Metal-Metal Interactions in Trinuclear Copper(II) Complexes [Cu3(RCOO)4(H2TEA)2] and Binuclear [Cu2(RCOO)2(H2TEA)2]. Syntheses and Combined Structural, Magnetic, High-Field Electron Paramagnetic Resonance, and Theoretical Studies

1 Metal-Metal Interactions in Trinuclear Copper(II) Complexes [Cu3(R-COO)4(H2TEA)2] and Binuclear [Cu2(RCOO)2(H2TEA)2]. Syntheses and Combined Structural, Magnetic, High-Field EPR and Theoretical Studies. Andrew Ozarowskia*, Carmen J. Calzadob*, Raj Pal Sharmac, Santosh Kumarc, Julia Jezierskad, Celestino Angelie , Federico Spizzog, Valeria Ferrettie,f* aNational High Magnetic Field Laboratory, Florida State University, Tallahassee, Florida 32310 United States bDepartamento de Química Física, Universidad de Sevilla. c/ Prof. García González, s/n. 41012- Sevilla. Spain. cDepartment of Chemistry, Panjab University, Chandigarh-160014, India dFaculty of Chemistry, Wroclaw University, 50-383 Wroclaw , 14 F. Joliot Curie Str., Poland. eDipartimento di Scienze Chimiche e Farmaceutiche and fCentro di Strutturistica Diffrattometrica, Università di Ferrara. Via Fossato di Mortara 17-27, I-44121, Ferrara, Italy. gDipartimento di Fisica e Scienze della Terra, Università di Ferrara, Via Saragat 1, I-44122 Ferrara, Italy * e-mail: [email protected] (AO); [email protected] (CJC); [email protected] (VF) 2 Abstract The trinuclear [Cu3(R-COO)4(H2TEA)2] copper(II) complexes, where RCOO- = 2-furoate (1), 2-methoxybenzoate (2) and 3-methoxybenzoate (3, 4), as well as dimeric species [Cu2(H2TEA)2(RCOO)2]∙2H2O, have been prepared by adding triethanolamine (H3TEA) at ambient conditions to hydrated Cu(R-COO)2 salts. The newly synthesized complexes have been characterized by elemental analyses, spectroscopic techniques (IR and UV/Visible), magnetic susceptibility, single crystal X-ray structure determination and theoretical calculations, using a DDCI (Dedicated Difference Configuration Interaction) approach for the evaluation of magnetic coupling constants. In the 2-furoate (1) and 2-methoxybenzoate (2) the central copper atom lies on an inversion center, while in the two polymorphic forms of 3- methoxybenzoate (3,4) derivatives the three metal centres are crystallographically independent. The zero-field splitting parameters (zfs) of the trimeric compounds, D and E, were derived from high-field, high-frequency EPR spectra at temperatures ranging from 3 K to 290 K and were used for the interpretation of the magnetic data. It was found that the dominant interaction between the terminal and central Cu sites J12 is ferromagnetic in nature in all complexes, even if differences have been found between the symmetrical or quasisymmetrical 1-3 complexes and non-symmetrical complex 4, while the interaction between the terminal centers J23 is negligible. Keywords: Trinuclear copper(II) complexes; Binuclear copper(II) complexes, Carboxylato bridges; Alkoxo bridges; Crystal structure; EPR; Magnetism; DDCI calculations 3 1. Introduction Materials exhibiting magnetic properties with potential technological applications are characterized by a close association of magnetic subunits able to communicate with each other. From a synthetic point of view, the use of simple bridging ligands (e.g., methoxide, hydroxide, alkoxide, oxide, carboxylate) to obtain transition-metal complexes has led to the formation of molecular clusters that in some cases behave as ‘single-molecule magnets’ 1 . Among the transition metals, copper is a good candidate for assembly of multinuclear clusters due to its variable coordination number and geometry, which can be conjugated with the versatility of the employed ligands. Actually, besides magnetism 2 multinuclear copper(II) complexes have proved to be of considerable interest in many modern research areas such as catalysis 3 , DNA cleavage 4 , oxygen activation 5 and host-guest equilibria 6 . The choice of the polyfunctional ligands allows a rational approach to obtain predetermined molecular architectures and in this context carboxylates have proved to be very efficient bridging ligands. Copper (II)-carboxylate complexes may be monomeric, dimeric or polymeric because of diverse coordination modes of carboxylate group. Dinuclear triply bridged copper(II) compounds, for example, can exhibit a great diversity of topologies, in which the coordination around the central metal countinuously changes from regular trigonal bipyramid to regular square-based pyramid 7 . It was found that the coordination mode affects the intramolecular magnetic exchange phenomena and consequently the final magnitude of the magnetic coupling constant, switching the magnetic behavior from ferro- to antiferro magnetic; furthermore, such structural property can be fine-tuned by the use of nitrogen containing non-carboxylic ligands. Among them, triethanolamine is in all respects a versatile ligand, that can be found in its neutral form, H3L, or mono-, H2L–, di-, HL2–, or trideprotonated, L3–. We have recently shown that ortho-, meta- and para-methoxybenzoate Cu(II) complexes react with triethanolamine to give dimeric complexes of general formula [Cu2(H2TEA)2(RCOO)2]∙2H2O 8 , showing ferromagnetic or antiferromagnetic behavior depending on the structural properties of the alkoxo bridge. That study was a part of our research on complexes of copper(II) arylcarboxylates with nitrogen-donor ligands 9 . While further exploring the reactions of Cu(II) carboxylates with triethanolamine, we isolated new trinuclear Cu(II) complexes with formula [Cu3(R-COO)4(H2TEA)2], where RCOO- is 2- furoate (1), 2-methoxybenzoate (2) and 3-methoxybenzoate (3, 4). Thus, in the cases of the ortho- and meta-methoxybenzoates, we have prepared both the dimeric and trinuclear complexes. 4 Scheme1- Ligands used in this study Similar binuclear and trinuclear compounds based on hetero triply-bridged di- and trinuclear,8, 10 , 11 , 12 , 13 Cu(II) units with carboxylato and hydroxo (or alkoxo) bridges have been reported previously, most of them presenting ferromagnetic behavior. While the binuclear Cu(II) bis(-hydroxo) and bis(-alkoxo) show a clear correlation between structure and magnetism (correlation between J and the Cu-O-Cu bonding angle and the Cu-OR-Cu torsion angle) 14 , the rationalization of the magnetic behavior in systems containing hetero bridges is more difficult and usually based on the concept of the complementarity / countercomplementarity between the ligand and metal orbitals, concept introduced by Nishida et al. 15 and McKee et al. 16 to explain the effect of different bridging ligands on the magnetic coupling. The present work is aimed at synthesis, characterization of these new copper(II) triethanolamine complexes and studying their magnetic properties by a multifaceted approach combining X-ray structure determination, magnetic susceptibility measurements, high-field, high-frequency EPR spectroscopy and theoretical calculations. The results of these different methodologies provide information about the Heisenberg and zero-field splitting parameters, and insights into the main mechanisms involved in the magnetic coupling interaction inside a complex. To the best of our knowledge, this is the first time that the magnetic properties of trimeric Cu(II) complexes have been explored in fine details using HF EPR spectroscopy. Additionally, the high field EPR spectra of the corresponding dinuclear compounds8 have been recorded and their description completed. furan-2-carboxylate m-methoxybenzoate H3TEA 5 2. Experimental 2.1 Materials and instruments Analytical grade reagents were used throughout this work without any further purification. Carbon, Hydrogen and Nitrogen were determined micro-analytically by automatic Perkin Elmer 2400 CHN-elemental analyzer and copper was determined by standard literature methods 17 using the Vario EL III Element Analyzer, (Elementar Analysensysteme GmbH) for C, H, N and (ARL)-Model 3410-ICP Spectrometer for Cu. FT-IR spectrum of compound 1 was recorded as KBr pellet on PERKIN ELMER SPECTRUM RXFT-IR system (India). Infrared (IR) spectra of compounds 2-4 were recorded over the range 400–4000 cm–1in KBr pellets using Bruker 113v spectrophotometer. The NIR-Vis-UV diffuse-reflectance electronic spectra of compounds 1 - 4 as well as of the dimeric complexes were recorded on a Cary 500 Scan spectrophotometer with measure step of 10 cm-1 in the 5000-50000 cm-1 spectral range at room temperature. 