Ruthenium Olefin Metathesis Catalysts Bearing a Macrocyclic N‐Heterocyclic Carbene Ligand: Improved Stability and Activity

Abstract Formation of sterically hindered C−C double bonds via catalytic olefin metathesis is considered a very challenging task for Ru catalysts. This limitation led to the development of specialised catalysts bearing sterically reduced N‐heterocyclic carbene (NHC) ligands that are very active in such transformations, yet significantly less stable as compared to general purpose catalysts. To decrease the small‐size NHC catalysts susceptibility to decomposition, a new NHC ligand was designed, in which two sterically reduced aryl arms were tied together by a C‐8 alkyl chain. The installation of this macrocyclic ligand on the ruthenium centre led to the formation of an olefin metathesis catalyst (trans‐Ru6). Interestingly, this complex undergoes transformation into an isomer bearing two Cl ligands in the cis‐arrangement (cis‐Ru6). These two isomeric complexes exhibit similarly high thermodynamic stability, yet different application profiles in catalysis.


Introduction Results and Discussion
The results discussed above show that ruthenium catalysts bearing N-aryl substituted NHC ligands lacking steric hindrance in the ortho position of the arene fragment, although generally more potent in the formation of tetrasubstituted CÀ C double bonds, can also give rise to intramolecular CÀ H insertion, leading to the formation of metathesis-inactive ruthenium complexes.In our design, we assumed that covalently connecting (tying) two N-aryl "arms" by a hydrocarbon chain of a proper length will limit the rotational freedom of the aryl fragments and possibly make the corresponding catalyst less prone to CÀ H activation and thus more stable (Figure 1c).The new NHC ligand was designed by a formal connection of two methyl groups in o-tolyl substituents present in the known SITol NHC ligand of Ru2 (Figure 1b) by an alkyl linker of a given length. [9]With the help of Dreiding models [10] (see Supporting Information for details) we predicted the minimal length of the tether.This simple "modelling" eventually led to the practical synthesis of the NHC ligand precursor 6 in which two phenyl substituents in the SITol-type structure are linked by a C-8 hydrocarbon chain (Scheme 1).
Commercially available acid 3 was converted into methyl ester, which was then alkylated with 1,6-diiodohexane to produce 4 after hydrolysis, decarboxylation, and reduction of the nitro group (Scheme 1).Key macrocyclisation was conducted with glyoxal under optimised conditions, yielding the corresponding diimine in 90 % yield.This product was then conveniently reduced with sodium borohydride leading to a macrocyclic 1,2-diamine 5 in good yield (77 % over 2 steps).Initially, the formation of imidazolinium salt 6 under classical conditions was lower yielding (50 %, see Supporting Information for details).Although such a yield can be considered quite satisfactory in the case of a macrocyclic architecture, [11] we decided to further optimize this step.The use of microwave irradiation (MW) and a finely powdered NH 4 BF 4 salt allowed us to obtain salt 6 in an improved yield of 85 % after 50 minutes of reaction (Scheme 1).
Having the macrocyclic NHC ligand precursor 6 in hand, we attempted to obtain the planned catalyst.Deprotonation of 6 with potassium tert-pentoxide followed by the addition of the Grubbs 1 st generation complex ((PCy 3 ) 2 -(Cl) 2 Ru=CHPh) led to the formation of complex Ru6, as a pink-brown solid, in 55 % isolated yield (Scheme 2A).The complex was stable in a solid form for several months (stored under inert atmosphere), as well as in DCM solution.However, when its solution in DCM/MeOH/H 2 O (originally purple) was deposited on silica gel and slowly eluted with the same solvent mixture, we noticed that a new green fraction appeared.The green band was then eluted from silica with DCM/MeOH/H 2 O, to yield, after recrystallisation from DCM, the new complex, obtained as a green solid in 45 % yield (Scheme 2, route b).The same green complex can be made directly from the ligand precursor 6 and Grubbs 1 st generation complex, in 37 % over two steps (Scheme 2, route c).We were fortunate to grow single crystals suitable for X-ray measurement from both of these complexes, which, in addition to the analysis of 1 H, 13 C, 31 P, and DEPT NMR spectra, allowed us to unambiguously assign the structures of both obtained products.(Scheme 2B).As a result, we found that in the presence of SiO 2 and methanol, the initially formed trans-Ru6 isomerises to a complex where both Cl ligands are in the cis-arrangement (cis-Ru6).Interestingly, in the absence of SiO 2 , trans-Ru6 is stable in DCM solution.
To describe the differences in a solid state between the newly obtained complexes and the known o-tolyl-NHC Ru catalysts (those with reported crystallographic structures), we defined specific valence angles (α 1 and α 2 ; see Scheme 2 and Figure S5).The average of reported α 1 and α 2 angles in known SITol-type ligands is 120(1)°and 121(1)°(see Supporting Information for details), whereas these angles are significantly smaller for trans-Ru6 (114.1(2)°and118.5-(2)°) and for cis-Ru6 (116.7(1)°and119.9(2)°).This simple analysis shows that the introduction of the aliphatic chain connecting the o-tolyl groups in the NHC ligand not only forces their syn conformation and blocks their free rotation, but also moves these arms up (which is witnessed by smaller angles α 1 and α 2 ).
The trans-to cis-dichloro isomerisation process of Ru6 is quite intriguing and requires further comment.In 2004, the first derivatives of the Hoveyda-type catalyst (a complex featuring a chelating benzylidene ligand) with cis-dichloro configuration have been reported. [12]Since then, a large number of cis-dichloro structures have appeared in the literature, most of them belonging to the Hoveyda-type family of chelated complexes. [13]n the other hand, it is quite difficult to obtain a related cis-dichloro Grubbs-type catalyst (a Ru complex featuring a non-chelating benzylidene and two strongly σ-donating and bulky ligands, such as PCy 3 or NHC) using the same design principle. [14]13e, 15] Therefore, we decided to study the isomerisation of the title Grubbs complex Ru6 using DFT approach in four different solvent models (toluene, DCM, methanol, and water), focusing on the relative stabilities of the trans and cis isomers (for details see Supporting Information).Interestingly, the trans isomer was predicted to be slightly favoured over the cis isomer in toluene (by 2.7 kcal mol À 1 ), but in methanol the cis isomer was predicted to be favoured over the trans isomer (by 2.2 kcal mol À 1 ; for more details on the analysis of cis-trans isomerisation, see Supporting Information).Next, we considered that the concerted mechanism of isomerisation, where the trans isomer converts to the cis one through a single transition state, is operational in this case.Thus, the phosphine ligand (PCy 3 ) changes its position in a manner similar to the Berry pseudorotation mechanism. [16]This mechanism seems, according to our calculations, to be preferred over an alternative one-step mechanisms in which the phosphine ligand is moved close to the trans position with respect to one chloride ion, while the second chloride shifts to the trans position with respect to the Ru=C bond (for details see Supporting Information).The estimated Gibbs free energy for the transition state of isomerisation (25.6-25.8kcal mol À 1 depending on the solvent) is similar to the energy barrier for trans-cis isomerisation of a Hoveydalike system obtained earlier, [17] and are in good agreement with the observed experimental results, suggesting a very slow isomerisation. [18]eturning to the bench chemistry, we were pleased to see that both complexes featuring the macrocyclic, smallarms NHC ligand are very stable, not only in a solid form

