Heptacoordinated Molybdenum(VI) Complexes of Phenylenediamine Bis(phenolate): A Stable Molybdenum-Amidophenoxide Radical

The syntheses, crystallographic structures, magnetic properties and theoretical studies of two heptacoordinated molybdenum complexes with N,N’-bis(3,5-di-tert -butyl-2-hydroxyphenyl)- 1,2-phenylenediamine (H 4 N 2 O 2 ) are reported. A formally Mo(VI) complex 1 [Mo(N 2 O 2 )Cl 2 (dmf)] was synthesized by the reaction between [MoO 2 Cl 2 (dmf) 2 ] and H 4 N 2 O 2 , whereas the other Mo(VI) complex 2 [Mo(N 2 O 2 )(HN 2 O 2 )] was formed when [MoO 2 (acac) 2 ] was used as a molybdenum source. Both complexes represent a rare case of Mo(VI) ion without any multiply bonded terminal ligands. In addition, molecular structure, magnetic measurements, ESR spectroscopy and density functional theory calculations indicate that the complex 2 is the first stable Mo(VI)-amidophenoxide radical.


INTRODUCTION
Redox-active catechols and ortho-aminophenols are of great interest as non-innocent ligands for their ability to contribute to electronic properties typically associated with metal valence electrons. 1One example of such ligands is N,N'-bis(3,5-di-tert-butyl-2-hydroxyphenyl)-1,2phenylenediamine (H4N2O2) that can be viewed as a dimeric derivative of two bidentate orthoaminophenols. 2 Thus, it can act, once partially or fully deprotonated, as a multidentate ligand to form complexes with Cu and Zn 2 as well as with Ti and Zr. 3,4This potentially tetradentate ligand has rich electrochemical behavior and it can present five different oxidation states that are interrelated by one-electron transfer steps (see Chart 1).
Oxomolybdenum(VI) complexes of various multidentate nitrogen based ligands are known to behave as active catalysts in bioinspired oxotransfer reactions. 5,6,7In the current contribution, we wanted to combine the redox-active nature of the ligand H4N2O2 with the reactive MoO2 functionality to generate new potential oxotransfer catalysts.However, in our experiments the reaction of H4N2O2 with several MoO2 2+ sources did not yield the desired oxomolybdenum(VI) complex because the terminal oxo groups of the molybdenyl ion were removed upon ligand coordination.In the present paper, we report the syntheses, molecular structures and magnetic behavior of two new molybdenum complexes with the N2O2 ligand.The synthesized heptacoordinated complexes [Mo(N2O2)Cl2(dmf)] ( 1) and [Mo(N2O2)(HN2O2)] (2) are of interest due to their general coordination chemistry and the non-innocent behavior of the ligand.In support of the experimental work, we also carried out a comprehensive computational study to address the electronic nature of compounds 1 and 2. The stoichiometric reaction of [MoO2Cl2(dmf)2] with H4N2O2 in methanol or acetonitrile lead to a rapid formation of an intensely colored solution that afforded dark green shiny air-stable crystals in high yield (Scheme 1).The crystals are practically insoluble in common organic solvents or water, which prevents any NMR analyses.The infrared spectrum of the compound lacks the bands characteristic of a Mo=O function, which indicates the loss of molybdenyl oxo groups during complexation.Structural analysis by X-ray crystallography (see below) show that the solid state structure of compound 1 consists of separated neutral molecules of [Mo(N2O2)Cl2(dmf)].Thus, in the formation of complex 1, two metal-oxo bonds have been cleaved while two metal-chloride bonds have remained intact.This result was quite unexpected although similar reactivity with MoO2Cl2 derivatives has been observed earlier. 8,9In general, the cleavage of both metal-oxo bonds and the formation of a Mo(VI) compound without any multiply bonded terminal ligands is rare. 10As structurally comparable 2,2′-biphenyl-bridged bis(2-aminophenol) ligand 4,4′-di-tert-butyl-N,N′-bis(3,5-di-tert-butyl-2-hydroxyphenyl)-2,2′diaminobiphenyl (H4 t BuClip) is reported to react with MoO2(acac)2 to form [MoO2(H2 t BuClip)] where the diarylamines remain protonated and bind trans to the terminal oxo groups. 10In our studies, the elimination of both oxo moieties is probably due to the rigid geometry of the ligand system, which precludes the formation of the favorable cis-MoO2 structure.This and the relatively vague geometrical parameters of the product inspired us to study the bonding in detail (see below).

