Cobalt(II) Complexes Based on Benzylmalonate Anions Exhibiting Field-Induced Single-Ion Magnet Slow Relaxation Behavior

: Thereactionof(NBu 4 ) 2 Bzmal(whereBzmal 2 − isbenzylmalonatedianion)withCo(OAc) 2 · 4H 2 O gives the [Co(Bzmal)(EtOH)(H 2 O)] n 2D-polymer ( 1 ). The addition of 2,2 (cid:48) -bipyridine (bpy) to the starting system results in the [Co(Bzmal)(bpy) 2 ] · H 2 O · EtOH molecular complex ( 2 ). Their molecular and crystal structures were analyzed by single-crystal X-ray crystallography. An analysis of the static magnetic data supported by the SA-CASSCF / NEVPT2 calculations revealed the presence of easy-plane magnetic anisotropy in both complexes. The AC susceptibility data conﬁrm that both complexes show a slow ﬁeld-induced ( H DC = 1000 Oe) magnetic relaxation behavior.


Introduction
Cobalt coordination complexes exhibit redox [1,2], optical [3,4], and magnetic [5,6] properties that can be attractive for application in catalysis [7,8], biology, and medicine [9,10]. The magnetic behavior of cobalt(II) ions depends on the coordination environment, crystal field, and interaction with nearby paramagnetic ions [11]. It is known that the spin-orbit splitting in coordination compounds with "magnetic" metal ions arises as a result of structural distortions in the crystal and lowers various symmetries of the magnetic ion except octahedral. In this case, single-ion anisotropy occurs due to splitting in zero field (ZFS) and is a very important reason for the appearance of the properties of single-molecule/ion magnet (SMM/SIM) [6,12]. The crystal field determines the sign and value of the axial ZFS parameter (D, cm −1 ) [13]. Depending on the sign of D, two types of magnetic anisotropy are distinguished-namely, axial for negative D or easy planar for positive D. Negative D values are preferable for the formation of an energy barrier between two states with S = ±3/2 and the relaxation of magnetization [5]. There are some examples of mononuclear octahedral cobalt(II) complexes with easy plane magnetic anisotropy that exhibit field-induced slow magnetic relaxation behavior-i.e., SIM properties [14][15][16][17][18][19].
Here, we present the synthesis and structure of Co II coordination polymer with benzylmalonic acid dianions (Bzmal 2− ), [Co(Bzmal)(EtOH)(H 2 O)] n (1), and the product of polymer fragmentation by N-donor 2,2 -bipyridine (bpy) molecules, [Co(Bzmal)(bpy) 2 ]·H 2 O·EtOH (2). Magnetic measurement data combined with calculated data show that the presented compounds are SIM. These complexes are the first examples of Co II complexes with benzylmalonic acid anions and the first examples of SIM based on Co II malonates.

Synthesis of New Compounds
[Co(Bzmal)(EtOH)(H 2 O)] n (1). Co II acetate (0.1 g, 0.402 mmol) was added to a solution of a freshly prepared salt (NBu 4 ) 2 Bzmal (obtained from 1.072 mL water solution of tetrabutylammonium hydroxide (1.608 mmol) and benzylmalonic acid (0.156 g, 0.804 mmol)) in EtOH (40 ml). The reaction mixture was stirred with weak heating (t = 50 • C) for one hour. The resulting crimson solution was filtered to remove the cloudy precipitate and allowed to stand at room temperature for several days. The resulting crimson crystals are suitable for X-ray diffraction analysis. The crystals of 1 were filtered and dried in air at 25 • C. The yield of 1 is 0.063 g (50% counting per Co). Calc. (%) for C 12  [Co(Bzmal)(bpy) 2 ]·H 2 O·EtOH (2). Co II acetate (0.102 g, 0.410mmol) was added to a solution of 2,2 -bipyridine (0.256 g, 1.640 mmol) and freshly prepared salt (NBu 4 ) 2 Bzmal (obtained from 1.093 mL water solution of tetrabutylammonium hydroxide (1.640 mmol) and benzylmalonic acid (0.159 g, 0.820 mmol)) in EtOH (35 ml). The reaction mixture was stirred with weak heating (t = 60 • C) for one hour. The resulting orange solution was concentrated in a Schlenk flask to 15 ml and was left for one week. The resulting orange crystals are suitable for X-ray diffraction analysis. Crystals of 2 were filtered, washed by cold acetonitrile (−5 • C), and dried in air at 25 • C. The yield of 2 is 0.080 g

