2.1 Materials

Copper(II) chloride dihydrate, tetramethylammonium hydroxide, triethylamine, ethyl oxalyl chloride and 2-aminopyrimidine were purchased from commercial sources and used as received. The proligand (2-pyrimidyl)oxamic acid ethyl ester (HEtpmo) was prepared by reacting ethyl oxalyl chloride with 2-aminopyrimidine and triethylamine in 1:1:1 molar ratio under reflux at 80 °C during 4 h, the general procedure previously described for these type of ligands being followed [18]. Elemental analysis (C, H, N) were conducted by the Microanalytical Service of the Federal University of Goiás and the copper contents was determined by spectrophotometry with a Hitachi-Z8200 spectrophotometer.

2.2 Preparation of {[(CH₃)₄N]₂[Cu(C₂O₄)₂] · H₂O}_n (1)

Teramethylammonium hydroxide (0.441 cm³, 4.4 mmol) was added to a methanolic solution (10 cm³) of HEtpmo (0.20 g, 1.1 mmol) under continuous stirring. After 10 min, an ethanolic solution (5 cm³) of copper(II) chloride dihydrate (0.093 g, 0.55 mmol) was added dropwise. The small amount of a bluish solid that resulted was removed by filtration and the remaining deep blue solution was allowed to slowly evaporate in a hood at room temperature. X-ray quality turquoise crystals of 1 as blue prisms were grown after several weeks. They were collected by filtration, washed with a small amount of cold water and dried over filter paper. Yield: 45.0%. Anal. found: C, 35,37; H, 6.35; Cu, 15.59; N, 6.80%. Calc. for C₁₂H₂₆CuN₂O₉: C, 35.48; H, 6.40; Cu, 15.66; N, 6.90%. Main IR features (values in cm⁻¹, KBr pellet): 3326 m [ν(O–H)], 3020w, 2914w and 2837w [ν(C–H)], 1673s, 1652s and 1637s] [ν_as(OCO)], 1488 and 1418 [ν_s(OCO)], 1289 m [ν_s(C–O)] and 808w [δ(OCO)].

2.3 Physical measurements

An infrared spectrum of 1 was recorded with a Perkin Elmer Precesily Spectrum 400 FT-IR/FT-FIR spectrophotometer as a KBr pellet in the range 4000–400 cm⁻¹. Magnetic susceptibility measurements on a polycrystalline sample of 1 were carried out in the temperature range 1.9–300 K with a Quantum Design SQUID susceptometer, using applied magnetic dc fields of 0.1 T (T ≥ 50 K) and 500 G (T < 50 K). The corrections for the diamagnetism of the constituent atoms were estimated from Pascal's constants [19] as –204 × 10⁻⁶ cm³ mol⁻¹ [per mol of copper(II) ions]. Corrections for the temperature-independent paramagnetism [60 × 10⁻⁶ cm³ mol⁻¹ per Cu(II)] and for the simple holder were also applied.

2.4 Crystal structure determination and refinement

A prism-shaped single crystal of 1 with dimensions 0.12 × 0.08 × 0.07 mm³ was selected for the X-ray data collection. It was mounted on a glass fiber and afterwards positioned on the goniometer. Room temperature X-ray diffraction intensities (298 K) were measured using graphite-monochromated Mo Kα line (λ = 0.71073 Å), Enraf-Nonius Kappa-CCD diffractometer equipped with a 95 mm CCD camera on a κ-goniostat operating in φ-ω scans mode with κ offsets. All reflections were used to index the final unit cell. The acquisition of the X-ray diffraction frames was monitored by using the program COLLECT [20] and the raw dataset was treated with the HKL Denzo-Scalepack [21]. Because of the small size (medial crystal size x = 0.08 mm) and the not too high absorption coefficient (μ = 1.275 mm⁻¹), no absorption correction was applied. According to the International Union of Crystallography (IUCr) [22], there is no need of absorption correction if the product μx is not above 0.1. This product was 0.102 for the measured crystal of 1 by using Mo Kα radiation. Therefore, we have preferred to comply with IUCr recommendations even if the refinement statistic was satisfactory (Table 1).

