Non-Redox Assisted Oxygen–Oxygen Bond Homolysis in Titanocene Alkylperoxide Complexes: [Cp₂Ti^IV(η¹-OO^tBu)L]^+/0, L = Cl⁻, OTf⁻, Br⁻, OEt₂, Et₃P

Abstract

The titanium(IV) alkylperoxide complex Cp₂Ti(OO^tBu)Cl (1) is formed on treatment of Cp₂TiCl₂ with NaOO^tBu in THF at −20 °C. Treatment of 1 with AgOTf at −20 °C gives the triflate complex Cp₂Ti(OO^tBu)OTf (2), which is rapidly converted to the bromide Cp₂Ti(OO^tBu)Br (3) on addition of ⁿBu₄NBr. The X-ray crystal structures of 1 and 3 both show η¹-OO^tBu ligands. Complex 2 is stable only below −20 °C; ¹H, ¹³C, and ¹⁹F NMR spectra suggest that it also contains an η¹-OO^tBu ligand. Removal of the chloride from 1 with [Ag(Et₂O)₂]BAr'₄ (Ar' = 3,5-(CF₃)₂C₆H₃)) yields the etherate complex [Cp₂Ti(OO^tBu)(OEt₂)]BAr'₄ (4). Again, coordination of a fourth ligand to the Ti center indicates an η¹-OO^tBu ligand in 4. These peroxide complexes do not directly oxidize olefins or phosphines. For instance, the cationic etherate complex 4 reacts with excess Et₃P simply by displacement of the ether to form [Cp₂Ti(η¹-OO^tBu)(Et₃P)]BAr'₄ (5). Compounds 1–5 all decompose by O–O bond homolysis, based on trapping and computational studies. The lack of direct oxygen atom transfer reactivity is likely due to the η¹ coordination of the peroxide and the inability to adopt more reactive η² geometry. DFT calculations indicate that the steric bulk of the ^tBu group inhibits formation of the hypothetical [Cp₂Ti(η²-OO^tBu)]⁺ species.

Introduction

Metal-peroxide complexes have been proposed as reactive intermediates in a variety of oxidation reactions ranging from industrial to biochemical processes.^1,2 Coordination of the peroxide to the metal center activates the peroxide toward direct oxidation of substrates and/or toward O–O bond cleavage to give secondary oxidants. Transition metal alkylperoxide complexes, for instance, are suggested intermediates in catalytic epoxidation reactions from the industrial production of propylene oxide to the Sharpless titanium-tartrate chiral epoxidation.^{1, 3, 4} In biological systems, many metalloproteins are thought to utilize hydroperoxide intermediates, formed from H₂O₂ or from O₂ (+ 2e⁻ + H⁺).² Based on these systems, many biomimetic metal systems have been developed.^{1, 2, 5}

The oxidations of C–H bonds by metal/peroxide systems have traditionally been viewed as involving initial cleavage of the O–O bond. This can occur in either a homolytic or heterolytic manner (eqs 1,2; Fenton/Haber-Weiss-type mechanisms).^{1, 6} In both cases, the peroxide

$${\text{L}}_{\text{n}}{\text{M}}^{n+}\text{OOR}\to {\text{L}}_n{\text{M}}^{(n+1)+}=\text{O}+•\text{OR}$$ (1)

$${\text{L}}_{\text{n}}{\text{M}}^{n+}\text{OOR}\to {\text{L}}_n{\text{M}}^{(n+2)+}=\text{O}+{}^{-}\text{O}\text{R}$$ (2)

cleavage is typically viewed as requiring an increase in the oxidation state of the metal-center, to stabilize of the resulting metal-oxo species. Newcomb and Coon have recently proposed that an iron hydroperoxide species can oxidize C–H bonds by direct OH⁺ insertion without prior cleavage of the O–O bond (eq 3),⁷ which has sparked new debate into metalloenzyme oxidation

$${\text{L}}_{\text{n}}{\text{Fe}}^{\text{III}}\text{OOH}+\text{R}-\text{H}\to {\text{L}}_{\text{n}}{\text{Fe}}^{\text{III}}{\text{O}}^{-}+{\text{ROH}}_2^{+}$$ (3)

mechanisms.⁸ An interesting feature of eq 3 is that the oxidation state of the metal does not change. Therefore if such reactions are possible, they could occur even with non-oxidizable metal centers such as d⁰ ions. d⁰-Metal ions are widely used to activate peroxides but for oxygen atom transfer rather than C–H bond oxidation. The classic example is the selective epoxidation of terminal alkenes rather than oxidation of their weak allylic C–H bonds.^1,3 These reactions usually involve η²-peroxide or η²-alkylperoxide complexes, from which oxygen atom transfer directly gives an oxo or alkoxide complex (eq 4). η²-Peroxide complexes are more reactive than their η¹-isomers because they are more electronically activated and more sterically accessible.^1b–c

(4)

Reported in this article are the synthesis, characterization, and reactivity of new titanocene(IV) d⁰ η¹-alkylperoxide complexes Cp₂Ti(OO^tBu)L, where L is an anionic or neutral ligand. This class of compounds was studied in order to probe the reactivity of peroxide complexes containing a non-oxidizable metal center. Related hafnium complexes Cp*₂Hf(OO^tBu)R (Cp* = η⁵-C₅(CH₃)₅, R = alkyl) have been reported by Bercaw and coworkers⁹ and Cp₂Zr(OO^tBu)Cl has been very briefly mentioned by Schwartz and co-workers.^10,11 Titanium species have been widely used to catalyze peroxide reactions, and various intermediates have been suggested.¹² For example, the active oxidants “TiOOH” and even “Ti=O” have been proposed for reactions of titanium silicate molecular sieves using H₂O₂.¹³ Homogeneous titanium/^tBuOOH systems epoxidize alkenes and convert sulfides into sulfoxides, in some cases with high enantiomeric excess.^4,14 Closest to the chemistry described here, Cp₂TiCl₂ is reported to catalyze the ^tBuOOH oxidation of bis-homoallylic alcohols into tetrahydrofuranols and tetrahydropyranols.¹⁵ In few of these systems, however, has the reactive species been isolated and characterized in great detail.^4,14,16

The Cp₂Ti^IV(η¹-OO^tBu)L compounds described here are not direct oxygen atom transfer reagents towards phosphines and alkenes. Instead, they decompose at room temperature via homolysis of the peroxide bond even though formal oxidation of the metal center is not possible (eq 5). Homolytic O–O bond cleavage has previously been found in the decomposition of

$${\text{L}}_{\text{n}}{\text{M}}^{n+}\text{(OOR)}\to {\text{L}}_n{\text{M}}^{n+}(\text{O}•)+•\text{OR}$$ (5)

Cp*₂Hf(OO^tBu)R compounds.⁹ The lack of reactivity of the Cp₂Ti(η¹-OO^tBu)L complexes is attributed to their inability to form the more reactive and accessible η² conformer, even with very weakly binding ligands L. A preliminary account of part of this work has appeared.¹⁷

Results

I. Synthesis of Cp₂Ti^IV(OO^tBu)L complexes

Titanocene dichloride (Cp₂TiCl₂, Cp = η⁵-C₅H₅) reacts with 4 equiv of sodium tert-butylperoxide (NaOO^tBu) in THF at −20 °C to give the new titanocene tert-butylperoxide complex Cp₂Ti(OO^tBu)Cl (1) (eq 6). Low temperatures are required to prevent the

$${\text{Cp}}_2{\text{TiCl}}_2+{\text{NaOO}}^t\text{Bu}\underset{-20°\text{C}}{\overset{\text{THF}}{\to }}\underset{\mathbf{1}}{{\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu})\text{Cl}}+\text{NaCl}$$ (6)

decomposition of 1 as described below. Reactions using fewer than 4 equiv of NaOO^tBu result in a reduced yield and starting material still present in the reaction mixture. Complex 1 is isolated in 84% yield by removal of the THF solvent and extraction with hexanes to leave unreacted NaOO^tBu and NaCl behind. The yellow solid is >98% pure by ¹H NMR (using C₆Me₆ as an internal standard). This synthetic route follows the briefly mentioned Cp₂Zr(OO^tBu)Cl;^10,11 we have not been able to locate a procedure or characterization for this zirconium compound and our attempts to produce it have not been successful (as indicated by ¹H NMR). The related Cp*₂Hf(OO^tBu)R was prepared by proteolysis of Cp*₂Hf(H)R with anhydrous tert-butylhydroperoxide.⁹

Complex 1 has been characterized by NMR, IR, and mass spectroscopies, elemental analysis, and X-ray crystallography. ¹H and ¹³C{¹H} NMR spectra in THF-d₈ show the expected singlets for the Cp and ^tBu groups. IR spectra in CH₂Cl₂ solutions show a moderate intensity band at 819 cm⁻¹ not present in Cp₂TiCl₂ or NaOO^tBu which is tentatively assigned as the O–O stretch.