Synthesis of [Mo(N2O2)(HN2O2)] 2.
When [MoO2(acac)2] was reacted with H4N2O2 in methanol, a dark solution was formed upon which black shiny air-stable crystal of 2 deposited at ambient temperature.The compound is soluble in hydrocarbon solvents and ethers, but virtually insoluble in methanol.The reaction was repeated in different stoichiometries to have identical product in lower yields without any sign of a 1:1 complex.Similar crystals were obtained using the structurally analogous [MoO2(Heg)2] (Heg -= ethanediolate monoanion) as a starting material.The 1 H NMR spectrum of 2 does not offer any structural information as it shows only broad overlapping signals for the tert-butyl groups as well as for the hydrogen atoms in the aromatic rings.Similarly to 1, the infrared spectrum of 2 does not display any characteristic absorption for the Mo=O moiety.The structure of the compound was verified by X-ray crystallography (see below) to be a neutral molybdenum complex where two different ligands are coordinated to the metal.The protonation states of the ligands were verified by the observation of a peak in the electrospray mass spectrum at the mass expected for the empirical formula.Interestingly, both ESI(+) and ESI(-) mode gave similar peak patterns with the characteristic isotope distribution of the metal.It seems that the molecular cation is formed due to the removal of the odd electron, whereas the molecular anion is formed by the reduction of the metal or pairing of the odd electron.
Analytical samples of 2 were obtained from freshly prepared reaction mixtures as the material seems to metamorphose upon standing for a longer period of time.Although the physical appearance and unit cell parameters of the crystals remain unchanged over time, their diffraction intensities decrease significantly.This causes the refinement of the structure to fail due to the strong disorder of the ring atoms, which in turn suggests that the oxidation state of the ligand and/or metal can vary without any substantial changes in the overall molecular structure.
The cyclic voltammogram of 2 was measured in acetonitrile in the potential range from +2.0 to -1.7 V vs. Fc + /Fc.Three distinct one-electron oxidation waves (+0.41, +0.83 and +1.23 V) and three one-electron reduction waves (-0.01, -0.63 and -1.26 V) are seen within the solvent window (see Supporting Information).For comparison, the ligand-based redox potentials for [Zn(N2O2 ox )] are seen at +0.03 and +0.37 V for oxidation and at -0.64 and -1.29 V for reduction, respectively.Scheme 2. Formation of [Mo(N2O2)(HN2O2)] 2. The tert-butyl groups of the ligand N2O2 are omitted for clarity.

STRUCTURAL STUDIES
Crystals of 1 were obtained from the reaction mixture in acetonitrile.In the solid state structure (Figure 1), the molybdenum atom shares a plane with two oxygen atoms and two nitrogen atoms from the N2O2 ligand as well as with one oxygen donor from the coordinated dmf ligand.Two chlorides in axial positions complete the heptacoordinated environment around the metal center which is best described as a distorted pentagonal bipyramid. 11In principle, the protonation and oxidation level of the H4N2O2 ligand can be determined from high quality single crystal X-ray data as the C-C, C-N and C-O distances change systematically upon stepwise one-electron oxidation processes.1In 1, the C-C bond lengths within both the phenolic parts of the N2O2 ligand and the central ring fall in the range 1.38 -1.41 Å (Table 1), which does not allow unambiguous definition of the oxidation state of the ligand.Similarly, the C-N and C-O bond distances of 1.398 Å and 1.322 Å, respectively, indicate that the ligand oxidation state might be either -1 or -2 (Chart 1), while the observed metal to donor atom distances are characteristic for anionic phenoxide and amide ligands.Consequently, theoretical calculations at the DFT level were performed for a model system of 1 (see below).The single crystals of 2 were separated from the methanol solution of the ligand and [MoO2(acac)2] as described above.X-ray structure (Figure 2) showed that the asymmetric unit consists of two crystallographically independent molecules with comparable structural parameters (Table 1).In these molecules, both the oxo moieties and the acetylacetonato ligands have been replaced during complexation, the final product being a neutral heptacoordinated complex [Mo(N2O2)(HN2O2)] where the ligand displays two different coordination modes.One of the two ligands is fully deprotonated, whereas the other ligand has a dangling phenol part with an intact OH group.The formal oxidation state of the Mo center can again be estimated from the oxidation levels of these two different ligands.Similarly to complex 1, the ligand assembly is not unambiguous since the quality of the X-ray data does now allow an in-depth analysis of the geometrical parameters.Nevertheless, the tetradentate ligand seems to be analogous to a fully deprotonated [N2O2 red ] 4-(Chart 1) as in 1, whereas the tridentate ligand with a dangling phenol part can be best described as a triply deprotonated ligand [HN2O2 sq1• ] 2-(Chart 1).With these presumptions, the formal oxidation state of the metal center is Mo(VI).As the oxidation level assumed for the tridentate ligand involves one unpaired electron, the material should be paramagnetic.

