Conversion of Metal Pyrazolate/(Hydr)oxide Clusters into Nanojars: Solution vs Solid-State Structure and Magnetism

Nanojars are a class of anion binding and extraction agents composed of a series of [Cu(μ-OH)(μ-pz)]n (pz = pyrazolate; n = 26–36) supramolecular metal–organic complexes. In contrast to other anion binding agents amenable to liquid–liquid extraction, nanojars only form by self-assembly around the target anion, and guest-free nanojar hosts cannot be isolated. An extraordinary binding strength toward highly hydrophilic anions such as carbonate and sulfate was demonstrated by the inability of Ba2+ ions to precipitate the corresponding insoluble barium salts from nanojars. Herein, we provide an additional proof for the superior robustness of the nanojar framework based on competition experiments with other transition metal pyrazolate/(hydr)oxide complexes. In addition to the mass spectrometric characterization, we present variable-temperature nuclear magnetic resonance studies with an emphasis on the influence of the paramagnetic Cu2+ centers on 1H hyperfine shifts, along with X-ray crystallographic analysis of two polymorphs of (MePh3P)2[CO3⊂{Cu(OH)(pz)}27], including the highest (cubic) symmetry nanojar crystal lattice obtained to date as well as magnetism studies for the first time. Furthermore, we provide evidence for the first molybdate-incarcerating nanojars, [MoO4⊂{Cu(μ-OH)(μ-pz)}n]2– (n = 28, 31–33), formed by rearrangement from [MoVI8O12(μ-O)9(μ-pz)6(pzH)6·3pzH] in the presence of Cu2+ ions.


