Development of a Method for the Preparation of Ruthenium Indenylidene-Ether Olefin Metathesis Catalysts

The reactions between several derivatives of 1-(3,5-dimethoxyphenyl)-prop-2-yn-1-ol and different ruthenium starting materials [i.e., RuCl2(PPh3)3 and RuCl2(p-cymene)(L), where L is tricyclohexylphosphine di-t-butylmethylphosphine, dicyclohexylphenylphosphine, triisobutylphosphine, triisopropylphosphine, or tri-n-propylphosphine] are described. Several of these reactions allow for the easy, in-situ and atom-economic preparation of olefin metathesis catalysts. Organic precursor 1-(3,5-dimethoxyphenyl)-1-phenyl-prop-2-yn-1-ol led to the formation of active ruthenium indenylidene-ether complexes, while 1-(3,5-dimethoxyphenyl)-prop-2-yn-1-ol and 1-(3,5-dimethoxyphenyl)-1-methyl-prop-2-yn-1-ol did not. It was also found that a bulky and strong σ-donor phosphine ligand was required to impart good catalytic activity to the new ruthenium complexes.


Introduction
Metal alkylidene complexes have been the focus of intense research in synthetic chemistry [1]. Most notable are the Schrock molybdenum and tungsten alkylidene [2] and the Grubbs ruthenium alkylidene complexes [3], which are excellent catalysts for olefin metathesis and have enabled an astonishingly broad spectrum of applications in organic and polymer synthesis [4]. Molybdenum

OPEN ACCESS
alkylidene complexes find many important uses including in enantioselective metathesis reactions [5], while certain tungsten alkylidene systems were recently shown to be suitable catalysts for Z-selective metathesis [6]. Ruthenium alkylidene systems have been the most widely used olefin metathesis catalysts in industrial and academic laboratories, because they combine robustness, functional group tolerance, excellent activity, and good selectivity including enantioselectivity [5,[7][8][9] and Z-selectivity [10]. Many ruthenium-based olefin metathesis catalysts are commercially available including members of the ruthenium benzylidene (1) [11,12], benzylidene-ether (2) [13,14], and indenylidene (3) families ( Figure 1) [15,16]. The preparations of these complexes are not straightforward, but involve several steps and require crystallization of the catalysts to remove phosphine byproducts. For example, the most practical method to produce the 1st generation Hoveyda catalyst 2a consists of first preparing and isolating the 1st generation Grubbs complex 1a in a two-step, one-pot process [11], and subsequently reacting 1a with an alkoxystyrene molecule to give 2a and one equivalent of tricyclohexylphosphine (PCy 3 ) [13]. The end of the process requires isolating the catalyst 2a away from the liberated phosphine. In addition to being cumbersome, this catalyst synthesis is not atom economic, wasting several equivalents of triphenylphosphine (PPh 3 ) and one equivalent of the expensive PCy 3 ligand. Therefore, we became motivated to develop a direct, in-situ, and atom-economic method for the synthesis of olefin metathesis catalysts.
The preparation of the ruthenium indenylidene complex 3a is comparable with that of 1a in some respects, because it also involves a two-step, one-pot process followed by isolation of the catalyst away from the liberated PPh 3 . On the other hand, making 3a seems more attractive because it avoids the use of diazo compounds and does not require cooling of the reaction mixtures. However, an analysis of the history of 3a's discovery and additional studies reveals that its formation is not as straightforward as it initially appears. Indeed, the reaction between RuCl 2 (PPh 3 ) 3 and 1,1diphenylprop-2-yn-1-ol (HC≡CCPh 2 OH) was first studied by Hill and reported to yield allenylidene complex RuCl 2 (PPh 3 ) 2 (=C=C=CPh 2 ) [17]. Conversely, Nolan [18] and Fürstner [19] described that the same reaction under very similar conditions (reflux in THF for 2 h) yields the indenylidene complex RuCl 2 (PPh 3 ) 2 (PhInd) (where PhInd is a 3-phenyl-1-indenylidene fragment). Schanz found the preparation of RuCl 2 (PPh 3 ) 2 (PhInd) under the aforementioned conditions to be unreliable and showed that it affords the dinuclear species (PPh 3 ) 2 ClRu(µ-Cl) 3 Ru(PPh 3 ) 2 (=C=C=CPh 2 ) [20]. It is also interesting to note that [RuCl 2 (p-cymene)] 2 reacts with HC≡CCPh 2 OH in the presence of two equivalents of PCy 3 under similar conditions as above to give RuCl 2 (PCy 3 ) 2 (=C=C=CPh 2 ), which does not rearrange to RuCl 2 (PCy 3 ) 2 (PhInd) [18]. In contrast, the formation of RuCl 2 (PPh 3 ) 2 (PhInd) from the reaction of RuCl 2 (PPh 3 ) 3 and HC≡CCPh 2 OH is thought to go through an allenylidene-to-indenylidene rearrangement involving an electrophilic aromatic substitution of one of the allenylidene's phenyl groups [20]. A similar rearrangement was observed by Dixneuf in a cationic ruthenium allenylidene system [21].
Altogether, we were attracted by the use of propargyl alcohol derivatives to generate metal alkylidene complexes, but desired to design 1-phenylprop-2-yn-1-ol precursors that would favor the formation of metal indenylidene over metal allenylidene complexes. We reasoned that prop-2-yn-1-ol molecules containing an activated phenyl ring would be suitable precursors. We have previously communicated the convenient, atom-economic, and in-situ preparation of a new indenylidene-ether olefin metathesis catalyst using 1-(3,5-dimethoxyphenyl)-1-phenylprop-2-yn-1-ol [22]. Herein, we describe the path that led to the development of this synthesis as well as the reaction of other organic precursors with different ruthenium starting materials.

