	3.44(2)	358(57)	—	—	—	—	—	61
[Fe(L1)3]^2+	PF6^−	3.44^c	358^c	0.00^c	351^c	28.6(3)	91.5(8)	313(4)	61
[Fe(L4)3]^2+	CF3SO3^−	3.44^c	358^c	0.00^c	351^c	35.7(3)	101(1)	351(5)	61
[Fe(L6)3]^2+	BF4^−	4.10(3)	369(3)	0.00	342(3)	19.5(2)	71.3(6)	273(3)	This work
[Fe(L7)3]^2+	PF6^−	3.595(3)	368.8(2)	0.00	346.2(3)	23.01(2)	82.5(1)	279(1)	This work

---

The Curie constants C and TIP values found for [Fe(L6)₃]²⁺ and [Fe(L7)₃]²⁺ (Table 3, entries 5–6) are typical for the [FeN₆] chromophores previously investigated in [Fe(didentate)₃] complexes.^{16,61,97–102} One notes that C_HS > 3.00 cm³ K mol⁻¹ (spin only value) can be taken as the signature of some partially unquenched orbital momentum in this pseudo-octahedral environment. The thermodynamic parameters 19.5 ≤ ΔH_SCO ≤ 23 kJ mol⁻¹ and 71 ≤ ΔS_SCO ≤ 83 J mol⁻¹ K⁻¹ measured for [Fe(L6)₃]²⁺ and [Fe(L7)₃]²⁺ also match the usual range found for similar complexes in solution.^45,97,98 Interestingly, the pyrimidine rings in ligands L6 and L7 produce the weakest enthalpic change accompanying the spin crossover process in solution compared with pyridine in L1 and pyrazine in L4. Due to partial H/S compensation^101–105 established for SCO processes,¹⁰⁵ a satisfying ΔH_SCO = α·ΔS_SCO + β linear correlation is observed for the series of [Fe(Lk)₃]²⁺ complexes with α = 601(103) K and β = −24 936(9020) J mol⁻¹ (Fig. S35†). Consequently, the transition temperature T_1/2 = ΔH_SCO/ΔS_SCO = α + (β/ΔS_SCO) in CD₃CN decreases with ΔS_SCO (since β < 0) and reaches its lowest value for [Fe(L6)₃]²⁺ (273 K) followed by [Fe(L7)₃]²⁺ (279 K), [Fe(L1)₃]²⁺ (313 K) and [Fe(L4)₃]²⁺ (351 K, Table 3, column 9).

However, the unavoidable mer ↔ fac equilibrium (14) depicted in Fig. 4 implies that the above rough analysis of the SCO processes operating in [Fe(didentate)₃]²⁺ complexes in solution is of limited pertinence when the didentate ligand is non-symmetrical.^33,61 A pioneering attempt to obtain specific SCO thermodynamic contributions for fac-[Fe(Lk)₃]²⁺ and mer-[Fe(Lk)₃]²⁺ isomers used non-covalent lanthanide tripods to force facial arrangement around Fe^II as illustrated in the dimetallic triple-stranded helicate (Fig. 1). The associated thermodynamic data recorded in solution for the SCO processes occurring in LaFe-helicates, where the didentate units bound to Fe^II mirrors ligands L1 and L4, were taken as valuable models for fac-[Fe(Lk)₃]²⁺ complexes. The thermodynamic characteristics of mer-[Fe(Lk)₃]²⁺ could be then deduced from the χ^para_MT versus T plots recorded for standard solution mixtures according that are at hand at each temperature.⁶¹ We propose here a parallel approach which takes reasonably for granted that the Curie constants C and the TIP contributions are identical for the fac and mer isomers of the same complex. Moreover, one also assumes (still reasonably) that the thermodynamic data extracted from VT-NMR analysis for the isomerization process operating in [Zn(Lk)₃]²⁺ holds true for [Fe(Lk)₃]²⁺ (in other words ). With this in mind, eqn (16) can re-written as eqn (19) and the introduction of the thermodynamic constants K_mer→fac and K_SCO finally yields eqn (20) (see ESI Appendix 2 for details†).⁶¹

χ^para_MT = (x^fac_hs + x^mer_hs)·(C_hs + T·TIP_hs) + (x^fac_ls + x^mer_ls)·(C_ls + T·TIP_ls) (19)

The systematic application of the van't Hoff relationship ΔG° = –RTln(K) = ΔH° − TΔS° for estimating the various stability constants together with (i) the use of and gathered in Table 2 for L6 in CD₃CN and (ii) , allows the satisfying non-linear least-square fits with eqn (20) of χ^para_MT versus T plots for fac-[Fe(L6)₃]²⁺ and mer-[Fe(L6)₃]²⁺ (Fig. 7). In complete agreement with the previous analysis using a loosely constrained triple-stranded LaFe helicate for modeling fac-[Fe(Lk)₂]²⁺ (Lk = L1, L4),⁶¹ this alternative approach avoids the synthesis of the related helicates and confirms that the SCO enthalpies are only marginally larger for the facial isomer (≈1 kJ mol⁻¹). However, the SCO entropy is systematically more favorable for the meridional isomer mer-[Fe(L6)₃]²⁺ by ΔS^mer_SCO− ΔS^fac_SCO = 11(1) J mol⁻¹ K⁻¹, a value mirroring those previously reported for mer-[Fe(L1)₃]²⁺ (14(2) J mol⁻¹ K⁻¹) and mer-[Fe(L4)₃]²⁺, (11(1) J mol⁻¹ K⁻¹).⁶¹ In going from facial to meridional isomer in [Fe(Lk)₃]²⁺, H/S anti-compensation seems to be the rule since the spin transitions, which occur in mer-[Fe(Lk)₃]²⁺ are concomitantly favored over fac-[Fe(Lk)₃]²⁺ by (i) a lower enthalpy cost and a (ii) larger entropic gain (see Fig. 7).

The different enthalpic contributions are reminiscent of the concept of trans influence which is thought to strengthen Fe–N bonds in the facial isomer. The entropic preference exhibited by the meridional isomer might be relevant to solvation effects. Indeed, the primary change in solvation energy accompanying the low-spin to high-spin transition is produced by the expansion of [Fe(Lk)₃]²⁺ considered as a charged monopole.^106,107 The associated endergonic change ΔΔ_solvG_monopole ∝ −(z²_hs/R_hs − z_ls²/R_ls) modeled with Born equation¹⁰⁶ is expected to be very similar for both meridional and facial isomers since their cationic charges are identical (z^mer_hs = z^fac_hs = z^mer_ls = z^fac_ls = 2) and their ionic radii are very close (R_fac ≈ R_mer since V_fac ≈ V_mer). The minor, but non-negligible additional dipolar contribution to the solvation energy can be estimated with the help of Onsager equation to give ΔΔ_solvG_dipole ∝ −(μ_hs²/R_i,hs³ − μ_ls²/R_i,ls³).^108–111 Again, the radius of the spherical cavity R_i cut from the dielectric when the solute, taken as spherical, is immersed into the solvent, is similar for both isomers (R_i,fac ≈ R_i,mer), but the magnitude of the dipole moments are different with μ_fac² > μ_mer².¹¹² Since R_i,hs > R_i,ls, then ΔΔ_solvG_dipole > 0 and the contribution of dipole solvation disfavors the low-spin to high-spin spin transition more strongly for the facial isomer than for the meridional isomer as measured by ΔS^mer_SCO > ΔS^fac_SCO.