The structure was solved by direct methods with SHELXS-97 [23]. All non-hydrogen atoms were readily found from the electronic density map constructed by Fourier synthesis, in which the Cu, C, O and N atoms were clearly identified. The initial model was refined by the full-matrix least squares method on F² with SHELXL-97 [23] adopting anisotropic thermal parameters for all non-hydrogen atoms. The hydrogen atoms form the NMe₄⁺ cation were positioned stereochemically and they were refined with fixed individual isotropic displacement parameters [U_iso(H) = 1.5U_eq of both methyl carbons and water oxygen] by using a riding model with CH bond lengths of 0.96 Å. The positions of the hydrogen atoms from the water molecules were assigned from the electronic density map generated by Fourier difference and they were refined freely.

The ORTEP-3 program [24] was used within the WinGX software package [25] to deal with the processed crystallographic data and artwork representations. After the end of the refinement, the molecular structure of 1 was checked with MOGUL program [26]. This is a valuable program for analyzing geometric features of any structure. It searches for substructures of compounds deposited at the Cambridge Crystallographic Data Centre (CCDC) that are similar to those of a target. Comparisons of bond angles and lengths of 1 with the corresponding parameters of similar structures deposited in the Cambridge Structural Database (CSD, version 5.32 of November 2010 with August 2011 update) [27] were useful to strengthen interesting intramolecular features of the complex studied. Bond lengths and angles of 1 together with those for structurally related complexes are listed in Tables 2 and 3 whereas the X-H O (HC and O_w) intermolecular interactions of 1 are given in Table 4.

3.1 Synthesis and IR characterization

It is known that oxamato and oxamidato groups can undergo a copper(II)-assisted hydrolysis in basic media yielding oxalate [18,28]. This hydrolytic reaction transforms HEtpmo into oxalate and it accounts for the slow formation of single crystals of 1 from the basic aqueous solution containing copper(II) and the oxamate ligand (Scheme 1). When the synthesis was carried out by using oxalic acid instead of the monoxamic derivative, a blue powder resulted whose analytic data indicate that a different species has been obtained. It seems that the hydrolytic process can be considered as a suitable way to get X-ray quality crystals of oxalate-bridged copper(II) complexes that could not be obtained by the direct reaction between oxalate and the copper(II) ion.

The IR absorptions of the oxalate group in the spectrum of 1 [1673,1652 and 1637 cm⁻¹, (ν_as(COO) stretching vibrations), 1488 and 1418 cm⁻¹ (ν_s(COO) stretching vibrations) and 1289 and 808 cm⁻¹ (ν_s(CO) and δ(OCO) vibrations, respectively] suggest the presence of chelating and bis-chelating oxalato [29], a feature that has been demonstrated by the X-ray structure of this compound (see below).

3.2 Description of the structure

Compound 1 crystallizes in the centrosymmetric orthorhombic space group Pbca, with one copper(II) ion, two oxalate ligands, two tetramethylammonium counterions, and one free water molecule in the asymmetric unit (Fig. 1). One of the two oxalate groups in the asymmetric unit acts as a bis-bidentate ligand coordinating two b-glide plane symmetry-related copper(II) ions through its four oxygen atoms [O1A and O2A at Cu1 and O3A and O4A at Cu1ⁱ [symmetry code: (i) = ½ – x, ½ + y, z]. The other oxalate group adopts the bidentate coordination mode through the O1B and O2B atoms completing the six-fold coordination-sphere around the copper atom. The resulting [Cu(ox)₂]_n²ⁿ⁻ anionic unit grows along the crystallographic b axis as a zigzag copper(II) chain (Fig. 2). The copper(II)-copper(II) separation across the bridging oxalate is 5.5898(3) Å [Cu1 Cu1ⁱ] whereas the distances between copper(II) ions belonging to neighbouring chains are much longer: 8.9746(4) Å for the distance between chains packed parallel to the crystallographic c axis and related by 2₁ screw axis along the [100] direction [Cu1Cu1ⁱⁱⁱ; symmetry code: (iii) = –½ + x, ½ – y, – z]. It is striking to notice that these chains are not mirror images. They are only running on opposite directions parallel to the crystallographic b axis (Fig. 2). In addition, a-glide plane symmetry-related chains are also packed parallel to the crystallographic c axis. In this case, there is a separation of 10.2382(4) Å between the copper(II) ions on adjacent mirror chains [Cu1Cu1^iv; symmetry code: (iv) = –½ + x, y, ½ – z]. Finally, the interchain distances between translation symmetry-related ribbons packed along the crystallographic a axis are the largest ones, 10.2783(3) and 11.2757(4) Å for Cu1Cu1^v and Cu1Cu1^vi, respectively [symmetry code: (v) = 5 – x, ½ + y, z; (vi): 1 + x, y, z].