2.2 Synthesis Complex 1, [Cu3(2-furoate)4(H2TEA)2]. Sodium salt of 2-furoic acid was prepared in situ by dissolving 0.16 g (4 mmol ) of NaOH and 0.44 g (4 mmol) of 2-furoic acid in minimum amount of water (Eq.(R1)). 0.50 g (2 mmol) of hydrated copper sulphate, CuSO4.5H2O was dissolved in 10 mL of distilled water. On mixing the two solutions, copper(II) furanoate precipitated immediately, as shown by Eq. (R2). The precipitated product was filtered, washed with water followed by methanol and dried at room temperature (yield 85%). Triethanolamine was added dropwise to the suspension of copper(II) furoate in 20 mL of a methanol-water mixture (4:1 v/v). The addition was continued till a clear greenish blue colored solution was obtained. When the solution was allowed to evaporate slowly at room temperature, blue colored shiny crystals appeared after few days, in accordance with Eq. (R3), which were separated from the mother liquor and dried in air. 6 Other three trinuclear complexes, 2, 3 and 4 were prepared in a procedure in which previously reported analogous dimeric complexes were obtained.8 In the case of the ortho- and metamethoxybenzoates, first the trinuclear species precipitated from the mother liquor and subsequently, in some preparations the dimeric complexes crystallized in the filtrate. Very pure samples of either dimeric or trinuclear species could be obtained. Raw 2 was recrystallized from MeCN, while 3 and 4 were obtained by recrystallization of the corresponding crude product from methanolic solutions. 3 can also be prepared by recrystallization of 4 from hot methanol, but no good crystals could be obtained in this way. The unpredictability of the synthesis and formation of both tri- and binuclear complexes in the same reaction mixture was also reported by others.12b,13 It seems that the formation of either dinuclear or trinuclear products is determined by the conditions in the solution when the crystallization starts and the outcome is difficult to control. Once a dimeric species has been formed, it is not possible to recrystallize it into a trinuclear species, and no opposite process occurs. Possibility of the EPR identification of the products and purity assessment immediately after the synthesis was a big advantage in this work. No trinuclear paramethoxybenzoate and no pure dimeric furanoate could be obtained. It is thus clear that the reaction may occur in two ways. Interestingly, in all cases in which the trinuclear complexes were obtained in this paper, the raw intermediate carboxylate species were proven by EPR to be high-purity dimeric "paddlewheel" complexes [CuL2]2(H2O)2, similar to copper acetate monohydrate and showing typical EPR parameters (Table S3, Figs S6-S7). In the p-methoxybenzoic acid system, where only a dimer [Cu2(H2TEA)2(RCOO)2]∙2H2O was obtained, the intermediate was a mixture of two slightly different monomeric species (Fig S7). All complexes are soluble in methanol and in acetonitrile, but insoluble in water. Complex 1, [Cu3(2-furoate)4(H2TEA)2] decomposes at 185oC. FT IR (KBr) (νmax, cm-1): 3140(w), 2966(w), 2881(w), 1599(s), 1561(m), 1416(m), 1365(s), 1224(s), 1197(m), 1137(m), 1087(m), 1006(s), 901(m), 779(s), 621(s), 526(w), 487(s). Anal. calcd. for 7 C32H40Cu3N2O18 (%): C, 41.24; H, 4.29; N, 3.00; Cu, 20.46; Found: C, 41.05; H, 4.58; N, 2.92.1; Cu, 20.12. Complex 2, [Cu3(o-methoxybenzoato)4(H2TEA)2]∙2(C2H3N). FT IR (KBr) (νmax, cm-1): 3388(m, broad), 2929(m), 2881(m), 1602(s), 1564(s), 1409(s), 1382(s), 1250(s), 1084(m), 1026(m), 897(m), 831(w), 756(m). Anal calcd. for C48H62O18N4Cu3: (%) C 49.12, H5.32, N 4.77, Cu 16.24. Found: C 49.0. H 5.4, N 4.80, Cu 16.15. Complex 3, [Cu3(m-methoxybenzoato)4(H2TEA)2]. FT IR (KBr) (νmax, cm-1): 3223(m, broad), 2929(m), 1558(s), 1450(s), 1392(s), 1252(s), 1093(m), 1037(s), 908(m), 762(s). Anal. Calcd for C44H56O18N2Cu3: (%) C 48.41, H5.17, N 2.57, Cu 17.46. Found: C 48.3, H 5.22, N 2.49, Cu 17.55. Complex 4, [Cu3(m-methoxybenzoato)4(H2TEA)2]. FT IR (KBr) (νmax, cm-1): 3409(m, broad), 2941(m), 1603(m), 1554(s), 1452(m), 1390(s), 1248(m), 1286(w), 1083(m), 1045(m), 906(w), 773(m). Anal. Calcd. same as 3. Found (%) C 48.2, H 5.0, N 2.47, Cu 17.40 2.3 X-ray crystallography Single-crystal diffraction data for 1-4 were collected on a Nonius Kappa diffractometer equipped with a CCD detector with graphite-monochromatized MoKα radiation (λ = 0.71069 Å). Intensities were corrected for Lorentz, polarization and absorption 18 effects. The structures were solved by direct methods with the SIR97 suite of programs 19 and refinement were performed on F2 by full-matrix least-squares methods with all non-hydrogen atoms anisotropic. Hydrogen atoms of TEA ligands in 1 and of OH type in 2-4 were found in the difference Fourier map and refined isotropically; all other hydrogens were included on calculated positions, riding on their carrier atoms. All calculations were performed using SHELXL-97 20 implemented in the WINGX 21 system of programs 2.4 EPR and magnetic susceptibility Magnetic susceptibility measurements over the temperature range 1.8-300 K were performed at a magnetic field of 0.5 T using a Quantum Design SQUID MPMSXL-5 magnetometer. Correction for the sample holder, as well as the diamagnetic correction χD, which was estimated from the Pascal constants, 22 was applied. High-field, high-frequency EPR spectra at temperatures ranging from ca. 3 K to 290 K were recorded on a home-built spectrometer at the EMR facility of the NHMFL 23 with the microwave frequencies 52-416 GHz. The instrument is a transmission-type device and uses 8 no resonance cavity. The microwaves were generated by a phase-locked Virginia Diodes source, generating frequency of 13 ± 1 GHz, and equipped with a cascade of frequency multipliers to generate higher harmonic frequencies. A superconducting magnet (Oxford Instruments) capable of reaching a field of 17 T was employed. Additionally, X-band and Q- Band spectra were recorded on a Bruker ElexSys E500 instrument equipped with an NMR teslameter ER 036TM and a frequency counter E 41 FC (Faculty of Chemistry, Wroclaw University). 9 3. Results and Discussion 3.1. X-ray diffraction The ORTEPIII 24 views of complexes 1-4 and coordination polyhedra around each Cu atom are shown in Figs 1-4, respectively. Experimental details are given in Table 1 and the geometrical parameters for the most important interactions are reported in Table 2. Figure 1. ORTEP view and atom numbering scheme for [Cu3(furan-2-carboxylato)4 (H2TEA)2] (1) Thermal ellipsoids are drawn at the 40% probability level. Hydrogen bonds are drawn as dashed lines. C-H hydrogen atoms are not shown for clarity. Central and terminal Cu coordination polyhedra are also shown. 16 Table 1. Crystal data for trimeric copper(II) complexes 1 2 3 4 Chemical formula C32H40Cu3N2O18 C44H56Cu3N2O18. 