Magnetism of Manganese Complexes with Fluorinated Benzimidazole-Substituted Nitronyl Nitroxides

Abstract

A series of layered compounds of the formula {[Mn₂(radical)₃](ClO₄)}_n was obtained by a reaction (in methanol) of manganese(II) acetate with 2-(2-benzimidazolyl)-4,4,5,5-tetramethylimidazolidinyl-1-oxy-3-oxides fluorinated on the benzene ring and by successive addition of sodium perchlorate. This study showed that the magnetic properties of the complexes are sensitive to the number and arrangement of fluorine atoms in the paramagnetic ligands. It was found that the heterospin complex with 4-FBzIm-NN behaves as a magnet with Curie temperature T_C = 50 K, which is close to that of the {[Mn₂(BzIm-NN)₃](ClO₄)}_n complex containing a nonfluorinated ligand. Meanwhile, no 3D ordering was noted for complexes with difluorinated and 5-fluoro-substituted ligands.

1. Introduction

In the field of molecular-based magnets, the metal–radical approach invented by Rey and Gatteschi [1,2] in the late 1980s has proven to be particularly fruitful [3,4,5,6,7]. A fundamental basis of development of this approach was the study of the magnetic–structural correlations of a large number of diverse complexes of transition metal ions with radicals [8,9,10]. It has been shown that the use of free radicals as bridging ligands is particularly relevant, which allows the assembly of metal centers into 2D and 3D exchange-coupled systems [11,12]. The greatest progress in obtaining such systems was ensured by the use of functionalized nitronyl nitroxide radicals [13]. Firstly, due to the two symmetric NO groups where an unpaired electron is delocalized equally [14,15], these radicals can be employed as bridging ligands to generate extended metal–radical alternating systems [16,17]. A successful technique is to couple spins of different sizes to eventually achieve a nonzero resulting spin. Secondly, direct linking between the free radicals and the metal centers favors potent magnetic interactions. Finally, nitroxide radicals are highly persistent free radicals despite the presence of metal ions and can be prepared in nearly countless forms, thereby enabling the delicate design of a large panel of bridging magnetic ligands. Because stable nitronyl nitroxides can be synthesized with many substituents, their coordination chemistry is rich, encompassing d- and f-elements [18,19,20,21].

By means of nitronyl nitroxides, Rey’s and Gatteschi’s research groups have managed to obtain the first 1D (one-dimensional) metal–nitroxide compounds that behave like a magnet [22,23,24,25]. These metal–nitroxide 1D compounds have consequently shown excellent ferri- or ferromagnetic behavior along their strands and unfortunately low Curie temperatures (<20 K) owing to faint magnetic interactions between the chains. To attain higher Curie temperatures, it has been necessary to increase dimensionality of the radical–metal network of interactions. One possible way to create such a magnetic high-dimensional network is complexation of nitroxide polyradicals with metal hexafluoroacetylacetonates. For instance, by means of a series of nitroxide triradicals, a framework structure was implemented in the [Mn(hfac)₂]₃L₂ complex, which is ordered at 46 K as a ferrimagnet [26,27,28]. Using nitronyl nitroxide radicals, one of the present authors has designed another synthetic approach aimed at completely removing the auxiliary hexafluoroacetylacetonate ligands when full coordination surroundings of a metal center are exclusively available for nitroxide radicals for constructing a high-dimensional metal–radical network. The most relevant results have been achieved with the help of nitronyl nitroxides functionalized by benzimidazole or imidazole groups. For instance, in the case of 3d metals, as many as three radicals can be put around a metal and as many as four radicals in the case of ions of lanthanides. Moreover, subsequent to deprotonation, imidazole-substituted nitronyl nitroxides can serve as bridging bis-chelating ligands in the same manner as oxalates can for the assembly of extended networks while also acting as spin carriers. This methodology has been rather successful, in particular with manganese(II), for which 1D and 2D complexes have been obtained [18].