Isolation, structures and magnetic properties of [M(L6)₃]X₂ and [M(L7)₃]X₂ in the solid state (M = Fe, Ni, Zn and X = CF₃SO₃⁻, BF₄⁻, PF₆⁻, ClO₄⁻)

Stoichiometric M : Lk = 1 : 3 mixtures of L6 or L7 (10 mM) with Ni(BF₄)₂·6H₂O or Zn(CF₃SO₃)₂ or Fe(ClO₄)₂·xH₂O or Fe(BF₄)₂·6H₂O in acetonitrile were evaporated to dryness. The resulting microcrystalline powders were re-dissolved in acetonitrile or methanol and allowed to crystallize by evaporation, or by diffusion of tert-butyl methyl ether after metathesis in presence of a large excess (10 eq.) of (ⁿBu)₄NClO₄ or (ⁿBu)₄NBF₄ or (ⁿBu)₄NPF₆ to give primary salts [Ni(L6)₃](BF₄)₂·1.55CH₃CN·0.4H₂O, [Zn(L6)₃](CF₃SO₃)₂·0.6H₂O, [Fe(L6)₃](BF₄)₂·0.15CH₃CN·1.2H₂O, [Ni(L7)₃](BF₄)₂·0.85H₂O, [Zn(L7)₃](PF₆)₂·1.25H₂O, and [Fe(L7)₃](PF₆)_1.76(BF₄)_0.24·0.6H₂O complexes in 46–72% yield (Table S10†). Single crystals of sufficient quality for X-ray diffraction studies could be obtained for complexes 1–13 gathered in Fig. 8 (Tables S11–S40 and Fig. S36–S48†). Reminiscent to the empirical isolation of either pure facial or meridional isomers reported for [Fe(imidazolylimine)₃](Cl)(PF₆) salts with different solvent contents,^113,114 the thirteen crystal structures 1–13, which incorporate similar [M(Lk)₃]²⁺ cations with various poorly coordinating anions and solvent molecules, are statistically partitioned between six pure facial isomers (2, 4, 5, 7, 8, 9), six pure meridional isomers (3, 6, 10, 11, 12, 13) and one mixture of meridional/facial (44/56%) isomers (1, Fig. 8 and Tables S41 and S42†). All complexes display poorly distorted octahedral [MN₆] chromophores¹¹⁵ with bond lengths within the expected range.^116,117 The Fe–N bond lengths measured for [Fe(L6)₃](ClO₄)₂ (6), and [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7) at 180 K (2.03 to 2.15 Å, Table S41†) are longer than standard Fe_ls–N bonds^26,38 found in [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12) and [Fe(L7)₃](PF₆)_1.72(ClO₄)_0.28·CH₃OH (13) at 150 K (1.98–1.99 Å. Table S42†). This suggests that the iron complexes with ligands L6 exists as mixtures of low-spin and high-spin in 6 and 7 at 180 K, while those with L7 in 12 and 13 are mainly low-spin at 150 K.

Fig. 8 Molecular structures of the [M(Lk)₃]²⁺ cations in the crystal structures of [Zn(L6)₃](ClO₄)₂(C₂H₃N)_2.5 (1), [Zn(L6)₃](BF₄)₂·(CH₃CN)₂ (2), [Ni(L6)₃](ClO₄)₂ (3), [Ni(L6)₃](ClO₄)₂·C₂H₃N (4), [Ni(L6)₃](BF₄)₂·C₂H₃N (5), [Fe(L6)₃](ClO₄)₂ (6), [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7), [Zn(L7)₃](ClO₄)₂ (8) [Zn(L7)₃](PF₆)₂ (9) [Ni(L7)₃](ClO₄)_1.48(PF₆)_0.52·CH₃CN (10), [Ni(L7)₃](BF₄)₂·CH₃OH·(CH₃CN)_0.5 (11), [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12) and [Fe(L7)₃](PF₆)_1.72(ClO₄)_0.28·CH₃OH (13).

A second collection of X-ray diffraction data at 100 K for complexes [Fe(L6)₃](ClO₄)₂ (6, meridional isomer) and [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5 (7, facial isomer) indeed showed a significant decrease of the Fe–N bond length, which converged to 1.99 Å at low temperature as expected for pure low-spin [FeN₆] units (Table S43†). Interestingly, complex 6 at 100 K adopts the monoclinic crystal system with one meridional isomer per asymmetric unit, and transforms into a triclinic system with two different meridional complexes (noted as Fe–A and Fe–B) per unit cell at 180 K. A careful look at the bond lengths measured in the crystal structure at 180 K (Table S25†) indeed shows different average bond lengths with 2.01(1) Å for Fe–A and 2.08(3) Å for Fe–B units. A detailed variable-temperature X-ray diffraction analysis of mer-[Fe(L6)₃]²⁺ in 6 revealed two successive phase transitions, , accompanied by a two-steps spin crossover transition (Table S44†).

Complexes Fe–B (50% of total iron content) undergo the low-spin to high-spin transition within the 165–210 K range (Fig. 9a). The remaining Fe–A sites see their coordination spheres to be slightly shrunk when low-spin Fe–B is transformed into high-spin Fe–B. Then the Fe–A sites undergo their own spin transition in the 220–250 K range (Fig. 9a). As expected, this extension of the Fe–N bond lengths is accompanied by a rather regular increase of the volume of the unit cell with small bumps at 170 and 240 K (Fig. 9b), which are confirmed by differential scanning calorimetry (Fig. S49†).

Fig. 9 Evolution of (a) Fe−N bond lengths in the monoclinic phases (green) and in the triclinic phase with Fe−A (red) and Fe−B (blue) and (b) cell volumes of [Fe(L6)₃](ClO₄)₂ (6) in its crystalline state as a function of temperature during heating.

The distinction between two complex molecules with the same chemical formula, but exhibiting specific transitions in the same solid is not unprecedented and was theoretically rationalized by Sasaki and Kambara.¹¹⁸ Related experimental observations of this phenomenon within the SCO topics have been reported for both Fe^III complexes¹¹⁹ and Fe^II complexes,¹²⁰ which displayed an ‘intermediate-phase’ (IP) ordering phenomenon leading to a two-step spin-crossover behavior. The IP is defined by the limiting temperatures of the two-phase transitions and is usually formed of approximately equal proportions of LS and HS states. It is therefore not so surprising that paramagnetic χ^para_MT versus T plot obtained by SQUID measurements for [Fe(L6)₃](ClO₄)₂ (6, meridional isomers) in the solid state display two-successive spin transitions in the 100–300 K range (Fig. 10a). On the other side, [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5 (7, facial isomers) and [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12, meridional isomers) exhibit regular one-step spin transitions (Fig. 10b and c). Interestingly, [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7) exhibits a complete and abrupt spin transition, which allows the reliable estimation of a Curie constant C_ls = 0.0080(3) cm³ K mol⁻¹ and temperature-independent paramagnetism TIP_ls = 387(14) × 10⁻⁶ cm³ mol⁻¹ in the pure low-spin state and C_hs = 3.53(13) cm³ K mol⁻¹ TIP_hs = 494(18) × 10⁻⁶ cm³ mol⁻¹ in the pure high-spin state. Taking these values as reasonable approximations for all low-spin, respectively high-spin Fe^II complexes with ligands L6 and L7, it is possible to apply eqn (17) for transforming the plots of the molar paramagnetic susceptibility (χ_MT) versus temperature (T) collected in Fig. 10 into plots of high-spin mole fractions (x_hs) as a function of temperature (Fig. S50†).

