The Structural and Magnetic Properties of FeII and CoII Complexes with 2-(furan-2-yl)-5-pyridin-2-yl-1,3,4-oxadiazole

Two novel coordination compounds containing heterocyclic bidentate N,N-donor ligand 2-(furan-2-yl)-5-(pyridin-2-yl)-1,3,4-oxadiazole (fpo) were synthesized. A general formula for compounds originating from perchlorates of iron, cobalt, and fpo can be written as: [M(fpo)2(H2O)2](ClO4)2 (M = Fe(II) for (1) Co(II) for (2)). The characterization of compounds was performed by general physico-chemical methods—elemental analysis (EA), Fourier transform infrared spectroscopy (FT-IR), nuclear magnetic resonance (NMR) in case of organics, and single crystal X-ray diffraction (sXRD). Moreover, magneto-chemical properties were studied employing measurements in static field (DC) for 1 and X-band EPR (Electron paramagnetic resonance), direct current (DC), and alternating current (AC) magnetic measurements in case of 2. The analysis of DC magnetic properties revealed a high spin arrangement in 1, significant rhombicity for both complexes, and large magnetic anisotropy in 2 (D = −21.2 cm−1). Moreover, 2 showed field-induced slow relaxation of the magnetization (Ueff = 65.3 K). EPR spectroscopy and ab initio calculations (CASSCF/NEVPT2) confirmed the presence of easy axis anisotropy and the importance of the second coordination sphere.


Introduction
The main focus of magnetochemical research is situated in the study of molecular functional materials exhibiting bistability in which physical properties can be triggered by a change of temperature, pressure, light or magnetic field, such as spin crossover compounds (SCO) or single-molecule magnets (SMM), where the latter are nanomagnets characterized by the slow relaxation of the magnetization.The widespread basis for the study of these phenomena originates in its potential technological applications in many areas of everyday life.In this case, SMM compounds are studied as materials for storage devices, quantum computing, sensors, and spintronics [1,2].Additionally, SCO compounds may find potential in many applications such as pressure or optical switches [3], gas sensors [4], pressure, or temperature sensors [5].Therefore, the magnetic properties of 3d elements are of much interest to the scientific community.Among all of the 3d metals, Co(II) possesses great predispositions for the formation of compounds exhibiting SMM behavior due to relatively large spin-orbit coupling and the possibility to vary the coordination numbers from 2 to 8 [6].Likewise, Fe(II) ranks among elements whose compounds are commonly known for SCO behavior [7].The tuning of physico-chemical properties of both SMM and SCO phenomena can be achieved by means of several factors.In SMMs, complexes honing of magnetization reversal barrier quantified by U eff parameter and magnetic blocking temperature (T B ) is achieved by the strengthening of magnetic anisotropy of the easy-axis type.The modification of U eff is accomplished by increasing magnitude of the zero-field splitting parameter D, due to a simple relationship among them, U = |D|(S 2 − 1/4) for half-integer spin and U = |D|(S 2 ) for integer spin, and it holds that values of ZFS (zero-field splitting) parameters are dictated by the coordination number, the ligand