Alkoxy-Bridged Dicopper(II) Cores Meet Tetracyanonickelate Linkers: Structural, Magnetic, and Theoretical Investigation of Cu/Ni Coordination Polymers

Two heterometallic Cu(II)/Ni(II) coordination polymers, [Cu2(Hbdea)2Ni(CN)4]n (1) and [Cu2(dmea)2Ni(CN)4]n·nH2O (2), were successfully self-assembled in water by reacting Cu(II) nitrate with H2bdea (N-butyldiethanolamine) and Hdmea (N,N-dimethylethanolamine) in the presence of sodium hydroxide and [Ni(CN)4]2–. These new coordination polymers were investigated by single-crystal and powder X-ray diffraction and fully characterized by FT-IR spectroscopy, thermogravimetry, elemental analysis, variable-temperature magnetic susceptibility measurements, and theoretical DFT and CASSCF calculations. Despite differences in crystal systems, in both compounds, each dinuclear building block [Cu2(μ-aminopolyalcoholate)2]2+ is bridged by diamagnetic [Ni(CN)4]2– linkers, resulting in 1D (1) or 2D (2) metal–organic architectures. Experimental magnetic studies show that both compounds display strong antiferromagnetic coupling (J = −602.1 cm–1 for 1 and −151 cm–1 for 2) between Cu(II) ions within the dimers mediated by the μ-O-alkoxo bridges. These results are corroborated by the broken symmetry DFT studies, which also provide further insight into the electronic structures of copper dimeric units. By reporting a facile self-assembly synthetic protocol, this study can be a model to widen a still limited family of heterometallic Cu/Ni coordination polymer materials with different functional properties.


INTRODUCTION
−14 Considering the dependence of the magnetic interactions on the type of metal ion and the superexchange path, the selection of transition metals and bridging groups emerges as a key feature in the construction of heterometallic coordination systems. 15,16ystemic research has been conducted on the coordination polymers driven by the [M(CN) x ] (M = Fe, Co, Ni, Cu, or Pd; x = 4 or 6) linkers. 17In this context, cyanometalate anions including [Fe(CN) 6 ] 3− , 18−21 [Co(CN) 6 ] 3− , 11,22−27  [Ni-(CN) 4 ] 2− , 27−37 or [Pd(CN) 4 ] 2 −29,33,38 are particularly interesting given their ability to connect to other metal centers through a different number of cyanide groups, producing magnetically attractive heterometallic systems.However, the use of square-planar building blocks such as tetracyanonickellate(II) anions in the design of heterometallic systems containing multidentate aminoalcohols as stabilizing ligands has been underexplored.Such Cu/Ni systems can benefit from different coordination geometries of Ni(II) and Cu(II) centers 39 42 of which only two are the coordination polymers with dicopper(II) aminoalcoholate 1.171.42.49, 46 and an absorption correction was applied using the integrated multiscan absorption algorithm.For 2, the diffraction data were obtained on a Bruker APEX-II diffractometer with CCD detector using Mo Kα radiation (λ = 0.71073 Å).APEX programs were used for data collection, while SAINT and SADABS were applied for absorption correction. 47The crystal structures were solved by Direct Methods (SHELXT) 48 and refined by full-matrix least-squares on F 2 using SHELXL2014/4; 49 these are included in the package of programs, Olex2. 50All non-hydrogen atoms were refined with anisotropic displacement parameters.Hydrogen atoms were placed in calculated positions and treated by a mixture of independent and constrained refinements with U iso (H) values set to 1.2U eq of the attached carbon atoms and 1.5U eq for methyl hydrogen atoms (CCDC 2305435-2305436).Crystallographic data are summarized in Table 1.