2(C2H3N) C44 H56Cu3N2O18 C44 H56Cu3N2O18 Mr 931.28 1173.67 1091.56 1091.56 Space Group Triclinic, P-1 Monoclinic, P21/c Triclinic, P-1 Monoclinic, P21/c Z 1 2 2 4 a, b, c (Å) 8.5790 (2), 9.4128 (2), 11.6658 (3) 11.6469(2), 15.8467(3), 14.0542(3) 8.44460(10), 16.8601(2), 16.9374(3) 13.4231(1), 13.4250(1) 26.0803(3) α, β, γ (º) 97.114 (1), 99.975 (1), 92.867 (1) 90, 93.566(1), 90 89.0670(6), 81.1580(7) 82.6680(6) 90, 98.1420(4), 90 V (Å3) 918.19 (4) 2588.89(9) 2363.34(6) 4652.43(7) Dc/g cm-3 1.684 1.506 1.534 1.558 µ (mm−1) 1.80 1.30 1.41 1.44 Crystal size (mm) 0.35 × 0.23 × 0.17 0.55 x 0.29 x 0.20 0.20 x 0.15 x 0.09 0.47 x 0.12 x 0.10 No. of measured, independent and observed [I > 2σ(I)] reflections 13956, 4435, 3521 19508, 7547, 5642 25241, 11356, 8336 40773, 11114, 8061 Rint 0.045 0.049 0.036 0.062 R[F2 > 2σ(F2)], wR(F2), S 0.036, 0.108, 1.06 0.044, 0.130, 1.07 0.047, 0.119, 1.09 0.041, 0.102, 1.06 No. of reflections 4435 7547 11356 11114 No. of parameters 318 335 620 620 Δρmax, Δρmin (e Å−3) 1.07, −0.76 0.40, -1.08 0.70, -0.70 0.71, -0.87 17 Table 2. Structural parameters (Å , °) for hydrogen bonds D-H D...A H...A D-H......A 1 O2-H...O8 0.84(3) 2.665(4) 1.83(3) 172(3) O3-H...O1i 0.74(4) 2.818(2) 2.07(4) 174(5) Equivalent positions: (i) 2-x,1-y,-z 2 O2-H...O8 0.84(3) 2.660(3) 1.84(3) 162(3) O3-H...O7i 0.86 2.918(2) 2.09 161 Equivalent positions: (i) -x-1,-y,-z; 3 O2-H….O8 0.80(5) 2.693(4) 1.92(5) 163(4) O11-H...O17 0.87(3) 2.650(4) 1.80(3) 166(2) O3-H...O10i 0.85(3) 2.768(3) 1.97(3) 155(3) O12-H...O1ii 0.84(3) 2.754(3) 1.99(4) 150(3) Equivalent positions: (i) x+1,y,z; (ii) x-1,y,z 4 O2-H...O8 0.87(3) 2.609(3) 1.77(3) 164(3) O11-H...O17 0.89(4) 2.637(3) 1.77(4) 163(4) O3-H...O7i 0.84(2) 2.864(3) 2.05(2) 164(2) O12-H...O10ii 0.84(6) 2.908(3) 2.07(6) 170(5) Equivalent positions: (i) 2-x,-y,2-z; (ii) 1-x,y-1/2,3/2-z 18 Table 3. Geometrical parameters of the Cu-O-Cu bridge and fitted J12 values of TEA-Cu(II) trinuclear complexes. CSD code J12 /cm -1 Cu-O-Cu angle carboxylate ligand Ref. AWEQEZ 66 75.92 /105.17 thiophene carboxylate 12 AWEQID 66 73.91 / 108.40 73.05 / 107.76 thiophenyl acetate 12 AWEQUP 66 75.69 / 103.69 thiophenyl acetate (solvate) 12 FISJIB 84 76.37 / 104.33 Acetate 11 MEDHUZ 65 73.25 / 109.27 acetate (solvate) 13 MEDJAH 88.9 76.02 / 105.25 formate (solvate) 13 MEDJEL 70.8 77.23 / 105.89 Trifluoroacetate 13 1 77.1 75.37/106.05 furan-2-carboxylate This work 2 66.4 72.92/108.43 o-methoxy-benzoate This work 3 59.5 73.76/106.46 74.03/107.61 m-methoxy-benzoate This work 4 51.0 71.32/108.05 70.95/104.06 m-methoxy-benzoate This work 19 3.2. FT-IR Spectroscopy The infrared spectra of newly synthesized complex salts have been recorded in the region 400- 4000 cm-1 and tentative bands assignments have been made on the basis of earlier literature reports 25 . The spectra of the four compounds, shown in Fig. S9 of SI, are very similar. The broad bands in 3140-3409 cm-1 region were ascribed to ν(O-H) of triethanolamine. The band at 3140 cm-1 in complex 1 appears to be considerably shifted to lower frequency, possibly due to the involvement of O-H groups in strong hydrogen bonding. The absorption bands around 2900- 2800 cm-1 were assigned to asymmetric and symmetric C-H stretch of furoic/benzoic acids. The bands in the region 1603-1599 cm-1 were attributed to ν(C=C) of the furanoic/benzoate moiety. The splitting of bands of as(COO-) and s(COO-) indicated the presence of two types of arylcarboxylate ligands. The bands at 1599/1365 cm-1 in 1, 1562/1382 cm-1 in 2, 1558/1392 cm-1 in 3 and 1554/1390 cm-1 in 4 were assigned to as(COO) , s(COO) respectively. The Δν=as(COO-) -s(COO-) separation of 234, 180, 166 and 164 cm-1 in 1-4, respectively, indicates the presence of bidentate or bidentate bridging coordination mode of arylcarboxylate, which was finally confirmed by single crystal X-ray structure determination. The strong to medium bands in the region 1250-1000 cm-1 may be attributed to C-N, C-O strech of the triethanolamine ligand. 26 Finally, the IR bands in the 500-400 cm-1 region are due to the M-O and M-N stretching vibrations. 3.3. Magnetic susceptibility and HF EPR spectra Dimeric complexes. The magnetic susceptibility data of dimeric complexes [Cu2(H2TEA)2(RCOO)2]∙2H2O derived from the ortho-, meta- and paramethoxybenzoic acid have been reported previously.8 The copper ion environment in these dimeric complexes is similar to the environment of the terminal copper ions in the trinuclear systems which are the main subject of the present work. Their high-field EPR spectra were therefore recorded (Fig. 7) to help in interpretation of spectra of the trinuclear complexes. The m-methoxy and p-methoxy dimers are ferromagnetic with the exchange integrals of 101 cm-1, while the o-OCH3 dimer is 20 antiferromagnetic with J = -83 cm-1. These data refer to the Heisenberg-Dirac-VanVleck Hamiltonian in a form: Ĥ= - J Ŝ1• Ŝ2 (1) The differences in magnetic behavior were discussed in ref 8. The EPR spectra were interpreted in terms of the standard spin Hamiltonian expressed in the total spin S=1 of the system: Ĥ=  BB{g}Ŝ + D{Ŝz2 – S(S+1)/3} + E(Ŝx2 – Ŝy2) (2) Instead of using Hamiltonian (2), which is often referred to as the "giant spin Hamiltonian", one can use the spin operators of separate ions. Ĥ=  BB{g1}Ŝ1 +  BB{g2}Ŝ2 +D12{Ŝ1z Ŝ2z – Ŝ1 Ŝ2/3} + E12(Ŝ1x Ŝ2x – Ŝ1y Ŝ2y) (3) The relations between the parameters of these two spin Hamiltonians are (for dimers of S=1/2 ions only): 27 D12 = 2 D, E12 = 2 E (4) It will be shown in the following text that the {g} matrices in the dinuclear and trinuclear systems studied here were found to be non-coaxial with the zero-field splitting tensors. While the tensor quantities had to be used in calculations, the well-known scalar parameters gx, gy, gz, D and E are reported in the tables of results and an explanation of the notation used is necessary. The zero-field splitting part (terms with D and E) of eq 2 can also be expressed by using the zero-field splitting tensor {D}: Ĥzfs=Ŝ {D} Ŝ (5) Analogously, the zfs terms in (3) may be replaced by Ĥzfs=Ŝ1 {D12} Ŝ2 (6) The relation between {D12} and {D} is similar to eq 4 (again, in a dimer of S=1/2 ions only): {D12} = 2 {D} (7) In a system of coordinates in which the {D} tensor is diagonal, the relations between the diagonal tensor elements Dxx etc. and the scalar parameters D and E are D = (2Dzz – Dxx – Dyy) / 2 and E = (Dxx – Dyy)/2 (8) Formulas (8) relate in the same way either the diagonal elements of {D} to D, E, or the diagonal elements of {D12} to D12, E12. Because all our dimers are centrosymmetric, the g components of the "giant spin" Hamiltonian (2) are equal to the g components of the individual ions. The latter 21 is not valid for the trinuclear systems (see below), where extracting the g components of individual ions is much more challenging. Figure 7. Powder EPR spectra of the dimeric orthometa- and para-methoxybenzoato complexes [Cu2(H2TEA)2(RCOO)2]∙2H2O recorded at 30 K with the microwave frequency 328.8 GHz. The red traces are simulated with parameters from Table 4 and a 11 deg misalignment between the g matrix and the zfs tensor. Labels x, y, z indicate the molecular orientations at which transitions occur. Table 4. Experimental Spin Hamiltonian parameters and estimated anisotropic exchange contributions to the zero-field splitting, Dex, and Eex, for the dinuclear complexes [Cu2(H2TEA)2(RCOO)2]∙2H2O R J, cm-1 gx = gy gz D, cm-1 E, cm-1 Dex, cm-1 a Eex, cm-1 a m-OCH3C6H4 101 b 2.064 2.311 -1.066 -0.0393 -1.13 0.01 o-OCH3C6H4 -83 b 2.064 2.291 -0.907 -0.0495 -0.97 0 p-OCH3C6H4 101 b 2.063 2.318 -1.140 -0.0393 -1.20 0.01 C4H3Oc - 2.060 2.305 -1.075 -0.0374 -1.14 0.01 NO2C6H4 100 d 2.060 d 2.303 d -1.135 d -0.0430 d -1.20 0.01 aDex and Eex were obtained by subtracting Ddip = 0.06 cm-1 and Edip= -0.05 cm-1 from the experimental D and E (see below). bRef. 8. cThis work, observed as a contamination in a trinuclear complex sample. d Ref. 28 22 Table 4 lists the g components and zero-field splitting parameters D and E for the dimeric complexes together with the exchange integrals. Parameters of a dimer derived from nitrobenzoic acid and TEA 28 are also given for comparison. Although no pure dimer could be obtained in the furanoic acid system, it was observed as a contamination in a less successful preparation of the corresponding trinuclear complex (Fig. S1 in SI) and its EPR parameters were determined. High-Field EPR offers a possibility of determination of the sign of the zero-field splitting parameters as it affects the intensity pattern in the low temperature spectra.8, 29 Negative D was found for all dimeric complexes studied in this work. The g and zfs parameters of the ferromagnetic dimers in Table 4 span a rather narrow range and are also not much different for the one antiferromagnetic dimer. Interestingly, for a successful EPR simulation, 11 deg rotation of the g matrix about the y axis of the zfs tensor must be assumed for each dimer in Table 4. With this rotation, the spectra at any microwave frequency can be simulated with the same set of the g, D and E parameters. EPR Spectra and Magnetic Susceptibility of Trinuclear Complexes Spectra of the three centrosymmetric trimers 1, 2 and 3, as well as of AWEQEZ12 (spectrum shown in Fig. S2 of SI) are qualitatively similar, but show considerable variations in the zerofield splitting parameters (Table 5). All exhibit small E/D ratio, opposite to the asymmetric complex 4, for which E/D is close to 1/3 (see below). Spectra of complexes 3 and 4, which are both meta-methoxybenzoates, are compared in Fig. 8. Figure 9 shows the thermal dependence of the effective magnetic moment for compounds 1-4. The effective magnetic moment at room temperature is ca. 3.4 BM, which is slightly larger than expected for three isolated Cu(II) ions, (expected (3/2)gave, thus some 3.2 BM with a reasonable gaverage of 2.15). Upon temperature lowering, the magnetic moment increases to a maximum of 4.12 - 4.18 BM corresponding to full occupation of the coupled S=3/2 state - expected gave√(15/4) = 4.16. The interactions are thus of ferromagnetic character. At the lowest temperatures, the magnetic moment falls slightly due to the splitting of the ground S=3/2 state by the Zeeman and zero-field splitting interactions. The magnetic moment is slightly lower for 4 than for other ones (Fig. 9). Since the parameters 23 derived from EPR were used to interpret the magnetic data, the EPR spectra will be discussed first. Figure 8. HF EPR spectra of the symmetric complex 3 and the non-symmetric complex 4 recorded at 30 K with the microwave frequency 208.00 GHz. The magnified traces show the ‘forbidden’ MS = 2 and MS = 3 transitions appearing at roughly ½ and 1/3 of the normal resonance field, respectively. Labels X, Y and Z indicate the molecular orientations at which the respective ‘allowed’ transitions occur. Table 5. The coupled-spin Hamiltonian parameters for the trinuclear complexes Complex J12, J23 cm-1 a gxb gyb gzb D,b cm-1 E,b cm-1  ,  ,  ,c deg 1 77.1(5), -1.2(4) 2.236 2.145 2.070 0.686 -0.014 9, 40, -25 2 66.4(6), -6.8(4) 2.225 2.123 2.065 0.977 0.035 6, 35, -19 3 59.5(6), -6.8(4) 2.222 2.124 2.058 0.724 0.039 9, 37, -18 4 51(1), -4.7(6) - 2.15d 2.09 d 0.64 d 0.23 d - AWEQEZ e 66, 0e 2.228 2.133 2.067 0.710 -0.033 8, 40, -23 a J values refer to the spin Hamiltonian (9). b g, D and E values refer to the coupled-spin Hamiltonian (2) with S=3/2. c Main axes of the coupled {g} matrix were obtained by consecutive rotations of the main axes of the zfs tensor by angles  ,  and  , about the x, y and z axes, respectively. dPowder EPR spectra could not be simulated, parameters determined from the frequency dependencies of the resonance positions (Fig. S4). eJ value taken from Ref 12. (HF EPR spectrum shown in Fig S2). 0 1 2 3 4 5 6 7 8 9 Y Y Z Z Z 3 MS = 3 X 208 GHz, 30 K Y X Y Z MS = 2 4 Magnetic Induction, Tesla 24 Figure 9. The magnetic properties of complexes 1, 2, 3 and 4. Circles are the experimental data and red lines are calculated with the J12 and J23 parameters from Table 5. The D12 = D13, E12 = E13 parameters of spin Hamiltonian (9) were obtained by multiplying the “giant spin” Hamiltonian parameters D and E, as found from EPR (Table 5) by 3. In the magnetic data fitting, gaverage was allowed to vary and values of 2.136, 2.163, 2.167 and 2.085 were obtained for 1, 2, 3, and 4, respectively. No intermolecular interactions needed to be taken into account in the cases of 1, 2, and 3. For 4, the fit was significantly improved with the zJ parameter of 0.037 cm-1. Spin Hamiltonian parameters for the centrosymmetric trinuclear complexes. Out of our four trinuclear complexes, two (1 and 2) possess the inversion center and a third one (3) is very close to having one and will be also treated as centrosymmetric. The terminal copper ions (2 and 3) in the centrosymmetric systems are equivalent, the corresponding main axes of their {g} matrices are parallel and the corresponding g components of ions 2 and 3 are equal, like g2x = g3x etc. Also, both the isotropic and anisotropic interactions 1-2 and 1-3 must be equal. The spin Hamiltonian, which can be used to interpret both the magnetic and EPR data is27 3.4 3.6 3.8 4.0 4.2 2 1 050 100 150 200 250 300 3.4 3.6 3.8 4.0 Effective Magnetic Moment, B.M. Temperature, K 4 3 25 Ĥ = - J12(Ŝ1 Ŝ2 + Ŝ1 Ŝ3) – J23Ŝ2 Ŝ3 + + Ŝ1{D12} Ŝ2 + Ŝ1{D13} Ŝ3 + Ŝ2{D23} Ŝ3 (9) + BB{g1} Ŝ1 + BB{g2} Ŝ2 + BB{g3} Ŝ3 where {gi} represent the g matrices and {Dij} are the zero-field splitting tensors for the appropriate pairs. The first line in eq 9 is the Heisenberg Hamiltonian that describes the isotropic part of the metalmetal interactions. The eigenstates of this Hamiltonian are two doublet states D1 and D2 with ST=1/2 and a quartet state Q, (ST=3/2). ST is the total spin of the system, ŜT = Ŝ1 + Ŝ2 + Ŝ3. The relative energies can be expressed by J12 and J23 as: E (D1) = 3J12/2 E (D2) = J12/2 + J23 (10) E (Q) = 0 These energies determine the bulk magnetic properties. For EPR, on the other hand, the anisotropic metal-metal interactions and the Zeeman term (lines 2 and 3 in eq 9, respectively) are decisive. Figure 10. Experimental (blue) EPR spectra of 1 recorded at 30 K with the microwave frequencies as marked. Red: spectra simulated using spin Hamiltonian (2) with S=3/2 and parameters given in Table 5. The magnified traces in the top spectrum show the “forbidden” transitions, MS=2 and 3 which appear at roughly 1/2 and 1/3 of the magnetic induction in the center of the main spectrum, respectively. 