Spin Crossover in Three Mononuclear Iron (III) Schiff Base Complexes

Abstract

The synthesis, crystal structure, and magnetic properties of three new mononuclear complexes [Fe(R-LA)(L1)](BPh4), where R-LA²⁻ is a doubly deprotonated pentadentate Schiff base ligand and L1 is a monodentate benzimidazole or furopyridine ligand, are reported. Ligand- and anion-driven changes in crystal structures and magnetic behavior were investigated in terms of the magnetic susceptibility measurements and theoretical calculations.

1. Introduction

Spin crossover (SCO) of (pseudo)octahedral complexes of iron(III) manifests as spin transition between S = 1/2 (low-spin, LS) and S = 5/2 (high-spin, HS) states [1]. Typically, the iron(III) complexes exhibit SCO for several types of coordination environments such as {FeO₆}, {FeO₃S₃}, {FeS₆}, {FeN₂O₂S₂}, {FeN₃O₃}, or {FeN₆}, but the most explored are the {FeN₄O₂} complexes with tri-, tetra-, penta-, or hexadentate Schiff base ligands [2]. For the last decade our attention focused predominantly on the Fe(III) complexes with pentadentate Schiff bases originating from reactions between derivatives of various ortho-hydroxy salycilaldehydes and aliphatic triamines [3,4,5,6,7]. In these compounds, two basic types of SCO complexes can be recognized due to different kinds of amine used in the synthesis of Schiff base ligands: so-called symmetric ones in which derivatives of bis(3-aminopropyl)amine are used (H₂R-LA, R- substituted compounds of 4-azaheptamethylene-1,7-bis(salicylideneiminate), H₂LA) and asymmetric ones with N-(2-aminoethyl)-1,3-propanediamine (H₂R-LB, R- substituted compounds of 4-azahexamethylene-1,7-bis(salicylideneiminate), H₂LB Scheme 1). Initially, SCO was observed only for compounds with the general formula [Fe(R-LA)(L1)](BPh₄) [8] or [{Fe(R-LA)}₂(μ-L2)](BPh₄) [9], and the observed transitions were of gradual or even spin equilibrium character. One exception was found in the case of a mononuclear compound with R-LA = bis(3-methoxysalicylideneiminopropyl)methylamine and L1 = 4-aminopyridine, which exhibited rather cooperative SCO, possibly due to its crystal packing involving N–H···O and N–H···π non-covalent interactions between the complex cations [10]. Other interesting results were obtained for compounds [Fe(LA)(L1A)][M(dmit)₂]·CH₃CN, M = Ni, Pd, Pt, in which L1A (1-(pyridin-4-yl)-2-(N-methylpyrrol-2-yl) ethane) is a photoisomerable ligand and [M(dmit)₂]⁻ anions can act as molecular conductors [11]. Compounds of the general formula [Fe(LA)(L1B)](BPh₄) also contain photoisomerable ligands L1B = 3-phenylazopyridine or 4-phenylazopyridine [12]. Nevertheless, these complexes are not very interesting from a magnetic point of view because they exhibit only gradual spin transitions. This changed with the introduction of the shorter aliphatic chain in Schiff base ligands by substitution of the derivatives of bis(3-aminopropyl)amine by N-(2-aminoethyl)-1,3-propanediamine. Increased rigidity of the resulting ligands (supported also by the introduction of naphthyl instead of benzene rings) led to preparation of the [Fe(LB)(LP)] complexes (where LP stands for (1) pseduhalido ligand), which exhibited cooperative spin transitions, in some cases even accompanied by thermal hysteresis [3].

Inspired by the abovementioned results we decided to attempt preparation of the [Fe(LA)(L1)](BPh₄) complexes exhibiting cooperative SCO by using monodentate ligands, which could increase the rigidity and the number of significant non-covalent interactions in the resulting complexes. Therefore, we decided to use two different bulky monodentate ligands, benzimidazole (L1^a) and 1-benzofuro[3,2-c]pyridine (L1^b), together with two slightly different pentadentate ligands, H₂LA and H₂3EtO-LA (4-azaheptamethylene-1,7-bis(3-thoxy-salicylideneiminate, Scheme 1). We were successful in the preparation of three compounds: [Fe(3EtO-LA)(L1^a)](BPh₄) (1a), [Fe(3EtO-LA)(L1^b)](BPh₄)·CH₃OH (2a), and [Fe(LA)(L1^b)](BPh₄) (2b). Here, we report their crystal structure and magnetic properties.

