A 3D MOF based on Adamantoid Tetracopper(II) and Aminophosphine Oxide Cages: Structural Features and Magnetic and Catalytic Properties

This work describes an unexpected generation of a new 3D metal–organic framework (MOF), [Cu4(μ-Cl)6(μ4-O)Cu(OH)2(μ-PTA=O)4]n·2nCl-EtOH·2.5nH2O, from copper(II) chloride and 1,3,5-triaza-7-phosphaadamantane 7-oxide (PTA=O). The obtained product is composed of diamandoid tetracopper(II) [Cu4(μ-Cl)6(μ4-O)] cages and monocopper(II) [Cu(OH)2] units that are assembled, via the diamandoid μ-PTA=O linkers, into an intricate 3D net with an nbo topology. Magnetic susceptibility measurements on this MOF in the temperature range of 1.8–300 K reveal a ferromagnetic interaction (J = +20 cm–1) between the neighboring copper(II) ions. Single-point DFT calculations disclose a strong delocalization of the spin density over the tetranuclear unit. The magnitude of exchange coupling, predicted from the broken-symmetry DFT studies, is in good agreement with the experimental data. This copper(II) compound also acts as an active catalyst for the mild oxidation and carboxylation of alkanes. The present study provides a unique example of an MOF that is assembled from two different types of adamantoid Cu4 and PTA=O cages, thus contributing to widening a diversity of functional metal–organic frameworks.


■ INTRODUCTION
Over the last decades, the synthesis of metal−organic frameworks (MOFs) has seen a tremendous development with fascinating applications in catalysis, 1,2 magnetism, 3 biochemistry, 4 and materials science. 5 Among different transition-metal compounds, copper-based MOFs are particularly attractive, given the recognized significance of copper in molecular magnetism 6 and catalysis 7 and its presence in the active centers of different oxidation enzymes. 8 Hence, a good number of different bioinspired copper coordination compounds have been designed and applied in diverse catalytic transformations, 9−11 which also include the oxidative functionalization of alkanes (very abundant but inert hydrocarbons). 7,11 Cu-based MOFs with intriguing magnetic properties and related applications have also been reported. 6,12−14 To build new MOFs with unique structures and functional properties, the selection of appropriate organic linkers is important. Among a variety of organic linkers applied in MOF research, the cagelike aminophosphine 1,3,5-triaza-7-phosphaadamantane (PTA) and its P-oxide (1,3,5-triaza-7-phosphaadamantane 7-oxide, PTAO) are very interesting building blocks that feature a diamondoid geometry and several N,P-or N,O-sites for coordination. 15−17 Nevertheless, despite their considerable use in aqueous organometallic chemistry, PTA and PTAO are underexplored as building blocks for the design of MOFs. 15−17 This might be explained by a difficulty in realizing multiple N,P-or N,O-coordination modes of PTA or PTAO cages, respectively. 18−20 Therefore, the use of PTA O as a water-soluble and stable building block offers a prospective way toward the preparation of novel and structurally unique metal−organic architectures.
Following our research lines on the exploration of PTA and its derivatives in the design of new metal−organic architectures 18−22 and investigation of their functional properties, 23 4 ] n ·2nCl-EtOH·2.5nH 2 O (2) was initially obtained by a facile self-assembly reaction between copper (II) nitrate and PTAO in 2-chloroethanol/ethanol (v/v, 1/1) (route 1, Scheme 1). It was isolated as a red air-stable crystalline solid and then characterized by IR and EPR spectroscopy, and elemental, thermal, and X-ray diffraction analyses. In addition, in the course of the synthesis of 2, the formation of the ionic monocopper (II)

intermediate [H-PTAO] 2 [CuCl 3 (NO 3 )] (1) was observed ([H-PTAO] + is a protonated form of PTAO).
This hybrid inorganic−organic compound is considered as a precursor of 2, formed via a Cu-catalyzed dechlorination of 2chloroethanol (solvent component). 29−33 Certainly, the source of chloride ions in compound 2 is 2-chloroethanol. The use of this chlorinated solvent is essential for the synthesis of 1 and 2, as these products are not generated in similar reactions starting from CuCl 2 and PTAO in ethanol. Surprisingly, a similar reaction of CuCl 2 and PTAO in methanol instead of an ethanol/2-chloroethanol mixture leads to an amorphous, fine red powder of 2′ that has a different solvate system (route 2, Scheme 1). The synthetic protocols for both 2 and 2′ were optimized, maintaining the same molar copper/PTAO ratios. Unlike the synthesis of 2′ where methanol was used as a solvent, the preparation of 2 requires a mixture of ethanol and 2-chloroethanol.