32 through the central copper ion. A spectrum similar to that of 4 was obtained (Fig. S5), exhibiting a larger  of some 2.0 cm-1 and a high E/D ratio. Similarly as in the case of 4, no full EPR simulation was possible despite of the higher symmetry. Comparison to the literature EPR and magnetic data of similar systems The linear trinuclear complexes similar to ours were reported in many papers.10-13 While X-band EPR (~9.5 GHz) has mainly been applied, we found some papers reporting the HF EPR spectra.10a,b Surprisingly, no determination of the zero-field splitting parameters from HF EPR data was attempted. In X-Band EPR typically a resonance close to the effective g = 4.3 value is observed, accompanied by another one close to geff of 2.1.The former one is due to transitions within the |±1/2> Kramers doublet of the S=3/2 state, at the “perpendicular” orientation. The appearance of that feature indicates that the splitting between the Kramers doublets of the S=3/2 state is larger than the microwave quantum energy. The geff = 2.1 signal is a combination of transitions within the |±1/2> Kramers doublet of S=3/2 at the “parallel” orientation and those occurring within the excited S=1/2 state(s). In some cases, the g=4.3 signal was observed to shift towards the g=2.1 position when temperature was increased from 4.2 K to 80 K and above, which was explained by coalescence of the resonances due to the quartet and doublet states10a,e,13. In the X-Band spectra of our trinuclear complex 1, those g=4.3 and g=2.1 signals are observed (Figure 15) , but do not coalesce at higher temperatures, similarly as was observed in the very closely structurally related thiophenylcarboxylate-TEA trimer,12 meaning that there is no fast relaxation between the ground state and the excited doublet states. No coalescence effects are expected in our high-field spectra in which transitions between the |±3/2> and |±1/2> Kramers doublets are observed, allowing accurate determination of the D and E parameters. We have synthesized for comparative purposes complex [Cu3(thiophenyl-COO)4(H2TEA)2] (AWEQEZ) for which D = 0.83 cm-1 and gaverage=2.16 were determined from magnetic data.12 That compound is isostructural with 1. Analysis of its HF EPR spectra in this work (Fig. S2) led to D = 0.710 cm- 1, E= -0.033 cm-1 and also allowed determination of the relative orientation of the {D} tensor versus the {g} matrix (Table 5). 33 Figure 15. X-Band EPR spectrum of 1, recorded at 80 K. The spin Hamiltonian [9] parameters determined from the high-field spectra recorded at 30 K were used in the simulation. Compare to the high-field spectra of 1 (Figs 10 and 14). The magnetic properties of many trinuclear copper complexes similar to ours have been reported.10-13 In cases where only the J12 interaction has been considered, its magnitude has been found to be around 60-90 cm-1.11,12,13 When allowing non-zero J23 (terminal to terminal interaction) in the fitting procedures, it has always been found antiferromagnetic,10a,d-f in agreement with our experimental results. Sometimes, the magnitude of J23 was a surprisingly large fraction of J12, like 20%.10e In our complexes, attempts of forcing in the magnetic fitting J23 to positive values (ferromagnetic), or just to zero, caused a visible deterioration of the average deviation of the calculated magnetic data from the experimental ones. Moreover, the deviations took a systematic character being positive at low temperatures and negative at higher. This effect could not be compensated by assuming intermolecular interactions. The experimental results, both ours and those in the literature, thus hint at the antiferromagnetic character of J23, although the intuition and the theoretical calculations presented below indicate otherwise. The nature of the zero-field splitting In polynuclear copper(II) complexes there are two contributions to the zero-field splitting, one of which is caused by the magnetic dipole-dipole interaction and the other one is due to the anisotropic exchange interactions:27-29 D = Dex + Ddip, E = Eex + Edip (16) The theory of the latter has been in progress recently. 32 To estimate the exchange contribution, the dipolar part must be subtracted from the experimental zfs parameters (see last 2 columns of 0.2 0.4 0.6 0.8 1.0 Simulated, 80 K Experimental, 80 K Magnetic Induction, Tesla 34 Table 4). The largest component of the dipole-dipole interaction tensor is along the Cu-Cu direction, while the largest component of the exchange-related zfs interaction is expected to be parallel to the direction of gz,27-32 which in our dimeric complexes is roughly perpendicular to the Cu-Cu vector. When that direction is labeled "Z" and that of the C-Cu vector is "X" then the expressions for Ddip and Edip (referred to the giant spin Hamiltonian) are 33 Ddip = (2gz2 + 2gx2 - gy2)  B2/4R3Cu-Cu (17) Edip = -(2gx2 + gy2)  B2/4R3Cu-Cu These formulas are approximate, as they do not take into account the electron delocalization. A strong effect of delocalization was observed in a biradical system, in which the point dipole Ddip was too small compared to the experiment, while the value obtained by using the wavefunction was correct. 34 The effect was less pronounced for the dimeric copper acetate, where the pointdipole Ddip = -0.17 cm-1 was too large29 compared to the quantum-mechanical value of -0.12 cm- 1.32a Because of the approximate character of eq 17 and close similarity of both the structures and the g parameters of all our dimeric complexes, we will use for all of them the same Ddip = 0.06 cm-1 and Edip = -0.05 cm-1, which result from eq 17 with RCu-Cu of 2.92 Ǻ. It is thus seen that the dipolar contribution is minor compared to the overall D of -1 to -1.2 cm-1 (Table 4), and that the zero-field splitting related to the anisotropic exchange is almost axial (Eex << Dex). The exchange interaction anisotropy is caused by mixing of the excited states of a system to its ground state via the spin-orbit coupling.27,32 An excited state can only contribute to zfs if a non-zero matrix element of the L operator exists between that excited state and the ground state.27,32 The excited states of a pair to be considered in our complexes are those in which one of the copper ions is in its ground state dx2-y2 while another one is in an excited state – dxy, dxz or dyz. In the dimeric copper(II) complexes with the dx2-y2 ground state, Dex and Eex can be expressed as32a 2, , 2 2, , 2 2, , 2 2, , 2 2, , 2 22 22 22 22 22 22 22 22 22 22 4 1 4 1 4 1 4 1 2 yzyx yzyx xzyx xzyx ex yzyx yzyx xzyx xzyx xyyx xyyx ex E J E J E E J E J E J D                       (18) where the symbols Jx2-y2,n are the exchange integrals between the dx2-y2 ground orbital of one copper ion and an excited dn orbital of another, and  = -828 cm-1 is the spin-orbit coupling constant for Cu2+. The exchange integrals Jx2-y2,xz and Jx2-y2,yz are expected to be similar, as may 35 be inferred from Fig. 16 which shows the arrangement of these orbitals. Also, /Ex2-y2,xz and /Ex2-y2,yz must be equal because gx and gy are equal (eq 14). Accordingly, the Eex parameter is expected to be small (eq 18) in agreement with the data in Table 4. The exchange-related zfs interaction is thus axial (E << D) and the dipole-dipole interaction is axial as well, but because they operate in different systems of coordinates, their sum is not axial. The experimental E for dimers of this type cannot thus be equal to zero (Table 4). The situation is very different in the well-known dimeric 'paddlewheel' copper carboxylates, in which both the exchange-related and dipole-dipole zero-field splitting have their largest component along the Cu-Cu direction resulting in very small E values.29 (See also Table S3). On the contrary, very large E values, comparable to D were observed in dinuclear complexes containing copper ions in trigonal bipyramide environment and dz2 ground state. This was explained by the difference between the exchange integrals Jz2,xz and Jz2,yz which determine Dex and Eex in those compounds. 