Magnetic Properties.
Magnetic susceptibility measurements on crystalline samples of 1 and 2 were performed using a 6.5 T S700X SQUID magnetometer (Cryogenic Ltd.) over the temperature range from 4 to 300 K under an external magnetic field of 0.5 T. Polycrystalline samples (∼15 mg) were prepared and fixed in gelatin capsules.Each raw data file for the experimental magnetic susceptibility was corrected for the contributions of the sample holder, and the diamagnetic components of the constituent atoms were estimated from Pascal's constants as −301.4 × 10 −6 and −227 × 10 −6 emu/ mol for 1 and 2, respectively.
2.4.Broken-Symmetry DFT Studies.The ORCA 5.0.4 package 51,52 was used for all calculations.Unless stated otherwise, the B3LYP functional 53,54 with the ma-def2-TZVPP basis set 55,56 was used for all atoms for broken The Journal of Physical Chemistry C symmetry calculations.For single point and broken symmetry calculations, the SCF optimization convergence criteria were settled with the VeryTightSCF keyword, and integration grids of high density (Defgrid3 keyword) were employed.Optimization of geometries was performed using the ma-def2-SVP (preliminary calculations for H atoms optimization) or ma-def2-TZVP (final calculations) basis sets 55,56 for all atoms (default SCF convergence criteria and default integration grids were applied).The CASSCF studies were performed using the def2-TZVPP basis set for copper atoms and their first coordination sphere and def2-SVP basis set 56 for all other atoms.The AutoAux keyword 57 was used to generate other auxiliary basis sets in all cases.The C-PCM solvation model 58 with infinite dielectric constant was used as a rough model of a crystal field.The dispersion correction was introduced through the D4 model. 59,60Visualization of the molecular orbitals was performed by means of the Avogadro 1.2 program. 61,62Unless stated otherwise, the isosurfaces of the molecular orbitals are shown at the 0.02 au level.Extraction of exchange coupling constants was done through the equation J = −(E HS − E BS )/(⟨S 2 ⟩ HS − ⟨S 2 ⟩ BS ). 63Analysis of bond critical points and noncovalent interactions indexes based on reduced density gradients (RDG) 64 was performed using the Multiwfn 3.8 program. 65Selected ORCA outputs and Cartesian coordinates of the structural fragments used for the calculations are provided in Listings S1 and S5 (Supporting Information).S1.The structure of 1 is composed of onedimensional chains, running along the diagonal of the crystallographic ac plane.These 1D chains are assembled from the [Cu 2 (Hbdea) 2 ] 2+ units and [Ni(CN) 4 ] 2− linkers (Figure 1), so that the Ni atoms are 4-coordinate and have an almost ideal square-planar geometry.As expected for a strongly bound ligand (C�N), the Ni−C1 and Ni−C2 distances and the C1−Ni−C2 angle are 1.874(4) Å, 1.879(4) Å, and 88.73(17)°, respectively.The copper(II) centers in the dinuclear blocks are 5-coordinate and adopt a distorted square-pyramidal geometry.The Cu−O distances range from 1.907(3) to 2.338(3) Å, while the Cu−N bonds vary from 1.953(3) to 2.078(3) Å.The formation of a cyanide bridge can also be confirmed by FTIR spectroscopy.−69 The geometric index of distortion for a nickel center is equal to τ 4 = 0 (τ 4 ′ = 0, square-planar), whereas for the copper centers the corresponding index is equal to τ 5 = 0.21 [β = 171.16(13)°,α = 158.79(12)°], corresponding to a distorted square pyramid. 70,71Each resultant 1D chain features a 2C1 topology (Figure S6) and is surrounded by six additional chains with a spacing of 8.71−10.70Å, based on the distances between the adjacent Ni atoms (Figure 1).

