2.1 Synthesis

All chemicals were purchased from Sigma–Aldrich and were used as received.

HI_i-Pr verdazyl radicals have been prepared according to the procedure previously reported [11].

[Mg(I_i-Pr)₂·2H₂O]_n·0.35H₂O (I_i-Pr–Mg): 0.1 g (0.33 mmol) of verdazyl radical HI_i-Pr was dissolved in 15.0 mL of methanol. To this solution was added an excess amount 0.5 g (8.6 mmol) of freshly prepared Mg(OH)₂. The reaction mixture was stirred for 30 minutes, filtered, and the solution was allowed to crystallize. These crystals were further recrystallized from ethyl acetate to give homogeneous and pure crystals of I_i-Pr–Mg. The compound was characterized by elemental analysis, IR spectroscopy, and single-crystal XRD. Yield (%) 61, Elemental analyses. Anal. Calcd for C₃₀H_40.7MgN₈O_8.35 (M_r = 671.3): C, 53.68; H, 6.11; N, 16.69. Found: C, 52.57; H, 6.08; N, 16.19. IR (cm⁻¹): 483, 546, 657, 667, 711, 794, 1167, 1230, 1369, 1387, 1410, 1456, 1539, 1558, 1611, 1652, 2982, 3446.

[Mn(I_i-Pr)₂(CH₃OH)₄] (I_i-Pr–Mn): 0.1 g (0.4 mmol) of manganese (II) acetate tetrahydrate was dissolved in 5.0 mL of methanol. To this solution was added a solution of 0.24 g (0.8 mmol) of HI_i-Pr in 15.0 mL of methanol. To this reaction mixture was added 20.0 mL of toluene. The resulting reaction mixture was stirred for 3 hours and thereafter was concentrated under reduced pressure. Toluene (20.0 mL) was added to the concentrated solution and stirred for another 3 hours. All of the solvent was evaporated under reduced pressure and the obtained microcrystalline product was recrystallized from methanol to yield red crystals of Mn(II)-verdazyl compound. These crystals were characterized by elemental analysis, IR spectroscopy, and single-crystal XRD. Yield (%) 57, Anal. Calcd for C₃₄H₅₂N₈MnO₁₀ (M_r = 787.32): C, 51.84; H, 6.65; N, 14.22. Found: C, 51.41; H, 6.57; N, 14.33. IR (cm⁻¹): 481, 656, 710, 791, 1168, 1229, 1370, 1387, 1404, 1456, 1545, 1610, 1684, 2978, 3416.

[Zn(I_i-Pr)₂(CH₃OH)₄] (I_i-Pr–Zn): the zinc(II) analogue I_i-Pr–Zn was obtained using the same procedure as for I_i-Pr–Mn but the crystals were not suitable for single-crystal XRD. Yield (%) 48.2, Anal. Calcd for C₃₄H₅₂N₈O₁₀Zn (M_r = 796.31): C, 51.16; H, 6.57; N, 14.04. Found: C, 51.27; H, 6.37; N, 14.70. IR(cm⁻¹): 484, 546, 656, 710, 790, 871, 1018, 1128, 1168, 1229, 1301, 1368, 1386, 1557, 1610, 1685, 2934, 2978, 3419.

[Ni(I_i-Pr)₂(CH₃OH)₄] (I_i-Pr–Ni): 0.1 g (0.4 mmol) of nickel (II) acetate tetrahydrate was dissolved in 5.0 mL of methanol. To this solution was added a solution of 0.24 g (0.8 mmol) of HI_i-Pr in 15.0 mL of methanol. The resulting reaction mixture was stirred for 20 minutes and allowed to crystallize to yield small crystals of I_i-Pr–Ni. These crystals were characterized by elemental analysis, IR spectroscopy and single-crystal XRD. IR spectrum is given in Annexure. Yield (%) 55, Anal. Calcd for C₃₄H₅₂N₈NiO₁₀ (M_r = 790.32): C, 51.59; H, 6.62; N, 14.16. Found: C, 50.49; H, 6.57; N, 14.32. IR (cm⁻¹): 657, 710, 789, 1018, 1168, 1229, 1369, 1387, 1550, 1610, 1685, 2979, 3420.

