Spin-Crossover in Iron(II) Complexes of N,N′-Disubstituted 2,6-Bis(Pyrazol-3-yl)Pyridines: An Effect of a Distal Substituent in the 2,6-Dibromophenyl Group

Abstract

A series of new bis(pyrazol-3-yl)pyridines (L^R) N,N′-disubstituted by 4-functionalized 2,6-dibromophenyl groups have been synthesized to study the effect of a distal substituent on the spin-crossover (SCO) behaviour of the iron(II) complexes [Fe(L^R)₂](ClO₄)₂ by variable-temperature magnetometry, NMR spectroscopy, and X-ray diffraction. The SCO-assisting tendency of the substituents with different electronic and steric properties (i.e., the bromine atom and the methyl group) in the para-position of the 2,6-dibromophenyl group is discussed. Together with earlier reported SCO-active iron(II) complexes with N,N′-disubstituted bis(pyrazol-3-yl)pyridines, these new complexes open the way for this family of SCO compounds to emerge as an effective ‘tool’ in revealing structure–function relations, a prerequisite for successful molecular design of switchable materials for future breakthrough applications in sensing, switching, and memory devices.

1. Introduction

Metal complexes or frameworks [1] with the transition metal ions able to reversibly switch between low-spin (LS) and high-spin (HS) electronic configurations as a response to an applied stimulus (such as temperature, pressure, light irradiation, or electric or magnetic field [1]) are promising candidates to be incorporated into various materials and devices [2,3]. This ability for a spin-crossover (SCO) between the two configurations with different optical, magnetic, electrical and mechanical properties is exploited in displays [4], switches [5], sensors [6], memory devices [7], thermometers [8], and contrast agents [9] in magnetic resonance imaging [2,3]. The required SCO behavior, which is most often observed in N6-coordinated complexes of iron(II) [10,11], is tailored for the listed applications [12] by chemical modifications to the heterocyclic N-donor ligands [13,14,15,16,17,18,19,20,21,22]. Recognizing them for the ‘truly molecular’ design [12] of SCO compounds (free from crystal packing [23,24] or substate [25,26] effects) relies on structure–function relations [12,13,14,15,16,18] identified for selected series of metal complexes in their systematic studies by an available solution-state technique such as NMR spectroscopy [12]. The results of these studies agree on a rather [27] general tendency of bulky groups close to the donor nitrogen atoms to stabilize the HS state of the metal ion through the steric demand that makes its LS state more unlikely, with a less straightforward role for other, more remote substituents [12].

Among many different ligands found in SCO complexes, such as a popular family of bis(pyrazol-1-yl)pyridines (1-bpp) [12], isomeric bis(pyrazol-3-yl)pyridines (3-bpp) [28] induce a SCO tuned by counterions or solvents via hydrogen bonds with the unsubstituted NH groups [29,30,31,32,33]. The latter, however, render the spin state of the metal ion environment-dependent [34,35] in the resulting complexes that are otherwise HS [12,24,36,37,38,39], thus precluding their further evolution as SCO compounds [12] until very recently [27].

We have proposed [27] a ligand design to obtain the first SCO-active iron(II) (and cobalt(II) [40]) complexes of N,N′-disubstituted 3-bpp by introducing bulky ortho-groups into N-phenyl substituents. Following this research, here we report a series of 3-bpp ligands with 2,6-dibromophenyl groups functionalized in its para-position by another bromine atom and a methyl group with different electronic and steric properties (Scheme 1). The latter allowed us to study the effect of a distal substituent on the spin-state behavior of the iron(II) complexes in an attempt to make new SCO compounds from a 3-bpp family that are amenable to molecular design.

2. Materials and Methods

Synthesis. All synthetic manipulations were carried out in air. Solvents were purchased and purified by distilling from conventional drying agents under an argon atmosphere prior to use. 2,6-Bis(1,3-dioxo-3-ethoxypropyl)pyridine was synthesized as reported previously [41]. 2,6-dibromo- and 2,4,6-tribromophenylhydrazines were synthesize from commercially available anilines using a standard diazotization protocol with a subsequent reduction with SnCl₂ [42]. 2,6-dibromo-4-methyl-phenylhydrazine was also synthesized from the corresponding aniline using a standard diazotization protocol described above. 2,6-dibromo-4-methyl-aniline was synthesized by bromination of p-toluidine with Br₂ in glacial acetic acid [43]. Analytical data (C, H and N contents) were obtained with a Carlo Erba model 1106 microanalyzer.

Synthesis of 2,6-bis(1,3-dioxo-3-ethoxypropyl)pyridine. To a mixture of 2,6-pyridinedicarboxylic ethyl ester (2.23 g, 10 mmol) and ethyl acetate (2.44 mL, 25 mmol) in dry THF (25 mL), sodium hydride was added (1.8 g, 50% in mineral oil). The reaction mixture was refluxed for 4 h and then allowed to cool. The volatiles were evaporated, and the residual solid was washed with diethyl ether (2 × 30 mL), filtered, and dried. The crude product was dispersed in water (30 mL), and the resulting alkaline solution was treated with 1 M hydrochloric acid until it became slightly acidic (pH 5). The precipitated solid product was filtered, washed with water, dried, and finally recrystallized from a mixture of diethyl ether and hexane. The yield was 78% (2.4 g) ¹H NMR (CDCl₃, 400 MHz; a mixture of diketo and keto-enol forms): (ppm) = 1.21–1.33 (t + t + t, 6H, CH2CH3), 4.15–4.28 (q + q, 4H, CH₂CH₃), 4.14 (s, CH2 of the diketo form), 6.36 (s, CH of the keto-enol form), 7.35 (t, p-CH of the diketo form), 8.03 (t, p-CH of the keto-enol form), 8.08 (d, m-CH of the diketo form), 8.24 (d, m-CH of the keto-enol form), and 12.40 (s, OH of the keto-enol form), lit. [41].