The presence of different solvent molecules in the crystal lattice does not change the PXRD patterns of the 2 and 2′ samples obtained via routes 1 and 2. In addition, a diffractogram simulated for 2 from single-crystal X-ray data and after removal of solvent molecules shows a good match with the experimental PXRD patterns (Figure 1). For further studies, crystalline samples of 2 obtained by route 1 were used. Compound 1 was isolated as an orange crystalline solid and structurally characterized, revealing an unprecedented type of [CuCl 3 (NO 3 )] 2− anion, as confirmed by a search of the CSD (Cambridge Structural Database). 34 Interestingly, in both 1 and 2, the copper centers are five-coordinate and are simultaneously bound by three Cl − ligands. The presence of hydroxo and oxo ligands in 2 is associated with hydrolysis of the aqueous Cu(II) nitrate or chloride starting materials, 35 which might be accelerated by the presence of base (PTAO) and alcohol solvent medium (EtOH for 2; MeOH for 2′). . Powder X-ray patterns for 2 (route 1) and 2′ (route 2). From the top down: experimental patterns for 2′ (gray) and 2 (black) and calculated patterns for 2 (red) and 2 after removal of solvent molecules (orange). The high-angle reflections for 2 (black and red) are shown in detail.
To get further insight into the structure of MOF 2, we carried out its topological analysis by applying the concept of an underlying net. 36 After first round of simplification ( Figure  4a), all of the bridging ligands were reduced to centroids. Further simplification of this net was performed by treating the Cu 4 SBUs as 4-connected cluster nodes (Figure 4b). These nodes, along with the 4-connected Cu2 nodes and the 2connected μ-PTAO linkers, form a uninodal 4-connected framework with an nbo (NbO) topology and point symbol of (6 4 .8 2 ) (Figure 4b).  Magnetic Studies. The magnetic properties of 2 were investigated over the 1.8−300 K temperature range. Plots of magnetic susceptibility χ m and χ m T product vs T (χ m is the molar magnetic susceptibility for five Cu II ions) are given in Figure 5. The magnetic susceptibility of 2 increases with cooling, together with a simultaneous systematic increase in χ m T from 1.93 cm 3 K mol −1 (3.93 μ B ) at 300 K to 2.62 cm 3 K mol −1 (4.58 μ B ) at 1.8 K. Following prior data on Cu II tetramers with the general formula Cu 4 OX 6 L 4 (X = halogen, L = ligand), 37−45 the μ eff vs T pattern allowed the classification of such compounds into two groups: (i) those that exhibit increasing μ eff with decreasing temperature 41−44 and a maximum followed by a rapid decrease as the temperature is lowered further and (ii) those for which μ eff continuously decreases on lowering the temperature. The increase in χ m T (μ eff ) upon cooling indicates that the interactions between the copper (II) ions are ferromagnetic. The values of the Curie and Weiss constants determined from the relation χ m −1 = f(T) over the 1.8−300 K temperature range are equal to 1.92 cm 3 mol −1 K and 3.2 K, respectively. The positive value of the Weiss constant also confirms the occurrence of ferromagnetic interactions between the copper(II) centers in 2. To confirm the nature of the ground state of 2, we investigated the variation of the magnetization, M, with respect to the field at 2 K. The results are shown in Figure 5 (see inset), where the molar magnetization M is expressed in μ B . The compound does not reach the saturation in the applied field range, and the magnetization at 5 T is equal to 4.50 μ B . The magnetization data were compared with the sum of the Brillouin function of five isolated copper (II) ions. The experimental values for 2 are slightly greater than that expected for five independent S = 1/2 systems. These results also confirm the ferromagnetic coupling between the neighboring copper (II) ions.