Crystals of 1 suitable for X-ray diffraction were grown from a saturated toluene solution at −5 ºC. Structure solution showed a typical bent-metallocene structure (Tables 1, 2; Figure 1). The tert-butylperoxo ligand is bound to the titanium center through only one oxygen, as indicated by the long Ti • • • O(2) distance of 2.952(2) Å and the open Ti–O(1)–O(2) angle of 121.5(1)°. Overall the structure is similar to that of the ethoxide analog, Cp₂Ti(OEt)Cl.¹⁸ The Ti–O(1) bond distance in 1 of 1.9090(14) Å is somewhat longer than the 1.855(2) Å Ti–OEt bond distance. The orientation of both the OO^tBu and OEt ligands has the α-substituent (O^tBu, Et) out of the equatorial plane, which allows π-donation from the oxygen p orbital to the empty 1a₁ orbital of the Cp₂Ti(Cl) metallocene fragment.¹⁹ π-Donation is supported by the Ti–O(1) bond distance in 1 being ~0.1 Å shorter than that predicted for a Ti–O σ bond on the basis of covalent radii (1.99–2.05 Ų⁰). This π-donation may also be the cause of the longer Ti–Cl distances in 1 and Cp₂Ti(OEt)Cl (2.396(1), 2.405(1) Å) versus that in Cp₂TiCl₂ (2.364(3) Å).²¹ The O(1)–O(2) distances of 1.467(2) Å in 1 and 1.489(12) Å in Cp*₂Hf(OO^tBu)Et⁹ are both within the range of typical peroxide bond distances (1.42 to 1.50 Å).²² The Cl-Ti-O(1)-O(2) dihedral angle in 1 of 79.71(11)° is larger than the C(1)-Hf-O(1)-O(2) dihedral angle of 70.9(7)° in the hafnium analog. The only other structurally characterized titanium-alkylperoxide complex is an [(η²-tert-butylperoxo)titanatrane]₂ dimer,¹⁶ which has Ti–O(1) and O–O bond distances [1.913(3) Å and 1.469(3) Å] that are nearly identical to those in 1. However, the Ti–O(2) bond distance of 2.269(2) Å and the Ti–O(1)–O(2) angle of 83.2(2)° for the η²-peroxide are much smaller – by nearly 0.7 Å and 40° – than in the η¹-bound titanocene complex 1.

Table 1.

Single Crystal X-ray Diffraction collection and refinement data for Cp₂Ti(OO^tBu)X.

	Cp2Ti(OOtBu)Cl (1)	Cp2Ti(OOtBu)Br (3)
empirical formula	C14H19Cl1O2Ti1	C14H19Br1O2Ti1
formula weight	302.64	347.10
temperature (K)	130(2)	130(2)
wavelength (Å)	0.71073	0.71073
crystal description	prism	prism
color	orange	orange
crystal system	monoclinic	monoclinic
space group	P 21/c	P 21/c
unit cell dimensions
a (Å)	6.4150(2)	6.6130(3)
b (Å)	12.0410(5)	11.8830(5)
c (Å)	19.0220(10)	19.0890(10)
α(°)	90	90
β (°)	108.7020(14)	109.0891(19)
γ (°)	90	90
volume (Å3)	1391.74(10)	1417.57(11)
Z	4	4
density (Mg/m3)	1.444	1.626
absorption coefficient (mm–1)	0.798	3.414
F(000)	632	704
crystal size (mm)	0.48 × 0.26 × 0.14	0.10 × 0.10 × 0.10
reflections for indexing	741	269
θ range for data collection (°)	3.57–28.28	3.26–28.30
index ranges	−6≤ h ≤ 6	−8≤ h ≤ 8
	−14≤ k ≤ 15	−13≤ k ≤ 14
	−25≤ l ≤ 25	−25≤ l ≤ 25
reflections collected	5103	5637
unique reflections	3017	3311
Rint	0.0447	0.0508
completeness to θ = 25.00°	96.7%	98.8%
absorption correction	semi-empirical	semi-empirical
max/min transmission	0.8965/0.7007	0.7265/0.7265
refinement method	full-matrix least-squares	full-matrix least-squares
data	3017	3311
Parameters (Restraints)	166 (0)	166 (0)
goodness-of-fit on F2 (S)	1.047	0.969
final R1 (I > 2σ)	0.0393	0.0391
wR2 (all data)	0.1083	0.0741
largest diff. peak and hole (e/Å3)	0.384/–0.427	0.528/–0.540

Table 2.

Selected bond distances (Å) and angles (°) for Cp₂Ti(OO^tBu)X (X = Cl, 1; Br, 3).^a

	1	3		1	3
Ti-Cl/Br	2.396(1)	2.568(1)	Cp(1)-Ti-Cl/Br	105.2(2)	105.1(2)
Ti-O(1)	1.909(2)	1.922(2)	Cp(2)-Ti-Cl/Br	106.2(3)	106.2(3)
O(1)-O(2)	1.467(2)	1.472(3)	Cp(1)-Ti-Cp(2)	132.7(3)	132.7(2)
O(2)-C(11)	1.458(2)	1.457(3)	Cp(1)-Ti-O(1)	110.4(2)	110.7(3)
C(11)-C(12)	1.522(3)	1.520(4)	Cp(2)-Ti-O(1)	99.7(2)	100.0(3)
C(11)-C(13)	1.520(3)	1.522(4)	Cl/Br-Ti-O(1)	97.31(4)	96.43(7)
C(11)-C(14)	1.525(3)	1.515(4)	Ti-O(1)-O(2)	121.45(10)	120.65(14)
Cp(1)-Ti-O(1)-O(2)	29.6(2)	29.3(2)	O(1)-O(2)-C(11)	107.65(13)	107.23(19)
Cp(2)-Ti-O(1)-O(2)	172.4(3)	172.7(2)	O(2)-C(11)-C(12)	110.38(16)	110.3(2)
Cl/Br-Ti-O(1)-O(2)	79.7(3)	79.5(3)	O(2)-C(11)-C(13)	101.53(16)	101.4(2)
Ti-O(1)-O(2)-C(11)	156.6(3)	155.9(2)	O(2)-C(11)-C(14)	110.34(16)	110.6(3)

^aCp(#) is the centroid of the Cp ring.

Figure 1.

ORTEP diagram of Cp₂Ti(OO^tBu)Cl (1), with thermal ellipsoids drawn at 30% probability.

