Synthesis and Reactivity of Oxo-Peroxo-Vanadium(V) Bipyridine Compounds

Abstract

The vanadium(IV) compound [V^IVO(OH)(^tBu₂bpy)₂]BF₄ (V^IVO(OH)) (^tBu₂bpy = 4,4′-di-tert-butylbipyridine) is slowly oxidized by O₂ in ethereal solvents to give the oxo-peroxo compound [V^VO(O₂)(^tBu₂bpy)₂]BF₄ (V^VO(O₂)) in excellent yield. This and related compounds were fully characterized by NMR, IR, and optical spectroscopies, mass spectrometry, elemental analyses, and an X-ray crystal structure of the 4,4′-dimethylbipyridine analog, [V^VO(O₂)(Me₂bpy)₂]BF₄. Monitoring the reaction of V^IVO(OH) with O₂ in THF/acetonitrile mixtures by ¹H NMR and optical spectroscopies surprisingly shows that the initial product is the cis-dioxo compound [V^V(O)₂(^tBu₂bpy)₂]BF₄ (V^VO₂), which then converts to V^VO(O₂). Reaction of V^IVO(OH) with ¹⁸O₂ gives ca. 60% triply ¹⁸O labeled V^VO(O₂). The mechanism of formation of V^VO(O₂) is complex and may occur via initial reduction of O₂ at vanadium(IV) to give a superoxo-vanadium(V) intermediate, autoxidation of the THF solvent, or both. That V^VO₂ is generated first appears to be due to the ability of V^IVO(OH) to act as a hydrogen atom donor. For instance, V^IVO(OH) reacts with V^VO(O₂) give V^VO₂. V^VO(O₂) is also slowly reduced to V^IVO(OH) by the organic hydrogen atom donors hydroquinone and TEMPOH (2,2,6,6-tetramethylpiperidin-1-ol) as well as by triphenylphosphine. Notably, the peroxo complex V^VO(O₂) is much less reactive with these substrates than the analogous dioxo compound V^VO₂.

Introduction

Oxo-peroxo-vanadium(V) compounds are of interest because of their presence in the haloperoxidase enzymes found in fungi and marine algae,¹ and their involvement in stoichiometric and catalytic oxidations of organic substrates.^2,3 These complexes are typically prepared by reacting vanadium(IV) or vanadium(V) complexes with hydrogen peroxide or alkylperoxides.⁴ For catalytic applications, it is desirable to use O₂ as the oxidant, and there have been a few reports of the formation of oxo-peroxo-vanadium(V) products by reaction of O₂ with vanadium(IV) compounds.^5–8 These reactions are complicated by the mismatch in redox stoichiometry: oxidation of V^IV to V^V is a one-electron change while reduction of dioxygen to peroxide is a two-electron process.

A particularly detailed study of a reaction of a vanadium(IV) complex with O₂, by Kelm and Krüger, showed using ¹⁸O₂-labelling that both peroxo oxygens in the oxo-peroxo-vanadium(V)-(py₂(NMe)₂) product originated from dioxygen. A V^V-superoxo intermediate was also observed by EPR.⁶ It was proposed that this species abstracts a hydrogen atom from the THF solvent followed by deprotonation of the resulting V^V-hydroperoxo species to give the final V^V–peroxo product. Similar mechanisms have been proposed by Kosugi et al. for autoxidation of their tris(3,5-diisopropylpyrazolyl)borate vanadyl complex,⁵ and by Egdal et al. for their divanadium-phenoxyl-amine-pyridyl complexes.⁸ Kotchevar et al. have also proposed reduction of O₂ by a vanadium(IV) (bis-cyclopentadienyl)(acetylacetonate) complex to give a vanadium(V) superoxo intermediate, which initiates peroxidation of lipids in liposomal membranes.⁹ In contrast, Tajika et al. attribute formation of a V^VO(O₂) product with a tris(2-pyridylmethyl)amine (TPA) ligand simply to peroxide formed in the solvent.⁷ Autoxidation of solvent¹⁰ to form catalytically active peroxo-vanadium species has also been suggested in the catalytic epoxidation of olefins,¹¹ and alkane oxidation.¹²

Here we report the oxidations of vanadium(IV)-oxo-hydroxo-bipyridine compounds by O₂ in THF to give oxo-peroxo-vanadium(V) products of the formula [V^VO(O₂)(4,4'-R₂bpy)₂]BF₄ (R = ^tBu, Me, H). Bipyridine is a classic ligand for vanadium(IV) and vanadium(V) but the prior literature in this area has some inconsistencies, as described below. We came across these issues as part of our studies of hydrogen atom transfer reactions interconverting dioxo-vanadium(VI) and oxo-hydroxo-vanadium(IV) complexes.¹³ This report discusses the nature of the O₂ reactions and their mechanisms, and describes reactions of the peroxo complex [V^VO(O₂)(4,4'-^tBu₂bpy)₂]BF₄ with triphenylphosphine and hydrogen atom donors. This system is unusual because the vanadium(IV) compound itself can act as a hydrogen atom donor and because the peroxovanadium(V) complex is remarkably stable.