Synthesis of 2,6-dibromo-4-methylaniline. p-Toluidine (8 g, 74.66 mmol) was dissolved in 100 mL of glacial acetic acid. A solution of liquid bromine (7.7 mL, 149.32 mmol) in 25 mL of acetic acid was added dropwise to the resulting toluidine solution at 0–5 °C. The mixture was stirred for 1 h at room temperature, and then 100 mL of water was added. The precipitate was filtered and washed with 1 M aqueous solution of sodium hydroxide. The wet product was recrystallized from aqueous ethanol to give pale red crystals. The yield was 16.22 g (82%) ¹H NMR (CDCl₃, 400 MHz): δ(ppm) = 2.21 (s, 3H, CH₃), 4.38 (br. s., 2H, NH₂), 7.20 (s., 2H, m-Ph-H), lit. [43].

General procedure for synthesis of 4-substituted-2,6-dibromoarylhydrazine. A suspension of an appropriate aniline (30.0 mmol) in 25 mL of HCl (38%, aqueous) was cooled to −10 °C. A solution of sodium nitrite (2.17 g, 31.5 mmol) in 5 mL of water was then added dropwise while keeping the temperature below −5 °C. After stirring the reaction mixture for 1 h, SnCl₂ (75 mmol) in 30 mL of concentrated HCl was added dropwise at −5 °C. The resulting suspension was heated to r.t. and then stirred at this temperature for 1 h. A yellowish precipitate was filtered, washed with 100 mL of methylene chloride and HCl, and dried under vacuum. The obtained product was sensitive to light, so it was stored as a hydrochloride salt in a dark vial and converted into a freebase form with a 1 M aqueous solution of sodium hydroxide right before use.

2,4,6-tribromophenylhydrazine [44]. Yield: 3.30 g (61%). ¹H NMR (DMSO-d6, 400 MHz): δ(ppm) = 4.33 (s, 2H), 5.78 (s, 1H), 7.74 (s, 2H).

2,6-dibromophenylhydrazine [45]. Yield: 4.38 g (55%). ¹H NMR (DMSO-d6, 300 MHz): δ(ppm) = 7.6 (d, ³J_H,H = 8.1 Hz, 2H, m-Ph-H), 6.92 (t, ³J_H,H = 8.1 Hz, 1H, p-Ph-H), 4.91 (br.s, 3H, NH).

(2,6-dibromo-4-methylphenyl)hydrazine. Yield: 5.79 g (69%). ¹H NMR (CDCl₃, 300 MHz): δ(ppm) = 7.25 (s, 2H, m-Ph-H), 5.34 (br.s, 1H, NH), 3.88 (br.s, 2H, NH₂), 2.21 (s, 3H, CH₃).

General procedure for synthesis of ligands L^H, L^Br and L^Me (Scheme 2). A mixture of diethyl 3,3′-(pyridine-2,6-diyl)bis(3-oxopropanoate) (0.307 g, 1 mmol) and an appropriate phenylhydrazine (2.3 mmol) was dissolved in 10 mL of acetic acid to produce an orange solution. After its heating under reflux for 8 h, a light-yellow precipitate was formed, which was then filtered, washed with acetic acid and water, and dried under vacuum. The product was used without further purification.

L^H.Yield: 0.576 g (81%). ¹H NMR (DMSO-d6, 400 MHz): δ(ppm) = 6.19 (s, 2H, Pz-CH), 7.43 (t, ³J_H,H = 8,1 Hz, 2H, p-Ph-H), 7.81 (br. s, 3H, Py), 7.87 (d, ³J_H,H = 8,1, 4H, m-Ph-H). ¹³C NMR (DMSO-d6, 101 MHz): δ(ppm) = 84.93 (s, 4-Pz), 118.50 (s, 3-Py), 124.9 (s, 4-Ph), 132.91 (s,1-Ph), 133.08 (s, 2-Ph), 136.74 (s, 3-Ph), 137.75 (s, 4-Py), 151.77 (s, 3-Pz), 151.70 (s, 2-Py), 154.61 (s, 5-Pz). Calc. for (C23H13Br4N5O2): C, 38.85; H, 1.84; N, 9.85; Found: C, 38.91, H, 1.92, N, 9.59.

L^Br.Yield: 0.764 g (88%). ¹H NMR (DMSO-d6, 400 MHz): δ(ppm) = 6.19 (s, 2H, Pz-CH), 7.81 (s, 4H, m-Ph-H), 8.18 (br. s, 3H, Py). ¹³C NMR (DMSO-d6, 101 MHz): δ(ppm) = 84.64 (s, 4-Pz), 118.22 (s, 3-Py), 124.15 (s, 4-Ph), 125.41 (s,1-Ph), 134.72 (s, 2-Ph), 136.0 (s, 3-Ph), 137.42 (s, 4-Py), 151.22 (s, 3-pyraz), 151.69 (s, 2-Py), 154.27 (s, 5-Pz). Calc. for (C23H11Br6N5O2): C, 31.80; H, 1.28; N, 8.06; Found: C, 39.01, H, 1.37, N, 8.29.

L^Me.Yield: 0.569 g (77 %). ¹H NMR (DMSO-d6, 400 MHz): δ(ppm) = 2.40 (s, 6H, CH₃), 6.15 (s, 2H, Pz-CH), 7.70 (s, 4H, p-Ph-H), 7.78 (br. s, 3H, Py), 11.61 (br.s, 2H, OH, D2O-exchangeable). ¹³C NMR (DMSO-d6, 101 MHz): δ(ppm) = 20.13 (s, CH₃), 84.33 (s, 4-Pz), 117.89 (s, 3-Py), 123.88 (s, 4-Ph), 132.76 (s,1-Ph), 133.79 (s, 2-Ph), 137.08 (s, 3-Ph), 143.30 (s, 4-Py), 151.47 (s, 3-Pz), 151.57 (s, 2-Py), 154.13 (s, 5-Pz). Calc. for (C25H17Br4N5O2): C, 40.63; H, 2.32; N, 9.48; Found: C, 40.55, H, 2.37, N, 9.61.

General procedure for synthesis of complexes [Fe(L)₂](ClO₄)₂, L = L^H, L^Br and L^Me. Perchlorate salts are potential explosives and must be handled with care. One should avoid producing large quantities of these materials.