2. Materials and Methods

Chemicals were purchased from commercial sources (Sigma-Aldrich) and used as received. 1-benzofuro[3,2-c]pyridine was prepared according to a previously reported procedure [13,14]. Elemental analysis was carried out on a Flash 2000 (ThermoFisher Scientific, Waltham, MA, USA).

Magnetic susceptibility and magnetization measurements were done using a SQUID magnetometer (Quantum Design Inc., San Diego, CA, USA) from T = 2 K at B = 0.1 T. The magnetization data were taken at T = 2.0 and 4.6 K, respectively. Raw susceptibility was corrected, and diamagnetic corrections of the constituent atoms were estimated from Pascal constants. The effective magnetic moment was calculated as usual: μ_eff/μ_B = 798(χ’T)^1/2 when SI units are employed.

Single-crystal X-ray diffraction data were collected on an Oxford diffractometer Xcalibur2 (Oxford Diffraction Ltd., Oxford, UK) with a Sapphire CCD detector and fine-focused sealed tube (Mo Kα radiation, λ = 0.71073 Å) source and equipped with an Oxford Cryosystem nitrogen gas-flow apparatus. All structures were solved and refined (full-matrix least-squares on F_o²−F_c²) by using SHELXS2014 software [15].

2.1. Synthesis

2.1.1. Ligands

Neutral, pentadentate ligands H₂LA and H₂3EtO-LA were prepared by mixing the appropriate aldehyde and amine in a 2:1 molar ratio. Synthesis of all Schiff base ligands is analogous; therefore, only synthesis of H₂LA is presented in detail [9]. A methanol solution of salicylaldehyde (2.44 g, 20 mmol in 50 cm³) was combined with di(3-aminopropyl)amine (1.31 g, 10 mmol) and the mixture was refluxed for 60 min. The ligand was obtained as a yellow solution ready for subsequent use.

2.1.2. Mononuclear Precursors

Synthesis of all precursors was very similar [9]; therefore, only the synthesis of [Fe(LA)Cl] is described. The methanol solution of ligand H₂LA (10 mmol in 50 cm³) was combined with a solution of iron(III) chloride hexahydrate (2.70 g, 10 mmol) in 40 cm³ of methanol. The mixture was refluxed for 20 min, and then triethylamine (2.22 g, 22 mmol) was added to complete deprotonation of the Schiff base ligand. The resulting violet solution was refluxed for 30 min and left to cool slowly to room temperature when a dark violet micro-crystalline powder precipitated. This was filtered off using a fritted funnel, washed with methanol and diethylether, and dried.

2.1.3. Mononuclear Complexes 1a, 2a, and 2b

The complexes were prepared in the same manner by mixing 100 mg of the [Fe(LA)Cl] or [Fe(3EtO-LA)Cl] (0.193 mmol in the preparation of 1a and 2a, 0.233 mmol in the preparation of 2b) precursor complexes with a heterocyclic derivate (23 mg of L1^a for the preparation of 1a, 33/38 mg of L1^b for the preparation of 2a/2b) in a molar ratio of 1:1. After 30 min of reflux, the solution was filtered through paper filter into an equimolar amount of solution of NaBPh₄.

As an example, the synthesis of [Fe(3EtO-LA)(L1^a)](BPh₄) is described in detail. To a solution of [Fe(3EtO-LA)Cl] (100 mg in 30 cm³ of methanol), L1^a was added. The resulting violet solution was refluxed for 30 min and filtered into a solution of NaBPh₄ (66 mg in 5 cm³). Black crystals precipitated overnight. They were collected, washed with methanol and diethyl ether, and dried.

2.1.4. Elemental Analysis

1a: calcd (%) for C₅₅B₁Fe₁H₅₇N₅O₄, M_w = 918.7 g·mol⁻¹, C, 71.9; H, 6.3; N, 7.6. Found: C, 71.5; H, 6.1; N, 7.3. Yield = 67%.