Two-dimensional complexes with the general formula {[Mn₂(radical)₃]X}_n are composed of Mn–radical two-dimensional networks with a honeycomb-like structure, where each manganese is coordinately bound to three radicals while anions X are situated in between the layers. Research into magnetic behavior of {[Mn₂(radical)₃]X}_n has revealed that the Curie temperature is highly dependent on the structure of the paramagnetic ligand and counterion X. For {[Mn₂(BzIm-NN)₃](ClO₄)}_n, the Curie temperature is approximately 55 K, which is nevertheless the highest among metal–nitroxide complexes [18]. Meanwhile, the {[Mn₂(Im-NN)₃](ClO₄)}_n complex manifests the initiation of magnetic long-range ordering at <1.4 K, implying a pronounced dependence of the Curie temperature on the paramagnetic ligand’s structure. To determine how the between-layer spacing distances can affect the magnetic characteristics, substances containing n-alkylsulfate anions {[Mn₂(BzIm-NN)₃](X)}_n (X = C_nH_2n+1SO₄, where n = 10 to 14 or 18) have been prepared. Investigation into magnetic behaviors shows that the interlayer spacing profoundly affects the magnetic properties. Indeed, compounds at n = 10–14 act as magnets, with Curie’s temperature diminishing from 48 K to 30 K when the spacing between layers enlarges, but no ordering was noted for the compound at n = 18. To additionally characterize Mn–nitroxide complexes and to determine how variations in the paramagnetic ligand affect their structure and magnetic properties, we obtained here new complexes containing mono- and difluorinated anions 5-FBzIm-NN, 4-FBzIm-NN, 5,6-F₂BzIm-NN, and 4,5-F₂BzIm-NN (Scheme 1). Interest in the uses of such fluorinated paramagnetic ligands can be explained as follows: the fluorine atom features rather low polarizability and the most potent inductive effect (among other chemical elements) and at the same time is a weak acceptor in relation to hydrogen bond donors. Therefore, addition of fluorine atoms to radical ligands may be a good way to control magnetic couplings and noncovalent between-molecule interactions. In the present study, by means of this method and within our comprehensive investigation into structure–property relationships typical of polyfluorinated organic radicals and their metal complexes [29,30], we present the preparation and magnetic characteristics of the first {[Mn₂(radical)₃](ClO₄)}_n complexes containing fluorinated radicals.

2. Results and Discussion

2.1. Synthesis and Crystal Growth

Compounds {[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n (1), {[Mn₂(5-FBzIm-NN)₃](ClO₄)}_n (2), {[Mn₂(4,5-F₂BzIm-NN)₃](ClO₄)}_n (3), and {[Mn₂(5,6-F₂FBzIm-NN)₃](ClO₄)}_n (4) were synthesized following the procedure described previously to obtain either valence tautomeric compounds {[Mn₂(Im-NN)₃]X}_n with X = ClO₄ˉ, BF₄ˉ, or PF₆ˉ or compounds with long-range magnetic ordering {[Mn₂(BzIm-NN)₃]ClO₄}_n. To this end, a nitronyl nitroxide radical (4-FBzImH-NN, 5-FBzImH-NN, 4,5-F₂BzIm-NN, or 5,6-F₂BzIm-NN) was first mixed with manganese(II) acetate at a 3:2 ratio in methanol, where manganese(II) acetate acted both as a source of metal ions and a way to deprotonate the imidazole moiety. Then, the compounds were slowly precipitated as polycrystalline powders upon addition of sodium perchlorate. Despite numerous attempts, only powders of nano-sized crystals were obtained for compounds 1–4, thus preventing structure determination by single-crystal X-ray diffraction. Moreover, our attempts to use powder X-ray diffraction failed to provide a reliable crystal structure, even when the [Mn₂(Im-NN)₃] structural motif from the crystal structure of {[Mn₂(Im-NN)₃]ClO₄}_n was used as a reference building block.

Despite the impossibility of direct structural analysis of compounds {[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n (1), {[Mn₂(5-FBzIm-NN)₃](ClO₄)}_n (2), {[Mn₂(4,5-F₂BzIm-NN)₃](ClO₄)}_n (3), and {[Mn₂(5,6-F₂FBzIm-NN)₃](ClO₄)}_n (4), according to the available experimental data, they have the structure of 2D complexes. Their composition was confirmed by the data of the elemental analysis, characteristic properties, and specific magnetic behavior inherent only in cousin 2D complexes [18]. In addition, XRD powder patterns carried on compound 1 are consistent with the lamellar structure of {[Mn₂(BzIm-NN)₃](ClO₄)}_n (ESI).

