Chemically and Light-Driven Coordination-Induced Spin State Switching (CISSS) of a nonheme-iron complex

2.1 Synthesis of complexes 1 and 2

To obtain the target CISSS complex 1 the general method of Scheme 5 was used. Iron(II) perchlorate hydrate and the literature-known tridentate ligand bis(2-pyridylmethyl)benzylamine (3) [39] were stirred for four days until addition of the bidentate ligand bis(1-methylimidazole)ketone (bmik, 4) [40].

Scheme 5:

Synthesis of complex 1: (a) 1. MeOH, rt, four days, 2. 4, rt, 20 h.

Adding 4 to a solution of Fe(ClO₄)₂ * x H₂O in MeOH immediately led to precipitation of the homoleptic bmik complex, which was separated from the reaction solution. The desired complex 1 could then be obtained with a yield of 51%.

A six-step synthetic route was developed to generate the azpy-trident ligand 5 (Scheme 6). We started with the literature-known synthesis of 3-nitrosobenzoic acid (6) [41]. Azopyridine carbonic acid 7 was synthesized via the Mills reaction under basic conditions. Esterification of 7 led to the azo ester 8, and subsequent reduction with lithium aluminum hydride gave the corresponding azo alcohol 9. Azo chloride 10 could be obtained by refluxing in thionyl chloride. It turned out that the azo chloride is very unstable and has to be converted further directly. For this reason, we only removed excess thionyl chloride in vacuo and then carried out an in situ Finkelstein reaction with the literature-known bis(2-pyridylmethyl)amine (11) [39]. The photoswitchable azpy-trident ligand 5 could be generated this way with moderate yields.

Scheme 6:

Synthesis of complex 2: (a) Oxone, DCM/H₂O, rt, 4.5 h; (b) 3-aminopyridine, TMAH, pyridine, 1. 120 °C, 4.5 h, 2. rt, 18 h; (c) H₂SO₄, MeOH, reflux, 18 h; (d) 1 M LAH solution in THF, 1. 0 °C, 2. rt, two days; (e) thionyl chloride, 90 °C, 20 h; (f) Cs₂CO₃, TBAI, MeCN, 1. 90 °C, 24 h, 2. rt, three days; (g) 1. iron(II) perchlorate hydrate, MeOH, rt, 20 h, 365 nm, 2. 4, rt, 20 h.

The switching properties of azo compounds 5, 8, and 9 were investigated by UV/Vis and NMR spectroscopy (Supporting Information Figures S1–S3, S8–S10).

For the synthesis of complex 2, iron(II) perchlorate hydrate was first stirred with the azpy-trident ligand 5 for 20 h. In order to promote κ⁴-coordination of 5 (i.e., by the cis isomer), the reaction flask was irradiated with a wavelength of 365 nm. The bidentate ligand bmik 4 [40] was then added in a second step. Again, formation of the homoleptic bmik complex could not be avoided, such that always a mixture of both complexes was obtained. However, while the homoleptic complex precipitated at the bottom of the flask, the desired complex 2 crystallized at the top of the flask, allowing the two complexes to be separated manually. Moreover, due to irradiation with light with 365 nm during the reaction (see above) and a longer reaction time, the yield of the desired complex could be increased up to 31%.

We could not obtain single crystals for either complex 1 or complex 2 and therefore no scXRD data are available. It should therefore be mentioned that the structures given are based on those of related compounds and/or spectroscopic/physicochemical evidence.

2.2 Investigation of complex 1

Since the coordination of an axial ligand to 1 should lead to a diamagnetic octahedral complex (cf. Scheme 3), structural and electronic changes are to be expected. As a result, corresponding electronic absorption spectra should change as well. In order to obtain information on this issue, UV/Vis spectra of 1 were measured in both dichloromethane (DCM) and pyridine, as shown in Figure 1.

Figure 1:

Electronic absorption spectra of 1 in dichloromethane (DCM) (red, c = 0.034 mM) and in pyridine (black, c = 0.034 mM); pyridine titration: stepwise addition of pyridine leads to a decrease of the absorption band at 569 nm and evolution of an absorption band at 410 nm.

The spectrum of 1 in DCM (Figure 1, red) shows an absorption band at 330 nm (22,153 L mol⁻¹ cm⁻¹) and two weak shoulders at 359 and 375 nm (cf. Figure S5a in Supporting Information). For iron(II) complexes containing pyridine-based ligands MLCT absorption bands in this range are common [42]. In addition, an absorption band at 569 nm (1392 L mol⁻¹ cm⁻¹) is observed, which is attributed to MLCT transitions typical for iron(II) bmik complexes [43]. The spectrum of 1 in pure pyridine (Figure 1, black and Figure S5a), on the other hand, shows no absorption band at 569 nm, and the absorption band at 330 nm just shifts to 322 nm (21,936 L mol⁻¹ cm⁻¹).