Research Articles
but also in a solution.As can be seen from Figure 2, the observed stability significantly extended the stability of the structurally related small-NHC benzylidene complex Ru2.Surprisingly, the stability of the newly obtained complexes bearing "tied" aryl arms was as high as the one of complex Ru3 (it is known that the phosphine-free Hoveyda-Grubbstype complexes exhibit in general higher thermal stability comparing to the corresponding Grubbs' systems).The unique stability of phosphine-containing Ru6 complexes makes them potentially interesting in catalysis, especially under higher-temperature conditions (see below).
Although similar in stability at room temperature, the two isomers of Ru6 exhibited different catalytic properties.In the challenging ring-closing metathesis (RCM) leading to carbo-and heterocycles 2 a-2 f featuring tetrasubstituted CÀ C bonds, cis-Ru6 gave reproductively better results than its trans-isomer (Scheme 3, rows A-B).In these challenging reactions cis-Ru6 outdistanced also its closest structural analogue-the phosphine-containing catalyst Ru2 [3a] -offering 9 to 56 pp (percent points) higher conversions.Noteworthy, data reported in the literature [5] allows to compare Ru6 also with the state-of-the-art complex bearing a medium-size NHC, Ru7. [5,19]Interestingly, while this trend is preserved for the derivative of isopulgeol (2 f), the RCM of a diene 1 g-bearing one terminal and one geminally disubstituted CÀ C double bond, based on a sesquiterpene alcohol, bisabolol-leads to almost quantitative conversion for both isomers of Ru6.3a] Subsequently, we focused on even more challenging substrates, precursors of products 2 h-2 i. Requested by one of the Referees, we added to the pool of tested catalysts also the phosphine-free relatives-the known SITol Hoveydatype complex (Ru3), and its macrocyclic-NHC ligand-based analogue Ru8 (for its synthesis and crystal structure, see Supporting Information).To push the starters of this race to their limit of efficiency, we switched to more forcing conditions, consisting of using a microwave reactor and perfluorotoluene as a solvent. [20]We decided also to include in this study the model malonate substrate 1 a again, [21] which gave with cis-Ru6 the quantitative conversion, surpassing the efficiency of trans-Ru6, as well as Ru2, Ru3, and Ru8 under the same conditions (Scheme 3, row C).Also for another problematic substrate (1 h), that contains an electron-poor CÀ C double bond, we found that complex cis-Ru6 affords the highest conversions, exceeding the efficiency of other catalysts in a range from 9 to 34 pp (percent points).Noteworthy, in this case the SITol-bearing Ru2 (Grubbstype) and Ru3 (Hoveyda-type) were found to be completely ineffective.It should be noted that under the same conditions, the macrocyclic-NHC catalysts Ru6 and Ru8 gave conversions up to 35 % and 26 %, respectively.Importantly, the conversion toward 2 h could be further increased to 53 %, as demonstrated for cis-Ru6 by adding the catalyst portion-wise (2 × 2 mol %, Scheme 3, row C).In turn, the formation of chloro-substituted 2 i revealed also the advantage of catalysts bearing the macrocyclic-NHC ligand in comparison to their SITol analogues within the same catalyst type (Grubbs-type: Ru2 vs. Ru6 and Hoveydatype: Ru3 vs. Ru8).Similarly to the previous case, when cis-Ru6 was used in two portions (in total 2 × 2 mol %), the conversion toward chlorocyclohexene 2 i become significantly increased (to 73 %, Scheme 3, row C).
It should be noted that the beneficial properties of cis-Ru6 are accompanied by its latent character, well-illustrated during the formation of polyfunctional cycloalkenes 2 j and 2 k (Scheme 3 row D).In these two (rather straightforward) model RCM transformations [22] cis-Ru6 required a visibly longer time to achieve complete conversion compared to Ru2 and trans-Ru6 (Scheme 3, row D).The different performance of cis and trans-Ru6 complexes was also demonstrated in the RCM reaction of model challenging substrate 1 a conducted at room temperature (Scheme 3, row D).We observed that trans-Ru6 activates faster than cis-Ru6 (42 % conversion of 1 a after 4 h versus 3 %, respectively, for details see Figure S8 in Supporting Information), however, the latter is more stable in the reaction environment-being still present in the reaction mixture after 25 days.This leads to a much slower reaction at rt but better final yield compared with that obtained in the reaction catalysed by the trans isomer (Scheme 3, row D).Therefore, the superior properties of cis-Ru6 over trans-Ru6 in challenging metathesis transformations may originate from its slower activation, leading to higher stability in the reaction mixture.Interestingly, Hoveyda-type Ru8 led to very good results in tetrasubstituted CÀ C double bond formation not only at 120 °C but also at rt-needing of course much longer time (for details see Figure S9 in Supporting Information and compare results presented in Scheme 3, rows C and D). [23]ext, macrocyclic-NHC complexes were tested in crossmetathesis (CM) reactions (Scheme 4a-e).In a model reaction between challenging gem-disubstituted alkene 7 a [24] and (Z)-2,3-buten-1,4-diol diacetate (8 a) known complexes featuring SITol ligand, Ru2 and Ru3, gave the expected Scheme 3. Ring-closing metathesis scope and limitation study.The conversion was determined by GC (for substrates 1 a-e and 1 j-k) or by 1 H NMR (for substrates 1 f-i), for both techniques durene was used as an internal standard (for details see Supporting Information).[a] 0.5 mol %, 60 °C, PhMe, data taken from Ref. [5].[b] 0.1 mol %, 60 °C, PhMe, data taken from Ref. [5].[c] 4 × 0.25 mol %, 60 °C, PhMe, data taken from Ref. [5].
[d] Accompanied with the formation of higher cyclic and oligomeric products, for details, see Supporting Information.