Complex 1 reacts rapidly with one equiv of silver triflate (AgOTf) in THF-d₈ at −20 °C to quantitatively form the orange triflate complex Cp₂Ti(OO^tBu)OTf (2) by 1H NMR (eq 7).

$$\underset{\mathbf{1}}{{\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu})\text{Cl}}+\text{AgOTf}\underset{-20°\text{C}}{\overset{\text{THF}}{\to }}\underset{\mathbf{2}}{{\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu})\text{OTf}}+\text{AgCl}$$ (7)

Warming solution or solid samples of 2 above 0 °C results in decomposition to black material within minutes. This instability has prevented the isolation of pure 2. ¹H NMR spectra of 2 show singlets for the Cp and ^tBu groups downfield from those of 1 at the same temperature. The triflate ligand is observed as a quartet at δ120.4 ppm in the ¹³C{¹H} NMR (J_C–F = 318 Hz) and as a singlet at δ79.2 ppm by ¹⁹F NMR (THF-d₈, −20 °C), similar to values for other triflate complexes.^{23 19}F NMR spectra of solutions containing both 2 and excess AgOTf show two signals (δ79.2, δ78.6 ppm). Since AgOTf is completely dissociated in THF,²⁴ this indicates that the triflate is bound to the titanium center in 2 rather than free in solution. The bound OTf⁻ strongly suggests that the peroxide is still η¹-bound to Ti.

Treatment of in situ-generated 2 with one equiv of ⁿBu₄NBr at −20 °C forms the bromide analog of 1, Cp₂Ti(OO^tBu)Br (3), in 62 % yield based on 1 (eq 8). The highest yields of 3

$$\underset{\mathbf{2}}{{\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu})\text{OTf}}+{}^n\text{B}{\text{u}}_{\text{4}}\text{NBr}\underset{-20°\text{C}}{\overset{\text{THF}}{\to }}\underset{\mathbf{3}}{{\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu})\text{Br}}+{}^n\text{B}{\text{u}}_{\text{4}}\text{NOTf}$$ (8)

are obtained when pentane and Et₂O are used as solvents for preparative reactions (THF-d₈ is used for NMR reactions) and when the solutions are kept cold throughout. Complexes 1 and 3 are isomorphous (Table 2) and have very similar spectra. The formation of 3 from 1 via 2 confirms the characterization of 2 and shows that ligand metathesis occurs readily without disruption of the tert-butylperoxo ligand.

To avoid the anion coordination to titanium observed for 2, we have explored compounds with the B[3,5-(CF₃)₂C₆H₃)]₄⁻ anion (BAr'₄⁻). Reaction of 1 with one equiv of AgBAr'₄·xEt₂O²⁵ in CD₂Cl₂ at −20 °C quantitatively forms the ionic species [Cp₂Ti(OO^tBu)(OEt₂)]BAr'₄ by ¹H NMR (4; eq 9). The coordinated ether in 4 [(δ 3.61 (br q, 4H), 1.29 (t, 6H)] derives from the

$$\underset{\mathbf{1}}{{\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu})\text{Cl}}+{\text{AgBAr}}_{\text{4}}^{\prime }·x{\text{Et}}_2\text{O}\underset{-20°\text{C}}{\overset{{\text{CD}}_{\text{2}}{\text{Cl}}_2}{\to }}\underset{\mathbf{4}}{{[\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu}){\text{OEt}}_2{]\text{BAr}}_{\text{4}}^{\prime }}+\text{AgCl}$$ (9)

AgBAr'₄·xEt₂O; in our hands ether cannot be removed from this reagent without decomposition.^{26 1}H NMR spectra of reaction mixtures at −20 ºC show separate Et₂O resonances for 4 and for free ether in solution. Removal of the volatiles and addition of fresh CD₂Cl₂ results in loss of the free Et₂O peaks. Exchange of bound and free Et₂O is thus slow on the NMR timescale, which is surprising since ether is typically taken as a poor ligand.^1c The observation of a single, simple quartet for the methylene hydrogens of the bound ether indicates that there is rapid rotation about the Ti–OEt₂ bond. As found for the triflate complex 2, binding of a weak ligand is favored over an alternative structure with an η²-peroxide.

Generated in situ, 4 reacts with 1 equiv of Et₃P to give the phosphine complex [Cp₂Ti(OO_tBu)(Et₃P)]BAr'₄ (5, eq 10). Surprisingly, no oxidation of the phosphine is seen at −20

$$\underset{\mathbf{4}}{{[\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu}){\text{OEt}}_2{]\text{BAr}}_{\text{4}}^{\prime }}+{\text{PEt}}_{\text{3}}\underset{-20°\text{C}}{\overset{{\text{CD}}_{\text{2}}{\text{Cl}}_2}{\to }}\underset{\mathbf{5}}{{[\text{Cp}}_2\text{Ti}({\text{OO}}^t\text{Bu}){\text{PEt}}_3{]\text{BAr}}_{\text{4}}^{\prime }}+{\text{Et}}_{\text{2}}\text{O}$$ (10)

ºC by ¹H or ³¹P NMR, even in the presence of excess PEt₃. Instead, Et₃P simply displaces the bound Et₂O. The triflate analog of 5 is similarly formed on treatment of 2 with PEt₃. Like 2 and 4, thermal instability in solution and in the solid state has precluded isolation of pure 5, which has been characterized by NMR. The Cp resonance for 5 in the ¹H NMR is a doublet with ³J_H–P = 3 Hz, similar to other titanocene alkylphosphine complexes.²⁷ A gradient selected-HMBC 2D NMR experiment showed a correlation between the Cp ¹H NMR resonance at δ6.27 and the ³¹P NMR resonance at δ27.6, confirming this assignment.

II. Decomposition of Cp₂Ti^IV(OO^tBu)L complexes

All of the Cp₂Ti^IV(OO^tBu)L complexes decompose at ambient temperatures in solution. The decompositions and all the reactions described in this account were done under anaerobic and anhydrous conditions. Complex 1 decays in CD₂Cl₂ over a couple of hours to give t-butanol (96%) and a number of Cp-containing products in low yield, including Cp₂TiCl₂ (confirmed by spiking with authentic material and by mass spectrometry). Products were identified and quantified by ¹H NMR, integrating versus a C₆Me₆ internal standard. t-Butanol was further confirmed by vacuum transferring the volatiles from a reaction, acquiring a ¹H NMR spectrum, and spiking the sample with ^tBuOH. The hydroxyl resonance of the ^tBuOH was not observed in any reaction mixture in either ¹H or ²H NMR spectra, even after vacuum transfer, so it could not be determined whether the product is ^tBuOH or ^tBuOD. Other titanium products could not be identified. The t-butoxide complex Cp₂Ti(O^tBu)Cl is not among the products; it has been prepared independently from Cp₂TiCl₂ and NaO^tBu (see Supporting Information), Treatment of reaction mixtures with HCl(aq) or Me₃SiCl did not yield additional Cp₂TiCl₂. The decompositions of 2–5 in CD₂Cl₂ are quite similar to that of 1, except that a ~50% yield of Cp₂TiBr₂ is observed in the case of 3.

The decompositions of 1, 3, 4, and 5 in CD₂Cl₂ all follow first order kinetics, as monitored by ¹H NMR (Figure 2; 2 has only been generated in THF). The relative rate constants fall in the order: 4 > 3 ≅ 5 > 1 (Table 3). Overall, there is remarkably little variation among the four compounds, as the range in k_dec is less than an order of magnitude and the activation parameters are quite similar. Decomposition of 1 is slower in C₆D₁₂ (t_1/2 = 2.1 h) and C₆D₆ (1.0 h) than in CD₂Cl₂ (0.5 h, Figure 2).

Figure 2.

Plot of [1] versus time (using the average of both Cp and ^tBu peaks from ¹H NMR) at 303 K in C₆D₁₂ ( ), C₆D₆ ( ), and CD₂Cl₂ ( ) with exponential fits.

Table 3.

