graphic data is shown in Table 1. In each case, the structure is composed of a novel metallocyclic core {M 2 V 2 Cl 2 N 4 O 8 } (M=Co, Zn) incorporating four DEA ligands in two distinct ligating modes (Figure 1). The core of each cluster consists of two distorted {VO 5 N} octahedra sharing a single edge; and two distorted {MO 3 NCl} trigonal bipyramidal units, each sharing an edge with both vanadium octahedra (Figure 1b). Each vanadium octahedra contains the nitrogen atom (VÀN, 2.138 Å, 1; 2.121 Å, 2) and four bridging oxygen atoms of the {HN-(CH 2 ÀCH 2 ÀO) 2 } ligands; and a terminal oxygen atom (V=O t ; VÀO, 1.610 Å, 1; 1.593 Å, 2), which is a structural feature characteristic of polyoxoanions. The distorted trigonal bipyramidal geometry around the M (M=Co, Zn) centers is defined by a peripheral terminal chlorine atom (MÀCl, 2.282 Å, 1; 2.236 Å, 2); each M is also bound to the nitrogen atom (MÀN, 2.172 Å, 1; 2.127 Å, 2) and a single oxygen atom of the {HN-(CH 2 ÀCH 2 ÀO)(CH 2 ÀCH 2 ÀOH)} ligand (Figure S1 in the Supporting Information). The inorganic core may be described as composed of two face-sharing Cubans, each missing a vertex. The coordination geometries of cobalt and vanadium and orthogonal views of the cluster compound in 1 with space-filled representations of cobalt and vanadium are shown in Figure 2.

The isostructural clusters 1 and 2 crystallize in the monoclinic space group P2 1 /n. In the crystalline state, the clusters are joined by hydrogen bonds to form a 3D network structure (Figures 3 andS3). Each secondary amine ligand of M (M=Co, Zn) acts as a (hydrogen) donor to a hydroxy group oxygen of a neighboring cluster to form a hydrogen bond (N1ÀH1C•••O4). Each chlorine atom acts as a bifurcate hydrogen acceptor, accepting a hydrogen atom from a secondary amine ligand of V (N2ÀH2 C•••Cl1) on a neighboring cluster, while also In analogy to Type III POMs, which contain nucleophilic terminal oxygen atoms, the terminal chlorine atoms of 1 and 2 should coordinate a variety of transition metals, [16] facilitating (electro-)catalytic activity and the formation of coordination polymers. Much of the application scope of active clusters relies on cluster heterogenation, often through electrostatic cluster-support interactions. [17] The chemical topology of the clusters 1 and 2 anticipates facile heterogenation through soft chemical methods.

The infrared absorption spectra of 1 and 2 (Figures 4 andS4) contain bands associated with the oxometallate core and the DEA ligands. The likeness of the spectra is consistent with isostructural molecules differing only in the identity of a single atomic constituent. Typical IR absorption spectra of polyoxometalates contain strong bands associated with stretching modes of the terminal metal-oxygen bond (M=O t ), and weaker bands associated with bridging oxygen (MÀO b ÀM) vibrational modes. Band assignments are facilitated by spectral comparisons to previously studied DEA [18] and triethanolamine (TEA) [19] functionalized hexavanadate structures containing vanadium coordination octahedra like that observed in 1 and 2. The strong band at around 958 cm À1 is attributed to the symmetric stretch of the terminal vanadium-oxygen bond (ν sym V=O t ). 6 O 6 (OCH 2 CH 2 ) 2 NH) 6 ] + (DEA-NaV 6 ) [18] reveals distinct bands at 989, 642, 562, and 497 cm À1 . Because terminal bonds are expected to have the most (nearly) pure stretching vibrations, and thus the highest IR absorption, we provisionally attribute the strong band at 989 cm À1 (Figure 4) to the symmetric stretch of the CoÀCl bond of the Co-centered polyhedra. Comparison of the IR spectrum of 2 to that of DEAÀNaV 6 reveals distinct bands at 985, 639, 564, and 489 cm À1 . We provisionally attribute the band at 985 cm À1 (Figure 4) to the symmetric stretch of the ZnÀCl bond of the Zn centered polyhedra.