Magnetic Properties.
The magnetic properties of compounds 1 and 2 were measured on polycrystalline samples, the phase purity of which was confirmed by powder X-ray diffraction (Figures S3 and S4).As previously discussed, both compounds are made by alternating μ-O-alkoxo-bridged copper(II) dimers interlinked by diamagnetic [Ni(CN) 4 ] 2− .As a result, it is expected that the intradimer Cu(II)−Cu(II) interactions would dominate the magnetic behavior through magnetic coupling involving μ-O-alkoxo bridges.In both  Å), suggesting that the contribution of the cyanide bridging ligand to the magnetic coupling is negligible.For 1, the temperature dependence of the product of magnetic susceptibility (χ p ) and temperature, χ p T, referred to a copper dimer unit, is shown in Figure 3.At room temperature (T = 301 K), χ p T is 0.62 emu K mol −1 per dimer, which is lower than the spin-only value of 0.75 emu K mol −1 for two S = 1/2 uncoupled Cu(II) ions (eq 1), indicating the occurrence of magnetic exchange.Here, N is Avogadro's number, β refers to the Bohr magneton, g is the Landéfactor with value g = 2, and k stands for the Boltzmann constant.As the temperature decreases, χ p T rapidly decreases from 300 to about 170 K, and below this decrease it becomes slower until approximately 50 K.
−76 In fact, the [Ni(CN) 4 ] 2− linkers can only provide very weak magnetic interactions (see the Theoretical Calculations section), so strong antiferromagnetic coupling is predicted to be mediated by the μ-O-alkoxo bridges.
For 2, the temperature dependence of the χ p T product (Figure 3) only reaches 0.48 emu K mol −1 at room temperature (T = 300 K), which is much lower than the expected value for a system of two independent Cu(II) ions (S = 1/2), considering g = 2.00.As the temperature decreases, the system shows a decrease of χ p T to about 70 K at a slower pace compared to 1. Below this, χ p T smoothly decreases as the temperature goes down, reaching almost zero at 4 K, denoting the presence of an overall strong antiferromagnetic behavior.Furthermore, the fact that the high-temperature χ p T limit was not yet achieved at 300 K for compounds 1 and 2 is consistent with the presence of strong exchange couplings within the dicopper(II) units.Therefore, and similarly to 1, it is assumed that 2 behaves as an antiferromagnetically coupled dicopper-(II) system where the [Ni(CN) 4 ] 2− units do not play any significant role.
From the plot of χ p as a function of temperature (Figure 4) for 2, a clear behavior for a Cu dimer can be observed with a maximum near 280 K.At very low temperatures, a rapid increase of χ p as a commonly observed Curie tail for coupled antiferromagnetic systems indicates a small percentage of paramagnetic impurities.Accordingly, an equation considering three terms including a major contribution of two antiferromagnetically coupled S = 1/2 spins was proposed to analyze the temperature dependence of the magnetic susceptibility of 2 (eq 2).It is described by the Bleaney− Bowers equation 75 and based on the model of an isolated Heisenberg dimer of S = 1/2 ions with Hamiltonian interaction H = −2JS 1 S 2 (where S 1 = S 2 for Cu(II)), a temperature-independent paramagnetism (TIP) term, and a low-temperature Curie contribution (eq 2).Herein, J is the intradimer superexchange interaction parameter between two Cu(II) ions, ρ is the parameter accounting for the percentage    ( ) As shown in Figure 4, the determined fitting results are in very good agreement with the experimental data giving the best-fit parameters J = −151.0cm −1 and g = 2.08 with a Curie tail corresponding to 2.8% of paramagnetic impurities and a TIP term A of 7.8 × 10 −4 .The negative J value indicates the existence of antiferromagnetic interactions occurring between the two Cu(II) ions, as supported by the shape of the χ p T−T plot.The obtained g value is reasonable for Cu(II).By performing the same type of analysis for 1 (Figure 5), the absence of a maximum in χ p vs T in the range of measured temperatures is noted, suggesting that the antiferromagnetic interaction coupling constant (J) may be considerably larger than the exchange constant for 2. In fact, fitting eq 2 to the experimental data resulted in J = −602.1 cm −1 with g = 2.32, 4.3% of paramagnetic impurities, and a TIP contribution of 7.0 × 10 −4 .In fact, the exchange coupling within the dimer is stronger in this case compared to 2, confirming the presence of stronger antiferromagnetic interactions.
The strong antiferromagnetic behavior and the observed differences in J values are related to different geometric parameters around the copper(II) centers within the dimeric units.For hydroxo-and alkoxo-bridged copper(II) compounds, the magnetic properties are influenced by the intradimer angles (Cu−O−Cu) as well as the hinge distortion of the [Cu 2 (μ-O) 2 ] 2+ core or the out-of-plane shift of the carbon atom from the alkoxo bridge. 44,73,77−81 An antiferromagnetic coupling dominates for θ higher than 98°, while a ferromagnetic interaction is predominant for θ lower than 98°.Compounds 1 and 2 present Cu−O−Cu angles of 104.09(12)°and 101.80(7)°, respectively, which are in agreement with the correlations established for alkoxo-, carboxylate-, and hydroxobridged Cu(II) structures.For 1, a stronger antiferromagnetic coupling (J = −602.1 cm −1 ) is in line with a higher value of θ.