2.2. Magnetic Assays

Magnetic measurements were performed on bulk crystalline powders. It was found that {[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n (1) shows magnetic behavior that is different from that of complexes {[Mn₂(5-FBzIm-NN)₃](ClO₄)}_n (2), {[Mn₂(4,5-F₂BzIm-NN)₃](ClO₄)}_n (3), and {[Mn₂(5,6-F₂FBzIm-NN)₃](ClO₄)}_n (4). For 1, at 300 K in a magnetic field of 5 kOe, the χT product is 7.43 cm³∙K∙mol⁻¹, which is well below what could be expected (9.875 cm³∙K∙mol⁻¹) for the uncorrelated spins comprising two Mn²⁺ ions (S_Mn = 5/2) and three radicals (S_R = 1/2) [χT = g²(2S_Mn(S_Mn + 1) + 3S_R(S_R + 1))/8]. At the same time, the observed χT value is close to the expected (7.875 cm³∙K∙mol⁻¹) for S_tot = 7/2 spin units resulting from two Mn²⁺ ions antiferromagnetically coupled with three radicals [χT = g²(S_tot(S_tot + 1))/8]. Values of metal–nitroxide antiferromagnetic couplings can be predicted from data on magnetic measurements of molecular complexes, for example, the mer form of complex [Mn(BzIm-NN)₃](ClO₄), in which the Mn ion is coordinated by three chelating radicals. For this complex, the χT value on cooling decreases from 2.9 cm³∙K∙mol⁻¹ at 300 K and reaches a plateau whose χT value of ~1 cm³∙K∙mol⁻¹ corresponds to ground spin state S_tot = 1, as expected for one Mn(II) ion (S = 5/2) antiferromagnetically coupled with three radicals (S = 1/2). The fitting of the experimental magnetic susceptibility gave manganese–nitroxide coupling J/k_B = −113 K (H = −2JΣS_iS_j) [18,31]. In contrast to the molecular complex, for {[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n (1), the χT value increased upon cooling, thereby reaching a maximum, ~200 cm³∙K∙mol⁻¹, at 45 K in a magnetic field of 5 kOe or ~35 cm³∙K∙mol⁻¹ at 70 K at a magnetic field of 50 kOe (Figure 1). This behavior is similar to that shown by {[Mn₂(BzIm-NN)₃]ClO₄}_n and can be explained as follows: the alternating spins S = 5/2 and S = 1/2 (which are antiferromagnetically coupled within the 2D metal–radical frameworks) behave as ferromagnetically coupled S_tot = 7/2 effective units.

The inverse of magnetic susceptibility (1/χ) versus T proved to be almost linear in the temperature range 150–300 K, implying Curie–Weiss behavior with a θ value of 100 K and Curie constant C = 5.82 cm³∙K∙mol⁻¹ (Figure 2a). The magnetic susceptibility was fitted using the mixed quantum–classical Heisenberg model reported for a 2D honeycomb structure featuring alternating Mn²⁺ and Cu²⁺ ions [32]. Good agreement with experimental data was obtained in the temperature range 150–300 K, and the best fit gave antiferromagnetic interaction J/k_B = −76.9(4) K (H = −2JΣS_iS_j) with g_rad = 2.00(2) and g_Mn = 2.08(2). Below 150 K, the model could not be fitted to the experimental data because it does not take into account the 3D magnetic ordering. Figure 2b depicts zero field cool (ZFC) and field cool (FC) measurements of χ versus T for compound 1. They clearly indicate the onset of remanence below 50 K, which is characteristic of a magnet. This behavior suggests that a 3D ferromagnetic ordering took place, as observed previously in alternating Mn(II)–radical systems [18,25,33]. The first derivative on FC gave Curie temperatures T_c = 50 K. It is comparable to the Curie temperatures inherent in the {[Mn₂(BzIm-NN)₃]X}_n complex (T_c = 55 K).