Fig. 10 Plots of molar paramagnetic susceptibility (χ_MT) versus temperature (T) collected with increments of 1 K min⁻¹ in a constant magnetic field of 5000 Oe between 5–300 K for (a) [Fe(L6)₃](ClO₄)₂ (6, meridional isomers) and 4–400 K for (b) [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5 (7, facial isomers) and (c) [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12, meridional isomers). The red points represents the experimental data while the dashed black traces are built by using the fitted values of ΔH_sco, ΔS_sco, and γ (Table 4, see text).

According to the various shapes observed for x_hsversus T plots (Fig. S50†), the simple van't Hoff relationship (eqn (1)) is not adequate and should be replaced with eqn (2) where the adjustable interaction parameter γ takes into account intermolecular interactions.^43,46 However, x_hs cannot be expressed analytically as a function of T with eqn (2), but the temperature can be expressed as a function of x_hs in eqn (21).

Non-linear least squares fits using eqn (21) for the experimental 〈T; x_hs〉 couples measured for [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5 (7, facial) and [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12, meridional) yield the thermodynamic constants ΔH_sco and ΔS_sco and the cooperativity factor γ collected in Table 4 (columns 4 and 5) and which satisfyingly reproduce the experimental data (Fig. 10b, c and S50b, c†).

Table 4 Thermodynamic parameters (ΔH_sco, ΔS_sco, and γ) associated with the SCO process in [Fe(L6)₃](ClO₄)₂ (6), and [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7), and [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12)

Complex	[Fe(L6)3]^2+ (6)		[Fe(L6)3]^2+ (7)	[Fe(L7)3]^2+ (12)
	Fe–A	Fe–B
a Agreement factor .
Isomer	Meridional	Meridional	Facial	Meridional
ΔHSCO/kJ mol^−1	10.6(4)	3.58(6)	15.5(9)	13.45(1)
ΔSSCO/J mol^−1 K^−1	42(2)	20.1(3)	75(4)	39.5(4)
T 1/2/K	251(12)	178(4)	207(16)	341(3)
γ/kJ mol^−1	2.54(9)	2.12(3)	3.7(2)	1.71(2)
AF^a	4.61 × 10^−3		5.01 × 10^−4	8.51 × 10^−3

---

Since the energies of intermolecular interactions overpass those induced by the molecular SCO processes, ΔH_sco, ΔS_sco observed in the solid state for [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7), and [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12, Table 4) significantly deviate from those recorded in solution (Table 3). Consequently, the transition temperature found for fac-[Fe(L6)]²⁺ in acetonitrile (T_1/2 = 309(23) K in Fig. 7) does not match that found in the crystal structure of 7 (T_1/2 = 207(16) K). The systematic operation of positive cooperativity observed for [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5 (7, γ = 3.7(2) kJ mol⁻¹), and [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12, γ = 1.71(2) kJ mol⁻¹) in the solid state is more instructive and suggests that these crystal lattices are well-adapted for accommodating the larger high-spin form. The value of γ found for 7 in the solid state is slightly larger than 2RT_1/2 (γ > 2RT_1/2 = 3.4 kJ mol⁻¹) and we must consider the possibility that the latter complex might undergo a hysteretic transition between the low-spin and the high-spin states.^39,40 However, the record of magnetic susceptibility along a complete heating and cooling cycle showed no interpretable hysteresis, but some drift resulting from the evaporation of volatile interstitial solvent molecules (Fig. S51†).

Finally, molar magnetic susceptibilities (χ^para_M) recorded for mer-[Fe(L6)₃](ClO₄)₂ (6) at variable temperatures between 5–300 K indeed confirmed the crystallographic studies and the operation of two successive SCO transitions, each affecting 50% of the total iron complexes (Fig. 10a). The magnetic data could be rationalized with eqn (22), which considers an equimolar mixture of two complexes Fe–A (T_1/2 at high temperature) and Fe–B (T_1/2 at low temperature) in the solid state.

Introducing the single set of magnetic parameter C_ls = 0.008 cm³ K mol⁻¹, C_hs = 3.53 cm³ K mol⁻¹, TIP_ls = 387 × 10⁻⁶ cm³ mol⁻¹ and TIP_hs = 494 × 10⁻⁶ cm³ mol⁻¹ into eqn (22) transforms the experimental χ_MT values recorded for mer-[Fe(L6)₃](ClO₄)₂ (6, Fig. 10a) into total high-spin mole fractions x^tot_hs = x^A_hs + x^B_hs (Fig. S50a†). Assuming that the growth of high-spin fraction is solely assigned to x^B_hs in the 140–200 K range and to x^A_hs in the 225–300 K domain allows two separated non-linear least-square fits of the 〈T; x_hs〉 couples with eqn (21) for the two successive transitions detected in mer-[Fe(L6)₃](ClO₄)₂ (6, Fig. 10a). The associated approximated thermodynamic parameters are collected in Table 4 (columns 2 and 3) and show transition temperatures T_1/2(Fe–B) = 178(4) K and T_1/2(Fe–A) = 251(12) K in line with the phase transition observed by DSC (Fig. S48,†T_1/2 = 167 K and 241 K). Some slight positive cooperativity 2.1 ≤ γ ≤ 2.5 kJ mol⁻¹ operates within each spin state transition in mer-[Fe(L6)₃](ClO₄)₂ (6) and, as expected from the very limited cooperativity factors, the χ^para_MT versus T plots recorded for 6 and 12 for heating–cooling cycles show no hysteresis.

Conclusion

Even if it may appear tedious, there is currently no more efficient method to tune SCO properties of a [FeN₆] unit than the use of some semi-empirical trial/analysis/optimization procedures within a family of closely related complexes as proposed for the L1–L7 series. In this context, [Fe(L6)₃]²⁺ emerges to be the best candidate in term of both stability and spin transition temperature in solution, to be connected to luminescent lanthanide stains in metallosupramolecular architectures. These systematic studies (i) confirmed the unpredictability of SCO behaviors and associated thermodynamic parameters observed in solid state materials, and (ii) allowed us to focus our attention on the effect of mer ↔ facial isomerization on the spin state transitions operating for [Fe(didentate)₃]²⁺ in solution (didentate stands for a non-symmetrical diimine ligand). As expected from the minor thermodynamic trans influence reported for the formation of pseudo-octahedral complexes along the 3d series,^74,84 ΔH_SCO are slightly more positive, and thus unfavourable, for the facial isomers (ΔH^fac_SCO − ΔH^mer_SCO ≈ 1 kJ mol⁻¹). This trend is amplified by a larger drift originating from an entropic preference for SCO operating in the meridional isomers in solution ΔS^mer_SCO − ΔS^fac_SCO ≈ 11 J mol⁻¹ K⁻¹. Since the same entropic gap has been previously reported in triple-stranded FeLn helicates, where the facial isomer is prevented to transform into meridional isomer by the non-covalent lanthanide tripod,⁶¹ we conclude that, in solution, mer-[Fe_hs(didentate)₃]²⁺ is systematically entropically favored over fac-[Fe_hs(didentate)₃]²⁺ by dipolar solvation effects. With the [Fe(2-pyrimidine-benzimidazole)₃] units as programmed in [Fe(L6)₃]²⁺, we are now equipped for the exploration of dinuclear Fe^II–Ln^III complexes, the lanthanide-based luminescence of which could be modulated by Fe^II spin crossover at the molecular level.

Experimental section

General

Chemicals were purchased from Sigma-Aldrich and Acros and used without further purification unless otherwise stated. Dichloromethane, 1,2-dichloroethane, tert-butyl methyl ether and N,N-dimethylformamide were dried through an alumina cartridge. Silica-gel plates (Merck, 60 F₂₅₄) were used for thin-layer chromatography, SiliaFlash® silica gel P60 (0.04–0.063 mm), and Acros silica gel 60 (0.035–0.07 mm) were used for preparative column chromatography.