Reaction of 1-Phenylprop-2-yn-1-ol Derivatives with Different Ruthenium Starting Materials
Compound 4a was reacted with RuCl 2 (PPh 3 ) 3 in refluxing THF-d 8 . 31 P-NMR spectroscopy revealed that complete disappearance of RuCl 2 (PPh 3 ) 3 occurred after 3 h to give free PPh 3 (singlet at −4.9 ppm) and two new species in a ~6:1 ratio (singlet at 53.4 ppm for the major species and singlet at 27.5 ppm for the minor species). The methoxy groups of both the major and minor species seemed to be inequivalent, according to 1 H-NMR spectroscopy, suggesting that the products could be ruthenium indenylidene complexes. Nevertheless, a firmer elucidation of the nature of the organometallic products was hampered by the lack of information in the 1 H-NMR spectrum and the low intensities of the low-field signals in the 13 C-NMR spectrum. Thus, we began to explore the reaction of RuCl 2 (PPh 3 ) 3 with the 13 C-doubly labeled organic precursor 4a-13 C 2 in refluxing THF-d 8 . After 3 h of reaction, the 13 C{ 1 H} spectrum of the mixture features a doublet of doublets at 288.9 ppm ( 1 J C-C = 49.6 Hz; 2 J C-P = 12.2 Hz) for the 13 C  nucleus of the major species, consistent with the C  nucleus coupling to one 31 P nucleus and to the 13 C  nucleus. Further down field, the spectrum exhibits a doublet of triplets at 301.1 ppm ( 1 J C-C = 41.3 Hz; 2 J C-P = 14.2 Hz) for the 13 C  nucleus of the minor species, consistent with the C  nucleus coupling to two 31 P nuclei and to the 13 C  nucleus. The spectrum also exhibits a doublet of doublets at 143.8 ppm for the 13 C  nucleus of the major species and a doublet of triplets at 140.2 ppm for the 13 C  nucleus of the minor species, but lacks any resonance in the 250-220 ppm region. These data suggest that no appreciable amount of ruthenium allenylidene complex was formed and are consistent with the major species being a mono-phosphine ruthenium indenylidene complex 5a-13 C 2 and the minor species a bis-phosphine ruthenium indenylidene 5b-13 C 2 (Scheme 3) [17][18][19][20]. The 31 P-NMR spectrum shows a doublet of doublets at 53.4 ppm for the major species and a doublet of doublets at 27.5 ppm for the minor species. The configuration of the ligands around the ruthenium center, including whether the chlorides adopt a cis-or trans-arrangement, is not known based on the above data. 13