The presence of ferrimagnetic coupling within the layers was confirmed by magnetization values M, which increased asymptotically, reaching 6.82 μ_B, but did not attain the expected M = 7 µ_B saturation value for effective S_tot = 7/2 (Figure 3a). Upon cooling at 5 and 50 kOe, magnetization values M increased, reaching 6.50 and 6.82 μ_B, respectively; the shape of dependence M(T) itself was magnetic-field-dependent (Figure 3b). Similar magnetic behavior has also been documented for other 2D and 1D coordination polymer-based compounds based on Mn²⁺ ions and nitronyl nitroxide radicals and has been ascribed to spin-canting within the 2D layers [18,34].

The magnetic characterization suggests that complex {[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n (1) manifests ferrimagnetic behavior similar to that of previously described complex {[Mn₂(BzIm-NN)₃](ClO₄)}_n, containing a nonfluorinated paramagnetic ligand. Both substances undergo a long-range 3D weak ferromagnetic ordering with Curie temperature T_c = 50 and 55 K, respectively. Thus, the introduction of a fluorine atom into position 4 of the benzimidazole moiety in BzIm-NN had virtually no effect on the magnetic properties of layered complex {[Mn₂(BzIm-NN)₃](ClO₄)}_n. At the same time, as demonstrated below, the introduction of a fluorine atom at position 5 of the benzimidazole moiety dramatically affected the magnetic characteristics of the complex. For complex {[Mn₂(5-FBzIm-NN)₃](ClO₄)}_n (2), at 300 K within a magnetic field of 5 kOe, the χT product was 8.41 cm³∙K∙mol⁻¹, which is between the expected values 7.875 and 9.875 cm³∙K∙mol⁻¹, respectively, for the S_tot = 7/2 spin unit resulting from two Mn²⁺ ions antiferromagnetically coupled with three radicals and for two S = 5/2 and three uncorrelated spins S = 1/2. Upon cooling, the χT value rose to reach only ~18 cm³∙K∙mol⁻¹ at 30 K in a magnetic field of 5 kOe or ~13 cm³∙K∙mol⁻¹ at 55 K at a magnetic field of 50 kOe (Figure 4). It is evident that in complex 2, the maximum value of χT observed with the variation in temperature is substantially less than that in complex 1.

The inverse of magnetic susceptibility (1/χ) versus T was almost linear in the temperature range 150–300 K, pointing to Curie–Weiss behavior with a θ value of 42 K and Curie constant C = 7.20 cm³∙K∙mol⁻¹ (Figure 5a). The magnetic susceptibility was fitted in the temperature range 300–150 K using the mixed quantum–classical Heisenberg model [32], thereby affording J/k_B = −32.8(2) K (H = −2JΣS_iS_j) with g_rad = 2.00(2) and g_Mn = 2.08(2). Therefore, in complex 2, the antiferromagnetic exchange interactions are much weaker than those in the manganese complex with 4-FBzIm-NN. Moreover, the ZFC and FC measurements of χ(T) for compound 2 (Figure 5b), as well as measurements of the field dependence of the magnetization at 2 K (Figure 6), did not reveal a phase transition into a magnetically ordered state.

The magnetic behavior of manganese complexes 3 and 4 with difluorinated paramagnetic ligands 4,5-F₂BzIm-NN and 5,6-F₂BzIm-NN almost matched that of complex 2 with a 5-fluorine derivative (ESI). This finding means that the introduction of a fluorine atom into position 5 or two fluorine atoms into either positions 4 and 5 or positions 5 and 6 of the benzimidazole moiety of the paramagnetic ligand either leads to the disappearance of the magnetic phase transition that was observed in {[Mn₂(BzIm-NN)₃](ClO₄)}_n or this transition shifts to the temperature region below 2 K. On the other hand, the magnetic behavior of the {[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n complex containing a 4-fluorine derivative is almost identical to that of the {[Mn₂(BzIm-NN)₃](ClO₄)}_n complex containing a nonfluorinated ligand. Both complexes showed an effect of magnetic ordering at temperatures of 50–55 K.