In order to obtain information regarding spectral changes resulting from coordination of pyridine to complex 1, pyridine titration experiments were performed. Gradual addition of pyridine up to 10,000 equivalents of 1 in DCM results in a slight intensity increase of the MLCT band at 569 nm (Figure 1, blue, and Figure S6). Further addition of pyridine up to 100,000 equivalents (Figure 1, gray), however, leads to an intensity decrease of the same band, and upon adding larger amounts of pyridine this band disappears completely. On the other hand, above 50,000 equivalents of pyridine, an additional band appears at 410 nm, reaching its maximum intensity at about 300,000 equivalents (Figure 1, gray, and Figure S7). In the case of pyridine-containing N-donor ligands, this band has been observed before. Specifically, it was found that its position and intensity depend on the number of pyridine groups [42]. Together with the disappearance of the MLCT band at 569 nm, it thus can be assumed that the bmik ligand is replaced by pyridine. Above 300,000 equivalents of pyridine, the 410 nm band decreases in intensity and completely disappears in pure pyridine, indicating formation of a polypyridine complex (Figure 1, black). Taken together, it appears that the desired six-coordinate complex 1-py (Scheme 3) forms initially upon addition of pyridine to its five-coordinate precursor 1, but at higher amounts of pyridine, ligand exchange occurs.

To get insight into the change of the spin state of 1 upon addition of pyridine, Evans NMR susceptibility measurements were performed. This method takes advantage of the fact that the shift of a signal depends on the susceptibility of the surrounding solution [44, 45]. Two tubes, one inside the other, are used, both containing solvent and a probe, here tetramethylsilane. One tube also contains the paramagnetic sample, which induces a downfield shift of the TMS probe signal. Two TMS signals are then obtained due to the use of two tubes inside one another, and the magnetic gram susceptibility χ _g,para can be calculated according to Equation (1).

Here, Δf is the frequency shift of the reference (TMS, Hz), f is the frequency of the NMR spectrometer (Hz), m is the concentration of the sample (g/cm³), 3/4π is the demagnetization factor for a cylindric sample parallel to the high field of a cryomagnet [46], and χ _g,dia is the diamagnetic correction for the sample calculated from Pascal’s constants [47]. From (1), the molar magnetic susceptibility χ _m (cm³/mol) and the effective magnetic moment μ _eff (B.M.) are in turn obtained according to Equations (2) and (3):

where M is the molar mass and T the temperature.

It is important that both tubes are completely sealed airtight. It turned out that even the smallest amounts of oxygen lead to a paramagnetic shift and that different amounts of oxygen in the two tubes falsify the obtained shift. We therefore used a conventional J. Young NMR tube as the outer tube and employed a capillary as the inner tube, which was fused airtight. In order to apply this method to complex 1, we transferred a solution of this complex in DCM or pyridine into the outer sample tube and added 3% of TMS. The inner reference capillary only contained solvent and 3% of TMS.

Table 1 and Figure 2 summarize the ¹H NMR shifts resulting from a pyridine titration of 1 in DCM. For 1 dissolved in pure DCM a signal shift of 26.96 Hz is obtained. According to Equations (2) and (3), this corresponds to an effective magnetic moment of 5.22 B.M. which fits very well to a paramagnetic Fe(II) complex. Gradual addition of pyridine leads to a decrease in μ _eff; i.e., the samples become more diamagnetic. With the addition of 9000–10,000 equivalents of pyridine, a minimum value 3.30 B.M. is attained, and upon adding higher amounts of pyridine, the sample becomes more paramagnetic again. These results agree with the UV/Vis data and confirm that the desired six-coordinate complex 1-py is formed first, inducing a change of spin state. This way a CISSS effect of complex 1 could be demonstrated by Evans NMR measurements. Further addition of pyridine, however, leads to a ligand exchange probably going along with loss of the bmik coligand.

Figure 2:

Change of μ _eff during the Evans NMR pyridine titration of complex 1: plot of μ _eff as a function of added pyridine equivalents. The colors were chosen according to Figure 1: red corresponds to the initial spectrum, blue represents the coordination of pyridine to 1, and gray represents the ligand exchange with pyridine.

2.3 Investigation of complex 2

As the π–π* and n-π* bands of azo compounds change characteristically upon transition from the trans to the cis isomer, UV/Vis spectroscopy was employed to examine the switching behavior of complex 2 and compare it with that of the free ligand 5. In addition, structural changes due to coordination/decoordination should influence the electronic structure, so that a possible LD-CISSS effect should also be detectable by optical absorption spectroscopy.

Figure 3 shows the UV/Vis absorption spectra of photoswitchable ligand 5 (a) and the derived complex 2 (b) in DCM. In case of the free ligand 5, the initial spectrum (Figure 3a, black) shows the π–π* band at 319 nm (12,493 L mol⁻¹ cm⁻¹) and the n-π* band at 436 nm (319 L mol⁻¹ cm⁻¹), typical for the thermodynamically more stable trans isomer of azo compounds. As expected, after irradiation with light with 324 nm (Figure 3a, red; trans → cis) the intensity of the π–π* band decreases while that of the n-π* band increases (641 L mol⁻¹ cm⁻¹). The initial spectrum is obtained again by irradiation with a wavelength of 440 nm (Figure 3a, blue; cis → trans). A total of five switching cycles (trans → cis → trans) were run and no fatigue was detected. Moreover, by measuring the thermal relaxation after photoexcitation at 324 nm, a half-life of the cis isomer of 180 h was determined (Supporting Information, Figure S3).

