3.3. Molecular and Crystal Structures of Compounds 1–3
The structure of compound {[Mn(bpy)(C
2O
4)]·1.5H
2O}
n (
1), crystalizing in a monoclinic space group
C2/
c, consists of neutral [Mn(bpy)(C
2O
4)] units with the manganese(II) ions linked by oxalate groups to form
zigzag chains extending in the direction [110] (
Figure 1), and crystallization water molecules. The asymmetric unit contains, beside manganese(II) atom and bpy ligand, halves of two oxalate bridges (
Figure 1 and
Figure S1) and one and a half of the crystallization water molecules; one water molecule is located on a twofold axis (therefore, p.p. 0.5), while the other one is disordered about another twofold axis (two positions with p.p. 0.5).
The manganese(II) atom displays distorted octahedral coordination, involving two N atoms from the bipyridine molecule [2.2435(17) and 2.2435(17) Å] and four O atoms from two bridging bis(bidentate) oxalate groups (on average Mn–O = 2.17595 Å). The values of the Mn–N and Mn–O bond lengths (
Table S1) are in good agreement with those of similar 1D coordination polymer of manganese(II) ions, crystallizing without water, in space group
Pna2
1 [
43,
44].
The Mn1⋯Mn1i and Mn1⋯Mn1ii [symmetry operators: (i) −x, −y, −z; (ii) 1/2 − x, 1/2 − y, −z] distances across the bridging oxalate group are 5.6249(6) and 5.6656(6) Å, respectively. The shortest distances between two manganese(II) ions from two neighbouring chains is 7.6054(7) Å.
Water molecule O5 connects two neighbouring chains by hydrogen bonding to oxalate groups, while accepting two C–H∙∙∙O hydrogen bonds from bpy ligands of another two chains (
Table S2). Through these interactions, the 1D oxalate-bridged chains are self-assembled into a 3D supramolecular structure.
Bipyridine moieties of neighbouring chains stack in a zipper-like fashion, forming layers parallel to the plane (001) (
Figure 1;
Table S3).
Compound {[CrCu
3(bpy)
3(CH
3OH)(H
2O)(C
2O
4)
4][Cu(bpy)Cr(C
2O
4)
3]·CH
2Cl
2·CH
3OH·H
2O}
n (
2), crystalizing in
space group, comprises a 1D coordination anion [Cu(bpy)Cr(C
2O
4)
3]
nn− (
Cr2–Cu4) with alternating [Cr(C
2O
4)
3]
3− and [Cu(bpy)]
2+ units mutually bridged through an oxalate group, extending in the direction [100] (
Figure 2 and
Figure S2). Another chain (
Cr1–Cu3) is similar, but involves homodinuclear unit [Cu(bpy)(CH
3OH)(
µ-C
2O
4)Cu(bpy)(H
2O)]
2+ (
Cu1–Cu2) coordinated as a pendant group to a terminal oxalate oxygen (
Figure 2 and
Figure S2). In addition, the asymmetric unit contains one uncoordinated molecule of dichloromethane, methanol and water.
The copper(II) and chromium(III) atoms in the anionic chains [Cu(bpy)(
µ-C
2O
4)Cr(C
2O
4)
2]
nn−(
Cr1–Cu3 and
Cr2–Cu4) have a usual octahedral coordination: each Cr
3+ is coordinated by three oxalates, two having bis(bidentate) mode and one bidentate, and atoms Cu
2+ by four O atoms from two oxalate bridge and two N atoms of 2,2’-bipyridine (
Table S4). The Cu1 atom in the dimeric unit [Cu(bpy)(CH
3OH)(
µ-C
2O
4)Cu(bpy)(H
2O)]
2+ (
Cu1–Cu2) has a distorted square-pyramidal coordination with the bridging oxalate and a bpy moiety forming the basal plane; to the apical positions is bound a methanol molecule. Coordination of Cu2 is a severely Jahn–Teller distorted octahedron with the bridging oxalate and a bpy moiety forming the basal plane; a water molecule and oxalate oxygen atom O16 from the
Cr1–Cu3 unit are in apical positions. The bond Cu2–O16 is very elongated, its length being 2.712(2) Å (
Figure 2 and
Figure S2); it is significantly longer than the typical Cu–O covalent bond (1.98 Å) [
45], but it is shorter than the sum of the van der Waals radii (2.92 Å).