Preparation of pyrimidine-2-carboxylic acid (1)

Pyrimidine-2-carboxylic acid was synthesized according to a slightly modified reported procedure.¹²¹ Pyrimidine-2-carbonitrile (5.02 g, 47.76 mmol, 1 eq.) was stirred with 5 M NaOH (35 mL) at 55 °C for 3 h. The solution was neutralized (pH 2.0) with diluted aqueous HCl and evaporated to dryness. The resulting solid was suspended in acetonitrile, refluxed for 1 h, and filtered when hot to remove the insoluble salts. The filtrate was evaporated under vacuum to dryness. The resulting solid was re-dissolved in methanol and crystallized to give 3.21 g of pyrimidine-2-carboxylic acid (2, 25.8 mmol, yield 54%) as colourless thin needles. ¹H NMR (CDCl₃, 400 MHz, 298 K) δ/ppm: 7.61 (1H, t, ³J = 4.9 Hz), 9.00 (2H, d, ³J = 4.9 Hz), 10.62 (br).

Preparation of N-methyl-2-nitroaniline (2)

1-Chloro-2-nitrobenzene (30 g, 190.4 mmol, 1 eq.) and methylamine (197.7 mL 40% wt. in water, 2.28 mol, 12 eq.) were introduced into a Carius tube equipped with a magnetic stirrer and heated at 120 °C for 48 h. The excess of methylamine was rotary evaporated, and the residual brown oil was partitioned between CH₂Cl₂ (300 mL) and half-saturated aqueous NH₄Cl (300 mL). The organic layer was separated and the aqueous phase was further extracted with CH₂Cl₂ (3 × 150 mL). The combined organic extracts were dried over anhydrous Na₂SO₄, filtered and the solvent evaporated to dryness. The resulting red oil was purified by column chromatography (silica, CH₂Cl₂) to give N-methyl-2-nitroaniline (1, 178.57 mmol, yield 94%) as a deep red/orange oil, which was crystallized upon trituration. ¹H NMR (CDCl₃, 400 MHz, 298 K) δ/ppm = 8.13 (1 H, dd, ³J = 8.8 Hz, ⁴J = 1.6 Hz), 8.00 (1 H, bs), 7.43 (1 H, ddd, ³J = 8.8 Hz, ³J = 7.2 Hz, ⁴J = 1.6 Hz), 6.81 (1 H, dd, ³J = 8.6 Hz, ⁴J = 1.0 Hz), 6.62 (1 H, ddd, ³J = 8.4 Hz, ³J = 7.2 Hz, ⁴J = 1.2 Hz), 2.99 (3 H, s).

Preparation of N-methyl-N-(2-nitrophenyl)pyrimidine-2-carboxamide (3)

Thionyl chloride (7.41 mL, 12.15 g, 102.14 mmol, 9.26 eq.) was added dropwise to the mixture of pyrimidine-2-carboxylic acid (1, 1.37 g, 11.03 mmol, 1 eq.) dissolved dry 1,2-dichloroethane (35 mL). Dimethylformamide (300 μL) was added to the reaction mixture. The solution was refluxed for 5 h under an inert atmosphere of nitrogen, then cooled and evaporated to dryness. The residue was dissolved in 40 mL of dry 1,2-dichloroethane. N-Methyl-2-nitroaniline (2, 4.19 g, 27.57 mmol, 2.5 eq.) dissolved in 10 mL 1,2-dichloroethane was added dropwise to the solution. The solution was refluxed 15 h during which two portions of (iPr)₂NEt (Hünig's base, 3.5 mL, 20.09 mmol, 1.82 eq.) were added after 30 min and 12 h. The solution was concentrated under vacuum to afford a solid residue which was partitioned between CH₂Cl₂ (400 mL) and half saturated aqueous NH₄Cl (4 × 120 mL). The aqueous phase was further extracted with CH₂Cl₂ (3 × 50 mL). The combined organic phase was dried over anhydrous Na₂SO₄, concentrated under vacuum and purified by column chromatography (silica, CH₂Cl₂/MeOH 99.2 : 0.8) to give 2.62 g of N-methyl-N-(2-nitrophenyl)pyrimidine-2-carboxamide (3, 10.14 mmol, yield 92%) as a beige solid. ¹H NMR (CDCl₃, 400 MHz, 298 K) δ/ppm: mixture of rotamers A (86.5%) and B (13.5%): 3.41 (3H, s, B), 3.55 (3H, s, A), 7.08–7.75 (4H, m, A and B), 7.95 (1H, dd, ³J = 8.4 Hz, A), 8.10 (1H, dd, ³J = 8.6 Hz, B), 8.53 (2H, d, ³J = 4.8 Hz, A), 8.91 (2H, d, ³J = 4.9 Hz, B). ESI-MS (soft positive mode – MeOH + CHCl₃ + HCOOH)- [3 + H]⁺ = 259.3 (exp.) 259.3 (calc.), [3 + Na]⁺ = 281.4 (exp.) 281.2 (calc.), 213.4, 136.4, 119.3, 107.3 (fragments).

Preparation of 1-methyl-2-(pyrimidin-2-yl)-1H-benzo[d]imidazole (L6)

N-Methyl-N-(2-nitrophenyl)pyrimidine-2-carboxamide (3, 2.60 g, 10.06 mmol, 1 eq.) was dissolved in a mixture of EtOH : DMF (20 mL : 25 mL). Sodium dithionite (7 g, 40.20 mmol, 3.97 eq.) was added to the solution and the temperature of the mixture was raised to 60 °C when 20 mL of distilled water was added. After refluxing for 24 h, the mixture was neutralized using aqueous ammonia and the solvents were removed in vacuo. The concentrate was dissolved in CH₂Cl₂ (50 mL) and washed with water (3 × 200 mL). The aqueous layers were collectively re-extracted with CH₂Cl₂ (3 × 50 mL). The combined organic fractions were then concentrated under vacuum and purified by column chromatography (silica, CH₂Cl₂/MeOH 98 : 2) to yield L6 (5.86 mmol, yield 58%). The compound was crystallized as needles by slow diffusion of n-hexane into CH₂Cl₂. ¹H-NMR (CD₃CN, 400 MHz, 298 K) δ/ppm; 4.175 (3H, s), 7.29–7.33 (1H, ddd, ³J = 5.4–8.1 Hz, ⁴J = 1.2 Hz), 7.37–7.41 (1H, ddd, ³J = 6.8–8.2 Hz, ⁴J = 1.2 Hz), 7.47 (1 H, t, ³J = 4.8 Hz), 7.57 (1 H, ddd, ³J = 8.1 Hz, ⁴J = 0.9 Hz), 7.78 (1 H, ddd, ³J = 8.0 Hz, ⁴J = 0.8 Hz), 8.95 (2 H, d, ³J = 4.8 Hz). ¹³C-NMR (CD₃CN, 101 MHz, 298 K) δ/ppm; 159.59 (C_q), 158.48 (CH), 150.33 (C_q), 143.59 (C_q), 138.12 (C_q), 124.72 (CH), 123.46 (CH), 121.81 (CH), 121.27 (CH), 111.67 (CH), 33.28 (CH₃). ESI-MS (soft positive mode – MeOH + CHCl₃ + HCOOH): [L6 + H]⁺ = 211.3 (exp.) 211.09 (calc.), [L6 + Na]⁺ = 233.1 (exp.) 233.08 (calc.), 170.3 (fragments). Elemental analysis: calculated (%): C 68.56, H 4.79, N 26.65; found (%): C 68.39, H 4.69, N 26.82.