Figure 3:

Electronic absorption spectra of azpy-trident ligand 5 (a) (c = 0.051 mM) and of complex 2 (b) (c = 0.024 mM) in DCM: initial spectra are shown in black. After irradiation, spectra of the respective cis isomers (324 nm, 10 min, red) and trans isomers (440 nm, 10 min, blue) are obtained.

For comparison, electronic absorption spectra of complex 2 were recorded under identical conditions. Here, the initial spectrum of complex 2 (Figure 3b, black) also shows the typical bands (π–π*: 320 nm, 35,747 L mol⁻¹ cm⁻¹; n-π*: 435 nm, 2545 L mol⁻¹ cm⁻¹) of the azo group. In addition, shoulders at 335, 352, and 379 nm indicate CT bands of the metal complex. Compared to complex 1, the typical MLCT band for the bmik ligand appears to be missing in the range of 550–600 nm. However, the intensity in the region of the n-π* band is quite high (ε = 1500–3000 L mol⁻¹ cm⁻¹; see Insets). We suspect that the MLCT transition associated with the bmik ligand (see above) is, at least partially, shifted to higher energy with respect to 1 and gets superimposed with the n-π* transition. This could be due to the fact that complex 2, unlike complex 1, does not exhibit a fac-1 geometry of the tridentate ligand (cf. Scheme 2). The change in coordination geometry of the tridentate ligand may be induced by the replacement of the phenyl group by a more bulky azopyridine moiety. Positioning this group cis to the equatorial bmik ligand (cf. Scheme 2) possibly becomes sterically unfavorable in complex 2 whence the tridentate ligand in the latter complex adopts a fac-2 or mer coordination. In, e.g., the mer-configuration of the tridentate ligand the bmik ligand exhibits an axial-equatorial coordination (cf. Scheme 2) and the energy of one of the d-orbitals which are involved in a MLCT transition into the π*-orbitals of the bmik ligand is lowered by backbonding interactions with two pyridine moieties. In complex 1 no such interactions exist for these d-orbitals. The corresponding MLCT transition thus is shifted to higher energy. Similar considerations apply to the fac-2 geometry such that we cannot distinguish between fac-2 and mer. For simplicity, however, it is assumed in the following that the latter geometry applies to complex 2.

By irradiation of 2 with light with 324 nm, the corresponding cis isomer can be obtained (Figure 3b, red) whereby no impairment of switching capability due to coordination of the azpy-trident ligand 5 is noticeable. This could be due to the fact that in complex 2 the photoswitchable unit is linked via a methylene bridge, which precludes electronic coupling. In previous work, we have demonstrated that coordination of an azopyridine-functionalized ligand reduces the switchability of the latter which was ascribed to electronic effects [48]. Employing light of 440 nm wavelength it was also possible to switch the complex back to trans (Figure 3b, blue), with 94% trans isomer being obtained in comparison to the initial spectrum. No further decrease in trans fraction was observed over 5 switching cycles. Finally, a half-life of 84 h of the cis isomer of complex 2 was determined by following thermal relaxation after photoexcitation at 324 nm (Supporting Information, Figure S4), about 50% of the value obtained for the free ligand. The comparatively small reduction of the half-life also suggests a weak influence of the metal coordination on the photophysical properties of the azo group [49].

To conclude, the photoswitchable unit of complex 2 can be reversibly switched between its trans and its cis isomer. However, MLCT bands, which could indicate a LD-CISSS effect, are superimposed by bands of the azo unit and, therefore, UV/Vis spectroscopy does not provide information about whether the ligand is coordinating/decoordinating and/or whether the system is changing its spin state. In order to obtain insight into these issues, we investigated system 2 using Evans NMR susceptibility measurements. For this, we dissolved complex 2 (4.85 mM) in DCM in the outer sample tube and added 3% of TMS. The inner reference capillary only contained DCM and 3% of TMS. Since the trans configuration corresponds to the thermodynamically more stable isomer, a paramagnetic complex and a downfield shift of the TMS signal, respectively, are expected for the initial measurement according to Schemes 1 and 4.

Figure 4 shows the initial spectrum in black at the bottom. It exhibits a shift of 48.48 Hz, which according to Equations (2) and (3) results in μ _eff value of 3.92 B.M., clearly below the value of complex 1 in solution (see above). Therefore, it can be assumed that not only the paramagnetic species of 2 is present, but also partially a diamagnetic octahedral complex. Subsequently, we exposed the sample to 365 nm for 30 min to obtain the cis isomer. Due to an intramolecular LD-CISSS a highfield shift and a decrease of μ _eff were expected. However, with 61.80 Hz and resulting µ _eff of 4.38 B.M. a slight downfield shift and an increase of paramagnetism were observed. Switching back to the trans isomer using light of 440 nm for 30 min should yield a more paramagnetic complex. Again, the opposite was observed, if only to a small extent: with a total shift of 60.55 Hz and a resulting μ _eff value of 4.34 B.M. the sample apparently became more diamagnetic.