Scheme 3. Reaction between RuCl 2 (PPh 3 ) 3 and 4a-
The mixture of 5a-13 C 2 and 5b-13 C 2 in THF-d 8 obtained above was treated with three equivalents of PCy 3 per ruthenium atom at room temperature to give a mixture of organometallic species containing a new mono-phosphine ruthenium indenylidene complex 6a-13 C 2 as a major component (Scheme 4). 6a-13 C 2 is characterized by a doublet at 48.6 ppm ( 2 J C-P = 11.1 Hz) in the 31 P-NMR spectrum, and in the 13 C-NMR spectrum, by a doublet of doublets at 287.0 ppm ( 1 J C-C = 49.0 Hz; 2 J C-P = 11.2 Hz) for the 13 C  nucleus and a doublet at 129.2 ppm ( 1 J C-C = 49.0 Hz) for the 13 C  nucleus.
Complex 6a-13 C 2 can also be generated as the major product by reacting RuCl 2 (p-cymene)(PCy 3 ) with organic precursor 4a-13 C 2 upon reflux in THF-d 8 for 16 h (Scheme 5) [22]. This reaction also produces a minor product 6b-13 C 2 , whose 31 P-NMR spectrum features a doublet at 68.1 ppm ( 2 J C-P = 15.2 Hz) and whose 13 C-NMR spectrum shows a doublet of doublets at 256.2.0 ppm ( 1 J C-C = 47.8 Hz; 2 J C-P = 15.0 Hz) for the 13 C  nucleus and a doublet at 129.2 ppm ( 1 J C-C = 48.9 Hz) for the 13 C  nucleus. A mixture of non-labeled 6a/6b complexes was prepared on a gram scale by a similar procedure. However, efforts to isolate and purify the major and minor products 6a and 6b by silica gel chromatography and crystallization have been unsuccessful.
Bruneau and coworkers independently prepared a related complex by a similar approach utilizing 1,1-bis-(3,5-diisopropoxyphenyl)prop-2-yn-1-ol as an organic precursor [23]. The NMR spectroscopy data for their complex is very similar to those for our minor species (6b-13 C 2 ): The 31 P-NMR spectrum shows a resonance at 68.2 ppm and the 13 C-NMR spectrum shows resonances at 258.8 ppm ( 2 J C-P = 14.2 Hz) and 136.3 ppm for the 13 C  and 13 C  nuclei, respectively. Additionally, Bruneau and coworkers obtained a crystal structure of their complex showing it to be a mono-PCy 3 ruthenium indenylidene-ether complex where the phosphine and ether ligands are trans to each other. We propose that 6a and 6b may be isomers, where the minor species 6b possesses a structure similar to that of Bruneau's complex (trans-phosphine-ether arrangement) and where the major complex 6a adopts a cis-phosphine-ether arrangement as the more stable isomer, due to the reduced steric of the methoxy group in 6 compared to the bulkier isopropoxy group in Bruneau's complex. Similarly, a relatively unhindered ruthenium alkylidene-pyridine complex supported by a NHC ligand was shown to exist as a major isomer adopting a cis-NHC-pyridine arrangement and a minor isomer with a trans-NHC-pyridine configuration [24].
A solution of complexes 6a and 6b, prepared in situ using the non-labeled organic precursor 4a following a similar procedure as that shown in scheme 5, can be used without additional treatment to catalyze ring-closing metatheses (RCM) to produce 5-, 6-, and 7-membered disubstituted cycloalkenes with activities similar to those of the 1st generation Hoveyda catalyst 2a under standard conditions [22,25]. In order to explore the influence of the phosphine ligand on the activity of these in-situ generated ruthenium indenylidene-ether complexes, other RuCl 2 (p-cymene)(L) starting materials, where L is 3 ], or tri-n-propylphosphine [P(nPr) 3 ], were reacted with organic precursor 4a in refluxing THF-d 8 for 16 h. 31 P-NMR spectroscopy showed that each one of these reactions affords new organometallic species ( Table 1). The activity of these complexes in ring-closing metathesis is described in Section 2.3. Table 1. 31 P-NMR shifts for RuCl 2 (p-cymene)(L) and for the major product of the reaction between RuCl 2 (p-cymene)(L) and 4a. a