■ INTRODUCTION
−5 The three (or four) 1 charge-neutral constituent [cis-Cu II (μ-OH)(μ-pz)] x (Cu x ; x = 6−14, except 11) metallamacrocycles (rings) of nanojars are bound together by a multitude of hydrogen bonds and axial Cu−O interactions.The anion bound in the cavity of the nanojar by hydrogen bonding to its two smaller side rings appears to be an essential, additional supramolecular "glue" that ultimately allows for the formation of the nanojar host because an empty, guest-free nanojar could never be observed.As anion binding and extraction agents, nanojars exhibit an inverse Hofmeister bias 6,7 and preferentially bind anions with large hydration energies, such as CO 3  2− (ΔG h °= −1324 kJ/mol) and SO 4   2− (ΔG h °= −1064 kJ/mol). 8As measurements of binding constants by host−guest titration are precluded by the inability to isolate the empty host, the strength of anion binding by nanojars was assessed by competitive binding studies using Ba 2+ .Experiments conducted under both heterogeneous [using an aqueous solution of Ba(NO 3 ) 2 and an immiscible, dichloromethane or 2-methyltetrahydrofuran solution of nanojars] and homogeneous (using 2-methyltetrahydrofuran solutions of barium dioctyl sulfosuccinate and nanojars) conditions suggest very strong binding 3,9 as the incarcerated anion did not precipitate out as the highly insoluble BaCO 3 or BaSO 4 salts (K sp = 2.58 × 10 −9 and 1.08 × 10 −10 , respectively, in H 2 O at 25 °C). 10o further investigate the robustness of the nanojars, we hypothesized that other metal pyrazolate/(hydr)oxide clusters would transform into nanojars in the presence of Cu 2+ ions if the nanojars are more stable.Thus, we identified five known transition metal clusters that contain pyrazolate and (hydr)oxide ligands and tested their reactivity toward Cu 2+ ions (Figure 1).In all cases, the metal cluster was transformed into a mixture of (Bu   Structural Analysis by X-ray Crystallography.Two types of crystals (prisms and cubes) were obtained by vapor diffusion of pentane into a p-xylene/nitrobenzene solution of Cu n CO 3 (Bu 4 N + salt) in the presence of MePh 3 P + Br − as an additive, which correspond to two polymorphs of (MePh 3 P) 2 [CO 3 ⊂{Cu(OH)(pz)} 27 ]: one triclinic, P1̅ (1), with the nanojar unit on a general position, and the other cubic, I23 (2), with the nanojar unit centered on a C 3 rotation axis perpendicular to the Cu x (x = 6, 9, and 12) rings (Table S1 and Figures 3 and 4).Upon crystallization, both original Bu 4 N + counterions were replaced by MePh 3 P + in 1.In 2, at least one Bu 4 N + unit (adjacent to the Cu 6 ring) is replaced by a MePh 3 P + .The identity of the second counterion in 2 could not be determined due to extensive disorder with solvent molecules, but it is likely also MePh 3 P + due to the additional aromatic interactions it provides both with pz moieties and with other MePh 3 P + counterions within the crystal lattice.In 1, the second MePh 3 P + unit (adjacent to the Cu 9 ring) does interact with a neighboring MePh 3 P + unit across an inversion center, leading to a parallel pair of Ph groups [crystallographically imposed; centroid−centroid distance: 3.667(5) Å, plane-centroid distance: 3.293(6) Å] (Figure S4).In addition to the C 3 rotational symmetry, the nanojar framework in both 1 and 2 is close to being symmetrical relative to three σ mirror planes running along the C 3 axis at 60°from each other.
The structures of the nanojar frameworks in the two polymorphs 1 and 2 are similar to each other and to the one in a pseudopolymorph of 1 obtained earlier from toluene/ cyclohexane 4 and Bu 4 N + counterions, with slight differences observed in the geometry of the Cu x rings and the orientation of the CO 3 2− ion within the nanojar cavity (Figures 3 and 5, Tables S2−S9).These differences, as well as the loss of the C 3 symmetry in 1, appear to be caused by the orientation of the MePh 3 P + counterions relative to the nanojar.In both 1 and 2, one MePh 3 P + counterion is positioned close to the Cu 6 ring.As seen in Figure 3, however, the three phenyl groups of MePh 3 P + are arranged symmetrically above the Cu 6 ring only in 2. Here, the three equivalent Ph groups form one type of aromatic interaction with one set of the Cu  occupancy).Consequently, a different pattern of aromatic interactions arises in compound 1.The overall packing of nanojars in the crystal lattice is also very different in 1 compared to that in 2 (Figure 6).A less compact packing is observed in the case of the cubic symmetry lattice of 2, which displays large interstitial voids, also illustrated by the lower corresponding calculated crystal density of 1.395 vs 1.683 g• cm −3 in the case of 1.
The structural parameters (Cu−O and Cu−N bond lengths, trans and cis N−Cu−O angles, and Cu•••Cu distances) within the Cu n -rings of 1 and 2 are practically identical (Table S2).The positions of the Cu 9 ring relative to that of the Cu 12 ring and the bonding between these two rings in 1 and 2, however, are different.In 2, there are only three axial Cu  S2).Perhaps the most significant difference between 1 and 2 is the position of the incarcerated CO 3 2− ion, which is centered in a well-defined position within the nanojar cavity in 2 [12H bonds to the Cu 6 and Cu 9 rings shorter than 3.2 Å, averaging 2.838( 6) Å] but is disordered over two positions (0.82/0.18 occupancy) in 1 [for the major unit: 13H bonds to the Cu 6 and Cu 9 rings shorter than 3.2 Å, averaging 2.880( 6) Å].