Bond valence sum calculations [20] indicate metal oxidation states of V IV , Co II , and Zn II , consistent with the X-ray structural data and charge balance requirements of 1 and 2. Polyoxovanadates often occur as mixed valence (e. g., V IV ÀV V ) species and exclusively reduced vanadium (V IV ) is often observed in lownuclearity (e. g., n � 6) clusters functionalized with organic ligands, for example. [18,19,21] TGA profiles (Figures S6 andS7) of 1 and 2 overlay well until around 330 °C; decomposition initiates around 300 °C. This result is consistent with the identical nature of the organic moieties contained in each compound. 1 and 2 have comparable thermal stability to previously studied DEAand TEA-functionalized hexavanadate compounds. [18][19] Divergence of the TGA profiles following initial decomposition implies that heteroatom identity (Co/Zn) has a marked effect on the chemical transformations accompanying thermal decomposition. (See the Supporting Information for a detailed analysis of the TGA.)

Both compounds are stable in air as confirmed by FTIR. Reflectance spectra covering the UV, visible and a portion of the near-IR range of compounds 1 and 2 are plotted as pseudoabsorbances (1-R) in Figure 5. Broad bands are present in the UV, vis and NIR ranges for both compounds. In POMs, UV bands are associated with charge transfer processes between terminal oxygen atoms and (usually) octahedral metal centers; these are present in the V IV octahedra of 1 and 2. The redox activity of certain POM species is attributable to this process. [1b] We previously found both UV and visible absorption bands in hexavanadate clusters containing V IV coordination octahedra like those found in 1 and 2. [18][19] The shape and breadth of the UV absorption bands suggests an additional contribution from MÀCl (M=Co, Zn) charge transfer processes. Visible bands are associated with d-d and intervalence transitions arising from reduced metal cations (e. g., V IV ). [22] In this regard we observe that compound 1, containing (formally) Co (d 7 ) has a broader visible absorption spectrum than compound 2, which contains zinc with a (formally) full d-subshell. Thus the spectra are consistent with the structure and metal oxidation states of the compounds.

The magnetic susceptibilities of 1 and 2 are shown in Figure 6a. The exchange couplings are defined in Figure 6b, and axes systems for the V IV and Co II centers are found in Figure 6c. The susceptibility for 1 was successfully modeled (EasySpin [23] ) with

x,y,z = 2.03, 2.57, 2.52, J 1 = À5.35 cm À1 , J 2 ' = À12.58 cm À1 , J 2 '' = À7.48 cm À1 , D Co II = 25.1 cm À1 , and E/D = 0.04. The spin Hamiltonian (SH) parameters were also reasonably reproduced with multi-reference NEVPT2 calculations (see the Supporting Information for details) performed with ORCA. [24] The computationally predicted parameters were g V IV iso = 1.96, g Co II

x,y,z = 2.02, 2.39, 2.34, J 1 = À1.39 cm À1 , J 2 ' = À13.02 cm À1 , J 2 '' = À4.11 cm À1 , D Co II = 39.1 cm À1 , and E/D = 0.06. Furthermore, these parameters can be predicted to be g V IV iso = 1.96, g Co II x,y,z = 2.00, 2.42, 2.26, D Co II = 29.3 cm À1 , and E/D = 0.24 using ligand-field theory based on energies and spin-orbit coupling constants derived from ab-initio ligand field theory (AILFT). For 2, the experimental/computational/LFT SH parameters were found to be g V IV

x,y,z = 1.98, 1.99, 2.00/1.93, 1.98, 1.98/1.92, 1.98, 1.98 and J = À3.91/ À1.9 cm À1 . The isotropic g value extracted from the susceptibility is standard for V IV centers in pseudo-octahedral environments. In particular, the value of 1.95 observed here agrees nicely with the EPR-derived value of 1.97 observed on a tetravanadate system by Plass. [25] The exchange couplings observed (and predicted) for both 1 and 2 are all antiferromagnetic.