The structure of 2 is composed of tetracopper (II) [Cu 4 (μ-Cl) 6 (μ 4 -O)] cages and monocopper (II) [Cu(OH) 2 ] units assembled into a 3D metal−organic framework. The Cu1··· Cu1 separations in the {Cu 4 OCl 6 } core are approximately 3.13 Å, and the closest Cu···Cu distance between the Cu 4 unit and Cu2 atom is 6.718 Å. A significantly weaker intermolecular interaction is expected in comparison to an intramolecular coupling within tetracopper(II) units. The magnetic susceptibility of 2 was fitted according to eq 1, as the sum of two independent contributions, namely one due to the tetracopper-(II) blocks (Cu1) 4 (χ m(Cu4 unit) ) and one due to the isolated copper(II) site Cu2 (χ m(Cu2 atom) ), in addition to a possible temperature-independent term (χ Nα ), with a typical value for the copper(II) ion of 60 × 10 −6 cm 3 mol −1 .
The magnetic susceptibility per tetracopper(II) unit, χ m(Cu4 unit) , derived from the van Vleck formula assuming an equal g value for the four copper(II) ions, 46,47 is given by eq 2 where J is the intracage exchange parameter. The other symbols have their usual meaning. The contribution χ m(Cu2 atom) is expected to follow a Curie−Weiss model for the S = 1/2 spins (eq 4).
When the possible presence of intermolecular exchange is taken into account, eq 1 should be modified by including a molecular field correction term (zJ'). 48 This yields the eq 5: A least-squares fitting of the experimental data leads to the following values for 2: J = +20.1(1) cm −1 , zJ′ = 0.04(1) cm −1 , and g = 2.20(2) (R = 8.31 × 10 −4 ). The criterion applied to determine the best fit was based on the minimization of the sum of squares of the deviation: R = ∑(χ exp T − χ calc T) 2 / ∑(χ exp T) 2 . The calculated curve for 2 (solid line in Figure 5) matches well the experimental magnetic data in the whole temperature range. The obtained value of J indicates a ferromagnetic coupling within the tetracopper(II) cluster in 2. Additionally, the value of zJ′ revealed a very weak magnetic interaction between tetracopper(II) and monocopper (II) units, as expected for quite large distances between copper (II) ions in the crystal lattice.  Previous studies of hydroxo-, alkoxo-and phenoxo-bridged copper (II) compounds 49 indicate that the nature and the strength of the overall coupling in such systems can be influenced by the Cu···Cu distances and Cu−O−Cu angles. Generally, the longer the Cu···Cu distance, the weaker the exchange interaction. When the Cu−O−Cu angle is <97.5°, ferromagnetic interactions can be expected, whereas for angles higher than 97.5°the interaction is mostly antiferromagnetic, with an increasing magnitude as the angle increases. 49 Therefore, the ferromagnetic coupling in 2 seems to be surprising at first glance because the Cu−O−Cu angle is higher than 109°and should cause an antiferromagnetic coupling. 49 The structure of 2 reveals that copper (II) ions are additionally linked by chloride bridges. However, no simple magnetostructural relationship was established relating the value of the magnetic exchange constant J to the Cu−Cl−Cu bonding angle or Cu−Cl or Cu···Cu distances in chloro-bridged copper (II) systems. 50 This may be due to the large variation in structural features observed, such as a number of distinct coordination geometries, which involve different orbitals in the exchange pathway. On the basis of the theory of superexchange 51 and on the behavior of coupling parameters with bridging angles for some μ-hydroxo-copper(II) compounds, 52 the smaller Cu−Cl−Cu angle and the shorter Cu−Cu distance may disclose a ferromagnetic interaction in 2. Although the Cu−Cl−Cu angles and Cu···Cu distances are the most crucial geometrical parameters, the coupling constant can also be modulated by terminal ligands. The presence of N-donor apical ligands may also favor ferromagnetic coupling. 41,53 However, slight differences in structural features and 3D structure in the case of 2 may also cause some distinction in the magnetic properties from related complexes involving the {Cu 4 OCl 6 } core. Moreover, the sign and magnitude of the J value obtained from the magnetic calculations for 2 match the theoretical results (DFT).