3.3.Theoretical Calculations.−84 Although this approach suffers from several drawbacks, such as a strong dependency on the DFT functional and necessity of the spin decontamination procedure of the broken symmetry state, its straightforward nature makes it an important instrument in modern computational chemistry.−87 Since the solid-state crystal structures of 1 and 2 constitute coordination polymers, the selection of a proper 0D fragment of a structure is necessary prior to modeling the magnetic exchange.The structures feature dinuclear copper units where each copper atom is coordinated by one (1) or two (2) tetracyanonickelate blocks (Figures 1 and 2).Also, the structure of 2 contains water molecules that are located close to the ligand O-sites presumed to participate in the magnetic superexchange pathway.From these assumptions, the [Cu 2 (Hbdea) 2 {Ni(CN) 4 } 4 ] 2− (1a) and {[Cu 2 (dmea) 2 {Ni-(CN)} 4 ](H 2 O) 2 } 6− (2a) fragments were selected as starting points.The tetracyanonickelate block is expected to be in the diamagnetic low-spin state, 88,89 which agrees with the experimental measurements of magnetic susceptibility of 1 and 2. The broken symmetry calculations at the B3LYP/ma-def2-TZVPP level of theory suggested a strong antiferromagnetic coupling in both 1a and 2a, with the corresponding magnetic orbitals overlap integral S ab of 0.24941 and 0.17169, respectively (Figures 6 and 7).One can note the strong delocalization of these orbitals.The large overlap suggested the use of a spin-projected method by Yamaguchi et al. 63 to extract the final J values (see the Methods section).The predicted singlet−triplet gap for 1a is considerably higher than that for 2a (−657.40 and −238.41 cm −1 , respectively; the H ̂= −2JS ̂1S ̂2 formalism was used).This follows the geometry of the {Cu(μ-O) 2 Cu} cores where the Cu−O−Cu angle for 1a (103.9°) is slightly higher than that for 2a (101.8°). 90This tendency follows the disposition of experimental values for 1 and 2, where compound 1 shows a stronger exchange (Table 2).For comparative purposes, the broken symmetry calculations of fragments 1a and 2a were performed using the common DFT  The Journal of Physical Chemistry C functionals of meta-GGA and hybrid meta-GGA classes: M06-L, 91 r 2 SCAN, 92 TPSSh, 93 and PBE0 94 (entries a I −a VI in Table 2).As can be seen, all of the functionals stabilize the broken symmetry state and significantly overestimate the singlet− triplet gap (Table 2) as compared to the experimental data and the B3LYP functional.The TPSSh functional performs best after the B3LYP one.Earlier, the TPSSh functional was shown to outperform B3LYP for manganese coupled systems; 95 thus, for copper d 9 compounds it is expected to have similar performance.Apart from the functional, the choice of a spindecontamination scheme is an important step.The Yamaguchi et al. 62 expression routinely applied in the present study covers a broad range of interaction magnitudes, from weak to strong ones.However, the expression by Bencini, Noodleman, and Ruiz (BNR) et al. 96−98 J = −(E HS − E BS )/(S max (S max + 1)) results in a significant decrease of the predicted singlet−triplet gaps up to almost quantitative prediction of J constants for both 1a and 2a in the case of the M06-L functional (Table 2, a III ).Considering a large scattering of couplings predicted at different levels, the outstanding results demonstrated by the M06-L/BNR-expression level are likely a coincidence than a systematic result.It should be noted that the 2 kcal mol −1 precision can be viewed as excellent for DFT methods, while it corresponds to 700 cm −1 �a value higher than the magnitudes of typical exchange couplings.Thus, the broken symmetry calculations presented here describe the magnitudes of exchange couplings with a reasonable deviation from the experimentally determined values.
At the next step, we investigated the influence of structural and charge corrections on the magnitude of magnetic exchange.It is known that the coordinates of hydrogen atoms are poorly determined from the X-ray crystallographic data, where most often the X−H distances and angles are constrained due to a low scattering factor of hydrogen atoms.The respective constraints, however, not always correspond to the real geometry, 99,100 especially in the case of O−H groups.Hence, optimization of the H atom positions in the fragment may change the predicted magnitude of magnetic coupling.Positions of the H atoms in fragments 1a and 2a were optimized at the B3LYP/ma-def2-TZVP/D4 level, resulting in the expected elongation of all X−H bonds (fragments 1b and 2b).In the case of 2b, the hydrogen atoms of the water molecule underwent a rotation, enforcing the O−H•••O hydrogen bond with the alkoxy group (Figure 8) and making a weak hydrogen bond O−H•••NC with the cyanido group, as evidenced by the Laplacian as well as reduced density gradient (RDG) 64 plots (Figure 9).The electron densities in the respective bond critical points ρ(r BCP ) constitute 2.17 × 10 −2 and 8.8 × 10 −3 a.u., respectively.According to the relation proposed by Emamian et al., 101 these densities correspond to the binding energies of −34.7 and −16.7 kJ mol −1 , respectively.Thus, considering these hydrogen bonds only, the binding energy of a water molecule to a metal-complex framework can be considered as no less than −51.4 kJ mol −1 .The direct calculations of the binding energy (BE) at the B3LYP/ma-def2-TZVPP/D4 level resulted in the twice lower energy of −27.4 kJ mol −1 per water molecule (where BE AB = E AB − E A − E B ).