Z-Selective Olefin Metathesis Reactions Promoted by Tungsten Oxo Alkylidene Complexes

Abstract

Addition of LiOHMT (OHMT = O-2,6-dimesitylphenoxide) to W(O)(CH-t-Bu)(PMe₂Ph)₂Cl₂ led to WO(CH-t-Bu)Cl(OHMT)(PMe₂Ph) (4). Subsequent addition of Li(2,5-Me₂C₄H₂N) to 4 yielded yellow W(O)(CH-t-Bu)(OHMT)(Me₂Pyr)(PMe₂Ph) (5). Compound 5 is a highly effective catalyst for the Z-selective coupling of selected terminal olefins (at 0.2% loading) to give product in >75% yield with >99% Z configuration. Addition of two equivalents of B(C₆F₅)₃ to 5 led to catalyst activated at the oxo ligand by B(C₆F₅)₃. 5^.B(C₆F₅)₃ is a highly active catalyst that produces thermodynamic products (~20% Z).

Early in the development of olefin metathesis catalysts that contain tungsten, it was shown that metathetically more active and reproducible systems were produced when tungsten oxo complexes were deliberately employed or were present as impurities in WCl₆.¹ The possibility that oxo alkylidene complexes, e.g., W(O)(CHR)X₂ (where X is a chloride, alkoxide, etc.), are the true catalysts in at least some of the “classical” olefin metathesis systems became more likely when 1 (L = PMe₃ and other phosphines) was prepared and isolated in good yield.² Compound 1 was the first high oxidation state tungsten alkylidene complex that would both (i) metathesize terminal and internal olefins (in the presence of a trace of AlCl₃) and (ii) produce a new alkylidene that could be observed as a consequence of olefin metathesis.

(1)

By the time 1 was discovered, tantalum alkylidene complexes had been turned into functional olefin metathesis catalysts through use of alkoxides as ligands.³ Therefore, although some attempts were made to prepare a W(O)(CH-t-Bu)(OR)₂ species from 1, no bisalkoxide species could be isolated. In view of the synthetic problems encountered upon attempted alkylation of oxo complexes, including removal of the oxo ligand entirely,¹ⁱ and to protect alkylidenes against bimolecular decomposition, attention turned to the synthesis of imido alkylidene complexes of W and Mo, especially those containing a phenylimido ligand such as N(2,6-i-Pr₂C₆H₃).⁴ Consequently, interest in oxo alkylidene complexes in the last 25 years has been sparce.⁵

The most recent development in Mo and W imido alkylidene chemistry has been monoaryloxide monopyrrolide (MAP) complexes.⁶ One of the most interesting discoveries is the ability of some MAP catalysts to promote Z selective metathesis reactions as a consequence of the presence of a relatively “large” aryloxide and “small” imido group.⁷ The preferred metal for Z selective couplings of terminal olefins at this time appears to be tungsten and the most successful aryloxide ligand has been O-2,6-(2,4,6-triisopropylphenyl)₂C₆H₃ or OHIPT. (The more active molybdenum complexes⁸ appear to isomerize the Z product to E.) It has been proposed that the high steric demands of the OHIPT ligand force all metallacyclobutane substituents to one side of the metallacycle ring, and therefore allow only Z products to form. Since an oxo ligand is smaller than any NR ligand, the question arose as to whether MAP versions of tungsten oxo alkylidene complexes would be useful Z selective catalysts.

We chose to attempt to prepare W(O)(CH-t-Bu)(OHIPT)(Me₂Pyr) (where Me₂Pyr = 2,5-dimethylpyrrolide) from W(O)(CH-t-Bu)(PMe₂Ph)₂Cl₂ (1a), hoping that both PMe₂Ph ligands would dissociate from the metal in the crowded coordination sphere. The reaction between WO(CH-t-Bu)Cl₂(PMe₂Ph)₂ and LiOHIPT in benzene at 22 °C for 14h led to isolation of off-white WO(CH-t-Bu)Cl(OHIPT)(PMe₂Ph) (2) in 60% yield (equation 2). Two isomers of 2 are present in a 3:2 ratio according to ¹H, ¹³C, and ³¹P NMR spectra. Both are syn alkylidenes on the basis of J_CαH values for the alkylidene of 123 Hz (major isomer) and 117 Hz (minor isomer). The phosphine remains bound to tungsten on the NMR time scale (J_PW = 420 Hz and 379 Hz, respectively) at 22 °C. An X-ray crystal structure (see Supporting Information) revealed a distorted square-pyramidal geometry with the neopentylidene ligand in the apical position and the phosphine ligand trans to chloride. The alkylidene was found to be disordered over syn and anti orientations in a 91:9 ratio. The other isomer of 2 could be (for example) one in which the OHIPT and Cl ligands (eq 2) have switched positions.