In a 10 mL scintillation vial, iron perchlorate hexahydrate (0.072 g, 0.2 mmol) and an appropriate ligand (0.4 mmol) were mixed in 3 mL of THF and stirred for 1 h. As these manipulations were carried out in air and the iron(II) perchlorate hexahydrate contained an excess of water, the synthesis of the complexes was not a stoichiometric reaction so that a small amount of the ligands remained in the reaction mixture. To obtain them as high-purity samples for further magnetic measurements, they were purified by keeping the reaction mixture at −78 °C to produce crystalline products that were filtered and dried under high vacuum; the ligands can be recovered by evaporation of the remaining solution and by washing the residue with acetonitrile. No side reactions occurred in the reaction mixture as identified by NMR spectroscopy. The obtained complexes are stable on air for approximately 24 h. For longer storage, they should be kept in nitrogen atmosphere to prevent oxidation.

[Fe(L^H)₂](ClO₄)₂. Yield: 221 mg (66%). ¹H NMR (CD₃CN, 400 MHz): δ(ppm) = 9.68 (br.s, 8H, m-Ph-H), 11.26 (br.s, 4H, p-Ph-H), 17.20 (br.s, 4H, OH), 21.52 (br.s, 2H, p-Py-H), 45.92 (br.s, 4H, Pz-CH), 61.17 (br.s, 4H, m-Py-H). Calc. for (C₄₆H₂₆Br₈Cl₂FeN₁₀O₁₂): C, 32.95, H, 1.56, N, 8.35; Found: C, 33.22, H, 1.71, N, 8.64.

[Fe(L^Br)₂](ClO₄)₂. Yield: 198 mg (58%). ¹H NMR (CD₃CN, 400 MHz): δ(ppm) = 5.58 (br.s, 12H, CH₃), 9.37 (br.s, 8H, m-Ph-H), 17.44 (br.s, 4H, OH), 21.60 (br.s, 2H, p-Py-H), 44.44 (br.s, 4H, Pz-CH), 59,10 (br.s, 4H, m-Py-H). Calc. for (C₅₀H₃₄Br₈Cl₂FeN₁₀O₁₂): C, 34.66, H, 1.98, N, 8.08; Found: C, 34.87, H, 2.14, N, 8.35.

[Fe(L^Me)₂](ClO₄)₂. Yield: 179 mg (45%). ¹H NMR (CD₃CN, 400 MHz): δ(ppm) = 9.95 (br.s, 8H, m-Ph-H), 16.98 (br.s, 4H, OH), 22.00 (br.s, 2H, p-Py-H), 45.09 (br.s, 4H, Pz-CH), 60.01 (br.s, 4H, m-Py-H). Anal. Calc. for (C₄₆H₂₂Br₁₂Cl₂FeN₁₀O₁₂): C, 27.73, H, 1.11, N, 7.03; Found: C, 27.92, H, 1.29, N, 7.24.

X-ray crystallography. Single crystals of [Fe(L^H)₂](ClO₄)₂ and [Fe(L^Br)₂](ClO₄)₂ were grown from concentrated THF solutions kept in a Dewar flask with dry ice for 2 days, those of [Fe(L^Me)₂](ClO₄)₂, from a mixture of acetonitrile and diethyl ether on air. X-ray diffraction data were collected at 120 K with a Bruker APEX2 DUO CCD diffractometer, using the graphite monochromated Mo-Kα radiation (λ = 0.71073 Å) for [Fe(L^H)₂](ClO₄)₂ and [Fe(L^Br)₂](ClO₄)₂, and Cu-Kα radiation (λ = 1.54178 Å) for [Fe(L^Me)₂](ClO₄)₂. Using Olex2 [46], the structures were solved with the ShelXT structure solution program [47] using intrinsic phasing and refined using least-squares minimization. Hydrogen atoms of OH groups were located in difference Fourier synthesis. Positions of other hydrogen atoms were calculated, and they all were refined in the isotropic approximation in the riding model. Crystal data and structure refinement parameters for these complexes are given in

Table 1. Crystal data and structure refinement parameters for [Fe(L^H)₂](ClO₄)₂, [Fe(L^Br)₂](BF₄)₂ and [Fe(L^Me)₂](ClO₄)₂.

Parameter	[Fe(LH)2](ClO4)2	[Fe(LBr)2](ClO4)2	[Fe(LMe)2](ClO4)2
Formula unit	2C46H26Br8FeN10O4, 4ClO4, 11C4H8O	2C46H22Br12FeN10O4, 4ClO4, 11C4H8O, 4H2O	C50H34Br8FeN10O4, 4ClO4, 2C4H10O, 2CH3CN
Formula weight	4146.70	4850.01	1963.25
T, K	120	120	120
Crystal system	Monoclinic	Monoclinic	Monoclinic
Space group	C2/c	P21/c	P21/c
Z	2	2	4
a, Å	34.797(3)	15.914(2)	15.4750(3)
b, Å	12.9466(10)	17.779(2)	18.6741(4)
c, Å	23.6896(18)	30.696(4)	24.7555(5)
β, °	131.9720(10)	102.245(2)	90.3160(10)
V, Å3	7934.5(11)	8487.4(18)	7153.8(3)
Dcalc (g cm−1)	1.736	1.898	1.823
Linear absorption, μ (cm−1)	43.57	59.62	82.10
F(000)	4108	4736	3872
2Θmax, °	58	54	135
Reflections measured	89,248	89,299	100,303
Independent reflections	10,553	18,529	12,778
Observed reflections [I > 2σ(I)]	8445	8377	11,985
Parameters	504	1026	952
R1	0.0749	0.1152	0.0437
wR2	0.1667	0.3278	0.1086
GOF	1.044	1.083	1.055
Δρmax/Δρmin (e Å−3)	1.760/−1.551	1.822/−1.055	0.868/−0.673

. CCDC 1959676, 1959677, and 1959678 contain the supplementary crystallographic data for [Fe(L^H)₂](ClO₄)₂, [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂, respectively.