MAGNETIC PROPERTIES
Compound 1 showed only a diamagnetic signal as expected due to the even number of electrons.Compound 2, on the other hand, gave a paramagnetic signal with susceptibility χmol = 4.2×10 -7 m 3 /mol at 5 K (χmol = M•χv/ρ, where ρ is the density, M is molar mass and χv is volume magnetic susceptibility).The temperature dependence of the susceptibility was also measured and the data fitted to the Curie-law of localized moments, χmol = C/(T-θp).Figure 3 shows the χmolT vs. T plot of 2 (bottom) as well as the plots of χmol vs. T and 1/χmol vs. T (top).As can be seen from the figure, the inverse of susceptibility does not obey the Curie law, which means that the magnetic spin is not localized.However, χmol is not temperature independent either as would be expected for Pauli paramagnetism induced by a completely non-localized electron.
Consequently, the most probable explanation of the data measured for compound 2 is a partially localized unpaired electron that is causing the magnetic properties. 12The calculated magnetic moment (µeff) at the high temperature range (T > 150 K) is 1.80 B.M. whereas it is 1.25 B.M. at the low temperature range (T < 80 K).As the 'spin only' value for an unpaired electron is 1.73 B.M., the magnetic moment at the high temperature range clearly indicates the presence of only one unpaired electron.

ESR SPECTROSCOPY
X-band electron spin resonance (ESR) spectra of the solid compounds 1 and 2 were measured at room temperature as well as at 4 K, whereas the solution spectrum of 2 was measured in CH2Cl2 at room temperature.As expected, compound 1 did not show an ESR signal, while a solid sample of 2 gave an axial spectrum at 4 K with g = 2.0157 and g|| = 2.0047 (see Supporting Information).Thus, the calculated g = 0.011 and <g> = 2.012.In CH2Cl2, 2 produced a nearly isotropic ESR signal (S=½) giso = 2.0087 with minor asymmetry, possibly due to an unresolved hyperfine interaction.These values clearly indicate that the unpaired electron is predominantly on the orbitals of the ligand, in a similar fashion as in Ni(II) and Ru(II) semiquinone complexes, 13 and not on the metal as <g> varies from 1.938 to 1.952 in several octahedral Mo(V) compounds. 14Against this scenario, 2 is best described as a molybdenum(VI) complex with a semiquinone-type radical ligand.