The anti-coplanar orientation of the V IV centers in these systems are qualitatively expected to favor weak antiferromagnetic interactions based on the structural features of the molecule. We observe that the V=O t bonds are not purely coplanar with an O trans ÀVÀO t angle of � 169.7 and � 170.4 for 1 and 2 respectively. We also observe that the V IV 3d xy orbitals are twisted as evidenced by the NÀVÀOÀV torsions deviating from 180°by 20°and 19°for 1 and 2 respectively. These deformations result in non-zero overlap between the V IV magnetic orbitals which is supported by the corresponding orbital overlaps of 0.015/0.017 observed for 1 Zn /2 with broken symmetry PBE0/DKH-DEF2-QZVPP(V,Zn,Co)/DKH-DEF2-TZVP/ DKH solutions, where 1 Zn substitutes the Co II centers with Zn II centers. Our observed values for the VÀV exchange interactions are similar to that observed by Lu et al. on an anti-coplanar V IV dimer. [26] This work can also be compared with the work of Stuckart et al., [27] who studied the magnetic exchange interactions between octahedral V IV centers and square pyramidal V IV centers through a WO 4 2À unit. Interestingly, they observed an antiferromagnetic exchange coupling of À5.35 cm À1 , which is of similar magnitude to the value observed in this work. This highlights that the overlap between the 3d xy in our systems are weak enough to dampen the exchange interaction to the degree that it makes the superexchange pathway nearly equivalent to WO 4 2À pathways. All of the interactions reported in this work and referenced for comparison are significantly below the values regularly observed for bis(μ-OH) bridged V IV systems. [28] The interactions between the Co II centers and the V IV centers in 1 are inequivalent as predicted from the experimental structure where the CoÀV distances are 3.18 Å and 3.29 Å. It can also be observed that the superexchange pathway for the different exchange couplings (J 2 ', J 2 '') are inequivalent as illustrated in Figure 7. Inspection of Figure 7 suggests that J 2 ' will be stronger than J 2 '' due to the enhanced overlap between the V-3d xy orbital and the sp 2 hybridized μ2-O. Fraser et al. recently reported a Co/V Anderson wheel complex that likewise showed divergence of the exchange couplings between neighboring cobalt centers and adjacent edges on vanadium centers. [29]

In conclusion, we have prepared and fully characterized two isostructural organo-functionalized mixed-metal clusters