Each copper(II) ion in 1 adopts a distorted octahedral coordination geometry, a feature that contrast with the preferential four- or five-coordination of this metal ion in its oxalato-bridged complexes [5]. In the light of the CuO bond lengths from Table 2, the geometry around the copper(II) ion in 1 is better described as an elongated and statically distorted octahedron because of four short CuO distances [1.983(2), 1.973(2), 1.957(2) and 1.978(2) Å for Cu1O1B, Cu1–O2A, Cu1O2B and Cu1O4Aⁱⁱ, respectively; symmetry code: (ii) = ½ – x, – ½ + y, z] and two weaker CuO interactions [2.323(2) and 2.437(2) Å for Cu1O1A and Cu1O3Aⁱⁱ]. Distances similar to those observed in 1 are also observed for a tris(oxalato)copper(II) unit in a previous report where an elongated octahedral geometry around the copper(II) ion with four equatorial oxygen atoms and two axially bound oxygens at 1.99 and 2.32 Å, respectively, from the copper atom were suggested on the basis of EXAFS data [30]. In the related complex [Cu(en)₃]SO₄ [31]. EPR and XAS results demonstrated that the copper(II) ion geometry is a dynamically distorted octahedron at room temperature due to four short CuN lengths and two enlarged CuN oscillating distances, adopting a 4 + 2 coordination. Both geometry distortions of the copper(II) ion were attributed to the Jahn-Teller effect [30,31]. In this way, one can conclude that such an effect is responsible for the elongated and distorted octahedron geometry of 1. Moreover, it is important to note that the 4 + 2 octahedron geometry in 1 is a statically distorted octahedron rather than the dynamically one reported for the related complexes aforementioned, the two longer Cu1O1A and Cu1O3Aⁱⁱ bond distances being clearly different in 1.