Figure 4:

Shift of the TMS signals in ¹H NMR spectra measured by the Evans method of complex 2 with a concentration of 4.85 mM in DCM. From bottom to top: black shows the initial spectrum, red spectra are obtained after irradiation with light with 365 nm for 30 min, blue spectra are obtained after irradiation with light with 440 nm for 30 min and green spectrum after three days of relaxation time.

More trans isomer could be obtained after thermal relaxation for three days (Figure 4, green), leading to a more diamagnetic sample (52.06 Hz, 4.04 B.M.). Subsequent irradiation with wavelengths of 365 nm and 440 nm gave spectra following the previous observations (cf. Figure 4).

The obtained data suggest that the complex does not exhibit an intramolecular, but an intermolecular LD-CISSS effect. Here, the azopyridine of the switching unit coordinates in its trans configuration to another complex and decoordinates in its cis configuration (Scheme 7). The complex is then present as an octahedral diamagnetic trans isomer and as a square-pyramidal, paramagnetic cis isomer. Since an intermolecular LD-CISSS should be concentration-dependent, in contrast to the intramolecular LD-CISSS, we carried out Evans susceptibility measurements at different concentrations.

Scheme 7:

Assumed intermolecular LD-CISSS of complex 2: in trans configuration the switching unit coordinates to another complex and octahedral diamagnetic complexes are created (left); in cis configuration, the azo unit decoordinates and square pyramidal paramagnetic complexes are generated (right).

In case of a concentration-dependent intermolecular LD-CISSS, it is to be expected that the proportion of diamagnetic octahedral complex increases with increasing concentration. Table 2 collects the shifts obtained for the four different concentrations used in this experiment. As expected, the proportion of diamagnetic complex increases with increasing concentration. While at the lowest concentration of 1.25 mM with 4.87 B.M. mainly the uncoordinated paramagnetic complex appears to be present, less paramagnetic species is present at the highest concentration of 4.85 mM, leading to a µ _eff of 3.92 B.M. At each concentration we carried out irradiation and relaxation in analogy to Figure 4; i.e., after the initial measurement, the sample was irradiated with light with 365 nm for each concentration to obtain the cis isomer (Figure 5 and Supplementary Information). In agreement with Scheme 7, the concentration of uncoordinated, paramagnetic complex increased in all cases.

Figure 5:

Plot of μ _eff changes upon irradiation and relaxation for each concentration of complex 2: values for irradiation with light with 365 nm are shown in red, for irradiation at 440 nm in blue and for three days relaxation time in green respectively.

Subsequent exposure to 440 nm resulted again in more diamagnetic species. However, the μ _eff value of the respective initial measurement was not reached at any concentration (see above). Instead, after three days of thermal relaxation more diamagnetic species and correspondingly smaller values of μ _eff were obtained. At two concentrations (1.858 mM and 2.425 mM) even lower values compared to the initial ones were determined. Further irradiation with light with 365 nm and back switching at 440 nm corroborated the previous results. The Evans NMR data thus confirm the presence of a concentration-dependent, intermolecular LD-CISSS effect.

4.1 Synthesis of 3-(3-pyridylazo)benzoic acid (7)

To a solution of 3-aminopyridine (6.39 g, 67.9 mmol) in 60 mL pyridine 90 mL of a tetramethylammonium hydroxide solution (25 wt% in water) was added and the mixture was heated to 80 °C. 3-nitrosobenzoic acid (10.26 g, 67.9 mmol) was dissolved in 280 mL pyridine at 40 °C and then added dropwise to the 3-aminopyrdine solution. The reaction mixture was stirred for 4.5 h at 120 °C and additional 18 h at room temperature. After adjusting a pH of 3–4 by adding conc. hydrochloric acid the precipitated product was washed with water until the filtrate remained colorless and dried in vacuo. 3-(3-pyridylazo)benzoic acid was obtained as an orange solid. Yield: 14.32 g (63.0 mmol, 93%). Anal. calculated for C₁₂H₉N₂O₃: C, 63.43; H, 3.99; N, 18.49; found: C, 62.61; H, 4.19; N, 17.85. ¹H NMR (400.1 MHz, DMSO-d ₆): δ = 13.30 (s, 1H, OH), 9.16 (dd, ⁴ J = 2.4 Hz ⁵ J = 0.7 Hz, 1H, py H-2), 8.77 (dd, ³ J = 4.7 Hz, ⁴ J = 1.6 Hz, 1H, py H-6), 8.41–8.39 (m, 1H, benzyl H-2), 8.22 (ddd, ³ J = 8.2 Hz, ⁴ J = 2.4 Hz, ⁴ J = 1.6 Hz, 1H, py H-4), 8.19–8.13 (m, 2H, benzyl H-4,6), 7.76 (td, ³ J = 7.9 Hz, ³ J = 7.9 Hz, ⁵ J = 0.5 Hz, 1H, benzyl H-5), 7.64 (ddd, ³ J = 8.2 Hz, ³ J = 4.7 Hz, ⁵ J = 0.8 Hz, 1H, py H-5) ppm. ¹³C NMR (100.6 MHz, DMSO-d ₆): δ = 166.6 (Cq, 1C, C=O), 152.5 (CH, 1C, py C-6), 151.8 (Cq, 1C, benzyl C-1), 147.2 (Cq, 1C, py C-3), 146.5 (CH, 1C, py C-2), 132.4 (CH, 1C, benzyl C-6), 132.2 (Cq, 1C, benzyl C-3), 130.1 (CH, 1C, benzyl C-5), 127.6 (CH, 1C, C-4), 127.1 (CH, 1C, py C-4), 124.7 (CH, 1C, py C-5), 122.4 (CH, 1C, benzyl C-2) ppm. HR-ESI-MS: m/z calculated for C₁₂H₁₀N₃O₃ [M+H]⁺: 228.07675, 229.08011, found: 228.07644, 229.07983. IR (ATR): ṽ/cm⁻¹ = 2432 (br), 1695 (s), 1601 (w), 1425 (m), 1294 (s), 1272 (s), 1147 (m), 1039 (m), 819 (s), 759 (s), 692 (s), 678 (s), 639 (s).