Results

A. Synthesis and characterization of [V^VO(O₂)(4,4′-^tBu₂bpy)₂]BF₄, [V^VO(O₂)]

In an attempt to synthesize the dioxo-vanadium(V) compound [V^V(O)₂(^tBu₂bpy)₂]BF₄ [V^VO₂],¹³ the known vanadium(IV) hydroxo complex [V^IVO(OH)(^tBu₂bpy)₂]BF₄ [V^IVO(OH)]¹⁴ was stirred in THF while exposed to air. This followed a very similar procedure described by Brand et al. for the aerobic oxidation of the related vanadyl-chloride complex with unsubstituted bipyridine ligands, [V^IVO(Cl)(bpy)₂]ClO₄.¹⁵ In this study we have mostly used the 4,4'-^tBu₂bpy ligand because it gives complexes with higher solubility in organic solvents, and because the vanadium(IV) derivative V^IVO(OH) has an easily observed, characteristic ¹H NMR spectrum with four separate ^tBu resonances.¹³ Over three days at ambient temperatures under air, a yellow suspension of the partially soluble V^IVO(OH) in THF transforms into a red suspension, and addition of pentane yields a pale red solid in 97% yield. As described below, NMR, UV-Vis, IR and ESI-MS methods indicate this product is not the anticipated dioxo-vanadium(V) complex, but is instead an oxo-peroxo-vanadium(V) species, [V^VO(O₂)(^tBu₂bpy)₂]BF₄ [V^VO(O₂)] (eq 1). The peroxo complex is formed in high yield using either one atmosphere of air or one atmosphere of dry O₂, but no reaction is observed under an atmosphere of N₂. The reaction proceeds well in dry THF distilled from Na/Ph₂CO, or in unpurified (wet) THF from a bottle on the benchtop, which contains 250 ppm of the stabilizer BHT (butylated hydroxytoluene, 2,6-^tBu₂-4-Me-C₆H₂OH). Et₂O or CH₃CN/THF mixtures can also be used as the solvent, but no reaction is observed when V^IVO(OH) is exposed to an atmosphere of O₂ in CD₃CN or CD₂Cl₂ for three weeks. Thus V^IVO(OH) has been observed to convert to V^IVO(O₂) only when both O₂ and an ethereal solvent are present (although perhaps materials with weak C–H bonds other than ethers could also be effective).

(1)

¹H NMR spectra of V^VO(O₂) show that the compound is diamagnetic. The presence of 4 ^tBu and 12 bpy peaks indicates a C₁ symmetric complex containing two inequivalent 4,4'-^tBu₂bpy ligands (Figure 1). Assignments of the ¹H NMR spectrum were enabled by a combination of ¹H 2D COSY and 2D NOESY NMR spectra. The UV-Vis spectrum of V^VO(O₂) in CH₃CN shows a peak at 475 nm with an ε of 370 M⁻¹ cm⁻¹ (Figure 2), typical for a peroxo-to-vanadium charge-transfer transition.¹⁶ More intense peaks are also observed below 350 nm, presumably due to bipyridine π → π^* absorptions since similar bands appear in spectra of V^IVO(OH).

Figure 1.

¹H NMR of V^VO(O₂) in CD₃CN. The inset shows the ^tBu region, and * indicates the residual CD₂HCN solvent peak.

Figure 2.

UV-Vis spectrum of V^VO(O₂) in acetonitrile.

ESI-MS analysis of an acetonitrile solution of VO(O₂) from O₂ shows a base peak at 635.5 m/z, the predicted mass for the [V^VO(O₂)(^tBu₂bpy)₂]⁺ cation (Figure 3a). Isolation of the 635.5 peak followed by collision induced fragmentation yields one intense peak at 367.1 Dalton, corresponding to the loss of one ^tBu₂bpy ligand to give a [VO(O₂)^tBu₂bpy]⁺ fragment. FAB/MS using a nitrobenzyl alcohol matrix shows analogous results. The peak with an m/z of 619.6 in Figure 3a corresponds to a V^VO₂ impurity in this sample.

Figure 3.

ESI-MS spectra of reaction mixtures of (a) V^IVO(OH) with ¹⁶O₂ in THF and (b) V^IVO(OH) with ¹⁸O₂ in THF.

A suspension of V^IVO(OH) in THF was stirred under ca. 1 atm of ¹⁸O₂ (98.4 % enriched), to probe how many O atoms from O₂ are incorporated in the product. ESI-MS spectra of the resulting ¹⁸O-labeled product have their most intense peak at m/z = 641.4. This is 6 Dalton larger than the most intense peak observed for V¹⁶O(¹⁶O₂), indicating incorporation of three ¹⁸O atoms. Isolation and fragmentation of the 641 m/z peak yields a peak at 372.8 m/z, consistent with [V¹⁸O(¹⁸O₂)^tBu₂bpy]⁺ and further supporting the presence of three ¹⁸O atoms. There is a less intense peak at 639.6 m/z, corresponding to doubly ¹⁸O labeled V^VO(O₂), but no peaks are observed at m/z = 635 or 637. Thus all the V^VO(O₂) generated is at least doubly ¹⁸O labeled, and ~60 % is triply-labeled based on the peak heights. Residual V^VO₂ and V^IVO(OH) present in the reaction mixture give rise to peaks at 619.7, 621.6 and 622.5 m/z, indicating the presence of un-labeled and singly- and doubly-¹⁸O labeled materials.

V^VO(O₂) was also characterized by IR spectroscopy. Oxo-peroxo-vanadium complexes are known to have strong, distinct V=O and O–O IR stretches,³ and assignments of these stretches for V^VO(O₂) were confirmed by ¹⁸O labeling studies. The isotopic shifts are most clear in spectra obtained from dichloromethane solutions, as shown in Figure 4 below. The ¹⁶O compound shows a ν(V=¹⁶O) stretch at 963 cm⁻¹ and a ν(¹⁶O–¹⁶O) stretch at 927 cm⁻¹, typical values for oxo-peroxo-vanadium compounds.^3,17 In the complex prepared from ¹⁸O₂, which is predominantly triply ¹⁸O enriched, new peaks for ν(V=¹⁸O) and ν(¹⁸O–¹⁸O) appear at 919 and 874 cm⁻¹, respectively. These isotopic shifts agree well with those predicted from a simple diatomic oscillator model, 917 and 874 cm⁻¹. The formation of triply-labeled V^VO(O₂) was unexpected, but exchange of ¹⁸O labels between peroxide, water, and terminal oxo groups has been previously reported for oxo-peroxo-vanadium compounds.^6,18

Figure 4.

IR spectra of solutions of V^VO(O₂) (solid green line) and V¹⁸O(¹⁸O₂) (dashed blue line)in dichloromethane.