Research Articles
product within 70 % yield.Notably, 10-24 pp higher yields (80 and 94 %) were obtained with trans-Ru6 and cis-Ru6, respectively.The Hoveyda-type analogue Ru8 gave in the same reaction 81 % yield.Note that this CM was conducted under microwave irradiation, using previously established conditions (cf.Scheme 3).Both trans-and cis-Ru6 showed an advantage over commercial Ru2 in CM of sabinene (9 a), a natural bicyclic monoterpene (Scheme 4b).Interestingly, in CM of an even more challenging geminal substrate 10 a, [25] trans-and cis-Ru6 as well as Ru2 were found to be similarly productive, leading to yields of 57-67 % (Scheme 4c).In contrast, a derivative of a complex polyfunctional fusidic acid (11 a) reacted with 8 a relatively easily, giving product 11 b in a yield of 93-98 % with both trans-and cis-Ru6 and Ru2 (Scheme 4d).In CM of 12 a with an electron-deficient partner, [26] tert-butyl acrylate, latent cis-Ru6 was initially found to be less productive, however, at higher temperature, led to a very good yield (94 %) of the expected estrone derivative (E)-12 b, accompanied by only a tiny amount of the unwanted self-CM product 12 b' ("homodimer"). [27]inally, in a catalytic enyne cycloisomerisation, trans-Ru6 exhibited good reactivity, identical or slightly higher than the one displayed by benchmark SITol-bearing Ru2 (Scheme 4f, g).Interestingly, the more dormant cis-Ru6 required a higher temperature to operate (Scheme 4f, g, conditions d).
As the final test, we decided to perform the RCM reaction leading to sulphonamide 15 b, a relative of sildenafil -a drug sold inter alia under the trade-name Viagra™ (Scheme 5).Despite the similarity of substrate 15 a to a sulphonamide 2 b tested previously, we saw this substrate, due the presence of a number of Lewis basic groups in its structure, as a challenging target interesting in context of pharmaceutical chemistry.Applying a set of catalysts under previously established MW conditions (scale 0.1 mmol) leads to the following order of efficiency of the tested complexes: -Ru6 > trans-Ru6 > Ru2.Noteworthy, using cis-Ru6 in two 0.5 mol % portions allows to obtain analytically pure 15 b at larger scale (1.6 mmol) in 84 % of isolated yield without column chromatography.