Preparation of N-methyl-N-(2-nitrophenyl)pyrimidine-4-carboxamide (4)

Oxalyl chloride (55.33 mL, 14.0 g, 110.3 mmol, 10 eq.) was added dropwise to a mixture of pyrimidine-2-carboxylic acid (1.37 g, 11.03 mmol, 1 eq.) dissolved in dry dichloromethane (35 mL). Dimethylformamide (300 μL) was added to the mixture to catalyze the reaction. The solution was refluxed for 2 h under an inert atmosphere of nitrogen, then cooled and evaporated to dryness. The residue was dissolved in 40 mL of dry dichloromethane. N-Methyl-2-nitroaniline (2, 4.19 g, 27.57 mmol, 2.5 eq.) dissolved in 10 mL dichloromethane was added dropwise to the solution. The solution was refluxed for 15 h during which two portions of (iPr)₂NEt (Hünig's base, 3.5 mL, 20.09 mmol, 1.82 eq.) were added after 30 min and 12 h. The solution was concentrated under vacuum to afford a solid residue which was partitioned between CH₂Cl₂ (400 mL) and half saturated aqueous NH₄Cl (4 × 120 mL). The aqueous phase was further extracted with CH₂Cl₂ (3 × 50 mL). The combined organic phase was dried over anhydrous Na₂SO₄, concentrated under vacuum and purified by column chromatography (silica, CH₂Cl₂/MeOH 99 : 1) to give 2.62 g of N-methyl-N-(2-nitrophenyl)pyrimidine-4-carboxamide (4, 8.34 mmol, yield 76%) as a brown viscous liquid. ¹H NMR (CDCl₃, 400 MHz, 298 K) δ/ppm: mixture of rotamers A (15.1%) and B (84.9%): 3.49 (3H, s, B), 3.52 (3H, s, A), 7.34–7.77 (4H, m, A and B), 7.95 (1H, dd, ³J = 8.2 Hz, ⁴J = 1.4 Hz, B), 8.11 (1H, dd, ³J = 8.1 Hz, ⁴J = 1.1 Hz, A), 8.75 (1H, d, ³J = 5.1 Hz, B), 8.95 (1H, d, ³J = 5.0 Hz, A), 8.78 (1H, d, ³J = 1.2 Hz, B), 9.33 (1H, A). ESI-MS (soft positive mode – MeOH + CHCl₃ + HCOOH): [4 + H]⁺ = 259.3 (exp.) 259.3 (calc.), [4 + Na]⁺ = 281.3 (exp.) 281.2 (calc.), 213.4, 184.3, 136.3, 119.3, 108.3 (fragments).

Preparation of 1-methyl-2-(pyrimidin-4-yl)-1H-benzo[d]imidazole (L7)

N-Methyl-N-(2-nitrophenyl)pyrimidine-4-carboxamide (4, 2.60 g, 10.06 mmol, 1 eq.) was dissolved in EtOH : DMF (20 mL : 25 mL). Sodium dithionite (7 g, 40.20 mmol, 3.97 eq.) was added to the solution and the temperature of the mixture was raised to 60 °C when 20 mL of distilled water was added. After refluxing for 24 h, the mixture was neutralized using aqueous ammonia and the solvents were removed in vacuo. The concentrate was dissolved in CH₂Cl₂ (50 mL) and washed with water (3 × 200 mL). The combined aqueous layers were extracted with CH₂Cl₂ (3 × 50 mL). The organic fraction was then concentrated under vacuum and purified by column chromatography (silica, CH₂Cl₂/MeOH 98.5 : 1.5) to yield L7 (5.21 mmol, yield 52%). The compound was crystallized as needles by slow diffusion of n-hexane into CH₂Cl₂. ¹H-NMR (CD₃CN, 400 MHz, 298 K) δ/ppm; 4.31 (3H, s), 7.33 (1H, ddd, ⁴J = 1.2 Hz, ³J = 7.9 Hz), 7.41 (1H, ddd, ⁴J = 1.2 Hz, ³J = 8.0 Hz), 7.59 (1H, dt, ⁴J = 1 Hz, ³J = 8.2 Hz), 7.76 (1H, dt, ⁴J = 1 Hz, ³J = 8.1 Hz), 8.32 (1H, dd, ⁴J = 1.6 Hz, ³J = 5.2 Hz), 8.89 (1H, d, ³J = 5.2 Hz), 9.28 (1H, d, ⁴J = 1.6 Hz). ¹³C-NMR (CD₃CN, 101 MHz, 298 K) δ/ppm; 158.99 (CH), 158.38 (C_q), 148.79 (C_q), 143.51 (C_q), 138.73 (C_q), 125.13 (CH), 123.82 (CH), 121.38 (CH), 121.03 (CH), 111.78 (CH), 33.54 (CH₃). ESI-MS (soft positive mode – MeOH + CHCl₃ + HCOOH) [L7 + H]⁺ = 210.6 (exp.) 211.09 (calc.), [L7 + Na]⁺ = 233.1 (exp.) 233.08 (calc.), 82.4 (fragments). Elemental analysis: calculated (%): C 68.56, H 4.79, N 26.65; found (%): C 68.32, H 4.72, N 27.00.

Preparation [M(L6)₃]X₂ and [M(L7)₃]X₂ complexes (M = Fe, Ni, Zn and X = CF₃SO₃⁻, BF₄⁻, PF₆⁻, ClO₄⁻)

In a typical synthesis, 0.3 mmol (3 eq.) of the ligand L6 or ligand L7 dissolved in 2 mL of acetonitrile was added to 0.1 mmol (1 eq.) of Ni(BF₄)₂·6H₂O or Zn(CF₃SO₃)₂ or Fe(ClO₄)₂·xH₂O or Fe(BF₄)₂·6H₂O in 2 mL acetonitrile. The resulting mixtures were stirred under an inert atmosphere for 3 h for ensuring complete dissolution of all components and then evaporated to dryness to yield microcrystalline powders of the respective complexes. The powders were re-dissolved in acetonitrile or methanol and allowed to crystallize by evaporation, or diffusion of tert-butyl methyl ether into acetonitrile or methanol after metathesis in presence of a large excess (10 eq.) of (ⁿBu)₄NClO_4, (ⁿBu)₄NBF₄ or (ⁿBu)₄NPF₆, respectively, to give primary salts [Ni(L6)₃](BF₄)₂·1.55CH₃CN·0.4H₂O, [Zn(L6)₃](CF₃SO₃)₂·0.6H₂O, [Fe(L6)₃](BF₄)₂·0.15CH₃CN·1.2H₂O, [Ni(L7)₃](BF₄)₂·0.85H₂O, [Zn(L7)₃](PF₆)₂·1.25H₂O, [Fe(L7)₃](PF₆)_1.76(BF₄)_0.24·0.6H₂O complexes in 46–72% yield. Only non-perchlorate complexes were characterized by elemental analyses (Table S10†). Single crystals of sufficient quality for X-ray diffraction studies could be obtained and X-ray crystal structures could be solved for [Zn(L6)₃](ClO₄)₂(C₂H₃N)_2.5 (1), [Zn(L6)₃](BF₄)₂·(CH₃CN)₂ (2), [Ni(L6)₃](ClO₄)₂ (3), [Ni(L6)₃](ClO₄)₂·C₂H₃N (4), [Ni(L6)₃](BF₄)₂·C₂H₃N (5), [Fe(L6)₃](ClO₄)₂ (6), [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7), [Zn(L7)₃](ClO₄)₂ (8) [Zn(L7)₃](PF₆)₂ (9) [Ni(L7)₃](ClO₄)_1.48(PF₆)_0.52·CH₃CN (10), [Ni(L7)₃](BF₄)₂·CH₃OH·(CH₃CN)_0.5 (11), [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12) and [Fe(L7)₃](PF₆)_1.72(ClO₄)_0.28·CH₃OH (13).