The analogs of V^IVO(OH) with 4,4′-dimethyl-bipyridine and bipyridine ligands [V^IVO(OH)(Me₂bpy)₂]BF₄ and [V^IVO(OH)(bpy)₂]BF₄ show very similar chemistry.¹³ These complexes are insoluble in pure THF so their reactions with O₂ were performed in 1:1 CH₃CN/THF mixtures. Over 24 hours under air, these solutions changed from yellow to orange, and [V^VO(O₂)(Me₂bpy)₂]BF₄ and [V^VO(O₂)(bpy)₂]BF₄ were obtained in moderate yields after removal of solvent and washing with n-pentane. The NMR and UV-Vis spectra of the latter complex match those of the reported perchlorate salt, [V^VO(O₂)(bpy)₂]ClO₄, which was prepared following the more typical method of addition of H₂O₂ to an acidic mixture of V₂O₅ and ligand.¹⁹ For both [V^VO(O₂)(Me₂bpy)₂]BF₄ and [V^VO(O₂)(bpy)₂]BF₄, ¹H NMR spectra in CD₃CN or CD₂Cl₂ indicate diamagnetic, C₁-symmetric cations with two bipyridine ligands, similar to V^VO(O₂). ESI-MS and FAB-MS shows peaks consistent with [V^VO(O₂)(Me₂bpy)₂]⁺ and [V^VO(O₂)(bpy)₂]⁺, and the IR spectra of these compounds are similar to the spectrum of V^VO(O₂) as well. [V^VO(O₂)(Me₂bpy)₂]BF₄, for instance, shows a ν (V=O) stretch at 946 cm⁻¹ and a ν(O–O) stretch at 923 cm⁻¹.

X-ray quality crystals of [V^VO(O₂)(Me₂bpy)₂]BF₄ were obtained by slow diffusion of Et₂O into a concentrated CH₃CN solution of [V^VO(O₂)(Me₂bpy)₂]BF₄ in an NMR tube. The resulting structure suffers from whole molecule disorder, whereby a non-crystallographic 2-fold axis through the vanadium center rotates the oxo group into the peroxo group.^20a Because of this disorder, all non-hydrogen atoms, except for V1, were refined isotropically, but no restraints were used. The BF₄⁻ anion is also rotationally disordered and was treated using the SQUEEZE routine.^20b Half of the disordered cation is shown in Figure 5. Crystallographic information is given in Table 1 and selected bond lengths and angles for half of the disordered cation (which are within error of the values for the other half) are given in Table 2.

Figure 5.

ORTEP of the cation in [V^VO(O₂)(Me₂bpy)₂]BF₄, with hydrogen atoms omitted for clarity.

Table 1.

X-ray Crystallographic Data for [V^VO(O₂)(Me₂bpy)₂]BF₄.

Empirical Formula	C24H24BF4N4O3V
FW	554.22
crystal system	Monoclinic
Space group	P2(1)/c
a (Å)	14.190(5)
b (Å)	16.245(6)
c (Å)	10.876(4)
α (°)	90
β (°)	98.568(4)
γ (°)	90
Volume (Å3)	2479.1(16)
Z	4
density (g/cm3, calcd)	1.485
μ (mm−1)	0.465
λ (Å)	0.71073
Crystal size (mm)	0.15 × 0.12 × 0.10
Temperature (K)	130(2)
θ range (°)	2.51–25.57
Index ranges	−17 ≤ h ≤ 17
	−19 ≤ k ≤ 16
	−12 ≤ l ≤ 13
reflections collected	22202
Data / restraints / parameters	4548 / 0 / 261
Rint	0.0429
R1,wR2(I > 2σI)	0.0771, 0.2357
Goodness of fit	1.061

Table 2.

Selected bond lengths and angles for [V^VO(O₂)(Me₂bpy)₂]BF₄.

Bond Length (Å)		Angle (°)
V1-O1	1.869(7)	O3-V1-O1	104.4(3)
V1-O2	1.877(7)	O3-V1-O2	105.0(3)
V1-O3	1.556(7)	O1-V1-O2	44.0(3)
O1-O2	1.402(10)	N1-V1-N2	67.2(4)
V1-N1	2.198(12)	N3-V1-N4	73.6(4)
V1-N2	2.334(11)
V1-N3	2.263(9)
V1-N4	2.205(10)

The x-ray structure of [V^VO(O₂)(Me₂bpy)₂]BF₄ is consistent with all of the spectroscopic results discussed above and is similar to reported oxoperoxo-vanadium(V) structures.³ The O–O bond length of 1.402(10) Å is ~0.02–0.07 Å shorter than in most other oxoperoxo-vanadium(V) structures,³ however. This shorter than expected bond length may be due to the disorder between the oxo and peroxo moieties in the structure. Notably, the O–O bond length observed here is more reasonable than the previously reported, very short distances of 1.24 and 1.3 Å for [V^VO(O₂)(bpy)₂]ClO₄ and [V^VO(O₂)(phen)₂]ClO₄ (phen = 1,10 phenanthroline), respectively.¹⁹ The unusually short O–O distances in the previously reported structures are likely due to unrecognized disorder between the oxo and peroxo moieties, similar to what is observed here in [V^VO(O₂)(Me₂bpy)₂]BF₄. Additionally, the V=O bond length of 1.556(7) Å in [V^VO(O₂)(Me₂bpy)₂]BF₄ is ~ 0.04–0.06 Å shorter than is typical in oxoperoxo-vanadium(V) structures, which may also reflect the disorder. The strong trans effect of the oxo and peroxo groups causes a significant lengthening of the V-N bonds trans to these groups.