(2)

Treatment of 2 with Li(Me₂Pyr) in benzene at 60 °C for 16h led to formation of yellow W(O)(CH-t-Bu)(OHIPT)(Me₂Pyr) (3) in 80% isolated yield. An X-ray structure of 3 showed it to have a pseudotetrahedral geometry, a syn alkylidene, and an η¹-Me₂Pyr ligand (Figure 1). We had considered the possibility that PMe₂Ph was lost from the coordination sphere because the pyrrolide ligand was bound in an η⁵ fashion, thereby producing an 18 electron count at the metal. However, other steric factors alone appear to be sufficient to cause 14 electron 3 to be formed.

Figure 1.

Thermal ellipsoid drawing (50% probability) of syn-W(O)(CH-t-Bu)(OHIPT)(Me₂Pyr) (3). Hydrogen atoms have been omitted for clarity. Selected bond distances (Å) and angles (°): W1–C1 = 1.886(3), W1–O2 = 1.695(3), W1–O1= 1.868(2), W1–N1 = 2.001(2), W1–O1–C21 = 166.9(2), W1–C1–C2 = 136.7(3).

The analogous reaction between 1 and LiOHMT (OHMT = O-2,6-dimesitylphenoxide) in benzene at 22 °C for 3 h led to isolation of yellow WO(CH-t-Bu)Cl(OHMT)(PMe₂Ph) (4) in 70% yield (equation 3). As with 2, the ¹H NMR spectrum of the product contains two alkylidene doublet resonances that correspond to two isomers of 4 in a 87:13 ratio. The values of ¹J_CH (122 and 116 Hz) suggest that both isomers are syn alkylidenes. Addition of Li(Me₂Pyr) to 4 in toluene at −30 °C followed by stirring the mixture at 22 °C for 10h led to yellow W(O)(CH-t-Bu)(OHMT)(Me₂Pyr)(PMe₂Ph) (5) in 70% isolated yield. An X-ray structure of 5 (Figure 2) showed it to be a square pyramid with the syn neopentylidene in the apical position and the phosphine bound trans to the pyrrolide.

Figure 2.

Thermal ellipsoid drawing (50% probability) of syn-W(O)(CH-t-Bu)(OHMT)(Me₂Pyr)(PMe₂Ph) (5). Hydrogen atoms have been omitted for clarity. Solvent molecules are not shown. Selected bond distances (Å) and angles (°): W1–C1 = 1.900(3), W1–O2 = 1.717(2), W1–O1 = 1.964(2), W1–N1 = 2.074(2), W1–P1 = 2.580(1), W1–O1–C21 = 159.8(2), W1–C1–C2 = 141.0(2).

(3)

The PMe₂Ph ligand in 5 is partially dissociated at room temperature and rapidly exchanging on and off the metal. The alkylidene resonance is broad and its chemical shift is concentration dependent (8.57–9.14 ppm for 4 mM – 48 mM solutions in C₆D₆). Variable temperature ¹H and ³¹P NMR studies of a 20 mM solution of 5 in CD₂Cl₂ showed that the phosphine is “bound” below −30 °C as indicated by a sharp ³¹P signal corresponding to the coordinated ligand (1.80 ppm, ¹J_PW = 289 Hz). On the basis of the chemical shift for free and coordinated phosphine the value of the equilibrium constant for phosphine dissociation can be estimated as 0.015 M at room temperature. This value corresponds to 57% dissociation of phosphine in a 20 mM solution of 5 in C₆D₆.

Both 3 and 5 react with ethylene to give an unsubstituted metallacyclobutane complex (and t-butylethylene) that has a square pyramidal structure (presumably with the oxo ligand in the apical position) on the basis of chemical shifts of metallacycle protons in the range 0.7–4.5 ppm.⁹ The reaction of 3 with ethylene is relatively slow and what we propose is an intermediate square pyramidal β-t-butylmetallacyclobutane complex can be observed before free t-butylethylene is formed. In the case of compound 5, a methylidene complex is formed in addition to the unsubstituted square pyramidal metallacycle. In both systems the unsubstituted metallacycles slowly decompose over a period of 24 h to unidentified products. A square pyramidal metallacyclobutane made from imido alkylidenes has been proposed to be further from the transition state for olefin loss than is the alternative TBP metallacycle.^10a Extensive calculations have been performed on several high oxidation state metathesis systems that include metallacyclobutane complexes.^10b,c

Both 3 and 5 serve as initiators for the polymerization of 5,6-dicarbomethoxynorbornadiene (DCMNBD). The polymerization of 50 equiv of DCMNBD is relatively slow (hours) with 3 and propagation is faster than initiation. The resulting polymer is >99% cis, 90% syndiotactic. The polymerization of 50 equiv of DCMNBD with 5 is relatively fast (minutes) and all initiator is consumed. The resulting polymer is >99% cis, 98% syndiotactic.¹¹ These results suggest that the steric crowding in 3 is significantly greater than in 5 after phosphine is lost, and is in fact too great to allow formation of highly regular cis,syndiotactic-polyDCMNBD from 3.

Homocoupling of neat terminal olefins with 3 takes place slowly (hours) at room temperature. In contrast, 5 was found to be highly active and highly Z-selective (Table 1). A catalyst loading as low as 0.2 mol% yielded up to 86% conversion in 6 h for several of the six chosen substrates. No trans product could be observed in ¹H NMR spectra of the Z products (see Supporting Information).