field, and the symmetry of the coordination polyhedron.In general, higher |D|-values should yield higher U eff and correspondingly higher T B [8][9][10][11].However, large zero-field splitting parameter E causes an increase of the tunneling rate of the magnetization.In SCO compounds, a proper ligand field and cooperativity between neighboring molecules can allow a transition between low spin (LS) and high spin (HS) state and vice versa, which is triggered by the change of temperature or pressure, eventually by the light.Since many of these parameters in both SMM and SCO relate to the coordination environment, search for suitable organic ligands is crucial for improving properties of these molecular compound classes.Miscellaneous organic ligands have been employed to study or enhance the magnetic properties of Fe(II) and Co(II) coordination compounds.Such ligands frequently include five or six-membered heterocycles containing at least one nitrogen atom.Examples of these heterocycles cover substituted diazoles, triazoles, tetrazoles, pyridines, pyrimidines, pyrazines, triazines, and others [12][13][14].In our search for suitable ligands, we were inspired by numerous publications with interesting magnetochemical results encompassing substituted triazoles, and more specifically 4-amino-3,5-di-2-pyridyl-4H-1,2,4-triazole (abpt).Our co-workers participated in publishing Co(II) field-induced single-ion magnets incorporating the abpt ligand.The first publication in 2014 introduced [Co(abpt) 2 (tcm) 2 ] (tcm = tricyanomethanide), which was identified as the field-induced single-ion magnet with large ZFS parameters and positive D = 48 cm −1 .The energy value for spin reversal barrier U eff = 86.2K was the highest at the time among Co(II) complexes with transversal magnetic anisotropy [15].The research continued in series of two publications involving other pseudohalide analogs-[Co(abpt) 2 (solv) 2 ]X 2 (solv = H 2 O and X = tcap, solv = H 2 O and X = nodcm, solv = CH 3 OH and X = pcp) (tcap = 1,1,3,3-tetracyano-2-azapropenide, nodcm = nitrodicyanomethanide, pcp = 1,1,2,3,3-pentacyanopropenide), [Co(abpt) 2 (X) 2 ] (X = nca, NCSe, ndcm, N 3 ) (nca = nitrocyanamide, ndcm = nitroso-dicyanomethanide).The analysis of DC measurements revealed strong magnetic anisotropy and D in 31-41.4cm −1 range with the exception of [Co(abpt) 2 (N 3 ) 2 ] where D = −24.1cm−1 .Subsequent inquiry of AC data showed field-induced slow relaxation of the magnetization and U eff in the scope of 71.6-108 cm −1 [16,17].Thus, we have decided to explore 1,3,4-oxadiazoles, oxygen analogs of abpt, which are relatively unexplored in a magnetochemical area.Moreover, the fact that 4H-1,2,4-triazole analog 1,3,4-oxadiazole contains more electron-withdrawing oxygen instead of nitrogen should have a notable effect on magnetic properties.Generally, studies involving a 1,3,4-oxadiazole heterocycle focus on crystal engineering, which covers the design and preparation of building blocks, distinguished crystal structures, and understanding of intermolecular interactions in supramolecular structure.Moreover, in these studies, nitrogen of the 1,3,4-oxadiazole ring do not coordinate, and, instead, substituents (e.g., pyridine, pyrazine) in position 2 or 5 provide suitable donor atoms (Scheme 1).