B. Mechanistic experiments

To probe the mechanism of formation of V^VO(O₂), the reaction of V^IVO(OH) with O₂ in CH₃CN/THF solution was monitored by optical spectroscopy (Figure 6a and 6c). The THF was distilled from Na/benzophenone ketyl, so that no organic peroxides or peroxide inhibitors were present initially. After an induction period of ca. 1 hour, the V^IVO(OH) (λ_max = 400 nm) began to bleach without concomitant formation of any V^VO(O₂) for the next five hours, based on the absence of its characteristic optical band at 475 nm. The bleaching indicates formation of V^VO₂, as shown by the ¹H NMR experiments below. Only after ≥ 80% of the V^IVO(OH) disappeared did V^VO(O₂) form in observable amounts, and this peroxo product continued to form over the next 25 hours to give the final 98 % yield. ¹H NMR spectra of the resulting solution showed only V^VO(O₂), and what appeared to be THF oxidation products. GC-MS of the organic products after reaction showed formation of the γ-butyrolactone and 2-hydroxy-THF, consistent with the peaks observed by ¹H NMR. A similar reaction, with V^IVO(OH) under an atmosphere of O₂ in a cuvette but wrapped in aluminum foil and stored in the dark, gave a quantitative yield of V^VO(O₂) by optical spectroscopy when checked after 2 days. The induction period observed in these reactions varied considerably, ranging from hours to several days. The reaction of a 1:1 mixture of V^IVO(OH) and V^VO₂ (each 3 mM) in CH₃CN/THF with O₂ was also monitored by optical spectroscopy. After an induction period of several days, the V^IVO(OH) bleached, indicating its conversion to V^VO₂. All the V^VO₂ present was then converted to V^VO(O₂) in just one day. The cis-dioxo compound V^VO₂ therefore appears to be a kinetically competent intermediate for this reaction.

Figure 6.

UV-Vis spectra of reaction of (a) V^IVO(OH) with O₂ in THF/CH₃CN, (b) V^VO₂ with O₂ in THF/CH₃CN, and (c) a comparison of absorbance at 475nm versus time.

The reaction of V^IVO(OH) with O₂ was also monitored by ¹H NMR in THF-d₈/CD₃CN. By NMR, the broad, paramagnetic ^tBu peaks of V^IVO(OH)¹³ disappear over two days, forming predominantly the known cis-dioxo vanadium(V) complex [V^V(O)₂(^tBu₂bpy)₂]BF₄,¹³ with small (ca. 1%) amounts of V^VO(O₂) (eq 2). V^VO₂ then slowly disappears, quantitatively forming V^VO(O₂). This is consistent with the optical experiments above: V^IVO(OH) first bleaches to form V^VO₂, which is d⁰ and has no characteristic optical absorbance, then V^VO₂ is converted to V^VO(O₂).

(2)

To test for possible binding of O₂ to the vanadium(IV) complex V^IVO(OH), optical spectra were taken of a CH₃CN/THF solution of V^IVO(OH) before and immediately after O₂ was added. There is no perceptible difference between these spectra, indicating no significant formation of a dioxygen adduct under these conditions.

The experiments above implicate V^VO₂ as an important intermediate in the reaction of V^IVO(OH) with O₂. Therefore, its reactivity with O₂ has also been examined. Remarkably, solutions of V^VO₂ in CH₃CN/THF react with air to quantitatively form the peroxo complex V^VO(O₂) (eq 3). This reaction is significantly slower than the one starting from V^IVO(OH), requiring a week to proceed to completion (Figures 6b and 6c). An ¹H NMR spectrum of an aliquot of this solution in CD₃CN after complete reaction shows only V^VO(O₂), THF, and THF oxidation products. V^VO₂ is more rapidly converted to V^VO(O₂) in the presence of ca. 5 equivalents of the alkylperoxide ^tBuOOH in CD₃CN, by NMR.

(3)

We have also found that isolated V^IVO(OH) quantitatively reacts with V^VO(O₂) in THF/MeCN to give V^VO₂ over several hours (eq 4). In this comproportionation reaction, the vanadium center of V^IVO(OH) is oxidized to vanadium(V) while the peroxo ligand of V^VO(O₂) is reduced to an oxo group. This reaction was monitored by UV-Vis spectroscopy and the final vanadium products were confirmed by ¹H NMR. Interestingly, only one equivalent of V^IVO(OH) is consumed in the reduction of one equivalent of V^VO(O₂). In the opposite direction, V^IVO(OH) in CD₃CN reacts with ~1 equivalent of 30% aqueous H₂O₂ over three hours at room temperature to give V^VO(O₂) in good yield (by ¹H NMR, eq 5).

(4)

(5)

C. Reactivity of the peroxo complex [V^VO(O₂)(^tBu₂bpy)₂]BF₄ [V^VO(O₂)]

Solid V^VO(O₂) is very stable, lasting years on the benchtop, and solutions in CD₃CN over many days show no decomposition. V^VO(O₂) reacts with Ph₃P (1.5 equivalents) over a week in dry CD₃CN in a J. Young NMR tube, with the orange solution turning yellow. By both ¹H and ³¹P NMR spectroscopies, the PPh₃ is converted to Ph₃PO, with only Ph₃PO visible in ³¹P NMR spectra after one week. In the ¹H NMR spectra, broad peaks corresponding to the ^tBu groups of V^IVO(OH) grow in as V^VO(O₂) is reduced. A small amount of free ligand (ca. 10%) is also generated during the reaction, indicating that some other minor vanadium products must be formed. With 4.4 equivalents of Ph₃P, the reaction reaches completion in approximately 3.5 days and ca. 1.7 equivalents of Ph₃P are oxidized to Ph₃PO.

No V^VO₂ is observed in the NMR spectra of the reaction of V^VO(O₂) with PPh₃. This is surprising as most d⁰ peroxo complexes are reduced by phosphines to give the corresponding d⁰ oxo compounds.²¹ In this case, however, separate control experiments show that V^VO₂ reacts with Ph₃P faster than the peroxo complex, with the reactions reaching completion within 4 hours. This reaction gives 0.5 equivalents of Ph₃PO, approximately 0.5 equivalents of V^IVO(OH), free ligand and other unknown products with broad peaks in the ^tBu region, by ¹H NMR. Peroxo-vanadium(V) complexes are also known to transfer an oxygen atom to olefins to give epoxides,² but no reaction was observed between V^VO(O₂) and trans-stilbene in acetonitrile over one week at 25 °C.