Magnetic measurements. Magnetic susceptibility of the iron(II) complexes was measured with a Quantum Design MPMS-XL-7 SQUID magnetometer under the dc magnetic field of 0.1 T in the temperature range 2–300 K. Finely ground microcrystalline powders were immobilized in eicosane matrix inside a polycarbonate capsule. The data obtained were corrected for the eicosane, the capsule, the sample holder, and the diamagnetic contribution. The latter was done by using the Pascal constants [48]. Smoothing was performed by the Savitzky–Golay method with five points of window and polynomial order of two. The observed variable-temperature dc magnetic susceptibility data for [Fe(L^H)]₂(ClO₄)₂ were fitted in PHI software [49] using the spin Hamiltonian:

$$\widehat{H}=D\left({\widehat{S}}_z^2-\frac{{\widehat{S}}^2}{3}\right)+\beta H\cdot g_{iso}·\widehat{S}$$

NMR spectroscopy. ¹H and ¹³C NMR spectra were recorded from the solutions of the iron(II) complexes in CD₃CN and CD₃OD with Bruker Avance 300 and 600 FT-spectrometers (300.15 and 600.22 MHz ¹H frequency), respectively. The measurements were conducted using the residual signals of CD₃CN (¹H 1.94 ppm, ¹³C 118.26 ppm) or CD₃OD (¹H 3.31, 4.87 ppm, ¹³C 49.00 ppm).

Evans method. Magnetic susceptibility of the iron(II) complexes in acetonitrile-d₃ was evaluated by the Evans method [50,51] in the temperature range 235–325 K. For [Fe(L^H)₂](ClO₄)₂, the data were also collected in methanol-d₄ in the temperature range 190–330 K. In all cases, a Wilmad NMR tube with a coaxial insert was used. The inner (reference) tube was filled with acetonitrile-d₃ with approximately 1% of Me₄Si, and the outer tube, with the solution of the complex (~1–5 mg/cm³) in the same solvent with the same concentration of Me₄Si. The sample was kept at each temperature for 15 min before collecting the spectra to reach the thermal equilibrium. Molar magnetic susceptibility was calculated from the difference between the chemical shift of Me₄Si in pure acetonitrile-d₃ and its shift in a solution of the complex (Δδ in Hz) in this solvent using the following equation:

$${\chi }_M=\frac{\mathsf{\Delta }\delta M}{{\nu }_0{S}_{f}c}-{\chi }_M^{dia}$$

where M—molar weight of the complex, g/mol; ν₀—frequency of the spectrometer, Hz; S_f—shape factor of the magnet (4π/3); c—concentration of the complex, g/cm³; and χ_M^dia—molar diamagnetic contribution to the paramagnetic susceptibility calculated using the Pascal’s constant [48]. The concentration c was recalculated for each temperature in concert with the change in the solvent’s density ρ: c_T = m_sρ / m_sol, where m_s is the mass of the complex and m_sol is the mass of the solution.

Temperature-dependence of chemical shifts. Chemical shifts in ¹H NMR spectra of the iron(II) complexes in acetonitrile-d₃ were analyzed in the temperature range 235–345 K. For a compound that may exist in two spin states, the observed chemical shift of a given nucleus in the ¹H NMR spectrum is a weighted average of those for LS and HS species (${\eta }_{LS}$ and ${\eta }_{HS}$ are their populations):

$${\delta }_{obs}={\eta }_{LS}{\delta }_{LS}+{\eta }_{HS}{\delta }_{HS}$$

For the iron(II) complexes with the diamagnetic LS state, this chemical shift (in ppm) can be approximated as:

$${\delta }_{obs}={\eta }_{LS}{\delta }_{dia}^{LS}+{\eta }_{HS}\left({\delta }_{dia}^{HS}+{\delta }_{par}^{HS}\right)\approx {\delta }_{dia}+{\eta }_{HS}{\delta }_{par}^{HS}$$

Diamagnetic contribution ${\delta }_{dia}$ to the observed chemical shift, which is virtually the same for the LS and HS states, was taken as a chemical shift of the same nucleus in the NMR spectrum of a free ligand [52]. Therefore, paramagnetic contribution ${\delta }_{par}^{HS}$ was measured as a difference between the chemical shifts of the iron(II) complexes and those of the corresponding ligands. For systems in a pure spin state with low magnetic anisotropy [53,54] at high T (>200 K), the paramagnetic chemical shifts that are different for each nucleus in a complex follow the linear dependence on the inverse temperature: ${\delta }_{par}^{HS}=A+B{T}^{-1}$. As small paramagnetic shifts are measured less accurately, the hydrogen atoms that featured the largest chemical shifts (those of the pyrazolyl and meta-pyridine moiety; Scheme 1) were used in the fit by a regular solution model [53] to produce average thermodynamic parameters for the SCO-active iron(II) complexes.

The population of the HS state for these complexes was calculated based on the average deviations of the paramagnetic shifts of these hydrogen atoms from the values observed for the HS complex [Fe(L^H)₂](ClO₄)₂ at each temperature:

$${\eta }_{HS}=\frac{{\delta }_{par}}{{\delta }_{par}^F}$$

Quantum chemistry. Geometry optimization of the ligands L^R and model arylpyrazole pyridines was performed using the ORCA package, v. 4.2 [55,56], with a TPSSh functional [57,58,59] and a basis set def2-TZVP [60]. For the iron(II) complexes [Fe(L^R)₂]²⁺, X-ray diffraction geometries were chosen as a starting point for geometry optimization with the TPSSh functional, which provides a satisfactory estimate for the energy difference between the HS and LS states in (pseudo)octahedral iron(II) complexes [61], and a basis set def2-TZVP [60]. To speed up the calculations, the RIJCOSX approximation [62] with a def2/J fitting basis [63] set was used. Extra tight thresholds for forces and displacements were used.