Single Crystal X-ray Diffraction Analysis
The experimental array of reflections was obtained on a Bruker APEX II diffractometer (Bruker AXS, Madison, WI, USA) (two-coordinate CCD detector, MoKα radiation, λ = 0.71073 A, graphite monochromator). An absorption correction was applied empirically using the SADABS [34] program. Using Olex2 [35], the structures were solved with the ShelXT [36] structure solution program using Intrinsic Phasing and refined with the ShelXL [36] using Least Squares refinement on F 2 . Hydrogens atoms of methyl, methylene, aromatic fragments, and hydroxylic groups were calculated according to those idealized geometries and refined with constraints applied to C-H and O-H bond lengths and equivalent displacement parameters (U eq (H) = 1.2U eq (X). X: central atom of XH 2 group and aromatic ring; U eq (H) = 1.5U eq (Y); Y: central atom of YH 3 , OH group, and H 2 O). For complex 2, the contribution of the disordered solvent molecules (one ethanol molecule according to C,H,N-analysis) to the calculated structure factors was removed using a solvent mask [37]. The crystallographic parameters and refinement statistics are given in Table 1. CCDC numbers 2044684 (for 1) and 2044683 (for 2) contain the supplementary crystallographic data for the reported compounds. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Powder X-ray Diffraction
The purity of the compound samples was approved by PXRD (See supplement materials S1). The powder patterns were measured on a Bruker D8 Advance diffractometer (Bruker AXS, Madison, WI, USA) with a LynxEye detector in Bragg-Brentano geometry, with the sample dispersed thinly on a zero-background Si sample holder, λ(CuKα) = 1.54060 Å, θ/θ scan with variable slits (irradiated length 20 mm) from 5 • to 41 • 2θ, stepsize 0.02 • .

Magnetic Measurements
Magnetic susceptibility measurements were performed using a Quantum Design PPMS-9 susceptometer (Quantum Design, San Diego, CA, USA). For dc-magnetic susceptibility measurements, the 5000 Oe magnetic field was applied. The measurements have been performed in the 2-300 K temperature range. For ac-susceptibility measurements of all the samples, oscillating ac-magnetic fields of 5, 3, and 1 Oe within frequency ranges 10-100, 100-1000, and 1000-10000 Hz, respectively, have been applied. These settings allowed one both to avoid sample heating at low temperatures (which may occur when modulation amplitudes and frequency are high) and to obtain the best signal-to-noise ratio. All the magnetic measurements were performed on polycrystalline samples sealed in polyethylene bags and covered with mineral oil in order to prevent the field-induced orientation of crystallites. The paramagnetic components of the magnetic susceptibility χ were determined taking into account both the diamagnetic contribution evaluated from Pascal's constants and the contributions of the sample holder and mineral oil.