35 Figure 16. The orbital arrangement in the bridge unit of the dimeric complexes. Interactions between the dx2-y2 orbital of one ion and dxz of another (left) is expected to be similar to the interaction dx2-y2 – dyz (right). The Z axis (direction of gz) is perpendicular to the bridge plane. The situation in the trinuclear systems is much more difficult. The g matrix axes of the central copper ion are not parallel to the g axes of the terminal atoms, complicating calculations of the dipole-dipole related zfs. In view of the approximate character of the point-dipole model, this is not a major problem. The Ddip for interaction 1-2 calculated along the Cu1-Cu2 direction (referred to spin Hamiltonian 9) is -0.18 cm-1, and together with interaction 1-3 it will contribute just -0.06 cm-1 to the giant-spin D parameters in Table 5 (see also eq 12). Again, this is a minor contribution, and similarly as in the dimers, the anisotropic exchange must be responsible for most of the zfs. The exchange-related zero-field splitting in pairs Cu1-Cu2 and Cu1-Cu3 could be presumably analyzed in the same way as in the copper dimers.32 To our knowledge, theoretical calculations like those in ref 32a have not been attempted so far for any trinuclear system. 36 3.4. Theoretical calculations of the isotropic exchange interaction Calculations of the exchange interactions in our dimers have already been published8 and the following discussion will refer to the trinuclear complexes only. The trimers 1, 2 and 3 can be formally considered as a linear chain, where each Cu atom presents a square-planar coordination, CuO3N for terminal metal centers and CuO4 for the central Cu (Figure 17). In the analysis of the magnetic orbitals, this plane is considered the xy plane in a local axis system with each Cu at the origin and the x and y axis oriented along the Cu-O(N) bonds. The corresponding CuO3N and CuO4 planes form an angle of about 70º in 1, 64o in 2, and 66.8º, 67.3º in 3 (no exact inversion center). In each case, two protonated O atoms from the TEA molecule complete the metal coordination sphere as axial ligands, except for the central Cu of compound 4, with only one TEA protonated O atom in axial position with square-planar pyramid coordination. Each Cu atom bears one unpaired electron, placed on an in-plane Cu 3 dx2-y2 orbital. Figure 18 shows the active orbitals obtained from a CASSCF(3/3) calculation on the quartet ground state for the furan derivative 1, similar shapes are obtained for complexes 2 and 3. The two terminal CuO3N planes form a dihedral angle of 180º in 1-3, and 60º in 4. The Cu 3d magnetic orbitals on two neighboring centers are quasi-orthogonal (scheme 2), and accordingly, a ferromagnetic coupling between them is expected. In the case of compound 4, the relative orientation of Cu1 and Cu3 coordination planes is similar to that found in compounds 1-3, but the plane containing the Cu2 is rotated by 60º with respect to the plane containing the Cu3 center, and a different interaction pattern is expected for this center. The nature and amplitude of the interactions inside these four compounds are analyzed by means of ab initio calculations, where both the low-lying J23 J13 J12 J13 J12 J23 Figure 17. The spin interaction model for 1-3 (left) and 4 (right). The cores are shown containing three Cu atoms and their nearest neighbors. 37 quartet and doublet states are computed, with the aim of identifying the ground state and the energy separations among the states. The magnitude of the magnetic coupling constants between the terminal and central Cu atoms (J12 = J13) and between the two terminal Cu atoms (J23) will be extracted from the energy difference of the three lowest states, using the relations shown in equation 10. These relations are strictly valid for compounds 1 and 2 where centers Cu2 and Cu3 are related by an inversion centre. We also apply these equations to compound 3, although it presents a slight deviation with respect to Ci symmetry point group. In compound 4, Cu2 and Cu3 are no more equivalent (see scheme 3), and three different coupling constants can be distinguished, J12, J13 and J23. In this situation, equation 10 just provides a rough estimate of the J values, if any. The rigorous evaluation of the three J values requires combining the two energy differences (equation 10) with the information contained in the wavefunctions by means of an effective Hamiltonian. Details of the procedure can be found in Refs. 36 Usually |J12| >> |J23| and for J12 > 0, the quartet is the ground state. We have demonstrated in the past that among the available quantum chemistry methods, Dedicated Difference Configuration Interaction (DDCI) 37 is a reliable approach for the evaluation of magnetic coupling constants both in solid state and molecular magnetic systems. 38 , 39 , 40 , 41 , 42 , 43 , 44 The central idea is to build a configuration interaction matrix with all the configurations participating on the energy difference at second-order of perturbation. This excludes all the double inactive excitations (two-electron promotion from the inactive doubly occupied to the virtual empty set of MOs), which are by far the most numerous. The J values obtained in this way are in general in a good agreement with available experimental data. For the systems considered here, we have selected three active orbitals, which are essentially combinations of the singly occupied Cu 3d atomic orbitals, with delocalization tails on the neighboring atoms. Figures 18 and 19 show the active orbitals for 1 (similar to those of compounds 2-3) and for 4, respectively. These three orbitals contain three unpaired electrons and constitute the active space CAS(3/3). Additionally an extended CAS, CAS(7/5), containing also two occupied orbitals centered on the O bridging ligands has been employed. 38 Figure 18. Quartet state active orbitals of complex 1 obtained from a CASSCF(3/3) calculation. "a" and "b" correspond to the symmetric and antisymmetric combinations of the terminal Cu 3d orbitals, with small tails on ligands, "c" is essentially localized on the central Cu atom, and "d" is the sum of the three MOs, to show the relative orientation of the Cu 3d magnetic orbitals. The energy and wavefunction of the three magnetic states, D1, D2 and Q, have been evaluated by means of DDCI calculations on the basis of a minimal CAS(3/3), hereafter DDCI(3/3), and DDCI2 calculations on the extended CAS, referred as DDCI2(7/5). This space contains all the single excitations and the double excitations involving two active orbitals (occ2  act2 and act2 vir2). To reduce the computational cost of the calculations, the two external aromatic rings have been replaced by H atoms (C-H distance of 1.07 Å), while those bonded to the two carboxylate groups are explicitly taken into account. This model retains the intramolecular hydrogen bonds, and maintains the conjugation of the carboxylate C atoms. Difference dedicated molecular orbitals, 45 obtained from the diagonalization of the difference of the CAS+S density matrices of the quartet and the D1 doublet states, are employed to perform the final calculations. The most important feature of these dedicated MOs is that their eigenvalues can be considered as a measure of the participation of the MO on the description of the magnetic state gaps. By selecting only the most a b c d 39 participating MOs in the subsequent calculations, it is possible to considerably reduce the size of the CI matrix, and hence the diagonalization time, with a negligible effect on the computed energy gaps. 46 The core electrons of Cu atoms have been represented by means of pseudopotentials, combined with basis set [9s6p6d] with contraction (3s3p4d). 