3. Results and Discussion

The corresponding N,N′-disubstituted 3-bpp ligands L^R were obtained with high yields from 2,6-bis(1,3-dioxo-3-ethoxypropyl)pyridine [41] and 2,6-dibromo-4-R-phenylhydrazines (R = H, Br, Me) through a previously reported [27] one-step cyclization process. They were filtered from the reaction mixture after precipitation due to low solubility in glacial acetic acid and were used without further purification. The subsequent reaction with Fe(ClO₄)₂•6H₂O in THF produced the iron(II) complexes [Fe(L^R)₂](ClO₄)₂ as fine-crystalline powders. Their yellow color hinted on the HS state of the iron(II) ion [1], which was confirmed by variable-temperature magnetic measurements between 2 and 300 K (Figure 1). Note that perchlorate salts are potential explosives and have to be handled with care. At room temperature, the observed χT values for [Fe(L^R)₂](ClO₄)₂ are 3.04–3.51 cm³ K mol⁻¹ and close to the expected one for one iron(II) ion in the pure HS state with S = 2 (3.00 cm³ K mol⁻¹) and to those reported for HS iron(II) complexes of 3-bpp [27,37,64]. The χT values remain nearly constant on cooling the samples down to 30–50 K, with a drop to 1.40–1.74 cm³ K mol⁻¹ below these temperatures as a result of zero-field splitting effects, intermolecular antiferromagnetic interactions, and/or the saturation of magnetization rather than a spin-state switching [65]. Satisfactorily simulation of the data was obtained with isotropic values for the g-tensor g_iso and zero-field splitting energies ∣D∣ equal to 2.00–2.13 and 5.64–8.2 cm⁻¹, respectively; those are quite expected for N,N′-disubstituted 3-bpp [27,37].

Although typical of N,N′-disubstituted 3-bpp ligands [27], the observed ‘locking’ [37] of the complexes [Fe(L^R)₂](ClO₄)₂ in the HS state is, however, in a stark contrast with their behavior in solutions as probed by variable-temperature NMR spectroscopy. Two separate techniques were used for this purpose: the popular [12] Evans technique [50,51] and the less popular but more powerful [40] analysis of a temperature dependence of chemical shifts in the NMR spectra [40,66] that are collected with the same experimental setup (see experimental section). The former allows for measuring the magnetic susceptibility of a solution as an alternative to a more cumbersome magnetometry [12], while the latter allows for screening of paramagnetic complexes that may exist in mixtures of unknown concentrations [40]. The two approaches rooted in the widely available NMR spectroscopy are helpful tools in molecular design of SCO compounds [12,40] exploiting the spin state behaviour decoupled from crystal packing [23,24] or substrate [25,26] effects that sometimes block a SCO [23,24].

For [Fe(L^R)₂](ClO₄)₂, both methods show the complexes with the para-substituted 2,6-dibromophenyl groups to undergo a temperature-induced SCO between 235 and 325 K in acetonitrile solutions, while [Fe(L^H)₂](ClO₄)₂ remains HS even in methanol, which provides access to lower temperatures down to 190 K. The latter follows from the χT values being constant at ~3.3 cm³ K mol⁻¹ (Figure 2), which is close to the theoretical value for the HS iron(II) ion (3.00 cm³ K mol⁻¹), and from the linear dependence of the chemical shifts on the inverse temperature (Figure 3, left), which is a typical Curie behavior of a paramagnetic complex in one spin state [53,54]. For [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂, however, the χT values decrease from 3.03 and 3.54 cm³ K mol⁻¹ at room temperature to 2.40 and 2.99 cm³ K mol⁻¹ at 235 K (Figure 2), thus indicating a gradual depopulation of the paramagnetic HS state and the population of the diamagnetic LS state upon cooling. This is further supported by the chemical shifts in the corresponding NMR spectra that feature a significant deviation from the above linear dependence at temperatures below 285 K (Figure 3, right).

Fitting these sets of NMR-derived data separately (Figure 2 and Figure 3) by a regular solution model [53] resulted in midpoint temperatures T_1/2 of this SCO equal to 195–198 and 205 K for [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂, respectively, and the thermodynamic parameters (

Table 2. SCO parameters ¹ from the NMR data for [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂ in acetonitrile-d₃.

Parameter	[Fe(LBr)2](ClO4)2	[Fe(LMe)2](ClO4)2
T1/2, K	198	205
∆H, kJ/mol	18.7	21.1
∆S, J/mol	94.3	103.8

¹ Thermodynamic parameters are obtained by fitting the deviations of the chemical shifts in the NMR spectra of [Fe(L^Br)₂](ClO₄)₂ (R = Br, Me) from the same values for [Fe(L^H)₂](ClO₄)₂ by a regular solution model. For those obtained by fitting the Evans data, see Table S1 of ESI.

) typical for such measurements [16,67]. The similar midpoint temperatures of the two complexes, which agree nicely between the two different approaches, mirror similar steric effects of the para-substituents in the 2,6-dibromophenyl groups, as gauged by the volumes [68] of the bromine atom and the methyl group (34.3 and 32.2 ų), with the larger bromine atom predictably [12] favoring the HS state (Figure 4). In contrast, their opposite electronic effects cannot explain the SCO-assisting tendency (or the lack thereof) in [Fe(L^R)₂](ClO₄)₂.

For SCO-active iron(II) complexes of other N,N′-disubstituted 3-bpp ligands with the ortho-substituents in the N-phenyl groups (Scheme 1) that are generally smaller than the bromine atom in [Fe(L^R)₂](ClO₄)₂ [27], the midpoint temperatures T_1/2 in acetonitrile are between 240 and 270 K [27], the smaller substituent favoring the LS state. The smallest fluorine and hydrogen atoms, however, counterintuitively [27] produced HS complexes only, which was attributed [27,69] to the energy penalty associated with the loss of planarity between the 2,6-substituted N-phenyl group and the pyrazolyl moiety that are almost perpendicular to each other in the LS state.

To shed light on the appropriate geometry features that may be behind the SCO-activity of [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂, we resorted to X-ray diffraction data (Figure 5) for their single crystals collected from THF solutions or mixtures of acetonitrile and diethyl ether as solvates [Fe(L^H)₂](ClO₄)₂•5.5THF, [Fe(L^Br)₂](ClO₄)₂•5.5THF•2H₂O and [Fe(L^Me)₂](ClO₄)₂•2Et₂O•2CH₃CN with the lattice solvents ‘trapped’ by hydrogen bonds with the hydroxyl groups of the 3-bpp ligands L^R (O…O 2.534(12)–2.663(5) Å, OHO 153.8(9)–177.8(12)°). At the low temperature (120 K) of the X-ray diffraction experiments (

Table 3. Main geometric parameters and continuous symmetry measures ¹ as obtained from X-ray diffraction at 120 K for HS [Fe(L^H)₂](ClO₄)₂, HS [Fe(L^Br)₂](BF₄)₂ and HS [Fe(L^Me)₂](ClO₄)₂.