In the coordination sphere of copper(II) atom of 1, two of the four equatorial oxygen-atoms belong to two oxalate bridges related by b-glide plane symmetry (O2A and O4Aⁱⁱ) and the two others (O1B and O2B) are from the oxalate which does not bridge between two copper(II) atoms. All these four equatorial oxygens deviate considerably from the least-squares plane passing through them [O1B, O2A, O2B and O4Aⁱⁱ deviate –0.090(2) Å, 0.076(2) Å, 0.091(2) Å and –0.073(2) Å; rmsd of the four fitted atoms is 0.0063 Å]. The Cu1 atom is shifted by 0.0446(2) Å from the least-squares basal plane. The cis O_eqCuO_eq bond angles vary in the range 90.28(7)–94.24(7)° whereas the trans ones are 170.57(7)° (O1BCu1O4Aⁱⁱ) and 175.04(7)° (O2ACu1O2B). The O_eqCuO_ax bond angles span in the range 75.61(6) to 99.85(6)° and the axial O1ACu1O3Aⁱⁱ one is 170.72(5)° (Table 3). The short values of the bite of the chelating and bis-chelating oxalate ligands are at the origin of the main deviation from the ideal 90° or 180° bond angles at the copper environment in 1. The shortening of the C1AO1A (1.234(2) Å) and C2AO3A (1.236(3) Å) bond distances when compared to the related equatorial carbon-oxygen bonds [values in the range 1.262(3)–1.269(3) Å] is due to the strongest electrostatic interactions in the equatorial oxalate-oxygen to copper(II) bonds. Therefore, C1AO1A and C2AO3A bonds exhibit an increased CO double bond character rather than a major CO single bond feature of the enlarged bonds C1AO4A (1.262(3) Å), C1BO1B (1.269(3) Å), C2AO2A (1.266(3) Å) and C2BO2B (1.269(3) Å). Mean values of 1.25(2) Å and 1.26(3) Å were averaged by MOGUL in the Cambridge Structural Database (CSD) [27] on 170 and 16 structures with bonds equivalent to C1AO1A and C2AO3A of 1, respectively. MOGUL is a knowledge database of molecular geometry derived from CSD [27] that provides access to information on the preferred values of bond lengths, valence angles and torsion angles. Besides highlighting interesting geometrical features such as these, MOGUL searches have showed that all bond lengths and bond angles are in agreement with the expected values for a good X-ray diffraction structure refinement (Tables 2 and 3). In addition, MOGUL surveys have revealed an elongation of the bond distance between the two sp²-hybridized carbons of the oxalate bridge, which is rationalized by taking into account that this oxalate unit is four-coordinated to copper(II) and then there is a notable tension about the C1AC2A bond. This bond measures 1.566(3) Å in I whereas a mean value of 1.54(2) Å was averaged on 868 1-like compounds searched by MOGUL in the CSD [27]. On the other hand, the carbon-carbon equivalent bond in the non-bridging oxalate is 1.545(4) Å (C1BC2B bond length). This value matches to the MOGUL mean value averaged on 1485 structures similar to 1 (1.54(2) Å). Based on this MOGUL matching, it is possible to state that elongation of the C1BC2B bond length does not occur, which is agreement with the fact that the non-bridging oxalate is bidentate towards copper(II) and thus, it undergoes less tension on its C_sp²C_sp² bond than the bis-bidentate oxalate.

The three oxalate ligands are not coplanar: the octahedral coordination geometry around the Cu^II atom forces each oxalate ligand to lie almost perpendicular to each one. The least-squares planes through two b-glide plane symmetry-related oxalate bridges [the root-mean-square deviation (rmsd) of the six fitted atoms C1A, C2A, O1A, O2A, O3A and O4A is 0.0101 Å] form an angle of 68.38(3)°, while the least-squares plane of the other non-bridging oxalate fragment (rmsd of the six fitted atoms C1B, C2B, O1B, O2B, O3B and O4B is 0.0121 Å) forms angles of 72.33(5)° and 88.01(6)° with the corresponding planes of the bridging oxalate groups.

It is striking to note that the synthesis of a complex such as 1 with the Cu^II ion surrounded by three coordinated oxalate ligands in which the transition metal ion undertakes an octahedral coordination geometry is unpredicted because only three structures loading this framework are found in the CSD [27]. Likewise, a one-dimensional zigzag coordination polymer of six-coordinated copper(II) assembled with oxalate ligands is unlikely to be prepared using common inorganic preparative routes since no example of this is reported. In this work, such a structure is described for the first time. Oxalate bridges intercalates copper(II) ions along the [010] direction, resulting in a zigzag chain grown along the crystallographic b axis (Fig. 2).

Tetramethylammonium cations are also packed along the b axis, even though they are further arranged parallel to the a axis. Only van der Waals interactions keep them in contact along both directions. Therefore, 2D columns of tetramethylammonium running parallel to the [100] and [010] directions are assembled in the structure of 1. These columns intercalate between two copper(II) chains along the c axis, (Fig. 3). Moreover, the organic cations contribute to stabilize the crystal packing through weak CHO type interactions with both oxalate-oxygens and the crystallization water molecule (Table 4). These contacts connect the chains along the a and c axes, giving rise to a supramolecular 3D network. Likewise, water plays a dual role in the crystal assembly of 1. It is a non-classical hydrogen bonding acceptor from a given (CH₃)₄N⁺ cation in the C1A′H1A1O1W^v type interactions [symmetry code: (v) = –½ + x, y, ½ – z], C2A′–H2A3O1W, and C3A′H3A3O1w (Fig. 4). At the same time, each water molecule forms hydrogen bonds with the non-coordinated oxygens of two b-glide plane symmetry-related non-bridging oxalate units. The O1wH1WO4B hydrogen bond and the bifurcated O1w–H2WO4B^vi [symmetry code: (vi) = 1.5 – x, ½ + y, z] and O1wH2WO3B^vi ones connect water molecule to two coordination ribbons, acting as a cross-linker between two anionic chains parallel to the [100] direction (Fig. 4 and Table 4). Therefore, the non-coordinated oxalate-oxygen atoms are involved in hydrogen bonding interactions, which contribute to the stabilization of the supramolecular three-dimensional structure of 1. This reveals that there is a fine tuning between the formation of coordination bonds and intermolecular contacts in this noteworthy structure.