RCM Activity of the Ruthenium Indenylidene-Ether Complexes
As mentioned above, a solution of 6a/6b generated in situ efficiently promotes the formation of 5-, 6-, and 7-membered disubstituted cycloalkenes by RCM. Although the activity of our system is very similar to that of the 1st generation Hoveyda catalyst 2a in the RCM of diethyl diallylmalonate (DEDAM) under standard conditions [22,25], it is interesting to note that Bruneau's catalyst exhibits a longer initiation period (Figure 2) [23]. Indeed, at the 30 min time point, Bruneau's catalyst has converted about 33% of the substrate to product, while our catalyst has reached greater than 90% conversion. A possible explanation for this is that the bulkier indenylidene ligand of Bruneau's catalyst hinders the rotation around the Ru=C bond of the ruthenium indenylidene fragment, a rotation that may be necessary for the formation of the rotamer that initiates the olefin metathesis reaction. A similar rotation takes place in the Hoveyda catalysts, [28,29] whose activation is thought to involve dissociative and associative interchange pathways [30].
The activity of the solutions generated by the reaction of RuCl 2 (p-cymene)(L) and organic precursor 4a (Table 1) was compared to that of the solution of 6a/6b (L = PCy 3 ) in the RCM of DEDAM. The catalytic system 6a/6b bearing the PCy 3 ligand is the most active, reaching greater than 97% conversion within 30 min at 1.0 mol % catalyst loading ( Table 2; entry 1). The catalysts supported by P(tBu) 2 Me or P(iPr) 3 ligands are less effective than 6a/6b ( Table 2; compare entries 2 and 6 to entry 1), but are still able to achieve greater than 90% conversion within 60 min at 2.0 mol % catalyst loadings (Table 2; entries 3 and 7). Conversely, the complexes containing the P(Cy) 2 Ph, P(iBu) 3 or P(nPr) 3 ligands show very low activity in the RCM of DEDAM under the tested conditions (Table 2;    These results follow the ligand effects observed by Grubbs and coworkers, namely that larger and more electron-donating phosphine ligands lead to more active catalysts [31]. These trends are easily rationalized for the 1st generation Grubbs systems which enter the catalytic cycle via a dissociative substitution of a phosphine with an olefin [32]. Thus, the ligand dissociation is favored by the large steric hindrance and the strong trans influence of ligands such as PCy 3 , producing more active species. The explanation for the trends within the ruthenium indenylidene-ether set of catalysts is less intuitive. Nevertheless, it seems reasonable that phosphine ligands with stronger electron-donating abilities may turn over faster by accelerating the olefin binding step [33] or by stabilizing the metallacyclobutane intermediate and metathesis transition states [34], as it was originally proposed in the case of N-heterocyclic carbene (NHC) versus phosphine ligands [35]. It is also conceivable that bulkier phosphine ligands are required to influence the Ru=C bond rotation favoring the formation of the active rotamer and driving the reaction toward the less sterically hindered intermediates and transition states [33,34].