[Co(I_i-Pr)₂(CH₃OH)₄] (I_i-Pr–Co): the cobalt (II) analogue I_i-Pr–Co was obtained using the same procedure as for I_i-Pr–Ni but the crystals were not suitable for single-crystal XRD. Yield (%) 68, Elemental analyses. Anal. Calcd for C₃₄H₅₂N₈O₁₀Co (M_r = 791.31): C, 51.58; H, 6.62; N, 14.15. Found: C, 51.69; H, 6.43; N, 14.24. IR (cm⁻¹): 485, 547, 657, 711, 789, 871, 1018, 1168, 1229, 1369, 1387, 1558, 1610, 1685, 2978, 3420.

2.2 Physical measurements and instrumentation

Elemental analyses were carried out at the Microanalytical Service of the Faculty of Chemistry of the Grenoble University. IR spectra were measured using a Nicolet iS50FTIR spectrometer in KBr.

2.3 Crystallographic structure determination

Crystallographic measurements for compounds I_i-Pr–Mg, I_i-Pr–Mn, I_i-Pr–Ni were carried out using an Oxford-Diffraction Xcalibur E charge-coupled device (CCD) diffractometer equipped with graphite-monochromated Mo Kα radiation. The crystals were placed 40 mm from the CCD detector. The unit cell determination and data integration were carried out using the CrysAlis package of Oxford Diffraction [12]. The structure was solved by direct methods using Olex2 [13] software with the SHELXS [14] structure solution program and refined by full-matrix least-squares on F_o² with SHELXL-2015 [14]. Atomic displacements for non-hydrogen, nondisordered atoms were refined using an anisotropic model. For I_i-Pr–Mg, which crystallizes in Sohnke C2, space group inversion twinning was checked by the refinement of BASF parameter (the BASF is the refined parameter which indicates the twin component ratio); however, the value of −0.1 (4) for the Flack gave no conclusive answer. In addition, the “Solvent Mask” tool available in Olex2 was used to model the contribution of disordered solvent molecules to structure factors. The hydrogen atoms have been placed by Fourier Difference accounting for the hybridization of the supporting atoms and the possible presence of hydrogen bonds in the case of donor atoms. The molecular plots were obtained using the Olex2 program [13]. The main crystallographic data together with refinement details are summarized in Table 1, whereas the selected bond lengths and angles are presented in Table 2.

CCDC-1892270 for I_i-Pr–Mg, CCDC-1892067 for I_i-Pr–Ni, and 1892068 for I_i-Pr–Mn contain the supplementary crystallographic data for this article; these data can be obtained free of charge via http://www.ccdc.cam.ac.uk/conts/retrieving.html.

The powder XRD experiment at room temperature was performed on a D8 Advance Bruker AXS diffractometer using a Cu Kα source with an emission current of 36 mA and a voltage of 30 kV. Scans were collected over the 2θ = 5–45 range using a step size of 0.01° and a count time of 0.5 s/step.

2.4 Magnetism studies

Magnetic measurements were carried out on microcrystalline samples with a Quantum Design SQUID magnetometer (MPMS-XL). Variable-temperature (2–300 K) direct current magnetic susceptibility was measured under an applied magnetic field of 0.1 T. All data were corrected for the contribution of the sample holder and diamagnetism of the samples estimated from Pascal's constants [15,16]. The analysis of the magnetic data was carried out by fitting the χT and χ thermal variations including temperature-independent paramagnetism, impurity contribution (ρ), and intermolecular interaction (zJ′) [16–18].

3.1 Synthesis

The chemistry of HI_i-Pr with a bivalent metal is explored. I_i-Pr–Mg was obtained by reacting HI_i-Pr with an excess of freshly prepared Mg(OH)₂ in methanol according to the following reaction:

After filtration, methanol was evaporated and crystals suitable for XRD were grown from ethyl acetate.

For the 3d bivalent metal ions (Mn(II), Co(II), Ni(II), and Zn(II)), the starting materials were the hydrates of the metal (II) acetate. The overall reaction toward the final compounds thus appears as a substitution of the acetate ions by the carboxylate moiety of I_i-Pr⁻ according to the following reaction:

In the case of Co(II) and Ni(II) (I_i-Pr–Co and I_i-Pr–Ni), a direct substitution of acetic acid in methanol solution was performed. On the contrary, for Mn(II) and Zn(II), it is necessary to displace the equilibrium toward the desired compounds, I_i-Pr–Mn and I_i-Pr–Zn, by a double azeotropic distillation of acetic acid in toluene. This procedure is similar to the one used to vary the nature of the carboxylate in Mn₁₂ derivatives [19]. It should be noted that an analogous synthesis was undertaken starting from iron (II) and copper (II) acetate. Nonetheless, the redox activity of these metal ions toward verdazyl radicals [7,10] has modified the overall reaction scheme and prevented the synthesis of the corresponding analogues. According to X-ray analysis, the magnesium salt contains two coordination molecules of water.