Synthesis
Preparation of ligand 2-(furan-2-yl)-5-(pyridin-2-yl)-1,3,4-oxadiazole (fpo) is already described in the literature [30].However, we obtained fpo through a four-step reaction, which shows simplified reaction Scheme 2. In the first step, transformation of picolinic acid (PA) to methyl picolinate (I) occurs via esterification in methanol/sulphuric acid solution.In the second step, production of picolinic acid hydrazide (II) from I ensues in methanol/hydrazine hydrate solution.The third step involves the conversion of II into picolinic acid 2-(2-furanylmethylene)hydrazide (III) in a simple reaction with 2-furaldehyde in methanol.The last step shows oxidative cyclization of the imine bond in dimethylsulfoxide with iodine as an oxidation reagent and potassium carbonate as the base.Subsequently, dissolving M(II) perchlorate hexahydrate (M = Fe, Co) in methanol and adding the solution to fpo in methanol and reflux of the mixture results in the formation of the products.Within one week, yellow crystals of 1 and orange crystals of 2 appeared.Since all organic ligands are already recorded in literature, their synthesis was inspired by these protocols [31][32][33].To confirm the structure of I, II, III, and fpo, the measurements of IR and NMR spectra were employed.1.

Static and Dynamic Magnetic Properties
The variable temperature and field experimental magnetic data for 1 and 2 are plotted in Figure 3.The room temperature effective magnetic moments (µ eff ) adopt values 5.1 µ B for 1 and 4.3 µ B for 2. These values are a bit higher than the spin-only values for S = 2 (µ eff /µ B = 4.9) and for S = 3/2 (µ eff /µ B = 3.9) calculated for g = 2.0, which indicates a significant contribution of the spin-orbit coupling to the ground state.On lowering the temperature to 1.9 K, there is only a small decrease of the effective magnetic moment for 1 down to 4.7 µ B , whereas compound 1 exhibit a much larger drop of µ eff down to 3.5 µ B .These data suggest that magnetic anisotropy in 2 is much more pronounced in comparison to 1.This is also reflected in the isothermal magnetization data measured at T = 2 K saturating to M mol /N A µ B = 4.1 for 1 and to M mol /N A µ B = 2.2 for 2, when compared to the theoretical limit values of M mol /N A µ B →g•S ≈ 4 for 1 and 3 for 2.Moreover, there are no maxima of the molar susceptibility at low temperature for both compounds under study.Thus, we can exclude significant antiferromagnetic interactions in the solid state.Therefore, the experimental magnetic data were analyzed with the following spin Hamiltonian suitable for describing magnetic anisotropy.
where the single-ion ZFS term is described by axial D and rhombic E parameters, and the Zeeman term is defined in the α-direction of the magnetic field as B α = B(sin(θ)cos(ϕ), sin(θ)sin(ϕ), cos(θ)) [34].
The molar magnetization (M a ) was then calculated from the partition function (Z) for a given direction of magnetic field B α as: and the integral (orientational) average of the molar magnetization was calculated as: to properly simulate experimental powder magnetization data.To obtain trustworthy parameters, both temperature and field-dependent magnetic data were fitted simultaneously.Moreover, we have tested both possible signs of the D-parameter for both compounds 1 and 2. The value of |D|~1.2cm −1 for 1 was found, which is rather small, but, in the case of compound 2, |D| is around 20 cm −1 , which confirms large magnetic anisotropy.In addition, the significant rhombicity was observed in both compounds (E/D >> 0)-Tables 2 and 3.  a The values of the parameters are in cm −1 .b δ is the energy of the first excited ligand field term.c The ∆ 1-4 and ∆ 1-5 are energies of the fourth and fifth spin levels, respectively.
The large magnetic anisotropy in 2 encouraged us to measure AC susceptibility data for this compound.The zero-static magnetic field measurements result in a zero signal of the imaginary part of AC susceptibility.The application of small static DC field confirmed slow relaxation of the magnetization (Figure S1).Therefore, small static field (B DC = 0.1 T) was chosen to suppress the tunneling of the magnetization and to avoid induction of the magnetic dipolar interactions in solid state, which often cause the appearance of another relaxation pathway.Thus, the temperature-dependent AC susceptibility data were acquired for the range of frequencies 1-1500 Hz, and these data were further analyzed with the one-component Debye model based on Equation (4).
This resulted in isothermal (χ T ) and adiabatic (χ S ) susceptibilities, relaxation times (τ), and distribution parameters (α) (Table S1), and the Argand (Cole-Cole) plot (Figure 4).Then, the temperature dependence of the relaxation times was fitted to the combination of two-phonon Raman and Orbach relaxation processes, which is shown below.
This results in C = 44.8K −n s −1 , n = 2.19, τ 0 = 1.34 × 10 −10 s −1 , and U eff = 65.3K (Figure 4).It is worth mentioning that the other combinations of the relaxation processes were tested, like direct and Raman, or direct and Orbach, but these were unsuccessful.The found spin reversal barrier U eff = 65.3K = 45.4 cm −1 is in good agreement with a value of 48.3 cm −1 derived with fitted parameters from DC magnetic data (D = −21.2cm −1 , E/D = 0.32).The Raman parameter n is shifted to lower values than the theoretical value of 9, which is usually ascribed to the involvement of both acoustic and optical phonons in the relaxation [35].