2a: calcd (%) for C₆₀B₁Fe₁H₆₂N₄O₆, M_w = 1001.8 g.mol⁻¹, C, 71.9; H, 6.2; N, 5.6. Found: C, 71.6; H, 6.0; N, 5.2. Yield = 48%.

2b: calcd (%) for C₅₅B₁Fe₁H₅₀N₄O₃, M_w = 881.7 g.mol⁻¹, C, 74.9; H, 5.7; N, 6.4. Found: C, 74.5; H, 5.8; N, 6.1. Yield = 55%.

2.2. Theoretical Calculations

The theoretical calculations were carried out using the ORCA 4.1 computational package [16]. Three density functional theory (DFT) functionals, B3LYP [17,18,19], OPBE [20,21], and TPSSh [22,23], were used to optimize the molecular structures together with the polarized triple-ζ quality basis set def2-TZVP proposed by Ahlrichs and co-workers [24], where “verytightopt” optimization criteria were used in ORCA. The calculations utilized the RI approximation with the decontracted auxiliary def2/J Coulomb fitting basis set [25] and the chain-of-spheres (RIJCOSX) approximation to exact exchange [26,27] as implemented in ORCA. Increased integration grids (Grid5 and Gridx5 in ORCA convention) and tight SCF convergence criteria were used in all calculations. Moreover, the SCF stability test as implemented in ORCA was done for all final optimized geometries to verify whether the SCF solution was at a local minimum and not in a saddle point [28,29].

3. Results and Discussion

3.1. Crystal Structures

The crystal structures were determined by single-crystal X-ray diffraction for all three presented compounds 1a, 2a, and 2b, and these crystallized in triclinic (P-1 for 1a and 2a) and monoclinic (P2₁/c for 2b) space groups (

Table 1. Crystal data and details of structure determination.

Compound	1a	2a	2b
Formula	C55H57BFeN5O4	C59H58BFeN4O5	C55H50BFeN4O3
Formula weight	918.71	969.75	881.65
Crystal system	Triclinic	Triclinic	Monoclinic
Space group	P-1	P-1	P21/c
Cell parameters
a/Å	10.1648(7)	13.2869(4)	19.1978(10)
b/Å	15.2130(10)	13.6926(5)	11.7562(13)
c/Å	16.2320(10)	16.4083(6)	21.381(2)
α/°	81.990(6)	70.732(3)	90
β/°	72.359(6)	78.232(3)	110.458(8)
γ/°	86.049(6)	67.395(3)	90
V/Å3	2367.7(3)	2591.29(17)	4521.2(8)
Z	2	2	4
T/K	100(2)	190(2)	293(2)
Density, Dc/g cm−3	1.289	1.243	1.295
Abs. coefficient/mm−1	0.371	0.344	0.384
Data/restraints/param	8326/0/601	6630/3/644	7931/0/577
R1a, wR2b (all data)	0.0660, 0.1073	0.0545, 0.0867	0.0618, 0.0810
R1a, wR2b [I > 2σ(I)]	0.0379, 0.0978	0.0355, 0.0829	0.0340, 0.0763
Goodness of fit	1.049	1.028	0.954
CSD number	1939657	1939658	1939656

^a R₁ = ∑ (|F_o| – |F_c|)/∑|F_o|. ^b wR² = {∑[w(F²_o − F²_c)²]/∑[w(F²_o)²]}^1/2.

). All three compounds consist of complex Fe(III) cations charge balanced by BPh₄⁻ anions. In 2a, the additional methanol molecule is in its asymmetric unit, which is heavily disordered. It was not possible to model it reasonably; therefore, the SQUEEZE procedure [30] was used to subtract the corresponding electronic density.

The complex cations in 1a, 2a, and 2b consist of a pentadentate Schiff base ligand (3EtO-LA²⁻ in 1a and 2a, LA²⁻ in 2b) coordinated to the central Fe(III) atom, and the sixth coordination site is occupied by a monodentate N-donor heterocyclic ligand (L1^b in 2a and 2b, L1^b in 1a). The pentadentate Schiff base ligands coordinate iron centers in a cis-conformation of the oxygen atoms, which is typical for compounds with [Fe(R-LA)]⁺ cations (Figure 1) [31]. The monodentate ligand is in a trans-position to the secondary amine group of the pentadentate ligand.