4.2 Synthesis of 3-(3-pyridylazo)methyl benzoate (8)

To a suspension of 3-(3-pyridylazo)benzoic acid (14.32 g, 63.0 mmol) in dry MeOH (210 mL) 12 mL conc. sulfuric acid was added. After heating for 18 h under reflux the solvent was removed in vacuo. The residue was dissolved in 20 mL water and a pH of nine was adjusted by using a saturated sodium hydrogen carbonate solution. After extracting seven times with 50 mL DCM the combined organic phases were dried over MgSO₄ and the solvent removed in vacuo. The product was obtained as a red solid. Yield: 10.61 g (44.0 mmol, 70%). Anal. calculated for C₁₃H₁₁N₃O₃: C, 64.72; H, 4.60; N, 17.42; found: C, 64.40; H, 4.61; N, 16.99. ¹H NMR (400.1 MHz, DMSO-d ₆): δ = 9.14 (dd, ⁴ J = 2.4 Hz, ⁵ J = 0.6 Hz, 1H, py H-2), 8.76 (dd, ³ J = 4.7 Hz, ⁴ J = 1.6 Hz, 1H, py H-6), 8.37–8.35 (m, 1H, benzyl H-2), 8.20 (ddd, ³ J = 8.1 Hz, ⁴ J = 2.4 Hz, ⁴ J = 1.5 Hz, 1H, py H-4), 8.17 (ddd, ³ J = 7.9 Hz, ⁴ J = 2.2 Hz, ⁴ J = 1.3 Hz, 1H, benzyl H-4), 8.14 (ddd, ³ J = 7.7 Hz, ⁴ J = 1.7 Hz, ⁴ J = 1.2 Hz, 1H, benzyl H-6), 7.76 (td, ³ J = 7.9 Hz, ³ J = 7.9 Hz, ⁵ J = 0.5 Hz, 1H, benzyl H-5), 7.63 (ddd, ³ J = 8.2 Hz, ³ J = 4.7 Hz, ⁵ J = 0.5 Hz, 1H, py H-5) 3.91 (s, 3H, CH ₃) ppm. ¹³C NMR (100.6 MHz, DMSO-d ₆): δ = 165.5 (Cq, 1C, C=O), 152.5 (CH, 1C, py C-6), 151.7 (Cq, 1C, benzyl C-1), 147.1 (Cq, 1C, py C-3), 146.5 (CH, 1C, py C-2), 132.1 (CH, 1C, benzyl C-6), 130.9 (Cq, 1C, benzyl C-3), 130.2 (CH, 1C, benzyl C-5), 127.8 (CH, 1C, benzyl C-4), 127.0 (CH, 1C, py C-4), 124.6 (CH, 1C, py C-5), 122.2 (CH, 1C, benzyl C-2), 52.5 (CH₃, 1C, CH₃) ppm. HR-ESI-MS: m/z calculated for C₁₃H₁₂N₃O₃ [M+H]⁺: 242.09240, 243.09576, 244.09911, found: 242.09286, 243.09607, 244.09941. IR (ATR): ṽ/cm⁻¹ = 3075 (w), 3024 (w), 2964 (w), 2852 (w), 1725 (s), 1580 (m), 1425 (s), 1296 (s), 1274 (s), 1192 (s), 1149 (s), 1072 (s), 968 (m), 913 (m), 811 (m), 751 (s), 694 (s), 676 (s), 543 (m). UV/Vis: 0.060 mM in MeCN, 316 nm (π–π*, 12,833 L mol⁻¹ cm⁻¹), 430 nm (n-π*, 783 L mol⁻¹ cm⁻¹).