This polymer is the first structurally characterized compound in which copper(II) and chromium(III) centers are connected by bis(bidentate) oxalate group having one-dimensional arrangement [
4]; only one heterotrinuclear compound with these metals has been known, prepared by using the [Cr(C
2O
4)
3]
3− anion [
46]. Two dinuclear [
47,
48], one trinuclear [
49] and one tetranuclear [
50] oxalate-bridged compounds containing copper(II) and chromium(III) atoms were found in the literature, but they were synthesized without using tris(oxalato) building block.
The Cr1···Cu3 and Cr2···Cu4 distances across the oxalate bridges are 5.3236(5) and 5.3552(5) Å, respectively, which are significantly shorter than the corresponding one [5.4605(5) Å] in the known heterotrinuclear compound [
46]. The distance between copper(II) ions bridged by oxalate group in
Cu1–Cu2 unit is 5.1504(5) Å. It is somewhat longer than the analogous value (5.086 Å) in the most similar cation found in CSD [
4,
51].
Two polymeric chains are connected through hydrogen bonds: uncoordinated water molecule O32 acts as a proton donor towards two oxalate oxygens (O16 and O21) of two neighbouring symmetry-inequivalent chains (
Figure 3;
Table S2). A coordinated water molecule acts as a proton donor towards one oxalate oxygen (O27 of the chain
Cr2–Cu4) and towards the uncoordinated methanol, which in turn acts as a proton donor to O15 of the chain
Cr1–Cu3 (
Figure 3). Bipyridine moieties from neighbouring symmetry-inequivalent chains stack in a zipper-like fashion, forming layers parallel to the plane (011). (
Figure 4;
Table S3).
Compound {[CaCr
2Cu
2(phen)
4(C
2O
4)
6]·4CH
3CN·2H
2O}
n (
3) is a 3D coordination polymer comprising three different metal centers (Ca
2+, Cr
3+ and Cu
2+) oxalate bridged (
Figure 5 and
Figure S3). The chromium(III) atom is coordinated by six O atoms from three bridging oxalate moieties in octahedral fashion; calcium(II) is located on a twofold axis and is coordinated by eight O atoms from the four oxalates arranged as a dodecahedron, while copper(II) is coordinated by four N atoms of two phenanthroline ligands and one bridging oxalate (
Table S5).
Absolute configurations of Cr
3+ and Cu
2+ atoms is
Λ. Nodes of the 3D network are Ca
2+ atoms, which are connected with four Cr
3+ atoms through oxalate bridging ligands (
Figure 5); each [Cr(C
2O
4)
3]
3− group is bonded to two Ca
2+ atoms and to one [Cu(phen)
2]
2+ unit (
Figure 6 and
Figure 7). Thus the formed network can be reduced to an underlying graph of diamondoid (
dia) or (6
6) topology with Ca
2+ atoms as nodes and C
2O
4–Cr–C
2O
4 moieties as links (
Figure 5,
Figure 6 and
Figure S4). In the asymmetric unit one uncoordinated water molecule is present (which acts as a proton donor towards an oxalate oxygen;
Table S2) and two acetonitrile molecules. There are also π-interactions between phenanthroline moieties (
Table S3).
The distance between Cu2+ and Cr3+ metal centres bridged by the oxalate ligand is 5.5049(8) Å, while those between Ca2+ and Cr3+ across the oxalate bridges are 5.7043(7) and 5.7973(9) Å.
Fascinatingly, a search of the Cambridge Structural Database (CSD) [
4] does not find any oxalate-bridged compound containing a combination of these three metals, even non-oxalate-based.
Simultaneous thermogravimetric analysis (TG) and differential thermal analysis (DTA) have been used for studying the thermal properties of compounds
1–
3 as crystalline sample in nitrogen atmosphere, up to 1100 °C (
Figure S5, Table S6); due to various composition and structural arrangements studied systems show different thermal behavior. All compounds start to decompose almost immediately after the beginning of heating when the evacuation of crystal molecules happened. In all three compounds followed strong exothermic effects can be attributed to the release of
N-ligands and oxalate groups, when the main loss of mass takes place. The decomposition of
1 ends around 700 °C, when the constant mass is reached, with a black-coloured residue corresponding to Mn
2O
3. Probably due to a partial reduction of Cr
6+ to Cr
3+, followed by a small mass decrease and a weak endothermic maximum, the degradation of
2 and
3 ends around 960 and 900 °C, respectively, when the mass of the residue matches the mixture of oxides [
52].