To compare the overall shapes of the nanojar units in 1 and 2, the dihedral angles and the component fold and twist angles between the mean planes of pyrazolate moieties and adjacent Cu−O−Cu units (as defined earlier) 17 were analyzed.As shown in Table S3, the corresponding angle averages for the different Cu x rings are rather consistent in 1 and 2, although some significant differences are observed [notably for the twist angles in the Cu 6 rings, which vary from 0.91(16) to 10.30 (17)°in 1 compared to 0.2(3) to 1.7(2)°in 2].Furthermore, the dihedral angles and the component fold and twist angles between the mean planes of adjacent pyrazolate moieties were analyzed (Table S4).The most notable difference is for the twist angles in the Cu 9 rings, which range from 0.6(3) to 48.0(2)°in 1 compared to 11.2(3) to 49.8(7)°in 2. In terms of the geometry of the Cu x rings in 1 and 2, significant differences are observed for the deviations of the various Cu atoms from the Cu x mean planes (Table S5).
It is noteworthy that the non-centrosymmetric, bodycentered cubic space group I23 (prototype: Ga 4 Ni) 18 observed in the case of 2 is very rare: only 232 (0.018%) out of a total of almost 1.3 million structures in the Cambridge Structural Database (CSD) belong to this space group (2.7% of cubic structures, which represent 0.67% of all structures). 19The corresponding Schoenflies point group is T, with four C 3 and three C 2 axes as the symmetry elements.Within the crystal lattice of 2, nanojar units are centered on the C 3 axes and inbetween the C 2 axes.
Variable-Temperature 1 H NMR Characterization.For the NMR studies of carbonate nanojars, a sample of (Bu 4 N) 2 [CO 3 ⊂{Cu(OH)(pz)} n ] (n = 27, 29−31) was prepared directly from Cu(NO 3 ) 2 •2.5H 2 O, Bu 4 NOH, NaOH, and Na 2 CO 3 in THF. 4 The distribution of the nanojars of different sizes in this sample is shown in Figure S3.A sample of pure (Bu 4 N) 2 [CO 3 ⊂{Cu(OH)(pz)} 27 ] (used for the magnetism studies detailed in the next section) was obtained from a THF solution of Cu n CO 3 (n = 27, 29−31) by precipitation with hexanes.In contrast to the solid-state structure, which allows for C 3 as maximum rotational symmetry in the case of Cu 27 CO 3 (Figure 4), 1 H NMR indicates that in solution, all pz  moieties as well as OH groups are identical within each Cu x ring with the exception of the Cu 12 ring (Figure 7).For the latter, two sets of signals are observed for both the pz and the OH groups, due to the six of them being oriented toward the Cu 6 ring, whereas the other six are oriented toward the Cu 9 ring.Despite the paramagnetism of the Cu 2+ centers, which causes a drastic downfield shift of the pz proton peaks and an even more drastic upfield shift of the OH proton peaks, as well as a broadening of the peaks and loss of the J coupling between nuclei, relatively sharp peaks are observed in the 1 H NMR spectrum (especially for the pz protons, which are three bonds away from the Cu 2+ centers, whereas the OH protons are only two bonds away).This suggests strong antiferromagnetic interactions between Cu atoms of the nanojar.Indeed, magnetic susceptibility measurements described in the following section confirm the strong antiferromagnetic coupling in Cu 27 CO 3 .As seen in Figure 7, the extent of this coupling leading to a reduction of spin density is different for the various Cu x rings.The odd-membered Cu 9 ring appears to possess the largest spin density, indicated by the most prominent hyperfine shifts and the fastest relaxation.
To offer additional data complementing earlier 1 H NMR characterizations of the Cu n CO 3 mixture (n = 27, 29−31), 4 we now present individual chemical shift values for the variabletemperature 1 H NMR measurements in DMSO-d 6 over the 22−150 °C range (Table S10).The additional spectrum collected at ambient temperature after cooling the sample from 150 °C confirms that the nanojar mixture survives heating to 150 °C in DMSO-d 6 for 1 h without decomposition (Figures 8  and 9).However, the composition of the mixture changes.Initially, the mixture consists of substantial amounts of Cu 27 CO 3 , Cu 29 CO 3 (Cu 7+13+9 CO 3 , where 7, 13, and 9 indicate the component ring sizes), and Cu 31 CO 3 and small amounts of Cu 29 CO 3 (Cu 8+13+8 CO 3 ) and Cu 30 CO 3 .Upon heating, the larger nanojars (Cu 31 CO 3 and Cu 30 CO 3 ) gradually decompose and convert to Cu 8+13+8 CO 3 , which becomes the dominant species at 150 °C.The amount of Cu 7+13+9 CO 3 is also reduced at higher temperatures, although peaks corresponding to this nanojar are still discernible at 150 °C.
The NMR studies of the Cu n CO 3 nanojar mixture also reveal that in addition to the coupling between Cu atoms within Cu x rings, there is coupling between Cu x rings as well, indicated by the very different chemical shifts of the Cu 9 ring