EPR Spectra. The EPR spectra of a solid sample of 2 recorded in the X-band at room temperature and 77 K are essentially similar and additionally confirm the properties detected by the direct magnetic measurements ( Figure S3). In comparison with the spectrum at 77 K, the signals at 293 K are much sharper and stronger. The observed EPR spectrum is the average pattern of four copper (II) centers in the ligand field of a trigonal-bipyramidal {CuCl 3 NO} symmetry and one in an octahedral{CuO 6 } geometry. 54−57 Moreover, the ΔM S = 2 transition was detected in a low-field part of the spectrum, which confirms the exchange interaction in the tetracopper (II) [Cu 4 (μ-Cl) 6 (μ 4 -O)] cage. The spectra of 2 also display a poorly resolved line at 3200 G that can be attributed to monocopper (II) [Cu(OH) 2 ] units (g = 2.08) ( Figure S3).
DFT Calculations for Magnetic Properties of 2. Broken-symmetry DFT calculations were performed to evaluate the exchange couplings in 2. The calculation methodology was first tested on a series of reported binuclear copper complexes with a {Cu(μ-Cl)(μ-O)Cu} motif. These calculations showed a sufficient level of matching between experimental and calculated J values (Table S1) for the cases where the dimeric molecules are sterically isolated and do not contain volatile ligands (e.g., CH 3 OH or CH 3 O − ). The difference between the experimental and calculated exchange couplings for the cases with μ-OCH 3 could be understood if one considers alteration of the core structure during the sample preparation (e.g., grinding). A similar observation was recently made for a tetranuclear nickel complex, [Ni 4 (μ 3 -OCH 3 ) 4 (Piv) 4 (CH 3 OH) 4 ] (HPiv = pivalic acid), for which the magnetic properties were modeled by DFT calculations by considering the structure alteration after elimination of coordinated methanol molecules. 58 The diamagnetic substitution method (DSM), where certain paramagnetic sites are replaced with diamagnetic sites (Cu II → Zn II in our case), was applied to evaluate the magnetic coupling between the certain Cu II pairs in 2. The DSM is known to correctly estimate the exchange couplings in polynuclear clusters, 59−61 including very large ones, such as the Fe 34 core. 62 The coordination geometry of the {Cu 4 OCl 6 } core in 2 discloses two slightly different Cu···Cu separations of 3.126 and 3.138 Å in a [4 + 2] configuration. Hence, we attempted to calculate two J constants belonging to these separations. The {Cu 4 OCl 6 (PTAO) 4 } fragment was used, where the copper atoms as well as the atoms of the first coordination sphere were modeled at the def2-TZVPP level and all other atoms at the def2-SVP level. The results of the DSM calculations are summarized in Table S2, where three reported {Cu 4 OCl 6 } complexes (CSD refcodes CUQFID, JIWKAB, and WEXYON) 63−66 were also used for comparative purposes. The sign and magnitude of predicted exchange couplings strongly depend on Cu−O(Cl)−Cu angles ( Figure  6), as was accounted for earlier. 63,67 The ferromagnetic coupling is favored by a smaller Cu−O−Cu angle as well as a shorter Cu···Cu separation, while a Cu−Cl−Cu angle appears to have an opposite influence on the J value ( Figure  6). Most of the reported structures with the {Cu 4 OCl 6 } unit reveal a symmetry lower than the ideal T d , where the tetranuclear core shows a slight distortion. To take this fact into account, we attempted to calculate six independent J CuCu  II 4 OCl 6 X 4 } fragment (see Table S2 for details) on selected structural parameters, where two Cu II atoms are replaced with Zn II atoms.  (Table S2). The interaction between the "isolated" copper site Cu2 and the closest copper site from the Cu 4 unit was predicted to be weakly antiferromagnetic (Table S2 and Figure 7). Although the closest Cu···Cu distance between the Cu 4 unit and Cu2 atom is 6.718 Å, it is known that the couplings at a comparable distance could be quite strong (up to 35 cm −1 ). 68 In the case of 2, the calculations evidence that the superexchange interaction mediated by the PTAO ligand is weak but is not zero ( Table  S2).
The predicted magnitudes of exchange couplings using the DSM model are considerably higher than the experimental values (Table S2). The explanation of this inconsistency lies in strong spin delocalization within the {Cu 4 OCl 6 } cluster, which disables the possibility of a correct determination of the separate J integrals by DSM. With two copper atoms replaced with zinc atoms, the main spin density lies on the remaining copper atoms (Table 1). The spin population on bridging Cl and O atoms is considerably smaller, as expected for a typical Cu−X−Cu superexchange. However, the single-point calculations for a quintet state of {Cu 4 OCl 6 } fragments in various complexes reveal the spin density on a central μ 4 -O atom to be comparable to that for copper atoms (Table 1 and Figure 7). The effect of pronounced spin delocalization in {Cu 4 OCl 6 X 4 } complexes 69 was experimentally confirmed recently by polarized neutron diffraction. 70 Thus, the mutual influence of the copper shoulders is too high and the DSM model is not applicable in this case.