THEORETICAL STUDIES
Since the oxidation state of the ligands in 1 and 2 cannot be explicitly determined from the Xray diffraction data, DFT calculations were performed to shed light to the electronic structure and bonding of the ligands in these complexes.Generally in the non-innocent C6H4(NR)2-o and C6H4(O)(NR)-o species (R = H, alkyl, aryl), the discrimination between the completely reduced amido and phenoxido or oxidized imino and quinone formulation of the ligand is done by inspecting the trends in the experimentally determined C-C, C-N and C-O bond distances, which obviously requires very high quality X-ray diffraction data. 1,15,16In addition, the partially reduced radical o-benzosemiquinoneiminato form should be recognizable by comparison of these bond distances. 17As earlier studies have shown, 18 computational analyses provide in many cases crucial information when determining the electronic nature of the coordinating ligand and that of the entire complex, especially when the experimental structural data is inconclusive.
In order to determine the correct electronic states of the metals and ligands in both 1 and 2, the geometries of model complexes 1' and 2' (tert-butyl substituents replaced with methyl groups) were optimized using the PBE1PBE 19 functional and def2-TZVP 20 basis sets.The nature of the stationary points found was addressed by the subsequent calculation of two of the lowest eigenvalues of the Hessian matrix.Selected geometrical parameters of the optimized structures are given in Table 1.
For 1, there exist three plausible electronic states: a closed shell singlet 1'-S, a high-spin triplet 1'-T and a broken symmetry singlet diradical 1'-DR.Geometry optimizations were performed for all of these states and the results indicate that the triplet state is about 30 kJ/mol higher in energy than either of the two singlet states (in agreement with magnetometric measurements) and therefore will not be discussed further.From the two singlet states, the ground state is the diradical state 1'-DR which however is only 5 kJ mol -1 lower in energy than 1'-S.Considering such a small energy difference between these states, the diradical nature of complex 1 is relatively small, which is also evident from the calculated frontier molecular orbitals (MOs, see below) and from the spin distribution of the broken symmetry solution (Figure ).According to the Mulliken population analysis and the calculated spin density, 21 the formally unpaired electrons in 1'-DR are localized mostly at the nitrogen atoms of the ligand and at the molybdenum center.The diamagnetism in 1'-DR is then due to the relatively strong antiferromagnetic coupling between these electrons, which results in an S = 0 state.Table 1.Selected geometrical parameters of 1' and 2' as compared with the X-ray data.(For numbering scheme, see Figures 1 and 2).considering the observed disorder in the experimental X-ray data.For both 1'-S and 1'-DR, the largest deviation between the calculated and experimental bond lengths is about 0.04 Å and all important bond lengths in the complex (including those in o-phenylenediamine and oaminophenol rings) are reproduced with good precision (Rpar 22 values 0.0072 and 0.0045, respectively).The single biggest difference between the structures 1'-S and 1'-DR is in the Mo-N bond distances which are about 0.04 Å longer in 1'-DR and thereby in slightly better agreement with the experimental data.However, the differences in the optimized bond lengths of 1'-S and 1'-DR are in general too small in order for any definite conclusions to be drawn about the nature of the electronic ground state of 1.
The concept of metrical oxidation state (MOS) was introduced recently by Brown 23 and it has been used to quantify the formal oxidation state of non-innocent (oxidized) amidophenoxide or catecholate ligands coordinated to metal ions by examination of their geometrical parameters.
The MOS calculated for the experimental structure 1 is 1.55 (14).Corresponding values for the DFT optimized structures 1'-S and 1'-DR are 1.58 (17) and 1.47 (15), respectively.From these, the data for 1'-S is in slightly better agreement with the experimental value although the estimated standard deviations are very high, preventing definite conclusions being made.As explained by Brown, compounds with metal ions in high oxidation state and with two or fewer d electrons, such as molybdenum(VI) and vanadium(V) complexes, tend to have non-integer MOS values which originate from ligand to metal π-donation rather than from an antiferromagnetic coupling of electrons residing in separate orbitals. 23 To summarize, although the experimental structural data shows some discrepancies between the expected and observed bond lengths, the results from theoretical calculations indicate that the ground state of 1 is a singlet with a small diradical character, and which cannot therefore be fully described with a closed-shell configuration.The observed deviations in the geometrical parameters of the phenyl rings and Mo-N/O bonds can be attributed primarily to π-donation from the ligand to the high-valent molybdenum cation rather than actual electron transfer creating a lower oxidation state Mo(V) center.
In a similar fashion to 1, the geometrical parameters of the tridentate ligand in complex 2 suggest some variation in the oxidation state of the ligand which is also evident from the odd number of electrons in this complex.The plausible electronic states for 2 are a doublet (2'-D) and a quartet (2'-Q).Thus, DFT optimizations were performed to estimate the energy difference between the two states and the results show the quartet state to be over 80 kJ/mol higher in energy compared to the doublet.It is important to note that a broken symmetry doublet with two unpaired electrons at the ligand and one at the metal center (coupled antiferromagnetically as in 1'-DR) is also a plausible electronic configuration.However, we were not able to locate a minimum corresponding to such a state; the Mulliken populations in the HOMO (SOMO-1) orbital of 2'-D have also insignificant (only a few percent) contributions from the metal orbitals, thus excluding the possibility of a broken symmetry type state (see Supporting Information).
Hence, the ground state of complex 2 was inferred to be a pure doublet, which is fully supported by the data from magnetometric measurements (see above).
The theoretical model 2'-D reproduces the experimental geometrical features of 2 from good to excellent precision (Rpar value for bond distances from Table 1 is 0.0056, excluding the distance between the uncoordinated O52 and Mo1).The calculated data reproduce the key bond lengths around the metal center and show that the Mo-N(radical) bond length is significantly longer (over 0.1 Å) compared with the Mo-N(amido) bond.The structural parameters also show the overall shortening of bonds around the N43 center compared to the amido nitrogen as the latter bonds are about 0.02 to 0.07 Å longer than the former.In addition, small changes in the C-C and C-O bond distances of the neighboring phenyl groups of N43 indicate delocalization of the unpaired electron density over the whole aromatic system.The delocalization is also visible in the calculated spin distribution of 2' (Figure ) which, together with the Mulliken population analysis, shows the unpaired electron to be delocalized over the aromatic rings of the tridentate ligand with roughly one third of the total spin density attributable to the N43 atom.MOSs were calculated for the experimental structure of 2 (two separate molecules in the asymmetric unit) and for the theoretical model 2'-D.The MOS calculated for 2 from the average experimental bond distances is 1.50 (7), whereas the MOSs calculated separately for the individual molecules in the asymmetric unit are 1.42(10) and 1.58 (6).The corresponding MOS value for the DFT optimized structure is 1.30 (9) which differs slightly from the experimental data although the differences are again within 2.These results clearly point out the sensitivity of the MOS analysis to the X-ray data, which prevents an in-depth discussion of the electronic structure of 2. Similarly to 1, the calculated non-integer MOSs can be attributed to π-donation from ligand to the high-valent molybdenum cation (with the formal oxidation state Mo(VI)) rather than to any actual ligand to metal electron transfer.
Considered as a whole, the conducted experimental and theoretical investigations present an unambiguous picture of the electronic state and bonding in 2. To the best of our knowledge, this complex represents the first example of a stable high-valent molybdenum-amidophenoxide radical.[MoO2(Heg)2] were synthesized by literature procedures. 24,25,263b Other chemicals were used as purchased from commercial sources.The solvents used were of HPLC grade.All syntheses were done under ambient atmosphere.ESI-MS for 2 were measured in the positive and negative ion mode Bruker micrOTOF-Q spectrometer.The samples were injected as MeCN-water solutions.Cyclic voltammetry for 2 was recorded at ambient temperature using a platinum working electrode, a 1 mm diameter platinum counter electrode and a Ag/AgCl reference electrode.Samples were dissolved in MeCN containing 0.1 M of (Bu4N)ClO4 as the supporting electrolyte.The voltammograms were recorded at a scan rate of 100 mV s −1 while the potentials were measured in volts vs. the Fc + /Fc couple.