47 Atomic natural orbital (ANO) 48 type basis functions have been used for the rest of atoms, with contractions 4s3p1d for O and C atoms directly connected to metal atoms, 4s3p for O, N and C atoms in the external shell, and 2s for H atoms. MOLCAS 7.8 code 49 has been used for the evaluation of the atomic and molecular integrals and CASDI code 50 for the diagonalization of the CI matrices. A quartet ground state is found for compounds1-4, regardless the level of the calculation, in agreement with the experimental data and the relative orientation of the magnetic Cu 3d orbitals. The energy distribution of the magnetic states is in accordance with a Heisenberg behavior (Eq. 10). The calculated J values (in cm-1) obtained are collected in Table 5, together with the experimental ones. The dominant interaction is the nearest neighbor coupling, J12, as expected, with J23 being almost negligible for all the compounds. The comparison of the experimental and DDCI values indicates that DDCI(3/3) reproduces correctly the spin-state/energy profiles, while the splitting shows only a qualitative agreement, the theoretical estimates being only about 55- 60% of the experimental value. This is not an isolate case. In fact, while DDCI approach on a minimal active space has been successful in the evaluation of J on antiferromagnetic compounds, the same performance has not been observed for ferromagnetic compounds. And this is not a difficulty pertinent solely to DDCI. In fact, recent DFT calculations on hetero triply-bridged dinuclear Cu(II) compounds found similar difficulties to reproduce the ferromagnetic coupling.7 The results reproduce qualitatively the experimental behavior but not the fine details along the series of compound explored. 40 Table 6. Calculated vs fitted magnetic coupling constants (cm-1) in complexes 1-4. For compound 4, J13 in parentheses, for the rest, J12=J13. DDCI(3/3) DDCI2(7/5) Fitted Complex J12 J23 J12 J23 J12 J23 1 40.8 0.2 69.3 2.1 76.8 0.5 2 35.6 0.3 77.1 2.8 65.2 -7.0 3 34.4 2.2 72.8 2.7 58.4 -6.9 4 28.9 (1.8) -0.4 51 -5 AWEQEZ 37.8 0.3 71.6 -1.3 66 -- However, when the active space is extended including the bridging ligand orbitals which play a crucial role in the ferromagnetic coupling (here two occupied orbitals centered on the O bridging ligands), the agreement is remarkably improved, even when staying at the DDCI2 level (in this case, only the double excitations involving two active orbitals are considered). A detailed analysis of the key role of these bridging orbitals is out of the scope of this work, and will be discussed elsewhere.Errore. Il segnalibro non è definito. The improvement brought by such an approach is clearly shown in the results reported in Table 6 for the four compounds studied here and for compound AWEQEZ which is isostructural with 1 and shows similar trends that those observed for compounds 1-3. Scheme 2. Two views of the relative orientation of the active Cu 3d orbitals in compounds 1-3. 41 Scheme 3. Two views of the relative orientation of the active Cu 3d orbitals in compound 4. The dominant interaction between the terminal and central Cu sites J12 is quantitatively reproduced, showing in all cases a coupling ferromagnetic in nature. This behavior is in agreement with that found in mixed bridged binuclear Cu(II) compounds where Cu centers are bridged by alkoxide and carboxylate ligands. These two groups are countercomplementary following the concept introduced by Nishida et al.15 and McKee et al.16 in the sense that one of the ligand orbitals has the correct symmetry to mix with the combination of the Cu 3d orbitals to form an antibonding M-L orbital, while the other has not net overlap (scheme 4). As a result of this countercomplementarity, a ferromagnetic interaction between the metal centers is expected. The same concept has also been invoked previously to explain the magnetic behaviour of mixed -acetato -hydroxo dinuclear and trinuclear Cu(II) complexes.10d,13 Our calculations, however, indicate that these two ligands contribute differently to the Cu-Cu magnetic coupling. In fact, the carboxylate group is just a spectator in the coupling, by completing the coordination sphere of Cu and impossing a certain geometry and orientation of the two centers, while the alkoxide ligand is crucial for the coupling as demonstrated by the fact that the calculated J values agree with the experimental ones when just the two orbitals of the alkoxide ligands are included in the active space. Scheme 4. Overlap between Cu 3d orbitals and alkoxide and carboxylate orbitals. 48 Tomkowicz, Z.; Haase, W.; Fleischhauer, P.; Krupski, H.; Micklitz, H.; Friedt, O.; Ściesińska, E.; Ściesiński, J.; Sczaniecki, P. J. Magn. Magn. Mater. 2001, 236, 347-356. (c) Van Slageren, J.; Ahmedova, A.; Gatteschi, D.; Massa, C. A.; Pardi, L. A. Inorg. Chim. Acta 2003, 351, 59-62. (d) Gutierrez, L.; Alzuet, G.; Real, J. A.; Cano, J; Borras, J; Castiñeiras, A. Inorg. Chem. 2000, 39, 3608-3614. (e) Tomkowicz, Z,; Fleischhauer, P.; Haase, W.; Baran, M.; Szymczak, R.; Zaleski, A. J.; Jablonski, R. J. Magn. Magn. Mater. 1997, 172, 128-138. (f) Gehring, S.; Fleischhauer, P.; Paulus, H.; Haase, W. Inorg. Chem. 1993, 32, 54-60. ( 11 ) Tudor, V.; Kravtsov, V.Ch.; Julve, M.; Lloret, F.; Simonov, Y. A.; Averkiev, B. B.; Andruh, M. Inorg. Chim. Acta 2005, 358 , 2066-2072. ( 12 ) (a) Boulsourani, Z.; Tangoulis, V.; Raptopoulou, C. P.; Psycharis, V.; Dendrinou-Samara, C. Dalton Trans. 2011, 40, 7946-7956. (b) Panagoulis, D.; Pontiki, E. Skeva, E.; Raptopoulou, C.; Girousi S. D.; Hadjipavlou-Litina, C.; Dendrinou-Samara, C. J. Inorg. Biochem. 2007, 101, 623-634. ( 13 ) Escovar, R.M.; Thurston, J.H.; Ould-Ely, T.; Kumar, A.; Whitmire, K.H. Z. Anorg. Allg. Chem. 2005, 631, 2867-2876. ( 14 ) (a) Ruiz, E.; Alemany, P.; Alvarez, S.; Cano, J. J. Am. Chem. Soc. 1997, 119, 1297-1303. (b) Ruiz, E.; Alemany, P.; Alvarez, S.; Cano, J. Inorg. Chem. 1997, 36, 3683-2688. (c) Doyle, R. P.; Julve, M.; Lloret, F.; Nieuwenhuyzen, M.; Kruger, P. Dalton Trans. 2006, 2081-2088. ( 15 ) a) Nishida, Y.; Kida, S. J. Chem. Soc. Dalton Trans. 1986, 2633-2640. b) Nishida, Y.; Takeuchi, M.; Takahashi, K.; Kida, S. Chem. Lett. 1985, 631-634; c) Nishida, Y.; Takeuchi, M.; Takahashi, K.; Kida, S. Chem. Lett. 1983, 1815-1818. ( 16 ) McKee, V.; Zvagulis, M.; Reed, C. A. Inorg. Chem. 1985, 24, 2914-2919. ( 17 ) Malati, M. A. Experimental Inorganic Chemistry, 1st Ed, Harwood Publishing, Chichester, 1999. ( 18 ) Blessing, R. H. Acta Crystallogr. 1995, A51, 33-38. ( 19 ) Altomare, A.; Burla M.C.; Camalli, M.; Cascarano, G.; Giacovazzo, C.; Guagliardi, A.; Moliterni, A.G.; Polidori, G.; Spagna, R. J. Appl. Crystallogr. 1999, 32, 115-119. ( 20 ) Sheldrick, G. M. SHELXL97, Program for Crystal Structure Refinement, University of Göttingen, Göttingen, Germany, 1997. ( 21 ) Farrugia, L.J. J. Appl. Crystallogr. 1999, 32, 837-838. ( 22 ) (a) O’Connor, C. J. Prog. Inorg. Chem. 1982, 29, 203-283. (b) Bain, G. A.; Berry, J. F. J. Chem. Ed. 2008, 85, 532-536. ( 23 ) Hassan, A. K.; Pardi, L. A.; Krzystek, J.; Sienkiewicz, A.; Goy, P.; Rohrer, M.; Brunel, L.- C. J. Magn. Reson. 2000, 142, 300-312. ( 24 ) Burnett, M.N.; Johnson, C.K. ORTEPIII. Oak Ridge Thermal Ellipsoid Plot Program for Crystal Structure Illustrations, Report ORNL-6895, Oak Ridge National Laboratory, Oak Ridge, Tennessee, USA, 1996. 