Computational Details
Ab initio (post Hartree-Fock) calculations of ZFS parameters and g-tensor were performed based on state-averaged complete-active-space self-consistent-field (SA-CASSCF) wave functions complemented by N-electron valence second-order perturbation theory (NEVPT2) [38][39][40][41] using the ORCA program package (version 4.2.1) [42,43]. The SA-CASSCF/NEVPT2 calculations were performed with the geometry of the experimentally determined X-ray structures. The active space of the CASSCF calculations was composed of seven electrons in five d orbitals of Co 2+ ions (S = 3/2): CAS (7,5). The state-averaged approach was used, in which all 10 quartet (S = 3/2) and 40 doublets (S= 1/2) states were averaged with equal weights. The polarized triple-ζ-quality basis set def2-TZVP was used for all atoms. The ZFS parameters were calculated on the effective Hamiltonian theory [44], in which an approximation to the Breit−Pauli form of the spin-orbit coupling operator (SOMF approximation) was utilized. The splitting of the d-orbitals was analyzed within the ab initio ligand field theory (AILFT) [45,46], as implemented in the ORCA software.
Both the compounds were isolated as single crystals and polycrystals. The phase purity of the polycrystalline samples was confirmed by X-ray powder diffraction (see supplementary data, Figures S1.1 and S1.2). All the samples obtained were characterized by elemental analysis and IR spectroscopy. The IR spectra of the samples of complexes 1 and 2 show that they comprise benzylmalonate anions. The stretching vibrations of coordinated COO − groups are at the frequencies: 1634 cm −1 (ν as ) and 1441 cm −1 (ν s ) for 1 and 1596 cm −1 (ν as ) and 1441 cm −1 (ν s ) for 2. The presence of an aromatic substituent is supported by the stretching vibrations bands of C-H bonds at 3059 and 3023 cm −1 for 1 and 3077 and 3032cm −1 for 2 and C-C bonds of the ring at 1571 and 1496 cm −1 for 1 and 1567 and 1494 cm −1 for 2. The data obtained from our studies confirm the single-phase structure Crystals 2020, 10, 1130 5 of 14 of the samples. The background observed in the diffractogram of sample 2 indicates the presence of an amorphous impurity, a side product, or partially desolvated complex 2. Additionally, the composition of the product is confirmed by the results of the elemental analysis, which match the calculated gross formula. The compositions of the compounds match the formulas based on the X-ray diffraction results.

Magnetic Properties
The magnetic behavior of 1 and 2 was investigated under a 5000 Oe dc-magnetic field in the 2-300 K temperature range. The molar magnetic susceptibility temperature dependencies, χMT(T), are shown in Figure 3. The shape of the dependences is typical for Co 2+ complexes.
In both cases, the χMT values at 300 K (3.36 and 3.08 cm 3 /mol K for 1 and 2, respectively) are much higher than the spin-only value (1.875 cm 3 /mol K), which indicates an unquenched orbital contribution to the total magnetic momentum. The difference between the χMT magnitudes one can connect with the ligand environment. The χMT values are reducing with rising speed during the temperature decrease from 3.36 and 3.08 cm 3 /mol K (at 300 K) to 1.74 and 1.77 (at 2 K) for 1 and 2, respectively. Such a type of behavior is most likely due to the anisotropy of Co 2+ ions and the Zeeman effect caused by the applied field [60][61][62]. The experimental dc-magnetic susceptibility dependencies were approximated by the use of the PHI program with the set of parameters presented in Table 2 (also see Supplementary Data, Figure S5) [63]. The approximations showed positive D values for both compounds. In agreement with a positive D value, the perpendicular components of the g-tensor are larger than the parallel one. The coordination geometry of the cobalt(II) atom can be described as a distorted octahedron in Co II complexes with two 2,2 -bipyridine molecules and anions of other dicarboxylic acids: [Co(bpy) 2 (ox)] (ox 2− -oxalate dianion) [54], Co[(bpy) 2 (dpa)]·H 2 O (dpa 2− -2,2 -diphenic acid dianion) [55] and [Co(bpy) 2 (H 2 O)L]·H 2 O (L 2− -2,5-dicarbomethoxybenzene-1,4-dicarboxylic acid anion) [56] or with anions of monocarboxylic acids, for example pivalic (Hpiv (or tBuCO 2 H)): [Co(bpy) 2 (piv) 2