3.4. Magnetization Study of Compounds 1–3
Temperature dependence of magnetization M(T) for all three compounds was measured in different magnetic fields, and no splitting between ZFC and FC curves was observed. Field dependence of magnetization M(H) at different temperatures did not show any hysteresis. Therefore, in both kinds of experiment no irreversibility was observed down to a temperature of 2 K. Moreover, no sharp peaks were observed, excluding thus magnetic long range order in compounds 1, 2 and 3.
Molar magnetic susceptibility corresponding to one manganese ion per formula unit (f.u.) in {[Mn(bpy)(C
2O
4)]·1.5H
2O}
n (
1) is shown in
Figure 8. The inverse susceptibility was fitted first with the Curie–Weiss law:
in order to extract preliminary information about the spins and their interactions; χ
D is a constant that includes all temperature independent contributions,
C is the Curie constant,
and
Weiss temperature that gives the measure of the sum of exchange couplings which determine the local effective field acting on every spin. Other parameters have their usual meaning. Obtained values of the parameters are χ
D = 0.00166 ± 0.00009 emu mol
−1 Oe
−1, 𝐶 = (4.37 ± 0.03) emu K mol
−1 Oe
−1, and
θCW = (−17.8 ± 0.6) K. The effective magnetic moment is
corresponding to the magnetic moment of the Mn
2+ ion in the high spin state (
S = 5/2 and
g = 2.00). The negative Weiss temperature suggests antiferromagnetic (AFM) interactions between the magnetic centres and the approximate value of exchange interaction of 𝐽 = −3.05 K, taking into account two nearest neighbours (
z = 2) bridged by the oxalate ligand along the structural chains as the dominant factor. The existence of AFM interaction could be seen from the
χ(𝑇) dependence, where relatively broad maximum around 15 K is present, indicating the low dimensional magnetic structure (
Figure 8).
Using the Fisher formula [
53] for magnetic chains consisting of large spins (here S = 5/2),
measured data were successfully fitted. The obtained value for the intra-chain super-exchange interaction between the neighbouring Mn
2+ ions is 𝐽 = (−3.26 ± 0.01) K and 𝑔= (2.050 ± 0.004). It is interesting to note that the value obtained from Curie–Weiss fit is nearly the same as from the Fisher model being a much more accurate description of this system. The Curie–Weiss model is based on figureions along the chain are taken into account, additionally proves the negligible inter-chain magnetic interaction.
Before plotting data (
Figure 8) the contribution of paramagnetic impurities and uncompensated Mn
2+ ions have been subtracted. This amount is
q = 4.08%, which is determined by fitting the Fisher formula (2) together with the paramagnetic term added:
with
as the spin of Mn
2+ ion, and χ
F is from Equation (2).
If we take into account also the inter-chain interactions, within the frame of the mean field approximation,
where generally χ
0 is the susceptibility function of non-interacting spin, an insignificant change in parameters and fit-quality with the value of inter-chain interaction of 𝑗′ = (−0.02 ± 0.01) K with z’ = 2 has been obtained. Therefore, the 1D chains containing Mn
2+ spin can be assumed to be magnetically independent. Magnetic interaction between the chains is negligible because of the large distance of Mn
2+ ions from neighbouring chains (more than 7Å).
Field dependence of magnetization,
M(
H), present in the inset of
Figure 8, is typical for the antiferromagnets—slow linear increase of magnetization with the field, far away from saturation. The value of magnetization at 2 K and 5 T is
, which is far lower than the expected value of
for the spin 5/2.
The magnetic properties of similar oxalate-bridged 1D coordination polymer of manganese(II) were also studied [
43] and interaction was found to be −1.72, K with inter-chain Weiss parameter of −0.04 K. Compared to the known compound,
1 seems as being closer to the ideal 1D Heisenberg antiferromagnetic chain with stronger intra-chain interaction. This is due to a slight difference in the structural packing of these two compounds, as a consequence of the existence of crystalline water in
1.