4.3 Synthesis of 3-(3-pyridylazo)benzyl alcohol (9)

Under N₂-atmosphere, 3-(3-pyridylazo)methyl benzoate (10.61 g, 44.0 mmol) was dissolved in 150 mL dry tetrahydrofuran. A 1 M LiAlH₄-THF-solution (30.8 mL, 30.8 mmol, 0.7 eq.) was added dropwise while cooling with ice. The reaction mixture was stirred for two days at room temperature until quenching dropwise with water. The precipitate was filtered and washed with THF and the filtrate was extracted five times with 60 mL brine and five times with 60 mL THF. The combined organic phases were dried over MgSO₄ and the solvent removed in vacuo. The crude product was purified on silicagel by column chromatography (DCM/MeOH 10:1) and 3-(3-pyridylazo)benzyl alcohol was obtained as a red-orange solid. Yield: 4.29 g (20.1 mmol, 46%). Anal. calculated for C₁₂H₁₁N₂O: C, 67.59; H, 5.20; N, 19.71; found: C, 67.44; H, 5.29; N, 19.45. ¹H NMR (400.1 MHz, DMSO-d ₆): δ = 9.13 (d, ⁴ J = 2.1 Hz, 1H, py H-2), 8.75 (dd, ³ J = 4.7 Hz, ⁴ J = 1.5 Hz, 1H, py H-6), 8.19 (ddd, ³ J = 8.2 Hz, ⁴ J = 2.4 Hz, ⁴ J = 1.6 Hz, 1H, py H-4), 7.89 (ddd, ⁴ J = 2.1 Hz, ⁴ J = 1.6 Hz, ⁵ J = 0.7 Hz 1H, benzyl H-2), 7.82 (dt, ³ J = 7.4 Hz, ⁴ J = 1.7 Hz, ⁴ J = 1.7 Hz 1H, benzyl H-4), 7.63 (ddd, ³ J = 8.2 Hz, ³ J = 4.7 Hz, ⁵ J = 0.8 Hz, 1H, py H-5), 7.60–7.55 (m, 1H, benzyl H-5), 7.56–7.52 (m, 1H, benzyl H-6), 5.40 (t, ³ J = 5.8 Hz, 1H, OH), 4.63 (d, ³ J = 5.8 Hz, 2H, CH ₂) ppm. ¹³C NMR (100.6 MHz, DMSO-d ₆): δ = 152.1 (CH, 1C, py C-6), 151.9 (Cq, 1C, benzyl C-3), 147.3 (Cq, 1C, py C-3), 146.3 (CH, 1C, py C-2), 144.3 (Cq, 1C, benzyl C-1), 130.0 (CH, 1C, benzyl C-5), 129.2 (CH, 1C, benzyl C-6), 126.9 (CH, 1C, py C-4), 124.6 (CH, 1C, py C-5), 122.0 (CH, 1C, benzyl C-4), 119.7 (CH, 1C, benzyl C-2), 62.4 (CH₂, 1C, CH₂) ppm. HR-ESI-MS: m/z calculated for C₁₂H₁₂N₃O [M+H]⁺: 214.09749, 215.10084, 216.10420, found: 214.09762, 215.10076, 216.11301. IR (ATR): ṽ/cm⁻¹ = 3197 (br), 2922 (w), 2871 (w), 2832 (w), 1595 (w), 1574 (w), 1462 (w), 1435 (m), 1419 (m), 1329 (w), 1245 (w), 1123 (w), 1082 (w), 1047 (m), 1025 (m), 792 (m), 686 (s), 627 (m), 527 (m). UV/Vis: 0.085 mM in MeCN, 319 nm (π–π*, 16,868 L mol⁻¹ cm⁻¹), 430 nm (n-π*, 1112 L mol⁻¹ cm⁻¹).

4.4 Synthesis of 3-(3-pyridylazo)benzyl chloride (10)

Under N₂-atmosphere, a solution of 3-(3-pyridylazo)benzyl alcohol (3.83 g, 18.0 mmol) in thionyl chloride (200 mL) was heated for 20 h at 90 °C. Excess thionyl chloride was removed in vacuo to give 3-(3-pyridylazo)benzyl chloride as a dark red solid. Yield: a quantitative yield was assumed. ¹H NMR (400.1 MHz, DMSO-d ₆): δ = 9.32 (d, ⁴ J = 2.3 Hz, ⁵ J = 0.6 Hz, 1H, py H-2), 8.95 (dd, ³ J = 5.2 Hz, ⁴ J = 1.4 Hz, 1H, py H-6), 8.61 (ddd, ³ J = 8.3 Hz, ⁴ J = 2.3 Hz, ⁴ J = 1.4 Hz, 1H, py H-4), 8.02 (d, ⁴ J = 2.1 Hz, 1H, benzyl H-2), 8.00 (ddd, ³ J = 8.3 Hz, ³ J = 5.2 Hz, ⁵ J = 0.5 Hz, 1H, py H-5), 7.95 (ddd, ³ J = 7.8 Hz, ⁴ J = 2.0 Hz, ⁴ J = 1.4 Hz, 1H, benzyl H-4), 7.75–7.71 (m, 1H, benzyl H-6) 7.69–7.64 (m, 1H, benzyl H-5), 4.92 (s, 2H, CH ₂) ppm. ¹³C NMR (100.6 MHz, DMSO-d ₆): δ = 151.8 (Cq, 1C, benzyl C-3), 148.2 (Cq, 1C, py C-3), 147.3 (CH, 1C, py C-6), 141.7 (CH, 1C, py C-2), 139.5 (Cq, 1C, benzyl C-1), 133.1 (CH, 1C, benzyl C-6), 132.6 (CH, 1C, py C-4), 130.1 (CH, 1C, benzyl C-5), 126.7 (CH, 1C, py C-5), 123.6 (CH, 1C, benzyl C-4), 122.7 (CH, 1C, benzyl C-2), 45.3 (CH₂, 1C, CH₂) ppm. HR-ESI-MS: m/z calculated for C₁₂H₁₁ClN₃ [M+H]⁺: 232.06360, 233.06696, 234.06065, 235.06401, 236.06736, found: 232.06373, 233.06708, 234.06079, 235.06413, 235.95401. IR (ATR): ṽ/cm⁻¹ = 3040 (br), 2358 (br), 2084 (br), 1964 (br), 1725 (m), 1603 (m), 1529 (m), 1456 (m), 1264 (m), 1135 (m), 1000 (m), 915 (m), 813 (m), 684 (s), 619 (s), 529 (s).