V^VO(O₂) also reacts with hydrogen atom donors. The presence of 6 equivalents of hydroquinone reduces V^VO(O₂) to V^IVO(OH) in 90% yield (some free ligand is generated) in dry CD₃CN over 5 days, as monitored by ¹H NMR. The reaction reaches completion overnight at 50 °C with 10 equivalents of hydroquinone. 1.5 equivalents of hydroquinone are consumed per vanadium, and an equal amount of benzoquinone grows in over the course of the reaction, as does H₂O. Equation 8 shows the balanced reaction, with 3 reducing equivalents of hydroquinone consumed to reduce the peroxo ligand by 2e⁻ and the vanadium center by 1e⁻.

(8)

V^VO(O₂) is also reduced by 2 equivalents of the hydroxylamine 2,2,6,6-tetramethyl-N-hydroxypiperidine (TEMPOH). Over 24 hours, all of the TEMPOH is consumed while approximately 60% of V^VO(O₂) is reduced to V^IVO(OH). Three equivalents of TEMPOH would be required to fully reduce V^VO(O₂) to V^IVO(OH), but substoichiometric amounts of TEMPOH were used to avoid the competitive reaction of TEMPOH with V^IVO(OH).¹³

Discussion

A. Synthesis of the peroxo complex V^VO(O₂)

The formation of a vanadium(V) peroxo complex from a vanadium(IV) species and O₂ is a surprising reaction because of the mismatch in redox stoichiometry. The vanadium is oxidized by one electron and the dioxygen is reduced by 2e⁻. It is even more odd that V^VO(O₂) is formed quantitatively from the vanadium(V) dioxo complex V^VO₂ and O₂ in CH₃CN/THF, without any redox change at the vanadium. In these reactions, the ether solvent provides some or all of the reducing equivalents needed to form the peroxo ligand, as indicated by the formation of γ-butyrolactone and 2-hydroxy-THF from THF. In the absence of a reducing solvent, in CH₂Cl₂ or MeCN, V^IVO(OH) is unreactive with O₂. The formation of oxidized THF products upon reaction of a vanadium(IV) complex with O₂ has previously been reported,⁶ as have stoichiometric and catalytic aerobic oxidations of THF mediated by a variety of transition metal and main group complexes.²²

We originally investigated the reaction of V^IVO(OH) with O₂ based on the report of Brand et al. that [V^IV(O)Cl(bpy)₂]Cl in THF is oxidized by O₂ to [V^V(O)₂(bpy)₂]Cl. Their product was characterized by elemental analysis, IR and optical spectroscopies, and an X-ray crystal structure.¹⁵ Brand’s report contrasts with our observations that the related hydroxy complexes [V^IV(O)OH(R₂bpy)₂]BF₄ react with O₂ to yield oxo-peroxo complexes, and form dioxo species only as transient intermediates. In retrospect, it appears that the product prepared by Brand et al. most likely contained a substantial amount of the peroxo complex. Most notably, the optical spectrum of Brand et al.’s “[V^V(O)₂(bpy)₂]Cl” is reported to have a peak at 475 nm with a molar absorptivity of 340 M⁻¹ cm⁻¹, a transition characteristic of oxoperoxovanadium(V) compounds.³ The [V^VO(O₂)(bpy)₂]BF₄ reported here has this band at λ = 475 nm, ε = 370 M⁻¹ cm⁻¹. In contrast, the authentic cis-dioxo compound, [V^V(O)₂(bpy)₂]BF₄, is yellow with no distinct peaks above 350 nm.¹³ Our assignments have been confirmed by ¹H NMR spectra, which Brand et al. did not report. They do give an elemental analysis of the chloride hydrate, “[V^V(O)₂(bpy)₂]Cl·H₂O,” but this has a calculated elemental composition that is essentially indistinguishable from the peroxo compound [V^VO(O₂)(bpy)₂]Cl. The dioxo cation [V^V(O)₂(bpy)₂]⁺ has now been reported four times, with Brand et al.’s report being the first, the next two being hydrothermal syntheses where only single crystals were characterized.²³ In light of these issues, we recommend synthesizing the simple [V^V(O)₂(R₂bpy)₂]⁺ salts directly from aqueous vanadate and the bipyridine ligand, as we recently reported.¹³

B. Comments on the mechanism of formation of the peroxo complex V^VO(O₂)

The mechanism of conversion of V^IVO(OH) to V^VO(O₂) is clearly complex. As mentioned above, two types of mechanisms have been discussed for the oxidation of vanadium(IV) compounds by O₂. Initial binding of O₂ to a vanadium(IV) species could form a vanadium(V)-superoxide complex. Such a V^V(O₂^•−) species has been prepared electrochemically and observed by ESR by Kelm and Krüger, although it was generated by oxidation of a V^V-peroxo complex rather than from V^IV + O².⁶ If the chemistry described here proceeds via initial binding of O₂ to V^IV, this must be an unfavorable equilibrium since no spectral change is observed when O₂ is added to solutions of V^IVO(OH) and no reaction is observed in the absence of an ether solvent. One could imagine a pathway in which an unobserved vanadium(V)-oxo-superoxo intermediate such as “[(^tBu₂bpy)₂V^VO(O₂^•−)]²⁺” abstracts a hydrogen atom from V^IVO(OH), yielding V^VO₂ and V^VO(O₂) and H⁺.⁵ The V^VO(O₂) would then be reduced by another equivalent of V^IVO(OH) to form more V^VO₂. This reduction of V^VO(O₂) is likely to occur via formation of vanadium-hydroperoxo intermediates, which may generate small amounts of reactive organic species via homolysis of a VO-OH bond²⁴ or oxidation of THF. However, such a pathway does not explain the induction period or lack of autoxidation of V^IVO(OH) in less oxidizable solvents.

It has also been suggested that organic peroxides play a key role in the aerobic formation of vanadium peroxides from vanadium(IV) starting materials.^7,11,12 The above observations and the presence of a variable induction period for the bleaching of V^IVO(OH) is consistent with such a pathway, where unobserved reactive intermediates require some time to reach sufficient concentrations. The reaction does not appear to be initiated by reactive species already present in the solvent, such as an organic hydroperoxide, as it proceeds similarly in ‘benchtop’ THF and THF freshly distilled from benzophenone ketyl. It is not evident what process generates the reactive intermediates, as V^IVO(OH) is quite stable in THF and THF/CH₃CN mixtures in the absence of O₂. The observation of oxidized THF products is consistent with both this mechanism and the one in the previous paragraph.