Magnetic Properties
The magnetic behavior of 1 and 2 was investigated under a 5000 Oe dc-magnetic field in the 2-300 K temperature range. The molar magnetic susceptibility temperature dependencies, χ M T(T), are shown in Figure 3. The shape of the dependences is typical for Co 2+ complexes.
In both cases, the χ M T values at 300 K (3.36 and 3.08 cm 3 /mol K for 1 and 2, respectively) are much higher than the spin-only value (1.875 cm 3 /mol K), which indicates an unquenched orbital contribution to the total magnetic momentum. The difference between the χ M T magnitudes one can connect with the ligand environment. The χ M T values are reducing with rising speed during the temperature decrease from 3.36 and 3.08 cm 3 /mol K (at 300 K) to 1.74 and 1.77 (at 2 K) for 1 and 2, respectively. Such a type of behavior is most likely due to the anisotropy of Co 2+ ions and the Zeeman effect caused by the applied field [60][61][62]. The experimental dc-magnetic susceptibility dependencies were approximated by the use of the PHI program with the set of parameters presented in Table 2 (also see Supplementary Data, Figure S5) [63]. The approximations showed positive D values for both compounds. In agreement with a positive D value, the perpendicular components of the g-tensor are larger than the parallel one. As expected for the pseudo-octahedral complexes, two sets of split t2g and eg orbitals were found in 1 and 2 (Figure 4).    (1)) with the optimal set of parameters (see text).

Quantum Chemical Calculations and Griffith Hamiltonian Approach
It has been widely discussed that the zero-field splitting (ZFS) spin-Hamiltonian (SH) is not always applicable to the description of pseudo-octahedral Co(II) complexes due to the significant contribution of the unquenched orbital angular momentum [64][65][66]. According to the SA-CASSCF/NEVPT2 calculations of isolated monomer fragment of 1 (Figure 1a) and mononuclear complex 2 (Figure 2), the separation of ground and the first excited quartet states is less 1000 cm −1 (Table 3). In this case, the SH is not fully operative for these complexes, it should be substituted by more sophisticated approach based on the Griffith Hamiltonian (GH), which acts within the ground octahedral 4 T 1g -term of the Co ion (state with fictitious orbital angular momentum L = 1) and explicitly involves the orbital contributions: where λ is the spin-orbit coupling parameter which typically ranges from −180 cm −1 to −130 cm −1 , κ is the orbital reduction factor which can vary from 0.6 to 1.0 depending on the complex,L andŜ are the orbital angular momentum and spin operators. It is seen that the three low-lying spin-free energy levels for both complexes (Table 3) can be associated with the axial crystal field (CF) splitting of the octahedral 4 T 1g term of the Co(II) ion into the ground orbital singlet, which is strongly separated from excited orbital doublet (it undergoes further small splitting by the rhombic crystal field into two orbital singlets). The energy level splitting corresponds to the positive sign of the axial CF parameter ∆ ax in the Griffith Hamiltonian (GH), Equation (1) and are relatively weak (small |∆ rh |) for 1 and large rhombicity for 2. In the case of ∆ ax > 0, the SH is applicable and the magnetic anisotropy in such complexes can be described in terms of ZFS too. It is interesting to note that the coordination environment distortions of the Co(II) ion in 1 (tetragonal elongation) and the nature of d-AO splitting (Table 3) are also in good agreement with ∆ ax > 0. The coordination environment of the Co(II) ion in complex 2 is strongly distorted, and to a greater extent differs from the D 4h distorted octahedron.  Figure 3 shows DC magnetic properties of 1 and 2 described by GH (Equation (1)). In order to avoid overparameterization, we used the values of the axial and rhombic CF parameters obtained from the SA-CASSCF/NEVPT2 energies of the spin-free states: ∆ ax = 1005.25 cm −1 and ∆ rh = 118.75 cm −1 for 1 and ∆ ax = 713.65 cm −1 and ∆ rh = 209.05 cm −1 for 2. These values are then were fixed when fitting the DC magnetic properties in order to reduce the number of varied parameters in the GH to only two parameters, λ and κ. By the simultaneous fitting of the temperature dependence, we obtain the best-fit values λ = −146.4 cm −1 and κ = 0.997 for 1 and λ = −146.4 cm −1 and κ = 0.999 for 2, which fall within the typical range of parameters for the high-spin hexacoordinate Co(II) complexes. Figure 3 shows that the found best-fit parameters and the CF parameters obtained from the ab initio calculations provide quite a satisfactory description of the experimental DC data.
As expected for the pseudo-octahedral complexes, two sets of split t 2g and e g orbitals were found in 1 and 2 ( Figure 4).  As expected for the pseudo-octahedral complexes, two sets of split t2g and eg orbitals were found in 1 and 2 ( Figure 4).    Table 2 shows the results of the ZFS splitting parameters and the g-tensors calculations by the CASSCF/NEVPT2 method, which indicate the presence in both complexes of a strong easy plane-type magnetic anisotropy (D > 0). Both complexes have strongly anisotropic g-tensors; moreover, the perpendicular components of the g-tensors are larger than the parallel one, which is consistent with positive D values.