In the absence of structural data, it is difficult to explain why the introduction of a fluorine atom into position 5 of the benzimidazole moiety has such a strong effect on the magnetic properties of the complexes. Nonetheless, on the basis of data from previous studies, one can try to shed light on this phenomenon. Let us turn to crystal structures of {[Mn₂(BzIm-NN)₃](ClO₄)}_n that have been determined by Rietveld refinement of X-ray powder patterns (Figure 7) [18]. In {[Mn₂(BzIm-NN)₃](ClO₄)}_n, there are two sets of short contacts (less than the sum of van der Waals radii) between oxygen atoms of the ClO₄ anions located in the interlayer space and H atoms of the polymer layers {[Mn₂(radical)₃]⁺}_n. The first set consists of short contacts between the O atoms of the ClO₄⁻ ions and the H atoms of the methyl groups of the nitronyl nitroxide moiety: H27···O87 (2.368 Å), H19···O87 (2.517 Å), H70···O86 (2.548 Å), H52···O84 (2.095 Å), and H49···O84 (2.567 Å). The remaining O atoms form short contacts, notably, exclusively with the 5-H and 6-H atoms of the benzimidazole part of the molecule: H92···O84 (2.561 Å), H92···O86 (2.454 Å), H92···O87 (1.972 Å), H93···O84 (2.578 Å), H93···O86 (1.675 Å), H103···O85 (2.113 Å), H103···O87 (1.952 Å), and H111···O85 (2.200 Å). Among the last contacts, there are some that are considerably shorter, by 0.607, 0.768, or 1.045 Å, than the sum of the van der Waals radii of H and O atoms (2.72 Å) [35]. In other words, all ClO₄ anions are bound via strong noncovalent interactions with the 5-H and 6-H atoms of the benzimidazole moieties and, as a result, the perchlorate anions penetrate almost completely into the layers. It is obvious that the introduction of fluorine atoms at these positions should appreciably affect the arrangement of perchlorate ions in the interlayer space, thereby probably leading to enlargement of the spacing between the layers and in turn a decrease in the interlayer magnetic coupling of dipolar origin. The disappearance of the magnetic phase transition in complexes 2–4 and its presence in 1 at the Curie temperature of 50 K—which is similar to that in the {[Mn₂(BzIm-NN)₃](ClO₄)}_n complex (T_C = 55 K)—are consistent with this supposition.

3. Materials and Methods

3.1. Chemicals

2,3-Dihydroxylamino-2,3-dimethylbutane was synthesized following a previously reported procedure [36]. 2-(4-Fluorobenzo[d]imidazol-2-yl)- (4-FBzIm-NN), 2-(5-fluorobenzo[d]imidazol-2-yl)- (5-FBzIm-NN), 2-(4,5-difluorobenzo[d]imidazol-2-yl)- (4,5-F₂BzIm-NN), and 2-(5,6-difluorobenzo[d]imidazol-2-yl)-4,4,5,5-tetramethyl-4,5-dihydro-1H-imidazole-3-oxide-1-oxyl (5,6-F₂BzIm-NN) were prepared by condensation of 2,3-dihydroxylamino-2,3-dimethylbutane with 4-fluoro-, 5-fluoro-, 4,5-difluoro-, or 5,6-difluoro-1H-benzo[d]imidazole-2-carbaldehyde, respectively, as previously reported [37]. Other chemicals and solvents were of analytical grade and were used as purchased without further purification.

3.2. Preparation of 2D Manganese–Nitroxide Complexes

Safety notes: Metal perchlorates containing organic ligands can explode. Only a small amount of the complexes should be prepared, and they should be handled with great care.

{[Mn₂(4-FBzIm-NN)₃](ClO₄)}_n (1): To a solution of 4-FBzIm-NN (87.4 mg, 0.3 mmol) in 10 mL of methanol, a solution of Mn(CH₃CO₂)₂·4H₂O (49 mg, 0.2 mmol) in 10 mL of methanol was added. The resulting greenish-blue solution was stirred for 15 min, and then the solution was filtered and put at the bottom of a test tube, and a solution of NaClO₄ (36.7 mg, 0.3 mmol) in 20 mL of methanol was placed slowly on top of the first solution. A crystalline dark green powder was isolated by filtration after 5 days and washed several times with methanol cooled on ice and then washed with EtOH. Yield: 68 mg (63%). IR spectrum (KBr), ν/cm⁻¹: 3082, 2994, 2947, 1625, 1583, 1549, 1497, 1453, 1414, 1363, 1338, 1296, 1240, 1214, 1171, 1139, 1101, 1075, 974, 893, 870, 795, 742, 690, 625, 545, 463. Elemental analysis (%): C, 46.68; H, 4.57; Cl, 3.17; Mn, 10.31; N; 15.43. Calculated for C₄₂H₄₅ClF₃Mn₂N₁₂O₁₀ (%): C, 46.70; H, 4.20; Cl, 3.28; Mn, 10.17; N, 15.56.