49 ( 25 ) (a) Bellamy, L.J. The Infrared Spectra of Complex Molecules, 2nd Ed. Vol. 2, Chapman & Hall, London/New York, 1980. (b) Nakamoto, K. Infrared and Raman Spectra of Inorganic and Coordination Compounds, 5th Ed., John Wiley & Sons, New York , 1997. ( 26 ) (a) Das, M.; Shivashankar, S.A. Applied Organomet. Chem. 2007, 21, 15-25. (b) Topcu, Y.; Andac, O.; Yilmaz, V.T.; Harisson, W. J. Coord. Chem. 2002, 55, 805-815. ( 27 ) Bencini, A.; Gatteschi, D. EPR of Exchange Coupled Systems. Springer Verlag, Berlin, 1990. ( 28 ) Sharma, R.P.; Saini, A.; Monga, D.; Venugopalan, P.; Jezierska, J.; Ozarowski, A.; Ferretti, V. New J. Chem. 2014, 38, 437-447. ( 29 ) (a) Ozarowski, A. Inorg. Chem. 2008, 47, 9760-9762. (b) Ozarowski, A.; Szymanska, I. B.; Muziol, T.; Jezierska, J. J. Am. Chem. Soc. 2009, 131, 10279-10292. (c) Semenaka, V. V.; Nesterova, O. V.; Kokozay, V. N.; Dyakonenko, V. V.; Zubatyuk, R. I.; Shishkin, O. V.; Boca, R.; Jezierska, J.; Ozarowski, A. Inorg. Chem. 2010, 49, 5460-5471. ( 30 ) (a) Senchyk, G.A.; Lysenko, A.B.; Rusanov, E.B.; Chernega, A.N.; Jezierska, J.; Domasevitch, K.V.; Ozarowski, A. Eur. J. Inorg. Chem. 2012, 35, 5802-5813. (b) Aznar, E.; Ferrer, S.; Borras, J.; Lloret, F.; Liu-Gonzalez, M.; H. Rodriguez-Prieto, H.; Garcia-Granda, S. Eur. J. Inorg. Chem. 2006, 24, 5115-5125. ( 31 ) (a) Moriya, T. Phys. Rev. 1960, 120, 91-98. (b) Yoon, J.; Solomon, E. I. Coord. Chem. Rev. 2007, 251, 379-400. (c) Ferrer, S.; Lloret, F.; Pardo, E.; Clemente-Juan, J. M.; Liu- Gonzlez, M.; García-Granda, S. Inorg. Chem. 2012, 51, 985-1001. (d) Vasylevskyy, S. I.; Senchyk, G. A.; Lysenko, A. B.; Rusanov, E. B.; Chernega, A. N.; Jezierska, J.; Krautscheid, A.; Domasevitch, K. V. Ozarowski, A. Inorg. Chem. 2014, 53, 3642-3654. (e) Boča, R.; Herchel, R. Coord. Chem. Rev. 2010, 254, 2973-3025. ( 32 ) (a) Maurice, R.; Sivalingam, K.; Ganyushin, D.; Guihery, N.; de Graaf, C.; Neese, F. Inorg. Chem. 2011, 50, 6229-6236. (b) Gribnau, M. C. M.; Keijzers, C. P. Inorg. Chem. 1987, 26, 3413-3414. (c) Ross, P. K.; Allendorf, M. D.; Solomon, E. I. J. Am. Chem. Soc. 1989, 111, 4009- 4021. (d) Singh, S. K.; Rajaraman, G. Chem. Eur. J. 2014, 20, 5214-5218. (f) Atanasov, M.; Comba, P.; Hanson, G. R.; Hausberg, S.; Helmle, S.; Wadepohl, H. Inorg. Chem. 2011, 50, 6890- 6901. ( 33 ) Note that if the Z axis were chosen along the Cu-Cu direction, the formulas would be different: Ddip = -(4gx2 + gy2 + gz2)  B2/4R3Cu-Cu, Edip = (gz2 - gy2)  B2/4R3Cu-Cu ( 34 ) Riplinger, C.; Kao, J. P.Y.; Rosen, G. M.; Kathirvelu, V.; Eaton, S. S.; Kutateladze, A.; Neese, F. J. Am. Chem. Soc. 2009, 131, 10092-10106. ( 35 ) (a) Reger, D.L.; Pascui, A.E.; Smith, M.D.; Jezierska, J.; Ozarowski, A. Inorg. Chem. 2015, 54, 1487-1500. (b) Reger, D.L.; Pascui, A.E.; Foley, E.A.; Smith, M.D.; Jezierska, J.; Ozarowski, A. Inorg. Chem. 2014, 53, 1975-1988. (c) Reger, D.L.; Pascui, A.E.; Smith, M.D.; Jezierska, J.; Ozarowski, A. Inorg. Chem. 2012, 51, 11820-11836. (d) Reger, D.L.; Pascui, A.E.; Smith, M.D.; Jezierska, J.; Ozarowski, A. Inorg. Chem. 2012, 51, 7966-7968. ( 36 ) (a) Calzado, C. J.; Malrieu, J.P. Phys. Rev. B 2001, 63, 214520. (b) Calzado, C. J.; Malrieu, J.P. Eur. Phys. J. B 2001, 21, 375-381. (c) Calzado, C. J.; Malrieu, J.P. Phys. Rev. B 2004, 69, 0944351. (d) Calzado, C. J.; de Graaf, C.; Bordas, E.; Caballol, R.; Malrieu, J. P. Phys. Rev. B 50 2003, 67, 132409. (e) Malrieu, J.P.; Caballol,R.; Calzado, C.J.; de Graaf, C.; Guihery, N. Chem.Rev. 2014, 114, 429-492. ( 37 ) (a) Miralles, J.; Daudey, J. P.; Caballol, R. Chem. Phys. Lett. 1992, 198, 555-562; (b) Miralles, J.; Castell, O.; Caballol, R.; Malrieu, J. P. Chem. Phys. 1993, 172, 33-43. ( 38 ) (a) Oms, O.; Rota, J. B.; Norel, L.; Calzado, C. J.; Rousseliere, H.; Train C.; Robert, V. Eur. J. Inorg. Chem. 2010, 5373-5378. (b) Bandeira, N. A. G.; Maynau, D.; Robert, V.; Le Guennic, B. Inorg. Chem. 2013, 52, 7980-7986. (c) Vérot, M.; Rota, J. B.; Kepenekian, K.; Le Guennic B.; Robert, V. Phys. Chem. Chem. Phys. 2011, 13, 6657-6661; (d) Aronica, C.; Chastanet, G.; Pilet, G.; Le Guennic, B.; Robert, V.; Wernsdorfer, W.; Luneau, D. Inorg. Chem. 2007, 46, 6108-6119. ( 39 ) (a) Calzado, C. J.; Sanz, J.F.; Malrieu, J.P. J. Chem. Phys. 2000, 112, 5158-5167. (b) Calzado, C. J.; Evangelisti, S.; Maynau, D. J. Phys. Chem. A, 2003, 107, 7581-7588. ( 40 ) (a) Casanovas, J.; Illas, F. J. Chem. Phys. 1994, 100, 8257-8264. (b) Moreira, I. de P. R.; Illas, F.; Calzado, C. J.; Sanz, J.; Malrieu, J. P.; Ben Amor, N.; Maynau, D. Phys. Rev. B, 1999, 59, 6593-6596. (c) de Graaf, C.; Moreira, I. de P. R.; Illas, F.; Martin, R.L. Phys. Rev. B, 1999, 60, 3457-3464. (d) Muñoz, D.; Illas, F.; Moreira, I. de P. R. Phys. Rev. Lett. 2000, 84, 1579- 1582. ( 41 ) (a) Cabrero, J.; de Graaf, C.; Bordas, E.; Caballol, R.; Malrieu, J.P. Chem. Eur. J. 2003, 9, 2307-2315. (b) Queralt, N.; Taratiel, D.; de Graaf, C.; Caballol, R.; Cimiraglia, R.; Angeli, C. J. Comput. Chem. 2008, 29, 994-1003. ( 42 ) Suaud, N.; Gaita-Arino, A.; Clemente-Juan, J. M.; Sanchez-Marin, J.; Coronado, E. J. Am. Chem. Soc. 2002, 124, 15134-15140. ( 43 ) (a) Barone, V.; Cacelli, I.; Ferretti, A.; Monti, S.;Prampolini, G. J. Chem. Theory Comput. 2011, 7, 699-706. (b) Barone, V.; Cacelli, I.; Ferretti,A.; Monti, S.; Prampolini, G. Phys. Chem. Chem. Phys. 2011, 13, 4709-4714. ( 44 ) (a) Neese, F. J. Am. Chem. Soc. 2006, 128, 10213-10222. (b) Liakos, D. G.; Ganyushin, D.; Neese, F. Inorg. Chem. 2009, 48, 10572-10580. (c) Ganyushin, D.; Gilka, N.; Taylor, P. R.; Marian, C. M.; Neese, F. J. Chem. Phys. 2010, 132, 144111. (d) McNaughton, R. L.; Roemelt, M.; Chin, J. M.; Schrock, R. R.; Neese, F.; Hoffman, B.M. J. Am. Chem. Soc. 2010, 132, 8645- 8656. ( 45 ) (a) Miralles, J.; Caballol, R.; Malrieu, J.-P. Chem. Phys. 1991, 153, 25-37. (b) Calzado, C. J.; Malrieu, J.-P.; Cabrero, J.; Caballol, R. J. Phys. Chem. A 2000, 104, 11636-11643. ( 46 ) (a) Calzado, C. J. ; Evangelisti, S. Dalton Trans. 2014, 43, 2988-2996. (b) Rota, J.B. ; Calzado, C. J. ; Train, C. ; Robert, V. J. Chem. Phys. 2010, 132, 154702. (c) Rota, J. B.; Norel, L.; Train, C.; Ben Amor, N.; Maynau, D. ; Robert, V. J. Am. Chem. Soc. 2008, 130, 10380- 10385; (d) Castell, O.; Caballol, R.; García, V.M.; Handrick, K. Inorg. Chem. 1996, 35, 1609- 1615; (e) Calzado, C. J. ; Clemente-Juan, J.M. ; Coronado, E. ; Gaita-Ario, A. ; Suaud, N. Inorg. Chem. 2008, 47, 5889-5901. ( 47 ) Barandiarán, Z.; Seijo, L. Can. J. Chem. 1992, 70, 409-415. 51 ( 48 ) (a) Widmark, P.-O.; Malmqvist, P.-A.; Roos, B.O. Theor. Chim. Acta 1990, 77, 291-306. (b) Widmark, P.-O.; Persson, B.J.; Roos, B.O. Theor. Chim. Acta, 1991, 79, 419-432. (c) Pierloot, K.; Dumez, B.; Widmark, P.-O.; Roos, B.O. Theor. Chim. Acta 1995, 90, 87-114. ( 49 ) Aquilante, F.; de Vico, L.; Ferré, N.; Ghigo, G.; Malmqvist, P.-A.; Pedersen, T.; Pitonak, M.; Reiher, M.; Roos, B.O.; Serrano-Andrés, L.; Urban, M.; Veryazov, V.; Lindh, R. J. Comput. Chem. 2010, 31, 224-247. ( 50 ) CASDI program: Ben Amor, N. ; Maynau, D. Chem. Phys. Lett. 1998, 286, 211-220, package developed at the Laboratoire de Chimie et Physique Quantiques, Université Paul Sabatier, Toulouse (France). ( 51 ) (a) Sanchez Costa, J.; Bandeira, N. A. G; Le Guennic, B.; Robert, V.; Gamez, P.; Chastanet, G.; Ortiz-Frade, L.; Gasque, L. Inorg. Chem. 2011, 50, 5696-5705. (b) Moreno, J. M.; Ruiz, J.; Dominguez-Vera, J. M.; Colacio, E. Inorg. Chim. Acta 1993, 208, 111-115. (c) Plass, W.; Pohlmann, A.; Rautengarten, J. Angew. Chem., Int. Ed. 2001, 40, 4207-4210. (d) Valigura, D.; Moncol, J.; Korabik, M.; Pucekova, Z.; Lis, T.; Mrozinski, J.; Melnik, M. Eur. J. Inorg. Chem. 2006, 3813-3817. (e) Tang, J. K.; Costa, J. S.; Golobic, A.; Kozlevcar, B.; Robertazzi, A.; Vargiu, A. V.; Gamez, P.; Reedijk, J. Inorg. Chem. 2009, 48, 5473-5479. (f) Vaskova, Z.; Moncol, J.; Korabik, M.; Valigura, D.; Svorec, J.; Lis, T.; Valko, M.; Melnik, M. Polyhedron 2010, 29, 154-163. (g) Desplanches, C.; Ruiz, E.; Rodriguez-Fortea, A.; Alvarez, S. J. Am. Chem. Soc. 2002, 124, 5197-5205.(h) Le Guennic, B.; Ben Amor, N.; Maynau, D.; Robert, V. J. Chem. Theory Comput. 2009, 5, 1506-1510.