Parameter	[Fe(LH)2](ClO4)2	[Fe(LBr)2](ClO4)2	[Fe(LMe)2](ClO4)2
Fe-NPy, Å	2.098(5)	2.071(13)–2.123(12)	2.105(5)–2.109(5)
Fe-NPz, Å	2.242(5), 2.262(5)	2.208(12)–2.277(13)	2.225(5)–2.266(5)
Npz…Npz, Å	4.340(9)	4.303(18), 4.38(2)	4.311(6), 4.417(7)
NpzNpyNpz, °	110.6(3)	109.4(6), 112.5(6)	110.5(2), 111.9(2)
θ,°	82.43(10)	81.64(13)	81.58(10)
φ,°	166.5(3)	167.0(5)	166.31(18)
γ,°	82.9(3), 83.4(4) av. 83.1	82.8(6), 82.2(7), 78.0(7), 81.8(7) av. 81.2	79.8(2), 81.2(2), 77.1(2), 77.2(2) av. 78.8
β,°	15.0(3), 2.3(3)	11.8(6), 2.4(7), 12.8(6), 3.1(6)	14.7(2), 3.2(2), 8.5(2), 4.3(2)
S(Oh)	5.473	5.534	5.840
S(ebcT)	10.379	10.264	10.192
Br…Br, Å	3.615(2), 3.690(4), 3.904(19), 3.7094(13)	3.808(3), 3.579(3), 3.804(3), 3.630(3)	3.5644(9), 3.8028(10), 3.6351(11), 3.7500(11)

¹ N_pz and N_py are the coordinating nitrogen atoms from the pyrazol-3-yl and pyridine moieties; θ is the ‘twist’ angle between the least-squares planes of the two 3-bpp ligands; φ is the ‘rotatioN′ angle N_Py-M-N_Py; γ is the rotation angle of the N-phenyl group relative to the pyrazol-3-yl plane; β is the rotation angle of the pyrazol-3-yl moiety relative to the pyridine plane; S(O_h) and S(ebcT) are octahedral and edge-bicapped tetrahedral symmetry measures, respectively; Br…Br are the distances between the bromine atoms from possible intramolecular Br…Br interactions.

), the Fe-N bond lengths are typical for HS (pseudo)octahedral complexes of iron(II) in a N6-coordination environment [23] distorted towards an edge-bicapped tetrahedron [70], as judged by ‘twist’ θ and ‘rotatioN′ φ angles and continuous symmetry measures [70] relative to the two reference shapes, an ideal octahedron and an edge-bicapped tetrahedron. Of the three, Fe(L^H)₂](ClO₄)₂ matches the former better than the two other complexes, while [Fe(L^Me)₂](ClO₄)₂, matches the latter better than the two other complexes. In all cases, however, the observed differences are not conclusive enough to rationalize the different spin-state behavior of [Fe(L^R)₂](ClO₄)₂ in solutions.

This behavior cannot be also explained by the rotation of the 2,6-dibromophenyl group from the pyrazolyl plane (Figure 5) to adapt to bulky ortho-substituents proposed [27] as the reason for the SCO-assisting tendency of the N,N′-disubstituted 3-bpp ligands. Indeed, the rotation angles in [Fe(L^R)₂](ClO₄)₂ are between 77.1(2) and 83.4(4)° (Table 3), with the largest values unexpectedly adopted by the fully HS complex [Fe(L^H)]₂(ClO₄)₂, and the smallest by the SCO-active complex [Fe(L^Me)]₂(ClO₄)₂. Those fall nicely between the corresponding ranges for the earlier reported LS (67.6(5)–75.4(5)°) and HS (83.85(18)–89.6(2)°) iron(II) complexes of 3-bpp with ortho-substituted N-phenyl groups [27]. The rotation results in a severe distortion of the bis(pyrazol-3-yl)pyridine cores. The angles between the pyrazolyl and pyridine planes vary in a wide range of 2.3(3)–15.0(3)°, with the ligands in [Fe(L^Br)₂](ClO₄)₂ being slightly flatter in terms of an average and the largest value of this angle (Table 3). It also enables the 2,6-dibromophenyl groups to form intramolecular ‘parallel-displaced’ [71] stacking interactions with the pyridine moieties of the second 3-bpp ligand, which may [37] (or may not [17]) contribute to highly distorted molecular geometries of [Fe(L^R)₂](ClO₄)₂ often associated with SCO-inactive complexes [23,72]. As a result, there is no ‘terpyridine embrace’ packing motif [73] in the crystals of the solvates [Fe(L^H)₂](ClO₄)₂•5.5THF, [Fe(L^Br)₂](ClO₄)₂•5.5THF•2H₂O and [Fe(L^Me)₂](ClO₄)₂•2Et₂O•2CH₃CN (Figure 6) that is behind a cooperative SCO in some SCO-active metal complexes [74]. The geometrical parameters of these intramolecular stacking interactions, i.e., angles between the planes (1.0(4)–5.4(2)°), centroid-centroid and shift distances (3.577(5)–3.730(5) and 1.008(6)–1.214(5) Å), again reveal no correlation with the SCO-activity of [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂, as they are generally smaller in the former and are larger in the latter.

Other intramolecular interactions identified in the above solvates by the X-ray diffraction are similarly inconclusive, including those between the bromine ortho-substituents in the N-phenyl groups (Table 3) tentatively assigned to halogen bonds [75]. Based on interatomic distances Br…Br that should be smaller and angles C-Br…Br that should be closer to 180° for stronger halogen bonds, they seem to be slightly weaker in the less distorted complex [Fe(L^H)₂](ClO₄)₂; the appropriate average values are 3.74 Å and 104.0º as compared to 3.56–3.69 Å and 110.7–111.1° in the two others. The bromine atoms also form Br…pi interactions with the pi-systems of the pyrazolyl moieties, although the distances to these planes are almost the same at 3.3–3.4 Å in [Fe(L^H)₂](ClO₄)₂•5.5THF, [Fe(L^Br)₂](ClO₄)₂•5.5THF•2H₂O and [Fe(L^Me)₂](ClO₄)₂•2Et₂O•2CH₃CN.