Elemental analysis is in good agreement with the proposed composition, but in the case of Mg salt, some discrepancy may be observed, which probably can be associated with solvent loss before analysis. The IR spectra of all the 3d divalent metal ion derivatives I_i-Pr–M (M = Mn, Co, Ni, Zn; Fig. SI 1) are very similar and markedly different from that of the magnesium (II) derivatives I_i-Pr–Mg (Fig. SI 2). This strongly suggests that the composition and bond repartitions are similar within the I_i-Pr–M (M = Mn, Co, Ni, Zn) series and notably different in I_i-Pr–Mg. Several features in the IR spectra of the synthesized compounds compared with that of the starting radical HI_i-Pr can be detailed (Fig. SI 3). First, the fingerprint of the organic part of HI_i-Pr is retrieved in all of the five derivatives. On the contrary, significant differences appear in the vibrations associated with the carboxylate moiety of the radical. Upon deprotonation and complexation, the characteristic band of OH vibration is deeply modified because in HI_i-Pr the molecules form H-bonded dimers that forced the O-H bond, hence displaced the OH vibration down to 2680–2550 cm⁻¹ whereas the O-H bonds of the water and methanol molecules present in the divalent metal adducts, I_i-Pr–M, vibrate at much higher wavenumbers, for example, 3420 cm⁻¹, indicating the presence of much less efficient H-bonding. Another striking consequence of the coordination is the modification of ν_as (COO) and ν_s (COO), respectively, the asymmetrical and symmetrical carboxylate oscillation bands. The difference between the two bands Δν= ν_as (COO) − ν_s (COO) takes value from (1680–1404) 282 cm⁻¹ in I_i-Pr–Mn down to (1652–1410) 242 cm⁻¹ when the carboxylate moiety adopts a pseudobidentate coordination (I_i-Pr–Mg). These results indicate that the verdazyl moiety itself shall not be coordinated whereas the carboxylate moiety does. These results must be compared with those obtained by the single-crystal X-ray structural analysis.

3.2 X-ray diffraction

Fig. 1(a) shows the asymmetric unit of I_i-Pr–Mg. Two different modes of coordination of the radical to magnesium (II) are found in I_i-Pr–Mg. They are denoted as A and B. One radical is coordinated to Mg(II) by its carboxylate group in a bidentate κ(O,O′) coordination mode (A). The second radical is coordinated in a μ − 1κ(O);1κ(O′) syn–anti mode (B): one oxygen atom (O1B) is coordinated to Mg1 atom (Mg1-O1B = 2.031 Å) whereas the second oxygen atom of the carboxylate group (O2B) is connected to another, crystallographically equivalent, Mg atom (O2B-Mg1 = 2.037 Å). These bridges between the Mg(II) ions lead to the formation a one-dimensional (1D) coordination polymer (Fig. 1b).

The coordination sphere of the metal ion is completed by two water molecules leading to a highly distorted octahedral environment (Fig. 1c). The two intrachain hydrogen bonds between the coordinated water molecules and the bidentate radical further reinforce the 1D chain. Fig. 1(d) shows the crystal packing of the coordinated polymer. There is no π-stacking between the molecules of bidentate chelating radicals A. On the contrary, there are some π–π interactions between bridging ligands B. Fig. 1e shows that the stacking is far from being perfect because of both the bulkiness of the isopropyl groups borne by the verdazyl moieties and to the coordinated magnesium (II) ions onto the carboxylate groups. Centroid-to-centroid distance is of 3.911 and 5.613 (Å) for phenyl–verdazyl and verdazyl–verdazyl pairs, respectively, and angle between two stacked molecules equal to 18.7°(2).