Kinetic data for the decompositions of Cp₂Ti^IV(OO^tBu)L in CD₂Cl₂.

Compound	kdec (s−1)	ΔH‡ a	ΔS‡ b	ΔG‡ (298 K)a	range in Tc
Cp2Ti(OOtBu)Cl (1)	4.1 ± 0.2 × 10−4	24 ± 2	5 ± 2	22 ± 2	283 – 313
Cp2Ti(OOtBu)Br (3)	3.2 ± 0.3 × 10−4	23 ± 2	1 ± 2	22 ± 2	283 – 318
Cp2Ti(OOtBu)(OEt2)+ (4)	7.1 ± 0.3 × 10−3	22 ± 2	3 ± 2	21 ± 2	266 – 306
Cp2Ti(OOtBu)(PEt3)+ (5)	3.2 ± 0.2 × 10−4	24 ± 2	4 ± 2	23 ± 2	286 – 326

^akcal mol⁻¹

^bcal K⁻¹ mol⁻¹

^cTemperature range (K) of rate constants used in Eyring analysis.

III. Reactivity of Cp₂Ti^IV(OO^tBu)L complexes

In the presence of one equiv of Ph₃P, the decay of 1 occurs at roughly the same rate as the decomposition of 1 (both in CD₂Cl₂; see the kinetic studies below). Ph₃PO is formed in 98% yield, by ¹H and ³¹P{¹H} NMR and by mass spectrometry. Cp₂TiCl₂ is produced in 50% yield, together with a number of other unidentified Cp containing products. The total Cp integral in the ¹H NMR (δ7.0–5.8) remains constant over the course of the reaction, indicating that little paramagnetic material is formed (<5%). The ^tBu group in 1 is converted to ^tBuOH (23%), isobutylene (CH₂=CMe₂, 59%), and ^tBuCl (12%). These volatile products were most easily identified by vacuum transferring the volatiles to a new NMR tube, obtaining a ¹H NMR spectrum, and then spiking with authentic materials (^tBuCl was also confirmed by EI-MS). The product yields do not change substantially when the amount of Ph₃P is increased from 1 to 20 equivalents. Complexes 2–5 react similarly with excess Ph₃P in CD₂Cl₂ (THF-d₈ for 2) to give high yields of Ph₃PO. Excess PPh₃ added to solutions of 5 does not displace the PEt₃ bound to the Ti center.

In contrast, the reaction of 1 with one equiv of Et₃P in CD₂Cl₂ does not form any triethylphosphine oxide by ³¹P{¹H} NMR. Instead, the major phosphorus-containing product is the phosphinite Et₂PO^tBu, identified by EI-MS and its characteristic ³¹P chemical shift of δ109.²⁸ Et₂PO^tBu was independently synthesized from Et₂PCl, ^tBuOH, and Et₃N, confirming the ³¹P NMR chemical shift. The origin of this unusual product is discussed below. Et₂PO^tBu is slowly consumed as the reaction proceeds, yielding other unidentified phosphorus products.²⁹ The maximum observed yield of Et₂PO_tBu is 70% at which point there is roughly a 25% yield of its apparent decay products, so 1 + PEt₃ appears to form the phosphinite quantitatively. No substantial difference is observed in the Cp region of reaction ¹H NMR spectra when compared to spectra for the decomposition of 1. Reactions of 3 and 5 with Et₃P also give high yields of Et₂PO^tBu and its subsequent decomposition products. The decomposition of the phosphine complex 5 in the absence of added Et₃P does not form either phosphinite or phosphine oxide. Apparently the Et₃P initially present in 5 remains bound to titanium throughout decomposition.

The reaction of 1 with ⁿBu₃P forms both ⁿBu₂PO^tBu (78%) and ⁿBu₃PO (19%, by ¹H NMR integration). These structural isomers are easily distinguished by ³¹P{¹H} NMR.³⁰ Reaction of 1 and (PhO)₃P gives the mixed phosphite (PhO)₂P(O^tBu) by ³¹P NMR and MS (50%);³¹ the phosphate (PhO)₃PO was not detected.

The kinetics of the phosphine reactions were monitored by ¹H NMR in CD₂Cl₂ at 303 K (Table 4). The decay of 1 in the presence of one equiv of Ph₃P (both 40 mM) follows first order kinetics over 6 half lives, with k = (5.9 ± 0.3) × 10⁻⁴ s⁻¹. The first order kinetics is surprising under these apparently second-order conditions. This k is only 44% faster than that found for decomposition of 1 at 303 K [(4.1 ± 0.2) × 10⁻⁴ s⁻¹]. With 5 and 20 equiv of Ph3P, k = (7.2 ± 0.6) × 10⁻⁴ s⁻¹ and (6.9 ± 0.5) × 10⁻⁴ s⁻¹ which are only ~ 20% faster than the k observed for one equivalent. Thus the reaction is not first order in Ph₃P. In the presence of one equiv of Et₃P or (PhO)₃P, the first-order rate constants for decay of 1 are slower or comparable to the decomposition of 1 without added reagents: (1.6 ± 0.7) × 10⁻⁴ s⁻¹ [PEt₃], (3.2 ± 0.7) × 10⁻⁴ s⁻¹ [(PhO)₃P] vs. k_dec(1) = (4.1 ± 0.2) × 10⁻⁴ s⁻¹. In sum, while the presence of phosphine or phosphite affects the rate of decay of 1, the dependence on [R₃P] is closer to zero-order than first-order. As discussed below, the data are consistent with rate limiting O–O bond homolysis that does not involve the substrate. To probe the involvement of a radical chain pathway, the decomposition of 1 was examined in the presence of ⁿBu₃SnH and CBr₄; neither had a substantial effect on the rate constant.

Table 4.

Products and rate constants for reactions of Cp₂Ti(OO^tBu)Cl (1) in CD₂Cl₂.^a

Reaction	k × 104 (s-1)	R3PO	R2POtBu	tBuOHb	Me2C=CH2	tBuCl	Cp2TiCl2
1	4.1 ± 0.2	-	-	96%	n/o	n/o	5%
1 + 1 Ph3P	5.9 ± 0.3	98%	n/o	24%	59%	12%	50%
1 + 5 Ph3P	7.2 ± 0.6	96%	n/o	29%	56%	11%	48%
1 + 20 Ph3P	6.9 ± 0.5	97%	n/o	26%	57%	11%	44%
1 + 20 nBu3SnH	2.8 ± 0.8	-	-	98%	n/o	n/d	4%
1 + 1 Ph3P + 20 nBu3SnH	5.2 ± 0.6	23%	n/o	72%	n/o	n/d	4%
1 + 1 Et3P	1.6 ± 0.7	n/o	~95%c	3%	tr	tr	27%
1 + 1 P(OPh)3	3.2 ± 0.7	n/o	50%	28%	8%	3%	28%
1 + CBr4	3.5 ± 0.5	-	-	95%	tr	tr	34%d

^an/d = not determined; n/o = not observed by ¹H or ³¹P{¹H} NMR; tr = trace amount (<1%).

^bHydroxyl resonance not observed in ¹H or ²H NMR.

^cEt₂PO^tBu grows in to a maximum of 70% yield but is concurrently consumed; the yield reported is for Et₂PO^tBu plus its decomposition products.²⁹

^dMay contain Cp₂TiClBr or Cp₂TiBr₂.

Added cyclohexene, norbornene, trans-stilbene, dimethyl sulfide, or allyl alcohol do not significantly affect the decomposition of 1. No products of oxidation of these substrates were observed by ¹H NMR or by GC-FID. There are only slight changes in the ratios of the products of decomposition of 1 and the rate constant for decay of 1 is essentially unchanged. Similarly, the decomposition products from the triflate complex 2 are unaffected by the presence of cyclohexene, norbornene, or trans-stilbene.