As the above geometrical parameters of [Fe(L^R)₂](ClO₄)₂ were measured in a crystalline state, any correlations between them (or other parameters from Table 3) and the ability of the complexes [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂ to undergo a SCO in solutions may be hidden by crystal packing effects [23,24]. From the Hirshfeld surface analysis [76,77] that maps different types of intermolecular interactions on a ‘molecular’ surface of the cation [Fe(L^R)₂]²⁺ and from 2D fingerprint plots [78] that quantify their occurrence (Figure 7), the major contribution in the solvates [Fe(L^H)₂](ClO₄)₂•5.5THF and [Fe(L^Me)₂](ClO₄)₂•2Et₂O•2CH₃CN (39.4% and 36.4%, respectively) comes from numerous H…H contacts. In [Fe(L^Br)₂](ClO₄)₂, however, Br…H contacts contribute the most (36.8% vs. 16.2 and 21.1% in the above solvates) while the H…H contacts account for 22.6%. Similar values are found for O…H interactions, the strongest of them being hydrogen bonds between the hydroxyl groups of the 3-bpp ligands and the perchlorate anions or solvent molecules that appear as bright red areas on the Hirshfeld surfaces and as most intense areas on the 2D fingerprint plots. Combined with the O…H contacts produced by hydrogen atoms of 2,6-dibromophenyl, pyrazolyl or pyridine groups, they result in significant and rather similar contributions (23.7, 20.0 and 22.5%) to the overall crystal packing in [Fe(L^H)₂](ClO₄)₂•5.5THF, [Fe(L^Br)₂](ClO₄)₂•5.5THF•2H₂O and [Fe(L^Me)₂](ClO₄)₂•2Et₂O•2CH₃CN. For comparison, C…H contacts account for 9.3, 8.8 and 9.7% only.

To decouple the inherent structural features of [Fe(L^R)₂](ClO₄)₂ from these intermolecular interactions in [Fe(L^H)₂](ClO₄)₂•5.5THF, [Fe(L^Br)₂](ClO₄)₂•5.5THF•2H₂O, and [Fe(L^Me)₂](ClO₄)₂•2Et₂O•2CH₃CN, we performed quantum chemical calculations of the isolated cations [Fe(L^R)₂]²⁺, as crystal packing effects often prohibit a SCO [23,24] by, for example, inducing an additional distortion to the coordination geometry [23,24]. Indeed, a SCO does not generally [80] occur [81] in iron(II) complexes of 1-bpp [72] with the ‘rotatioN′ angle φ < 172° and/or the ‘twist’ angle θ < 76°, which is the case of [Fe(L^R)₂](ClO₄)₂ (Table 3). In the absence of X-ray diffraction data at higher temperatures due to low crystal quality, no conventional (or inverse [82]) SCO in any of the obtained solvates follows from their yellow color common for HS iron(II) complexes [1].

In the HS state, the isolated cations [Fe(L^R)₂]²⁺ feature less distorted and more consistent coordination geometries than those measured experimentally (Table 3 and Table S2), with weaker intramolecular interactions resulted from an expected overestimation of the Fe-N bond lengths. Indeed, the appropriate distances for the above stacking interactions, halogen bonds and Br…pi contacts increase by 0.2–0.5 Å while remaining almost the same between the cations. Upon switching to the LS state, the Fe-N bonds in [Fe(L^R)₂]²⁺ shorten to values typical [1] for LS complexes of iron(II) in an octahedron-like coordination environment with a minor distortion towards an edge-bicapped tetrahedron induced by the rigid 3-bpp ligands [24]. For [Fe(L^H)₂]²⁺, however, this distortion is more pronounced, as highlighted by continuous symmetry measures [70] and the ‘rotatioN′ φ and ‘twist’ θ angles. The latter parameter even becomes smaller in the LS state (Table S2). Its bis(pyrazol-3-yl)pyridine cores also remain significantly distorted, with an average angle between the heterocyclic planes of 8.5° as compared to 5.2 and 4.5° in [Fe(L^Br)₂]²⁺ and [Fe(L^Me)₂]²⁺, respectively.

On the other hand, the rotation angle of the 2,6-dibromophenyl group from the pyrazolyl plane is nearly the same, with the values larger by 6° than in the HS state, mirroring a more ‘perpendicular’ molecular geometry of LS iron(II) complexes with N,N′-disubstituted 1-bpp, 3-bpp, or similar ligands [27] favored by intramolecular ‘parallel-displaced’ [71] stacking interactions. The latter, however, does not straightforwardly follow from the appropriate angles between the two planes, the centroid-centroid, and shift distances that go from average values of 9.1–9.4°, 3.977–3.986, and 1.174–1.244 Å, respectively, to 10.0–13.6°, 3.837–3.893, and 1.013–1.518 Å, respectively, upon switching to the LS state. The cation [Fe(L^H)₂]²⁺ again stands out, as it features a decrease in the shift distances in contrast to the other two cations. A similar behavior is observed for the other two types of intramolecular interactions, the halogen bonds Br…Br and the interactions Br…pi. In [Fe(L^H)₂]²⁺, the average distance between the bromine ortho-substituents and to the pyrazolyl planes decreases by ~0.1 and 0.2 Å, respectively. In the cations [Fe(L^Br)₂]²⁺ and [Fe(L^Me)₂]²⁺, however, the former decreases by ~0.4 Å while the latter remains virtually the same at 3.6 Å.

Although none of the above geometric parameters clearly correlates with the SCO-assisting tendency (or the lack thereof) of the complexes [Fe(L^R)₂](ClO₄)₂, an overall more distorted [23,24] molecular geometry of the LS cation [Fe(L^H)₂]²⁺ that distinguishes it from the two others may be the reason for its LS state not being accessed in the entire temperature range of 190–325 K probed in solutions of [Fe(L^H)₂](ClO₄)₂. Indeed, the calculated energy difference between the two spin states of [Fe(L^H)₂]²⁺ strongly favors the HS state, which is preferred by 7.7 kcal/mol. The same values of 4.2 and 1.2 kcal/mol for the cations [Fe(L^Br)₂]²⁺ and [Fe(L^Me)₂]²⁺ agree nicely with the complexes [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂ undergoing a SCO with the midpoint temperatures of 195–198 and 205 K, indicative of a HS state stabilization [12] by the larger bromine atom.