The metal–ligand bond lengths (Figure 1) are close to the values typical for the LS state with the longest bonds observed between the iron atoms and secondary nitrogen atoms of the pentadentate ligands (in Å, 2.0293(19) in 1a, 2.0539(16) in 2a, 2.0843(15) in 2b) or the nitrogen atom of the heterocyclic monodentate ligand (in Å, 2.0105(18) in 1a, 2.0497(15) in 2a, 2.0990(16) in 2b). The Fe–N bonds involving the imino nitrogen atoms are rather shorter and similar for all three structures, ranging between 1.98 to 2.01 Å. The Fe–O bonds are even shorter: 1.87–1.89 Å. The angular distortion parameter Σ is rather small [32]: 17.4° (1a), 23.7° (2a), and 30.3 (2b).

The crystal structures of 1a, 2a, and 2b do not contain hydrogen bonding of significant strength. In 1a, the secondary amine group from the complex cation forms offset N–H···π non-covalent contact with the aromatic ring of the BPh₄⁻ anion. The shortest N···C distance is 3.446(3) Å (Figure S1 in supplementary materials). The N–H group from benzimidazole also forms N–H···π interaction with the aromatic ring of the BPh₄⁻ anion—the N···Cg distance is 3.227(4) Å (where Cg stands for the ring centroid).

In 2a, the interaction between complex cations is provided by very offset ring–ring interactions with a shortest C···C distance of 3.192(3) Å. Other non-covalent interactions include weak C–H···O and C–H···π contacts (Figure S2).

In 2b, the secondary amine group from the complex cation forms weakly offset N–H···π interaction with the aromatic ring of the BPh₄⁻ anion (the shortest N···C distance is 3.713(3) Å). Other non-covalent interactions in 2b are weak C–H···O and C–H···π contacts (Figure S3).

3.2. Magnetic Properties

The temperature dependence of the effective magnetic moment for 1a, 2a, and 2b is shown in Figure 2. All three compounds undergo spin crossover from the LS to the HS state (S = 1/2 → 5/2), which start above ca. 150 K. Evidently, the spin crossover is incomplete until 300 K, because room temperature values of the effective magnetic moment are less than the spin-only value for S = 5/2 and g = 2.0 (5.93 μ_B). Moreover, the low temperature values of μ_eff vary in the range ≈2–3 μ_B, which suggests that a small portion of iron(III) complexes stay in the HS state. This is also supported by the field dependence of molar magnetization measurements (measured at 2 and 4.6 K, Figure S4), which unequivocally confirm the LS ground state with a larger contribution of non-converted HS molecules in 1a and 2a (M_mol/N_Aμ_B = 1.5 in 1a, 1.4 in 2a at 2 K and 7 T) than in 2b (M_mol/N_Aμ_B = 1.1 at 2 K and 7 T).

The experimental data were analyzed with the help of the Ising-like model [33,34] having following Hamiltonian

$$\widehat{H}=\frac{\Delta }{2}\widehat{\sigma }-\gamma \langle \sigma \rangle \widehat{\sigma }$$

(1)

where σ is fictitious spin with eigenvalues −1 for LS and +1 for HS states, Δ is the energy difference between HS and LS states, γ stands for the cooperativeness of the system (γ > 0), and <σ> is the thermal average of the fictitious spin, which is calculated by solving the implicit equation

$$\langle \sigma \rangle =\frac{-1+r_{\mathrm{eff}}\exp \left[-\left(\Delta -2\gamma \langle \sigma \rangle \right)/kT\right]}{+1+r_{\mathrm{eff}}\exp \left[-\left(\Delta -2\gamma \langle \sigma \rangle \right)/kT\right]}$$

(2)

where r_eff is the effective degeneracy ratio of HS and LS states, and it incorporates both the spin and vibrational degeneracies of the respective spin states [35]. Then, the molar fraction of HS species, x’_HS, is computed as

$${x^{\prime }}_{\mathrm{HS}}=\frac{1}{2}\left(1+\langle \sigma \rangle \right).$$

(3)

In order to fit the experimental magnetic data, the overall susceptibility was calculated as

$${\chi }_{\mathrm{mol}}=({x^{″}}_{\mathrm{HS}}+x_{\mathrm{rHS}}){\chi }_{\mathrm{HS}}+\left(1-{x^{″}}_{\mathrm{HS}}-x_{\mathrm{rHS}}\right){\chi }_{\mathrm{LS}}$$