IV. Computational Studies

Cp₂Ti(OO^tBu)Cl (1), its homolysis to [Cp₂Ti(O · )Cl + ^tBuO · ], and several [Cp₂Ti(OOR)⁺] cations (R=H, Me, ^tBu) have been studied at the B3LYP level of density functional theory (DFT), with the 6–31G* basis set. Selected metrical data from computed gas phase structures are given in Table 5 below (see Supporting Information for complete details and for computations using the LANL2DZ basis set, which give similar results to those using the 6–31G* basis set). There is very good agreement between the calculated structure of 1 and that observed in the solid state. The largest discrepancy in a bond length is between the calculated Ti–O distance being 0.03 Å longer that that found in the X-ray structure. The calculated gas phase enthalpic barrier for O–O bond homolysis in 1 is 21.6 kcal mol⁻¹ as a lower limit. This is in good agreement with the experimental solution value H^‡ = 24 ± 2 kcal mol⁻¹. The calculations are discussed in more detail below.

Table 5.

Calculated gas-phase bond lengths (Å) and angles (°) at B3LYP/6–31G* optimized geometries.

	Cp2Ti(OOtBu)Cla	Cp2Ti(O · )Clb	Cp2Ti(OOH)+	Cp2Ti(OOMe)+	Cp2Ti(OOtBu)+
Ti–O(1)	1.880 (1.909)	1.699	1.936	1.928	1.917
Ti–O(2)	2.838 (2.952)	n/a	2.085	2.082	2.069
Ti–Cl	2.380 (2.396)	2.384	n/a	n/a	n/a
Ti-Cp(1)c	2.110 (2.052)	2.078	2.044	2.052	2.055
Ti-Cp(2)c	2.136 (2.072)	2.111	2.050	2.054	2.065
O(1)–O(2)	1.449 (1.467)	n/a	1.460	1.460	1.462
∠Ti–O–O	118.2 (121.5)	n/a	74.2	74.4	74.1
∠O–Ti–Cl	95.36 (97.31)	101.4	n/a	n/a	n/a
∠O–O–R	108.8 (107.7)	n/a	102.7	109.4	113.5
∠Ti-O-O-Rd	157.9 (156.6)	n/a	−113.6	−131.9	−148.2
twist anglee	n/a	n/a	3.4	7.2	11.7

^aSolid-state values from X-ray crystallography in parentheses.

^bUB3LYP.

^cDistance to Cp centroid.

^dTorsion angle.

^edeviation from orthogonality between the Cp^#-Ti-Cp^# plane (Cp^# = Cp centroid) and the O(1)-Ti-O(2) plane for η²-bound peroxides (see Figure 5).

Discussion

Given the broad interest in peroxide complexes, and titanium peroxide complexes specifically, it is surprising that few examples have been isolated and studied in detail.^{1, 2, 4, 13, 14} In this work, a series of titanocene t-butylperoxide complexes have been prepared starting from Cp₂TiCl₂ and NaOO^tBu. With well-characterized materials in hand, we have been able to probe the mechanisms of decomposition and various reactions.

I. Pathway for decomposition of peroxide complexes: O–O bond homolysis

The formation of the mixed phosphinites (Et₂PO^tBu from Et₃P, ⁿBu₂PO^tBu from ⁿBu₃P, and (PhO)₂PO^tBu from (PhO)₃P) are to our knowledge unprecedented reactions of metal alkylperoxo complexes. Typically, peroxo species rapidly oxidize phosphines and phosphites to phosphine oxides and phosphates, respectively.¹ The only reasonable way to form R₂PO^tBu compounds in these reactions is via phosphine trapping of the t-butoxyl radical. ^tBuO · adds to Et₃P and other phosphines at close to the diffusion limit to form metastable phosphoranyl radicals $[{\text{R}}_3\stackrel{•}{\text{P}}{\text{O}}^t\text{Bu}]$. The radical from Et₃P decomposes on the millisecond timescale by α-scission (α to the radical center), forming the phosphinite and Et · (eqs 11,12).³² α-Scission is the only observed pathway for $[{\text{Et}}_3\stackrel{•}{\text{P}}{\text{O}}^t\text{Bu}]$, even though the alternative β-scission to give Et₃PO and ^tBu · is much more thermodynamically favorable.

$${\text{PEt}}_3+{}^t\text{B}\text{uO}•\stackrel{k_{\mathit{trap}}=1.2\times {10}^9{\text{M}}^{-1}{\text{s}}^{-1}}{\to }[{\text{Et}}_3\stackrel{•}{\text{P}}{\text{O}}^t\text{Bu}]$$ (11)

$$[{\text{Et}}_3\stackrel{•}{\text{P}}{\text{O}}^t\text{Bu}]\stackrel{k_{\alpha -\text{scission}}\sim {10}^4{\text{s}}^{-1}}{\to }{\text{Et}}_2{\text{PO}}^t\text{Bu}+\text{Et}•$$ (12)

Addition of ^tBuO · to Ph₃P, however, is known to proceed by β-scission in $[{\text{Ph}}_3\stackrel{•}{\text{P}}{\text{O}}^t\text{Bu}]$ to give the phosphine oxide Ph₃PO. Presumably the greater strength of the Ph–P bond disfavors α-scission.^{32 n}Bu₃P reacts with ^tBuO · to give 80% ⁿBu₂PO^tBu and 20% ⁿBu₃PO,³³ and (PhO)₃P + ^tBuO · gives solely (PhO)₂PO^tBu.³⁴ In each case, these products from ^tBuO · quantitatively match the products observed on reaction of 1 with the respective R₃P. Unfortunately, no products resulting from the ethyl radical (eq 12), such as ethane or chloroethane, were detected in ¹H NMR spectra of reaction mixtures. Thus the reactions of 1 with R₃P involve ^tBuO · as the active oxidant; 1 does not oxidize the phosphines directly.²⁹ The formation of Et₂PO^tBu in reactions of 3 and 5 with Et₃P indicate a similar pathway in these cases. It should be noted that ^tBuOO · is not involved in the reactions of 1 since this peroxyl radical is known to quantitatively oxidize (PhO)₃P to the phosphate (PhO)₃PO,³⁵ which is not observed. ^tBuOO · in principle could be formed by Ti–O bond homolysis, as reported recently for Cp₂Ti(TEMPO)(Cl).³⁶

As a further test for the intermediacy of ^tBuO · , reactions of 1 with Ph₃P were run in the presence of tri-n-butyl tin hydride. ⁿBu₃SnH reduces ^tBuO · to ^tBuOH with a rate constant of 2.2 × 10⁸ M⁻¹s⁻¹,³⁷ almost an order of magnitude slower than the rate constant for Ph₃P + ^tBuO · (1.9 × 10⁹ M⁻¹ s⁻¹).³² Reaction of 1 + Ph₃P + 20 ⁿBu₃SnH yields 23% Ph₃PO, as compared to 98% in the absence of ⁿBu₃SnH, and the amount of ^tBuOH increases from 24% to 72% (Table 4). Consistent with the known rate constants, the ^tBuO · is trapped mostly, but not completely, by the 20-fold excess of ⁿBu₃SnH over the Ph₃P. The changes are not due to any direct reaction of 1 and ⁿBu₃SnH, as the decay of 1 is actually decelerated by the addition of the 20 equiv of ⁿBu₃SnH (Table 4).

t-Butoxyl radicals are reduced by Ph₃P to give ^tBu · .^{32a t}Bu · can form ^tBuCl by abstracting Cl · – from the CD₂Cl₂ solvent, 1, Cp₂TiCl₂ and/or other titanium products – or can form isobutylene by donating a H · to some mild oxidant in the solution (^tBu · has very weak C–H bonds). ^tBuCl and isobutylene are the dominant C₄ products observed from 1 + Ph₃P (Table 4). In the absence of phosphines, the ^tBuO · abstracts hydrogen from solvent or some other source to give high yields of ^tBuOH/^tBuOD. Apparently, this abstraction is faster than decomposition of ^tBuO · to acetone as none of this product is observed.³⁸ The 24% yield of ^tBuOH in the presence of Ph₃P likely results not from this H-atom abstraction pathway, but rather from oxidation of ^tBu · by 1 or some other species. In the presence of Et₃P, the ^tBu groups are retained in the Et₂PO^tBu product and very low yields of ^tBuOH, ^tBuCl, and CH₂=CMe₂ are found.