Electron Paramagnetic Spectroscopy
The X-band spectra of electron paramagnetic resonance (EPR) of complex 2 were measured in the temperature range from 2 K to 70 K to identify the type of the crystal-field anisotropy.The temperature evolution of the EPR spectra of complex 2 (Figure 5) shows a typical decrease of the signal intensity with increasing temperature for Co(II) complexes.Since there is no change of the shape of the spectra in the whole temperature range, a simplified effective spin S eff = 1/2 model due to the expected strong splitting between the ground and excited Kramers doublets was used for the analysis.This model assumes the mixing of higher excited states with the ground Kramers doublet as the consequence of the spin-orbit coupling, which yields highly anisotropic effective g-factors.The simulation of EPR spectra was performed within the EasySpin simulation package [36].The influence of the unresolved hyperfine coupling was included at first only by an anisotropic convolutional broadening.To fairly reproduce the experimental data, the anisotropic hyperfine interaction term A was then included together with an anisotropic convolutional broadening ∆B (full-width at half-height).The best agreement with the experiment was obtained using the set g 1 = 1.48, g 2 = 2.06, g 3 = 6.S2 (vide infra).Our attempts to simulate the EPR spectra using the spin-Hamiltonian formalism yielded the spectra similar to the experimental ones for both positive and negative values of the D-parameter with E/D ranging from 0.305 to 0.330.The values of the effective g-factors from the analysis of the EPR spectra using the effective spin S eff = 1/2 model allow revealing the type of the anisotropy (easy-axis or easy-plane) when compared with their theoretical prediction using Griffith-Figgis formalism or with ab initio calculations [37].Within the Griffith-Figgis formalism (with axial ∆ ax and rhombic ∆ rh crystal field term included), the calculated effective g-factors components, using a similar approach as in Reference [38], are restricted to g ≥ 2 for the positive axial field (easy-plane anisotropy).A very high value of g 3 = 6.60 is very close to the predicted g z component in the case of a negative axial field (easy-axis anisotropy) higher than 1500 cm −1 .In that case, the obtained g 1 , g 2 , and g 3 components of the effective g-factor can be assigned as g x = 2.06, g y = 1.48, and g z = 6.6.Although, in that case, the predicted difference between the ground and the first excited Kramers doublet ∆ 1-2 is approximately 150 cm −1 .The study of Titiš et al. shows that an additional deformation of the coordination octahedron, e.g., so-called scissoring as present in 2, will yield to the reduction of ∆ 1-2 [39].The experimental estimation of the effective g-factors and their anisotropy agrees better with the results of the ab initio calculations obtained for the second coordination sphere {[M(fpo) 2 (H 2 O) 2 ](ClO 4 ) 4 } 2− , which confirms the existence of the easy-axis anisotropy in 2 (Table S2).Furthermore, this result is supported by the presence of the Orbach relaxation process with an energy barrier close to the ∆ 1-2 (vide infra).

Theoretical Calculations
With the aim to resolve the ambiguity in fitted parameters from DC magnetic data, the multi-reference ab initio calculations based on the state-averaged complete active space self-consistent field method (SA-CASSCF) were performed for both complexes 1 and 2. Herein, the ORCA software package [40,41] was used for these CASSCF calculations with the active space defined as six electrons in five d-orbitals, CAS(6,5) for Fe II complex 1 and seven electrons in five d-orbitals, CAS (7,5), for Co II complex 2. In addition, the dynamic electronic correlation was handled by using the N-electron valence state perturbation theory (NEVPT2) [42][43][44].These calculations were undertaken on the mononuclear molecular fragments [M(fpo) 2 (H 2 O) 2 ] 2+ extracted from the experimental X-ray structures in which the atomic positions of all hydrogen atoms were optimized using the BP86 functional [45][46][47] together with the atom-pairwise dispersion correction method (D3BJ) [48,49].The results of CASSCF/NEVPT2 calculations are summarized in Tables 2 and 3 , which resulted only in much smaller changes in ZFS parameters than in the previous case.Figure 6 shows molecular fragments employed for calculations.All this points to the importance of the hydrogen bonds in a solid state in the molecular magnetism.To better visualize the impact of the second coordination sphere on the electronic structure and ZFS, the ab initio ligand field theory (AILFT) was utilized to calculate the energies of the d-orbitals in 1 and 2, as depicted in Figures 7 and 8. Evidently, the change of the molecular fragments {[M(fpo has a very significant impact on d-orbital's splitting, which is far from ideal O h symmetry.This is reflected in the energy levels of the ligand-field terms and, as expected, in ligand-field multiplets showing the zero-field splitting pattern (Figures 7  and 8).The importance and the impact of the second coordination sphere on the magnetism was also recently studied in other Co(II) complexes [50][51][52].In other words, the perchlorate induced hydrogen bonding alters the axial ligand field of aqua ligands, and, therefore, we decided to model this phenomenon in an alternative way.Thus, the bond distances Co-O in [Co(fpo) 2 (H 2 O) 2 ] 2+ varied from 1.9 to 2.5 Å and, for each geometry, the CASSCF/NEVPT2 calculations with CAS(7,5) were performed.The results of these calculations are shown in Figure 9. Evidently, the stronger axial ligand field led to small |D| values.On the contrary, a weak axial ligand field resulted in a negative D-parameter with small rhombicity.Generally, the magnetic anisotropy in the modelled system is very sensitive to small changes in the axial ligand field, which is manifested in abrupt changes of the rhombicity.