Caution! Dry perchlorates may explode and should be handled in small quantities and with the necessary precautions.^122,123

Spectroscopic and analytical measurements

¹H, ¹³C NMR and ¹H–¹⁵N HMBC spectra were recorded at 298 K on a Bruker Avance 400 MHz spectrometer. Chemical shifts are given in ppm with respect to TMS for ¹H and ¹³C and nitromethane for ¹⁵N. Spectrophotometric titrations were performed with a J&M diode array spectrometer (Tidas series) connected to an external computer. In a typical experiment, 25 cm³ of ligand L6 or ligand L7 in acetonitrile (∼2 × 10⁻⁴ M) was titrated at 298 K with a solution of Fe(CF₃SO₃)₂ and Ni(BF₄)₂·6H₂O or Zn(CF₃SO₃)₂ (∼2 × 10⁻³ M) in acetonitrile under an inert atmosphere. After each addition of 33 μL, the absorbance was recorded using Hellma optrodes (optical path length 0.1 cm) immersed in the thermostated titration vessel and connected to the spectrometer. Mathematical treatment of the spectrophotometric titrations was performed with factor analysis^67–69 and with ReactLab™ Equilibria (previously Specfit/32).^69,71,72 Pneumatically-assisted electrospray (ESI-MS) mass spectra were recorded from ∼1 × 10⁻⁴ M (ligands) and ∼1 × 10⁻³ M (complexes) solutions on an Applied Biosystems API 150EX LC/MS System equipped with a Turbo Ionspray source. Elemental analyses were performed by K. L. Paglia from the Microchemical Laboratory of the University of Geneva. Electronic spectra in the UV-Vis region were recorded at 293 K from solutions in CH₃CN with a PerkinElmer Lambda 1050 using quartz cells of 0.1 or 1.0 mm path length. Solid-state absorption spectra were recorded with a PerkinElmer Lambda 900 using capillaries. Solid-state magnetic data were recorded on MPMS 3 QUANTUM DESIGN magnetometers using magnetic fields of 1000–5000 Oe and at 1 K min⁻¹ rates within the 5–400 K range (5–300 K range for perchlorate containing complexes). The magnetic susceptibilities were corrected for the magnetic response of the sample holder and the diamagnetism of the compounds by using the approximation .¹²⁴

X-ray crystallography

Summary of crystal data, intensity measurements and structure refinements for ligands L6, L7 and complexes [Zn(L6)₃](ClO₄)₂(C₂H₃N)_2.5 (1), [Zn(L6)₃](BF₄)₂·(CH₃CN)₂ (2), [Ni(L6)₃](ClO₄)₂ (3), [Ni(L6)₃](ClO₄)₂·C₂H₃N (4), [Ni(L6)₃](BF₄)₂·C₂H₃N (5), [Fe(L6)₃](ClO₄)₂ (6), and [Fe(L6)₃](BF₄)₂·(C₅H₁₂O)_0.5·(C₂H₅N)_0.5 (7), [Zn(L7)₃](ClO₄)₂ (8) [Zn(L7)₃](PF₆)₂ (9) and [Ni(L7)₃](ClO₄)_1.48(PF₆)_0.52·CH₃CN (10), [Ni(L7)₃](BF₄)₂·CH₃OH·(CH₃CN)_0.5 (11), [Fe(L7)₃](PF₆)(BF₄)·CH₃OH (12) and [Fe(L7)₃](PF₆)_1.72(ClO₄)_0.28·CH₃OH (13) were collected in Tables S1, S4 and S8–S11.† Pertinent bond lengths, bond angles and interplanar angles were collected in Tables S2, S3, S5, S6, S15–S40† together with ORTEP views and the pertinent numbering schemes and intermolecular interactions gathered in Fig. S15–S20 and S36–S48.† The crystals were mounted on Hampton cryoloops with protection oil. X-ray data collections were performed with an Agilent SuperNova Dual diffractometer equipped with a CCD Atlas detector (Cu[Kα] radiation) or a XtaLAB Synergy-S diffractometer equipped with an hybrid pixel hypix arc 150 detector. The structures were solved by using direct methods.^125,126 Full-matrix least-square refinements on F² were performed with SHELX2014.¹²⁷ CCDC 2041614–2041630† contain the supplementary crystallographic data.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Financial support from the Swiss National Science Foundation is gratefully acknowledged (grant numbers: 200020_178758 and 206021_183324).