To trace the observed differences in the solution behaviour of the complexes [Fe(L^R)₂](ClO₄)₂ to the ligand design, the electronic and steric effects of the para-substituents in the N-phenyl groups were probed by charges of coordinating nitrogen atoms [15] and distances between them [13] in the free ligands L^R and by energy profiles to the rotation of the N-phenyl groups [27]. The calculated charges [15] are virtually the same in the three ligands (

Table 4. Selected geometric parameters ¹ and NBO charges of coordinating nitrogen atoms in the ligands L^R as obtained from TPSSh/def2-TZVP calculations.

Parameter	LH	LBr	LMe
Fe-NPy, Å	4.794	5.085	5.002
Fe-NPz, Å	2.795/2.771	2.811/2.902	2.763/2.893
Npz…Npz, Å	118.9	125.7	124.4
Npy…Npz, Å	−0.22	−0.22	−0.22
NpzNpyNpz, °	−0.37	−0.38	−0.37
q(NPz), e	81.1/88.9	84.7/75.7	80.8/81.2
q(NPy), e	4.5/19.2	15.5/36.0	12.2/39.8
γ,°	4.794	5.085	5.002
β,°	2.795/2.771	2.811/2.902	2.763/2.893

¹ γ is the rotation angle of the N-phenyl group relative to the pyrazol-3-yl plane; β is the rotation angle of the pyrazol-3-yl moiety relative to the pyridine plane.

) but their geometries are not. While the distances between the pyrazolyl and pyridine nitrogen atoms in L^Br and L^Me fall into the ranges for SCO-active iron(II) complexes, the ligand L^H features smaller values associated with the LS state [13], as opposed to the entirely HS complex [Fe(L^H)₂](ClO₄)₂. Large separations between the pyrazolyl nitrogen atoms result from the rotation of one of the pyrazol-3-yl moieties relative to the pyridine plane, which is less pronounced in L^H. Note, however, a typical configuration of the free 3-bpp ligands from the available X-rays diffraction data [41,83,84,85], with these nitrogen atoms pointing towards the para-position of the pyridine moiety which undermines the validity of the distance-based correlation proposed for the bidentate ligands [13].

As should be common for a distal substituent, the energy profiles for the rotation of the para-substituted 2,6-dibromophenyl groups are identical for the three model compounds (Figure 8), with a monotonical increase up to ca. 21.0 kcal/mol at 0° expected [27] for the large bromine atoms in the ortho-positions [27]. For the para-substituents, however, the size cannot explain the SCO-assisting tendency or the lack thereof by the energy penalty to rotation (although they have the same effect as the bulky ortho-substituents in the N-phenyl groups of the 3-bpp ligands on the spin-state behavior of their iron(II) complexes [27]).

4. Conclusions

A combined study of new iron(II) complexes [Fe(L^R)₂](ClO₄)₂ with 3-bpp ligands N,N′-disubstituted by 4-functionalized 2,6-dibromophenyl groups allowed for probing the effect of a distal para-substituent on the spin-state behavior of these complexes and determining the reasons behind it using X-ray diffraction, variable-temperature magnetic measurements, and NMR spectroscopy (the latter being the method of choice for ‘truly molecular’ design [12] of SCO compounds). An absence of a thermally-induced SCO in the case of the unsubstituted 2,6-dibromophenyl group in L^H agrees with an accepted [12] tendency for the HS state stabilization of bulky [27] substituents close to the metal ion. It, however, cannot explain the SCO-assisting tendency of the substituents in its para-position unambiguously identified by two separate approaches in NMR spectroscopy (i.e., the popular Evans method [50,51] and the analysis of chemical shifts in variable-temperature NMR spectra [40]).

The chosen para-substituents, i.e. the bromine atom and the methyl group, have different electronic characteristics, although they hardly affect the spin state of the iron(II) ion, as expected for such distal substituents and supported by the charges of the coordinating nitrogen atoms [15] in the free ligands. In contrast to previously discarded [27,69] electronic reasons behind the LS state stabilization by mesityl-substituted 1-bpp, a possible explanation by intramolecular stacking interactions [37] tuned by the electronic effects [86] does not seem viable. Indeed, the bromine atom and the methyl group should have opposite electronic effects. However, the resulting SCO-assisting tendency of the 3-bpp ligands L^Br and L^Me is almost the same as judged by similar midpoint temperatures T_1/2 of the SCO in solutions of [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂. The latter also mirrors similar steric demands of these substituents (as gauged by their sizes), with the larger bromine atom [12] favoring the HS state in compliance with the common wisdom in SCO research [12]. However, their effect on the spin state through the rotation of an appropriately substituted 2,6-dibromophenyl group [27,69] is expectedly negligible, as follows from the calculated energy variation in the corresponding phenylpyrazole molecules.

Other possible reasons could be intramolecular dispersion interactions between the bromine atom or the methyl group and the pyridine moiety of the second ligand in the complex [Fe(L^R)₂](ClO₄)₂ favoring the LS state [87] or, as an alternative, steric clashes between them. The latter, counterintuitive finding is, however, in an agreement with a previously stated idea that the LS state of the iron(II) complex of a mesityl-substituted 1-bpp is a result of the mesityl groups avoiding steric clashes [69].

Either way, the para-substituents introduced at the 2,6-dibromophenyl groups induce a SCO, as do bulky (but not too bulky [27]) substituents in the ortho-positions of the N-phenyl groups. Although the exact reason behind their SCO-assisting tendency in the N,N′-disubstituted 3-bpp ligands is still unknown, it is probably an intricate combination of the above aspects of the ligand and the complex design and is clearly rooted in the planar shape of the N-phenyl group [27].

With an observed SCO in solution of [Fe(L^Br)₂](ClO₄)₂ and [Fe(L^Me)₂](ClO₄)₂, a series of SCO-active complexes of N,N′-disubstituted 3-bpp contains seven representatives [27,40], thus paving the way for this family of SCO compounds amenable to functionalization at other positions of the ligand and screening [40] by NMR spectroscopy to become as instructive as the most populated [12] 1-bpp family in revealing structure–function relations for the successful molecular design of SCO compounds.