The initially observed product of air oxidation of V^IVO(OH) is V^VO₂ (eq 2, Figure 6a), but this is not mechanistically very informative. V^VO₂ could form directly, but it is also possible that the initial product is V^VO(O₂) which rapidly comproportionates with V^IVO(OH) to give V^VO₂ (eq 4). This is a unique feature of this system, probably related to the ability of V^IVO(OH) to act as a hydrogen atom donor. In previous work,¹³ V^IVO(OH) was shown to have a weak O-H bond, with a bond dissociation free energy (BDFE)²⁵ of only 70.6 kcal mol⁻¹. For instance, V^IVO(OH) reversibly transfers a hydrogen atom to the stable phenoxyl radical 2,6-di-tert-butyl-4-methoxy-phenyl (ArO^•) forming an equilibrium with V^VO₂ and the corresponding phenol. Thus the reduction of V^VO(O₂) by V^IVO(OH) is very similar to the reductions of V^VO(O₂) by the organic H-atom transfer agents TEMPOH and hydroquinone.

The conversion of V^VO₂ to V^VO(O₂) with O₂, with its unusual stoichiometry of 2e⁻ reduction of O₂ without change in the vanadium oxidation state (eq 3, Figure 5b), must also have a complex mechanism. Again the data are consistent with the reaction being mediated by small concentrations of reactive oxygen species. In particular, V^VO₂ is consumed faster when it is formed as an intermediate in reactions of V^IVO(OH), in which reactive species have presumably built up to significant concentrations. The observation that ^t BuOOH converts V^VO₂ to V^VO(O₂) implies that this transformation could be accomplished by other hydroperoxides, such as the autoxidation intermediate 2-hydroperoxy-THF.²⁶ Additionally, V^VO₂ may be capable of abstracting H^• from the THF solvent to start an autoxidation radical chain as V^VO₂ is known to stoichiometrically abstract H^• from xanthene and dihydroanthracene under anaerobic conditions. These species, however, have very weak C–H bonds (bond dissociation enthalpies ~ 80 kcal mol⁻¹), close to that of V^IVO(OH) and substantially weaker than the α-C-H bonds of THF (BDE 92 kcal mol⁻¹).²⁷ Consistent with these bond strengths, no reaction occurs between V^VO₂ and THF over days in the absence of O₂, as monitored by NMR or UV-Vis spectroscopies. Still, an undetectable amount of reaction could be sufficient to generate the small amount of reactive species needed to initiate the autoxidation reaction.

Ultimately, all of the processes described above could be occurring during the oxidation of V^IVO(OH) by O₂. In this light, it is remarkable that V^VO(O₂) is generated in essentially quantitative yield, which highlights the stability of this peroxovanadium complex.

C. Reactivity with Organic Substrates

Peroxo-vanadium compounds, like related d⁰ peroxo compounds of molybdenum, tungsten, and titanium, are commonly used as oxygen atom transfer reagents.^2,21,28 Still, reactions of phosphines with vanadium-oxo-peroxo complexes have only rarely been described. We have found two reports of V(O)(O₂)L_n species transferring one oxygen atom to Ph₃P to give Ph₃PO and a cis-dioxo-vanadium product.^5,29 In the vanadium-catalyzed oxidation of triarylphosphines by H₂O₂ in acidic MeCN/H₂O, however, the anionic diperoxo compound V^VO(O₂)₂⁻ is the reactive species while the monoperoxo and cis-dioxo species do not contribute to the observed reaction rate.³⁰

In the system studied here, V^VO(O₂) oxidizes Ph₃P to Ph₃PO over a few days at ambient temperatures, but the product is V^IVO(OH) not V^VO₂. This is because PPh₃ reacts more quickly with V^VO₂ than with the peroxo complex, as shown by independent experiments. Mostly likely, V^VO₂ is deoxygenated by Ph₃P to give a transient oxovanadium(III) species which comproportionates with a second V^VO₂ to give two molecules of vanadium(IV).³¹ This pathway, summarized by eqs 6 and 7, accounts for the stoichiometry of 1.5 equiv of Ph₃PO per V^VO(O₂) and requires 0.5 equivalent of adventitious water to form V^IVO(OH). Consistent with this suggested pathway, Zhang and Holm have reported that V^VOCl₄⁻ reacts with PhP₃ to give V^IIICl₄(CH₃CN)₂⁻ and Ph₃PO, and that V^III can accept an oxygen atom from iodosylbenzene or V^VO₂Cl₂.³² It is interesting and unusual that V^VO₂ is more reactive with Ph₃P than is V^VO(O₂), reaching completion in just a few hours instead of days.

(6)

(7)

Reynolds and Butler have reported that oxygen-atom transfer reactivity of V-, Mo- and W-peroxo compounds correlate with both the peroxo O–O stretching frequency and the λ_max of the peroxide-to-metal charge transfer absorption in electronic absorption spectra.²⁸ They note that metal monoperoxo compounds are much poorer oxygen-atom transfer reagents than metal diperoxo compounds and that V(V) peroxo complexes are generally less reactive than Mo(IV) or W(IV) peroxides. The observed O–O bond length of 1.402 Å for [V^VO(O₂)(Me₂bpy)₂]BF₄ is slightly shorter than is typical for peroxo-vanadium complexes.³ Additionally, both the ν(O–O) of 927 cm⁻¹ and the λ_max of 475 nm for V^VO(O₂), while typical for a vanadium(V) monoperoxo complex, are relatively high for metal-peroxo complexes. Taken together, these data suggest that the O–O bond in V^VO(O₂) is relatively strong and not very activated towards attack of substrates.