4.5 Synthesis of [N,N-bis(2-pyridylmethyl)]-3-(3-pyridylazo)benzylamine (azpy-trident ligand) (5)

A solution of 3-(3-pyridylazo)benzyl chloride (0.32 g, 1.36 mmol) in 10 mL acetonitrile was added to a solution of bis(2-pyridylmethyl)amine (0.27 g, 1.36 mmol) and cesium carbonate (0.80 g, 2.45 mmol, 1.8 eq.) in 5 mL acetonitrile. After adding TBAI (0.17 g, 0.45 mmol, 0.33 eq.) the reaction mixture was stirred for 24 h at 90 °C and then three days at room temperature. The precipitate was filtered and washed with acetonitrile and the collected filtrates extracted three times with 20 mL H₂O and the aqueous phase three times with 20 mL DCM. The collected organic phases were dried over MgSO₄ and the solvent removed in vacuo. The crude product was purified using flash column chromatography (DCM/MeOH, MeOH 0 → 20%) to give the azpy-trident ligand as a red-brown oil. Yield: 0.10 g (0.25 mmol, 18%). Anal. calculated for C₂₄H₂₂N₆: C, 73.07; H, 5.62; N, 21.30; found: C, 72.60; H, 6.06; N, 21.08. ¹H NMR (400.1 MHz, DMSO-d ₆): δ = 9.12 (dd, ⁴ J = 2.4 Hz, ⁵ J = 0.7 Hz, 1H, azo-py H-2), 8.75 (dd, ³ J = 4.6 Hz, ⁴ J = 1.6 Hz, 1H, azo-py H-6), 8.50 (ddd, ³ J = 4.9 Hz, ⁴ J = 1.8 Hz, ⁵ J = 0.9 Hz, 2H, py H-6), 8.19 (ddd, ³ J = 8.2 Hz, ⁴ J = 2.4 Hz, ⁴ J = 1.6 Hz, 1H, azo-py H-4), 7.95 (t, ⁴ J = 1.6 Hz, ⁴ J = 1.6 Hz 1H, benzyl H-2), 7.79 (td, ³ J = 7.6 Hz, ³ J = 7.6 Hz, ⁴ J = 1.8 Hz, 2H, py H-4), 7.82–7.77 (m, 1H, benzyl H-4), 7.68–7.64 (m, 1H, benzyl H-6), 7.63 (ddd, ³ J = 8.2 Hz, ³ J = 4.7 Hz, ⁵ J = 0.8 Hz, 1H, azo-py H-5), 7.61–7.57 (m, 1H, benzyl H-5), 7.59 (d, ³ J = 7.8 Hz, 2H, py H-3), 7.26 (ddd, ³ J = 7.5 Hz, ³ J = 4.9 Hz, ⁴ J = 1.2 Hz, 2H, py H-5), 3.79 (s, 2H, benzyl CH ₂), 3.78 (s, 4H, py CH ₂) ppm. ¹³C NMR (100.6 MHz, DMSO-d ₆): δ = 158.9 (Cq, 2C, py C-2), 152.1 (CH, 1C, azo-py C-6), 152.0 (Cq, 1C, benzyl C-3), 148.8 (CH, 2C, py C-6), 147.3 (Cq, 1C, azo-py C-3), 146.3 (CH, 1C, azo-py C-2), 140.6 (Cq, 1C, benzyl C-1), 136.6 (CH, 2C, py C-4), 132.3 (CH, 1C, benzyl C-6), 129.4 (CH, 1C, benzyl C-5), 126.9 (CH, 1C, azo-py C-4), 124.6 (CH, 1C, azo-py C-5), 122.8 (CH, 1C, benzyl C-2), 122.6 (CH, 2C, py C-3), 122.2 (CH, 2C, py C-5), 121.5 (CH, 1C, benzyl C-4), 59.1 (CH₂, 2C, py CH₂), 57.0 (CH₂, 1C, benzyl CH₂) ppm. EI-MS: m/z calculated for C₂₄H₂₃N₆ [M+H]⁺: 395.1, 396.1, 397.2, found: 395.1, 396.1, 397.1. HR-ESI-MS: m/z calculated for C₂₄H₂₃N₆ [M+H]⁺: 395.19787, 396.20123, 397.20458, found: 395.19777, 396.20111, 397.20440. IR (ATR): ṽ/cm⁻¹ = 3054 (w), 3038 (w), 3001 (w), 2984 (w), 1605 (m), 1579 (w), 1480 (m), 1445 (m), 1428 (m), 1313 (w), 1246 (w), 1161 (w), 1130 (w), 1063 (w), 1018 (w), 845 (m), 766 (s), 733 (s), 705 (s), 612 (s). UV/Vis: 0.051 mM in MeCN, 319 nm (π–π*, 12,493 L mol⁻¹ cm⁻¹), 436 nm (n-π*, 641 L mol⁻¹ cm⁻¹).