Chains containing manganese ions have been of interest for a long time, and they were synthesized using other superexchange-bridges. One example is chloride-bridged chain with dimethylammonium ions separating them [
54], having considerably stronger intra-chain interaction of −6.9 K and inter-chain interaction of −0.5 K. Although this
J indicates a stronger interaction than in
1, considerably weaker interaction between the chains (−0.02 K) makes our system much closer to the ideal 1D Heisenberg antiferromagnet.
Temperature dependence of susceptibility,
χ(
T), of compound {[CrCu
3(bpy)
3(CH
3OH)(H
2O)(C
2O
4)
4][Cu(bpy)Cr(C
2O
4)
3]·CH
2Cl
2·CH
3OH·H
2O}
n (
2) in field of 1000 Oe (inset in
Figure 9) shows no phase transition, but the rapid/sharp increase at temperatures below 30 K indicates the presence of ferromagnetic interactions (FM). From the crystal structure we can assume that it belongs to low-dimensional magnetic structures, i.e., magnetic chains of copper(II) and chromium(III) ions bridged by oxalate ligand. Fitting the reciprocal susceptibility data with Curie-Weiss law, equation (1), following parameters: 𝐶 = (4.43 ± 0.02) emu K mol
−1 Oe
−1,
, and
emu mol
−1 Oe
−1 have been obtained. One formula unit contains 4 Cu
2+ and 2 Cr
3+ ions, so that the Curie constant should be
emu K mol
−1 Oe
−1, where
g-factors
, and
have been taken [
55]. Measured value is somewhat lower, indicating that not all of the spins participate in this sum of free paramagnetic ions’ contributions. This discrepancy between calculated by the Curie constant and the observed value will be explained below, by analysing the temperature-dependence of
. However, a positive Weiss temperature surely confirms the existence of FM interactions with the average intra-chain interaction of the order between 1 and 10.
In
Figure 9, the temperature dependence of
shows the minimum at 100 K, where, after the initial rapid decline with increasing temperatures,
starts to slowly increase. This increase can be explained by looking more closely at the crystal structure of compound
2; one formula unit contains 4 Cu
2+ and 2 Cr
3+ ions, from two chains [Cu(bpy)(
µ-C
2O
4)Cr(C
2O
4)
2]
nn−(
Cr1–Cu3 and
Cr2–Cu4) and one dimeric unit [Cu(bpy)(H
2O)(
µ-C
2O
4)Cu(bpy)(CH
3OH)]
2+ (
Cu1–Cu2) (
Figure 2). This homodinuclear cation can be regarded as magnetically independent from the other magnetic centres, since a terminal oxalate oxygen atom from the
Cr1–Cu3 chain coordinated to copper(II) ion of this dinuclear unit, does not provide the exchange bridging path. Such dimers usually have a strong antiferromagnetic interaction [
26,
27,
56,
57], so that the spin of a dimer is 0. Assuming that is the case here, the increase of
as the temperature rises can be understood as the start of the decoupling of a dimer with very strong antiferromagnetic exchange intradimer interaction (few hundreds of Kelvins). From this viewpoint, the underestimated Curie constant obtained from the Curie–Weiss fit is understandable. By measuring the higher temperatures, where the dimer (
Cu1–
Cu2) is decoupled, the calculated value of Curie constant would be obtained, but then the compound would not be thermally stable. In
Figure 9, the contribution of the decoupling dimer is indicated with the ΔM(Cu–Cu), and it can be seen that at 300 K the
does not yet achieve the value where all the spins in the formula unit are decoupled (2 Cr
3+ and 4 Cu
2+ ions per f.u.), but as it still rises it can be assumed that at higher temperatures it will saturate at this value.