References

1. L. Pauling, J. Am. Chem. Soc., 1931, 53, 1367–1400.
2. L. Pauling, Phys. Rev., 1931, 37, 1185–1186.
3. L. Pauling, J. Am. Chem. Soc., 1932, 54, 988–1003.
4. C. D. Corryell, F. Stittand and L. Pauling, J. Am. Chem. Soc., 1937, 59, 633–642.
5. L. Cambi and L. Szego, Ber. Dtsch. Chem. Ges., 1931, 64, 2591–2598.
6. L. Cambi, L. Szego and A. Cagnasso, Atti Accad. Lincei, 1932, 15, 266–271.
7. L. Cambi, L. Szego and A. Cagnasso, Atti Accad. Lincei, 1932, 15, 329–335.
8. H. Bethe, Ann. Phys., 1929, 3, 133–208.
9. C. K. Jorgensen, Modern Aspects of Ligand Field Theory, North Holland Publishing Company, Amsterdam-London, 1971.
10. B. N. Figgis and M. A. Hitchman, Ligand Field Theory and Its Applications, Wiley-VCH, New York, Chichester, Weiheim, Brisbane, Singapore, Toronto, 2000.
11. D. M. P. Mingos, Struct. Bonding, 2016, 172, 1–18.
12. A. Hauser, Top. Curr. Chem., 2004, 233, 49–58.
13. Y. Tanabe and S. Sugano, J. Phys. Soc. Jpn., 1954, 9, 766–779.
14. P. Gütlich, Y. Garcia and H. A. Goodwin, Chem. Soc. Rev., 2000, 29, 419–427.
15. S. Alvarez, J. Am. Chem. Soc., 2003, 125, 6795–6802.
16. N. Deorukhkar, T. Lathion, L. Guénée, C. Besnard and C. Piguet, Chemistry, 2020, 2, 231–252.
17. R. J. Deeth, C. M. Handley and B. J. Houghton, in Spin-Crossover Materials, ed. M. A. Halcrow, Wiley, Chichester, 2013, vol. 17, pp. 443–454.
18. F. Neese, Coord. Chem. Rev., 2009, 253, 526–563.
19. C. J. Cramer and D. G. Truhlar, Phys. Chem. Chem. Phys., 2009, 11, 10757–10816.
20. S. Rackwitz, W. Klopper, V. Schünemann and J. A. Wolny, Phys. Chem. Chem. Phys., 2013, 15, 15450–15458.
21. L. J. K. Cook, R. Kulmaczewski, R. Mohammed, S. Dudley, S. A. Barrett, M. A. Little, R. J. Deeth and M. A. Halcrow, Angew. Chem., Int. Ed., 2016, 55, 4327–4331.
22. K. P. Kepp, Inorg. Chem., 2016, 55, 2717–2727.
23. M. Swart and M. Gruden, Acc. Chem. Res., 2016, 49, 2690–2697.
24. S. Rodriguez-Jimenez, M. Yang, I. Stewart, A. L. Garden and S. Brooker, J. Am. Chem. Soc., 2017, 139, 18392–18396.
25. P. Gütlich and H. A. Goodwin, Top. Curr. Chem., 2004, 233, 1–47.
26. M. A. Halcrow, Polyhedron, 2007, 26, 3523–3576.
27. B. Weber, in Spin-Crossover Materials, ed. M. A. Halcrow, Wiley, Chichester, 2013, vol. 2, pp. 55–76.
28. J. Olguin and S. Brooker, in Spin-Crossover Materials, ed. M. A. Halcrow, Wiley, Chichester, 2013, vol. 3, pp. 77–120.
29. J. Olguin and S. Brooker, Coord. Chem. Rev., 2011, 255, 203–240.
30. S. Brooker, Chem. Soc. Rev., 2015, 44, 2880–2892.
31. H. L. C. Feltham, A. S. Barltrop and S. Brooker, Coord. Chem. Rev., 2016, 344, 26–53.
32. R. W. Hogue, S. Singh and S. Brooker, Chem. Soc. Rev., 2018, 47, 7303–7338.
33. H. S. Scott, R. W. Staniland and P. E. Kruger, Coord. Chem. Rev., 2018, 362, 24–43.
34. K. S. Kumar, Y. Bayeh, T. Gebretsadik, F. Elemo, M. Gebrezgiabher, M. Thomas and M. Ruben, Dalton Trans., 2019, 48, 15321–15337.
35. J. Olguin, Coord. Chem. Rev., 2020, 407, 213148.
36. Spin Crossover in Transition Metal Compounds I-III, in Topics in Current Chemistry, ed. P. Gütlich and H. A. Goodwin, Springer-Verlag, Berlin-Heidelberg-New York, 2004, vol. 233–235.
37. M. A. Halcrow, Spin-Crossover Materials, John Wiley & Sons, Ltd, Chichester, United Kingdom, 2013.
38. H. Phan, J. J. Hrudka, D. Igimbayeva, L. M. L. Daku and M. Shatruk, J. Am. Chem. Soc., 2017, 139, 6437–6447.
39. O. Kahn, Molecular Magnetism, VCH Publishers, Weinheim, 1993, pp. 59–69.
40. P. Guionneau, J.-F. Létard, D. S. Yufit, D. Chasseau, G. Bravic, A. E. Goeta, J. A. K. Howard and O. Kahn, J. Mater. Chem., 1999, 9, 985–994.
41. M. Mikolasek, G. Félix, W. Nicolazzi, G. Molnar, L. Salmon and A. Bousseksou, New J. Chem., 2014, 38, 1834–1839.
42. C. Enachescu, C. M. Nishino and S. Miyashita, in Spin-Crossover Materials, ed. M. A. Halcrow, Wiley, Chichester, 2013, vol. 18, pp. 455–474.
43. M. Sorai, Bull. Chem. Soc. Jpn., 2001, 74, 2223–2253.
44. B. Weber and F. A. Walker, Inorg. Chem., 2007, 46, 6794–6803.
45. M. P. Shores, C. M. Klug and S. R. Fiedler, in Spin-Crossover Materials, ed. M. A. Halcrow, Wiley, Chichester, 2013, vol. 10, pp. 281–302.
46. C. P. Slichter and H. G. Drickamer, J. Chem. Phys., 1972, 56, 2142–2160.
47. M. Sorai and S. Seki, J. Phys. Chem. Solids, 1974, 35, 555–570.
48. A. Bousseksou, G. Molnar and G. Matouzenko, Eur. J. Inorg. Chem., 2004, 4353–4369.
49. A. Bousseksou, G. Molnar, L. Salmon and W. Nicolazzi, Chem. Soc. Rev., 2011, 40, 3313–3335.
50. P. Gutlich, A. B. Gaspar and Y. Garcia, Beilstein J. Org. Chem., 2013, 9, 342–391.
51. G. Molnar, L. Salmon, W. Nicolazzi, F. Terki and A. Bousseksou, J. Mater. Chem. C, 2014, 2, 1360–1366.
52. M. D. Manrique-Juarez, S. Rat, L. Salmon, G. Molnar, C. M. Quintero, L. Nicu, H. J. Shepherd and A. Bousseksou, Coord. Chem. Rev., 2016, 308, 395–408.
53. A. B. Gaspar and M. Seredyuk, Coord. Chem. Rev., 2014, 268, 41–58.
54. A. B. Gaspar, V. Ksenovontov, M. Seredyuk and P. Gütlich, Coord. Chem. Rev., 2005, 249, 2661–2676.
55. K. S. Kumar and M. Ruben, Coord. Chem. Rev., 2017, 346, 176–205.
56. Y. S. Meng and T. Liu, Acc. Chem. Res., 2019, 52, 1369–1379.
57. M. Darawsheh, L. A. Barrios, O. Roubeau, S. J. Teat and G. Aromi, Angew. Chem., Int. Ed., 2018, 57, 13509–13513.
58. A. Kawamura, J. Xie, J. N. Boyn, K. A. Jesse, A. J. McNeece, E. A. Hill, K. A. Collins, J. A. Valdez-Moreira, A. S. Filatov, J. W. Kurutz, D. A. Mazziotti and J. S. Anderson, J. Am. Chem. Soc., 2020, 142, 17670–17680.
59. H. J. Shepherd, C. M. Quintero, G. Molnar, L. Salmon and A. Bousseksou, in Spin-Crossover Materials, ed. M. A. Halcrow, Wiley, Chichester, 2013, vol. 13, pp. 347–373.
60. T. Lathion, A. Furstenberg, C. Besnard, A. Hauser, A. Bousseksou and C. Piguet, Inorg. Chem., 2020, 59, 1091–1103.
61. T. Lathion, L. Guénée, C. Besnard, A. Bousseksou and C. Piguet, Chem. – Eur. J., 2018, 24, 16873–16888.
62. C. Piguet, B. Bocquet and G. Hopfgartner, Helv. Chim. Acta, 1994, 77, 931–942.
63. C. E. Hadden, Magn. Reson. Chem., 2005, 43, 330–333.
64. K. Nakamoto, J. Phys. Chem., 1960, 64, 1420–1425.