Materials and Methods
Chemicals for syntheses were purchased from Sigma-Aldrich (St. Louis, MO, USA), Across Organics (Geel, Belgium), or Alfa Aesar (Kandel, Germany) and used without any further purification.Reactions were monitored on aluminium TLC sheets pre-coated with silica gel 60 (SIL G/UV 254 , 0.2 mm, Macherey-Nagel, Düren, Germany).Ligand fpo was purified by column chromatography on Merck silica gel 60 (0.015-0.040 nm, Darmstadt, Germany) and the reaction scheme of organic syntheses drawn by a BIOVIA draw [53].Elemental analysis of chemical composition was acquired by the Flash 2000 (Thermo Scientific, Waltham, MA, USA) analyser.IR measurements were performed on spectrometer Jasco FT/IR-4700, data interpreted in Spectragryph, and assigned with the help of the table of known characteristic vibration frequencies [54,55].NMR (Varian, Palo Alto, CA, USA) experiments 1 H and 13 C were conducted on 400 MHz Varian spectrometer using CDCl 3 or d6-DMSO solvent and processed in the iNMR program [56].Magnetic measurements on PPMS Dynacool (Quantum Design, Quantum Design, San Diego, CA, USA) and SQUID magnetometer XL-7 (Quantum Design).X-ray experiments were carried out on a four-circle κ-axis Xcalibur2 diffractometer equipped with a CCD detector Sapphire2 (Rigaku Oxford Diffraction, Yarnton, UK).The CrysAlis software package (version 1.171.39.9g,Rigaku Oxford Diffraction, Yarnton, United Kingdom) was used for data collection and reduction [57].The structures for both 1 and 2 were solved by the SHELXT program incorporated in the wingx program package [58,59].Refinement based on intensities was performed using the SHELXL program [60].All non-hydrogen atoms in both 1 and 2 were refined anisotropically.Perchlorate anions in both compounds were found to be disordered in two positions and the ratio between both parts was 0.83:0.17for 1 and 0.82:0.18for 2. Hydrogen atoms in the ligand were placed in the calculated positions with isotropic thermal parameters tied with the parent atoms (U(H) = 1.2U(C)).The positions of the water hydrogen atoms for the aqua ligands were found in different Fourier maps and their isotropic thermal parameters were tied with the parent atoms (U(H) = 1.2U(O)).The structural figures were drawn using the Diamond software [61].Crystal data and the final parameters of the structural refinements for both 1 and 2 are summarized in Table 1.The EPR spectra were studied using Bruker ELEXSYS II E500 X-band spectrometer (Bruker BioSpin GmbH, Rheinstetten, Germany) with an operating frequency of 9.4 GHz equipped with ESR910 helium flow-type cryostat (Oxford Instruments plc, Abingdon, UK).The powdered sample was mixed with Apiezon N grease (M&I Materials Ltd, Manchester, UK) and attached to the Suprasil sample holder (Wilmad-LabGlass, Vineland, NJ, USA).

Preparation of the Complexes
Preparation of coordination compounds was performed as follows: Fe(II) (with a small amount of ascorbic acid) or Co(II) perchlorate hexahydrate (0.234 mmol) in 10 mL of methanol was added dropwise to fpo (100 mg, 0.469 mmol) in 10 mL of methanol.After an hour of reflux, the solution was filtered and left to evaporate slowly.Square shaped crystals formed within one week.

Conclusions
Our work focused on preparation and magneto-chemical analysis of new 1,3,4-oxadiazole based coordination compounds.Thus, isolated coordination compounds [M(fpo) 2 (H 2 O) 2 ](ClO 4 ) 2 (M = Fe(II) for (1); Co(II) for ( 2)) may serve as potential building blocks for preparation of polynuclear compounds.The data showed large rhombicity in both complexes and, in the case of 2, field-induced slow relaxation of the magnetization with an energy barrier U eff = 65.3K was detected.The axial type of the magnetic anisotropy was also confirmed in 2 by employing X-band EPR.Detailed theoretical investigation based on the CASSCF/NEVPT2 calculations revealed the importance of the second coordination sphere formed by the hydrogen bonding between perchlorate anions and aqua ligands on the magnetic anisotropy parameters D and E. The hydrogen bond tunes the axial ligand field and, thus, this kind of non-covalent contacts can serve as a trigger between the easy-plane and the easy-axis type of the magnetic anisotropy.