(4)

where x_rHS is the mole fraction of the residual high-spin state at low temperature, x″_HS is the rescaled high-spin fraction calculated from the Ising-like model as x″_HS = x′_HS(1 – x_rHS), and the molar susceptibility values for LS and HS states were calculated by the Curie–Weiss law as

$${\chi }_{\mathrm{LS}}=\frac{N_{\mathrm{A}}{\mu }_0{\mu }_{\mathrm{B}}^2{S}_{\mathrm{LS}}(S_{\mathrm{LS}}+1)}{3k}\frac{g_{\mathrm{LS}}^2}{T-{\mathsf{\Theta }}_{\mathrm{LS}}}$$

(5)

$${\chi }_{\mathrm{HS}}=\frac{N_{\mathrm{A}}{\mu }_0{\mu }_{\mathrm{B}}^2{S}_{\mathrm{HS}}(S_{\mathrm{HS}}+1)}{3k}\frac{g_{\mathrm{HS}}^2}{T-{\mathsf{\Theta }}_{\mathrm{HS}}}.$$

(6)

To summarize, the variation of Ising-like model parameters (Δ, γ, ρ_εϕϕ) leads to temperature variation of the HS mole fraction, which is subsequently utilized to calculate the temperature variation of the overall molar susceptibility and, hence, the effective magnetic moment. Due to the fact that the spin crossover is not finished until 300 K, the g_HS value was fixed to 2.0, which is a typical value for HS octahedral iron(III) complexes. Also, the Weiss constant of the HS state was set to zero, because the low temperature data are dominated by the LS fraction. Then, we are left with these additional free parameters: g_LS, x_rHS, and Θ_LS. Finally, a fitting procedure was applied to find the best parameters describing the experimental magnetic data, and the values of the parameters are listed in

Table 2. The magnetic parameters for 1a, 2a, and 2b ^a..

Compound	gLS	ΘLS	xrHS	Δ (K)	γ (K)	reff	ΔH (J mol−1)	ΔS (J K−1 mol−1)	T1/2 (K)
1a	2.34	−4.3	0.15	930	116	35.0	7730	29.6	262
2a	2.01	−1.5	0.10	1200	95	150	9975	41.7	239
2b	2.01	−0.8	0.05	1033	127	17	8592	23.6	365

^a the thermodynamic parameters were calculated as $\Delta H=N_{\mathrm{A}}\Delta$, $\Delta S=R\ln r_{\mathrm{eff}}$, and $T_{1/2}=\Delta H/\Delta S$.

. The spin transition temperatures T_1/2 are increasing in the following order: 2a < 1a < 2b.

3.3. Theoretical Calculations

In order to evaluate the impact of various coordinated heterocyclic ligands on the spin crossover properties of pentacoordinate Schiff base iron(III) complexes, density functional theory (DFT) calculations were employed. Here, we tested three DFT functionals, B3LYP, OPBE, and TPSSh, which were selected by benchmark studies to be suitable functionals for the study of spin crossover phenomena [36,37,38]_. The molecular geometries of the complex cations [Fe(R-LA)(L1)]⁺ of 1a, 2a, and 2b were optimized for LS (doublet) and HS (sextet) states. Furthermore, we added another three analogous SCO complexes, namely, DAVCEJ, DETBEJ, and KISJUS, to address the robustness of this theoretical approach. The respective donor–acceptor distances calculated with the TPSSh functional are summarized in

Table 3. The interatomic donor–acceptor distances for density functional theory (DFT)-optimized geometries of [Fe(R-LA)(L1)]⁺ of 1a, 2a, and 2b and DAVCEJ, DETBEJ, and KISJUS using the TPSSh functional ^a..