65. C. Piguet, J.-C. G. Bünzli, G. Bernardinelli, C. G. Bochet and P. Froidevaux, J. Chem. Soc., Dalton Trans., 1995, 83–97.
66. S. Xu, J. E. T. Smith and J. M. Weber, Inorg. Chem., 2016, 55, 11937–11943.
67. E. R. Malinowski and D. G. Howery, Factor Analysis in Chemistry, Wiley, New York, Chichester, 1980.
68. H. Gampp, M. Maeder, C. J. Meyer and A. D. Zuberbuehler, Talanta, 1985, 32, 1133–1139.
69. H. Gampp, M. Maeder, C. J. Meyer and A. D. Zuberbuehler, Talanta, 1986, 33, 943–951.
70. B. R. Hall, L. E. Manck, I. S. Tidmarsh, A. Stephenson, B. F. Taylor, E. J. Blaikie, D. A. Vander Griend and M. D. Ward, Dalton Trans., 2011, 40, 12132–12145.
71. M. Maeder and P. King, Analysis of Chemical Processes, Determination of the Reaction Mechanism and Fitting of Equilibrium and Rate Constants, in Chemometrics in Practical Applications, ed. K. Varmuza, InTech., 2012, ISBN: 978-953-51-0438-4,  DOI:10.5772/31896.
72. ReactLab™ Equilibria (previously Specfit/32), Jplus Consulting Pty. Ltd., Palmyra WA 6957, Australia: http://jplusconsulting.com/products/reactlab-equilibria/.
73. L. Alderighi, P. Gans, A. Ienco, D. Peters, A. Sabatini and A. Vacca, Coord. Chem. Rev., 1999, 184, 311–318 , http://www.hyperquad.co.uk/hyss.htm.
74. L. Aboshyan-Sorgho, T. Lathion, L. Guénée, C. Besnard and C. Piguet, Inorg. Chem., 2014, 53, 13093–13104.
75. R. M. Smith and A. E. Martell, Critical Stability Constants, Plenum, New York, 1982, vol. 5, pp. 235–253.
76. M. Borkovec, J. Hamacek and C. Piguet, Dalton Trans., 2004, 4096–4105.
77. J. Hamacek, M. Borkovec and C. Piguet, Dalton Trans., 2006, 1473–1490.
78. C. Piguet, Chem. Commun., 2010, 46, 6209–6231.
79. S. W. Benson, J. Am. Chem. Soc., 1958, 80, 5151–5154.
80. G. Ercolani, C. Piguet, M. Borkovec and J. Hamacek, J. Phys. Chem. B, 2007, 111, 12195–12203.
81. G. Calzaferri, Phys. Chem. Chem. Phys., 2017, 19, 10611–10621.
82. H. Irving and R. J. P. Williams, J. Chem. Soc., 1953, 3192–3210.
83. P. W. Dimmock, P. Warwick and R. A. Robbins, Analyst, 1995, 120, 2159–2170.
84. B. J. Coe and S. J. Glenwright, Coord. Chem. Rev., 2000, 203, 5–80.
85. M. A. Robinson, D. H. Busch and J. D. Curry, Inorg. Chem., 1963, 2, 1178–1181.
86. A. B. P. Lever, Inorganic Electronic Spectroscopy, Elsevier, Amsterdam-Oxford-New York-Tokyo, 1984, p. 126.
87. S. M. Hart, J. C. A. Boeyens and R. D. Hancock, Inorg. Chem., 1983, 22, 982–986.
88. B. N. Figgis and M. A. Hitchman, Ligand Field Theory and its Applications, Wiley-VCH, New York, NY, USA, 2000.
89. B. Weber and F. A. Walker, Inorg. Chem., 2007, 46, 6794–6803.
90. A. A. Pavlov, G. L. Denisov, M. A. Kiskin, Y. V. Nelyubina and V. V. Novikov, Inorg. Chem., 2017, 56, 14759–14762.
91. D. F. Evans, J. Chem. Soc., 1959, 2003–2005.
92. T. H. Crawford and J. Swanson, J. Chem. Educ., 1971, 48, 382–386.
93. J. Löliger and R. Scheffold, J. Chem. Educ., 1972, 49, 646–647.
94. D. Ostfeld and I. A. Cohen, J. Chem. Educ., 1972, 49, 829.
95. D. H. Grant, J. Chem. Educ., 1995, 72, 39–40.
96. C. Piguet, J. Chem. Educ., 1997, 74, 815–816.
97. H. Toftlund, Coord. Chem. Rev., 1989, 94, 67–108.
98. H. Toftlund, Monatsh. Chem., 2001, 132, 1269–1277.
99. S. Rodriguez-Jimenez, A. S. Barltrop, N. G. White, H. L. C. Feltham and S. Brooker, Inorg. Chem., 2018, 57, 6266–6282.
100. S. Singh and S. Brooker, Inorg. Chem., 2020, 59, 1265–1273.
101. C. Piguet, Dalton Trans., 2011, 40, 8059–8071.
102. D. M. Ford, J. Am. Chem. Soc., 2005, 127, 16167–16170.
103. E. B. Starikov, J. Appl. Solution Chem. Model., 2013, 2, 240–245.
104. O. A. Khakhel and T. P. Romashko, J. Phys. Chem. A, 2016, 120, 2035–2040.
105. K. P. Kepp, Inorg. Chem., 2016, 55, 2717–2727.
106. P. W. Atkins and A. J. MacDermott, J. Chem. Educ., 1982, 59, 359–360.
107. G. Schreckenbach, Chem. – Eur. J., 2017, 23, 3797–3803.
108. L. Onsager, J. Am. Chem. Soc., 1936, 58, 1486–1492.
109. D. V. Matyushov, J. Chem. Phys., 2004, 120, 1375–1382.
110. C. J. Cramer and D. G. Truhlar, Acc. Chem. Res., 2008, 41, 760–768.
111. L. E. Johnson, S. J. Benight, R. Barnes and B. H. Robinson, J. Phys. Chem. B, 2015, 119, 5240–5250.
112. L. Babel, T. N. Y. Hoang, L. Guénée, C. Besnard, T. A. Wesolowski, M. Humbert-Droz and C. Piguet, Chem. – Eur. J., 2016, 22, 8113–8123.
113. Y. Furusho, H. Goto, K. Itomi, H. Katagiri, T. Miyagawa and E. Yashima, Chem. Commun., 2011, 47, 9795–9797.
114. D. Furushou, T. Hashibe, T. Fujinami, K. Nishi, H. Hagiwara, N. Matsumoto, Y. Sunatsuki, M. Kojima and S. Iijima, Polyhedron, 2013, 52, 1489–1498.
115. SHAPE is a free software developed by M. Llunell, D. Casanova, J. Cirera, P. Alemany and S. Alvarez, and available online at http://www.ee.ub.edu/.
116. A. G. Orpen, L. Brammer, F. H. Allen, O. Kennard, D. G. Watson and R. Taylor, J. Chem. Soc., Dalton Trans., 1989, S1–S83.
117. F. H. Allen, O. Kennard, D. G. Watson, L. Brammer, A. G. Orpen and R. Taylor, J. Chem. Soc., Perkin Trans. 2, 1987, S1–S19.
118. N. Sasaki and T. Kambara, Phys. Rev. B: Condens. Matter Mater. Phys., 1989, 40, 2442–2449.
119. B. J. C. Vieira, J. T. Coutinho, I. C. Santos, L. C. J. Pereira, J. C. Waerenborgh and V. da Gama, Inorg. Chem., 2013, 52, 3845–3850.
120. D. Chernyshov, M. Hostettler, K. W. Tornroos and H. B. Burgi, Angew. Chem., Int. Ed., 2003, 42, 3825–3830.
121. P. Lian, H. Wei, C. Zheng, Y. Nie, J. Bian, Z. Bian and C. Huang, Dalton Trans., 2011, 40, 5476–5482.
122. W. C. Wolsey, J. Chem. Educ., 1973, 50, A335–A337.
123. J.-L. Pascal and F. Favier, Coord. Chem. Rev., 1998, 178–180, 865–902.
124. G. A. Bain and J. F. Berry, J. Chem. Educ., 2008, 85, 532.
125. G. M. Sheldrick, Acta Crystallogr., Sect. A: Found. Crystallogr., 2008, 64, 112–122.
126. M. C. Burla, M. Camalli, B. Carrozzini, G. L. Cascarano, C. Giacovazzo, G. Polidori and R. Spagna, Acta Crystallogr., Sect. A: Found. Crystallogr., 1999, 55, 991–999.
127. G. M. Sheldrick, Acta Crystallogr., Sect. C: Struct. Chem., 2015, 71, 3–8.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 2041614–2041630. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0dt03828h

---