(2), which contain an unprecedented oxometallacyclic {M 2 V 2 Cl 2 N 4 O 8 }(M=Co, Zn) framework incorporating diethanolamine ligands via two distinct metal-ligand bonding modes, previously unobserved. The clusters contain terminal oxygen and chlorine atoms and have UV-vis absorption spectra consistent with multiple UV charge-transfer processes. In 1, there are four weakly coupled spin centers, where the isotropic exchange couplings are defined as J 1 , J 2 ', and J 2 ''. These couplings are J 1 = À5.4 cm À1 , J 2 ' = À12.6 cm À1 , and J 2 '' = À7.5 cm À1 . The ground state multiplicity of 2 is an open-shell singlet with an isotropic exchange coupling of À3.9 cm À1 . The vanadium centers are best described as a V IV centers, and the cobalt centers are high-spin Co II centers. Less orbital destabilization was observed due to weaker interaction of Cl À ligand on Co than what was observed for O 2À ligand on V centers. • 6H 2 O was synthesized according to a literature method. [30] Tetrabutylammonium trihydrogendecavanadate hexahydrate (0.095 g, 0.05 mmol), cobalt(II) chloride hexahydrate, CoCl 2 • 6H 2 O (Sigma-Aldrich, ACS grade; 0.126 g, 0.53 mmol) and diethanolamine, HN(CH 2 CH 2 OH) 2 (Fisher Chemical, Laboratory grade; 0.263 g, 2.49 mmol) were mixed in a 23 mL Teflon-lined Parr reaction vessel. The vessel was heated to 145 °C, held for 4 h at 145 °C, and then cooled to room temperature at the rate of ca. 5 °C h À1 . Three 3-mL portions of ethanol were added to the dark brown gelatinous product with gentle stirring. A solid product was obtained by filtration and dried under ambient conditions. The final product, in the form of purple crystals, was separated from a light brown co-product mechanically. The product yield was ca. 25.62, H 5.11, N 7.47; found: C 26.02, H 5.06, N 7.40. Characterization: Infra-red spectra were recorded on a Perkin Elmer Spectrum 100 FTIR spectrometer. An Ocean Optics HR4000 spectrometer controlled by Ocean Optics Spectra Suite Spectrometer Operating Software was used to obtain reflectance spectra of powdered samples of 1 and 2. In-house constructed dark chamber fitted with UV-Vis-NIR light source (Mikropack, DH-2000-BAL). Reflectance was measured against a Labsphere certified reflectance standard (USRS-99-010) and is reported as a pseudo-absorbance obtained by subtracting the reflectance (R) from unity. A Mettler Toledo TGA 2 SF/1100/735 thermogravimetric analyzer was used to obtain TGA profiles of 1 and 2 (N 2 , 100 mL min À1 ) at a heating rate of 5 °C min À1 in the 25-1000 °C temperature range. X-ray diffraction patterns were measured using a Cu Kα radiation source (λ = 1.54 Å) on a Bruker D2 Phaser spectrometer equipped with LynxEye linear position-sensitive detector. Data was collected in the 5-100°(2θ) range with a 0.03 2θ step and a 7.0 s/step dwell time. Single-crystalbased simulated X-ray powder diffraction patterns corresponded to the empirical X-ray powder diffraction patterns of each compound. Single crystal X-ray data were collected using synchrotron radiation (0.8854 Å; compound 2; 0.41328 Å) at Advanced Photon Source (APS), Argonne National Laboratory (ANL). Crystals suitable for X-ray diffraction were mounted on a glass fiber and transferred to a Bruker D8 platform goniometer equipped with a PILATUS3 X 2 M (Si 1 mm) detector. Using APEX III software package, data were integrated using SAINT V8.38A and scaled with a multi-scan absorption correction. [31] The crystal was kept at 295 K (compound 2; 100 K) during data collection. Using Olex2, the structure was solved with the XT structure solution program [32] using intrinsic phasing and refined with the XL refinement package [33] using Least Squares minimization. All non-hydrogen atoms were refined anisotropically. [34] Deposition Numbers 2081072 (for 1) and 2142846 (for 2) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.

The magnetic moment vs. temperature was measured by superconducting quantum interference device (SQUID) magnetometer MPMS XL at the Center for Nanoscale Materials (CNM) at Advanced Photon Source (APS), Argonne National Laboratory. Both zero field cooling and field cooling was performed at a temperature range of 2-300 K under helium and liquid nitrogen to achieve an ultracool system under reciprocating sample option (RSO) at magnetic field H = 1000 Oe. All moments were corrected for diamagnetic susceptibilities using Pascal's constants. [35] In addition, a term was used in fitting that allowed for temperature independent paramagnetism. The magnetic susceptibility was fit with the Heisenberg, Dirac, Van-Vleck (HDVV) spin Hamiltonian from 8 to 300 K, with the spin Hamiltonians shown in Equations (S1) and (S2) for 1 and 2 respectively. See the Supporting Information for further details.

The ORCA electronic structure suite, version 5, was used for all calculations. [36] The highly efficient resolution of identity (RI) and "chain or spheres" (COSX) approximations for the Coulomb and Exchange integrals respectively were employed (RIJCOSX [37] ). Grimme's D4 [38] dispersion correction was applied for all geometry optimizations. The DEF2-TZVP [39] basis set was employed for all geometry optimizations, and single point calculations used the relativistically recontracted DKH-DEF2-QZVPP basis set for all transition metals, and the relativistically recontracted DKH-DEF2-TZVP basis set for all other atoms. An automatically generated auxiliary basis set [40] was used for all single-point evaluations, which is the so-called "AutoAux" technique in ORCA parlance. Single point calculations also employed the scalar relativistic DKH2 [41] Hamiltonian, and picture change effects were included in the calculation of spin-orbit coupling. NBO calculations were performed with NBO 7.0. [42] All DFT calculations employed the PBE0 [43] hybrid functional. Broken-symmetry (BS) solutions were obtained using the "FlipSpin" technique as implemented in ORCA. See the Supporting Information for further details.