Variable-temperature NMR experiments conducted on Cu n CO 3 (n = 27, 29−31) indicate different temperature dependence of the chemical shifts (δ) for different Cu x rings (Figure 10).The peaks corresponding to the Cu 9 (and to a lesser extent Cu 7 and Cu 8 ) rings display large variations with temperature, becoming less paramagnetically shifted at higher temperatures, which indicates Curie behavior (δ ∝ 1/T).The δ values of all other Cu x rings show only slight variations with temperature, except for the OH signal of the Cu 13 ring in Cu 7+13+9 , which shows anti-Curie behavior becoming more paramagnetically shifted at higher temperatures.This last observation, which suggests transfer of spin density from one Cu x ring to another at higher temperatures, along with the fact that the δ values of Cu 9 rings in different nanojars display different temperature dependences (for example, the OH signal of Cu 9 ring spans ∼14 ppm in Cu 6+12+9 but only ∼7 ppm in Cu 7+13+9 in the 22−150 °C range) as well as different hyperfine shifts, is further evidence for the magnetic coupling between Cu x rings in nanojars.The corresponding Curie plots of the 1 H NMR data, obtained by plotting δ as a function of 10 3 /T, are shown in Figure S7.
Magnetic Properties.The DC magnetic susceptibility data for (Bu 4 N) 2 [CO 3 ⊂{Cu(OH)(pz)} 27 ] are shown as χ M T vs T from 310 to 1.8 K (under an external field of 0.1 T) in Figure 11.The χ M T value at 300 K is about 4.8 cm 3 mol −1 K, which corresponds to an effective magnetic moment of 6.2 μ B .This value is significantly lower than the theoretical value of μ eff = 9.0 calculated for 27 noninteracting Cu 2+ ions (assuming g = 2).The χ M T value steadily decreases as the temperature drops and finally reaches a value of 0.37 cm 3 mol −1 K at 1.8 K (μ eff = 1.7), which suggests a possible S = 1/2 ground state.Thus, the magnetic susceptibility measurements indicate strong antiferromagnetic coupling between the Cu 2+ ions, which cannot be overcome by thermal perturbation, even at room temperature.Both pyrazolate and O(H) ligands are known to mediate such coupling. 20The field-dependent magnetization increases from 0 to 5 T due to the field-induced population of states with higher spins (Figure 12).The close to linear increase in magnetization without saturation (corresponding to an S = 1/2 spin ground state) at high magnetic fields suggests spin frustration (competing antiferromagnetic interactions that cannot be simultaneously satisfied).Similar magnetic behavior was observed in a Cu II 27 cluster based on o-vanillinoxime, 21 as well as in Cu II 12 and Cu II 20 keplerates. 22For [CO 3 ⊂{Cu(OH)(pz)} 27 ] 2− , both an S = 1/2 ground state and spin frustration can be rationalized based on its structure comprising three Cu x rings (x = 6, 9, and 12).Thus, in the even-membered Cu  A more compact packing of nanojar units is observed in this cubic crystal lattice than in its lower symmetry (triclinic, P1̅ ) polymorph.As suggested by the NMR experiments in solution, strong antiferromagnetic coupling between the paramagnetic Cu 2+ centers is observed in the case of (Bu 4 N) 48 mmol) were added, and the reaction mixture was stirred in a stoppered flask at room temperature for 3 days.Then, the reaction mixture was filtered and added to hexanes (500 mL) under stirring.The blue precipitate was filtered out and rinsed with hexanes.Yield: 0.7501 g.The purity of the sample was confirmed by ESI-MS (Figure S3) and 1 H NMR (Figure 7).
Mass Spectrometry.Mass spectrometric analysis of the nanojars was performed with a Waters Synapt G1 HDMS instrument using electrospray ionization (ESI).10 −4 −10 −5 M solutions were prepared in CH 3 CN using either solids or aliquots taken from solutions.Samples were infused by a syringe pump at 5 μL/min, and nitrogen was supplied as the nebulizing gas at 500 L/h.The electrospray capillary voltage was set to −2.5 or +2.5 kV, respectively, with a desolvation temperature of 110 °C.The sampling and extraction cones were maintained at 40 and 4.0 V, respectively, at 80 °C.Reported m/z values represent averages of the observed isotopic distributions.
Magnetism.Magnetic data were collected on a Quantum Design MPMS-XL SQUID magnetometer equipped with a 7 T magnet.DC magnetic susceptibility measurements were carried out under a 1000 Oe magnetic field from 310 to 1.8 K, and magnetization measurements were operated at 5 K with an external field from 0 to 5 T. The polycrystalline sample used for the measurements was embedded in solid eicosane to prevent torqueing.Pascal's constants were used to determine the magnetic correction, which was subtracted from experimental data to obtain the molar susceptibility.
X-ray Crystallography.Single crystals of 1 and 2 were grown at room temperature by vapor diffusion of pentane into a nitrobenzene/ p-xylene solution of Cu n CO 3 (n = 27, 29−31) in the presence of MePh 3 PBr.Once removed from the mother liquor, the crystals are extremely sensitive to solvent loss under ambient conditions and were quickly mounted under a cryostream (150 K) to prevent decomposition.X-ray diffraction data were collected from a single crystal mounted atop a MiTeGen micromesh mount under Fomblin oil with a Bruker AXS D8 Quest diffractometer equipped with a Photon II charge-integrating pixel array detector (CPAD) using graphite-monochromated Mo-K α (λ = 0.71073 Å) radiation.The data were collected using APEX4, 23 integrated using SAINT, 24 and scaled and corrected for absorption and other effects using SADABS. 25The structures were solved by employing direct methods using ShelXS 26 or ShelXT 27 and refined by full-matrix least-squares on F 2 using ShelXL. 28C−H hydrogen atoms were placed in idealized positions and refined by using the riding model.Further refinement details and thermal ellipsoid plots (Figures S5 and S6) are provided in the Supporting Information.