−104 The broken symmetry calculations of the hydrogenoptimized fragments 1b and 2b revealed the resembling J values compared to those obtained for 1a and 2a (Table 2).Moreover, further alterations of the structure by adding protons to simulate coordination of the cyanido ligands and compensate negative charge, as well as truncating the n-butyl groups for 2 or replacing its tetracyanonickelate blocks with HCN and CH 3 CN, affected the magnitude of the predicted exchange coupling to a very limited extent (Table 2), keeping  Unless stated otherwise, the constants were calculated at the B3LYP/ma-def2-TZVPP level using H ̂= −2JS ̂1S ̂2 spin Hamiltonian.The H atoms positions were optimized in all cases where protons were added or ligands transformed.b Optimization was performed at the B3LYP/ma-def2-TZVP/D4 level.c 6 and 12 protons were added to the models 1b and 1c, respectively.d The water molecules were eliminated from the model 2b without further optimizations.e two protons added to the Ni−CN groups in the trans-position relative to copper centers.f The same as footnote e but with truncation of n-butyl groups to methyl ones.The def2-TZVPP basis set was used for copper atoms and first coordination sphere (including C atom of the cyanido groups), and def2-SVP was used for all remaining atoms., where the water molecules were generated by using the coordinates of the respective alkoxy-bridging atoms of the ligand (Figure S8).Only a very weak antiferromagnetic coupling of −0.33 cm −1 was predicted (magnetic orbital overlap S ab = 0.00447) in this way, suggesting that the dinuclear copper blocks are rather magnetically isolated.
According to the single point open-shell DFT calculations at the B3LYP/ma-def2-TZVPP level, the unpaired electrons are located at the d x 2 −y 2 copper orbitals in all studied cases, as expected from the square-pyramidal coordination geometries around the copper centers.The fragment 1f containing two protons at the terminal cyanido groups and truncated n-butyl groups was used for further comparative studies.The broken symmetry calculations using the B3LYP functional and ma-def2-TZVPP basis set for all atoms reveal J = −657.27cm −1 (Table 2).Because of the absence of a negative charge in 1f, the diffuse function of the basis sets becomes not necessary and can be excluded.Further, a small involvement of [Ni(CN) 4 ] blocks into the magnetic exchange suggests the possibility of using the large basis set on the copper atoms and their nearest surroundings only.Thus, the calculations using the def2-TZVPP basis set for copper and the first coordination sphere, keeping all other atoms at the def2-SVP level, resulted in J = −666.17cm −1 , a value that is very close to all others calculated for 1 (Table 2).The corresponding magnetic orbitals of 1f (Figure S9) are of 58.5% of d x 2 −y 2 character and involve 5.6% and 3.3% of p x and p y atomic orbitals of bridging oxygen atoms for each of the α and β magnetic orbitals.The restricted open shell (ROKS) calculations resulted in the singly occupied frontier molecular orbitals (SOMO) of shape similar to that of unrestricted corresponding orbitals (Figure 9), which contain 64% and 54% of d x 2 −y 2 copper atomic orbitals, the latter involving 8.6 and 10.2% of the oxygens of p x and p y atomic orbitals, respectively, showing some degree of delocalization over the bridging atoms.The energy gap between the ROKS/ SOMO orbitals is 9622 cm −1 .
The state-averaged SA-CASSCF calculations for the 1f and 2c models in the smallest possible (2,2) active space resulted in the ferromagnetic ground state with the first singlet 209.3 and  The Journal of Physical Chemistry C 256.9 cm −1 above, respectively.The active space orbitals are of the same shape as the ROKS/SOMO and localized orbitals and are of d x 2 −y 2 character as well (Figure S9).Although in this case the singlet−triplet gap is smaller for the model of complex 1, which follows the experimental tendency (Table 2), the inverse ground state accounts for incorrect energies of the singlet−triplet splittings obtained for the very limited CAS-(2,2) model.The involvement of all copper d-orbitals into the active space changes the energy gap drastically, predicting the correct antiferromagnetic ground state for CAS (18,10) for 1f with the first triplet state is just 44 cm −1 above that.Such an underestimation is known for pure multireference HF, where the lack of electron correlation prevents quantitative prediction of the exchange couplings.Although the multireference PT methods can partially recover the correlation energy, the true correspondence of theoretical and experimental exchange coupling in the multireference framework can be achieved only by using the DDCI3 methods that involve excitations out of the active space (CAS). 105,106Considering a little expected improvement from NEVPT2 and related methods, as well as extraordinary computational resources required by the DDCI ones, we have limited our studies to a pure CAS(18,10) model of 1f.
The ten orbitals and their relative energies obtained by the SA-CAS(18,10) calculations for 1f are depicted in Figure 10.The orbitals, corresponding to the magnetic ones (Figure S9), are those having the highest energies and containing 73.4 and 71% d x 2 −y 2 atomic orbitals.The respective energy gap is 5707 cm −1 (Figure 10), being close to that predicted by the DFT/ ROKS calculations (9622 cm −1 ).While the first ferromagnetic state was always found have a single configuration in all cases (>97% of the [11] configuration for the highest orbitals, Figure 10), the antiferromagnetic ground spin state is always constructed of an approximately equal mixture of [20] and [02] determinants.This contrasts significantly with the UHF DFT broken symmetry model that describes this spin state as a single reference one, where unpaired electrons located on copper centers have opposite spins.The state-specific CAS-(18,10) calculation for singlet and triplet states of 1f revealed a similar multireference nature of the singlet state (Listing S4), where the ground state is a singlet and the singlet−triplet gap is 60.7 cm −1 .