The data all indicate the presence of ^tBuO · as the primary oxidant, formed by homolytic cleavage of the O–O bond (Scheme 2). Unimolecular O–O bond homolysis is the rate limiting step, as the rate constants for decomposition (k_dec) with various added substrates over a range of concentrations vary by less than a factor of two (Table 4). The variation in k_dec is likely due to changes in the stoichiometry of decomposition. For instance, if the ^tBu · radical generated from 1 in the presence of Ph₃P itself consumes one equivalent of 1, then the observed rate constant for the decay of 1 would be twice the rate of homolysis. Rate limiting homolysis is further supported by the small positive ΔS^‡ values for the decompositions of 1, 3, 4, and 5 (Table 3). A radical chain mechanism is not likely because of the reproducible simple first-order kinetic behavior and because the decomposition of 1 is little affected by the oxidative and reductive radical traps CBr₄ and ⁿBu₃SnH (Table 4). The ease of homolysis in 1 is consistent with the tentatively assigned O–O stretching frequency, 819 cm⁻¹ since this is significantly lower than v(OO) in ^tBuOO^tBu (920 cm^{−1 39}). Structural parameters, however, do not show much difference in 1 vs. ^tBuOO^tBu, either in the O–O distance (1.467(2) vs. 1.478(3) Å) or the X–O–O–C torsion angle (156.6(1) vs. 164.1(7)°).⁴⁰ The DFT calculated gas phase enthalpic barrier to homolysis in 1 of 21.6 kcal mol⁻¹ is also consistent with the experimental value in solution of 24 ± 2 kcal mol⁻¹. This barrier to homolysis is 10 kcal mol⁻¹ lower than the 34 kcal mol⁻¹ determined for ^tBuOO^tBu and 16–19 kcal mol⁻¹ lower than organic hydroperoxides, which typically have a bond strength of 40–43 kcal mol^−1.1b

Scheme 2.

Mechanism for the decomposition of Cp₂Ti^IV(OO^tBu)L in the presence of Et₃P or Ph₃P.

O–O bond homolysis is a common mode of decomposition of metal alkylperoxide complexes, as in the Haber-Weiss mechanism (and the Fenton reaction for Fe²⁺ + H₂O₂).^1,6a Que and co-workers have generated Fe(TPA)(OO^tBu) at low temperature and determined that it undergoes homolysis to ^tBuO · and an iron(IV)-oxo intermediate (TPA = tris(2-pyridylmethyl)-amine).⁴¹ A series of cobalt(III) alkylperoxide complexes have been prepared by Mascharak et al. and shown to undergo homolysis upon mild heating.⁴² In contrast, the copper(II) tert-butylperoxide complex Cu(OO^tBu)(HB(3-^tBu-5-ⁱPrpz)₃) does not appear to undergo homolysis (which is calculated to have a high barrier) but rather acts as an electrophilic oxidant.⁴³ The only previous examples of d⁰ peroxide complexes undergoing homolysis are the hafnocene derivatives Cp*₂Hf(OO^tBu)Cl and Cp*₂Hf(OO^tBu)(C₆H₅) reported by Bercaw et al.⁹ In the presence of 9,10-dihydroanthracene, the latter compound converts to Cp*₂Hf(C₆H₅)OH and ^tBuOH.⁹ The barrier for homolysis in Cp*₂Hf(OO^tBu)Cl is ΔH^‡ = 22.6 kcal mol⁻¹, similar to what is seen for 1–5.

Homolyses of the O–O bonds in Cp₂Ti^IV(OO^tBu)L (1–5) and Cp*₂Hf^IV(OO^tBu)L are surprising because, unlike the cases above, the oxyl-metal product cannot be stabilized by oxidation of the metal center. Homolysis of 1 generates ^tBuO · and a titanyl species “Cp₂Ti(O · )Cl”, following eq 5 above. While the fate of the t-butoxyl radical is amply documented (see above), there is essentially no experimental information about the titanyl radical product Cp₂Ti(O · )Cl. The titanyl might have been expected to add to olefins, to transfer its oxygen atom to Et₃P to form Et₃PO, or to be trapped by ⁿBu₃SnH but these are not observed. This contrasts with the trapping of putative “Cp*₂Hf(O · )Cl” by 9,10-dihydroanthracene to give the hydroxide.⁹ It is possible that Cp₂Ti(O · )Cl is highly reactive and is rapidly converted to a myriad of products, making it difficult to trap with external reagents.

The nature of the implicated titanyl intermediate, Cp₂Ti(O · )Cl has been probed by DFT calculations using the UB3LYP functional (Table 5). The computed Ti–O distance of 1.699 Å is 0.18 Å shorter than the Ti–O bond in 1 and is a reasonable value for a ‘normal’ titanyl (L_nTi^IV=O) complex. Cp*₂Ti(=O)(4-phenylpyridine), for instance, has d(Ti=O) = 1.665(3) Å as measured by x-ray crystallography.⁴⁴ This suggests the presence of some Ti–O π bonding in Cp₂Ti(O · )Cl, although no specific π-bonding orbital is evident in the valence MOs. The Mullikan atomic spin populations (α spin - β spin; see Supporting Information) indicate that the oxyl oxygen carries the largest spin density (0.58), followed by the titanium (0.22) and four of the Cp carbon atoms (0.12 – 0.19). These four carbon atoms are close to the Ti–O π orbital perpendicular to the Ti-O-Cl plane and such an interaction may be responsible for their spin densities. The Ti–Cp interaction in Cp₂Ti(O · )Cl, appears to be perturbed relative to 1, based on the ~0.03 Å increase in the Ti–Cp(centroid) distances (although as above this is not evident from the valence MOs). In sum, the calculations suggest that the putative titanyl intermediate could be described as a mixture of resonance structures Cp₂Ti(O · )Cl and (Cp · )(Cp)Ti(=O)(Cl).

II. Attempted conversion to η²-alkylperoxide complexes; the limited reactivity of 1–5

Compounds 1–5 are remarkably inert toward oxygen atom transfer for a peroxide complex, particularly as a complex of a Lewis-acidic, early transition metal. We attribute this lack of reactivity to the η¹ coordination mode of the alkylperoxide ligand. These compounds all have a 16-electron count – considering the peroxide as a σ-only ligand – and could in principle convert to an η² structure without ligand loss. Instead, they approach an 18-e⁻ configuration through π-donation from the peroxide α-oxygen, as indicated by the short Ti–O distances and the orientation of the η¹-peroxide ligand. Removal of the chloride ligand in 1 does not form a cationic η²-complex – even the weak ligands triflate and Et₂O bind to the titanium centers in 2 and 4 in preference to an η²-complex.