3.3 Magnetic properties

The magnetic properties of 1 in the form of χ_MT against T plot [χ_M is the magnetic susceptibility per copper(II) ion] are shown in Fig. 5. At room temperature, χ_MT is equal to 0.40 cm³ mol⁻¹ K, a value which is as expected for a magnetically non-interacting spin doublet. This value remains practically constant upon cooling until T = 150 K and it further increases continuously to reach 0.69 cm³ mol⁻¹ K at 1.9 K. Because of the one-dimensional character of 1, we have analyzed its magnetic data through the numerical expression proposed by Baker and Rushbrooke [32] for a ferromagnetically coupled uniform chain of spin doublets [Eq. (1)]

In such an expression, J is the exchange coupling between adjacent copper(II) ion of the chain parameter, g is the average Landé factor and N, β and k have their usual meaning. Least-squares fit of the susceptibility data lead to the following values: J = +1.14 cm⁻¹, g = 2.08 and R = 1.6 × 10⁻⁵ (R is the agreement factor defined as ∑iχMTobsi−χMTcalci2/∑iχMTobsi2. The calculated plot matches well the magnetic data in the whole temperature range explored.

The ferromagnetic coupling found in 1 is weak and it corresponds to a case of accidental orthogonality between the interacting magnetic orbitals through the out-of-plane exchange pathway, a situation analogous to that previously observed in μ-chloro, di-μ-chloro and μ-oxamato dicopper(II) complexes where this exchange pathway is involved [33,34]. Simple orbital symmetry considerations allow to visualize the origin of this ferromagnetic interaction across the bridging oxalato in 1. Let us focus on a dicopper(II) fragment of this chain. The unpaired electron on each copper(II) ion in 1 is described by a d(x²–y²) type metal centered magnetic orbital [the x and y axes being roughly defined by the Cu1O2A and Cu1O1B bonds] which is located in the equatorial plane and the two magnetic orbitals are parallel to each other and perpendicular to the oxalate plane (Scheme 2).

The exchange pathway involves the oxalate O2AC2AO3A and its symmetry-related O1AC1AO(4a) set of atoms. According to Kahn's model [35], the coupling constant J in a dicopper(II) unit can be decomposed into two terms, one positive (ferro-, J_F) and the other negative (antiferromagnetic, J_AF), the expression being J = J_F + J_AF. In such a model, the value of the negative term is proportional to the square of the overlap integral (S²) between the two metal centered magnetic orbitals. The poor overlap between the two parallel magnetic orbitals through the two OCO oxalate set of atoms would lead to a weak antiferromagnetic coupling. However, in the case where this overlap is zero (the accidental orthogonality occuring in 1), a weak ferromagnetic coupling is predicted.

In that respect, the magneto-structural data of the oxalato-bridged dicopper(II) complexes listed in Table 5 show that the values of the bond angle at the axial copper to oxalate-oxygen (β) and the axial copper to oxalate-oxygen bond length are the main structural factors governing the nature of the magnetic coupling for the topology shown in Scheme 2. In the light of this Table, one can conclude that the frontier between the ferro- and antiferromagnetic coupling in this family of complexes corresponds to a value of β equal to 109.5°. Values of β greater than this one will favour the antiferromagnetic coupling whereas smaller ones will give rise to ferromagnetic interactions. Finally, it deserves to be pointed out that this magneto-structural result was first observed by Castillo and García-Coucerio during their doctoral work [42].