Conclusion
An N-heterocyclic carbene (NHC) ligand bearing two sterically reduced aryl arms connected by a C-8 linker was obtained.Installation of this macrocyclic ligand on a ruthenium centre led to the formation of a ruthenium benzylidene complex trans-Ru6.Interestingly, in the presence of methanol and silica, this complex undergoes isomerisation into cis-Ru6, an isomer containing two Cl ligands in a cis-arrangement, a geometry not very typical for classical Grubbs-type phosphine-containing benzylidene Ru complexes.Importantly, both of these complexes are significantly more stable than the structurally related small-NHC catalyst Ru2, and generally exhibit a very good activity profile in olefin metathesis.In addition, cis-Ru6 that displays a dormant character and can be assumed to be a latent metathesis catalyst, is particularly active in the challenging formation of tetrasubstituted CÀ C double bonds.Because the formation of sterically hindered olefins was always the Achilles' heel of ruthenium metathesis catalysts, the trait exhibited by cis-Ru6 seems to be of interest.
The results described herein show that, while it is well known that ruthenium catalysts bearing N-aryl substituted NHC ligands lacking steric hindrance in the ortho position of the arene substituent gave rise to intramolecular CÀ H insertion, leading to the formation of metathesis-inactive ruthenium complexes, connecting of these two aryl arms by a hydrocarbon linker can significantly stabilise the resulted catalyst.We believe that this observation is of importance for the further development of olefin metathesis catalysts.
Supporting Information: Detailed experimental procedures, copies on NMR spectra, computational methods, Cartesian coordinates of DFT-optimized structures, and crystallographic details.Crystal structures have been deposited with the Cambridge Crystallographic Data Centre CCDC). [28]The data can be obtained free of charge via www.ccdc.cam.ac.uk/structures.

Figure 1 .
Figure 1.a) Challenging formation of a tetrasubstituted CÀ C double bond by olefin metathesis and a popular Grubbs' NHC-bearing catalyst Ru1.b) Selected complexes Ru2-Ru4 featuring sterically reduced NHC ligands and one of multiple decomposition pathways of these catalysts.[6]c) Proposed catalyst Ru6 bearing an aryl-tied NHC ligand.

Figure 2 .
Figure 2. Catalysts stability in DCM-d 2 under argon at rt. Determined by 1 H NMR, with 1,3,5-trimetoxybenzene as an internal standard.Lines are visual aid only.