AC Magnetic Measurements
In order to find out whether compounds 1 and 2 formed by anisotropic Co 2+ are single-ion magnets-i.e., exhibit a slow relaxation of magnetization-their magnetic dynamic were probed by measuring the ac-magnetic susceptibility.
In the zero dc-magnetic field, the out-of-phase values are negligible at 2 K for ac-frequencies in the range from 10 to 10000 Hz ( Figures S6.1 and S6.2). The absence of considerable χ"(ν) signals for complexes 1 and 2 most likely originates from the significant contribution from the quantum tunneling of the magnetization (QTM), resulting in fast relaxation. For minimizing the effect of QTM, non-zero dc-fields up to 5000 Oe have been applied. This resulted in the appearance of the significant out-of-phase signals on the χ"(ν) dependencies. Based on this data, the optimal value of the dc-field strength (at which the relaxation rate is the smallest) was selected as 1000 Oe for both complexes ( Figures S6.1 and S6.2).
Frequency dependences of the in-phase and out-of-phase components of the ac-magnetic susceptibility for complexes 1 and 2 taken under optimal H dc field are shown on Figures S6.3 and S6.4, respectively.
The corresponding χ"(ν) isotherms were approximated by using the generalized Debye model. This yielding temperature dependences of relaxation time (τ vs. 1/T) shown on Figure 5. Overall non-linear course of these dependences evidences contribution of non-Orbach magnetization relaxation mechanisms in both the cases. The increase in the intensity of the χ"(ν) signal from 2 to 3 K that was observed for 1 may originate from the collective behavior caused by the weak dipole-dipole or exchange interactions between the Co 2+ ions [67,68].  (2) and Raman+QTM (1) relaxation mechanisms.
The fact, that the whole data range could be well approximated by using only the Raman or the sum of Raman and QTM relaxation mechanisms, suggesting that the Orbach relaxation mechanism does not participate in relaxation.

Conclusions
The combination of benzylmalonate anions and cobalt(II) ions led to the formation of 2D polymeric structure 1 as a result of the chelate-bridging function of the dicarboxylic dianion. The above reaction in the presence of the chelate N-donor ligand bpy gave mononuclear molecule 2, in which the benzylmalonic dianion also acts as a chelate ligand. The geometry of the coordination environment of cobalt ions (CoO 6 for 1 and CoN 4 O 2 for 2) in the complexes corresponds to a distorted octahedron. The analysis of the DC magnetic data with SH and GF showed that the magnetic anisotropy of both complexes can be described in terms of ZFS (D > 0) as well as CF (∆ ax > 0, ∆ rh > 0). Ab initio calculations indicated the presence of a strong easy plane-type magnetic anisotropy (D > 0) and strongly anisotropic g-tensors in both complexes. These data confirm the results of the analysis of the magnetization relaxation mechanisms for the complexes in question, the Raman mechanism for 1 and a combination of the Raman and QTM mechanisms for 2.  Table S3: H-bonds for 1 and 2; Table S4: Selected parameters of π-π intermolecular interactions in 2; Figure