CONCLUSIONS
Solid-state ESR spectra were recorded at 4 K, whereas the solution ESR of 2 (in CH2Cl2) was recorded at ambient temperature.The magnetic properties were measured in a SQUID magnetometer with 70 mg and 48 mg samples of 1 and 2, respectively, sealed in plastic nonmagnetic straws.The temperature dependence from 5 to 300 K was measured in a 0.1 T magnetic field using 10 K steps and the field dependence was measured at 5 K between -2 T and 2 T using 50 mT steps.The susceptibility of sample 2 was determined at 5 K from a linear fit to the M(B) data.
Preparation of 1.To a solution of [MoO2Cl2(dmf)2] (173 mg, 0.50 mmol) in acetonitrile (5 ml) was added 260 mg (0.50 mmol) of the ligand precursor dissolved in 10 ml of the same solvent.X-ray Crystallographic Details.Crystals suitable for single-crystal X-ray measurements were obtained directly from the reaction mixtures.The crystallographic data for compounds 1 and 2 are summarized in Table 2 along with other experimental details.The data sets were collected at 223 K (1) or at 173 K (2) with an Enraf Nonius Kappa CCD area-detector diffractometer with the use of graphite monochromated Mo-Kα radiation (λ = 0.71073 Å).Data collection was performed by using φ and ω scans, and the data were processed by using DENZO-SMN v0.93.0 27 .SADABS 28 absorption correction was applied for complex 2. The structures were solved by direct methods using SHELXS-97 29 and full-matrix least-squares refinements on F 2 were performed using SHELXL-97 29 .All figures were drawn with Diamond 3. 30 For all compounds, the heavy atoms were refined anisotropically, whereas all hydrogen atoms were included at the calculated distances with fixed displacement parameters from their host atoms (1.2 or 1.5 times of the host atom).Computational Details.All calculations were performed using the Turbomole 6.3 program package. 31The geometries of the complexes were optimized using the PBE1PBE 19 density functional and Ahlrichs' def2-TZVP 20 basis sets.The nature of stationary points found was assessed by calculating the two lowest eigenvalues of the Hessian matrix.Mulliken population analyses were performed as implemented in the Turbomole 6.3 code.The program gOpenMol 32 was used for visualizations of spin density and molecular orbitals.

Figure 3 .
Figure 3. Plot of susceptibility χmol vs. T (top) and χmolT vs. T (bottom) of compound 2 in a 0.1 T

Figure 4 .
Figure 4. Calculated spin density distribution of complex 1'-DR.Red and blue denote excess

Considering complex 1 ,Figure 5 .
Figure 5. Frontier MOs of 1'-S and 1'-DR.The Mulliken populations of the shown orbitals are

Table 2 .
Summary of crystallographic data for 1 and 2 at 223 K and 173 K, respectively.