The J integrals in a polynuclear complex could be estimated by knowing the energies of the high-spin (E HS ) and brokensymmetry (E BS ) states, using the approach neglecting the correction for spin projection, developed by Ruiz. 71 In this model, E BS − E HS = (2S 1 S 2 + S 2 )J 12 . 72 For Cu II ions with S 1 = S 2 = 1/2 this formula simplifies to E BS − E HS = J 12 . The accuracy of this approach has been confirmed by its successful use toward modeling of the exchange couplings in polynuclear complexes, including large-nuclearity cages. 73 As the calculation of the complete {Cu 4 OCl 6 (PTAO) 4 } fragment at the def2-TZVPP level is time-consuming, we investigated the possibility of using basis sets of reduced precision, as well as the use of truncated PTAO ligands. The respective results are summarized in Table S3. As can be seen, the nature of the substituent has a strong influence on the electron configuration of the whole molecule. The choice of a basis set was also crucial. We finally found that the {Cu 4 OCl 6 (NMe 3 ) 4 } fragment (where the coordinates of N and C atoms for NMe 3 groups are from the truncated PTA O ligands, with generated hydrogen atoms) and def2-TZVP basis set for all atoms give a Δ 12 energy gap similar to that for the {Cu 4 OCl 6 (PTAO) 4 }/def2-SVP/def2-TZVPP combina-   76 This compound does not exhibit spin delocalization, and its magnetic properties were precisely fitted with the support of a multifrequency, high-field EPR technique. By using the Ruiz approach, an overall behavior and exchange model symmetry can be predicted with sufficient precision (Table S4 and Figure 9). Next, we tested the same strategy on the fully asymmetric complex [Cu 4 OCl 6 (L 2 ) 4 ] (CSD refcode CUQFID) bearing a tetranuclear core resembling that in 2. 63 Although the predicted symmetry and distribution of the energy levels differ from those determined experimentally, the bulk magnetic behavior is correctly reproduced (Figure 9). For 2, all of the exchange interactions were predicted to be ferromagnetic, with the symmetry very close to T d assumed by the experimental model (Table S4). The difference in magnitudes between calculated and experimental couplings could be due to the simplified structural fragment used for the calculations, while the magnetic exchange it strongly dependent on the nature of the ligands in apical positions at the copper centers within a Cu 4 unit (Table S3). Also, the featureless magnetic curve of 2 ( Figure 5) prevents fitting with more than one exchange coupling constant due to the risk of overparameterization, The atom coordinates are obtained from an X-ray analysis: B3LYP/G functional, def2-TZVPP basis set for metal atoms and first coordination environment, def2-SVP set for all other atoms. b d(Cu1···Cu2) = 3.138 Å. c d(Cu1···Cu3) = 3.126 Å. d The numbers mean the copper atoms bridged by the respective chlorine atom. e The numbers indicate the copper atom coordinated by the respective nitrogen atom.   Table 4 for details. The magnetically isolated paramagnetic copper site Cu2 in 2 was not included in the simulation. All of the curves were adjusted by applying g = 2.0. The reported magnetic curve for FEVYAH shows the distinct decay of the magnetic moment at low temperature due to zero-field splitting, 76 not accounted for in the present study. neglecting in this way some possible slight differences between the individual coupling constants.
Catalytic Studies. Following our interest in the metalcatalyzed oxidative functionalization of alkanes under mild conditions, 77−79 we applied compound 2 as a catalyst for the oxidation and carboxylation of gaseous and liquid alkanes (i.e., propane and cyclohexane) to alcohols and ketones or carboxylic acids.