The reactions of V^VO(O₂) with the hydrogen atom donors hydroquinone and TEMPOH also occur much more slowly than analogous reactions of V^VO₂, with rates that differ by an order of magnitude or more at similar concentrations.¹³ We have noted previously that the dioxo compound is an unusually slow hydrogen atom acceptor,¹³ so it is interesting that the oxo-peroxo complex is even less reactive. Tajika et al. have reported similar results for the reduction of tris(3-pyridyl)methylamine oxo-peroxo-vanadium and dioxo-vanadium compounds with 2,6-di-tert-butylphenol and α-terpinene.³³ The first step of these reactions could be V^VO(O₂) abstracting a hydrogen atom from the substrate to give a hydroperoxo-vanadium(IV) intermediate, which is probably slow because it is significantly uphill (eq 8).³⁴ A similar step has been proposed in the intramolecular oxidation of isopropanol to acetone by peroxovanadium(V).³⁵

(8)

Conclusions

Oxo-peroxo-vanadium(V) complexes of the formula V^VO(O₂)(4,4'-R₂bpy)₂]BF₄ (R = ^tBu [V^VO(O₂)], Me, or H) have been prepared by the air oxidation of vanadium(IV)-oxo-hydroxo complexes [V^VO(OH)(4,4'-R₂bpy)₂]BF₄ (R = ^tBu [V^IVO(OH)], Me, or H) in THF. The unusual stoichiometry, with vanadium being oxidized by 1e⁻ but O₂ undergoing 2e⁻ reduction, is accounted for by the concomitant oxidation of the THF solvent. The autoxidation of V^IVO(OH) appears to involve an initial induction period, followed by close to quantitative formation of the cis-dioxo-vanadium(V) species ([V^VO₂(4,4'-^tBu₂bpy)₂]BF₄, V^VO₂) before the final V^VO(O₂) product forms. Remarkably, V^VO₂ also undergoes autoxidation in THF to the peroxo complex V^VO(O₂), without change in the vanadium oxidation state. V^VO(O₂) is reduced by the weak hydrogen atom donor V^IVO(OH) to give V^VO₂. V^VO(O₂) is also reduced by the organic hydrogen atom donors hydroquinone and TEMPOH, as well as the oxygen atom acceptor Ph₃P, all yielding V^IVO(OH). The peroxo species V^VO(O₂) is significantly less reactive with these organic reductants than the corresponding dioxo species V^VO₂, indicating that peroxo-vanadium(V) species are not always as reactive as often assumed.

Experimental Section

Reagent grade chemicals were obtained from Aldrich, Eastman Organic or Strem and used without further purification. Reagent grade solvents were purchased from Fisher Scientific or EMD chemicals. Anhydrous acetonitrile (<10 ppm H₂O) was purchased from Honeywell Burdick & Jackson, sparged with argon and plumbed from a steel keg directly into the glovebox. ¹⁸O₂ (98.4 % ¹⁸O enrichment) gas was obtained from Isotec laboratories. Deuterated solvents and H₂¹⁸O (95% ¹⁸O-enrichment) were obtained from Cambridge Isotope Laboratories. CD₂Cl₂ was dried by stirring over CaH₂ followed by vacuum transfer and stored in a dark bottle in a glovebox. CD₃CN was dried over CaH₂, vacuum transferred to P₂O₅ for 30 min, then CaH₂ again and transferred to a clean flask and stored in a glovebox. THF and THF-d₈ were stored over Na/ketyl and freshly vacuum transferred before use. TEMPOH³⁶ was prepared following literature procedures. Hydroquinone was recrystallized from acetonitrile.

The vanadium(IV) compounds [V^IVO(OH)(4,4′-R₂bpy)₂]BF₄ were prepared following the literature procedure.¹³ Compounds were stored in a N₂ glovebox and reactions were performed under N₂ using standard glovebox and Schlenk techniques. ¹H and ¹³C NMR spectra were recorded on Bruker Avance or DRX spectrometers (300 and 500 MHz) at ambient temperatures and are referenced to a solvent peak. Electrospray ionization mass spectra (ESI/MS) were obtained on a Bruker Esquire-LC ion trap mass spectrometer as solutions in acetonitrile, and an Agilent 5973 was used to acquire GC-MS data. IR spectra were recorded as KBr pellets or CH₂Cl₂ solutions in a cell with NaCl windows (as noted) using either a Perkin-Elmer 1720 or a Bruker Vector 33 FT-IR spectrometer. UV-vis spectra were recorded using a Hewlett-Packard 8453 spectrometer.

[V^VO(O₂)(4,4′-^tBu₂bpy)₂]BF₄ [V^VO(O₂)]

V^IVO(OH) (200 mg, 0.28 mmol) was mixed with THF (200 mL) in a round bottom flask that was sealed with a septum and vented to air with a needle. The stirred mixture changed from yellow to red in color over the course of 3 days. The solvent was then removed via rotovap, and the solids re-dissolved in CH₂Cl₂. Dropwise addition of n-pentane gave a red precipitate, which was isolated by vacuum filtration and dried in vacuo to give 0.331 g (98 % yield) of red-orange product. Anal. Calcd for [V^VO(O₂)]•H²O, C₃₆H₅₀BF₄N₄O₄V: C, 58.39; H, 6.81; N, 7.65. Found: C, 58.45; H, 6.57; N, 7.69. ¹H NMR (CD₃CN): one ^tBu₂bpy ligand; δ1.24, 1.59 (each s, 9H, ^tBu); 6.72, 9.64 (each d, 6 Hz, 1H, 6,6′-H); 7.10, 8.17 (each dd, 6, 2 Hz, 1H, 5,5′-H); 8.21, 8.59 (each d, 1 Hz, 1H, 3,3′-H); 2^{nd t}Bu₂bpy ligand: 1.34 , 1.60 (each s, 9H, ^tBu); 7.59, 10.36 (each d, 6 Hz, 1H, 6,6′-H); 8.12, 7.43 (each dd, 6, 2 Hz, 1H, 5,5′-H); 8.43, 8.59 (each d, 2 Hz, 1H, 3,3′-H). ¹³C{¹H} NMR (CD₃CN): δ30.11, 30.12, 30.35, 30.44 [C(CH₃)₃]; 36.24, 36.40, 36.96 (2C) [C(CH₃)₃]; 120.77, 122,04, 122.07, 122.85, 124.22, 125.71, 126.31, 126.62, 147.18, 148.76, 151.03, 151.22, 153.42, 153.79, 153.82, 155.33, 166.15, 167.08, 168.75 (bpy). IR(KBr pellet, in cm⁻¹): 965 s ν(V=O), 585cm⁻¹ m and 558 w ν(V–O), 927 s ν(O–O), 1616 vs ν(C=C, C=N), 2962 vs ν(CH₃), 3121 m ν(bpy C–H). UV-Vis [λ/nm, (ε/M⁻¹ cm⁻¹)]: 475(370), 311(24000), 253(91000). ESI/MS (CH₃CN): 635 [V^VO(O₂) - BF₄]⁺, 619 [V^VO(O₂) - O - BF₄]⁺.