Field dependence of magnetization,
M(
H), shows rapid increase of magnetization at lower fields and saturation at high fields (
Figure 10). From the value of saturation magnetization,
, it could be concluded that the ground state spin projection in one formula unit is 4. This shows that all spins in both chains (
Cr1–Cu3 and
Cr2–Cu4) point in the same direction (direction of applied magnetic field) in their ground state and confirms the persistence of strong AFM dimers (
Cu1–Cu2). To check this assumption we have plotted (
Figure 10) the Brillouin function for different cases, first for the case in which all magnetic ions are magnetically independent (red dashed line), then the case where the spin in dimer is 0 and two Cr
3+ and two Cu
2+ ions are independent (green dash-dot line) and, finally, the case where we have one spin of 4 in formula unit (Brillouin function for spin S = 4) (blue line). The best agreement with the data shows the Brillouin function for spin 4, but the measured susceptibility is still above the Brillouin function, indicating strong enough FM interactions to produce large spin correlations along [Cu(bpy)(
µ-C
2O
4)Cr(C
2O
4)
2]
nn−chains. In the inset of
Figure 10 the field dependence of magnetization
M(
H) is shown for temperatures 2, 5, 10 and 30 K and, as expected, the field dependence is becoming linear with increasing temperature, but even at temperature of 30 K the measured magnetization is higher than that for a case of paramagnetic spins of two Cr
3+ and two Cu
2+ (pink line in the inset of
Figure 10 represents the sum of Brillouin functions for Cr
3+ and Cu
2+ ions), and this difference comes from the finite
M(
H) response of
Cu1–Cu2 dimer.
From the crystal structure of compound {[CaCr
2Cu
2(phen)
4(C
2O
4)
6]·4CH
3CN·2H
2O}
n (
3) (
Figure 5), it can be assumed the existence of magnetic dimers of Cu
2+ and Cr
3+ ions connected through the oxalate bridge, since Ca
2+ is diamagnetic, thereby stopping the propagation of further exchange pathways. Temperature dependence of susceptibility shows no phase transition and
suggests the ferromagnetic interaction between magnetic moments (
Figure 11). Magnetic measurements were modelled using PHI software (Chilton Group, Manchester, UK.) [
58] with Hamiltonian for a [Cu
IICr
III] dimer:
where the zero-field splitting (ZFS) of Cr
3+ was also taken into account with component characteristic for axially distorted octahedral coordination. It has been found that the exchange constant is
, with
g-factors
,
, and the axial ZFS parameter
. An additional check of this fit was performed with a program developed in Python, which was as successful as in other complex cases [
59]. Ferromagnetic super exchange in Cu
II–C
2O
4–Cr
III magnetic units can be understood within the orthogonality of the magnetic orbitals. Namely, Cu
2+ ion has an axially distorted configuration, and one unpaired electron on the
orbital oriented toward the 4 nearest atoms (N1, N2, O2 and N4) in the coordination environment (
Figure S3). It interacts with the occupied orbitals of the oxalate group producing a magnetic orbital of
σ-character. Ion Cr
3+ in the octahedral configuration has three unpaired electrons on the
,
and
orbitals interacting with other occupied orbitals of the oxalate group, producing magnetic orbitals of
-character. The overall superexchange interaction is expected to be ferromagnetic due to orthogonality of all these magnetic orbitals [
48]. In other compounds with oxalate-bridged [Cu
IICr
III] dimer where the symmetry is such that the ferromagnetic coupling can be achieved due to orthogonality of magnetic orbitals, the ferromagnetic exchange is of comparable value [
46,
47,
48].
Using the mean field approach within PHI software for interactions between the dimers (Equation (4)), it was found that the small intermolecular interaction between dimers does exist, in amount
. This could mean that despite the large distance between dimers and the presence of diamagnetic Ca
2+, dimers still interact and interaction could come through the small but finite spin polarization of Ca
2+ orbitals which could produce super-exchange over this nearly diamagnetic bridge, as was already observed and also studied theoretically in the –Cr
III–O–Nb
V–O–Cr
III– complex [
60,
61]. This motivates further investigation and the calculation of the spin densities, which is beyond the scope of the present work.
The existence of ferromagnetic interaction could also be seen from the field dependence of magnetization, inset in
Figure 11, as the measured magnetization at 2K (black dots), is above the values expected for the independent spins and given by the sum of Brillouin functions for spin 1/2 and 3/2 (pink line). As expected, with the increasing temperature, the difference between the measured magnetization and the magnetization of the independent spins decreases, but is still present at 30 K, meaning that dimers are not completely decoupled yet.