CONCLUSIONS
In this study, we further explored a convenient self-assembly method for the generation of two new coordination polymers 1 and 2 in water.The obtained compounds are built from the dicopper(II)−aminoalcoholate blocks and the tetracyanonickelate(2−) linkers.Both coordination polymers were fully characterized by standard methods, and their structures were established by single-crystal X-ray diffraction.The latter revealed the formation of 1D chains in 1 and 2D metal− organic layers in 2.
The magnetic behavior of CPs 1 and 2 was investigated in detail, and the obtained results indicate that both compounds display strong antiferromagnetic coupling (J = −602.1 cm −1 for 1 and −151.0 cm −1 for 2) between Cu(II) ions within the dimers.The broken symmetry DFT calculations supported strong antiferromagnetic couplings for both CPs.A range of model fragments was studied, showing irrelevance of the predicted exchange on the positions of H atoms as well as the nature of the substituents.The CASSCF calculations confirmed the antiferromagnetic coupling within the {Cu(μ-O) 2 Cu} cores, where the singlet ground spin state has multireference character.Both DFT and CASSCF studies suggested that tetracyanonickelate blocks are not involved in the magnetic coupling between the copper centers nor act as superexchange bridges between dimeric blocks.
In summary, this work widens the growing family of cyanometalate-driven coordination polymers, showing that The Journal of Physical Chemistry C such compounds can be assembled from simple chemicals and by using self-assembly methods in water.Further research on extending the family of related heterodimetallic coordination polymers is currently underway in our laboratory.