4.6 Synthesis of [Fe(trident)(bmik)](ClO₄)₂ (1)

Under N₂-atmosphere a solution of bis(2-pyridylmethyl)benzylamine (0.58 g, 2 mmol) in 2.5 mL dry methanol was added to a solution of iron(II) perchlorate hydrate (0.51 g, 2 mmol) in 1 mL dry methanol. After stirring for four days at room temperature a solution of bis(1-methylimidazole)ketone (0.19 mg, 1 mmol) in 2.5 mL dry methanol was added. The reaction mixture was stirred for additional 20 h at room temperature. The resulting light blue solid was filtered and washed with dry methanol. The filtrate was reduced to 2 mL and stored at 8 °C until the product precipitated as a dark blue solid. The product was filtered, washed with dry MeOH and dry diethyl ether and dried in vacuo. Yield: 0.38 g (0.52 mmol, 52%). Anal. calculated for C₂₈H₂₉Cl₂FeN₇O₉: C, 45.80; H, 3.98; N, 13.35; found: C, 46.21; H, 4.03; N, 13.05. HR-ESI-MS: m/z calculated for C₂₈H₂₉FeN₇O [M]²⁺: 267.58860, 268.09028, 268.59196, 269.09363; found: 267.58846, 268.09013, 268.59181, 269.08756. IR (ATR): ṽ/cm⁻¹ = 3135 (w), 2966 (w), 2928 (w), 2865 (w), 1768 (w), 1640 (w), 1608 (w), 1487 (w), 1427 (w), 1292 (w), 1168 (w), 1082 (m), 1025 (m), 900 (m), 766 (m), 707 (w), 620 (m), 603 (w), 419 (w), 289 (w). UV/Vis: 0.034 mM in DCM, 330 nm (22,153 L mol⁻¹ cm⁻¹), 569 nm (1392 L mol⁻¹ cm⁻¹).

4.7 Synthesis of [Fe(azpy-trident)(bmik)](ClO₄)₂ (2)

Under N₂-atmosphere a solution of the azpy-trident ligand (0.10 g, 0.25 mmol) in 3 mL dry methanol was added to a solution of iron(II) perchlorate hydrate (0.064 g, 0.25 mmol) in 1 mL dry methanol. After stirring for 20 h at room temperature, a solution of bis(1-methylimidazole)ketone (0.024 mg, 0.13 mmol) in 1 mL dry methanol was added. The reaction mixture was stirred for additional 20 h at room temperature. The both resulting blue and red-brown solids were separated manually. The red-brown product was washed with dry methanol and dry diethyl ether and dried in vacuo. Yield: 33.4 mg (0.040 mmol, 31%). Anal. calculated for C₃₃H₃₂Cl₂FeN₁₀O₉: C, 47.22; H, 3.84; N, 16.69; found: C, 47.69; H, 4.22; N, 16.24. HR-ESI-MS: m/z calculated for C₃₃H₃₂FeN₁₀O [M]²⁺: 319.10729, 320.10495, 320.60662, 321.10829; found: 319.10691, 320.10452, 320.60617, 321.10472. IR (ATR): ṽ/cm⁻¹ = 3136 (w), 3073 (w), 2946 (w), 1642 (w), 1607 (w), 1573 (w), 1485 (w), 1424 (w), 1294 (w), 1193 (w), 1164 (w), 1079 (m), 1022 (m), 901 (w), 815 (w), 766 (m), 705 (m), 621 (m), 605 (w), 418 (w), 286 (w), 214 (w). FT-Raman (solid): ṽ/cm⁻¹ 3068 (w), 1585 (w), 1493 (m), 1458 (s), 1200 (s), 1140 (w), 1039 (w), 1024 (w), 1001 (w), 933 (w). UV/Vis: 0.024 mM in DCM, 320 nm (π–π*, 35,747 L mol⁻¹ cm⁻¹), 435 nm (n-π*, 2545 L mol⁻¹ cm⁻¹).