In contrast to numerous Cu-catalyzed systems for alkane oxidation, the reactions catalyzed by 2 do not require the presence of an acid promoter. 77−81 The oxidation of an inert light alkane such as propane in the presence of 2 leads to a mixture of isopropyl alcohol, acetone, n-propanol, and propanal with a total yield of 14% based on C 3 H 8 and a TON (turnover number; e.g. moles of product per mole of catalyst) value of up to 280 (Table 2). Isopropyl alcohol and acetone are the main products due to the more reactive nature of the secondary C atom of the propane molecule. The best TONs are observed when higher substrate pressures (loadings) are used. At a lower propane pressure, there is less substrate present in the system and thus the overall product yield is higher, although the TON values are lower ( Table 2). The propane oxidation is rather quick, resulting in the maximum product yields within a few hours. The product yields observed after the 24 h reaction time slightly decrease due to partial overoxidation. The effect of the acid promoter is not significant in the oxidation of propane, leading to comparable yields (Table 2, entries 4 and 5); only a minor improvement was observed in the presence of a TFA promoter. The fact that the catalyst is active without an acid promoter represents a feature of this system. The activity achieved herein is considerable when the high inertness of gaseous propane and mild reaction conditions applied are taken into account.
The oxidation of cyclohexane catalyzed by 2 results in ∼15% yields (based on C 6 H 12 ) of the products, cyclohexanol and cyclohexanone, already after 10 min of the reaction ( Figures  S10 and S11). The presence of 10 equiv (relative to the catalyst amount) of TFA (trifluoroacetic acid, known as an efficient promoter for Cu-containing catalytic systems) does not affect either the reaction rate or the total product yield ( Figure S10). Cyclohexane is transformed to a mixture of cyclohexanol (main product) and cyclohexanone with a total yield of 17%. The reaction rate is very high, resulting in a TOF (turnover frequency, i.e. moles of products per mole of catalyst per hour) value of ∼500 h −1 and a total yield of 14.4% after 7 min.
We also studied the effect of the catalyst amount on the maximum initial reaction rate (W 0 ) and the total product yield in the cyclohexane oxidation catalyzed by 2 ( Figure S11). When the catalyst amount is augmented from 1.25 to 10 μmol, the W 0 value increases with a nonlinear dependence. This indicates a reaction order above 1, suggesting that more than one Cu-containing catalytically active species participates in the rate-limiting reaction step. The same effect was also observed when multinuclear Fe-based catalytic systems were used for alkane oxidation. 82−84 When the catalyst amount in the reaction solution was lowered from 10 to 1.25 μmol, the maximum TON grew from 68 to 298.
Compound 2 was also screened as a catalyst for the mild single-pot carboxylation of cyclohexane or propane to form cyclohexanecarboxylic acid or a mixture of butyric acids, respectively ( Table 3). The reaction protocols involve the treatment of the alkane with CO (carbonyl source), H 2 O (hydroxyl source), and K 2 S 2 O 8 (oxidant and radical initiator) in an aqueous acetonitrile solution at 60°C. Thus, the carboxylation of C 6 H 12 yields 30% of C 6 H 11 COOH, along with minor amounts of cyclohexanone (1.2%) and cyclohexanol (0.3%) as a result of partial alkane oxidation. Propane is transformed to a mixture of isobutyric (main product) and nbutyric acids with a total yield of 28% (Table 3). In the present work, the activity of 2 in the carboxylation of hydrocarbons 78,85 is comparable to those of some other systems reported in our earlier studies (Table S5 in the Supporting Information).

■ CONCLUSIONS
In summary, we synthesized a unique 3D metal−organic framework which is driven by two different types of Cu II 4 and Cu II blocks and μ-PTAO linkers. The present compound extends a very limited family of MOFs assembled from cagelike aminophosphine oxide linkers, which are still poorly used for the design of metal−organic architectures. In addition, an unusual Cu-catalyzed dechlorination of 2-chloroethanol was observed upon treatment of copper (II)  Variable-temperature magnetic susceptibility and EPR studies indicate a ferromagnetic interaction between the neighboring copper (II) ions in the adamantoid Cu 4 unit. Moreover, compound 2 acts as an efficient catalyst for the oxidation and hydrocarboxylation of alkanes under mild conditions. The present work also provides a unique example of a functional MOF that is assembled from two different types of adamantoid Cu 4 and PTAO cages.