Supplementary Materials:
The following are available online at http://www.mdpi.com/1420-3049/25/2/277/s1. Figure S1: In-phase χ real and out-of-phase χ imag molar susceptibilities for 2 at zero static magnetic field and in non-zero static field.Table S1: Parameters of one-component Debye model for 2 derived according Eq.4 in main text.Table S2: The g-values for the ground state Kramers doublet calculated for the effective spin S eff = 1/2 for 2 derived from CASSCF/NEVPT2 calculations.CCDC 1972873-1972874 contain the supplementary crystallographic data for this paper.

Figure 2 .
Figure 2. The supramolecular 2D network formed through hydrogen bonds among perchlorates and aqua ligands in the crystal structure of 1.Only atoms involved in the hydrogen bonds are labelled.

Figure 3 .
Figure 3. Magnetic data for 1 and 2 shown as the temperature dependence of the effective magnetic moment and as the isothermal magnetizations measured at T = 2, 5, and 10 K.The empty circles represent the experimental data.Red full lines represent the fitted data using Equation (1) with D = −1.23 cm −1 , E/D = 0.31, g iso = 2.09 for 1 and D = −21.2cm −1 , E/D = 0.32, g xy = 2.13, g z = 2.37 for 2.

Figure 4 .
Figure 4. AC susceptibility data for 2. A: in-phase χ and out-of-phase χ molar susceptibilities at the applied external magnetic field B DC = 0.1 T (full lines are only guides for eyes).B: frequency dependence of in-phase χ and out-of-phase χ molar susceptibilities fitted with one-component Debye's model using Equation (4) (full lines).C: The Argand (Cole-Cole) plot with full lines fitted with Equation (4) and on the right of the fit of resulting relaxation times τ with the combination of Raman and Orbach processes (red line) using Equation (5).
60, A 1 = 140 MHz, A 2 = 165 MHz, A 3 = 160 MHz, ∆B 11 = 22 mT, ∆B 12 = 30 mT, and ∆B 13 = 7 mT.It should be noted that the spin-Hamiltonian formalism does not allow us to estimate the value of the D parameter for large values (only E/D ratio).The fitted g-factors are in good agreement with the parameters derived from ab initio calculations in Table

Figure 5 .
Figure 5. Temperature evolution of the EPR spectra of 2 measured with the excitation frequency of 9.4 GHz (solid lines).The simulation using the effective spin S eff = 1/2 model (black dashed line) with the parameters described in the text is included.

Figure 7 .
Figure 7.The graphical output of the CASSCF/NEVPT2 calculations with CAS(6,5) for the mononuclear molecular fragments of 1. Plot of the d-orbitals splitting calculated by the ab initio ligand field theory (AILFT) (A), low-lying ligand-field terms (B), and ligand-field multiplets-Kramers doublets (C).The quintet, triplet, and singlet spin states are shown in distinct colours.

Figure 8 .
Figure 8.The graphical output of the CASSCF/NEVPT2 calculations with CAS(7,5) for the mononuclear molecular fragments of 2. The plot of the d-orbitals splitting calculated by ab initio ligand field theory (AILFT) (A), low-lying ligand-field terms (B), and ligand-field multiplets-Kramers doublets (C).The doublet and quartet spin states are shown in red and green, respectively.

Figure 9 .
Figure 9.The graphical output of the CASSCF/NEVPT2 calculations with CAS(7,5) for the mononuclear molecular fragments of [Co(fpo) 2 (H 2 O) 2 ] 2+ of 2 for varying d(Co-O) bond distances.The plot of the d-orbitals splitting calculated by ab initio ligand field theory (AILFT) (A), four low-lying Kramers doublets (B), and the zero-field splitting parameters D and E (C and D).

Table 2 .
Comparison of the spin Hamiltonian parameters for 1 derived from CASSCF/NEVPT2 calculations done in ORCA and the experimental magnetic data a .
a The values of the parameters are in cm −1 .bδ is the energy of the first excited ligand field term.cThe∆ 1-5 and ∆ 1-6 are energies of the fifth and sixth spin levels, respectively.

Table 3 .
Comparison of the spin Hamiltonian parameters for 2 derived from CASSCF/NEVPT2 calculations done in ORCA and the experimental magnetic data a .