General
NMR spectra were recorded on a Bruker 400 MHz NMR spectrometer running Xwin-NMR software. Chemical shifts are reported in parts per million (ppm) downfield from tetramethylsilane (TMS) with reference to internal solvent for 1 H-NMR and 13 C-NMR spectra. Chemical shifts are reported in parts per million (ppm) downfield from H 3 PO 4 for 31 P-NMR spectra. All glassware was oven dried. Unless noted otherwise, all reactions were conducted under an atmosphere of argon (in an argon-filled glove box or under argon using Schlenk techniques). All organic solvents were dried by passage through solvent purification columns containing activated molecular sieves. All other commercial chemicals were used as obtained. Organic precursors 4a and 4a-13 C 2 [22], and RuCl 2 (p-cymene)(L) [36] [where L = PCy 3 , P(tBu) 2 Me, PCy 2 Ph, P(iBu) 3 , P(iPr) 3 , and P(nPr) 3 ] were prepared according to literature procedures.

Preparation of 1-(3,5-Dimethoxyphenyl)-1-methylprop-2-yn-1-ol (4c)
Powdered potassium carbonate (27.3 g, 197.5 mmol) was added to a suspension of 3,5-dihydroxyacetophenone (5.0 g, 32.9 mmol) in acetone (50 mL) and the mixture was stirred vigorously for 20 min. Methyl iodide (8.4 mL, 134.9 mmol) was added dropwise and the reaction mixture was refluxed for 16 h. The mixture was filtered and the solid washed with acetone. Water (100 mL) was added to the filtrate, and the product was extracted with ether (3 × 100 mL). The combined organic layers were washed with water (200 mL) before being dried with anhydrous magnesium sulfate. The filtrate was dried in vacuo to afford 3,5-dimethoxyacetophenone as a dark red oil in 93% yield.  197.76, 160.88, 139.10, 106.18, 105.34, 55.59, 26.72. A dry 100 mL reaction flask equipped with stir bar was charged with 3,5-dimethoxyacetophenone (0.5 g, 2.8 mmol) and anhydrous THF (10 mL) inside the glove box, capped with a septum, and taken out of the glove box. The mixture was placed in a 0 °C ice bath. A 0.5 M solution of ethynylmagnesium bromide in THF (9 mL, 4.5 mmol) was added dropwise under stirring. The reaction mixture was allowed to warm up to room temperature and stirred for 16 h. A 10% aqueous solution of NH 4 Cl (20 mL) was added and the mixture stirred for 30 min. The product was extracted with ether (3 × 30 mL) and the combined organic layers were washed with water (50 mL) and brine (100 mL) before being dried with anhydrous magnesium sulfate. The filtrate was dried in vacuo to afford 1-(3,5-dimethoxyphenyl)-1-methylprop-2-yn-1-ol as a brown oil in 84% yield. 1   The reaction between RuCl 2 (PPh 3 ) 3 and 4a-13 C 2 was set up following the same procedure used for the reaction between RuCl 2 (PPh 3 ) 3 and 4a (see sub-section 3.3.1.). The reaction mixture was analyzed by 13 C and 31 P-NMR. 13  3.3.4. General Procedure for the Reaction between RuCl 2 (p-cymene)(L) and Organic Precursors 4a A 2 mL vial was charged with RuCl 2 (p-cymene)(L) (50 mg), the organic precursor 4a (1.1 equiv), and THF-d 8 (0.5 mL). The mixture was then transferred to a J-Young tube, which was capped, brought out of the glove box, and placed in an oil bath set at 70 °C for 16 h. 31 P-NMR spectra were recorded (see Table 1 in Results and Discussion).