Figure 1 .
Figure 1.Illustration of the structure of various metal pyrazolate/(hydr)oxide clusters transformed into nanojars in the presence of Cu 2+ ions.

Figure 3 .
Figure 3.Comparison of the crystal structures of (MePh 3 P) 2 [CO 3 ⊂{Cu(OH)(pz)} 27 ] in 1 and 2 (top and sideviews).Green and blue dotted lines indicate hydrogen bonds and axial Cu•••O interactions, respectively.Lattice solvent molecules, C−H bond H atoms, and one of the counterions are omitted for clarity, and only the major component is shown for disordered moieties.C atoms of MePh 3 P + are shown in green.

Figure 4 .
Figure 4. Illustration of the 3-fold symmetry of the nanojar scaffold in 2.

Figure 5 .
Figure 5.Comparison of the Cu 6 , Cu 12 , and Cu 9 rings in 1 and 2 (only Cu atoms are shown).

Figure 6 .
Figure 6.Comparison of the packing diagrams (along the a axis) of 1 and 2. C−H bond H atoms, counterions, and solvent molecules are omitted for clarity, and only the major component is shown for disordered moieties.

Figure 7 .
Figure 7. 1 H NMR spectrum in DMSO-d 6 at ambient temperature of pure (Bu 4 N) 2 [CO 3 ⊂{Cu(OH)(pz)} 27 ].Inset shows a close-up of the OH region of the spectrum as well as a drawing of the Cu 6 ring, illustrating the equivalency of the pz and OH groups within Cu x rings.

Figure 8 .
Figure 8. Variable-temperature 1 H NMR spectra of the Cu n CO 3 (n = 27, 29−31) nanojar mixture in DMSO-d 6 , showing pyrazolate proton signals in the 21−38 ppm window.The temperatures shown are the target temperatures of the probe.

Figure 10 .
Figure 10.Temperature-dependent variation of the chemical shifts (δ) of different Cu x ring protons in various Cu n CO 3 nanojars in DMSO-d 6 .