■ EXPERIMENTAL SECTION
Materials and Methods. All synthetic work was performed in air. Copper(II) nitrate trihydrate (POCh) and solvents (anhydrous ethanol, 2-chloroethanol; Sigma-Aldrich) were obtained from commercial sources. 1,3,5-Triaza-7-phosphaadamantane 7-oxide (PTAO) was synthesized in accord with literature methods. 86,87 The infrared spectra (4000−400 cm −1 ) were recorded on a Bruker Vertex 70 FT−IR instrument using KBr pellets. Elemental analyses and thermal analyses were performed using a CHNS Vario EL CUBE apparatus and a TG-DTA Setaram SETSYS 16/18 instrument (corundum Al 2 O 3 , 100 μL crucible, N 2 atmosphere, heating rate of 10°C/min), respectively. Powder X-ray diffraction (PXRD) analyses were performed on a Bruker D8 ADVANCE diffractometer using Cu Kα radiation (λ = 1.5418 Å) filtered with Ni. The diffractograms were recorded with a step size of 2θ = 0.016°over the range 2θ = 5−60°a nd ratio 0.5. The calculated pattern was obtained from the singlecrystal XRD data using the MERCURY CSD 3.9 package. In catalytic studies, gas chromatography (GC) analyses were carried out on an Agilent Technologies 7820A series gas chromatograph (detector, flame ionization; carrier gas, He; capillary column, BP20/SGE, 30 m   88 The structures were solved by direct methods with SHELXT-2014/5 and refined with full-matrix least-squares techniques on F 2 with SHELXL-2018/3. 89,90 The void and difference electron density maps were generated with Olex2-1.3.0. 91 The nitrate (N7, O3−O5) and chloride (Cl4) anions in 1 were modeled as being substitutionally disordered with site occupations of 0.559(11) and 0.441(11), respectively. The structure of 2 contains large voids of 1412 Å 3 occupied by solvent molecules ( Figure S1). The peaks observed in the difference Fourier map were modeled as chloroethanol and water molecules. The bond distances and angles in the chloroethanol model were restrained to ideal values. The H atoms of the water molecules were not localized. All other hydrogen atoms in 1 and 2 were placed at calculated positions and refined using the riding model with U iso = 1.2U eq .
Crystal data for 1: Crystallographic data for the structures reported in this paper have been deposited with the Cambridge Crystallographic Data Centre as CCDC-2064900 (1) and CCDC-2064901 (2).
Magnetic and Electron Paramagnetic Resonance (EPR) Studies. The magnetic properties were investigated over the temperature range of 1.8−300 K on a Quantum Design MPMS3 SQUID magnetometer. Solid-state EPR spectra of 2 were recorded on a Bruker ELEXSYS E500 CW-EPR spectrometer in X-band at 298 and 77 K. The magnetization of powdered sample 2 was measured over the 1.8−300 K temperature range using a Quantum Design SQUID-based MPMSXL-5-type magnetometer. The superconducting magnet was generally operated at a field strength ranging from 0 to 5 T. Sample measurements were made at a magnetic field of 0.5 T. The SQUID magnetometer was calibrated with a palladium rod sample. Corrections are based on subtracting the sample-holder signal, and the χ D contribution was estimated from Pascal's constants. 92 DFT Calculations. "Broken-symmetry" 72,93,94 calculations were carried out by using the B3LYP/G functional 94−98 with the TZVPP basis set for the copper atoms and coordination sphere and SVP for all other atoms (unless stated otherwise), using the ORCA 4.2.1 package 99 with integration grids Grid4. For some cases the chainof-spheres RIJCOSX approximation was applied, with the support of the auxiliary basis set def2/J. 100 The diamagnetic substitution method (DSM) 71,101 was used to "switch off" the certain paramagnetic Cu II centers, which were replaced with Zn II . The X-ray atom coordinates of 2 were used without geometry optimization, unless stated otherwise. In the respective cases, geometry optimization was performed using the BP86 functional with the TZVP basis set for the metal and SVP for all other atoms. The dummy H atoms (used for the generation of structure fragments) were generated by using the Avogadro 1.2.0 program. 102 The energies of the high-spin and broken-symmetry states were used to extract the J value in copper dimers, according to the formalism J AB = −2(E HS − E BS )/(S A + S B ) 2 . 103−106 The MAGPACK program was used to calculate the magnetic susceptibilities and energy states. 107 The isosurfaces of spin densities were drawn using the VESTA 3.5.2 program. 108 ■ ASSOCIATED CONTENT
Additional experimental, synthetic, and spectroscopic data, details of the crystal structure studies, additional figures and tables as described in the text (PDF)