Figure 3 .
Figure 3. Temperature dependence of the χ p T product at 0.5 T for 1 (squares) and 2 (circles).

Figure 4 .
Figure 4. Temperature dependence of the χ p product at 0.5 T for 2. The black solid line represents the fit according to the Bleaney− Bowers model (eq 2) with J = −151 cm −1 (see text).

Figure 5 .
Figure 5. Temperature dependence of the χ p T product for 1.The solid line represents a fit with the Bleaney−Bowers model (eq 2) with J = −602.1 cm −1 (see text).

Figure 6 .
Figure 6.Isosurfaces of the unrestricted corresponding magnetic orbitals for 1a obtained from B3LYP/ma-def2-TZVPP broken symmetry calculations.H atoms are omitted for clarity.The {Cu(μ-O) 2 Cu} core is aligned in the xy plane.

Figure 7 .
Figure 7. Isosurfaces of the unrestricted corresponding magnetic orbitals for 2a obtained from B3LYP/ma-def2-TZVPP broken symmetry calculations.H atoms are omitted for clarity.The {Cu(μ-O) 2 Cu} core is aligned in the xy plane.

g
Truncation of n-butyl groups to methyl ones.h Protons added instead of Ni(CN) 3 groups.i Methyl groups were added instead of Ni(CN) 3 ones with optimization of positions of all H atoms and the C atom of methyl groups.The Journal of Physical Chemistry C the J values calculated for the full crystallographic coordinates 1a and 2a most close to those determined experimentally.The possibility of [Ni(CN) 4 ] 2− block to transmit superexchange interactions was studied by constructing a model fragment {[Cu(H 2 bdea)(H 2 O)][Ni(CN) 4 ][Cu(H 2 bdea)-(H 2 O)]} 2+

Figure 8 .
Figure 8. Plots of the Laplacian of the electron density in the real space, ∇ 2 ρ(r), showing the O−H•••O hydrogen bonds between the uncoordinated water molecules and bridging oxygen atoms of the {Cu(μ-O) 2 Cu} core before (left, model 1a) and after (right, model 1b) optimization of H atoms' positions.

Figure 9 .
Figure 9. Left: plot of the Laplacian of the electron density in the real space, ∇ 2 ρ(r), showing the hydrogen bond between the uncoordinated water molecules and cyanido group after optimization of the positions of H atoms in 1b.Right: plot of the RDG function for the same projection.The color scheme corresponds to the sign(λ 2 )ρ(r) function with the second largest eigenvalue of the Hessian of electron density, λ 2 , at the respective points.

Figure 10 .
Figure 10.Top: isosurfaces of the molecular orbitals of the CAS(18,10) active space for 1f with the relative energies and populations (25 and 20 roots were considered for singlet and triplet states, respectively).H atoms are omitted for clarity.Left bottom: the diagram showing the relative energies of the above orbitals in the same order.Right bottom: largest configuration contributions to the lowest singlet and triplet states.The def2-TZVPP basis set was used for copper atoms and first coordination sphere (including the carbon atoms of Cu−CN groups) and the def2-SVP basis set for all other atoms.

Table 1 .
Crystal Data and Refinement Parameters for 1 and 2

Table 2 .
Experimental and DFT Calculated Exchange Coupling Constants a