{[Mn₂(5-FBzIm-NN)₃](ClO₄)}_n (2): To a solution of 5-FBzIm-NN (43.7 mg, 0.15 mmol) in 4 mL of methanol, we added a solution of Mn(CH₃CO₂)₂·4H₂O (25 mg, 0.1 mmol) in 3 mL of methanol. The resulting greenish-blue solution was filtered and put at the bottom of a test tube. Then, step-by-step, 3 mL of methanol and a solution of NaClO₄ (18.4 mg, 0.15 mmol) in 5 mL of methanol were added on top of the first solution. The reaction mixture was kept at 5 °C for 10 days. A crystalline dark green powder was isolated by filtration and then washed several times with ethanol. Yield: 32.8 mg (61%). IR spectrum (KBr), ν/cm⁻¹: 3083, 2994, 2945, 1599, 1578, 1481, 1451, 1415, 1372, 1337, 1296, 1260, 1214, 1176, 1139, 1112, 1019, 961, 871, 807, 776, 734, 655, 621, 591, 546, 452. Elemental analysis (%): C, 46.44; H, 4.15; Cl, 3.19; Mn, 10.23; N; 15.19. Calculated for C₄₂H₄₅ClF₃Mn₂N₁₂O₁₀ (%): C, 46.70; H, 4.20; Cl, 3.28; Mn, 10.17; N, 15.56.

Complexes {[Mn₂(4,5-F₂BzIm-NN)₃](ClO₄)}_n (3) and {[Mn₂(5,6-F₂FBzIm-NN)₃](ClO₄)}_n (4) were obtained similarly using 4,5-F₂BzIm-NN and 5,6-F₂BzIm-NN, with yields 53% and 68%, respectively. Their elemental analysis data and properties matched the structure of complexes of layered polymeric design. Complex 3: Elemental analysis (%): C, 43.81; H, 3.57; Cl, 3.91; Mn, 9.31; N; 14.13. Calculated for C₄₂H₄₂ClF₆Mn₂N₁₂O₁₀ (%): C, 44.48; H, 3.73; Cl, 3.13; Mn, 9.69; N, 14.82. Complex 4: Elemental analysis (%): C, 43.92; H, 3.52; Cl, 3.47; Mn, 9.70; N; 14.23. Calculated for C₄₂H₄₂ClF₆Mn₂N₁₂O₁₀ (%): C, 44.48; H, 3.73; Cl, 3.13; Mn, 9.69; N, 14.82. Both complexes are not soluble in EtOH and only slightly soluble in cold MeOH.

3.3. Magnetic Measurements

These measurements were performed on a PPMS-9 magnetometer (Quantum Design, San Diego, CA, USA) in the temperature range 2–300 K. The analyses were performed on polycrystalline samples using polyethylene capsules as a sample holder in the range 2–300 K. Field dependences of magnetization were determined at different temperatures (2, 4, and 6 K) in the range 0–50 kOe. The data were corrected for diamagnetism of the sample holder and of the constituent atoms using Pascal’s constants.

4. Conclusions

We succeeded in the synthesis of new complexes of manganese with fluorinated benzimidazole-substituted nitronyl nitroxides. The complexes belong to a family of layered compounds based on 2D metal–radical frameworks {[Mn₂(radical)₃]⁺}_n interleaved with layers of ClO₄ anions for electroneutrality. Magnetic characterization showed that all the complexes feature ferrimagnetic behavior within the 2D metal–radical frameworks, owing to alternation of antiferromagnetically coupled spins (S_Mn2+ = 5/2 and S_radical = 1/2). Among the synthesized heterospin compounds, only the complex with 4-FBzIm-NN undergoes a long-range 3D weak ferromagnetic ordering with Curie temperature T_C = 50 K. These results indicate that the magnetic phase transition in lamellar complexes [Mn₂(radical)₃(ClO₄)]_n is sensitive to the introduction of fluorine atoms into the aromatic part of the paramagnetic ligands.