General Procedure for the Reaction between RuCl 2 (p-cymene)(PCy 3 ) and Organic Precursors 4b and 4c
A 2 mL vial was charged with RuCl 2 (p-cymene)(PCy 3 ) (30 mg, 0.051 mmol, 1.0 equiv), the organic precursor 4b or 4c (0.097 mmol, 1.9 equiv), and THF-d 8 (0.5 mL). The mixture was then transferred to a J-Young tube, which was capped and brought out of the glove box. The reactions were monitored at room temperature, 40 °C and 70 °C by 1 H-NMR and 31 P-NMR spectroscopy (see Results and Discussion).

and Organic Precursors 4b and 4c
RuCl 2 (PPh 3 ) 3 (30 mg, 0.031 mmol, 1.0 equiv) was weighed in a 2 mL vial and the organic precursor 4b or 4c (0.065 mmol, 2.1 equiv) was weighed in a separate 2 mL vial. The organic precursor was dissolved with THF-d 8 (0.5 mL) and this solution was transferred to the vial containing the RuCl 2 (PPh 3 ) 3 starting material. The mixture was then transferred to a J-Young tube, which was capped, brought out of the glove box, and placed for 3 h in an oil bath set at 70 °C. The reactions were analyzed by 1 H-NMR and 31 P-NMR spectroscopy (see Results and Discussion).

General Procedure for the RCM of Diethyl Diallylmalonate (DEDAM)
A 0.1 M stock solution of DEDAM was prepared in the glove box by dissolving DEDAM (60 mg, 0.25 mmol) in 2.44 mL of CD 2 Cl 2 . A portion of this 0.1 M DEDAM solution (0.5 mL, 50 µmol) was transferred to a NMR tube equipped with a screw-cap septum top. Separately, a 4 mL conical vial was charged with RuCl 2 (p-cymene)(L) (0.085 mmol) and 4a (25 mg, 0.094 mmol, 1.1 equiv). The vial was then filled with THF to a 1.0 mL mark (calibrated) before dropping a spin vane in the solution. The vial was sealed, brought out of the glove box, and placed under stirring for 16 h in an oil bath set at 70 °C.
A portion of this solution (6.0 µL for 1.0 mol %, and 12 µL for 2.0 mol % ruthenium loading) was added to the 0.1 M solution of DEDAM in CD 2 Cl 2 (0.5 mL, 50 µmol) by injecting it with through the septum using a syringe outside the glove box. The NMR tube was then placed in an oil bath regulated at 40 °C and the reaction mixture was analyzed by 1 H NMR spectroscopy after a period of time (30 and 60 min). The extent of conversion of the RCM reaction was determined by comparing the ratio of the integrals of the methylene protons in the substrate (dt, 2.61 ppm) with those in the product (s, 2.98 ppm).

Conclusions
Reactions between several derivatives of 1-(3,5-dimethoxyphenyl)-prop-2-yn-1-ol and different ruthenium starting materials (i.e., RuCl 2 (PPh 3 ) 3 and RuCl 2 (p-cymene)(L), where L is tricyclohexylphosphine, di-t-butylmethylphosphine, dicyclohexylphenylphosphine, triisobutylphosphine, triisopropylphosphine, or tri-n-propylphosphine) have been explored and have led to the development of a straightforward method for the preparation of new ruthenium indenylidene-ether ring-closing metathesis catalysts. The method involves the reaction between RuCl 2 (p-cymene)(PCy 3 ) and 1-(3,5dimethoxyphenyl)-1-phenylprop-2-yn-1-ol in refluxing THF and possesses many advantages. First, it is a one-step process from the starting materials. Second, it does not require the use of difficult-tohandle diazo compounds and does not need to be conducted at low temperature. Third, it consumes only one equivalent of the expensive PCy 3 ligand and is altogether very atom-economic. Fourth, it does not produce any inhibiting byproducts, allowing the catalyst solution to be used without further treatment. The resulting catalyst promotes the formation of 5-, 6-, and 7-membered disubstituted cycloalkenes with activities comparable to that of the commercial 1st generation Hoveyda catalyst under standard conditions.
Finally, a study of the effects of the phosphine ligand on the ring-closing metathesis activity of these new ruthenium complexes revealed that larger and more electron-donating phosphine ligands lead to more efficient catalysts.