Table 1.

Conversions of Neat Terminal Olefins to Homocoupled >99% Z Metathesis Products Promoted by 5.^a

time	substrate/cat loading (mol%)
	S1/ 0.2 %	S2/ 0.2 %	S3/ 0.2 %	S4/ 0.2 %	S5/ 0.2 %	S6/ 1 %
10 min	28%	44%	65%	-	28%	-
30 min	39%	67%	75%	-	39%	-
1 h	47%	79%	75%	2%	47%	10%
6 h	66%	86%	-	-	73%b	-
24 h	72%	88%	-	11%	-	59%

^aS1 = 1-octene, S2 = allylbenzene, S3 = allylboronic acid pinacolate ester, S4 = allylSiMe3, S5 = 1-decene, S6 = Methyl-10-undecenoate.

^bThe aliquot was taken after 7 h.

Only a small increase in conversion was found for >6 h reaction times, which suggests that the majority of the catalyst has decomposed at this stage. Decomposition of a catalyst prior to isomerization of the Z product to E can be a desirable feature of the coupling reaction. The reactions were run on a 200 mg scale in a closed vessel with a volume of ~20 mL. Homocoupling of 1-decene at 0.5 Torr did not show a significant increase in turnover compared to the reaction carried out under 1 atm of nitrogen. We ascribe the relatively low turnover in the case of allylTMS (S4) to steric issues, and in the case of methyl-10-undecenoate (S6, at 1% catalyst loading) to ester binding to W. These results should be compared with those obtained employing a W(N-3,5-Me₂C₆H₃) catalyst system.^7f

A long-standing question in classical olefin metathesis catalyst systems based on tungsten has been the role of a Lewis acid. One might expect that a Lewis acid in 5 could bind to the oxo ligand and thereby create a more electrophilic metal center and more reactive catalysts. Indeed we find that addition of Lewis acids to 5 significantly speeds up metathesis reactions. For example, addition of two equivalents of B(C₆F₅)₃ to 5 resulted in a catalyst that converted 90% 1-octene to 7-tetradecene in one hour at 22 °C (0.2 mol% loading). However, the 7-tetradecene is only 20% Z. Since pure Z-7-tetradecene (in C₆D₆) is isomerized to a 78:22 mixture of E- and Z-tetradecene by 1 mol% 5 in the presence of 2 equiv of B(C₆F₅)₃ in 15 minutes, any Z-7-tetradecene that is formed initially in the homocoupling reaction should be isomerized rapidly to a 4:1 E:Z mixture.

Addition of two equivalents of B(C₆F₅)₃ to 5 led to formation of (Me₂PhP)[B(C₆F₅)₃] and 5^.B(C₆F₅)₃. The B(C₆F₅)₃ in 5^.B(C₆F₅)₃ is labile at room temperature as demonstrated by a broadened alkylidene signal in the ¹H NMR spectrum at 7.30 ppm. The ¹H NMR spectrum of a 45 mM solution of 5^.B(C₆F₅)₃ at −60 °C shows a sharp alkylidene resonance at 7.06 ppm. An X-ray structure of 5^.B(C₆F₅)₃ showed that B(C₆F₅)₃ is coordinated to the oxo ligand (Figure 3). The W1–O2–B1 unit is bent (W1–O2–B1 angle is 159.9(1)°). The W1–O2 distance is elongated (1.759(2) Å) relative to that in 5 (1.717(2) Å) or in 3 (1.695(3) Å) and is slightly shorter than in reported B(C₆F₅)₃ adducts of tungsten oxo complexes.¹² A relatively weak coordination of B(C₆F₅)₃ to the oxo is also indicated by the B1–O2 bond length (1.571(3) Å), which is longer than in any B(C₆F₅)₃ adducts of transition metal oxo complexes (1.484(3)–1.558(2) Å) in the literature. The average values of the C–B–C and O–B–C angles (112.6° and 106.1°, respectively) also suggest that B(C₆F₅)₃ is relatively weakly coordinated.

Figure 3.

Thermal ellipsoid drawing (50% probability) of W(O)(B(C₆F₅)₃)(CH-t-Bu)(OHMT)(Me₂Pyr) (5^.B(C₆F₅)₃). Hydrogen atoms have been omitted for clarity. Selected bond distances (Å) and angles (°): W1–C1 = 1.868(2), W1–O2 = 1.759(2), W1–O1 = 1.860(2), W1–N1 = 1.968(2), B1–O2 = 1.571(3), W1–O1–C21 = 150.9(1), W1–C1–C2 = 155.4(2).

We conclude that tungsten oxo alkylidene complexes are effective Z-selective catalysts for the metathesis coupling of terminal olefins. We ascribe the selectivity to the small size of the oxo ligand relative to OHMT, the low rate of isomerization of the initial Z product relative to coupling of terminal olefins, and decomposition of the active catalyst under the conditions employed.