I_i-Pr–Mn and I_i-Pr–Ni are isostructural (Table 1). The structure of the latter is presented in detail in Fig. 2. This compound is formed of isolated mononuclear complexes. In the complex, the nickel (II) ion is coordinated by four methanol molecules and two radicals (Fig. 2a). The two radicals are coordinated to the metal ion by the carboxylate group in a monodentate mode. The second oxygen atoms of the two carboxylate groups (O2) are engaged in an intramolecular hydrogen bond with the hydrogen atoms (H4) of two coordinated methanol molecules. The two remaining methanol ligands form additional hydrogen bonds with oxygen atoms of the carbonyl groups of a verdazyl ring (Fig. 2b). The intermolecular interactions between the neighboring complexes are reinforced by π-stacking between the organic parts of the radicals arranged in a head-to-tail configuration (Fig. 2b). Centroid-to-centroid distance is of 3.940 for phenyl–verdazyl and 6.069 (Å) verdazyl–verdazyl pairs, respectively. These two intermolecular interactions lead to the formation of well-isolated 1D chain of complexes, which propagate along the c-axis (Fig. 2c and d).

The structure of the mononuclear complexes observed in I_i-Pr–Mn and I_i-Pr–Ni is very similar to that of the [M(OOCCH₃)₂(H₂O)₄] [20] supporting the substitution scenario proposed for the reaction. On the basis of the similarity of the IR spectra of I_i-Pr–Mn, I_i-Pr–Co, I_i-Pr–Ni, and I_i-Pr–Zn, as well and on the powder XRD (see Supplementary Information), this scenario should hold for I_i-Pr–Co and I_i-Pr–Zn.

3.3 Magnetic properties

Fig. 3 shows the temperature variation of magnetic susceptibility of I_i-Pr–Mg. The χ_MT product is worth 0.72 cm³ mol⁻¹ K at room temperature. This is close to the value expected for the sum of two noninteracting radicals (0.75 cm³ mol⁻¹ K for g = 2.00). Upon cooling, the χ_MT product decreases continuously with an increasing rate when the temperature is decreased. This evolution indicates that the exchange interactions between the two spin bearers are dominated by antiferromagnetic interactions.

With this diamagnetic metal ion, exchange interaction should mainly originate from π–π interactions between the radicals. Following the structure analysis of I_i-Pr–Mg (Fig. 1), such interactions exist between radicals labeled B whereas the A radicals are not coupled. Accordingly, the magnetic susceptibility is fitted using the following equation:

From fitting the value of antiferromagnetic coupling constant was found to be J = −17.56 cm⁻¹ and g = 1.96 (2). The low value of the g factor, as compared with the one expected for an organic radical, is probably due to the presence of a diamagnetic amorphous impurity. The constant of intermolecular interaction is −0.68 (1) cm⁻¹.

The thermal variations of χT product for the series of 3d series compounds I_i-Pr–M (M = Mn(II), Co(II), Ni(II), Zn(II)) is presented in Fig. 4. The values at the room temperature are consistent with theoretical value expected for two independent radical and one bivalent metal ion. Upon cooling, the χT products undergo a monotonous decrease. For I_i-Pr–Mn, I_i-Pr–Ni, and I_i-Pr–Zn, this temperature dependence suggests the presence of antiferromagnetic interaction whereas the much marked evolution for I_i-Pr–Co is a characteristic trademark of the strong spin orbital contribution of the octahedral Co(II) ion. Given that the radical is coordinated through its carboxylate, hence the long distance between the radical and the metal ion, and the similarity of the structure (Fig. 2) and thermal dependence of the χT products (Fig. 4) for I_i-Pr–Mn, I_i-Pr–Ni, and I_i-Pr–Zn, it clearly appears that the exchange interaction mainly originates from the π-stacking of the verdazyl moieties (Fig. 2b). Accordingly, the magnetic susceptibility data for I_i-Pr–M (M = Mn(II), Co(II), Ni(II), Zn(II)) can be written as

The values obtained for the exchange interaction parameter of the π-stacked radicals all along the I_i-Pr–M (M = Mn, Co, Ni, Zn) series fall below −2.4 cm⁻¹. They are in line with the value found for one of the polymorphs of HI_i-Pr [11] where a comparable highly head-to-tail arrangement with large verdazyl–verdazyl distances known to provoke a dramatic decrease in the exchange interaction [7] was found because of the steric hindrance imposed by the isopropyl side chains of the verdazyl moiety.