η²-Complexes of ^tBuOO⁻ are thought to be common intermediates in reactions mediated by compounds of titanium and other early metals,^1–5 so there must be some feature(s) specific to the metallocene structure that disfavor η²-coordination. Electronically, η²-coordination prevents π-donation from the peroxide ligand because the frontier orbitals of the bent metallocene are localized in the "wedge" plane.¹⁹ Thus a species such as [Cp₂Ti(η²-OOR)⁺] could only achieve a 16-e⁻ configuration. Perhaps more important is the unfavorable steric interaction between the Cp ligands and the ^tBu substituent in [Cp₂Ti(η²-OO^tBu)⁺]. This steric clash is indicated by the DFT-calculated structures for [Cp₂Ti(η²-OOR)]⁺ with R = H, Me, and ^tBu. The optimized geometries (Figure 5, Table 5) show increasing distortions of the peroxide ligand with increasing steric bulk of the R group. The peroxide twists out of the equatorial plane of the metallocene: the deviations from orthogonality between the O-Ti-O and Cp^#-Ti-Cp^# planes (Cp^# = Cp centroid) are 3.4º, 7.2º, and 11.7º for R = H, Me, ^tBu, respectively. In addition, the R group is pushed away from the metal, with increasing Ti–O–O–R torsion angles of 114º, 132º, and 148º. Remarkably, the η²-form [Cp₂Ti(η²-OO^tBu)]⁺ is calculated to be 1 kcal mol⁻¹ higher in enthalpy than the unsaturated[Cp₂Ti(η¹-OO^tBu)]⁺ isomer. This is consistent with even very weak ligands giving [Cp₂Ti((η¹-OO^tBu)(L)]^+/0 rather than an (η² form. In essence, the bulk of the ^tBu group prevents the formation of the (η² isomer. In light of this, catalytic systems using titanocene dichloride and ^tBuOOH seem likely to involve cleavage of a Cp ring to allow for (η²-coordination and the resulting reactivity observed.^12,15

Figure 5.

Geometry optimizations for [Cp₂Ti(η²-OOR)⁺], R = H, Me, and ^tBu (B3LYP/6–31G*).

Conclusions

A series of titanocene complexes with (η¹-tert-butylperoxide ligands, Cp₂Ti^IV((η¹-OO^tBu)L have been prepared [L = Cl (1), OTf (2), Br (3), Et₂O (4), PEt₃ (5)]. Compounds 1–5 all decompose at ambient temperature and phosphine trapping studies indicate a mechanism of O–O bond homolysis for 1, 3, and 5. Homolysis generates ^tBuO · which reacts with Et₃P to give Et₂PO^tBu. Facile O–O homolysis is surprising since the titanium(IV) centers are d⁰ and cannot be oxidized; peroxide bond homolyses in transition metal complexes typically are facilitated by concomitant oxidation of the metal. Compounds 1–5 do not oxidize alkenes or phosphines, which was unexpected in light of the wide use of titanium compounds as catalysts for oxidations where peroxides are the terminal oxidants. Homolysis and the lack of direct oxygen atom reactivity in 1–5 are likely due to the inability of the complexes to form more reactive (η²-conformers, apparently for steric reasons.

Experimental

General Procedures

All manipulations were performed under an argon or nitrogen atmosphere, using standard high vacuum line or inert atmosphere glove box techniques unless otherwise noted. All glassware was flame-dried under vacuum immediately before use. Protio (Fisher Scientific) and deuterio solvents (Cambridge Isotope) were dried and degassed over Na/Ph₂CO (pentane, cyclohexane, hexanes, benzene, toluene, Et₂O, THF) or CaH₂ (CH₂Cl₂) and vacuum transferred immediately before use. Cp₂TiCl₂ (99+ %) and AgOTf (99 %) (both Strem) were used as received. AgBAr'₄·xEt₂O was synthesized following literature procedures.25 ⁿBu₄NBr (Aldrich) was ground finely in a mortar and pestle and dried under vacuum. Following a related procedure,⁴⁵ NaOO^tBu was synthesized by precipitation from equimolar amounts of HOO^tBu (~5.5 M in decane over 4Å molecular sieves, >97%) and NaO^tBu (>97%, both from Fluka) in THF, filtered, dried in vacuo and stored in a dessicator.

NMR spectra were obtained at 300 K (unless otherwise noted) on Bruker Avance DRX-499, AV-500 or DMX-750 spectrometers. NMR spectra were obtained using either J. Young-valved sealable or flame-sealed NMR tubes and are referenced to residual solvent peaks for ¹H and ¹³C, external 85% H₃PO₄ (in a sealed capillary) for ³¹P, or external CFCl₃ for ¹⁹F. Mass spectra were performed in EI⁺ ionization mode using a direct inlet probe (hot stage) on Kratos Profile HV-3 (low resolution) or JEOL HX-110 (high resolution) mass spectrometers. FT-IR spectra were obtained on a Bruker VECTOR 22/N-C spectrophotometer using a NaCl solution cell. Elemental analyses were performed by Atlantic Microlab (Norcross, GA).

CAUTION: All the metal-peroxide complexes described here should be prepared and handled in small quantities, stored in an inert atmosphere below 0 °C and manipulated with Teflon-coated spatulas. Complex 1 has been observed to undergo spontaneous exothermic decomposition in the solid state.

Cp₂Ti(OO^tBu)Cl (1)

In the air, Cp₂TiCl₂ (100 mg, 0.40 mmol) and NaOO^tBu (180 mg, 1.60 mmol) were ground together carefully using a mortar and pestle. The mixed solids were placed in a swivel-frit assembly and evacuated on a vacuum line. THF (50 mL) was vacuum transferred onto the solids at −78 °C, and the reaction mixture was stirred at −20 °C for 2 h. The THF was removed in vacuo at ambient temperature and hexanes (50 mL) vacuum-transferred in. The resulting solids were filtered and washed with excess hexanes until the filtrate was no longer yellow in color. Removal of the hexanes, addition of 10 mL pentane and filtration gave yellow 1 (102 mg, 84%). X-ray quality crystals were obtained by cooling a saturated, filtered toluene solution of 1 at −5 °C for 24 h. ¹H NMR (THF-d₈, −20 °C): 6.35 (s, 10H, C₅H₅), 1.09 (s, 9H, C(CH₃)₃). ¹³C{¹H} NMR: (THF-d₈, −20 °C), 117.3 (s, C₅H₅), 82.4 (s, OOCMe₃), 27.0 (s, OOC(CH₃)₃). IR (CH₂Cl₂, cm⁻¹): 3008 s (CH); 1451 s, 1445 s, 1206 s (^tBu); 1018 w (CO); 815 m (OO). MS: For ¹²C₁₄¹H₁₉³⁵Cl¹⁶O₂⁴⁸Ti (M⁺): calcd, 302.05530; found, 302.05544. Anal. Calcd for C₁₄H₁₉ClO₂Ti: C, 55.62%; H, 6.34%; Fnd: C, 55.71%; H, 6.33%.

Cp₂Ti(OO^tBu)OTf (2)

THF (25 mL) was vacuum transferred onto 1 (100 mg, 0.33 mmol) and AgOTf (85 mg, 0.33 mmol) at −78 °C and the reaction mixture was stirred at −20 °C for 2 h. The THF was removed in vacuo at 0 °C and Et₂O (50 mL) was added. The suspension was stirred at −20 °C for 0.5 h, filtered at −78 °C to remove the AgCl and any residual AgOTf, and immediately the Et₂O was removed in vacuo at −20 °C, yielding 2 as an orange solid. Complex 2 is unstable at temperatures over 0 °C, decomposing quickly both in solution and as a solid. ¹H NMR: (THF-d₈, −20 °C), 6.58 (s, 10H, C₅H₅), 1.17 (s, 9H, C(CH₃)₃). ¹³C{¹H} NMR: (THF-d₈, −20 °C), 120.4 (q, J_C–F = 318 Hz, OSO₂CF₃), 118.8 (s, C₅H₅), 83.6 (s,OOCMe₃), 26.7 (s, OOC(CH₃)₃). ¹⁹F NMR: (THF-d₈, −20 °C), 79.2 (s, OSO₂CF₃).