Method	Compound	Fe–O	Fe–Nim	Fe–Nam	Fe–Nhetero
X-ray analysis	1a (LS)	1.8896/1.8951	1.9608/1.971	2.0293	2.0105
	DAVCEJ (LS)	1.869/1.880	1.961/1.974	2.091	2.000
	DAVCEJ01 (HS)	1.908/1.920	2.072/2.096	2.257	2.146
	KISJUS (LS, 100 K)	1.863/1.880	1.946/1.954	2.023	1.982
TPSSh (LS)	1a	1.875/1.892	1.949/1.972	2.047	2.010
	2a	1.872/1.880	1.979/1.984	2.054	2.050
	2b	1.888/1.889	2.000/2.010	2.084	2.099
	DAVCEJ	1.875/1.890	1.953/1.968	2.105	2.008
	DETBEJ	1.872/1.899	1.953/1.978	2.044	2.016
	KISJUS	1.867/1.896	1.949/1.972	2.044	1.991
TPSSh (HS)	1a (HS)	1.931/1.950	2.104/2.113	2.282	2.201
	2a (HS)	1.934/1.936	2.105/2.109	2.281	2.229
	2b (HS)	1.939/1.942	2.104/2.107	2.276	2.225
	DAVCEJ (HS)	1.938/1.945	2.095/2.098	2.345	2.202
	DETBEJ (HS)	1.938/1.939	2.105/2.109	2.276	2.228
	KISJUS (HS)	1.935/1.945	2.107/2.109	2.286	2.174

^a distances in Å.

; for other functionals, please see Tables S1 and S2. The experimental X-ray data are available for the low-spin state of 1a, DAVCEJ, and KISJUS, whereas DAVCEJ01 is only available in a high-spin state structure because the reported room temperature X-ray data of DETBEJ and KISJUS01 correspond to incomplete spin transition at this temperature with respect to the magnetic data.

Figure 3 shows a comparison of DFT-optimized donor–acceptor distances with the experimental data. It is evident that none of these DFT functionals provided perfect results: both B3LYP and OPBE overestimated Fe–N_am and Fe–N_het distances; Fe–N_im was overestimated by B3LYP and underestimated by OPBE. Moreover, the Fe–N_am distance of DAVCEJ (HS) is extremely long. It seems to us that the best results were provided by the TPSSh functional. Indeed, only this functional correctly predicted the LS ground state for the studied compounds according to the energy comparison of HS and LS isomers depicted in Figure 4. However, there is large variation in the energy difference, E_HS–E_LS, between 1.92 kcalmol⁻¹ for 2a and 12.91 kcalmol⁻¹ for DETBEJ. Obviously, the E_HS–E_LS energy difference does not correlate with the T_1/2 value of the studied compounds, and this inconsistency could also be ascribed to variation of the entropy within this series. However, all the studied compounds have a very similar composition of the Fe^III complex cation and have BPh₄⁻ as the counterion; thus, it can be anticipated that variation of the entropy for the whole series should be minute. Thus, it is most likely that the large variation in the E_HS–E_LS energy difference is due to imperfectness of the DFT functional and/or due to neglect of the crystal packing and non-covalent intermolecular interactions in the solid state, which cannot be reproduced with geometry optimization of [Fe(R-LA)(L1)]⁺ cations in vacuum.

4. Conclusions

In this article we reported the synthesis, crystal structure, and magnetic properties of three new iron(III) complexes with two different pentadentate Schiff base ligands and monodentate heterocyclic ligands. The main motivation for this research was to increase the rather low cooperativity of spin crossover behavior which was typically observed for this group of compounds previously. In general, it is well established that by introducing stacking interactions, the cooperativity of SCO systems might be enhanced. Therefore, relatively large, rigid, monodentate ligands (benzimidazole or 1-benzofuro[3,2-c]pyridine) capable of forming stacking interactions were deliberately introduced into the synthesis of the reported compounds. However, the prepared compounds did not exhibit any regular patterns of significant stacking interactions. The magnetic properties of the compounds showed that our attempt to increase SCO cooperativity was not successful, and the compounds exhibited weakly cooperative SCO without thermal hysteresis but with rather high critical SCO temperatures (T_1/2 = 262 (1a), 239 (2a), 365 (2b) K). DFT calculations were also employed in the study using three functionals, B3LYP, OPBE, and TPSSh. The best agreement with the experimental structures was found for the TPSSh functional, and also, only this functional identified LS isomers lower in energy than HS isomers. However, the general trend in the energy separation of HS and LS isomers and the spin crossover transition temperature T_1/2 found from the experimental data was not fully recovered, which most probably points to the importance of the intermolecular interactions.