[V^V(¹⁸O)(¹⁸O₂)(4,4′-^tBu₂bpy)₂]BF₄

Dry THF (10 mL, stored under N₂, over sodium benzophenone) was vacuum transferred to a 25 mL round-bottom flask containing V^IVO(OH) (15 mg, 0.021 mmol). This mixture was frozen with a liquid nitrogen/pentane bath (−131°) and the flask was exposed to ca. 1 atmosphere of ¹⁸O (1 L, 98.4 atom% ¹⁸O) and sealed. After stirring for 24 hours, the yellow mixture became orange. Pentane (10 mL) was vacuum transferred onto the THF mixture, and the solvents were poured off to give a red-orange solid which was dried under vacuum. IR (KBr pellet, cm⁻¹): 920 s ν(V=¹⁸O), 569 m and 521 m ν(V–¹⁸O), 876 s ν(O–O). (CH₂Cl₂ solution, cm⁻¹): 919 s ν(V=¹⁸O), 874 s ν(O–O). ESI/MS (CH₃CN): 641.3 [V¹⁸O₃(^tBu₂bpy)₂]⁺, 639.7 [V¹⁸O₂ ¹⁶O(^tBu₂bpy)₂]⁺, 623.3 [V¹⁸O₂(^tBu₂bpy)₂]⁺, 621.9 [V¹⁸O¹⁶O(^tBu₂bpy)₂]⁺.

[V^VO(O₂)(4,4′-Me₂bpy)₂]BF₄

A yellow solution of [V^IVO(OH)(4,4′-Me₂bpy)₂]BF₄ (100 mg, 0.185 mmol) in MeCN/THF (1:1 v/v, 200 mL) was stirred overnight while open to air, turning orange. The solvent was removed by rotovap to give a red orange solid which was washed with THF and dried in vacuo (0.060 g 60 % yield). Anal. Calcd for [V^VO(O₂)(4,4′-Me₂bpy)₂]BF₄, C₂₄H₂₄BF₄N₄O₃V: C, 52.01; H, 4.36; N, 10.11. Found: C, 51.86; H, 4.28; N, 9.91. ¹H NMR (CD₃CN): δ2.30, 2.44, 2.74, 2.75(each s, 3H, Me); 6.65, 7.51, 9.59, 10.30 (each d, 5.2–5.7 Hz, 1H, 6,6′-bpy H); 6.90, 7.23, 7.93, 7.98 (each d, 5.0–5.8 Hz, 1H, 5,5′-bpy H); 8.05, 8.30, 8.45, 8.47 (each s, 1H, 3,3′-bpy H). ¹³C{¹H} NMR (CD₃CN): δ 21.36, 21.92, 21.99 (2 C) [CH₃]; 124.14, 125.42 (2 C), 126.26, 128.05, 129.38, 129.94, 130.27, 146.94, 148.49, 150.85, 150.89, 153.03, 153.48, 153.57, 154.11, 155.00, 155.23, 156.87, 157.61 (bpy). IR(KBr pellet, cm⁻¹): 946 s ν(V=O), 923 s ν(O–O), 1619 vs ν(C=C, C=N). UV-Vis [λ/nm, (ε/M⁻¹ cm⁻¹)]: 472(364). ESI/MS (CH₃CN): 467 [VO₃(4,4′-Me₂bpy)₂]⁺, 451 [VO₂(4,4′-Me₂bpy)₂]⁺.

[V^VO(O₂)(bpy)₂]BF₄

A yellow solution of [V^IVO(OH)(bpy)₂]BF₄ (100 mg, 0.207 mmol) was dissolved in MeCN/THF (1:1 v/v, 200 mL) and stirred for 40 hours while open to air, turning grapefruit red in color. The solvent was removed by rotovap to give a red orange solid which was washed with THF and dried in vacuo (0.04 g, 40% yield). Anal. Calcd for [V^VO(O₂)(bpy)₂]BF₄•1/2H₂O, C₂₀H₁₇BF₄N₄O_3.5V: C, 47.37; H, 3.38; N, 11.05. Found: C, 47.66; H, 3.10; N, 11.02. ¹H NMR (CD₃CN): δ6.86, 7.70, 9.81, 10.53 (each d, 4.6–5.3 Hz, 1H, 2,2′-bpy H); 7.11, 7.43, 7.91 (each t, 1H, 3,3′-bpy H); 8.14 m, 8.21 m, 8.5 d, 8.64 m, 8.59 m (2H each, remaining bpy H). ¹³C{¹H} NMR (CD₃CN): δ123.69, 125.02, 125.10, 125.85, 127.65, 129.06, 129.55, 129.84, 141.59, 142.52, 143.92, 144.56, 147.66, 147.66, 149.28, 150.95, 151.71, 153.35, 153.73, 154.42, 155.38. IR(KBr pellet, cm⁻¹): 953 s ν(V=O), 923 s ν(O–O), 1607 vs ν(C=C, C=N). UV-Vis [λ/nm, (ε/M⁻¹ cm⁻¹)]: 474(370).