Cp₂Ti(OO^tBu)Br (3)

Complex 2 was generated as above, quickly dissolved in THF (25 mL), and at −78 °C ⁿBu₄NBr (106 mg, 0.33 mmol) was added. The reaction mixture was stirred at −20 °C for 2 h and the THF was removed in vacuo at ambient temperature. Et₂O (25 mL) was added and the resulting solids were filtered and washed with excess Et₂O until the filtrate was no longer yellow. The volume of Et₂O was reduced to ~5 mL and pentane (5 mL) was added, yielding 71 mg of yellow 3 (62 % based on 1). X-ray quality crystals were obtained by the slow evaporation of a saturated CH₂Cl₂ solution of 3 at −5 °C for 48 h. ¹H NMR (THF-d₈, −20 °C): 6.40 (s, 10H, C₅H₅), 1.08 (s, 9H, C(CH₃)₃). ¹³C{¹H} NMR (THF-d₈, −20 °C): 117.1 (s, C₅H₅), 82.7 (s,OOCMe₃), 27.0 (s, OOC(CH₃)₃). IR (CH₂Cl₂, cm⁻¹): 2985 m (CH); 1449 w, 1361 w, 1188 w (^tBu); 1016 w (CO); 819 m (OO). MS: For ¹²C₁₄¹H₁₉⁷⁹Br¹⁶O₂⁴⁸Ti (M⁺): calcd, 346.00478; found, 346.00486. Anal. Calcd for C₁₄H₁₉BrO₂Ti: C, 48.45%; H, 5.52%; Fnd: C, 48.23%; H, 5.48%.

[Cp₂Ti(OO^tBu)OEt₂][BAr'₄] (4)

An NMR tube was charged with 10.0 mg (0.033 mmol) of 1 and 36.9 mg (0.033 mmol) of AgBAr'₄·2Et₂O and ~0.7 mL of CD₂Cl₂ was vacuum transferred in at low temperature. The NMR tube was sealed and, keeping it at or below −20 ºC, shaken thoroughly and then placed vertically for 30 min to allow the AgCl to settle. ¹H NMR: (CD₂Cl₂, −20 °C), 7.73 (br s, 8H, o-Ar'), 7.58 (s, 4H, p-Ar'), 6.51 (s, 10H, C₅H₅), 3.61 (br q, 4H, ³J_HH = 7 Hz, O(CH₂CH₃)₂), 1.29 (t, 6H, ³J_HH = 7 Hz, O(CH₂CH₃)₂), 1.16 (s, 9H, C(CH₃)₃). ¹³C{¹H} NMR: (CD₂Cl₂, −20 °C), 161.5 (q, J_C–B = 50 Hz, B–ipso-Ar'), 134.8 (s, o-Ar'), 129.0 (qq, J_C–F = 32, 3 Hz, m-Ar'), 124.8 (q, J_C–F = 272 Hz, CF₃), 118.6 (s, p-Ar'), 117.8 (s, C₅H₅), 85.0 (s, OOCMe₃), 66.8 (s, Ti-O(CH₂CH₃)₂), 26.1 (s, OOC(CH₃)₃), 15.7 (br s, Ti-O(CH₂CH₃)₂).

[Cp₂Ti(OO^tBu)Et₃P][BAr'₄] (5)

Complex 4 was prepared as above except that the reaction was run in the glovebox freezer in a 1 dram vial with the reagents, solvent and apparatus pre-cooled to −20 ºC. CD₂Cl₂ (~0.7 mL) was added to the solids, the reaction mixed, and after 10 min filtered through two 13 mm nylon-membrane syringe filters (0.45 μm pore size, Gelman Acrodisc) using a 1 mL glass syringe. The filtrate was injected into a J. Young valved sealable NMR tube containing Et₃P (5 μL, 0.033 mmol) at −20 ºC. ¹H NMR: (CD₂Cl₂, −20 °C), 7.73 (br s, 8H, o-Ar'), 7.58 (s, 4H, p-Ar'), 6.27 (d, J_H–P = 3 Hz, 10H, C₅H₅), 1.96 (br m, 6H, P(CH₂CH₃)₃), 1.10 (dt, J_H–P = 7 Hz, J_H–H = 8 Hz, 9H, P(CH₂CH₃)₃), 1.16 (s, 9H, C(CH₃)₃). ¹³C{¹H} NMR: (CD₂Cl₂, −20 °C), 162.0 (q, J_C–B = 50 Hz, B–ipso-Ar'), 135.0 (s, o-Ar'), 129.0 (qq, J_C–F = 31, 3 Hz, m-Ar'), 124.8 (q, J_C–F = 272 Hz, CF₃), 117.8 (s, p-Ar'), 115.6 (d, J_C–P = 4 Hz, C₅H₅), 85.0 (s,OOCMe₃), 26.0 (s, OOC(CH₃)₃), 17.3 (d, J_C–P = 13 Hz, P(CH₂CH₃)₃), 9.0 (br s, P(CH₂CH₃)₃). ³¹P{¹H} NMR: (CD₂Cl₂, −20 °C), 27.6 (s).

Et₂PO^tBu

Et₂PCl (100 mg, 0.80 mmol, Acros Organics) was added to a stirring solution of Et₃N (809 mg, 8.00 mmol) and ^tBuOH (593 mg, 8.00 mmol) in dry benzene (10 mL). After stirring overnight, a white precipitate (presumably Et₃NHCl) was filtered off and the volatiles were removed in vacuo, leaving Et₂PO^tBu as a thick oil. ³¹P{¹H} NMR (CD₂Cl₂): 109.6 ppm. MS: 162 m/z.

¹H NMR kinetics

In a typical reaction, a vial was charged with 1 (10.0 mg, 0.033 mmol), C₆Me₆ (1.0 mg) as an internal standard, and Ph₃P (8.7 mg, 0.033 mmol). CD₂Cl₂ (0.5 mL) was added and the resulting solution was rapidly transferred to a J. Young-valved sealable NMR tube, sealed, and quickly frozen in liquid N₂. The NMR tube was warmed to room temperature at the spectrometer and spectra were acquired every 5 min for at least 3 h, employing a 10 second delay between pulses for accurate integration. Peaks in each spectrum were integrated individually vs. the internal standard and the residual solvent peak using WinNuts.

X-ray crystallographic studies

Crystals were mounted on a glass capillary with oil at −143 °C. Intensity data were collected on an Enraf-Nonius KappaCCD diffractometer equipped with a fine focus Mo-target X-ray tube. The data were integrated and scaled using hkl-SCALEPACK (hkl-2000 for 3).⁴⁶ These programs apply a multiplicative correction factor (S) to the observed intensities (I): S = (e^{−2B (sin2θ)/λ2})/scale. S is calculated from the scale and the B factor, which is determined for each frame and is then applied to I to give the corrected intensity (I_corr). Solution by direct methods (SIR 97) produced a complete heavy atom-phasing model.⁴⁷ All hydrogen atoms were placed using a riding model. All non-hydrogen atoms were refined anisotropically by full-matrix least squares (SHELXL-97, Table 2).⁴⁸

Computational Studies

Geometries were optimized and vibrational analyses were performed at the density functional (DFT) level of theory using the 6–31G* basis set.⁴⁹ The hybrid B3LYP functional was employed, which combines Becke’s gradient-corrected exchange functional⁵⁰ with the gradient-corrected correlation functional of Lee, Yang and Parr.⁵¹ The vibrational analyses were used to confirm the nature of the stationary points, and the unscaled vibrational frequencies were used to compute thermal contributions to enthalpies. All of the calculations were carried out with the Gaussian 03 suite of programs.⁵²

Supplementary Material

SI1. Supporting Information.

Metrical data for DFT calculated structures and intermediates, Eyring plots for the decomposition of 1 and 3–5 in CD₂Cl₂, and CIF files for 1 and 3.

Scheme 1.

Synthetic routes to complexes 2–5 from 1.